全 文 ：Received 20 Oct. 2003 Accepted 12 Dec. 2003
Supported by the National Natural Science Foundation of China (49971006, 39893360) and the Knowledge Innovation Program of The
Chinese Academy of Sciences (KSCX2-1-06B).
* Author for correspondence. E-mail: .
http://www.chineseplantscience.com
Ancestral Area Analysis of the Genus Caragara (Leguminosae)
ZHANG Ming-Li*
(Institute of Botany, The Chinese Academy of Sciences, Beijing 100093, China)
Abstract: Caragana has a temperate Asian distribution. Based on the phylogeny and 13 distributional
areas of this genus, its ancestral area was studied via the ancestral area analysis suggested by Bremer
(1992), Ronquist (1994) and Hausdorf (1997). The results indicate that three areas, Far East-Northeast
China, Altai-Sayan and North China-Qinling Mountains (Mts) are likely the ancestral areas. Linking to the
viewpoints of the Holarctic origin for north temperate flora, Far East-Northeast China seems more likely to
be the ancestral area. According to the three ancestral areas isolated geographically and the analysis in
the present study, as well as former biogeographical analysis of this genus, it is suggested that Caragana
speciation is mainly in the mode of vicariance rather than dispersal, and dispersed is often in short
distance.
Key words: Caragana ; ancestral area; distribution pattern; evolution
Caragana , belonging to the Papil ionoideae
(Leguminosae) and comprising about 70 species (Polhill,
1981), has a temperate Asian distribution (Wu,1993), mainly
located in the cold and drought area of northwestern
Qinghai-Xizang Plateau of China. The species of this ge-
nus have distinctive morphological variation and ecologi-
cal adaptation especially to drought and cold. Some spe-
cies occurring in grassland, desert or alpine mountain
meadow, could form dominant and constructed species in
shrubs, sometimes even solely occupied the whole com-
munity (Wu,1980). Komarov (1908; 1945) studied the East
Asian and Mongolian floristic geography by using five
genera Caragana , Clematoclethra, Codonopsis,
Epimedium and Nitraria among which Caragana appeared
as the key genus. After Komarov’s works, Caragana has
been treated as a hot spot by the plant taxonomists in China,
Russia, Mongolia and other countries (Pojarkova, 1945;
Moore, 1962; 1968; Sanchir,1979; 1980; Liu,1993; Zhao,1993).
In terms of the origin of the genus, Komarov (1908; 1945)
pointed out that the genus was originated from the south-
eastern East Asia, and C. sinica was the primitive species
with pinnate leaves containing two pairs of leaflets. From
the southeastern East Asia, this species evolved north-
ward and westward respectively into one group with pin-
nate leaves containing many pairs of leaflets and into an-
other group with palmate leaves. Komarov (1908; 1945) con-
sidered that the flora of Mongolia and Central Asia were
originated from East Asia. Moore (1968) discussed the ori-
gin of the genus and found C. sinica was a triploid species
(2n = 24) from chromosome data, suggesting that it could
not be a primitive species. Most species with pinnate leaves
containing many pairs of leaflets and foliage rachis decidu-
ous were diploid (2n = 16) and likely formed a primitive
group. Moore (1968) inferred that the southern Balkash Lake
was the place of origin in terms of the groups at different
evolutionary stages in his phylogenetic tree, and dispersed
eastward to the Asian Pacific shore and westward to the
southern Europe. However, both Sanchir (1979) and Zhao
(1993) regarded C. arborescens morphologically as a primi-
tive species, and its distributional areas, Siberia, as the place
of origin. Having summed up the species morphology, pol-
len morphology and chromosome data, I set up a phyloge-
netic tree using cladistics (Zhang, 1997a), and analyzed the
area relationship with analytical biogeographical methods
such as component analysis, minimum spanning tree, clus-
ter analysis and parsimonious analysis of endemicity
(Zhang, 1996; 1998). The results showed that C. arborescens
and its related group with pinnate leaves containing many
pairs of leaflets were primitive, and originated from the east-
ern Siberia. However, the analytical biogeographical meth-
ods usually focus on the area relationship. While the distri-
bution pattern, especially ancestral area, has not been di-
rectly studied. Therefore, in this study attention was paid to
locate the place of origin by using the ancestral area analysis.
