Content uploaded by Thaweesakdi Boonkerd
Author content
All content in this area was uploaded by Thaweesakdi Boonkerd
Content may be subject to copyright.
A phenetic study of Cassia sensu lato (Leguminosae-Caesalpinioideae:
Cassieae: Cassiinae) in Thailand
T. Boonkerd
1
, S. Pechsri
1
,and B. R. Baum
2
1
Department of Botany, Faculty of Science, Chulalongkorn University, Bangkok, Thailand
2
Agriculture & Agri-Food Canada, Eastern Cereal and Oilseed Research Centre, Ottawa, Ontario, Canada
Received December 3, 2003; accepted November 18, 2004
Published online: May 9, 2005
ÓSpringer-Verlag 2005
Abstract. Cassia L. sensu lato, a large heteroge-
neous genus of flowering plants, occurs naturally
in the tropics around the world. Recent works
based on floral morphology have supported a
division of this genus into three genera, namely
Cassia L. s. str., Chamaecrista Moench and Senna
Mill. In order to investigate this new classifica-
tion, 508 specimens of 18 taxa of the genus
Cassia s.l. grown in Thailand were analyzed using
cluster analysis and canonical discriminant anal-
ysis. The total 32 vegetative and reproductive
morphological characters were employed in these
analyses. In cluster analysis, Cassia s.l. can be
separated into four groups, respectively viz.
Chamaecrista,Senna alata,Senna and Cassia s/
str. The four-cluster grouping is discussed. From
a canonical discriminant analysis using the
four-cluster grouping as a priori groups, it can
be concluded that Cassia s. str., Senna, and
Chamaecrista are indeed distinct taxa. The three
most important characters that separate the three
genera are filament length, fruit length, and ovary
stalk length. These quantitative characters, to-
gether with some qualitative characters, were
useful in constructing an identification key to
these genera. Among the three genera, it was also
found that Senna is rather a heterogeneous taxon.
The difference between the studied species was
discussed.
Key words: Cassia sensu lato, Chamaecrista, cluster
analysis, discriminant analysis, morphometric
analysis, Senna.
The genus Cassia s.l. erected by Linnaeus
(1753), including some 650 species of trees,
shrubs and herbs and has a pantropical
distribution, but very few in Asia (Irwin and
Barneby 1981, 1982; Willis 1988). It also has a
long history of economic importance, espe-
cially as medicinal plants (Burkill 1935). The
genus is characterized by having even-pinnate
leaves with stipules of various types; flowers
slightly irregular, solitary, racemose or panic-
ulate; stamens 5 or 10, frequently unequal;
fruits sessile or stalked, terete or flattened and
often partitioned crosswise (Irwin and Barneby
1982, Larsen et al. 1984). Bentham (1871) in
his revision of the genus indicated that it is
comprised of three distinct groups. Recently,
three separate genera, i.e. Cassia sensu stricto,
Chamaecrista Moench, and Senna Mill. were
recognized based on characters of filaments
and the presence or absence of bracteoles.
(Irwin and Barneby 1981, 1982; Lock 1988;
Larsen 1993; Hou et al. 1996; Mabberley 1997;
Singh 2001). These three genera belong to the
tribe Cassieae and subtribe Cassiinae of the
Leguminosae-Caesalpinioideae.
Many attempts have been made to investi-
gate the taxonomic status of these segregated
genera. For example, Tucker (1996) investi-
gated and compared features of inflorescence
and floral organ initiation and development
among one species of Cassia sensu stricto, six
species of Senna, and two species of Chamae-
crista. She concluded that difference in floral
ontogeny supported the segregation of these
three genera. Recently, Ghareeb et al. (1999)
worked on seed proteins, chromosome number,
and morphological characters of 10 species of
Cassia belonging to subgenera Fistula and
Senna. The result obtained from their work
supported the segregation of the genus Cassia
into three distinct taxa as was suggested by
Bentham (1871). Likewise, a phylogenetic
analysis of nucleotide sequence in chloroplast
trnL intron of the subtribe Cassiinae supported
the segregation of Chamaecrista and Senna
from Cassia s.l. (Bruneau and Forest 2001).
Although restricted to material from Thai-
land and the understanding of its flora, the
present research aimed to find additional
evidence to support the segregation of Cassia
s. str., Chamaecrista and Senna from Cassia s.l.
as suggested by Irwin and Barneby (1982) by
using morphometric analyses.
Materials and methods
Specimen collections. Eighteen taxa of Cassia s. str.,
Chamaecrista and Senna growing in Thailand were
used for morphometric analyses. These include
indigenous, introduced and cultivated species. All
specimens were collected from known localities in
Thailand (Larsen et al. 1984) during June 2000 to
May 2002. They were determined to genus and
species based on the key to species by Larsen et al.
