See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/263074101

Diversity, composition and density of trees and shrubs in agroforestry

homegardens in Southern Ethiopia

Article  in  Agroforestry Systems · December 2013

DOI: 10.1007/s10457-013-9637-6

CITATIONS READS

52 1,684

4 authors, including:

Tesfaye Abebe Amdie Frank Sterck

Hawassa University Wageningen University & Research
35 PUBLICATIONS   570 CITATIONS    152 PUBLICATIONS   6,072 CITATIONS   

SEE PROFILE SEE PROFILE

Frans Bongers
Wageningen University & Research
381 PUBLICATIONS   24,414 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Ecology and management of pine resin production in a biosphere reserve in Chiapas, MX View project

The hydraulic properties and water use for tropical lianas and co-occurring trees View project

All content following this page was uploaded by Frans Bongers on 08 May 2015.

The user has requested enhancement of the downloaded file.

Diversity, composition and density of trees
and shrubs in agroforestry homegardens in
Southern Ethiopia

Tesfaye Abebe, F. J. Sterck,

K. F. Wiersum & F. Bongers

Agroforestry Systems
An International Journal incorporating
Agroforestry Forum

ISSN 0167-4366
Volume 87
Number 6

Agroforest Syst (2013) 87:1283-1293

DOI 10.1007/s10457-013-9637-6

1 23
Your article is protected by copyright and all
rights are held exclusively by Springer Science
+Business Media Dordrecht. This e-offprint
is for personal use only and shall not be self-
archived in electronic repositories. If you wish
to self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com”.