Bremer (1992) proposed a method to estimate the center
of origin or ancestral area. Ronquist (1994; 1995) critiqued
Bremer’s method based on the Camin-Sokal’s irreversible
parsimony, and suggested his method on the basis of Fitch’s
Acta Botanica Sinica
植 物 学 报 2004, 46 (3): 253-258
Acta Botanica Sinica 植物学报 Vol.46 No.3 2004254
reversible parsimony. Bremer (1995) gave an explanation of
his method. After combining both Bremer’s (1992) and
Ronquist’s (1994) methods, Hausdorf (1997) proposed a
weighted ancestral area analysis. So far, there are three
methods for quantitative estimation of the ancestral area.
In the present investigation, the ancestral area and distri-
bution pattern of Caragana were analyzed by using a com-
bination of all three methods.
1 Materials and Methods
1.1 Phylogenetic tree
I have constructed a phylogenetic tree or cladogram of
72 species of the genus by using 24 characters comprising
20 morphological, three pollen and one chromosomal char-
acters (Zhang,1997a). The twenty morphological charac-
ters are all good characters in the genus. Based on this
cladogram, two reduced cladograms are obtained by re-
ducing some species, just like our previous component
analysis (Zhang, 1998) (Figs.1,2).
1.2 Areas
The distribution region of Caragana was divided into
13 areas (Zhang,1998) according not only to the floristic
Fig.1. A reduced cladogram from Zhang (1997a) including 12
areas. A, Far East-Northeast China; B, Altai-Sayan; C, Mongolia
Plateau; D, North China-Qinling Mts; E, Hengduan Mts; F,
Himalayas; G, Balut-Afghanica; H, Central Asia; I, Tianshan Mts;
J, Pamir-Alai; K, Europe-West Sibirica; L, Turan; M, East Asia
Subtropical.
Fig.2. A reduced cladogram from Zhang (1997a) including 13
areas. A-M are the same as in Fig.1.
regionalization of China (Wu and Wu, 1996) and that of the
world (Takhtajan, 1986) (Fig.3), but also to the distribu-
tional characteristics of Caragana, for instance, C. sinica
appeared in both the eastern and southern China. These
two regions (Wu and Wu,1996) were treated as an area,
namely Subtropical East Asia area. Thirteen areas could be
regarded as endemic area to Caragana because there are
endemic species in each area, and there are obvious differ-
entiation among the areas in terms of the flora (Wu and Wu,
1986) and vegetation (Wu,1980).
1.3 Ancestral area analysis
It is necessary to label the species distribution area into
the two reduced cladograms (Figs.1, 2). And then the indi-
ces of G, L and G/L of Bremer’s (1992) method, of S and RP
of Rounquist’s (1994) method, of GSW, LSW and PI of
Hausdorf’s (1997) method were calculated. In principles,
Bremer’s method and Hausdorf’s method follow the crite-
ria that the areas are positionally plesiomorphic in the area
cladogram being more likely as parts of ancestral area than
the positionally apomorphic branches. Ronquist’s method
estimates the area as ancestral area by the number of nec-
essary steps under the assumption that this area was the
ancestral area. The most probable ancestral area is with the
minimum number of the steps.
255ZHANG Ming-Li: Ancestral Area Analysis of the Genus Caragara (Leguminosae)
Fig.3. Caragana distribution and 13 areas (A-M), of which A, B and D were the ancestral
areas in this paper. These three areas are just Caragana arborescens distribution. A-M are the
same as in Fig.1.
In Bremer’s, Ronquist’s and Hausdorf’s methods, G
(gain) means number of necessary gains forward Camin-
Sokal parsimony; L (loss) means number of necessary
losses under reverse Camin-Sokal parsimony; S means num-
ber of necessary steps if the area was the ancestral area;
RP means S values rescaled to a maximum value as the
reciprocal of the S values multiplied by the smallest S value;
GSW means number of weighted gain steps; LSW means
number of weighted loss steps; PI=GSW/LSW.
If the G/L, RP and PI values of the area are larger, the
area will more likely be the ancestral area.
2 Results and Discussion
2.1 Ancestral area
The calculated results for Caragana are shown in Table
1 and Table 2 corresponding to Figs.1, 2, respectively. Three
areas, Far East-Northeast China, Altai-Sayan, and North
China-Qinling Mts, were shown to be the ancestral areas
as they have the highest G/L, RP and PI values. Addition-
ally Balut-Afghanica has also large values of G/L and PI,
G/L=0.50, PI=0.36. In Table 2, both G/L and PI have in
common the first five largest values, which correspond to
five areas, i.e., Far East-Northeast China, Altai-Sayan, North
China-Qinling Mts, Hengduan Mts, and Balut-Afghanica.