(1984); Hou et al. (1996), and classified to genus
according to Irwin and Barneby (1982) as shown in
Table 1. Specimens of each species were confirmed
by comparison to the voucher herbarium specimens
deposited at BCU and BKF (Table 1). The voucher
specimens from this study have been deposited at
BCU (Herbarium abbreviations according to Holm-
gren and Holmgren 2004).
Data analysis
Quantitative characters. Five hundred and
eight (508) specimens were used for all analyses.
Thirty two quantitative characters of both vegeta-
tive and reproductive parts (Table 2) were sub-
jected to cluster and canonical discriminant
analyses (data matrix is available from the corre-
sponding author).
A sequential, agglomerative, hierarchical and
nested (SAHN) clustering (Sneath and Sokal 1973)
was performed using average taxonomic distance
and the unweighted pair-group method with arith-
metic averages (UPGMA) implemented in
NTSYSpc version 2.10 (Rohlf 2000) to place
individual specimens into groups. To reduce the
effects of different scales of measurement for
different characters, the values for each character
were standardized using procedure STAND.
A subset of characters that maximized differ-
ences among the groups determined by cluster
analysis was selected by stepwise discriminant
analysis. To characterize mean differences among
the species canonical discriminant analysis was
used to acquire insight into group differences and
to estimate character weights from correlations
between canonical variables and original variables,
the procedure CLASSIFY in SPSSpc-FW, release
9.0 (Anonymous 1998) was used. In comparison,
the resulted groupings from cluster analysis and the
18 taxa of the Cassia s.l. were used as a priori group
for a series of discriminant analysis.
To summarize the range of variation between
and within the segregated groups on each charac-
ter, univariate analysis was performed. Boxplots of
the most important characters were carried out
using SPSSpc-FW (Anonymous 1998).
Qualitative characters. Qualitative characters
of the three segregate genera are tabulated and
compared with the results from numerical analyses
for their importance in discriminating the 18 taxa.
Subsequently, the best characters for separating the
segregate genera as suggested by DELTA were
used to construct a key to genera (Dallwitz et al.
1993).
Results and discussion
Quantitative characters
Cluster analysis. The dendrogram separated
the 508 specimens into four groups at the
1.30 level of average taxonomic distance.
The first group consists of two species, i.e.
Chamaecrista pumila and Chamaecrista
leschenaultiana. The second group is solely
Senna alata. The third group is composed of
ten species of Senna, excluding Senna alata
and Senna spectabilis. The fourth group is
comprised of five species: Cassia fistula,
C. javanica,C. grandis,C. bakeriana, and
Senna spectabilis.
From the cluster analysis it can be seen that
Senna is a rather heterogeneous taxon, whereas
Cassia s. str. and Chamaecrista are rather
homogeneous. The taxonomic status of Senna
alata and Senna spectabilis is worth future
investigation.
Table 1. List of 18 taxa of the three segregated genera from Cassia s.l.
No. Taxa Specimens examined
1. Cassia javanica L. Herb. Trip 893, S. Poothong 4, S. Pechsri 54 (BCU); K. Larsen
et al. 30873, 33596 (BKF)
2. C. grandis L.f. S. Pechsri 52 (BCU); T. Santisuk 1627, Th. S. et al. 21 (BKF)
3. C. bakeriana Craib W. Busapavanija 20, S. Pechsri 53 (BCU); F. Konta et al. 4085
(BKF)
4. C. fistula L. S. Poothong 41, Sinchai 653, S. Pechsri 51 (BCU); Char-
arnmayu 434 (BKF)
5. Chamaecrista pumila (Lam.)
K. Larsen
S. Pechsri 67 (BCU); G. Murata et al. 3863, J.F. Maxwell 86–
1020 (BKF)
6. Ch. leschenaultiana (DC.)
O.Deg.
S. Pechsri 50, 68 (BCU); Deer 331, T. Smitinand 4966 (BKF)
7. Senna obtusifolia (L.) H.S.
Irwin & Barneby
S. Pechsri 77 (BCU)
8. S. alata (L.) Roxb. S. Pechsri 55, C. Sombongse 7, U. Damsri 47 (BCU);
J.F. Maxwell 94–1220 (BKF)
9. S. hirsuta (L.) H.S. Irwin
& Barneby
S. Pechsri 60 (BCU); FRDU & P.C. van Welzen 77, Prayun 3
(BKF)
10. S. sophora (L.) Roxb. BNS 630, S. Pechsri 78, S.P. 62 (BCU); S. Unjai 98,
H.M. Burkill 1276 (BKF)
