Geographical distribution, tree density and fruit production of Tamarindus indica L. (Fabaceae) across three ecological regions in Benin

Abstract

Introduction. There has been increasing interest in the domestication potential of indigenous fruit trees. Nevertheless, our understanding of how these species' abundance and yield of fruit is altered by ecological conditions, which is critical to foresee realistic sustainable management plans, is limited. Material and methods. We used local ecological knowledge, presence/absence data and quantitative methods to examine the effect of ecological conditions on the distribution, abundance and yields of tamarind trees (T. indica) across three ecological regions in Benin, West Africa. Results and discussion. Rural communities’ knowledge on the species’ ecological range was congruent with scientific findings. The natural distribution of tamarind individuals was limited to the Sudanian and the Sudano-Guinean regions and their density declined with increasing moisture, being highest (2 trees·km–2) in the Sudanian region and lowest in the Guineo-Congolian region (scarce). On the other hand, fruit and pulp mass and number of seeds per fruit varied significantly, being higher in the Guineo-Congolian wetter region. However, no significant variation occurred among ecological regions for estimated overall fruit yields per tree. This might indicate that tamarind trees tend to invest in a small number of very large fruits under wetter conditions and a very large number of small fruits under dryer conditions. Conclusion. The results showed that semi-arid lands would best suit T. indica domestication. Nevertheless, its productivity could be higher under wetter conditions. Because of its affinity for gallery forests, we recommend thorough studies on its capacity to survive the
increasing drought in its current ecological range.

Full text

Original article
Fruits, vol. 66 (2) 53
Geographical distribution, tree density and fruit production of Tamarindus
indica L. (Fabaceae) across three ecological regions in Benin.
Abstract –– Introduction. There has been increasing interest in the domestication potential
of indigenous fruit trees. Nevertheless, our understanding of how these species' abundance and
yield of fruit is altered by ecological conditions, which is critical to foresee realistic sustainable
management plans, is limited. Material and methods. We used local ecological knowledge,
presence/absence data and quantitative methods to examine the effect of ecological conditions
on the distribution, abundance and yields of tamarind trees (T. indica) across three ecological
regions in Benin, West Africa. Results and discussion. Rural communities’ knowledge on the
species’ ecological range was congruent with scientific findings. The natural distribution of
tamarind individuals was limited to the Sudanian and the Sudano-Guinean regions and their
density declined with increasing moisture, being highest (2 trees·km–2) in the Sudanian region
and lowest in the Guineo-Congolian region (scarce). On the other hand, fruit and pulp mass
and number of seeds per fruit varied significantly, being higher in the Guineo-Congolian wetter
region. However, no significant variation occurred among ecological regions for estimated ove-
rall fruit yields per tree. This might indicate that tamarind trees tend to invest in a small number
of very large fruits under wetter conditions and a very large number of small fruits under dryer
conditions. Conclusion. The results showed that semi-arid lands would best suit T. indica
domestication. Nevertheless, its productivity could be higher under wetter conditions. Because
of its affinity for gallery forests, we recommend thorough studies on its capacity to survive the
increasing drought in its current ecological range.
Benin / Tamarindus indica / population distribution / population density /
fruiting / indigenous knowledge
Répartition géographique, densité des arbres et production de fruits de
Tamarindus indica L. (Fabaceae) dans trois régions écologiques au Bénin.
Résumé –– Introduction. Le potentiel de domestication des arbres fruitiers indigènes a suscité
un intérêt croissant. Néanmoins, la façon dont l’abondance de ces espèces et leur rendement
en fruits sont affectés par les conditions écologiques, élément essentiel pour prévoir des plans
de gestion durable réalistes, est peu documentée. Matériel et méthodes. Nous avons utilisé
des connaissances écologiques locales, des données de présence / absence et des méthodes
quantitatives pour étudier l'effet des conditions écologiques sur la répartition, l'abondance et
les rendements des tamariniers (T. indica) dans trois régions écologiques du Bénin, en Afrique
de l'Ouest. Résultats et discussion. La connaissance de l'aire de répartition écologique par les
communautés rurales a été en accord avec les résultats scientifiques. La répartition naturelle des
tamariniers a été limitée aux régions soudanaise et soudano-guinéenne et leur densité a diminué
avec l’accroissement du taux d'humidité; la densité a été la plus élevée (2 arbres·km–2) dans la
région soudanaise et la plus basse dans la région guinéo-congolaise (présence rare). D’autre
part, la masse de pulpe et de fruits ainsi que le nombre de graines par fruit ont sensiblement
varié; ces caractères ont été les plus élevés dans la région humide guinéo-congolaise. Cepen-
dant, aucune variation significative n’est apparue d’une région écologique à l’autre en ce qui
concerne les rendements globaux de fruits par arbre. Cela pourrait indiquer que les tamariniers
tendent à produire un petit nombre de gros fruits en milieu plus humides et un grand nombre
de petits fruits en milieu plus sec. Conclusion. Nos résultats pourraient suggérer que les zones
semi-arides pourraient mieux convenir à la domestication de T. indica. Néanmoins, la produc-
tivité de l’espèce pourrait être plus élevée en conditions plus humides. Sur la base de son affinité
pour les forêts-galeries, nous recommandons des études approfondies sur sa capacité à survivre
à des stress hydriques croissants dans son environnement écologique actuel.
Bénin / Tamarindus indica / distribution des populations / densité de
population / fructification / connaissance indigène
Lab. Appl. Ecol., Fac. Agron.
Sci., Univ. Abomey-Calavi, 01
BP 526, Cotonou, Benin,
bfandohan@gmail.com
Geographical distribution, tree density and fruit production
of Tamarindus indica L. (Fabaceae) across three ecological
regions in Benin
Belarmain FANDOHAN*, Achille Ephrem ASSOGBADJO, Romain Glèlè KAKAÏ, Brice SINSIN
* Correspondence and reprints
Received 19 February 2009
Accepted 3 June 2009
Fruits, 2011, vol. 66, p. 53–62
© 2011 Cirad/EDP Sciences
All rights reserved
DOI: 10.1051/fruits/2010043
www.fruits-journal.org
RESUMEN ESPAÑOL,p.62