Therefore, these areas are indicated as ancestral areas.
However, in Table 2 the larger values of RP were found
only in North China-Qinling Mts, Hengduan Mts, and Balut-
Afghanica indicating that these areas were ancestral areas.
This result is different from that
coming from G/L and PI. And in
Table 2, North China-Qinling Mts
has the largest values of G/L, RP
and PI obtained from the three
methods. In Table 1, RP and PI
values indicate that North China-
Qinling Mts has the highest
values, but the highest G/L val-
ues were obtained in Far East-
Northeast China and Altai-Sayan.
Which area is the most likely
ancestral area among Far East-
Northeast China, Altai-Sayan,
and North China-Qinling Mts?
Due to some primitive species of
Caragana occurring in Boreal
Siberia, the origin of boreal tem-
perate flora was located in the
eastern and middle Siberia
(Budantsev,1992). So, the balanced conclusion to the place
of origin between Caragana and boreal temperate flora
would be inferred as the eastern Siberia, namely Far East-
Northeast China. Previous results of analytical
biogeography, Zhang (1998) also showed that Far East-
Northeast China was the ancestral area of Caragana.
The 13 areas were divided into two clusters, East Asia
and Tethys, by using some analytical biogeographical
methods, such as component analysis, minimum spanning
tree (MST), parsimony analysis of endemicity (PAE) and
cluster analysis (Zhang,1996; 1998). East Asia cluster is
composed of Far East-Northeast China, Mongolia Plateau,
North China-Qinling Mts, Hengduan Mts, Himalayas and
East Asia Subtropical. Other areas belong to Tethys cluster.
Why Mongolia Plateau was included in the East Asia
cluster? Since Mongolia Plateau especially Nei Mongol and
Northeast China-North China are geographically linked, and
both share same species, such as C. microphylla, therefore,
it would be reasonable, referring to the Caragana distribu-
tion pattern and in the present cladogram, to put Mongolia
Plateau in the East Asia cluster.
Because some areas with larger values belong to either
East Asia or Tethys, which of the two clusters is primitive
in the Caragana distribution pattern? For ancestral com-
parison here the average values of G/L and PI of the two
area clusters are calculated (Tables 1, 2).
All the above values indicate that East Asia is more
likely primitive, its related G/L and PI values are always
larger than those of Tethys. Consequently Kamarov’s view
Acta Botanica Sinica 植物学报 Vol.46 No.3 2004256
Table 1 An ancestral area analysis of Caragana based on Fig.1
G L G/L S RP GSW LSW PI
A 1 1 1.00 14 0.93 1.00 1.00 1.00
B 1 1 1.00 14 0.93 1.00 1.00 1.00
C 1 4 0.25 15 0.87 0.50 2.50 0.20
D 4 5 0.80 13 1.00 2.03 1.62 1.26
E 1 5 0.20 15 0.87 0.33 2.83 0.12
G 2 4 0.50 15 0.87 0.83 2.33 0.36
H 1 8 0.13 15 0.87 0.17 3.95 0.04
I 1 8 0.13 15 0.87 0.17 3.95 0.04
J 1 4 0.25 15 0.87 0.50 2.50 0.20
K 1 7 0.14 15 0.87 0.20 3.78 0.05
L 1 8 0.13 15 0.87 0.17 3.95 0.04
M 1 6 0.17 15 0.87 0.25 3.08 0.08
G (gain), number of necessary gains forward Camin-Sokal parsimony; L (loss), number of necessary losses under reverse Camin-Sokal
parsimony; S, number of neccessary steps if the area was the ancestral area; RP, S values rescaled to a maximum value of 1 by inverting them
and multiplying by the smallest S value; GSW, number of weighted gain steps; LSW, number of weighted loss steps; and PI = GSW/LSW (Bremer,
1999; Ronquist, 1994; Habisdorfs, 1997). East Asia average G/L = (1.00+0.25+0.80+0.20+0.17)/5 = 0.484; Tethys average G/
L = (1.00+0.50+0.13+0.13+0.25+0.14+0.13)/7 = 0.326; East Asia average PI = (1.00+0.2+1.26+0.12+0.08)/5 = 0.532; Tethys average PI
= (1.00+0.36+0.04+0.04+0.20+0.05+0.04)/7 = 0.247. A-M are the same as in Fig.1.
