Chloroplast-DNA Phylogenetics and Biogeography in a Reticulating Group: 
Study in Poa (Poaceae) STOR ? 
Robert J. Soreng 
American Journal of Botany, Vol. 77, No. 11. (Nov., 1990), pp. 1383-1400. 
Stable URL: 
http://links.jstor.org/sici?sici=0002-9122%28199011%2977%3All%3C1383%3ACPABIA%3E2.0.CO%3B2-Y 
American Journal of Botany is currently published by Botanical Society of America. 
Your use of the JSTOR archive indicates your acceptance of JSTOR' s Terms and Conditions of Use, available at 
http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you 
have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and 
you may use content in the JSTOR archive only for your personal, non-commercial use. 
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at 
http://www.jstor.org/journals/botsam.html. 
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or 
printed page of such transmission. 
JSTOR is an independent not-for-profit organization dedicated to creating and preserving a digital archive of 
scholarly journals. For more information regarding JSTOR, please contact support@jstor.org. 
http: //www .j stor. org/ 
ThuOct 19 15:41:05 2006 
Amer. J. Bot. 77(11): 1383-1400.   1990. 
CHLOROPLAST-DNA PHYLOGENETICS AND BIOGEOGRAPHY IN 
A RETICULATING GROUP: STUDY IN POA (POACEAE)1 
ROBERT J. SORENG 
L. H. Bailey Hortorium, Cornell University, Ithaca, New York 14853 
Cladistic analysis of Poa chloroplast DNA (cpDNA) restriction sites tested previously hy- 
pothesized relationships within the genus. Forty-six taxa representing 19 sections or groups and 
three subgenera of Poa and two out-group genera, Puccinellia and Bellardiochloa, are analyzed. 
Five major and several minor cpDNA groups are identified. The cpDNA cladogram is generally 
congruent with the subgeneric taxonomy of Poa. Exceptions are reclassified or are discussed in 
terms of character incompatibilities and possible reticulation events. The cpDNA tree detected 
relationships among sections that were unresolved using traditional character sets and provides 
a basis for polarization of morphological character states. An assessment of biogeographic events 
based on the cpDNA tree suggests: 1) Poa originated in Eurasia; 2) at least six groups of species 
independently colonized North America; and 3) two of the latter groups colonized South Amer- 
ica, and one closely related group colonized New Zealand and Australia. The cpDNA tree 
provided a conservative estimate of the number of amphi-neotropical disjunctions when com- 
pared to the known number of species disjunctions. 
Species of the genus Poa are native to arctic, 
temperate, and high-elevation-tropical regions 
around the world. The genus is diverse, in- 
cluding some 500 species that occur in a wide 
range of habitats. Although taxonomically dif- 
ficult, and historically a catch-all, the genus as 
delimited today is considered by most agros- 
tologists to be morphologically coherent. Clay- 
ton and Renvoize (1986, p. 101) in their re- 
vision of all grass genera concluded that "Poa 
is an extremely uniform genus for which there 
is no satisfactory infrageneric classification. Its 
taxonomy is rendered difficult by the dearth of 
useful discriminatory characters and compli- 
cated by the widespread occurrence of apomixy 
and introgression." Bor (1952, pp. 7, 8) pref- 
aced a major work on Poa with, "The system- 
atic treatment of the species ... is one of the 
most bewildering and difficult of taxonomic 
studies. While many species are clear-cut and 
can be recognized at a glance, there are groups 
1
 Received for publication 29 December 1989; revision 
accepted 25 May 1990. 
The author thanks Khidir Hilu, VPISU, in whose lab 
many of the cpDNA isolations, restriction digests, and 
southern transfers were performed; Michael Clegg, U.C. 
Riverside, and Jeff and Jane Doyle, Cornell, in whose labs 
probings were done; Jerrold I Davis, Cornell, for assistance 
with lab work and data analysis; James Rawson for per- 
mission to use the Pennisetum probes, which were pro- 
vided by Michael Clegg; Mike Frohlich, Jerrold I Davis, 
and two anonymous reviewers for comments on the manu- 
script; and many individuals and companies named in 
Table 1 who generously provided plant material, especially 
the USDA Plant Introduction Station. This study was prin- 
cipally supported by NSF grants BSR 8700204 to RJS and 
BSR 8696101 to Jerrold I Davis. 
of species about which one can only conclude 
that their evolutionary history has been so 
complex that they do not lend themselves to 
systematic treatment by present taxonomic 
methods. One cannot rely upon a single char- 
acter to separate species in such groups, but 
combinations of more or less variable char- 
acters must be used.. . ." Stebbins (1950, p. 
405) stated that, "when this genus is better 
known, it may have to be regarded as a single 
huge polyploid complex, which is in part purely 
sexual, in part facultatively apomietic, and 
which contains in addition obligate apomicts." 
Such assertions leave considerable doubt as to 
whether any satisfactory, much less phyloge- 
netic, classification can be achieved. 
Although subgeneric treatments and species 
alignments in Poa have converged (Edmond- 
son, 1978, 1980 [Europe]; Nicora, 1978 [South 
America]; Tzvelev, 1983 [USSR]; Soreng, 1985 
[North America]), higher level reltionships are 
rarely based on explicit hypotheses of character 
evolution. I have found that many morpho- 
logically distinguishable groups of Poa are also 
coherent in their ecological tolerances and 
breeding system characteristics. However, 
character states used to unite groups or infer 
cladistic relationships mostly are continuous 
rather than discrete, and often are inconstant. 
The main objective of this paper is to com- 
pare a cladistic analysis of chloroplast DNA 
(cpDNA) restriction sites (RS) with current 
subgeneric classification. A second objective is 
to use cpDNA RS as markers of geographic 
radiation. 
As an independent test of phylogeny where 
1383 
1384 AMERICAN JOURNAL OF BOTANY [Vol. 77 
relationships may be reticulate, analysis of 
cpDNA RS offers advantages (Palmer, 1987). 