54 Fruits, vol. 66 (2)
B. Fandohan et al.
1. Introduction
Until recently, little interest has been
shown in indigenous multipurpose tree spe-
cies compared with their industrial timber
counterparts. Available information gener-
ally lacks adequate quantitative analysis for
the development of economic opportunities
based on local resources as an alternative to
excessive import of exotic products. Thus,
initiatives in agroforestry seek to integrate
indigenous trees whose products have tra-
ditionally been gathered from natural forests
into tropical farming systems [1]. This is also
to provide marketable products from farms
that will generate income for resource-poor
rural households. As such, it has led to an
increasing interest in the domestication
potential of some traditionally valued agro-
forestry trees in Africa, i.e.,Detarium micro-
carpum Harms [2], Irvingia gabonensis
(Aubry Lecomte ex O'Rorke) Baill. [3] and
Sclerocarya birrea (A.Rich.) Hochst.) subsp.
caffra (Sond.) [4], to cite just a few.
The growing number of studies on the
domestication potential of indigenous trees
have provided information about morpho-
logical and genetic diversity [5, 6], potential
productivity [7] and nutritional and medici-
nal properties [8]. However, as local envi-
ronmental conditions influence variation in
plants [9], domestication programs should
also consider the impact of ecological fac-
tors on distribution, abundance and fruiting
potential of the targeted species. Such
aspects are poorly explored but critical for
identification of priority sites for planting
and genetic improvement. In addition, such
information may highlight the extent to
which natural populations could supply
market demand with products (i.e. fresh
fruit and pulp) depending on ecological
conditions. Although a number of studies
have demonstrated that the distribution,
abundance and fruiting potential of indige-
nous multipurpose trees depend on envi-
ronmental conditions [7, 10], the effect of
environmental conditions may vary with
plant species. In addition, the accuracy of
abundance estimation may be strongly
dependent on the method used and robust
statistical methods should be used when-
ever possible.
T-Square sampling [11] is a distance-
based sampling method which has been
used in ecology to estimate sizes and den-
sities of many plant populations [12]. It is a
robust method that can provide informative
results even if no prior information exists
about the randomness or the uniformity of
the spatial pattern of the species studied,
and is especially suited to large-scale pop-
ulation studies [13]. T-square sampling can
therefore be used to evaluate the abundance
of indigenous fruit trees or multipurpose
trees nationwide or regionwide. In addition,
recent ethnoecological studies advocate an
integration of local ecological knowledge
and ecological studies to advance our
understanding of ecological processes and
develop better plans for sustainable
resource use [14]. To enable such integra-
tion, it is important to evaluate the congru-
ence of local ecological knowledge with
available scientific findings [14].
Tamarindus indica L. is a semi-ever-
green tree widespread in the tropics (Africa,
Asia, Madagascar, South America), featuring
prominently in riparian habitats [15]. In
efforts to enhance the species’ genetic
resource conservation and utilization, it has
recently been identified as one of the top
ten agroforestry tree species to be priori-
tized for future crop diversification pro-
grams and development in sub-Saharan
Africa [16]. There has been extensive inter-
est in tamarind's biochemical, medicinal
and nutritional properties, reproductive
biology, morphology, cultivation and genet-
ics [17]. Studies on the species in Asia, Africa
and Latin America have documented its
domestication potential in terms of socio-
cultural and nutritional values, and aptitude
for seed and vegetative propagation [17].
Tamarind’s global distribution map has
given strong evidence of its plasticity in the
tropics [18]. Studies on the species’ native
populations in Africa have provided data on
biochemical analysis of its fruits [19], genetic
diversity, phenology, reproductive biology
including pollinators, biometrical characters
of seeds and seedlings [20–23], and impact
of habitat type on its conservation status
[24]. However, little scientific information is
available on how ecological conditions
locally influence its distribution, abundance

T. indica in three ecological regions of Benin
Fruits, vol. 66 (2) 55
and productivity in Africa. In this study, we
used LEK and ecological studies to examine
the distribution and productivity of tama-
rind trees in Benin as a case study. The fol-
lowing questions were addressed: (1) What
is the distribution range and abundance of
T. indica in relation to the ecological con-
ditions it lives in? (2) How do ecological and
tree-level factors alter its productivity of
fresh fruits, pulp and seeds?
2. Materials and methods
2.1. Study area
Our study was conducted in Benin (West
Africa). Ongoing work on the ecological
niche of T. indica in Benin has confirmed
the presence of tamarind individuals in the
three ecological regions of the country
(the Sudanian, Sudano-Guinean and sub-
humid Guineo-Congolian regions [25]).
Thus, this study was conducted nationwide
(114 673 km²) [26].
In the Sudanian region (9°45’–12°25’ N),
rainfall is unimodal. Mean annual rainfall is
often less than 1000 mm, the relative humid-
ity varies between 18% and 99% (highest in
August) and temperature varies between
24 °C and 31 °C. The Sudanian region has
hydromorphic soils, well-drained soils and
lithosols.
In the Sudano-Guinean region (7°30’–
9°45’ N), rainfall is unimodal, from May to
October, and lasts for about 113 days with
an annual total rainfall varying between
900 mm and 1110 mm. Mean annual tem-
perature ranges from 25 °C to 29 °C, and the
relative humidity from 31% to 98%. The soils
in this zone are ferruginous.
In the Guineo-Congolian region (6°25’–
7°30’ N), rainfall is bimodal with a mean
annual rainfall of 1200 mm. Mean annual
temperature varies between 25 °C and 29 °C
and the relative humidity varies between
69% and 97%. The soils are either deep fer-
rallitic or rich in clay.
The principal ethnic groups are the
Dendi, Bariba, Fulani, Waama and Gour-
mantché in the Sudanian region; the Fon,
Tchabè, Nago, Agnin and Idatcha in the
Sudano-Guinean region; and the Yoruba,
Adja, Tori, Holi, Péda and Xwla in the
Guineo-Congolian region. We focused on
these 16 aforementioned ethnic groups dur-
ing this study because they were the most
ancient in their locations throughout the
study area.
2.2. Ethnoecological survey
Our survey aimed at documenting local
perceptions about the distribution of tama-
rind trees in Benin. We applied semi-struc-
tured individual interviews to five traditional
hunters / healers and five Fulani herders
randomly sampled within each of the
77 state districts of Benin (770 respond-
ents). The sampling was limited to persons
of at least fifty, so that only experienced
informants were questioned. We targeted
Fulani herders because they are culturally
associated with the species and usually keep
the fruits during migrations (personal field
observation). The healers and hunters were
chosen because of their local reputations in
plant knowledge. These socio-professional
groups are more likely to hold very inform-
ative knowledge about the past and present
distribution of the species. The following
questions were asked: (i) As an indicator of
its presence in earlier times, we asked
whether tamarind had a local name in the
language of the most ancient indigenous
ethnic groups of each district; (ii) Is the spe-
cies currently found in the natural vegeta-
tion? If yes, in which habitats? (iii) Was the
species present in the native vegetation in
earlier times? If it was present in earlier times
but is not currently present, then why has
it disappeared?
2.3. Characterization
of the distribution patterns
and density of tamarind
In each of the 77 state districts, we made
a visual confirmation of the species’ pres-
ence/absence within protected areas, for-
ests and farmlands, following unfixed
transect lines across natural vegetation, with
the help of transhumants or hunters recom-
mended by local leaders.