1 23
 Author's personal copy
Agroforest Syst (2013) 87:1283–1293
DOI 10.1007/s10457-013-9637-6
Diversity, composition and density of trees and shrubs
in agroforestry homegardens in Southern Ethiopia
Tesfaye Abebe • F. J. Sterck • K. F. Wiersum
F. Bongers
Received: 28 March 2013 / Accepted: 5 August 2013 / Published online: 13 August 2013
Ó Springer Science+Business Media Dordrecht 2013
Abstract Diversity of trees and shrubs in agricul-
tural systems contributes to provision of wood and
non-wood products, and protects the environment,
thereby, enhancing socioeconomic and ecological
sustainability of the systems. This study characterizes
the diversity, density and composition of trees in the
agroforestry homegardens of Sidama Zone, Southern
Ethiopia, and analyses physical and socioeconomic
factors influencing diversity and composition of trees
in the systems. A total of 144 homegardens were
surveyed from 12 sites. In total, 120 species of trees
Electronic supplementary material The online version of
this article (doi:10.1007/s10457-013-9637-6) contains supple-
mentary material, which is available to authorized users.
T. Abebe (&)
College of Agriculture, Hawassa University, P.O. Box 5,
Hawassa, Sidama, Ethiopia
e-mail: tesfayeabe@yahoo.com; tesfayea@hu.edu.et
F. J. Sterck  F. Bongers
Centre for Ecosystem Studies, Wageningen University
and Research Centre, P.O. Box 47, 6700 AA Wageningen,
The Netherlands
e-mail: frans.bongers@wur.nl
K. F. Wiersum
Forest and Nature Conservation Policy Group,
Wageningen University and Research Centre, P.O. Box
47, 6700 AA Wageningen, The Netherlands
e-mail: freerk.wiersum@wur.nl
•
and shrubs were recorded of which, 74.2 % were
native to the area. The mean number of tree species per
farm was 21. Density of trees varied between sites with
mean values ranging from 86 to 1,082, and the overall
average was 475 trees ha-1. Four different crop-based
enset (Enset ventricosum (Welw.) Cheesman)-coffee
homegarden types were recognized and they differed
not only in the composition of major crops but also in
the diversity, density and composition of trees. The
composition, diversity and density of trees is influ-
enced by physical and socioeconomic factors. The
major physical factors were geographical distance
between sites and differences in altitude of farms. The
most important socioeconomic factors were farm size
and access to roads. Tree species richness and density
increased with farm size. Increased road access
facilitated marketing opportunities to agricultural
products including trees, and lead to a decline in the
basic components of the system, enset, coffee and
trees. In the road-access sites, the native trees have
also been largely replaced with fast growing exotic
species, mainly eucalypts. The decrease in diversity of
trees and perennial components of the system, and its
gradual replacement with new cash and annual food
crops could jeopardize the integrity and complexity of
the system, which has been responsible for its
sustenance.
Keywords Determinants of on-farm tree
diversity  Enset-coffee homegardens  On-farm
density of trees  Landuse changes  Sidama Zone
123
 Author's personal copy
1284
Introduction
Homegardens are multispecies agroecosystems where
different herbaceous and tree crops as well as trees are
managed in integration (Kumar and Nair 2006). They
provide numerous economic benefits to rural commu-
nities in the tropics and also serve as intermediary for
biodiversity conservation. In some cases, homegarden
agroforestry practices were found to be equally
effective as natural forests in conservation of tree
species diversity (Bardhan et al. 2012). Two types of
homegardens can be recognized in the tropics. The
first and common one consists of small, supplementary
food production units around houses, such as the
homegardens of Java (Wiersum 1982; Soemarwoto
1987), where there are additional farm fields for food
production. The second type is composed of extended
farm systems located around the house(s) from where
farmers derive their subsistence and cash needs, and
where they do not have additional land in other land
use systems. Examples of such homegardens are
reported from the highlands of Uganda (Oduol and
Aluma 1990), Tanzania (Rugalema et al. 1994) and
Southern Ethiopia (Abebe et al. 2006, 2010). The
homegardens presented in this paper belong to the
second category.
The homegardens of the highlands of Southern
Ethiopia have coffee as a dominant crop, like many of
their counterparts in other eastern African countries.
What makes them different is the unique association of
coffee with the co-dominant native crop enset (Enset
ventricosum (Welw.) Cheesman). Enset, the false
banana, is a perennial herbaceous crop and a staple
food for about 15 million people in the southern and
south-western parts of Ethiopia. These homegardens
have a complex multistorey structure where different
species of crops, trees and livestock are integrated in
an intimate association. While enset is the staple food,
other food crops, including maize, vegetables, root and
tuber crops such as sweet potato and yam, pulses,
fruits and other cereals are also grown (Abebe et al.
2010). Coffee is the major cash crop, but in some areas
khat (Catha edulis (Vahl.) Forssk. ex. Endl.) is
increasingly important as cash crop. These homegar-
dens are also called ‘Enset-coffee homegardens’ after
the two dominant crops. Trees and shrubs are very
important components of the enset-coffee homegar-
dens. They provide the households with wood and
non-wood products and cash, and also play a vital role
123
Agroforest Syst (2013) 87:1283–1293
in maintaining and enhancing the physical environ-
ment needed to sustain crop production.
Homegardens have been considered as sustainable
and productive agricultural systems supporting very
dense populations as compared to other agricultural
systems (Soemarwoto and Conway 1991; Torquebiau
1992; Kumar and Nair 2006; Michon et al. 2007).
However, recent trends in homegardening are indi-
cating changes in their structure and composition, and
this has been attributed to changes in the overall
farming system and livelihood strategies of homegar-
deners (Wiersum 2006; Peyre et al. 2006). The
changes in the structure, diversity, density and com-
position are often associated with changes in the
functions of homegardens. For instance, access to
market leads to transformation of subsistence gardens
to commercial ones (Abdoellah et al. 2006; Wiersum
2006). This transformation also affects the tree
components of homegarden: diversity and density of
trees decreases as a result of the farmers’ focus on
high-value crops (Wiersum 1982; Jensen 1993). The
level of household resources, such as land holding, is
also expected to influence diversity and density of
trees because resource poor farmers attach priority to
subsistence crops for their survival (Abebe et al. 2006;
van der Wal and Bongers 2013).
A recent study on homegardens of Southern
Ethiopia (Abebe et al. 2006, 2010) has shown a
changing trend in the structure, composition and
function of the gardens owing to improved market
access and land fragmentation. Production of cash
crops and annual food crops is expanding at the
expense of the other homegarden components (Abebe
and Bongers 2012). Due to the changes in the
composition of major crops, four different enset-
coffee homegarden types were identified: (a) enset-
coffee-maize, (b) enset-coffee-maize-sweet potato,
(c) enset-coffee-maize-khat, and (d) enset-coffee-
maize-pineapple-khat (Abebe et al. 2006, 2010).
However, it is not clear what effects this homegarden
dynamics has on the diversity and density of trees. It is
also not known whether these different homegarden
types vary in tree species composition, diversity and
density. The present study aims at characterizing the
diversity of trees and shrubs in the agroforestry
homegardens of Southern Ethiopia, and analyzes the
factors that influence tree diversity and density. More
specifically, this paper answers the questions (a) What
is the diversity and density of trees at site, farm and
 Author's personal copy
Agroforest Syst (2013) 87:1283–1293
farm unit levels? (b) Do homegarden types differ in
tree diversity and density? (c) Do sites differ in tree
species composition? and (d) Which factors determine
tree diversity and density in these systems?
Materials and methods
The study area
The research was conducted in Sidama administrative
zone, Southern Nations, Nationalities and Peoples’
Regional State (SNNPRS), Ethiopia (Fig. 1). It was
selected for the study due to its representativeness with
respect to the enset-coffee homegarden production
systems as well as the prevailing population pressure
in the highlands (Abebe 2005). Sidama is located
within 5°450 –6°450 N and 38°–39°E, and covers a total
area of 7,672 km2 (SZPEDD 1997). It has a population
of about 3.0 million people, out of which 93 % is
rural. The altitude of Sidama Zone ranges from 500 to
3,500 m.a.s.l., but the sub-humid to humid, warm
subtropical climatic zone (1,500–2,300 m.a.s.l.) is the
most important in terms of productivity and preva-
lence of the homegardens. This agroecological zone
receives an annual rainfall of 1,000–1,600 mm, and
enjoys a mean annual temperature of 15–20 °C. It
constitutes 54 % of the total land area of Sidama Zone
(SZPEDD 1997).
The study was carried out in 12 sites, known as
Kebele1 (Table 1) selected from four districts where
homegardening is practised extensively. The 12 sites
were selected purposefully to represent variations in
market access and cropping systems. From each site,
12 households were selected, four from each wealth
category of farmers (poor, medium and rich) making a
total of 12 households per site and 144 households for
the whole study. The reason for considering wealth
category was based on the assumption that the
variation in household resources could influence tree
management and utilization practices. To select rep-
resentative households, wealth ranking of all house-
holds in each site was made with key informants, and
then from each group four households were selected
randomly. The criteria used in wealth ranking of
1 units/fields from one household) and farm unit. The
A ‘‘Kebele’’ is the lowest administrative unit in Ethiopia. It
covers an area of about 800 hectares usually inhabiting 400–800
households, and it is administered by elected farmers.
1285
Fig. 1 Map of Sidama administrative zone in Southern
Ethiopia
households were, (a) farm size, (b) number of
livestock, (c) area of cash crops especially coffee
and khat, and (d) number of matured enset plants.
Data collection
Data were collected from 144 sample household
farms/homegardens. We evaluated physical and socio-
economic farm variables that were expected to affect
diversity, density and composition of trees. Among the
physical factors, geographical distance, altitude and
slope are considered important as they lead to different
site and micro-site conditions that could affect the
diversity and composition of trees. Thus, sites as far
apart as 80 km were selected, and farms with different
altitudes (1,540–2,040 m.a.s.l.) and slope (0–55 %)
were represented in the samples. Proximity to roads
and market were expected to enhance marketability of
agroforestry products and thus farms at different
distances from roads (0.04–26 km) and local markets
(0.02–6.0 km) were selected to assess their effect on
diversity and density of trees. At farm level, farm size,
available labour force, involvement of family mem-
bers in off-farm activities as well as presence of
woodlots are expected to influence tree diversity and
composition and they were represented by selecting
households with different economic status.
Data were collected at two levels; farm (=all farm
homegardens often display a mosaic of patches or
farm units which are distinct from one another because
123
 Author's personal copy
1286 Agroforest Syst (2013) 87:1283–1293