The slow evolution of chloroplast genomes 
provides phylogenetic resolution at high tax- 
onomic ranks independent of previously hy- 
pothesized character transitions (except for 
out-group choice). Uniparental (predominantly 
maternal) inheritance and absence of inter- 
molecular recombination disallow reticulation 
in cpDNA phylogenies, as opposed to mor- 
phological, plant mitochondrial-DNA, or nu- 
clear gene phylogenies. The possible influence 
of hybridization in the species phylogeny can 
be evaluated from congruence between cp- 
DNA cladograms and traditional species 
groups, and by subsequent experimentation 
(e.g., by isozyme analysis). The present study 
provides an evaluation of congruence with cur- 
rent classification and morphological data. 
A recent phylogenetic analysis of cpDNA RS 
in the subfamily Pooideae (including, among 
other genera; Arctagrostis, Bellardiochloa, Bri- 
za, Catabrosa, Chascolytrum, Dactylis, Fes- 
tuca, Microbriza, Puccinellia, Sclerochloa, Ses- 
leria, and Torreyochlod) suggested Poa is 
monophyletic (Soreng, Davis, and Doyle, in 
press). In this study, infrageneric relationships 
of Poa are assessed using Bellardiochloa and 
Puccinellia as out-groups: Traditionally recog- 
nized taxa were mostly supported, although 
some are in need of reevaluation in light of 
new cpDNA data. Several instances of putative 
reticulate evolution of chloroplast genomes are 
revealed. The cpDNA cladogram was consis- 
tent with a biogebgraphic scenario involving 
the origin and primary diversification of Poa 
in Eurasia; spread of at least six groups to North 
America, two of which occur in South Amer- 
ica. It also supports a close relationship be- 
tween the major New World group and a group 
occurring in Australia and New Zealand. 
MATERIALS AND METHODS 
i 
The sample?The 46 taxa examined are list- 
ed in Table 1, along with phromosome num- 
bers, mode of reproduction, and classification. 
Sampling was designed to include sexually re- 
producing species from as many diploid taxa 
as possible, as many sections as available, and 
duplicates within sections or groups if possible. 
Species were also selected to test hypotheses 
of the origin1 of dioecy andgeographic con- 
nections of New World Poa. Several known 
apomictic species' were included in order to 
learn more about their origins. In four cases 
cpDNA was in short supply and different ac- 
cessions were pooled (Poa cusickii 2977, 2993, 
2994; P. fendleriana subsp. fendleriana; P. f. 
subsp. albescens; P. labillardierei and P. sie- 
beriana [two closely related Australian spe- 
cies]). In three cases RS data from different 
accessions of the same species were pooled (af- 
ter checking for synapomorphies) to make 
complete sets (P. arctica, P. cusickii, P. pra- 
tensis). All other terminal taxa represent a sin- 
gle individual or population or adjacent local 
populations. Multiple accessions were screened 
for intraspecific variability of RS in P. arctica, 
P. cusickii, P. fendleriana, and P. pratensis. 
Seven samples of P. secunda sensu lato (s.l.) 
were checked. 
I describe the sampling of species in some 
detail as it is critical to the interpretation of 
the results. Diploids are included from four of 
11 sections in which they are known. Among 
the polyploids, Poa annua (Ochlopoa) is an 
infrasectional allopolyploid (Tutin, 1957). 
Morphological comparisons suggest that the 
apomicts, Poa alpina, P. arctica, P. chamae- 
clinos, P. palustris, and P. nemoralis are also 
derived from within their respective sections. 
Other apomicts sampled are likely to involve 
wider (intersectional) parentage; P. bulbosa, P. 
compressa, P. pratensis, P. wheeleri (Soreng, 
1986), and P. secunda s.l. Samples of the re- 
maining species were from putatively sexual 
populations, except in P. fendleriana subsp. 
fendleriana, where one sexual and two apo- 
mictic populations were pooled (Soreng, 1986). 
The breadth of taxonomic variation sampled 
in the genus is of concern. In this regard, all 
but two monotypic (Leptophyllae, Nanopoa) 
and one small section (Nivicolae) of Poa cur- 
rently recognized in Europe and the USSR were 
sampled. In North America I have sampled all 
of the major and most of the minor groups of 
Poa. From South America I have sampled two 
major groups and two minor groups (the latter 
sampled from Mexico), leaving an antarctic 
island group (P. flabellata (Lam.) Raspail, P. 
cqokii, P. rammossisima Hook, f.) and the small 
subgenus Andinae s.s. (which is questionably 
distinct from group IVB species) unsampled 
there. A regional gap exists in Southeast Asia 
both in the sample and in our knowledge of 
sectional affinities of the species. However, I 
believe that many if not most species of that 
region can be accommodated within sections 
found in the USSR. My estimation, derived 
frorn a survey of the Gray and U.S. National 
Herbarium collections and the literature, is that 
species have been sampled from three-quarters 
of the sectional level groups in the genus. 
Two outgroups were used. Puccinellia dis- 
tant, a polyploid, has restrictipn patterns that 
are nearly identical to those of other species of 
that genus, including diploids (Davis and So- 
November 1990] SORENG?POA CHLOROPLAST DNA 1385 
! 
f 
s 
I 
a 
o 
?g 
s 
a 
o 
0* 
g 
<3 
S 
e 
?"?* 
<3 
?a I 
o 
is 
?S 
a o a 
I 
I 
?si I &j 
-1-7 O    tfl 
?O W UJ   3 
?s la  . >H 
> (N w CO 
03 
? to 
ac 
O v 
Si 
p 
1 S g a-a 
S"'1^ 
<*j "P    .,   -   ? ei 
? a "^ "i "5    <*2 
a u D s ^ .g \D 
3 <N <N <N ?'   "9 "1 UJ ?i      <, ?' 
C3 
S 
<U 5, 
SI- 
'S a ta a d ? 13   Ci 
si 
_? <: ?4 SI 
!* 
55 ^ m o 
p 
B 
t- as o 
3 2 
o so "n 3 
vs rt O O c> m <N 
m ro CO OH 
t- 
rf <N E oo 
o ?i y o\ 
?c -c OH <N 
SO 
?<* 
un 
(N 
en 
. ?*    . 
CI jo IS 
& d i & O o 
+* V | s 
?a s P 
?? P 
.s 
Mi 23     <8 
P 
a = so. 
{ O O 
iilL 
0 H 
6 3 
M 13 ? 