11. S. surattensis (Burm.f.)
H.S. Irwin &
Barneby subsp. surattensis
C. Thanakorn 5, S. Pechsri 66, S.P. 74 (BCU); P. Suvarnkoses,
P. Hampanond (BKF)
12. S. surattensis (Burm.f.)
H.S. Irwin & Barneby
subsp. glauca
Sinchai 663, S. Pechsri 66 (BCU); J.F. Maxwell 88–1181, De
230 (BKF)
13. S. occidentalis (L.) Link A. Chutinthorn 20, T. Jonganurak 152, S. Pechsri 61 (BCU);
K. Bunchuai 120, D. Bunpheng 1 (BKF)
14. S. tora (L.) Roxb. O. Thaithong 201, V. Srisuvanatach 4, S. Pechsri 77 (BCU);
C. Phengklai et al. 3320, K. Larsen et al. 34189 (BKF)
15. S. timoriensis (DC.)
H.S. Irwin & Barneby
Herb. Trip 638, BNS 520, S. Pechsri 79 (BCU); P.B. 65,
T. Smitinand 8651 (BKF)
16. S. garrettiana (Craib)
H. S. Irwin & Barneby
O. Thaithong 249, B. Na Songkhla 260, S. Pechsri 59 (BCU);
FTP. 31411,Luang Vanpruk 53 (BKF)
17. S. spectabilis (DC.)
H.S. Irwin & Barneby
S. Pechsri 56 (BCU); H. & G.C. 148, N. Fukuoka 62004 (BKF)
18. S. siamea (Lam.)
H.S. Irwin & Barneby
C. Siwasilp 9, K. Sridith 183, S. Pechsri 57 (BCU); J.F. Max-
well 86–495, C Phengklai et al. 3751 (BKF)
Canonical discriminant analysis
The four-cluster groupings. Twenty six out of
thirty-two characters were used in this analysis
(Table 2). Canonical variable 1 is 99% corre-
lated with the all characters in analysis and the
variance explained by it is 60.2% (Table 3). It
is most highly associated with the following 9
characters FTL (32), OVL (28), POLL (13),
OSL (27), PCL (17), FML (23), PTL (20),
FLD (14), and PTW (21). The variance
explained by canonical variable 2 is 23.8% of
the total. This axis is most highly associated
with 11 characters ATL (25), TLW (8), PED
(2), BTWP (9), SPL (18), STL (30), ATD (26),
BTW (16), TLL (7), SPW (19), and LWR (10).
The third canonical variable explains 16.0% of
the total variance. This axis is most highly
associated with character NOL (6), PET (1),
RCL (3) and OVD (29). According to stepwise
discriminant analysis six variables, viz. RCD
(4), DBLP (5), LS (11), LMW (12), BTL (15),
and FMD (24) were not selected for further
use in canonical discriminant analysis
(Table 2).
The canonical plot on the two canonical
axes (Fig. 1) shows that the four-cluster
groupings are not distinct. The four entities
separated into three groups on canonical axis
1. Group 1 is composed of Cassia s. str. and
Senna spectabilis. Group 2 is composed of two
species of Chamaecrista. Group 3, the largest
group, is composed of the ten taxa of Senna (3)
and Senna alata (4). However, Senna alata (4)
did separate from the ten taxa of Senna (3) on
canonical axis 2. This result indicated the
heterogeneity within the genus Senna, namely
Senna alata tends to be a subgroup of the
genus Senna.
Groupings of the 18 taxa. The canonical
plot on the two canonical axes (Fig. 2) shows
that the eighteen taxa are separated into 3
groups on both canonical axis 1 and axis 2.
Group I (1–4) is composed of 4 species of
Cassia s.str., namely C. javanica (1), C. grandis
(2), C. bakeriana (3), and C. fistula (4); whereas
group II consists of Chamaecrista pumila (5)
and Chamaecrista leschenaultiana (6); and
group III is the largest group (7–18) including
all 12 taxa of Senna.
The nature of the groups differences is
shown by the pooled within canonical structure
(Table 2). Canonical variable 1 is 99.7%
correlated with the thirty characters and the
variance explained by it is 42.1% (information
available on request). It is most highly associ-
ated with character FML (23). Canonical
variable 2 explains 19.0% of the total variance.
This axis is most highly associated with TLW
(8), ATL (25), POLL (13) and PED (2). The
two variables RCD (4) and LMW (12) were
not selected by stepwise discriminant analysis
to be used in further canonical discriminant
analysis (Table 2).