Table 1 Tree diversity of farms at the research sites (Kebeles)

Site (PA) Number of tree species Number of trees Shannon’s index (H0 ) Evenness (E)
Total Mean SD per farm per ha

Setamo 52 19.7cd 8.9 528b 309bc 1.62ab 0.58ab

ab b abc ab
Shoyicho 66 28.4 9.6 912 592 1.49 0.46abc
Qumato 31 14.2de 4.2 187b 113c 1.21bc 0.47abc
a a a ab
Belesto 72 33.4 13.8 3,791 1,081 1.53 0.46abc
cd b bc ab
Lela Honcho 52 19.7 5.0 442 315 1.48 0.51abc
de b c ab
Tesso 58 14.0 6.7 125 86 1.60 0.64a
bc b ab bc
Sheyicha 62 22.9 8.6 1,699 728 1.29 0.43bc
Ferro I 66 28.5ab 7.4 729b 347bc 1.88a 0.58ab
cde b a bc
Ferro II 58 18.2 7.2 559 1082 1.27 0.46abc
bc b bc bc
Tula Aposto 69 25.3 9.6 572 454 1.21 0.39c
e b bc c
Chefasine 32 11.5 4.3 401 398 0.86 0.37c
de b bc ab
Abela Tula 43 12.8 5.3 312 198 1.44 0.59ab
Mean (n = 144) 120 20.7 10.3 855 475 1.41 0.50
F test (P) \0.001 \0.05 \0.001 \0.01 \0.05
In each site 12 farms were analysed
Note Means followed by different letters are significantly different at P \ 0.05, according to Duncan’s Multiple Range Test (DMRT)

of the dominant crop grown on it. Such patch
formations in homegardens have also been reported
by Mendez et al. (2001), who used the term ‘micro
zones’ to describe them. The major farm units
considered in this study were, (a) home and grazing,
(b) enset, (c) coffee, (d) woodlots, (e) maize, (f) khat,
(g) sugarcane, (h) sweet potato and (i) pineapple.
Total area of the farm, and area of each farm unit
was measured and the different tree species grown in it
counted and listed using local and botanical names.
For those species that were difficult to identify at field
level, plant specimens were collected and submitted to
the National Herbarium in Addis Ababa for identifi-
cation. For the purpose of species richness calcula-
tions, all tree and shrub species in the farms were
considered. But in counting the population (individ-
uals) of each tree species, only trees with a minimum
breast-height diameter (dbh) of 5 cm were taken into
account. It should be noted that the homegardens
stated in this paper are equivalent to a farm system.
Data analyses
Analyses of the data were carried out using quantita-
tive and qualitative methods. To determine tree
species richness of each farm and farm unit, the total
number of tree species on the farm and farm unit were
calculated. This index does not indicate the relative
proportion or abundance of a particular species in the
farm. Hence, models that incorporate both richness
and the evenness of relative abundance were required.
Shannon index (Shannon and Wiener 1949) and
Evenness measure (E), which are commonly used
tools for these purposes (Pielou 1969; Magurran 1988;
Huston 1995), were calculated.
Shannon diversity index (H0 ) is calculated using the
P
formula, H0 = - pi ln pi (Magurran 1988), where pi
is the proportion of individuals composed of species i.
As a measure of heterogeneity, Shannon’s index takes
into account the evenness of abundance of species
(Peet 1974). However, an additional measure of
evenness (E), which compares observed distribution
with the maximum possible even distribution of the
number of species in the sample (Pielou 1969) was
calculated. The measure of evenness (E) is the ratio of
observed diversity to maximum diversity and it is
calculated as, E = H0 /Hmax, = H0 /ln S (Magurran
1988). E has values between 0 and 1.0, where 1.0
represents a situation in which all species are equally
abundant.
From these calculations, species richness and
heterogeneity as well as density of trees were charac-
terized for each farm. Density of trees refers to the
number of individual trees and shrubs per hectare and

