Morphological Variability of Pods of Four Faidherbia ...
Morphological Variability of Pods of Four Faidherbia albidis
Provenances in Senegal
Ij t; ,s. /. > ,
!1
j II% .’ [,
M.H. Chevalier’, M. Sogna2, A.S. Sarr2, and P. Danthul
&*:; ,
..Y, ?
*f ‘f
Abstract
Pods of four Faidherbia albida provenances from Senegal, each with 20-30 progenies, were ‘an-
alyred for morphological variability. Pod traits (size and form) were measured and interpreted
using analysis of variante, multivariate analysis, principal component analysis, factorial analysis
of correspondences, and discriminant factor analysis. Pod dimensions exhibited the greatest mor-
phological variability among the traits measured. Intra-provenance divers@ was generally signifi-
tant, SO churacterization of geographic origins was dificulr. Nevertheless, morpho!ogical
characteristics of the Bode pods contrasted significantly with those of Kagnobon. The Ovadiour
pods were signijcantly differentfrom the otkr provenances.
Introduction
were used from the following four provenances by
separate progenies: the Merina Dakar provenance
Over the last decade, management of tree genetic re-
from northem Senegal (28 progenies); the Ovadiour
sources in Senegal has diminished. A number of
re-
provenance from the Groundnut Basin (30 pro-
cent projects have begun the task of conserving,
genies); the Kagnobon provenance from Casamance
protecting, and improving these resources (FAO
(20 progenies); and the Bode provenance from Casa-
1980). Faidherbia albida, due to its value in
mance (22 progenies). One hundred progenies in total
Senegalese agriculture, was one of the priority spe-
and ten pods per progeny were studied.
cies chosen for the program (FAO 1980; Lou~pe
Nine traits were measured and four indices were
1989). Diverse studies have since been carried out by
calculated from tbese (Table 1). Data were interpreted
international organizations throughout the range of
using analysis of variante. principal components and
this species. ‘The Direction des recherches sur les
correspondence! analysis, and discriminant factor
productions forestières (DRPP)/Institut sénégalais de
analysis.
recherches agricoles (ISRA) has implemented a pro-
gram to collect and evaluate genetic resources of
F. albida in Senegal.
Resuits
Tbis study focused on the biometry of pods. Prior
work has shown significant variability in pod form
Pod size for a11 tbe provenances was extremely vari-
and size of different acacia species as well as of
able. This was due to large differences in outaide pod
F. albida (Nongonierma 1977).
lengrh (112-270 mm) inside pod length (44412 mm),
distance between pod tips (12-71 mm), pod width
Materials and Methods
(17-67 mm), and pod mass (292-989 mg). pods con-
tained 7-23 seeds. Among the important indices cal-
In 1990,7 provenances of b, albia’a from northern to
culated, the spiraling index ranged from 0.17-0.90
southem Senegal were collected. Seeds and pods
with an average: of 0.53, the surface area index from
1. Centre technique forestier tropical
(ClFf’)finstitut
s6négalais de recherches agricoles
(ISRA), B.P. 2312, Dakar, Senegal.
2. Direction des recherches sur les productions foresti$res (DRPP)/lnstitot
sén6galais de recherches agricoles (ISRA), B.P. 2312, Dakar,
Senegal.
Chevatier,
M.H., Sogna, M., Sarr, A.S., and Danthu, P. 1992. Morphological variability
of pods of four Fuidherbia ahi& pmvms in
Senegal. Pages 67-70 in Fnidherbia albida in the West African semi-arid tropics: proceedings of a workshop, 22-26 Apr 1991. Niiy, Niger
(Vamknbeldt,
RJ., ed.). Patancheru, A.P. 502 324, India: International Crops Research Institute for
the Semi-Arid Tropics; aod Nairobi, Kenya:
lmematiooal
Centre for Research in Agmforestry.
,__r.l.l--ll_
~_--
-
----
-I-__I
--
Table L Measured traits and derived indices for
1613-9502 mm* and the ratio of width to average
length from 0.13-0.54.