Based on the results of cluster analysis
together with canonical discriminant analyses
it is reasonable to segregate the 18 taxa of
Cassia s.l. into three distinct genera, i.e. Cassia
s.str., Chamaecrista, and Senna. As can be seen
from Fig. 2, the phenetic gaps between groups
are larger than those within groups. If these
three genera are delimited on phylogenetic
evidence (not presented here), then the result-
ing groupings are more or less phenetically
cohesive. However, Senna shows some heter-
ogeneity. Two species of Senna, i.e. Senna alata
and Senna spectabilis seem to fall into two
distinct subgroups within the Senna.
The three segregated genera. Univariate F
values of the 25 characters used in canonical
discriminant analysis as well as their pooled
within canonical structure of the three segre-
gated genera are shown in Table 2. Canonical
variable 1 (Axis 1) is most highly associated
with characters FML (23), FTL (32), OSL
(27), OVL (28), PCL (17), FLD (14), PTL (20),
PTW (21), DBLP (5), STD (31) and OVD (29)
in descending order of the absolute values of
the correlation (Table 2). Canonical variable 1
is 98.8% correlated with these characters and
the variance explained by it is 63.8% (infor-
mation available on request). A high canonical
correlation indicates a strong relationship
between the specific linear combination of
variables (Baum and Bailey 1984). This axis
Table 2. Thirty two characters, with their methods of scoring used in the study of Cassia s.l. Univariate F values of the different character used in
the canonical discriminant analysis and pooled within canonical structure using (I) 4 clustering groupings, (II) 18 taxa groupings, and (III) 3
segregated genera groupings
No. Abbreviation Characters (I) (II) (III)
F-value AXIS 1 AXIS 2 AXIS 3 F-value AXIS 1 AXIS 2 AXIS 3 F-value AXIS 1 AXIS 2
1. PET
a
Petiole length in mm 115.98 0.08 0.05 0.18 116.28 0.06 0.10 )0.16 117.08 0.05 0.12
2. PED
b
Petiole diameter in mm 568.90 0.20 0.26 )0.19 222.67 0.12 0.20 0.12 206.78 0.09 0.12
3. RCL
b
Rachis length in mm 148.05 0.11 0.06 )0.14 467.63 0.16 0.13 0.44 87.71 0.1 0.05
4. RCD Rachis diameter in mm 128.73 0.11 0.12 )0.05 145.42 0.11 0.14 )0.03 94.83 0.05 0.09
5. DBLP
c
Distance between 1
st
and
2
nd
leaflet pair
90.99 0.09 0.09 )0.06 344.00 0.15 0.13 0.09 86.23 0.08 0.05
6. NOL
a
Number of leaflet 97.88 )0.02 )0.09 )0.18 602.10 )0.03 )0.23 0.56 83.02 0 )0.12
7. TLL
a
Terminal leaflet length
in mm
150.62 0.11 0.12 )0.08 247.76 0.13 0.16 0.01 114.85 0.08 0.09
8. TLW
a
Terminal leaflet width
in mm
566.27 0.20 0.28 0.04 849.25 0.24 0.48 )0.14 578.60 0.14 0.25
9. BTWP
a
Distance from base to
the widest point of leaflet
340.87 0.12 0.23 )0.16 170.40 0.09 0.17 0.04 119.90 0.08 0.1
10. LWR
a
Terminal leaflet length
to width ratio
37.75 )0.01 )0.11 0.02 124.90 )0.02 )0.10 0.07 17.04 )0.02 )0.04
11. LS
c
Terminal leaflet shape
(calculated by BTW/TLL)
23.99 )0.07 0.03 )0.03 146.63 )0.05 )0.05 0.01 23.13 )0.03 )0.05
12. LMW
e
Lamina width in mm 158.52 0.10 0.13 )0.07 256.12 0.11 0.15 0.04 127.67 0.08 0.1
13. POLL
a
Petiolule length in mm 421.93 0.21 0.15 0.10 438.94 0.22 0.28 )0.06 647.27 0.18 0.23
14. FLD
a
Flower diameter in mm 146.71 0.14 )0.04 0.01 480.80 0.26 0.08 0.13 197.13 0.13 0.04
15. BTL
c
Bracteole length in mm 135.59 0.01 0.15 )0.15 364.40 0.04 0.18 0.34 29.48 0.03 0.06
16. BTW
a
Bracteole width in mm 107.78 0.03 0.17 )0.07 590.73 0.00 0.29 0.51 39.09 0.01 0.08
17. PCL
a
Pedicel length in mm 266.58 0.17 )0.09 0.12 562.52 0.28 0.08 0.07 210.41 0.14 0.06
18. SPL
a
Largest sepal length in mm 184.90 0.06 0.22 )0.06 223.58 0.01 0.23 0.14 103.86 0.02 0.13
19. SPW
b
Largest sepal width in mm 88.97 0.07 0.11 0.09 152.23 0.01 0.21 0.04 223.23 0.06 0.19