123
 Author's personal copy
Agroforest Syst (2013) 87:1283–1293 1287

per farm. The values obtained at site and farm levels
were statistically analyzed using descriptive statistics
and one way ANOVA. Post-hoc pairwise comparisons
were made with Fisher’s least significance difference
(LSD) tests at a = 0.05 to isolate group means that
show significant differences.
Similarity between sites was calculated using
Jaccard’s similarity, given by 1-Jaccard dissimilarity
index as implemented in the R-package ‘Vegan’
(Oksanen et al. 2012). The similarity values vary
between 1 (sites being completely the same) and 0
(sites being completely different).
Four types of the enset-coffee homegarden agro-
forestry systems were identified on the basis of the
composition of major crops (Abebe et al. 2006). Tree
species diversity as well as tree density of farms were
compared among these types. The variation in species
composition among the sites was described using
detrended correspondence analysis (DCA, as imple-
mented in CANOCO version 4.5 (Ter Braak and
Smilauer 2002). The effects of physical and socioeco-
nomic environments on the diversity indices were
determined through multiple linear regression analy-
sis. The dependent variables (Species richness, Shan-
non index (H’), Evenness index (E), number of trees
per farm and number of trees per hectare) were
regressed with the physical environments (altitude
and slope of farms) and socioeconomic environments
(distance to markets and major roads, farm size, area of
woodlots, number of farm labour force and involve-
ment in off-farm activities). F Tests were used to detect
the level of significance. Statistical Package for Social
Sciences-SPSS version 17 (SPSS Inc. 2008) was used
for most analyses, while the similarity analyses were
done using R (R Development Core Team 2012).
camaldulensis Dehn. (92.4 %), Persea americana
Miller. (86 %), Milletia ferruginea (Hochst.) Baker
(85 %), and Euphorbia candelabrum Trem. & Kots-
chy. (85 %) (Appendix 1). On the contrary, 19 tree
species were very rare each occurring only in one of
the farms. Out of the total number of woody species,
74.2 % were indigenous.
Tree species richness varied across sites (Kebeles)
(F test, P \ 0.001) (Table 1). Among the sites,
Belesto, from Aleta Wondo district had the highest
number of species (72) accounting for 61 % of the
total number. Farm level species richness is also
highest at the same site. At this site, the average
number of tree species per farm was 33.4. For the
whole samples in the study, the number of tree species
per farm varied from 4 to 55 with an overall average of
21. Shannon’s index of diversity (H’) showed a mean
value of 1.41 (range: 0.86–1.88), while the measure of
evenness (E) was 0.50 (range: 0.37–0.64). The average
density of trees varied widely among the sites with
mean values ranging from 86 (Tesso) to 1,082 (Ferro
II) trees ha-1, the overall average being 475 tree-
s ha-1 (Table 1).
Diversity of trees at farm units
Tree species richness of the different farm units varied
widely (Table 2). The coffee units, which cover the
largest proportion of the farms (36.6 %), are the
richest in tree species (13.3) followed by enset (8).
Coffee units/fields, even when small, are associated
with more trees but units of pineapple, khat, sugar-
cane, sweet potato and maize of any size always have
few tree species or no trees at all. The home and

Table 2 Mean number of tree species associated with the

different unit types
Results
Farm spatial units n Number of species SD
Farm level diversity and density of trees and shrubs Home and grazing 144 7.20 4.98
Enset 142 8.03 4.83
In total 120 tree and shrub species were recorded in the Coffee 140 13.27 8.28
agroforestry homegardens of Sidama Zone, Southern Woodlots/trees 69 7.57 7.80
Ethiopia. Frequency of occurrence of the species Maize 59 4.08 2.28
across homegardens was rather variable, but six Khat 42 2.46 1.45
species occurred in over 85 % of the homegardens.
Sugarcane 28 2.64 1.45
Coffea arabica L. was the most frequent species
Sweet potato 27 3.58 1.51
occurring in all (100 %) of the farms. It is followed by
Pineapple 25 2.40 1.48
Cordia africana Lam. (95.1 %), Eucalyptus

123
 Author's personal copy
1288 Agroforest Syst (2013) 87:1283–1293

grazing unit include the homestead and its front yard species richness and number of trees of farms increased
which is used for grazing. In the front yards, native significantly (P \ 0.001) with farm size. As area of
timber species such as Podocarpus falcatus (Thunb.) woodlots increased, the density of trees increased
Mirb. are kept scattered in wide spacing so that they significantly (P \ 0.001) but the diversity (H’) and
provide shade to humans and animals, without Evenness decreased. Distance to local markets did not
hindering the passage of sufficient light for proper have any effect on the diversity and density of trees.
growth of the grass. Boundaries of the front yards,
which are often boundaries of the farm, are planted
with live fences of E. candelabrum Trem. & Kotschy. Discussion
or Cupressus lusitanica Mill., and closely spaced rows
of eucalypts and sometimes Podocarpus trees. The Tree species diversity of farms
density of trees along the boundaries is high but the
number of species is small. The tree species richness of Sidama homegardens
Woodlots (tree units) were encountered on 44 % of (120) is very high compared to similar highland
the sample homegardens. An average of 7.6 tree agroecosystems such as the Chagga homegardens in
species occurred in the woodlots, and this is because Tanzania, where a total of 53 woody species were
the woodlots are dominated by eucalypts. reported (Fernandes et al. 1984). Homegardens in
tropical highlands are expected to have lower diversity
Tree composition and diversity of homegarden compared to those in lowland humid tropics because
types the latter enjoy very high rainfall and temperature.
This indeed generally is the case: 129 tree species in
Tree species composition varied strongly among Kerala, India (Kumar et al. 1994), 168 in the Peruvian
farms. A DCA showed large differences in species
composition among individual farms (Fig. 2), and
similarity in the composition of tree species decreased
with increasing geographical distance and elevation
difference (Fig. 3). Part of the (dis)similarity among
farms is related to the homegarden type. In the DCA
the different types of enset-coffee homegardens
clearly spreaded (Fig. 2). Tree species richness varied
significantly (P \ 0.001) across homegarden types:
the subsistence-oriented enset-coffee-maize-sweet
potato type had the highest number of tree species
(25.3) followed by enset-coffee-maize type (24.4)
(Table 3). Also, the density of trees varied signifi-
cantly (P \ 0.001) with enset-coffee-maize type ho-
megardens having the highest density of trees
(636 trees ha-1). The remaining two cash crop ori-
ented homegarden types have the lowest number of
tree species and tree density.