F. albida pods, Dakar, Senegal, 1991.
Analysis of variante done showed a significant
Abbre-F
provenance effect for a11 the variables with the excep-
Parameter
viatian
tion of density index and number of angles (Tables 2
Traits
and 3). Differences between provenances were highly
Pod mass (mg)
PMS
significant for length between pod tips, pod width,
Outside length (dorsal cleft) (mm)
LDC
number of seeds per pod, and surface area index.
Inside length (sutura1 cleft) (mm)
LSC
Certain variables characterize each provenance. For
Length between pod tips (mm)
LEX
example, the Ovadiour provenance had the largest
Pod width (mm)
I
and straightest pods among the provenances studied
Number of seeds per pod
”
as well as the fewest number of seeds per pod.
Thicknesst
PTH
In order to determine correlations between vari-
Number of angles
NBA
ables, multivariate analyses were carried out using
Form*
FOR
correlation matrices of a11 the variables. A first anal-
Indices
ysis was done on the traits measured; the second was
Mean length (mm)
(LDC + LSC) i- 2 LOMO
done on the calculated indices.
Surface index (mm*)
LOMO + PWD
SUR:F
Spiraling index
LEX +LSC
SPIR.
Dcnsity index
LOMO + PMS
PDEIN
Principal Components and Correspondence
Form index
PWD + LOMO
FORM
Analysis
1. Thickaess indices are: 1 = thin; 2 = average; 3 = fleshy.
2. Angle indices are: 1 = sickle-shaped; 2 = wavy, straight;
Principal components analysis is a descriptive statisti-
3 = spirale& 4 = tightly spiraled.
cal method which is often used to examine the
inter-
Table 2. Means, standard errors, and CV for seven measured F. albida pod characters, Dakar, Senegal,
1991.
Pod characters
PMS’
LSC
LEX
P W D
Provenance
(mg)
I:mm)
(mm)
(mm
GPP
PTH
Kagnobon
Mean
664
179
73.3
42.7
26.2
2.3
18.6
SE (~0
167
40
15.4
8.0
5.5
1.0
2.4
c v (%)
26
22
21
19
21
44
13
Ovadiour
Mean
545
172
74.4
25.3
46.9
2.0
12.6
SE (~0
102
12
11.9
2.6
12.0
0.9
3.6
c v (%)
19
7
16
10
26
43
28
Bode
Mean
489
154
61.0
33.7
24.4
1.6
15.3
SE 6~)
121
29
12.5
13
2.6
0.7
3.0
c v (%)
25
19
21
39
11
41
20
Merina
Mean
507
157
67.9
44.4
30.2
2.0
17.5
SE 03
102
18
10.3
7.8
4.9
0.7
2.9
c v (%)
20
11
15
18
16
36
17
Total
Mean
542
165
69.4
36.0
33.1
2.0
15.8
SE (0)
134
27
13.5
11.6
12.0
0.8
3.9
c v (%)
25
16
19
32
36
43
25
1. See Table I for abbreviation codes.
68
Table 3. Means, standard errors, and CV for 5 calculated Fddherbiu albida pod characters, Dakar,
Senegal, 1991.
Fbd characters
LOMO’
SURF
Provenance
(mm)
(mm21
SPIR
PDEN
FDRM
Kagnobon
Mean
126
3326
0.61
0.20
0.22
SE (01
21
1005
0.15
0.04
0.05
c v (%)
22
30
25
21
23
Ovadiour
Mean
123
5832
0.35
0.23
0.38
SE (0)
10
1754
0.07
0.04
0.09
c v (%)
8
30
19
16
23
Bode
Mean
107
2613
0.56
0.23
0.23
SE (0)
19
543
0.19
0.05
0.05
cv (%)
18
21
33
21
20
Mer ina
Mean
113
3395
0.66
0.23
0.23
SE (a)
12
641
0.10
0.05
0.05
c v (%)
11
19
16
20
20
Total
Mean
117
3941
0.53
0.22
0.28
SE (01
19
1710
0.18
0.04
0.09
cv (%)
16
43
34
20
32
1. See Table 1 for abbreviation codes.
relationships among several variables. The first prin-
Correspondence
analysis was also performed,
cipal component defines a new variable that explains
since this method permits use of the qualitative as
as much as possible of the variability in the original
well and the quantitative variables. The latter are
data. The second principal component is made to be
transformed into classes. This analysis yielded results
independent of the first and in such a way that it
that were identical to those of the principal compo-
explains as much as possible of the variability that
nents. The quantitative variables were used in the lirst
remains. Often the first two components explain a
component. The qualitative variables were used in the
large proportion of the total variability and the data
second component and did net help in separating the
are then usually displayed in a two dimensional plot.
different groups of data.