56 Fruits, vol. 66 (2)
B. Fandohan et al.
Then, we performed a sampling plan for
tamarind density estimation based on the
ten phytogeographical subdivisions of the
three ecological regions [27]. We randomly
selected one to two sites (protected areas
and surrounding farmlands) within each
phytogeographical district where tamarind
natural populations were observed
(table I). The density of the species was esti-
mated using the T-Square sampling
method. Each phytogeographical district
was represented by 45–90 sampled points
for density estimation (overall 630 points
for 112622 km²), following a protocol
already published [28]. To address any
underlying heterogeneity in tamarind trees’
pattern, a stratification of the vegetation and
a systematic sampling approach were per-
formed. Thus, each selected site was first
stratified into three habitats: forest, savan-
nah and farmland. We established 15 points
in each habitat. The 15 points were distrib-
uted on five azimuth transects of 15 km, so
that each transect was divided into
3 equally spaced marks (5 km) (Si); then,
from each marked point, the distance (xi)to
the nearest tamarind individual (T1) was
measured using a global positioning system
(GPS Garmin 76). Next, the distance (yi)
between T1and its nearest neighbor (T2)
situated in the ‘half-plane’ excluding Siwas
measured [29] (figure 1). We used a GPS
instead of a tape measure or a decameter
since “x” and “y” distances were often over
500 m, except in gallery forests.
2.4. Estimating average yield
of fresh fruit, pulp and seeds
Data were collected on a total of 65 trees
sampled among the trees that we located
when estimating tree density. Thirty trees
were sampled in both the Sudanian and
Sudano-Guinean regions while only five
trees were selected in the Guineo-Congolian
Table I.
Sampling plan for assessing geographical distribution, tree density and fruit production of Tamarindus indica L.
in three ecological regions of Benin.
N° Ecological regions Phytogeographical district State district Study site Coordinates
Lat. E Long. N
1 Sudanian Atacora Chain Tanguiéta Pendjari National Park 1° 44’ 11° 07’
2 Mekrou-Pendjari Karimama W National Park 2° 54’ 12° 06’
3 Mekrou-Pendjari Tanguiéta Pendjari National Park 1° 30’ 11° 25’
4 Borgou North Banikoara W National Park 2° 40’ 11° 55’
5 Borgou North Pehunco Upper Alibori Forest 2° 17’ 10° 59’
6 Sudano-Guinean Borgou South Tchaourou Monts Kouffé Forest 2° 08’ 9° 00’
7Borgou South Kalalé Three Rivers Forest 3° 19’ 10° 50’
8 Bassila Bassila Penessoulou Forest 1° 64’ 9° 21’
9 Zou Djidja Gbadagba Forest 2° 01’ 7° 57’
10 Guineo-Congolian Oueme Valley Zogbodomey Locoly Swamp Forest 2° 36’ 7° 10’
11 Plain Kétou Kétou-Dogo Forest 2° 60’ 7° 73’
12 Plain Houéyogbé Houéyogbé Forest 1° 85’ 6° 50’
13 Pobè Pobè Pobè Forest 2° 68’ 7° 07’
14 Coast Ouidah Ahozon Forest 2° 16’ 6° 37’
Figure 1.
T-Square sampling method
[tamarind trees (T); sampling
point (mi); distances labeled x
and y; planes labeled L and R;
region of interest (Ω,)] (adapted
from Diggle [29]).

T. indica in three ecological regions of Benin
Fruits, vol. 66 (2) 57
region. The lower number in the latter
region was due to the scarcity of tamarind
trees in that region. For consistency and to
make realistic comparisons, sampling was
limited to trees with a diameter at 1.3 m
above ground (D130) of at least 20 cm. Sam-
pled trees were measured for D130, total
height and crown diameter. Crown areas
were computed according to a method pub-
lished elsewhere [30]. For each fruit-bearing
tree, fruits were harvested and weighed
together.
To estimate fruit and pulp mass and
number of seeds per fruit, 30 fruits were
sampled per fruit-bearing tree as described
elsewhere [31]. After weighing, fruits were
oven-dried at 65 °C for 48 h to obtain dry
mass. Dried fruits were broken and the con-
tent extracted (pulp + seeds + fibers). Pulp
was removed by soaking in water. The
remainder (seeds and fibers) was oven-
dried at 65 °C for 48 h and weighed. Pulp
mass was computed as: dry mass of the fruit
minus dry mass of the remainder. Finally,
the number of seeds per fruit was also
counted.
Climatic data for over 30 years (1978–
2008) were obtained for each sampling site
within ecological regions from the World-
clim data [32].
2.5. Data analysis
For ethnoecological data, we computed
a response rate per question using the fol-
lowing formula [33]: F=[100×(S/N)],
where Sis the number of responses to a par-
ticular question and Nis the total number
of persons interviewed.
The density of T.indica in each phyto-
geographical district was calculated accord-
ing to the following formula [29]:
, where: λis
the estimate of the number of adult trees
(D130 ≥10 cm) per km², mis the number of
sampling points, xiis the distance from the
ith point to the nearest tamarind (T1) tree,
yiis the distance between T1and its closest
neighbor (T2) in the opposite plane which
does not contain the ith sampling point, and
πis approximately equal to 3.14.
The distribution and abundance of
T. indica were then mapped using Arcview
software.
The mass of the pulp content (wp) in each
fruit was computed using the following for-
mula: wp =wPi–wRi, where wp is the pulp
content of a fruit for a given tree; wPiis the
total dry mass of the fruit (i); and wRiis the
total dry mass of seeds, fibers and husk of
the fruit (i). Overall yield of pulp and
number of seeds of the sampled trees were
estimated based on total yield of fresh fruits
and mean values obtained for fresh mass
and dry mass of fruit.
The climatic index of Mangenot (IM),
which is a measure of water availability [27],
was computed for each sampled site as fol-
lows:
, where Pis the mean
annual rainfall (mm); MSis the mean rainfall
of dry months (rainfall less than 50 mm); nS
is the number of dry months; is the max-
imum of annual relative humidity (%) ; and
is the minimum of annual relative
humidity (%). This index provides a better
quantitative assessment of climatic condi-
tions than any single climatic variable and
was used here as the indicator of the envi-
ronmental conditions in each ecological
region [27].
Because productivity of fresh fruits and,
pulp and seed values were not normally dis-
tributed (Ryan-Joiner test of normality) [34],
the non-parametric test of Kruskal-Wallis
was applied to describe and compare the
three ecological regions. To investigate the
correlation between ecological conditions,
tree characteristics (D130, height, crown
area) and differences in tamarind produc-
tivity, a Standardized Principal Component
Analysis (PCA) was first performed on the
productivity traits of the species (pulp mass,
number of seeds per fruit, overall yield of
fruits and pulp, and number of seeds per
λm
π
-----1
x1
2
i1=
m
∑1
2
---y1
2
i1=
m
∑
×
-----------------------------------------------
×=
I
M
P
100
---------MSU
X
++
nS 500
Un
---------+
--------------------------------------
=
Ux
Un