Table 2 An ancestral area analysis of Caragana based on Fig. 2
G L G/L S RP GSW LSW PI
A 1 2 0.50 19 0.84 1.00 2.00 0.50
B 1 2 0.50 19 0.84 1.00 2.00 0.50
C 1 3 0.33 19 0.84 0.50 2.50 0.20
D 4 7 0.57 16 1.00 1.98 1.93 1.03
E 3 7 0.43 18 0.89 0.75 3.03 0.25
F 1 5 0.20 20 0.80 0.25 3.58 0.07
G 2 4 0.50 18 0.89 0.83 2.33 0.36
H 2 9 0.22 19 0.84 0.29 3.71 0.08
I 1 9 0.11 20 0.80 0.13 3.72 0.03
J 1 3 0.33 19 0.84 0.50 2.50 0.20
K 1 8 0.13 20 0.80 0.15 3.59 0.04
L 1 9 0.11 20 0.80 0.13 3.72 0.03
M 1 7 0.14 19 0.84 0.17 3.45 0.05
East Asia average G/L = (0.50+0.33+0.57+0.43+0.20+0.14)/6 = 0.362; Tethys average G/L = (0.50+0.50+0.22+0.11+0.33+0.13+0.11)/7 =
0.271; East Asia average PI = (0.50+0.20+1.03+0.25+0.07+0.05)/6 = 0.35; Tethys average PI = (0.50+0.36+0.08+0.03+0.20+0.04+0.03)/
7 = 0.177. Abbreviations are the same as in Table 1.
point (Kamarov, 1908; 1945), that Caragana was origi-
nated from East Asia and then dispersed into Mongolia
and Central Asia, is reasonable.
2.2 Distribution pattern
The ancestral areas of Caragana are likely to be Far
East-Northeast China, Altai-Sayan, and North China-Qinling
Mts according to our results from the ancestral area
analysis. These three areas are just the distribution region
of Caragana arborescens which is the most primitive spe-
cies in Caragana (Zhang,1997a) (Fig.3). Because these ar-
eas are far isolated geographically, the speciation and di-
versification mode in Caragana could be explained on the
basis of vicariance. Dispersion can only be considered as
subordinate and is in short distance (Zhang,1997b; 1998),
for instance, the distribution pattern of Qinghai-Xizang
(Tibetan) Plateau and Himalayas (Zhang,1997b).
Concerning the time of origin, the Bering Land Bridge
connected the East Asia and North America, broke-up in
Pliocene (Hamilton,1983; McKenna,1983). If the time of
origin of Caragana was before the Pliocene, it should ap-
pear in North America via the Bering Lund Bridge before
Pliocene where the two regions were connected. However,
Caragana species are naturally absent in North America,
so, according to its restricted temperate Asian distribution
pattern and its absence in North America, the time of origin
and differentiation of Caragana seem quite late, and could
be inferred to Pliocene or latter.
2.3 Method of ancestral area analysis
Some methods of analytical biogeography, such as com-
ponent analysis, minimum spanning tree, parsimonious
257ZHANG Ming-Li: Ancestral Area Analysis of the Genus Caragara (Leguminosae)
analysis of endemicity and cluster analysis (Myers and
Giller,1988), focus on the area relationship rather than on
the identification and originiation of the ancestral area, ex-
cept for component analysis. As ancestral area analysis is
emphasized on estimation of ancestral areas (Bremer,1992;
1995), it provides some numerical indices, therefore, it is a
suitable method.
According to the three method of ancestral area analy-
sis as well as the results of combined use of these methods
in the present study, Hausdorf’s (1997) method as derived
from Bremer’s (1992) gives an ancestral area weighted esti-
mation in which PI is more precise than Bremer’s (1992) G/
L to ancestral area. Moreover, Bremer’s (1992) and
Hausdorf’s (1997) methods are different from Ronquist’s
(1994) method (Table 2). Compared with these three methods,
the Nothofagus analysis resulting from Swenson et al. (2000)
had also shown the same conclusion.
Acknowledgments: I deeply appreciate to K Bremer for
his help during my visit in Uppsala University. Many thanks
are also given to F. Ronquist and J. Wen for his/her help, to
three anonymous reviewers for their valuable comments.
References:
Bremer K.1992. Ancestral areas: a cladistic reinterpretation of the
center of origin concept. Syst Biol, 41:436-445.