1?1 >> 
P. o a 3 (3 3 O P & P e P &> +H to 
ca 1 C3 "3. a o p 
|H UH 
P P 
a 0 3 
to 3 3 s s s 
p p p 
& a C3 & 
<N 
^ 
00 rf rN 00 
<N -* <N 
Os OO 
in os 
SOSOM 
vs in rf 
o  o  o O   0)   o 
.? .? .2 
'O 'o 'O 
s?? 
rs] 00 00 
3 
ca 
?a 
JO 
p 
rlJ.D0 
00 , 
? I 
;?! 
"??3 
?5 a, 
I- 
> o 
.2? 
H3 03 
e 
P 
w 
?? 
?^ 
-So 
=3     .   a 
? N ^ 05     a. 
s3 g _l/3 
B<rN ? S T ^ u -i ? 
s *?' s; 
g &s 
O C 
os ,^r5 
at < 
?Il 111 
?2 s  ?H 3 
?  -a 
? 1 
?srii 
P ^ s 
S ? s 
?ss ? ? 
?-5 S 
| 
a 
i 
is 
(SO 
3 
P 
| 
5 J <U 
??| 
ft, ft, 
<3   ? 
S s 
-2 s 
03 
Pi 
.i-g 
?a -a b u 
S " 
Si 
a 
> i to    tn0>    O 
u3 
a a co, 
S S ? 
I Ss-c 
O^        (?c^^^^^^ 
Is 
? a,a, a. 
SO 
1386 AMERICAN JOURNAL OF BOTANY [Vol. 77 
5,1/3 t/1 t*) ,? </5 
M,i>-H-Hrt(N~-        .   CO   -H 
CO 
CO   CO 
00 so 
CO 00 
CN CO 
00 so 
CN CN 
</-> 
J3 O co ?* 
rt CO  CO 
CN CO CO 
t- 
os 
CN 
*     *     4 
<N O ?* ?* ^ 
fO M IN M M 
00  ^H lO 
B U j> > 
o- o- i B B % a a a a 
1 o o o o d d 1 a o ess o o o 
?t-i ?is .a +-? o u u u 
so 
.3 <+H    <*H rt ui C*-l   C(H   *4H a 13 13 35    CO 0. 13 U <D U ?   ai   m P s 
o a 
+ + 
on 
cs 
3 
1 
o 
a 
o 
"9 
3 
?S 
s 
a 
M T3 13 
as g 
::   <u <3 
8 S.S 
a x> 
O 
u- CN 
>.       r 
a     i.: ?  ?? 
?ill 
a la 
ill 
Q  G  g 
*s ^S c ll 
csl .en a .? 
's c <u o ,s cj; 
-o o o c * s 
a ^ * 
?f 1 .? ? 
u c s*. 
5 s 
CN 
111 
ESI 
?a- 
OS 
OS 
CN 
os r- 
CT\ OS 
CN CN 
OS   -H 
O 00 
CO -H 
CN CN 
O oo >o 
CO rt OS 
CN CN CN 
o" O" -H        so 
oo r- CN      <n 
hna        CO 
^H CN CN        CO 
j? & 
tu V o o 
??a ^ ji Xl 
a 5 
?! u u _   (U   u 
.? .? .? 
'o *5 *U 
Q 
o 
S3, 
as 
3 i CN 00 00 vo 00   ? so 
?* CN CN w> CN   S ""> 
s& ?   CO CD   CD ... 
g S9 > > 3 
13 CN 
o 
4 
s 
o 
?a 
o I 
B 
1 &JH ? i 
S S 8 a 
'S R F 
iJ 8=3:5 
.    2?   M 
O      ^ CN co - 
CN        CN CN CN I 
r-^      oo      ( 
CN        CN        i 
?? ??   lo 
^   to 
X) 
OS       O 
CN CO 
<3 i 
?till 
o 
s 
i 
<3 
O    M    tij ill 
si ? 
1^? 
November 1990] SORENG?POA CHLOROPLAST DNA 1387 
?S 
s 
a 
3 
'?3o 
II 
111 111 
?I1 
o <*) 
a 
| 
J3 
?3 
O 
| 
a 
o 
i 
0 
v) vi w &o t/3 &o w 
z? 3 
OH 
CD 
PH |>       in ^H <N ^ ^ <N 
m
 J2 I     IS I-41   &  -tn   *?!    Sit   P 2 a 
1 
1    1 
?a p ? 
SO* 
I- d^^l^^S00^^ 
1 
s 
3?|D ?0 
5 ?? ~ I 2
1.** 
e t: h a ^ PL, ir> 
CO 
< 
OH <N 
rn 
<N 
32 
00 ?> -. 
2<^ 
~H m ^H     T?1 CJ IN 
Od 
ri-* 
rn vo (N O I   u-> (N Cl (N ?* ~H o (N Cl o m CH ?* ~H c<i 
1 .B.a B B o o 
o 
o 
?'   O 
<-- ~H 
I   I (N <N 
y3 t? 
.3    .3- 
o 
a 
o 
rii 
I 
o 
2 0 
ft ft 
B B 
o o 
u u 
II 
o o 
u u 
Si la 
W5     CO CA    35 
.a 
e 
o 
?g-g 
s s 
o o 
a ft 
<J. 
4 <N 
t- ?* 
oo oo      ^ 
J a 
o 
E77.. 
si 0 
gjj-s 
S5SSg Bo a 
"? ^3 IS 3  3^ 111 is* 5 a !i 
(S ' 
? 13 ! 
13
    5 /-Si 
5S5 
?IIS ;|SI 
, o. 
<d5' 
1388 AMERICAN JOURNAL OF BOTANY [Vol. 77 
reng, unpublished data). Bellardiochloa vio- 
lacea is a diploid member of a small genus 
included within Poa by Clayton and Renvoize 
(1986),butnotbyEdmondson(1978),Tzvelev 
(1983), or Soreng, Davis, and Doyle (in press). 