In this study a principal component analysis was
conducted on both the traits measured and on the
derived indices. The first two axes on the measured
Discriminant Factorial Analysis
traits explained 62.1% of the variation. A display of
the data showed that the progenies of Ovadiour could
In order to determine the relationship of progenies to
be distinguished from the remaining progenies. A dis-
their provenances of origin and to test the provenance
tinguishable contrast also existed between the Bode
effect determined by the analyses of variantes, a dis-
and Kagnobon progenies. The results from an anal-
criminant factorial analysis was done on the traits
ysis of the derived variables did not give results that
measured, Only 71% of the progenies were definitely
were as clear as with the original variables.
classed in their provenance of origin (Table 4).
69
*-
.-- ----
-,--__--.--- ,-I--II-
-_----
Table 4. Classification of progenies into provenances by the discriminant factorial analysis~perforkxd on
measured traits.
.
Allocated provenance
Correctly
Provenance of origin
Kagnobon
Ovaldiour
Bode
Merina
classified (%)
Kagnobon
15’
0
3
2
75.02
Ovadiour
3
122
5
0
73.3
Bode
2
0
13
7
59.1
Merina
3
0
4
21
75.a
1 . Values are numbers of progeny classified in each group, e,g., in the first row of data, 15 Kagnobon progenies were correctly
classified in the analysis. Five were incorrectly classified.
2. Percentage of progeny correctly classified in its provenance of origin.
Among the provenances, Bode was the most het-
studied. Also, pods from provenances located at the
erogeneous since 41% of its progeny were classed as
northern and southem extremes of the Senegalese
Kagnobon and Merina. None of the progenies of
range of F. ulbida Will be analyzed. Finally, the com-
Bode, Kagnobon or Merina resembled those of Ova-
parison of morphological variability and that ob-
diour. Ovadiour progenies exhibited specific traits in
tained from enzymatic markers of the same origin
terms of pod width, length between extremities, pod
Will permit a more complete evaluation of genetic
mass, and number of seeds (Table 2).
resources of F. albida in Senegal.
Acknowledgment. We thank the Laboratoire de gé-
Discussion
nétique et amélioration des plantes, ORSTOM
(Dakar) where we performed the statistical analyses.
The great variability in F.
pod morphology
shown by this study confirmed results of a study by
Nongonierma (1977).
The following traits revealed the greatest diversity
Keferences
of F. afbidu pods collected throughout the Senegalese
range of the species: inside and outside pod length
FAO. 1980. Ressources genétiques d’essences arbo-
between pod tips; pod width; pod mass, and number
rées des zones arides et semi-arides. FAO/IBPGR
of seeds per pod. The different form indices (qualita-
Project. Rome, Italy: Food and Agriculture Organiza-
tive or calculated variables) and number of angles,
tion of the United Nations/International
Board for
provided lime benefit to the analysis. In any case, it
Plant Genetic Resources. 130 pp.
was very difficult, at the time of measuring, to class
the often variable and complex forms of the observed
Louppe, D. 1989. Influence de Fuidtzerbiu ulbidu sur
pods.
le rendement agricole-nouvelle contribution. Pre-
Intra- and interprovenance variability were of the
sented at the Workshop on ‘Forêt: Environnement et
same order; therefore, it was very difficult to charac-
Développement,’ Dakar, Senegal, 22-26 May 1989.
terize provenances according to their geographic
ori-
Dakar, Senegal: Institut des sciences de
I’environne-
gin. Nevertheless, pod size of the Bode provenance
ment/University of Dakar.
contrasted with that of the Kagnobon provenance.
Pods of the Ovadiour provenance were charac-
Nongonierma, A. 1977. Contribution à l’étude bio-
teristically stubby. T O verify if this particular tr.ait of
systematique du genre Acacia Miller en Afrique occi-
the Ovadiour provenance is unique to its location,
dentale. V. Caractères hiométriques des fruits.
three other provenances in the same region Will be
Bulletin IFAN 39:695-787.