20. PTL
a
Largest petal length in mm 175.01 0.15 0.01 )0.05 468.26 0.25 0.14 0.19 179.80 0.12 0.06
21. PTW
a
Largest petal width in mm 110.95 0.11 0.07 )0.03 205.42 0.14 0.13 0.13 141.56 0.1 0.08
Table 2. (Continued)
No. Abbreviation Characters (I) (II) (III)
F-value AXIS 1 AXIS 2 AXIS 3 F-value AXIS 1 AXIS 2 AXIS 3 F-value AXIS 1 AXIS 2
22. PSL
a
Petals stalk length in mm
(largest petal)
92.09 0.09 0.07 )0.07 126.92 0.06 0.14 0.09 66.97 0.04 0.09
23. FML
a
Filament length in mm
(largest fertile stamen)
230.32 0.16 )0.07 )0.09 1481.15 0.52 )0.17 0.08 1241.27 0.34 )0.06
24. FMD
d
Filament diameter in mm
(largest fertile stamen)
71.91 0.13 0.03 )0.10 88.44 0.06 0.07 0.10 71.80 0.1 0.02
25. ATL
a
Anther length in mm
(largest fertile stamen)
276.30 0.02 0.30 0.00 539.46 )0.03 0.42 0.30 200.79 )0.03 0.18
26. ATD
b
Anther diameter in mm
(largest fertile stamen)
281.27 0.13 0.18 )0.15 287.64 0.12 0.20 0.28 124.20 0.07 0.1
27. OSL
a
Ovary stalk length in mm 297.16 0.19 )0.04 )0.08 533.27 0.31 0.01 0.10 713.85 0.26 0.01
28. OVL
a
Ovary length in mm 531.76 0.26 0.01 )0.11 772.52 0.35 0.14 0.02 487.88 0.21 0.08
29. OVD
a
Ovary diameter in mm 21.39 0.02 0.05 )0.07 54.64 0.02 0.02 0.11 7.54 0.03 0.01
30. STL
a
Style length in mm 117.60 )0.01 0.19 0.03 251.98 )0.02 0.22 0.21 63.19 )0.02 0.1
31. STD
b
Style diameter in mm 164.26 0.13 )0.08 )0.08 97.37 0.09 0.01 0.06 70.06 0.07 0.02
32. FTL
a
Fruit length in mm 658.89 0.29 )0.03 )0.09 779.46 0.39 0.10 0.06 905.07 0.29 0.07
Note: Character followed by superscript alphabet indicates character entered in stepwise discriminant analysis and subsequently used in canonical
discriminant analysis:-
a
: (I), (II) and (III);
b
: (I) and (II);
c
: (II) and (III);
d
: (II);
e
: (III)
is most important for separating Chamaecrista,
and Senna from Cassia s.str. Canonical vari-
able 2 explains 36.2% of the total variance.
This axis is most highly associated with TLW
(8), POLL (13), SPW (19), ATL (25), SPL (18),
PET (1), NOL (6), BTWP (9), STL (30), LMW
(12), PSL (22), TLL (7), BTL (15), LS (11) and
LWR (10) in descending order in a corre-
sponding manner. The F values (Table 2) very
nearly reflect the association with canonical
Axis 1 because of its high correlation and high
amount of variance explained. The seven
variables PED (2), RCL (3), RCD (4), SPW
(19), FMD (24), ATD (26), and STD (31) were
not selected for further used in canonical
discriminant analysis (Table 2).
Table 4 shows means and standard devia-
tions of 32 quantitative characters of Cassia s.
str., Chamaecrista and Senna. Boxplots of the
four most important characters that segregate
these three genera are presented in Fig. 3. It
can be seen that filament length (FML) is the
most important character.
Classificatory discriminant analysis
Twenty five of the thirty two quantitative
characters (Table 2) were determined by step-
wise discriminant analysis to be important for
discriminating between the three segregated
genera. The resulting linear discriminant func-
tion classified 100% of the specimens correctly
when using the suite of the first two important
characters for canonical axis 1–2 based on 40,
37 and 3 observations for each segregated
genus. Because of these, the linear discriminant
function (Table 5) can be used for identifica-
tion of specimens of Cassia s.l. To use the
discriminant function in Table 5 for generic
identification, multiply each character score by
its coefficient in each column. Calculate the
total for each column, the column with the
highest total is a genus to which the specimen
belongs. Though this method of identification
is different from traditional keys, it can be used
in a complementary manner.