58 Fruits, vol. 66 (2)
B. Fandohan et al.
tree). The principal components obtained
from the PCA were then correlated with
D130, height, crown area of tamarind trees
and the climatic index of Mangenot using
Pearson’s correlation coefficient. This
method was used to avoid evident correla-
tion between the productivity traits of the
species that could bias the estimation of the
degree of link they have with the dendro-
metric and ecological parameters.
3. Results
3.1. Local perception on past
and present distribution of tamarind
All the respondents knew the species
because of their profession. From this sur-
vey, it appeared that except for the Yoruba
ethnic group, tamarind had no local name
within the Guineo-Congolian region
(table II). According to the respondents
from the Sudano-Guinean and Guineo-Con-
golian regions and all questioned Fulani
herders (78% of respondents), the current
distribution of the species is limited in the
south by the Sudano-Guinean region. How-
ever, the species was said to have totally dis-
appeared or to be fading out of some
locations where it used to be abundant due
to tree felling for urban construction and
agricultural purposes in that region (near its
borders with the Guineo-Congolian region).
Other human practices (debarking for med-
icine, pruning for fodder and fruit harvest)
were also mentioned as threats to the spe-
cies. In state districts within the Sudano-Gui-
nean region, local ethnic groups generally
associated tamarind trees with Fulani peo-
ple. Furthermore, in this region, Fulani peo-
ple still have the tendency to appropriate the
harvest from any tamarind, even if trees are
on farmers’ land. Tamarind fruits are impor-
tant for the Fulani as a laxative during tran-
shumance trips.
3.2. Characterization
of the distribution patterns
and abundance of tamarind
Tamarind is widespread in Benin (fig-
ure 2). It was identified in all state districts
except those in the phytogeographical dis-
trict of Oueme-valley. However, as sug-
gested by the survey on local ecological
knowledge, tamarind trees are extremely
rare in the native vegetation within the
Guineo-Congolian region. Observed trees
were reported to have been planted 20-
60 years ago. On the other hand, in the
Sudanian and Sudano-Guinean regions,
tamarind stands (with gregarious patterns of
trees) were only found along water courses
in gallery forests. Beyond gallery forests, the
species was observed as isolated individuals
and was sometimes shrub-shaped in open
savannah ecosystems.
Density of tamarind adult trees declined
from the Sudanian region (approximately
2 trees per km²) to the Guineo-Congolian
region (totally absent) (figure 2). Based on
the densities obtained, estimation of tama-
rind adult tree abundance per phytogeo-
graphical district was estimated (table III).
The overall number of tamarind adult trees
in Benin was estimated at 68,114 trees,
meaning approximately an average of
0.6 trees per km² for the whole country.
Table II.
Tamarind local names according to ethnic groups in Benin.
Ecological region Ethnic group Names
Sudanian Bariba Môkôssô - Môhôhô
Dendi Bobosséi - Bosséi
Fulani Djêtami
Gourmantché Bu pugubu
Waama Pusika
Sudanian-Guinean Agnin Gougnémou
Fon Djêvivi
Idatcha Arinran
Nago Kaïma
Tchabè Adjagbon
Guineo-Congolian Adja –
Holi –
Ouémè –
Péda –
Xwla –
Yoruba Adjagbon

T. indica in three ecological regions of Benin
Fruits, vol. 66 (2) 59
Figure 2.
Distribution patterns and
density of tamarind according
to the phytogeographical
subdivision of the three
ecological regions of Benin.