70
Provenances in Senegal
Ij t; ,s. /. > ,
!1
j II% .’ [,
M.H. Chevalier’, M. Sogna2, A.S. Sarr2, and P. Danthul
&*:; ,
..Y, ?
*f ‘f
Abstract
Pods of four Faidherbia albida provenances from Senegal, each with 20-30 progenies, were ‘an-
alyred for morphological variability. Pod traits (size and form) were measured and interpreted
using analysis of variante, multivariate analysis, principal component analysis, factorial analysis
of correspondences, and discriminant factor analysis. Pod dimensions exhibited the greatest mor-
phological variability among the traits measured. Intra-provenance divers@ was generally signifi-
tant, SO churacterization of geographic origins was dificulr. Nevertheless, morpho!ogical
characteristics of the Bode pods contrasted significantly with those of Kagnobon. The Ovadiour
pods were signijcantly differentfrom the otkr provenances.
Introduction
were used from the following four provenances by
separate progenies: the Merina Dakar provenance
Over the last decade, management of tree genetic re-
from northem Senegal (28 progenies); the Ovadiour
sources in Senegal has diminished. A number of
re-
provenance from the Groundnut Basin (30 pro-
cent projects have begun the task of conserving,
genies); the Kagnobon provenance from Casamance
protecting, and improving these resources (FAO
(20 progenies); and the Bode provenance from Casa-
1980). Faidherbia albida, due to its value in
mance (22 progenies). One hundred progenies in total
Senegalese agriculture, was one of the priority spe-
and ten pods per progeny were studied.
cies chosen for the program (FAO 1980; Lou~pe
Nine traits were measured and four indices were
1989). Diverse studies have since been carried out by
calculated from tbese (Table 1). Data were interpreted
international organizations throughout the range of
using analysis of variante. principal components and
this species. ‘The Direction des recherches sur les
correspondence! analysis, and discriminant factor
productions forestières (DRPP)/Institut sénégalais de
analysis.
recherches agricoles (ISRA) has implemented a pro-
gram to collect and evaluate genetic resources of
F. albida in Senegal.
Resuits
Tbis study focused on the biometry of pods. Prior
work has shown significant variability in pod form
Pod size for a11 tbe provenances was extremely vari-
and size of different acacia species as well as of
able. This was due to large differences in outaide pod
F. albida (Nongonierma 1977).
lengrh (112-270 mm) inside pod length (44412 mm),
distance between pod tips (12-71 mm), pod width
Materials and Methods
(17-67 mm), and pod mass (292-989 mg). pods con-
tained 7-23 seeds. Among the important indices cal-
In 1990,7 provenances of b, albia’a from northern to
culated, the spiraling index ranged from 0.17-0.90
southem Senegal were collected. Seeds and pods
with an average: of 0.53, the surface area index from
1. Centre technique forestier tropical
(ClFf’)finstitut
s6négalais de recherches agricoles
(ISRA), B.P. 2312, Dakar, Senegal.
2. Direction des recherches sur les productions foresti$res (DRPP)/lnstitot
sén6galais de recherches agricoles (ISRA), B.P. 2312, Dakar,
Senegal.
Chevatier,
M.H., Sogna, M., Sarr, A.S., and Danthu, P. 1992. Morphological variability
of pods of four Fuidherbia ahi& pmvms in
Senegal. Pages 67-70 in Fnidherbia albida in the West African semi-arid tropics: proceedings of a workshop, 22-26 Apr 1991. Niiy, Niger
(Vamknbeldt,
RJ., ed.). Patancheru, A.P. 502 324, India: International Crops Research Institute for
the Semi-Arid Tropics; aod Nairobi, Kenya:
lmematiooal
Centre for Research in Agmforestry.
,__r.l.l--ll_
~_--
-
----
-I-__I
--
Table L Measured traits and derived indices for
1613-9502 mm* and the ratio of width to average
length from 0.13-0.54.
F. albida pods, Dakar, Senegal, 1991.
Analysis of variante done showed a significant
Abbre-F
provenance effect for a11 the variables with the excep-
Parameter
viatian
tion of density index and number of angles (Tables 2
Traits
and 3). Differences between provenances were highly
Pod mass (mg)
PMS
significant for length between pod tips, pod width,
Outside length (dorsal cleft) (mm)
LDC
number of seeds per pod, and surface area index.