Qualitative characters
Table 6 summarizes eleven qualitative mor-
phological characters of Cassia s. str.,
Chamaecrista and Senna. The most useful
characters to separate these three genera are
the filament length and extent of filament
curvature. The filament length was classified
into three groups, i.e. very short, medium and
long. This character is supported by the result
from canonical discriminant analysis in that it is
an important character to discriminate the three
genera. So filament length together with the
extent of filament curvature was used to
construct a key to these three segregated genera.
Key to genera
The following is a simplified key to the
segregate genera of Cassia s.l. (data from 18
taxa) using both quantitative and qualitative
characters (Tables 2 and 6) based on the
output obtained from DELTA (Dallwitz et al.
1993).
1a. The longest filament recurved, more than
2.5 cm long ...................... 1. Cassia s. str.
1b. The longest filament straight, less than
1 cm long................................................ 2
2a. Ovary stalk more than 0.9 mm long, fruit
more than 9 cm long ...................2. Senna
2b. Ovary stalk less than 0.9 mm long, fruit
less than 9 cm long..........3. Chamaecrista
Table 3. Summary of canonical discriminant function of four clustering groupings of Cassia s.l.
Function Eigenvalue % of variance Cumulative % Canonical
correlation
1 46.31 60.19 60.19 0.99
2 18.35 23.85 84.04 0.97
3 12.28 15.96 100.00 0.96
Conclusion
Two techniques of morphometric analysis
were used to investigate the taxonomic status
of 18 taxa in the genus Cassia s.l. in
Thailand. We conclude that the results from
the numerical taxonomic study as well as the
Fig. 1
Canonical Axis 1
20100-10-20-30
Canonical Axis 2
20
10
0
-10
-20
Canonical Axis 1
20100-10-20-30
Canonical Axis 3
10
0
-10
-20
1
2
4
3
4
23
1
(a)
(b)
Fig. 1. Canonical discriminant analysis of 18 taxa of Cassia s.l., using four-cluster groupings as priori groups
(from cluster analysis). (a) - axes 1 and 2, (b)-axes1and3(1-Cassia s. str. and Senna spectabilis,2-
Chamaecrista,3-Senna,4-Senna alata)
comparison of qualitative characters of Cas-
sia s.l. provide justification for recognition of
the segregation of the three genera, namely
Cassia s. str., Chameacrista and Senna from
Cassia s.l. This result, although based on
material from Thailand only, is in agreement
with Irwin and Barneby (1981, 1982) who
used characteristic of filaments and the
presence or absence of bracteoles as diag-
nostic characters. From the overall result of
the analyses it can be seen that Senna is
rather a heterogeneous genus, whereas Cassia
Fig. 2. Canonical discriminant analysis of 18 taxa, using 18 taxa as priori groups (a)-axes1and2,(b)-axes1
and 3 (1–4 Cassia s. str., 5–6 Chamaecrista,7–18Senna), see Table 1 for species
s. str. and Chamaecrista are rather homoge-
neous genera. Among the 12 Senna species,
only S. timoriensis and S. tora are indigenous
to Thailand and the Malesiana region (Hou
et al. 1996), while the rest are American
species (Irwin and Barneby 1982). S. alata
and S. spectabilis are rather distinct based on
the results of cluster analysis and canonical
discriminant analyses. The incongruity be-
tween the 12 Senna species was probably
due, at least in part, to their placement in
different taxonomic sections. Irwin and Bar-
neby (1982) placed S. alata and S. spectabilis
under section Senna and section Peiranisia of
the American Cassiinae, respectively, whereas
most American species used in this study
were placed under sections Chamaefistula and
Psilorhegma.
Table 4. Means and standard deviation of 32 quantitative characters of the three segregated genera from
Cassia s.l.