DNA isolation and RS analysis?Purified 
cpDNA was isolated from species listed in Ta- 
ble 1 using minor modifications (Soreng, Da- 
vis, and Doyle, in press) of a nonaqueous ex- 
traction procedure (Dally and Secund, 1989) 
or modifications (Hilu, 1988) of a sucrose gra- 
dient method (Saltz and Beckman, 1981). The 
latter method usually yielded 5-10 jug of suf- 
ficiently pure cpDNA for restriction work per 
30-100 g of fresh tissue. Although some deg- 
radation of the DNA was apparent with both 
methods, higher (0.5-1 Mg/g) and more con- 
sistent yields were obtained by the nonaqueous 
extractions using only 2-15 g fresh weight of 
leaves. The latter procedure was particularly 
successful in liberating cpDNA from mature, 
sclerified material. 
Digests were performed according to Be- 
thesda Research Labs' (BRL) instructions us- 
ing 14 enzymes that recognize six base pair 
(bp) sequences: BarriH I, Bel I, Bgl II, Cla I, 
EcoR I, Hind III, Hpa I, Kpn I, Pst I, Pvu II, 
Sal I, Sea I, Sma I, and Xho I. Digested DNAs 
were size fractionated electrophoretically in 
agarose (1.2-1.0% for frequent cutters, 0.7- 
0.5% for rare cutting enzymes). DNA was 
transferred from gels to Zetaprobe nylon mem- 
branes (BRL) for probing by the methods of 
Southern (1975) or Reed and Mann (1985). 
Probing was performed using cloned fragments 
of the Pennisetum americanum cpDNA library 
(Thomas et al., 1984) as reported by Soreng, 
Davis, and Doyle (in press). 
Restriction digest patterns were dia- 
grammed from photos of the ethidium bro- 
mide stained gels and fragment sizes calculated 
from lambda Hind III or Pst I standards. The 
general order of single digest fragments was 
indicated by their homology with specific Pen- 
nisetum cpDNA probes. Fragments hybridiz- 
ing to adjacent probes, and subsequent RS 
changes among cpDNA fragments of different 
species within probe regions, were used to ar- 
range fragments in linear order. The entire 
chloroplast genome was mapped in this man- 
ner for Hind III, Hpa I, Kpn I, Pst I, Pvu II, 
Sal I, and Sma I (Fig. I), For the remainder 
of the enzymes, RS were localized within and 
between probe regions (Table 2). Possible in- 
sertion/deletion (I/D) events (fragments less 
than 300 bp in length were not usually resolved 
on gels) were not included in the formal cla- 
distic analysis, but their presence was scored 
(Table 3) for comparison with the RS cladistic 
structure. 
Shared RS were scored as presence/absence 
data (Fig. 1) and analyzed cladistically by im- 
plicit-enumeration of all most parsimonious 
trees using HENNIG-86 (Farris, 1988). Dollo 
parsimony (PHYLIP 3.0, "Dollop" option: 
Felsenstein, 1987) was invoked to examine the 
effects on tree structure of excluding parallel 
RS gains. Shortest trees generated by each 
method were analyzed with CLADOS (K. Nix- 
on, unpublished computer program) to ex- 
amine the distribution and homoplasy of RS, 
investigate alternative trees, and produce cla- 
dograms. 
RESULTS 
Forty-four taxa of Poa from 19 sections and 
groups, distributed in three of four traditional 
subgenera, and two outgroups were examined 
for RS variation using 14 restriction enzymes. 
Of some 16,000 RS examined, about 400 are 
distinct (constituting circa 0.5% of the chlo- 
roplast genome), 127 vary within the study 
group (Fig. 1), and 43 are synapomorphic with- 
in Poa. Approximate chloroplast coordinates 
(relative to the Oryza sequence) of RS are pre- 
sented for some restriction enzymes (Fig. 1); 
other RS were localized within probe regions 
(Table 2). 
The RS presence/absence data (Fig. 1) were 
analyzed in a cladistic framework. Eleven 
equally parsimonious cladograms were found. 
Excluding autapomorphies, these are 87 steps 
long with a consistency index (C.I.) of 0.84. 
One of these trees, that which most closely 
agreed with the strict consensus tree, is pre- 
sented in Fig. 2. Alternative trees are discussed 
below. Sixty-one events scored as unique I/D 
or RS that produced undetected small frag- 
ments (Table 3) are generally congruent with 
the cladogram and add support to the recog- 
nition of certain groups. 
A second cladistic analysis was performed 
using Poa eminens as an out-group within Poa, 
as is suggested by Tzvelev's (1983) treatment 
of the genus. This resulted in ten trees with 
essentially the same structure as Fig. 2 that 
were (excluding autapomorphies) 55 steps long 
with a C.I. of 0.91. This demonstrates that 
about half of the homoplasy evident in Fig. 1 
resulted from the inclusion of the out-group 
genera. 