What factors influence diversity and density

of trees in the homegardens?
Physical and socioeconomic factors influenced farm
level species diversity and density of trees (Table 4).
Farms located near roads had less tree species, lower
diversity and lower density. The total number of trees
as well as their density increased with elevation. Tree
Fig. 2 Detrended correspondence analysis (DCA) ordination
of farms based on tree species composition and abundance
(n = 144). Symbols indicate homegarden prototypes, based on
crop species (Abebe et al. 2006). Stars Type A, homegardens of
enset-coffee-maize type; open circles Type B, homegardens of
enset-coffee-maize-chat type; filled Type C, homegardens of
enset-coffee-maize-sweet potato type; open squares Type D,
homegardens of enset-coffee-maize-chat-pineapple type

123
 Author's personal copy
Agroforest Syst (2013) 87:1283–1293 1289

Fig. 3 Similarity in
composition of tree species
in relation to elevation
(altitudinal) difference and
geographical distance of the
sites (n = 66). Jaccard
similarity index is given,
which is 1 - D, where D is
Jaccard dissimilarity as
implemented in the
R-package ‘vegan’
(Oksanen et al. 2012)

Table 3 Mean values of Species richness, Shannon and Evenness indices and density of trees at the different types of the enset-
coffee agroforestry homegardens
Homegarden types Mean no. of tree species per Shannon index Evenness index Number of trees
farm (H0 ) (E)
per farm per ha
a a ab a
A. Enset-coffee-maize (n = 84) 24.4 1.51 0.50 776 636a
b a ab ab
B. Enset-coffee-maize-chat 12.2 1.15 0.48 357 298ab
(n = 24)
C. Enset-coffee-maize-sweet potato 25.3a 1.21a 0.39b 572ab 454ab
(n = 12)
D. Enset-coffee-maize-Chat-Pineapple 14.1b 1.41a 0.55a 156b 100b
(n = 24)
Mean (n = 144) 20.7 1.41 0.50 855 475
F test P \ 0.001 P \ 0.05 n.s. P \ 0.05 P \ 0.001
Note Means followed by different letters are significantly different at P \ 0.05, according to Duncan’s Multiple Range Test (DMRT)

Amazon (Padoch and Jong 1991) and 179 in West Java
(Soemarwoto 1987), but also 69 in the compound
farms of Nigeria (Okafor and Fernandes 1987) and 83
in Nicaragua (Mendez et al. 2001).
The mean number of tree species per farm was 21,
which again is high compared to other studies: Kerala
homegardens have 11–39 tree species (Kumar et al.
1994), Rwandan homegardens have 12–34 (Biggelaar
and Gold 1996). The high number of tree species in
Sidama homegardens, 74 % of which are native to the
region, indicates the significant role of these agrofor-
estry systems in the conservation of genetic diversity,
as it was reported also in homegarden studies in other
regions (Michon et al. 1983; Alvarez-Buylla Roces
et al. 1989; Soemarwoto and Conway 1991; Kessy
1998; Bardhan et al. 2012; Jose 2012; Negash et al.
2012). These trees serve several functions. They
provide firewood, timber, wood for different purposes
(local construction, farm implements, household uten-
sils), fodder, food, medicine, and they also play
beneficial ecological roles such as erosion control and
soil fertility improvement (Appendix 1). The presence
of many tree species and shrubs of different utilities
could, therefore, contribute to diversification of tree
products and sustenance of agricultural systems.
The diversity indices constructed to compute
heterogeneity of the tree species generally showed
that few species are more abundant than others. On the
whole, evenness of tree species was 50 % of what it
could have been under uniform distribution. Such a
distribution is expected in agroecosystems, since
farmers have preferences to more valuable species.
The evenness values obtained here are comparable to
values reported by Kumar et al. (1994) for the
homegardens of Kerala where evenness values ranged
from 0.24 to 0.71.