Character Taxon
Cassia s. str. Chamaecrista Senna
mean ±SD Mean ±SD mean ±SD
PET 36.2631 18.2345 3.8525 0.7221 37.1629 11.9930
PED 2.2440 0.5679 0.4868 8.383E-02 1.9243 1.2612
RCL 237.5676 49.4325 41.5023 13.5656 159.2748 119.3745
RCD 1.1034 0.4059 0.1840 5.222E-02 1.0318 0.3974
DBLP 30.1230 18.6362 1.6910 0.5082 20.3514 9.6249
NOL 10.8148 4.5388 21.7750 12.1032 7.8472 4.5691
TLL 82.5166 40.2344 5.4210 1.8974 68.3171 24.6552
TLW 34.2088 20.2352 1.7368 0.6736 29.0934 15.7664
BTWP 40.0712 13.4028 3.6878 1.4144 35.0998 13.6913
LWR 2.5715 0.4019 3.2635 0.6755 2.5919 0.7743
LS 0.5286 0.1224 0.6703 8.589E-02 0.5322 0.1280
LMW 170.7142 81.9850 10.7447 3.6797 143.7132 47.5668
POLL 3.3627 1.9539 .1743 4.618E-02 2.0839 0.9063
FLD 55.7750 22.0650 10.7597 2.9021 36.1681 9.2589
BTL 5.8210 1.9303 2.4795 0.7606 7.1250 6.6278
BTW 1.6795 1.0800 0.5400 0.3544 3.3782 4.3109
PCL 39.9533 18.1242 5.1718 2.2624 18.6303 8.7829
SPL 8.6944 2.0158 5.0085 1.5501 10.0673 2.2803
SPW 4.9784 1.3593 1.6555 0.4644 7.1254 1.8723
PTL 27.3270 10.0861 5.4537 1.4280 19.3028 5.0911
PTW 15.6959 4.9002 4.5355 1.3371 12.8928 3.2766
PSL 2.1002 0.5917 0.5918 0.1755 2.1851 0.9258
FML 38.0605 10.5138 1.5068 0.3458 4.4009 2.1751
FMD 0.7584 0.1626 0.2180 0.1166 0.6518 0.2752
ATL 3.9769 1.0043 1.8483 0.6917 6.5528 1.9388
ATD 2.4082 0.3427 0.6505 8.973E-02 2.0937 0.7001
OSL 10.4541 3.4854 0.5075 0.1441 2.1222 1.1345
OVL 26.8907 6.5333 3.1630 1.1575 11.5989 4.9568
OVD 0.9756 0.1450 0.8198 0.2149 0.9169 0.2398
STL 2.7469 1.0880 1.6400 0.4011 4.3666 2.0799
STD 0.6406 0.1262 0.3345 9.803E-02 0.4819 0.1666
FTL 435.9197 114.9881 40.8465 12.4728 147.0867 52.0012
The present findings reveal some heteroge-
neity within the genus Senna. The treatment of
Senna alata as a subgenus may not be justified,
because the plot of points in the canonical
ordination, observed on axis 2 (Fig. 1a) which
is 97% correlated with all the variables and
with variance explained of 24%, does not
significantly isolate it from the remainder of
the Senna species, and furthermore Senna as a
whole is scattered. Thus the taxonomic status
of the genus Senna could benefit from a
multivariate morphometric approach based
on worldwide material that is beyond the
scope of this study.
Fig. 3. Boxplots of the four most important characters of Cassia s.l. (1-Cassia s. str., 2-Senna,3-Chamaecrista),
unit in mm
Table 5. To identify a specimen to one of the three genera, multiply each character score by its coefficient
in each column; compute the total for each column; the column with the highest total is the genus to which
the specimen belongs. See Table 2 for character abbreviation
Character Cassia s. str. Chamaecrista Senna
PET 0.958 0.7 1.078
DBLP 0.767 0.919 5.45E-02
TLL )0.37 1.444 )0.501
BTWP )9.3 )7.634 )7.832
LWR 96.878 63.211 113.37
LS 672.146 610.881 592.206
LMW )3.87E-02 )0.372 )0.253
FLD 2.52 1.776 1.922
SPL 14.307 7.04 10.714
PTL )4.389 )2.162 )4.791
PTW )2.527 1.073 1.213
PSL )37.829 )21.386 )26.573
ATL )16.933 )6.445 )4.351
OVD 3.337 16.197 16.576
STL )9.582 )1.826 )6.434
LNOL 237.059 169.929 153.073
LTLW 852.139 515.852 836.349
LPOLL )47.06 )88.718 )51.383
LBTL 110.17 56.813 89.317
LBTW 1.011 )29.924 )6.354
LPCL 74.595 2.77 64.491
LFML 117.562 8.322 11.163
LOSL 23.99 )27.575 )21.085
LOVL 80.61 )4.122 50.745
LFTL 229.758 189.042 208.413
(Constant) )1386.45 )680.913 )1120.1
Table 6. Comparison of eleven qualitative morphological characters of the three segregated genera from
Cassia s.l.
Character Cassia s. str. Chamaecrista Senna
habit tree under-shrub shrub or tree
leaf upper-surface pubescent, glossy glossy hirsute, glossy
foliar gland absent present absent or present
inflorescence raceme; lateral or
axillary
few; axillary few, raceme; axillary
or terminal
petal color red, pink, yellow yellow yellow
staminode absent absent absent or present
filament medium to long;
recurved
very short; straight medium; straight
enlargement at middle
of filament
absent or present absent absent
style glabrous or pubescent glabrous glabrous
fruit terete or cylindrical;
glabrous or pubescent
flat; glabrous flat, terete, swollen, or
winged; glabruos or hirsute
We are grateful to anonymous reviewers for
their valuable comments on the manuscript. This
work was partially supported by a grant from
Chulalongkorn University, Bangkok, Thailand and
the TRF/ BIOTECH Special Program for Biodi-
versity Research and Training program (grant No.