60 Fruits, vol. 66 (2)
B. Fandohan et al.
3.3. Estimating tamarind productivity
of fresh fruit, pulp and seeds
according to ecological regions
Significant differences were observed
among ecological regions for individual fruit
and pulp mass, and number of seeds per
fruit (p< 0.001; H = 314.96; H= 216.99 and
H = 35.57, respectively; table IV). Fruits
from the Guineo-Congolian region had the
highest values for fruit and pulp mass and
number of seeds per fruit. In contrast, no sig-
nificant differences were found between
ecological regions regarding overall yields
in fruits, pulp and number of seeds per tree
(p= 0.05; H = 1.79; H = 3.55; H = 2.57,
respectively). However, the coefficients of
variation (cv) suggested considerable tree-
to-tree variation for the majority of the inves-
tigated variables (table IV).
Table III.
Abundance of tamarind according to the phytogeographical districts within three ecological regions of Benin.
Ecological region Phytogeographical district Area
(km²)
Density
(trees·km–2)
Abundance1
(trees)
Sudanian Atacora Chain 6,905.46021 0.25 1,726.365
Mekrou-Pendjari East 10,579.0816 2.04 21,581.326
Mekrou-Pendjari West 6,261.25419 2.04 12,772.959
Borgou North 32,676.6419 0.51 16,665.087
Sudano-Guinean Borgou South 9,599.48146 0.15 1,439.922
Bassila 24,726.4271 0.50 12,363.214
Zou 10,431.6667 0.15 1,564.750
Guineo-Congolian Plain-East 2,071.22318 Very rare –
Plain-West 5,098.4805 Very rare –
Coast 686.07915 Extremely rare –
Pobé 1,613.20103 Extremely rare –
Oueme Valley 1,973.0031 Absent –
1Abundance = area × density.
Table IV.
Mean and coefficient of variation of yields of fruits, pulp and seeds of tamarind according to their location in one
of the three ecological regions in Benin.
Ecological
region
Fruit mass Pulp mass per fruit Number of seeds per
fruit
Fruit yield per tree Pulp yield per tree Seed yield per tree
Mean
(g)
Coefficient
of variation
(%)
Mean
(g)
Coefficient
of variation
(%)
Mean Coefficient
of variation
(%)
Mean
(kg)
Coefficient
of variation
(%)
Mean
(kg)
Coefficient
of variation
(%)
Mean Coefficient
of variation
(%)
Sudanian 11.29 40.57 4.1 51.22 8.12 28.20 34.42 46.19 12.39 40.07 24758 45.18
Sudano-
Guinean
17.54 37.69 5.51 45.55 7.68 31.51 39.07 125.75 12.89 105.62 20550 125.76
Guineo-
Congolian
48.85 26.64 14.59 30.23 9.73 20.76 74.73 94.53 22.42 91.16 14889 89.01
Significance1< 0.0001 < 0.0001 < 0.0001 0.41 0.26 0.049
1According to the Kruskal-Wallis non-parametric test.

T. indica in three ecological regions of Benin
Fruits, vol. 66 (2) 61
The Principal Component Analysis (PCA)
performed on productivity characteristics
showed that the first two axes explained
85.83% of the observed variations in fruit
and pulp mass and number of seeds
(table V). The first axis is the productivity
axis; it shows a positive relationship with
and between all the productivity character-
istics. The correlation between this axis and
the dendrometric and ecological parame-
ters shows a relatively significant and posi-
tive correlation between IM(the climatic
index of Mangenot, which is a measure of
water availability) and productivity charac-
teristics. This means that fruit and pulp
mass, number of seeds per fruit, and overall
yield of fruits, pulp and number of seeds
per tree increased with higher IMvalues
(i.e., wetter regions). Correlation between
the first axis component of the PCA and
D130, height and crown area were not sig-
nificant (P> 0.05) and cannot be used here
to explain the observed variations. The sec-
ond axis did not take into account produc-
tivity traits of T. indica species and was not
used to assess and describe the link
between productivity of the species and the
dendrometric and ecological parameters.
4. Discussion
Our findings provide insight into the
effect of ecological conditions on the distri-
bution, abundance and productivity of tam-
arind. They also reflect the effect of human
disturbance on its distribution. The nonex-
istence of local names for the species in
some ethnic groups suggests a very recent
contact with the species and thus its very late
migration to their locations. This matches
the hypothesis which assumes the species
to have a dry ecosystem origin and suggests
its very late introduction into humid regions.
By contrast, the affinity of the species for
water courses (gallery forests) suggests that
habitats with less arid conditions better suit
its establishment and expansion. The pre-
cise origin of T .indica is still under debate.
Remnants of orchards dating back from
400 BC are known from Egypt [35], but Bud-
dhist scriptures refer to it as 650 BC [21]. What
is broadly accepted is that the species have
a tropical African, Madagascarian and Asian
origin [22]. Considering biogeographical
regions in Africa, tamarind is most common
in the Sahelian and the Sudanian ecological
regions [25]. Nonetheless, it is established in
Table V.
Correlation between productivity characteristics, tree-level and ecological parameters of
tamarind trees in Benin, and the axes 1 and 2 of Principal Component Analysis factors
(in brackets is the proportion of variation explained by each axis, expressed in
percentage).
Characteristics Axis 1
(73.93%)
Axis 2
(11.9%)
Fruit yield per tree 0.954 -0.243
Pulp yield per tree 0.937 -0.322
Seed yield per tree 0.686 -0.409
Fruit mass 0.745 0.320
Pulp mass per fruit 0.669 0.23
Number of seeds per fruit 0.544 0.139
Diameter at 1.3 m above ground (D130) 0.023 -0.400*
Height 0.257 -0.307
Crown area 0.318 -0.367
Climatic index of Mangenot (IM) 0.534** 0.531**
** significant at p< 0.01.