123
 Author's personal copy
1290 Agroforest Syst (2013) 87:1283–1293

Table 4 Multiple stepwise linear regression of tree diversity and density indices with physical and socioeconomic factors
Factors Species richness No. of trees No. of Shannon’s Evenness (E)
(S) per farm trees (ha-1) index (H0 )

Model (Adjusted R2) 0.53*** 0.46*** 0.26*** 0.20*** 0.23***

Physical environment
Altitude of the farm ns 0.15* 0.24** ns ns
Slope of the farm 0.15* ns ns ns ns
Socio economic environment
Distance to markets ns ns ns ns ns
Distance to major roads 0.27*** 0.19* 0.22** 0.22** ns
Farm size 0.40*** 0.39*** ns ns ns
Area of woodlot 0.14* 0.40*** 0.41*** -0.46*** -0.40***
Farm labour force ns ns ns ns ns
Involvement in off-farm activities 0.14* ns ns ns ns
Values indicate standardized Beta coefficients
ns not significant
*, **, *** Significant (F test) at P \ 0.05, \0.01, \0.001, respectively

Density of trees Diversity of trees in farm units

The number of trees per hectare ranged from 86 to
1,082 with an average of 475. The high density of trees
in some farms is attributed to trees (mainly eucalypts)
planted in very high densities on boundaries and in
block arrangements. In some farms, dense live fences
of E. candelabrum Trem. & Kotschy. and Cupressus
lusitanica Mill. are also found. Eucalypts are har-
vested in short rotations for their long and slender
posts and, thus, they are grown in high densities. On
the other hand, trees dispersed in the farms generally
have low densities in order to reduce light competition
with agricultural crops. The high density of trees in
Sidama homegardens is in agreement with earlier
reports from other regions. In the Kandyan gardens of
Sri Lanka, Perera and Rajapaske (1991) have reported
that the number of trees with a diameter of 5 cm and
above was 92–3,736 trees ha-1 with 70 % of the
gardens containing 500–1,500 trees ha-1. Jacob and
Alles (1987) have also reported density of
65–1,700 trees ha-1. In the highlands of Rwanda,
where ecological and demographic factors are more or
less similar to the study area, an average density of
731–1,689 trees ha-1 was reported (Biggelaar and
Gold 1996).
The dominant coffee and enset units together with
home and grazing lands cover about 80 % of the farm
areas. The remaining 20 % of the farm is composed of
other patches of crops of which maize takes a share of
about 10 %. Nevertheless, due to the huge dominance
of the integrated coffee-enset units in relation to small
size of the farms, the multistorey structure is more
evident and, hence, the term homegarden. In addition
to their coverage of large areas, which could increase
the chance of accommodating more species, deliberate
planting and management of trees occurs in the coffee
and enset units because the trees are used as shade, and
for soil fertility maintenance. On the other hand, units
of maize, khat and pineapple have few associated tree
species, since shade trees are deliberately reduced to
allow more light for the crops.
Composition of tree species among sites
There are basic sets of tree species that are common
in the homegardens of Sidama, but differences
occur in composition of the species. The variation
is mainly attributed to differences in altitude, size
and distance to road of farms. The altitudinal