T145008).
References
Anonymous (1998) SPSS for Windows Release
9.0., Standard version, SPSS Inc., Chicago.
Baum B. R., Bailey L. G. (1984) Taxonomic studies
in wall barley (Hordeum murinum sensu lato) and
sea barley (Hordeum marinum sensu lato). 2.
Multivariate morphometrics. Canad. J. Bot. 62:
2754–2764.
Bentham G. (1871) Revision of the genus Cassia
Linn. Trans. Linn. Soc. 27: 503–591.
Bruneau A., Forest F. (2001) Phylogenetic rela-
tionships in the Caesalpinioideae (Leguminosae)
as inferred from chloroplast trnL intron se-
quences. Syst. Bot. 26(3): 487–514.
Burkill I. H. (1935) A dictionary of the economic
products of the Malay Peninsula. The University
Press, Oxford.
Dallwitz M. J., Paine T. A., Zurcher E. J. (1993
onwards) User’s Guide to the DELTA System: a
General System for Processing Taxonomic
Descriptions. 4th edition, available on the inter-
net at http://biodiversity.uno.edu/delta/.
Ghareeb A., Khalifa S. F., Fawzi N. (1999)
Molecular systematics of some Cassia species.
Cytologia (Tokyo) 64(1): 11–16.
Holmgren P. K., Holmgren N. H. (2004) Index
Herbariorum (online at http://www.nybg.org.bs-
ci/ih/ (2003, September 30).
Hou D., Larsen K., Larsen S. S. (1996) Caesalpinia-
ceae. In: Kalkman C., Stevens P. F., Kirkup D. W.,
deWildeW.J.J.O.,NooteboomH.P.(eds.)Flora
Malesiana. Vol. 12, part 2. Rijksherbarium/ Hortus
Botanicus, Leiden.
Irwin H. S., Barneby R. C. (1981) Cassieae. In:
Polhill R. M., Raven P. H. (eds.) Advances in
Legume Systematics; Part 1: 97–106. Royal
Botanic Gardens, Kew.
Irwin H. S., Barneby R. C. (1982) The American
Cassiinae. A synoptic revision of Leguminosae
tribe Cassieae subtribe Cassiinae in the New
World. Mem. New York Bot. Gard. 35(1):
1–918.
Larsen K., Larsen S. S., Vidal J. E. (1984) Legumi-
nosae-Caesalpinioideae. In: Smitinand T., Larsen
K. (eds.) Flora of Thailand 4(1): 102–123.
Larsen K. (1993) Note on the nomenclature of
Cassieae (Leguminosae: Caesalpinioideae) in
Malaysia. Nordic J. Bot. 13(4): 403–404.
Linnaeus C. (1753) Species Plantarum. vol.1. The
Gresham Press, London.
Lock J. M. (1988) Cassia s.l. (Leguminosae-
Caesalpinioideae) in Africa. Kew Bull. 43(2):
333–342.
Mabberley D. J. (1997) The Plant-book. A portable
dictionary of the vascular plants. Cambridge
University Press.
Rohlf F. J. (2000) NTSYSpc, Numerical Taxon-
omy System, Version 2.1 Setauket, N.Y.: Exeter
Software.
Singh V. (2001) Monograph on Indian subtribe
Cassiinae (Caesalpiniaceae). Jodhpur, Scientific,
278 pp.
Sneath P. H. A., Sokal R. R. (1973) Numerical
taxonomy: The principles and practice of numer-
ical classification. W. H. Freeman and Com-
pany, San Francisco.
Tucker S. C. (1996) Trends in evolution of floral
ontogeny in Cassia sensu stricto, Senna, and
Chamaecrista (Leguminosae: Caesalpinioideae:
Cassieae: Cassiinae); A study in convergence.
Amer. J. Bot. 83 (6): 687–711.
Willis J. C. (1988) A dictionary of the flowering
plants and ferns. 8th Edition, revised by H.K.
Airy Shaw, Cambridge University Press, Cam-
bridge.
Addresses of the authors: Dr. Thaweesakdi
Boonkerd (e-mail: Thaweesakdi.B@Chula.ac.th)
and Mr. Sahanut Pechsri, Department of Botany,
Faculty of Science, Chulalongkorn University,
Bangkok 10330, Thailand. Dr. Bernard R. Baum
(e-mail: baumbr@agr.gc.ca) Agriculture & Agri-
Food Canada, Eastern Cereal and Oilseed
Research Centre, Neatby Building, 960 Carling
Avenue, Ottawa, Ontario, K1A 0C6, Canada.