62 Fruits, vol. 66 (2)
B. Fandohan et al.
more humid zones and in coastal regions
[36]. Hypotheses have assumed that the cli-
matic variation observed from 20,000 to
10,000 Before Present and from 2,800 to
2,000 Before Present [37], and the subse-
quent replacement of dense forests of Equa-
torial Africa by savannah landscapes might
have allowed the natural establishment of
some savannah tree species (i.e., Adansonia
digita L., Vitellaria paradoxa C.F. Gaertn.,
Borassus aethiopum Mart.) within zones of
higher rainfall [7]. The latter authors have
also suggested that high density of these
species within the highest rainfall zone of
Benin may have to do with the Dahomey
Gap phenomenon which dried the humid
green forest block. But tamarind might have
colonized these regions before those climate
change periods. During the last two dec-
ades, scientific interest has increased in the
Holocene vegetational history of West Africa
[38–40], and the Dahomey gap in Benin
more specifically [41]. However, none of
these studies, including studies in areas
where tamarind is currently found, have
mentioned T. indica, suggesting a very late
migration of the species. This overlaps with
our results on local ecological knowledge.
However, the absence of T. indica from
paleobotanical findings might be linked to
its erratic occurrence at very low frequen-
cies, which might result in a large under-rep-
resentation compared with other taxa. In
addition, considering that tropical trees are
often ecologically highly plastic (at least
within the tropics), their natural distribution
may rather be strongly correlated to their
natural dispersion strategies than local var-
iation in environmental conditions. The cor-
relation between spatial patterns of plant
species and their dispersion strategies has
been highlighted [42]. Since tamarind is a
zoochorous species (monkeys and humans
are its natural dispersal agents), its distribu-
tion out of its native areas might be linked
to monkey and human migrations. It has
been illustrated that gene flow related to
Fulani migrations might explain the low
inter-population genetic diversity noticed
within tamarind populations in West Africa
[21]. The prevalence of some economically
important tree species in the savannahs of
Africa north of the equator, combined with
paleobotanical and historical evidence, are
strong indicators of human involvement in
tree dispersal [43]. However, Fulani herders’
passings (i.e., transhumance) are very recent
in some Sudano-Guinean localities where
natural and very old tamarind trees were
observed, and thus transhumance cannot
explain the presence of these trees (Sinsin,
pers. commun.). The contribution of human
migrations to the distribution of T. indica
may be addressed through the study of the
pattern of lineages among the species’ pop-
ulations across the main migration corridors,
using chloroplastic DNA analyses [21]. With
regard to the observed gregarious pattern of
tamarind trees along water courses (i.e., gal-
lery forests), two questions may need to be
further addressed: which are the factors or
events underlying the affinity of a species
typical of arid lands for wetter habitats (e.g.,
gallery forests)? Was the paleo-climate
which favored the establishment of the spe-
cies wetter than the current climate it is fac-
ing? And, in the case of a positive answer,
how far could the adaptability of the species
allow its persistence in arid ecosystems?
The findings from this study confirmed
the species’ erratic distribution in Benin. The
marked reduction of tamarind tree density
between phytogeographical districts within
the Sudanian and the Sudano-Guinian
regions might be due to tree felling and mor-
tality of natural regeneration caused by field
burning for agricultural purposes. The nat-
ural range of the species (especially in the
phytogeographical districts of Borgou North
and Borgou South, as well as Zou) is char-
acterized by extensive agriculture such as
cotton (Gossypium spp.) production. In
addition, according to our field surveys, in
Borgou South and Zou the species is less
important to the people except for the
Fulani tribes. Thus, tamarind trees were not
spared when land was cleaned for agricul-
tural purposes. Though it is frequently
observed on hillsides, tamarind was found
in low density in landscapes dominated by
hills and mountains such as the phytogeo-
graphical districts of the Atacora chain and
partly that of Zou. This observation contrasts
with findings on baobab trees [7] and may
be linked to the preference of tamarind for
well-drained and deep soils [17]. Three rea-
sons may explain the presence of tamarind
on hillsides: (i) tamarind seeds might have

T. indica in three ecological regions of Benin
Fruits, vol. 66 (2) 63
been disseminated by monkeys; and/or (ii)
when dry periods start, hillsides have plen-
tiful grass plants and are ideal pasture lands
for cattle, so Fulani herders might have dis-
seminated tamarind seeds while roaming
with their cattle; (iii) cattle might also have
contributed to the species' dispersion since
they also feed on its fruits.
Productivity (fruit and pulp mass, and
number of seeds per fruit) of tamarind trees
significantly varied with ecological condi-
tions. This is consistent with other studies
that have shown that environmental varia-
bles can affect fruit size and shape, and ker-
nel mass [7, 38]. However, estimated overall
fruit yields per tree did not significantly dif-
fer across ecological regions, suggesting that
tamarind trees may tend to invest in a small
number of very large fruits under wetter
conditions and a very large number of small
fruits under dryer conditions.
Fresh fruit yield per tamarind tree in this
study was 3 to 6 times lower than the figure
in East Africa [(150 to 300) kg] [36]. Produc-
tivity, including overall fruit yield per tree,
was weakly correlated to tree-level factors
(diameter, height and crown area) and the
results did not show a linear increase in pro-
ductivity with tree size, in contrast with
other studies [44, 45]. However, a positive
correlation was observed with the climatic
index of Mangenot (IM), suggesting produc-
tivity to be relatively higher under more
humid conditions. Successful fruiting in one
year is often followed at the cost of vegeta-
tive growth and some woody plants alter-
nate supra-annual schedules of low and
high production years [46]. This may explain
the high tree-to-tree variation revealed by
the coefficients of variation (cv). It can also
explain why some trees with greater diam-
eter or crown area showed very low fruit
yield. This also provides strong evidence
that ecological conditions which are much
less variable than production itself are not
the only determinants of variation in tama-
rind productivity [47]. For example, disease,
herbivores, adverse weather such as high-
speed winds, particularly during key phe-
nological events such as pollination or fruit
development, seasonal fire [48], and severe
pruning can reduce yield [49]. Variation in
productivity may also reflect genetic differ-
ences [50]. In addition, because tree produc-
tivity may greatly fluctuate with time, a
single census study can hardly provide an
accurate estimation. Besides, the results on
seed production irrespective of ecological
regions indicate that the often observed lack
of regeneration in the species' populations
is not driven by lack of seed production.
Other factors such as insect feeding pressure
on seeds and soil degradation and vegeta-
tion fire could be involved.
5. Conclusion
Our study has highlighted the difficulty in
understanding the current distribution of
tamarind and the need for greater research
on this topic. It has also provided useful pre-
liminary information on the variation in fruit
yield in tamarind. Savannah landscapes
where natural populations of the species
exist (i.e., the phytogeographical districts
within the Sudanian and the Sudano-Guin-
ean regions) are obviously suitable for tam-
arind plantation. However, the affinity of the
species for gallery forests elicits further inter-
est in its capacity to survive the increasing
drought stress in its current ecological range.
Multi-year census studies are needed to con-
sistently model the impact of environmental
characteristics (including soil characteris-
tics), dendrometric characters, genetic vari-
ation and human harvesting pressure
(pruning) on the inter-annual variation in
fruit yield in order to understand the species’
productivity better.
Acknowledgments
This study was supported by Domestica-
tion and Development of Baobab and Tam-
arind (EU-DADOBAT project). We
especially thank Patrick Van Damme for
comments on an earlier version of the man-
uscript. We are very grateful to the Fulani
herders, local hunters and healers for their
help during the field survey and the anon-
ymous reviewers for their useful comments
on improving the manuscript.