The major structure of the cpDNA phylog- 
eny of Poa is relatively robust. Out-group anal- 
ysis using data from seven restriction enzyme 
(RE) digests demonstrated Poa to be mono- 
phyletic (see introduction). The cladistic struc- 
November 1990] SORENG?POA CHLOROPLAST DNA 1389 
[ Unmapped Character Numbers ] [ Happed Character Numbers ] 
11111111111111111111111111111 
11111111112222222222333333333344444444445555555555666666666677777777778888888888999999999900000000001111111112222222223 
12345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456780124567890 
BamH I        Bel. I   Bat II Cla I  EcoR I        Xho I Hind III        Hpa I Pst I Pvu II Sma I 
ton I     Sal I 
Taxa 1111 1    1        1 
11111111 11111111112 111111111     110077655544443322 0721763095432164 721866633221 
12345678901234567125456789123456789012345678901234567812345678901234567812345681765357655320815663988349162957348020046504362070 
1. 11000110111001001 10010111101011000010 10110111101111111011010000000000100000011001011110111010000000101110101001 
2. 1000010000000100101110111110100100101110000010 10110011101110111111100000110000000011000001011001011001100010000010110000 
3. 11000100000001001011111111101101101011110000100000111110110011101110111100110000000000000010000010110101011111010010100010110000 
4  11000100000001001 10110110101111000010 10110011101110111111100000000001010010000010110101011110000010100010110000 
5. 111010 0110---0111100  
6. 11000100000001001 10110110101111000010 10110011101110111111110000000000000010000010110101011111000010100010110000 
7. 11000101000011000 10110110100111100010 10110011001110110111110000000100000010100111110101011001000010100000110000 
8. 11000100100001011 10110110101111000110 10110011101110110111110011001100000010000111110101011111100011100010110000 
9. 1100010010000110101110111011110110101111000010000101101011001111111011011111000000010000001000011 1010101111111--100010110000 
10. 11000100100001101011101110111101101011110000100001011010110011111110110111110000000100000010000111100101010111111110100010110000 
11. 1100010010000110101010111011110110101111000010000101101011001111111011011111000000010000001000011 101010111111110100010110000 
12. 111011001000011011010011001111011010111101001100000110101110111110101101.111100000001000000100001111001010111111101--100010110000 
13. 1110110010000110110100110011110110101111010011000001101011001111101011011111000000010000001000011 10110100010110000 
14. 111011001000011011010011001111011010111101001100000110101100111110101101 000001000000100001111001010111111101--100010110000 
15. 111011001000011010111011101111011010110100001100000110101100111110101101111100000001000000100001111001010111111101--100010110000 
16. 11101100100001101111100110111101101011110000110010011011010011111010110111110000000100000010000111100101 10110  
17. 11101100100001101 11110110101111000011 101100111110101101 0000010000001000011 101  
18. 1110110010000110111110111011110110101111000011 10110011111010110111110000000100000010000111100101 101--100010110000 
19. 1110110010000110111110111011110110101111000011100001101011001111101011011111000000010000001000011 101011111110110100010110000 
20. 11101100100001101 1111011010111100001100000110101100111110101101111100000001000000100001111001010111111001--100010110000 
21. 11101100100001101 11110110001111000011 10010001111010010111110000000100000010000111100101 101--100010110000 
22. 11101100100001101111101010111101101011110000110000011010110011111010010111110000000110000010000111100101 10110100010110000 
23. 11101100100001101 111100101011110000110000001010110011111010110111110000000100000010000111100101 101--100010110000 
24. 1110110010000110111010111011110110101111000011000001101011001111101011011111000000010000001000011 101011111110110  
25. 11101100100001101 11110110101111000011 101100111110101101 000001000000100001111001010111111001--100010110000 
26. 1110110010000110111110111011110110101111001011000001001011001111101011011111000000010000000000011 101011111110110100010110000 
27. 11101000100001101 11110110101111000011 101100111110101101 101  
28. 1110110010000110111110111011110110101111000011000001101011001111101011011111000000010000001000011 101  
29. 11101100100001101111101110111101101011110000110000011010110011111010110111110000000100000010000111100101 10110100010110010 
30. 111011001000011011111011101111011010111100001100000110101100111110101101111100000001000000100001111001010111111101--100010110000 
31. 1110110010000110111110111011110110101111000011000001101011001111101011011111000000010000001000011 101 10110100010110000 
32. 11111100100001101 11110110101111000011 10110011111010110111110000000100000010000111100101 101--100010110000 
33. 1110110010000110111110111011110110101111000011010001101011001111101011011111000000010000001000011 10110100010110000 
34. 11000100100000001 10111110101111000010 10110011111110110111110100000100000010000111100101 00110100110110000 
35. 11000100100000001 10111110101111000010 10100011111110110111110100000100000010000111100101011111100110100111110000 
36. 01000100100000001 10111110101111000010 1011001111111011011111010000010000001000011 001--100110110000 
37. 11000100100000001 10111110101111000010 101100111111101101 0000010000001000011 1010111111001  
38. 11000100100000001 1010001111111011011111010000010000001000011 101 10  
39. 1100010010000000101110111010111110101111000010 10100011111110110111110100000100000010000111100101 00110100110110000 
40. 11000100100000001 10110110101111000000 0011001011111010011011000000010000001000011 101 00110  
41. 11000100100000001 10110110101111000000 1100 001  
42. 11000100100100001 10110110101111000000 00110010111100100110110000000100000010000111100101001111100110100010110000 
43. 11000100100000001 10110110101111000000 001100101111101001 0000010000001000011 101011111100110  
44. 11000100100000001 00110110101111000000 00110010111110100110110000000100000010000111100101 00110100010110000 
45. 11000100100000001011101110001101101011110000000000111000110010111110100110110000000100000010000111100101011111100110110010110000 
46. 10000100100000001 10110110111111000010 101100101101101101 000001000011100001111001010111111001--000010110100 
Fig. 1. Apomorphic restriction sites among species of Poa and out-groups. Taxa numbered as in Table 1. RS arranged 
by enzyme; character numbers correspond to Fig. 2; RS arranged within probe regions numbered as in Table 2; mapped 
RS coordinates correspond to approximate order relative to the Oryza cpDNA sequence, beginning at the junction of 
the large single copy region with the inverted repeat nearest psbA gene and proceeding toward the rbcL gene. Data 
codes are: 0 = absent, 1 = present, ? = no data. A A marks the first character in each enzyme RS set. 
ture within Poa did not change when the two rest of Poa and fixation of the branching pattern 
data sets were merged, but fewer shortest trees in group IV as depicted in Fig. 2. However, 
were obtained. Based on these data it was not the five major groups with their included spe- 
possible to resolve the trichotomy between Poa, cies remained unchanged in all analyses. 