123
 Author's personal copy
Agroforest Syst (2013) 87:1283–1293
difference (1,520–2,040 m.a.s.l.) is large enough to
show variation in the composition of natural
vegetation. For instance, in the low lying and
relatively warmer and drier sites, dryland species
such as Faurea rochetiana (A. Rich.) Chinov. ex
Pichi-serm., and Combretum molle R. Br. Ex G.
Don are common but these species are absent in the
higher altitudes. On the other hand, tree species
such as Ocotea kenyenis Kosterm., Olea capensis
ssp. hochstetteri (Bak.) Fris & P.S. Green, Polys-
cias fulva (Hiern.) Harms. and Erythrina spp. that
are common in the higher altitudes are very rare or
absent in the lower altitudes. These variations
reflect ecological adaptations of the species to
particular sites. Likewise, the composition of some
native timber species such as Ekbergia capensis
Sparrman, Syzigium guineense (Willd.) DC., and
Olea capensis near highways is very low, due to
intensive exploitation.
Comparison of the diversity indices among the
different types of the enset-coffee homegarden
systems indicated a clear difference in tree species
richness, evenness and density of trees between the
widely occurring enset-coffee-maize systems and
the other derivatives. The enset-coffee-maize system
is original type and the other types are derived from
it, mainly due to new marketing opportunities for
cash crops and shrinking farm size. The diversity
values were generally lower in the systems where
the composition of the basic crops, enset and coffee,
decreased due to an increased share of new cash
and food crops. The change in composition is
largely the effect of market access opened through
development of road infrastructure. Several reports,
such as Marten and Abdoellah (1988), Jensen
(1993), Peyre et al. (2006), Wiersum (1982, 2006)
have shown that increased market opportunities lead
to changes in composition of species in homegar-
dens. Proximity of the market in itself did not affect
diversity and density of trees and this is because the
local markets are not important market outlets for
wood and wood products. The composition of tree
species also showed a distinct relationship among
the different types of the homegardens. Due to a
combination of factors such as altitude, geograph-
ical distance and road access, the sites have evolved
into distinct land use types which are similar not
only in the composition of the major crops but also
in the composition of tree species.
1291
What factors influence the diversity and density
of trees in farms?
Diversity and density of trees varied among sites and
households mainly due to local socioeconomic and
physical conditions. The most important of these
factors were farm size, area of woodlots, proximity to
major roads (highways) and altitude.
Larger farms had more trees and tree species. This
is because small-holder farmers concentrate on fewer
species of greater utility and allocate more of their
land to food crops, while large holders can afford to
include different types of trees. The pattern of
increasing tree species richness with increasing land
holding was also reported for other homegarden
systems (Kumar et al. 1994; Biggelaar and Gold
1996; Mendez et al. 2001). Farm size did not affect the
density of trees. This contradicts with earlier reports
(Nair and Sreedharan 1986; Kumar et al. 1994;
Biggelaar and Gold 1996) who found that density of
trees declined with increasing farm size. The large
share of densely stocked eucalypts in these farms
could have contributed to high overall density of trees.
With increasing area of woodlots, the density of
trees increased, but evenness decreased significantly.
Area of woodlot is highly correlated (r = 0.41) with
proportion of eucalypts, and eucalypts are often
planted in high densities to be harvested in a short
rotation cycle of 3–8 years. Therefore, densely
stocked eucalypts trees that dominate the woodlots
have contributed to the high density of trees. On the
other hand, the high dominance of eucalypts in the
woodlots resulted in a highly uneven population of
trees leading to a low evenness value. Thus, while tree
species richness is heavily influenced by farm size, the
density of trees is heavily associated with the size of
woodlots.
Distance of a farm to major roads influences tree
species diversity and density. Farms close to main road
have fewer tree species, lower density and less
homogeneous tree population. This is because the
roads have increased market access to the farmers who
sell different wood products on the roadside that are,
then, transported to traders and consumers in big
towns. Thus, proximity to roads has given a better
market access to wood products than the physical
proximity to local markets. Increased marketing of
wood products results in more intensive exploitation
of trees in the farms, and this could lead to reduced
123
 Author's personal copy
1292
diversity and density of trees. Here, tree planting is
often carried out with fast growing and highly
demanded species, mainly eucalypts, whose domi-
nance results in highly uneven population of tree
species. Such effects of road access on the diversity
and homogeneity of trees has also been reported from
elsewhere (Kaya et al. 2002).
In conclusion, woody species diversity of the enset-
coffee agroforestry homegardens of Sidama is gener-
ally high, but the diversity and density of trees varies
within and between sites. Among the farm units,
woodlots and boundary plantings have very high
density sometimes exceeding 10,000 trees per hectare
while trees scattered in crop fields have very low
density. At present, a declining trend in tree species
diversity is being observed particularly on sites that
have access to roads. Improved market access facil-
itated by the roads has lead to the decline in the share
of the basic components, enset, coffee and trees. The
decline in the share of these perennial components and
their replacement particularly with annual crops is
likely to reduce the ecological benefits derived from
these integrated and complex systems. Hence,
attempts to improve these homegardens should not
significantly affect the integrated nature and perennial
components of the systems. Research and develop-
ment efforts should rather aim at incorporating new
crops into the system without affecting its integrity.
Acknowledgments We express our gratitude to the World
Agroforestry Centre (formerly, the International Centre for
Research in Agroforestry (ICRAF), for financing the field
research, and Wageningen University for a PhD sandwich
scholarship to TA. Also we thank the farmers and agricultural
offices of the four districts in Sidama Zone (Aleta Wondo, Dale,
Dara and Hawassa Zurya) for all their support and unreserved
cooperation during data collection.
References
Abdoellah OS, Hadikusumah HY, Takeuchi K, Okubo S, Pari-
kesit (2006) Commercialization of homegardens in an
Indonesian village: vegetation composition and functional
changes. In: Kumar BM, Nair PKR (eds) Tropical home-