64 Fruits, vol. 66 (2)
B. Fandohan et al.
References
[1] Leakey R.R.B., Simons A.J., The domestica-
tion and commercialization of indigenous
trees in agroforestry for the alleviation of
poverty, Agrofor. Syst. 38 (1998) 165–176.
[2] Kouyaté A.M., Van Damme P., Caractères mor-
phologiques de Detarium microcarpum Guill.
et Perr. au sud du Mali, Fruits 57 (2002) 231–
238.
[3] AtanganaA.R., Ukafor V., Anegbeh P., Asaah E.,
Tchoundjeu Z., Fondoun J-M., Ndoumbe M.,
Leakey R.R.B., Domestication of Irvingia
gabonensis: 2. The selection of multiple traits
for potential cultivars from Cameroon and
Nigeria, Agrofor. Syst. 55 (2002) 221–229.
[4] Leakey R.R.B., Shackleton S., du Plessis P.,
Domestication potential of marula (Scleroca-
rya birrea subsp. caffra) in South Africa and
Namibia: 1. Phenotypic variation in fruit
traits, Agrofor. Syst. 64 (2005) 25–35.
[5] Bouvet J-M, Kelly B., Sanou H., Allal F., Com-
parison of marker- and pedigree-based
methods for estimating heritability in an
agroforestry population of Vitellaria paradoxa
C.F. Gaertn. (shea tree), Genet. Resour. Crop
Evol. 55 (2008) 1291–1301.
[6] Ekué M.R.M., Gailing O., Finkeldey R., Transfe-
rability of Simple Sequence Repeat (SSR)
Markers developed in Litchi chinensis to Bli-
ghia sapida (Sapindaceae), Plant. Mol. Biol.
Rep. 27 (2009) 570–574.
[8] Leakey R., Fuller S., Treloar T., Stevenson L.,
Hunter D., Nevenimo T., Binifa J., Moxon J.,
Characterization of tree-to-tree variation in
morphological, nutritional and medicinal
properties of Canarium indicum nuts, Agro-
for. Syst. 73 (2008) 77–87.
[7] Assogbadjo A.E., Sinsin B., Codjia J.T.C., Van
Damme P., Ecological diversity and pulp,
seed and kernel production of the baobab
(Adansonia digitata) in Benin, Belg. J. Bot.
138 (1) (2005) 47–56.
[9] Schlichting C.D., Pigliucci M., Phenotypic evo-
lution: a reaction norm perspective, Sinauer
Assoc., Sunderland, Mass., U.S.A., 1998.
[10] Gaoue O.G., Ticktin T., Impacts of bark and
foliage harvest on Khaya senegalensis
(Meliaceae) reproductive performance in
Benin, J. Appl. Ecol. 28 (2008) 34–40.
[11] Diggle P.J., Robust density estimation using
distance methods, Biometrika 62 (1) (1975)
39–48.
[12] Young L.J., Young, H., Statistical ecology: a
population perspective, Kluwer Acad. Publ.,
Boston, U.S.A., 1998.
[13] Diggle P.J., Besag J., Gleaves J.T., Statistical
analysis of spatial point by means of distance
methods, Biometrics 32 (1976) 659–667.
[14] Gaoue O.G., Ticktin T., Fulani knowledge of
the ecological impacts of Khaya senegalensis
(Meliaceae) foliage harvest in Benin and its
implications for sustainable harvest, Econ.
Bot. 63 (3) (2009) 256–270.
[15] Maundu P.M., Ngugi G.W., Kabuye C.H.S.,
Traditional food plants of Kenya, Natl. Mus.
Kenya, Nairobi, Kenya, 1999.
[16] Eyog Matig O., Gaoué O.G., Dossou B.,
Réseaux "Espèces Ligneuses Alimentaire",
C. R. Prem. Réun. Réseau, 11–13 déc. 2000,
CNSF, Ouagadougou, Burkina Faso, Inst.
Int. Ressour. Phytogénét., 2002.
[17] El-Siddig K., Gunasena H.P.M, Prasad B.A.
Pushpakumara D.K.N.G., Ramana K.V.R.,
Vijayanand P., Williams J.T., Tamarind, Tama-
rindus indica L, Southampt., Cent. Underutil.
Crops, Southampt., U.K., 2006.
[18] Bowe C., Haq N., Quantifying the global envi-
ronmental niche of an underutilized tropical
fruit tree (Tamarindus indica) using herba-
rium records, Agric. Ecosyst. Environ. (2010)
doi:10.1016/j.agee.2010.06.016.
[19] Soloviev P., Niang T.D., Gaye A., Totte A.,
Variabilité des caractères physico-chimiques
des fruits de trois espèces ligneuses de
cueillette récoltés au Sénégal : Adansonia
digitata,Balanites aegyptiaca et Tamarindus
indica, Fruits 59 (2004) 109–119.
[20] Diallo B.O., Biologie de la reproduction et
évaluation de la diversité génétique chez une
légumineuse : Tamarindus indica L. (Caesal-
pinioideae), Univ. Montp. II, Sci. Tech. Lan-
guedoc, thèse, Montp., France, 2001.
[21] Diallo B.O., Joly H.I., Mckey D., Hossaert-
Mckey M., Chevallier M.H., Genetic diversity
of Tamarindus indica populations: Any clues
on the origin from its current distribution?,
Afr. J. Biotechnol. 6 (7) (2007) 853–860.
[22] Diallo B.O., Mckey D., Chevallier M-H., Joly
H.I., Hossaert-Mckey M., Breeding system
and pollination biology of the semi-domesti-
cated fruit tree, Tamarindus indica L. (Legu-
minosae: Caesalpinioideae): Implications for
fruit production, selective breeding, and
conservation of genetic resources, Afr. J.
Biotechnol. 7 (22) (2008) 4068–4075.