Puccinellia, and Bellardiochloa, in strict con- The following sections or groups of Poa are 
sensus trees. Within Poa, three basal groups represented within the five major cpDNA 
(I?III) and two major derived cpDNA sister groups. Group I: Arctopoa, Sylvestres; Group 
groups (IV, V) were identified. Groups IV and II: Caespitosae; Group III: Ochlopoa; Group 
V share three to four (character 26 is not fixed IV: Australopoa, Dasypoa, Dioicopoa sensu 
in position) derived RS and four I/D events, stricto (s.s.), Diversipoa, Homalopoa, Macro- 
but are separated by five RS and 141/D events, poa, Madropoa, Nervosae, Poa, Punapoa; 
Dollo  parsimony,  which prohibits  parallel Group V: Abbreviatae, Bulbosae, Coenopoa, 
gains, resulted in two changes within Poa: Oreinos, Secundae, Stenopoa, Tricopoa. Most 
placement of P. alpina as a sister group to the of the variation among the parsimony trees 
1390 AMERICAN JOURNAL OF BOTANY [Vol. 77 
TABLE 2.    Unmapped Poa chloroplast DNA restriction site changes 
No.' Enzyme Regionb Change0 
1 
2 
3 
4 
5 
6 
7 
8 
9 
10 
11 
12 
13 
14 
15 
16 
17 
18 
19 
20 
21 
22 
23 
24 
25 
26 
27 
28 
29 
30 
31 
32 
33 
34 
35 
36 
37 
38 
39 
40 
41 
42 
43 
44 
45 
46 
47 
48 
49 
50 
51 
52 
53 
54 
55 
56 
57 
58 
59 
60 
61 
62 
63 
64 
65 
Bam HI 
Belli 
Bgl II 
Clal 
EcoRI 
7 1.95 + 0.64 - 2.6 36 
5 8.1 + 1.3 - 9.3 2,46 
5 4.65 - 0.67 + 4.0 12-33 
5 4.0 - 2.2 + 1.7 32 
5 3.5 - 0.5 + 3.0 12-33 
5 3.0 + 4.4 - 7.5 27 
5 4.4 - 2.2 + 2.3 1 
5 4.4 - 3.6 + 0.8 7 
5-4 17.5 + 2.5 - 20.1 2-7 
4 17.5 - 4.4 + (6.4 + 4.7) 1 
4-3 17.5 - (4.4 + 6.4) + 4.7 1 
4-3 17.5 - 13.0 + 4.45 42 
3 1.7 -> 1.4 + 0.4 7 
3-6 2.6 + 1.05 - 3.9 34-46 
2 6.65 -> 0.76 + 5.7 9-33 
2 6.7 -> 3.4 + 3.5 8 
2 6.7 + 3.45 -> 10.1 7 
6 2.9 -> 1.75 + 1.25 12-14, 16-33 
5 2.2 + 2.05 -> 4.25      ' 12-14 
5-4 4.3 + 7.5 - 11.8 (partial?) 11,24 
3 2.5 + 1.84 -? 4.5 12-14 
3 1.84 - 0.88 + 0.95 3 
3 1.84 + 6.2 -? 8.75 16 
3-2 1.75 + 3.1 - 4.9 22 
3-2 3.1 + 2.5 -> 5.6 12-14 
2-1 3.55 + 6.9 -> 10.45 (7-)9^16 
1 5.5 + 4.0 - 9.6 44,45 
7 2.9 -> 2.6 + 0.3 9-33 
6 4.1 + 0.8 - 4.8 1 
5 0.8 + 7.7 -? 8.6 2 
5 7.7 -> 5.7 + 2.0 34-39 
5 0.7 + 6.9 - 7.7 23 
5 6.9 - 1.5 + 5.5 2 
5 6.9 -> 2.4 + 4.1 1 
5-4 6.9 + 3.1 - 10 21 
4 3.1 - 2.4 + 0.7 46 
4 3.5 + 1.78 - 5.4 7 
3 1.6 + 1.2 - 2.9 1 
3 1.2 + 3.4 -> 4.6 15 
3 3.4 + 1.1 -? 4.5 2 
3-6 16.5 - 8.6 + 7.2 7 
3-6 16.5 -> 11.5 + 5.0 12-14 
3-6 16.5 - 12.0 + 4.6 26 
3-6 16.5 -> 14.5 + 2.2 8 
2 2.2 + 1.3 -> 3.5 4CM5 
1 5.0 - 4.1 + 1.0 12-33 
5 12.5 - 11.55 + 0.95 19 
5 3.95 -> 1.95 + 1.85 33 
4 8.0 - 5.35 + 2.55 16 
3 5.0 -? 4.1 + 1.0 9-11 
3 0.7 + 1.03 -? 1.75 9-33 
2 0.83 + 1.45 -> 2.35 23 
2 1.5 + 3.45 - 1.18 + 3.77 26 (inversion?) 
6 5.6 -> 3.7 + 1.9 3 
7 1.3 (IR) + 0.1 & 0.18 - 1.58 & 1.4 40-45 
7 2.6 - 1.8 + 0.8 16 
7 1.07 + 0.3 -> 1.37 16,21 
7 2.6 + 0.4 - 3.0 35, 38-39 
6 1.8 -> 0.57 + 1.03 12 
6 2.3 - 1.9 + 0.5 1 
6 2.3 + 0.74 - 3.0 21 
5 0.74 + 0.74 - 1.59 40-46 
4 3.4 - 3.0 + 0.4 7 
4 5.9 -? 4.2 + 1.9 9^16 
4 1.0 + 2.0 - 2.81 46 
November 1990] SORENG?POA CHLOROPLAST DNA 1391 
TABLE 2.   Continued 
Enzyme Region6 Change0 Taxa" 
66 
67 
68 
69 
70 
71 
72 
73 
74 
75 
76 
77 
78 
4 2.0 + 0.3 - 2.31 
3 6.8 + 1.0 -? 8.0 
3 6.8 + 1.9 - 5.8 + 
3 1.8 + 1.28 -> 3.1 
3 1.1 + 1.24 -> 2.34 
2 1.65 + 2.4 - 4.5 
2 2.4 + 4.3 - 6.4 
4 12.5 + 3 + 7.9 - 
4 3.0 + 7.9 -? 10.8 
3 9.6 + 7.6 - 17.6 
2 3.9 + 16.5 - 20.4 
2 20.4 ?11+8.4 
2 16.5 - 10.8 + 6.5 
2.6 
21.5 
12-33 
42 
1 
21,22 
40-45 
7^16 
1 
3 
3,40-45 
1 
2,5-6 
6 
34-39 
a
 RS are numbered in order by enzyme and probe region. 
b
 Chloroplast probe regions are coded as follows with included map coordinates [increments = 1 kb, beginning at 
the juncture of the large single copy region nearpsbA and proceeding toward rbcL]: 1 = 0-12; 2 = 12-26; 3 = 25-43; 
4 = 43-57; 5 = 57-78; 6 = 78-96, 117-135; 7 = 96-117. 
c
 Changes are unpolarized; arrows indicate the state in the taxa listed. 
d
 Taxa are numbered as in Table 1. 
occurred in group IV. The structure of each of 
these groups is discussed below. 