gardens: a time-tested example of sustainable agroforestry.
Springer, Dordrecht, pp 233–250
Abebe T (2005) Diversity in homegarden agroforestry systems
of Southern Ethiopia. Ph.D. Dissertation, tropical resource
management papers no. 59, Wageningen University, the
Netherlands
123
Agroforest Syst (2013) 87:1283–1293
Abebe T, Bongers F (2012) Land-use dynamics in enset-based
agroforestry homegardens in Ethiopia. In: Arts B, van
Bommel S, Ros-Tonen M, Verschoor G (eds) Forest people
interfaces: understanding community forestry and biocul-
tural diversity. Wageningen Academic Publishers, Wa-
geningen, pp 69–85
Abebe T, Wiersum KF, Bongers F, Sterck F (2006) Diversity
and dynamics in homegardens of Southern Ethiopia. In:
Kumar BM, Nair PKR (eds) Tropical homegardens: a time
tested example of sustainable agroforestry. Springer Sci-
ence, Dordrecht, pp 87–103
Abebe T, Wiersum KF, Bongers F (2010) Spatial and temporal
variation in crop diversity in agroforestry homegardens of
Southern Ethiopia. Agrofor Syst 78:309–322
Alvarez-Buylla Roces ME, Lazos Chavero E, Garcia-Barrios JR
(1989) Homegardens of humid tropical region in Southeast
Mexico: an example of an agroforestry cropping system in
a recently established community. Agrofor Syst 8:133–156
Bardhan S, Jose S, Biswas S, Kabir K, Rogers W (2012)
Homegarden agroforestry systems: an intermediary for
biodiversity conservation in Bangladesh. Agrofor Syst
85(1):29–34
Biggelaar C, Gold MA (1996) Development of utility and
location indices for classifying agroforestry species: the
case of Rwanda. Agrofor Syst 34:229–246
Fernandes ECM, Oktingati A, Maghembe J (1984) The Chagga
homegardens: a multistoreyed agroforestry cropping sys-
tem on Mt. Kilimanjaro, Northern Tanzania. Agrofor Syst
2:73–86
Huston MA (1995) Biological diversity: the coexistence of
species on changing landscapes. Cambridge University
press, New York
Jacob VJ, Alles WS (1987) Kandyan gardens of Srilanka. Ag-
rofor Syst 5:123–137
Jensen M (1993) Productivity and nutrient cycling of Javanese
homegarden. Agrofor Syst 24:187–201
Jose S (2012) Agroforestry for conserving and enhancing bio-
diversity. Agrofor Syst 85(1):1–8
Kaya M, Kammesheidt L, Weidelt HJ (2002) The forest garden
system of Saparua island, central Maluku, Indonesia, and
its role in maintaining tree species diversity. Agrofor Syst
54:225–234
Kessy JF (1998) Conservation and utilization of natural
resources in the East Usambara forest reserves: conven-
tional views and local perspectives. Ph.D. thesis, Wagen-
ingen University
Kumar BM, Nair PKR (eds) (2006) Tropical homegardens: a
time-tested example of sustainable agroforestry. Springer,
Dordrecht
Kumar MB, George SJ, Chinnamani S (1994) Diversity, struc-
ture and standing stock of wood in the homegardens of
Kerala in Peninsular India. Agrofor Syst 25:243–262
Magurran AE (1988) Ecological diversity and its measurement.
Croom Helm, London
Marten GD, Abdoellah OS (1988) Crop diversity and nutrition
in West Java. Ecol Food Nutr 21:17–43
Mendez VE, Kok L, Somarriba E (2001) Interdisciplinary
analysis of homegardens in Nicaragua: micro-zonation,
plant use and socioeconomic importance. Agrofor Syst
51:85–96
 Author's personal copy
Agroforest Syst (2013) 87:1283–1293
Michon G, Bompard J, Hecketsweiler P, Ducatillon C (1983)
Tropical forest architectural analysis to agroforests in the
humid tropics: the examples of traditional village-agro-
forests in west Java. Agrofor Syst 1:117–129
Michon G, De Foresta H, Levang P, Verdeaux F (2007)
Domestic forests: a new paradigm for integrating local
communities’ forestry into tropical forest science. Ecol
Soc 12(2):1. http://www.ecologyandsociety.org/vol12/iss2/
art1/
Nair MA, Sreedharan C (1986) Agroforestry farming systems in
the homesteads of Kerala, Southern India. Agrofor Syst
4:339–363
Negash M, Yirdaw E, Luukkanen O (2012) Potential of indig-
enous multistrata agroforests for maintaining native flo-
ristic diversity in the south-eastern Rift Valley escarpment,
Ethiopia. Agrofor Syst 85(1):9–28
Oduol PA, Aluma JRW (1990) The banana (Musa spp.)-Coffee
robusta: traditional agroforestry system of Uganda. Agro-
for Syst 11:213–226
Okafor JC, Fernandes ECM (1987) Compound farms of South
Eastern Nigeria: a predominant agroforestry homegarden
system with crops and small livestock. Agrofor Syst
5:153–168
Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR,
O’Hara RB, Simpson GL, Solymos P, Stevens MHH,
Wagner H (2012) Vegan: Community Ecology Package. R
package version 2.1-21. http://r-forge.r-project.org/
projects/vegan/. Accessed 21 March 2013
Padoch C, Jong W (1991) The house gardens of Santa Rosa:
diversity and variability in an Amazonian agricultural
system. Econ Bot 45:166–175
Peet RK (1974) The measurement of species diversity. Annu
Rev Ecol Syst 5:285–307
Perera AH, Rajapakse NRM (1991) A baseline study of
Kandayan gorest gardens of Srilanka: structure, composi-
tion and utilization. For Ecol Manag 45:269–280
Peyre A, Guidal A, Wiersum KF, Bongers F (2006) Homegar-
den dynamics in Kerala, India. In: Kumar BM, Nair PKR
(eds) Tropical homegardens: a time-tested example of
sustainable agroforestry. Springer, Dordrecht, pp 87–103
View publication stats
1293
Pielou EC (1969) An introduction to mathematical ecology.
Wiley, New York
R Development Core Team (2012) R: a language and environ-
ment for statistical computing. R Foundation for Statistical
Computing, Vienna
Rugalema GH, Okting’ati A, Johnson FH (1994) The home-
garden agroforestry systems of Bukoba district, North-
Western Tanzania. 1. Farming systems analysis. Agrofor
Syst 26:53–64
Shannon CE, Wiener W (1949) The mathematical theory of
communication. The University of Illinois Press, USA
Soemarwoto O (1987) Homegadens: a traditional agrforestry
system with a promising future. In: Steppler H, Nair PKR
(eds) Agroforestry: a decade of development. International
Council for Research in Agroforestry (ICRAF), Nairobi,
pp 157–170
Soemarwoto O, Conway GR (1991) The Javenese homegarden.
J Farming Syst Res Extension 2:95–117
SPSS (Statistical Package for Social Sciences) (2008) SPSS
statistics 17.0. SPSS Inc., Chicago
SZPEDD (Sidama Zone Planning and Economic Development
Department) (1997) Sidama administrative zone: a socio-
economic profile. Awassa, Ethiopia
Ter Braak CJF, Smilauer P (2002) CANOCO reference manual
and CanoDraw for Windows user’s guide, software for
canonical community ordination (version 4.5). Micro-
computer Power, Ithaca
Torquebiau E (1992) Are tropical agroforestry homegardens
sustainable? Agric Ecosyst Environ 41:189–207
van der Wal H, Bongers F (2013) Bio-social and bio-numerical
diversity and home garden size in Tabasco, Mexico. Ag-
rofor Syst 87:93–107
Wiersum KF (1982) Tree gardening and Taungya on Java:
examples of agroforestry techniques in the humid tropics.
Agrofor Syst 1:53–70
Wiersum KF (2006) Diversity and change in homegarden cul-
tivation in Indonesia. In: Kumar BM, Nair PKR (eds)
Tropical homegardens: a time-tested example of sustain-
able agroforestry. Springer, Dordrecht, pp 13–24
123