Inside length (sutura1 cleft) (mm)
LSC
Certain variables characterize each provenance. For
Length between pod tips (mm)
LEX
example, the Ovadiour provenance had the largest
Pod width (mm)
I
and straightest pods among the provenances studied
Number of seeds per pod
”
as well as the fewest number of seeds per pod.
Thicknesst
PTH
In order to determine correlations between vari-
Number of angles
NBA
ables, multivariate analyses were carried out using
Form*
FOR
correlation matrices of a11 the variables. A first anal-
Indices
ysis was done on the traits measured; the second was
Mean length (mm)
(LDC + LSC) i- 2 LOMO
done on the calculated indices.
Surface index (mm*)
LOMO + PWD
SUR:F
Spiraling index
LEX +LSC
SPIR.
Dcnsity index
LOMO + PMS
PDEIN
Principal Components and Correspondence
Form index
PWD + LOMO
FORM
Analysis
1. Thickaess indices are: 1 = thin; 2 = average; 3 = fleshy.
2. Angle indices are: 1 = sickle-shaped; 2 = wavy, straight;
Principal components analysis is a descriptive statisti-
3 = spirale& 4 = tightly spiraled.
cal method which is often used to examine the
inter-
Table 2. Means, standard errors, and CV for seven measured F. albida pod characters, Dakar, Senegal,
1991.
Pod characters
PMS’
LSC
LEX
P W D
Provenance
(mg)
I:mm)
(mm)
(mm
GPP
PTH
Kagnobon
Mean
664
179
73.3
42.7
26.2
2.3
18.6
SE (~0
167
40
15.4
8.0
5.5
1.0
2.4
c v (%)
26
22
21
19
21
44
13
Ovadiour
Mean
545
172
74.4
25.3
46.9
2.0
12.6
SE (~0
102
12
11.9
2.6
12.0
0.9
3.6
c v (%)
19
7
16
10
26
43
28
Bode
Mean
489
154
61.0
33.7
24.4
1.6
15.3
SE 6~)
121
29
12.5
13
2.6
0.7
3.0
c v (%)
25
19
21
39
11
41
20
Merina
Mean
507
157
67.9
44.4
30.2
2.0
17.5
SE 03
102
18
10.3
7.8
4.9
0.7
2.9
c v (%)
20
11
15
18
16
36
17
Total
Mean
542
165
69.4
36.0
33.1
2.0
15.8
SE (0)
134
27
13.5
11.6
12.0
0.8
3.9
c v (%)
25
16
19
32
36
43
25
1. See Table I for abbreviation codes.
68
Table 3. Means, standard errors, and CV for 5 calculated Fddherbiu albida pod characters, Dakar,
Senegal, 1991.
Fbd characters
LOMO’
SURF
Provenance
(mm)
(mm21
SPIR
PDEN
FDRM
Kagnobon
Mean
126
3326
0.61
0.20
0.22
SE (01
21
1005
0.15
0.04
0.05
c v (%)
22
30
25
21
23
Ovadiour
Mean
123
5832
0.35
0.23
0.38
SE (0)
10
1754
0.07
0.04
0.09
c v (%)
8
30
19
16
23
Bode
Mean
107
2613
0.56
0.23
0.23
SE (0)
19
543
0.19
0.05
0.05
cv (%)
18
21
33
21
20
Mer ina
Mean
113
3395
0.66
0.23
0.23
SE (a)
12
641
0.10
0.05
0.05
c v (%)
11
19
16
20
20
Total
Mean
117
3941
0.53
0.22
0.28
SE (01
19
1710
0.18
0.04
0.09
cv (%)
16
43
34
20
32
1. See Table 1 for abbreviation codes.
relationships among several variables. The first prin-
Correspondence
analysis was also performed,
cipal component defines a new variable that explains
since this method permits use of the qualitative as
as much as possible of the variability in the original
well and the quantitative variables. The latter are
data. The second principal component is made to be
transformed into classes. This analysis yielded results
independent of the first and in such a way that it
that were identical to those of the principal compo-
explains as much as possible of the variability that
nents. The quantitative variables were used in the lirst
remains. Often the first two components explain a
component. The qualitative variables were used in the
large proportion of the total variability and the data
second component and did net help in separating the
are then usually displayed in a two dimensional plot.
different groups of data.