Groups I-III? The set of species included in 
these groups was constant, but the basal struc- 
ture of the genus remains tentative. The P. 
alpina and P. annua accessions possessed many 
autapomorphies, suggesting independent di- 
vergences of their respective sections. The po- 
sitions of these taxa were consistent among all 
global parsimony trees. However, as noted 
above, P. alpina moved to the basal-most po- 
sition in Poa under the Dollo constraint. In all 
global parsimony trees, the sections Sylvestres 
and Arctopoa were depicted as a sister group 
to P. alpina and P. annua and groups IV and 
V. Poa eminens, although united with the Syl- 
vestres by one RS loss, bears no apparent mor- 
phological, geographical, or ecological rela- 
tionship to the group. Poa eminens, which 
occurs along arctic coastal-strand, is rhizom- 
atous and halophytic (as are other species of 
subgerms Arctopoa), whereas Sylvestres are ces- 
pitose, temperate forest, montane species that 
share advanced characteristics of subgenus Poa. 
Although these sections shared one RS loss, 
this could represent parallelism (parallel losses 
being more frequent than parallel gains). Their 
otherwise similar RS patterns may stem in part 
from incomplete sampling in the latter group. 
Further resolution within cpDNA groups I-III 
requires additional RS analysis involving more 
species of Poa and more closely related genera. 
Group IV?This group included three sharp- 
ly denned subgroups; "A" {Poa and Macro- 
pod), "C2" (Dioicopoa), arising from within 
or as a sister group to "Cl" (a diverse group 
of species including Dasypoa, Diversipoa, 
Homalopoa, gynodioecious, gynomonecious, 
and dioecious species, among others). A fourth 
group, "B" (including Australopoa), is sepa- 
rated from group C by a single RS gain in the 
latter group. Poa sieberiana lacks a Bel I RS 
that is synapomorphic for IVC, yet no more 
can be said about this RS since not all taxa 
have been tested. Moreover, it could represent 
a reversal. Group IV subdivisions were sup- 
ported by eight RS between groups A and C, 
and C2 was distinguished from Cl by four RS 
synapomorphies. 
Five of the 11 shortest cladograms occurred 
within group IVC1. Alternative trees placed P. 
bigelovii, P. chamaeclinos, P. piperi, R. rhi- 
zomata, P. tracyi, and P. hybrida in differing 
combinations, as sister taxa to the remainder 
of Cl or as derived from within the group. 
Most of the alternative cladograms arose from 
gaps in the data that result in the first four taxa 
attaching within Cl, or between Cl and IVA. 
A Pvu II RS absence whose polarity is uncertain 
caused P. tracyi and P. hybrida to join the tree 
either above or below the IVC1 node. This RS 
was present in all other samples of group IVC 
and A, in P. eminens (group I), and in species 
of two other Pooideae tribes (Soreng, Davis, 
and Doyle, in press). Because losses are more 
likely than parallel gains (Debry and Slade, 
1985) it seems best to consider this a loss after 
a single gain, at least within group IVC. This 
results in the arrangement of group IV taxa 
seen in Fig. 2 and in Dollo trees. If the presence 
of this RS is homologous among the taxa, then 
it must have originated deep in the phylogeny 
1392 AMERICAN JOURNAL OF BOTANY [Vol. 77 
TABLE 3.   Poa chloroplast DNA insertion/deletion events 
Enzyme 
No.' Regionb Changec Taxa" 
BamHI 
1 6 1.2 -? 1.5 = 0.3 1 
2 5 4.65 - 4.8 = 0.15 1 
3 = 14 5 4.45 - 4.25 = 0.2 9-11 
4= 13 5 4.45 -? 4.35 = 0.1 16 
5 = 39, 49 5-4 2.6 -> 2.5 = 0.1 9-33 
6 = 34 3 3.5 -? 3.15 = 0.35 1 
7 3 1.73 - 1.74 = 0.01 1,3,4,29,33 
8 = 62 3 1.73 - 1.7 = 0.03 12-14 
9 = 61 3-2 1.05 - 1.18 = 0.13 1 
10 2 6.7 - 6.65 = 0.05 9^16 
11 = 36, 64, 69 7 4.95 -'4.8 = 0.15 34-45, 46? 
Bel I 
12 5 2.05 - 1.97 = 0.08 3 
13 = 4 5 2.05 - 1.95 = 0.1 16 
14 = 3 5 2.05 -? 1.85 = 0.2 9-10 
15 = 55 2 0.76 -? 0.85 = 0.09 26,28 
Bgl II 
16 7 1.5 - 1.55 = 0.05 3 
17 7 3.98 - 3.85 = 0.13 7 
18 7 2.9 - 2.65 = 0.25 9-33 
19 6 1.52 - 1.55 = 0.03 3 
20 5 0.8 - 0.78 = 0.02 46 
21 5-4 6.9 -? 7.0 = 0.1 4 
22 5-4 6.9 - 6.3 = 0.6 2 
23 = 79 5-4 6.9 - 6.6 = 0.3 7-8, 19, 24 
24 = 50 4 3.1 - 3.05 = 0.05 12-14 
25 4 3.1 -> 3.2 = 0.1 7 
?   26 = 54 4 3.12 - 3.1 = 0.02 1-2, 9-33, 46 
27 4 3.5 - 3.49 = 0.01 1 
28 4 1.78 - 1.76 = 0.02 1 
29 4 1.78 - 1.79 = 0.01 9-11, 15, 17-18 
20-26, 28-33 
30 4 1.78 - 1.8 = 0.02 12-14, 16, 19, 27 
31 4 1.78 - 1.81 =0.03 4 
32 4 1.23 - 1.20 = 0.03 40-45 
33 = 46 4 0.95 - 0.94 = 0.01 9-33 
34 = 6 3 3.4 - 3.6 = 0.2 1 
35 3-2 16.5 -> 17.0 = 0.5 46 
36= 11,64,69 2-1 3.4 - 3.25 = 0.15 34-46 
37 2-1 2.17 - 2.19 = 0.02 1 
38 2-1 2.17 - 2.18 = 0.01 13 
Clal 
39 = 5, 49 5 4.05 -? 3.95 = 0.1 9-33 
40 4 0.7 -? 0.86 = 0.16 15 
41 =72 4 0.65 -> 0.56 = 0.1 12-14 
42 2-1 3.45 -? 3.8 = 0.35 3 
EcoRI 
43 7 2.6 -> 2.57 = 0.3 3, 6, 20 
44 6 3.66 - 3.65 = 0.01 3, 6, 9-33 
45 6 3.66 - 3.45 = 0.21 1,2 
46 = 33 5 2.35 - 2.33 = 0.02 9-33 
47 = 58 5 2.35 - 2.32 = 0.03 34-46 
48 = 59? 5 1.55 - 1.59 = 0.04 7 
49 = 5, 39 5-4 3.4 -> 3.27 = 0.16 9-33 
50 = 24 4 4.2 - 4.15 = 0.05 12-14 
51 4 2.1 - 2.31 =0.2 12-29,31-33 
52 4 2.1 - 2.32 = 0.21 1 
53 4 2.1 - 2.15 = 0.05 8 
54 = 26 4 1.4 -> 1.38 = 0.02 7-33, 46 
55= 15 2 4.5 - 4.4 = 0.1 26 
November 1990] SORENG?POA CHLOROPLAST DNA 1393 
TABLE 3.   Continued 
Enzyme 
No." Rcgionb Change" Taxa" 
56 
57 
2 
2 
4.5 -4.37 = 0.13 
4.3 - 4.35 = 0.05 
34-37, 39 
34-46 
Hind. Ill 
58 = 47 
59 = 48 
60 
61 =9 
62 = 8 
63 
64= 11,36,69 
65 
(70-72) 
(52-61) 
(37^13) 
(32-34) 
(29-32) 
(13-16) 
(8-13) 
(8-13) 
2.05 - 2.00 = 0.05 
9.4 - 9.2 = 0.2 
7.05 - 7.1 =0.05 
2.55 - 2.65 = 0.1 
3.55 - 3.5 = 0.05 
3.6 - 4.0 = 0.4 
5.5 - 5.4 = 0.1 
5.5 - 5.6 = 0.1 
34-46 
7 
4, 35-^5, 46? 