T. indica in three ecological regions of Benin
Fruits, vol. 66 (2) 65
[23] Diallo B.O., Joly H.I., McKey D., Hossaert-
McKey, M., Chevallier M.H., Variation des
caractères biométriques des graines et des
plantules de neuf provenances de Tamarin-
dus indica L. (Caesalpinioideae), Fruits 65 (3)
(2010 153–167.
[24] Fandohan B., Assogbadjo A.E., Glèlè Kakaï,
R., Sinsin, B., Van Damme P., Impact of habi-
tat type on the conservation status of tama-
rind (Tamarindus indica L.) populations in the
W National Park of Benin, Fruits 65(1) (2010)
11–19.
[25] White F., The vegetation of Africa, UNESCO
Paris, France, Nat. Resour. Res. 20 (1983)
1–356.
[26] Anon., Troisième recensement général de la
population et de l'habitation (RGPH-3),
Résultats définitifs : Caractéristiques géné-
rales de la population, Inst. Natl. Stat. Appl.
Econ. (INSAE), Cotonou, Bénin, 2003.
[27] Adomou C.A, Sinsin B., van der Maesen
L.J.G., Phytosociological and chorological
approaches to phytogeography: a meso-
scale study in Benin, Syst. Geogr. Plants 76
(2006) 155–178.
[28] Grais R.F., Coulombier D., Ampuero J., Lucas
M.E.S., Barretto A.T., Jacquier G., Diaz F.,
Balandine S., Mahoudeau C., Brown V., Are
rapid population estimates accurate? A field
trial of two different assessment methods,
Disasters 30 (3) (2006) 364–376.
[29] Diggle P.J., Statistical analysis of spatial point
processes, Second ed., Arnold, Lond., U.K.,
2003.
[30] Lamien N., Tigabu M., Guinko S., Oden P.C.,
Variations in dendrometric and fruiting cha-
racters of Vitellaria paradoxa populations
and multivariate models for estimation of
fruit yield, Agrofor. Syst. (2007) DOI 10.1007/
s10457-006-9013-x.
[31] Leakey R.R.B., Fondoun J.-M., Atangana A.,
Tchoundjeu Z., Quantitative descriptors of
variation in the fruits and seeds of Irvingia
gabonensis, Agrofor. Syst. 50 (2000) 47–58.
[32] Hijmans R.J., Cameron S.E., Parra J.L., Jones
P.G., Jarvis A., The WorldClim interpolated
global terrestrial climate surfaces, Vers. 1.3.,
http://biogeo.berkeley.edu/, 2004.
[33] Kouyaté A.M., Aspect ethnobotaniques et
étude de la variabilité morphologique, biochi-
mique et phénologique de Detarium micro-
carpum Guill et Perr. au Mali, Univ. Ghent,
Thèse, Ghent, Belg., 2005.
[34] Ryan T.A., Joiner B.L., Normal probability
plots and tests for normality: technical report,
Univ. Park, Stat. Dep., Pa. State Univ., State
Coll., Pennsylvania, U.S.A., 1976.
[35] Aubréville A., Flore forestière soudano-gui-
néenne, AOF-Cameroun-AEF, Soc. Ed. ,
Marit. Colon., Paris, France, 1950.
[36] Jama B.A., Mohamed A.M., Mulatya J., Njui
A.N., Comparing the ‘‘Big Five’’: A fra-
mework for the sustainable management of
indigenous fruit trees in the drylands of East
and Central Africa, Ecol. Indic. 8 (2) (2008)
170–179.
[37] Maley J., Brenac P., Vegetation dynamics,
palaeoenvironments and climatic changes in
the forests of West Cameroon during the last
28,000 years, Rev. Palaeobot. Palynol. 99
(1998) 157–188.
[38] Salzmann U., Waller M., The Holocene vege-
tational history of the Nigerian Sahel based
on multiple pollen profiles, Rev. Palaeobot.
Palynol. 100 (1998) 39–72.
[39] Salzmann U., Are savannas degraded
forests? A Holocene pollen record from the
Sudanian zone of NE Nigeria, Veg. Hist.
Archaeobot. 9 (2000) 1–15.
[40] Salzmann U., Hoelzmann P., Morczineck I.,
Late Quaternary climate and vegetation of
the Sudanian zone of NE Nigeria, Quat. Res.
58 (2002) 73–83.
[41] Salzmann U., Hoelzmann P., The Dahomey
gap: An abrupt climatically induced rain forest
fragmentation in West Africa during the late
Holocene, Holocene 15 (2) (2005) 190–199.
[42] Collinet F., Essai de regroupement des princi-
pales espèces structurantes d’une forêt
dense humide d’après l’analyse de leur répar-
tition spatiale (Forêt de Paracou-Guyane),
Lab. Biom., Génét. Biol. Popul., UMR CNRS
5558, Group. Intérêt Sci. Sylvolab, Guyane,
France, 1997.
[43] Maranz S., Wiesman Z., Evidence for indige-
nous selection and distribution of the shea
tree, Vitellaria paradoxa, and its potential
significance to prevailing parkland savanna
tree patterns in sub-Saharan Africa, north of
the equator, J. Biogeogr. 30 (2003) 1505–
1516.
[44] WadtL.H.O., Kainer K.A., Gomes-Silva D.A.P.,
Population structure and nut yield of Berthol-
letia excelsa stand in southwestern Amazo-
nia, For. Ecol. Manag. 211 (2005) 371–384.
[45] Zuidema P.A., Boot R.G.A., Demography of
the Brazil nut tree (Bertholletia excelsa) in the

T. indica in three ecological regions of Benin
Fruits, vol. 66 (2) 66
Bolivian Amazon: Impact of seed extraction
on recruitment and population dynamics, J.
Trop. Ecol. 18 (2002) 1–31.
[46] Kelly D., Sork V.L., Mast seeding in perennial
plants: why, how, where?, Annu. Rev. Ecol.
Evol. Syst. 33 (2002) 427–447.
[47] Koenig W.D., Knops J.M.H., Patterns of
annual seed production by Northern Hemis-
phere trees: a global perspective, Am. Nat.
155 (2000) 59–69.
[48] Stephenson A.G., Flower and fruit abortion:
Proximate causes and ultimate functions,
Annu. Rev. Ecol. Evol. Syst. 12 (1981)
253–279.
[49] Ghosh S.N., Bera B., Kundu A., Roy S., Dutta
Ray S.K., Fruit production and quality impro-
vement in aonla (Emblica Officinalis Gaertn.)
through canopy management, J. Agric. Tech-
nol. 3 (8) ser. n°21 (2009) 40–43.
[50] Assogbadjo A.E., Kyndt T., Sinsin B., Gheysen
G., Van Damme, P., Patterns of genetic and
morphometric diversity in Baobab (Adanso-
nia digitata) populations across different cli-
matic zones of Benin (West Africa), Ann. Bot.
97 (2006) 819–830.
Benin / Tamarindus indica / distribución de la población / densidad de la
población / ructificación/ conocimiento indígena