In this study a principal component analysis was
conducted on both the traits measured and on the
derived indices. The first two axes on the measured
Discriminant Factorial Analysis
traits explained 62.1% of the variation. A display of
the data showed that the progenies of Ovadiour could
In order to determine the relationship of progenies to
be distinguished from the remaining progenies. A dis-
their provenances of origin and to test the provenance
tinguishable contrast also existed between the Bode
effect determined by the analyses of variantes, a dis-
and Kagnobon progenies. The results from an anal-
criminant factorial analysis was done on the traits
ysis of the derived variables did not give results that
measured, Only 71% of the progenies were definitely
were as clear as with the original variables.
classed in their provenance of origin (Table 4).
69
*-
.-- ----
-,--__--.--- ,-I--II-
-_----
Table 4. Classification of progenies into provenances by the discriminant factorial analysis~perforkxd on
measured traits.
.
Allocated provenance
Correctly
Provenance of origin
Kagnobon
Ovaldiour
Bode
Merina
classified (%)
Kagnobon
15’
0
3
2
75.02
Ovadiour
3
122
5
0
73.3
Bode
2
0
13
7
59.1
Merina
3
0
4
21
75.a
1 . Values are numbers of progeny classified in each group, e,g., in the first row of data, 15 Kagnobon progenies were correctly
classified in the analysis. Five were incorrectly classified.
2. Percentage of progeny correctly classified in its provenance of origin.
Among the provenances, Bode was the most het-
studied. Also, pods from provenances located at the
erogeneous since 41% of its progeny were classed as
northern and southem extremes of the Senegalese
Kagnobon and Merina. None of the progenies of
range of F. ulbida Will be analyzed. Finally, the com-
Bode, Kagnobon or Merina resembled those of Ova-
parison of morphological variability and that ob-
diour. Ovadiour progenies exhibited specific traits in
tained from enzymatic markers of the same origin
terms of pod width, length between extremities, pod
Will permit a more complete evaluation of genetic
mass, and number of seeds (Table 2).
resources of F. albida in Senegal.
Acknowledgment. We thank the Laboratoire de gé-
Discussion
nétique et amélioration des plantes, ORSTOM
(Dakar) where we performed the statistical analyses.
The great variability in F.
pod morphology
shown by this study confirmed results of a study by
Nongonierma (1977).
The following traits revealed the greatest diversity
Keferences
of F. afbidu pods collected throughout the Senegalese
range of the species: inside and outside pod length
FAO. 1980. Ressources genétiques d’essences arbo-
between pod tips; pod width; pod mass, and number
rées des zones arides et semi-arides. FAO/IBPGR
of seeds per pod. The different form indices (qualita-
Project. Rome, Italy: Food and Agriculture Organiza-
tive or calculated variables) and number of angles,
tion of the United Nations/International
Board for
provided lime benefit to the analysis. In any case, it
Plant Genetic Resources. 130 pp.
was very difficult, at the time of measuring, to class
the often variable and complex forms of the observed
Louppe, D. 1989. Influence de Fuidtzerbiu ulbidu sur
pods.
le rendement agricole-nouvelle contribution. Pre-
Intra- and interprovenance variability were of the
sented at the Workshop on ‘Forêt: Environnement et
same order; therefore, it was very difficult to charac-
Développement,’ Dakar, Senegal, 22-26 May 1989.
terize provenances according to their geographic
ori-
Dakar, Senegal: Institut des sciences de
I’environne-
gin. Nevertheless, pod size of the Bode provenance
ment/University of Dakar.
contrasted with that of the Kagnobon provenance.
Pods of the Ovadiour provenance were charac-
Nongonierma, A. 1977. Contribution à l’étude bio-
teristically stubby. T O verify if this particular tr.ait of
systematique du genre Acacia Miller en Afrique occi-
the Ovadiour provenance is unique to its location,
dentale. V. Caractères hiométriques des fruits.
three other provenances in the same region Will be
Bulletin IFAN 39:695-787.
70