Bremer K.1995. Ancestral areas: optimization and probability.
Syst Biol, 44:255-259.
Budantsev L Y. 1992. Early stages of formation and dispersal of
the temperate flora in the boreal region. Bot Rev, 58:1-48.
Hamilton W. 1983. Cretaceous and Cenozoic history of the north-
ern continents. Ann Miss Bot Gard, 70:440-458.
Hausdorf B. 1997. Weighted ancestral area analysis and a solution
of the redundant distribution problem. Syst Biol, 47:445-456.
Komarov V L. 1908. Generis Caragana monographia. Acta Horti
Petrop, 29:77-388.
Komarov V L. 1945. V.L. Komarov Opera Selecta. Mosquae:
Acad Sci Press, USSR. 159-342.
Liu Y-X . 1993. Caragana. Flora Reipublicae Sinicae. Tomus 42
(1). Beijing: Science Press. 17-67. (in Chinese)
McKenna M C. 1983. Holarctic landmass rearrangement, cosmic
events, and cenozoic terrestrial organisms. Ann Miss Bot Gard,
70:459-489.
Moore R J. 1962. On the origin Caragana sinica. J Arnold
Arboretum, 43:203-214.
Moore R J. 1968. Chromosome numbers and phylogeny in
Caragana (Leguminosae). Can J Bot, 46:1513-1522.
Myers A A, Giller P S. 1988. Analytical Biogeography. London:
Academic Press.
Pojarkova A I. 1945. Caragana. Komarov V L. Flora of USSR,
Vol. Ⅱ. Moscow and Leningrad: Acad Sci Press, USRR. 327-
368.
Polhill R M. 1981. Galegeae. Polhill R M, Raven P H. Advances
in Legume Systematics, Part 1. London: Royal Botanical
Garden, Kew. 357-363.
Ronquist F. 1994. Ancestral areas and parsimony. Syst Biol, 43:
267-274.
Ronquist F. 1995. Ancestral areas and revisited. Syst Biol, 44:
572-575.
Sanchir Cz. 1979. Genus Caragana Lam. (Systematics, geography,
phylogeny and economic significance in study on flora and
vegetation of P.R. Mongolia. Vol.1. UlanBotaor: Academic
Press.
Sanchir Cz. 1980. Outline of Caragana Lam. Species. Trans Inst
Bot Acad P. R. Mongolia, 4:106-123.
Swenson U, Hill R R, McLoughlin S. 2000. Ancestral area analy-
sis of Nothofagus (Nothofogaceae) and its congruence with
the fossil record. Aust Syst Bot, 13:469-478.
Takhtajan A. 1986. Floristic regions of the world. Berkeley:Uni-
versity of California Press.
Wu C-Y. 1993. The areal-types of Chinese genera of seed plants.
Acta Bot Yunnan , 5(Suppl.):1-139. (in Chinese with English
abstract)
Wu Z-Y. 1980. China Vegetation. Beijing: Science Press. (in
Chinese)
Wu Z Y, Wu S G. 1996. A proposal for a new floristic kingdom
(realm)— the E. Asiantic Kingdom,its delineation and
characteristics. Zhang A L, Wu S G. Floristic Characteristics
and Diversity of East Asian Plants. Beijing and Berlin: China
Higher Education Press and Springer-Verlag. 3-42.
Zhang M-L. 1996. A numerical analysis of endemism for Caragana
(Fabaceae). Zhang A L, Wu S G. Floristic Characteristics and
Diversity of East Asian Plants. Beijing and Berlin:China Higher
Education Press and Springer-Verlag. 223-227.
Zhang M-L . 1997a. A reconstructing phylogeny in Caragana
(Fabaceae). Acta Bot Yunnan ,19:331-341. (in Chinese with
English abstract)
Zhang M-L . 1997b. The geographical distribution of the genus
Caragana in Qinghai-Xizang Plateau and Himalayas. Acta
Phytotax Sin, 35:136-147. (in Chinese with English abstract)
Zhang M-L . 1998. A preliminery analytical biogeography in
Caragana (Fabaceae). Acta Bot Yunnan , 21:1-11. (in Chinese
with English abstract)
Zhao Y-Z. 1993. Taxonomic study of the genus Caragana from
China. Acta Sci Nat Univ Intramongolicae , 24:631-653. (in
Chinese with English abstract)
(Managing editor: HAN Ya-Qin)