1 
12-14 
8 
34-45 
13 
Hpal 
66 
67 = 81 
68 = 78 
(100-102) 
(52-53) 
(24-29) 
2.7 - 2.6 = 0.1 
0.9 - 0.84 = 0.06 
6.4 - 6.6 = 0.2 
1 
14 
3-8 
Kpnl 
69= 11,36,64 
70 
71 
(5-9) 
(0-9) 
(0-9) 
5.0 -> 4.9 = 0.1 
8.0 - 7.9 = 0.1 
8.0 - 8.1 =0.1 
34-46 
3-8, 34-46 
9-33 
Pstl 
72 = 41 
73 
74 
(51-54) 
(37^5) 
(34-37) 
2.7 - 2.6 = 0.1 
9.0 - 9.2 = 0.2 
3.1 - 2.9 = 0.3 
12, 14 
9-45 (73 = 35 taxon 46) 
3^16 
.PvwII 
75 = 76 (30-35) 5.3 - 5.8 = 0.5 1 
Sal I 
.    76 = 75 
77 
78 = 68 
(26-34) 
(23-29) 
(23-29) 
6.6 - 6.9 = 0.3 
7.1 - 6.9 = 0.2 
7.1 - 7.3 = 0.2 
1 
40 
3-8 
Sma I 
79 = 23 (54-61) 7.2 - 7.1 =0.1 3-8, 19 
Seal 
80 
81 =67 
82 
(39-59) 
(39-59) 
3 
4.0 - 3.7 = 0.3 
4.0-4.1 =0.1 
1.4- 1.55 = 0.15 
11 
14 
9-33 
Xhol 
83 (52-55) 3.7 - 3.65 = 0.05 9-33 
a
 Insertion/deletion events are numbered in order by enzyme and probe region, or, where mapped, their inclusive 
map coordinates are given in parentheses. Equivalent events detected in other digests are indicated by an "=" sign. 
b
 Chloroplast regions are coded as in Table 2. 
" Changes are not polarized, arrows indicate the state in the listed taxa. 
d
 Taxa are numbered as in Table 1. 
of the subfamily Pooideae and may have been 
lost as many as ten times. 
Group V?Group V includes two parts: "D" 
(encompassing the P. bulbosa accession, and 
the group Secundae), and "E" (encompassing 
species of the Abbreviatae, Oreinos, Stenopoa 
s.s., and Tricopoa, with Coenopoa basal). The 
union of group V, marked by only one RS, was 
supported by four different I/D events. 
DISCUSSION 
Interpretation of the cpDNA phytogeny of 
Poa?In this section I discuss congruence be- 
tween the cpDNA tree (Fig. 2) and traditional 
classification of the genus Poa. Strict cladistic 
analysis cannot represent the full evolutionary 
history of reticulately inherited characters. In 
Poa polyploidy and apomixis (which facilitate 
introgression and perpetuation of hybrids) are 
1394 AMERICAN JOURNAL OF BOTANY [Vol. 77 
common and diploid species rare, numerous 
hybrids occur naturally or have been synthe- 
sized (Knobloch, 1968), and reticulate origins 
of certain species groups have been postulated. 
Although cladistic analysis can be applied as 
an exploratory tool in reticulating groups, lim- 
ited attempts to use it on gross morphological 
and anatomical characters of Poa (Soreng, un- 
published data) were confounded by homo- 
plasy, some of which evidently derived from 
reticulation; thousands of shortest trees were 
generated that exhibited low consistency (0.3- 
0.4), and all major nodes collapsed in strict 
consensus trees. The nature of evolution in Poa 
to some extent is responsible for numerous 
equally parsimonious and often highly contra- 
dictory solutions to formal cladistic analyses 
based on morphological characteristics. One 
way to circumvent theoretical and practical 
limitations to cladistic analysis where reticu- 
lation is suspected is to define relationships by 
applying parsimony analysis to linearly inher- 
ited characteristics such as cpDNA RS. Cau- 
tion is needed in interpretation of results where 
lineage sorting, hybrid origins, and introgres- 
sion may have occurred (Palmer et al., 1983; 
Doyle, Doyle, and Brown, in press), but post- 
facto search for inconsistencies with other types 
of data can expose possible hybrids. The pres- 
ent ? cladistic analysis of cpDNA RS in Poa 
revealed a relatively stable phylogenetic struc- 
ture with low homoplasy. 
One measure of the correspondence of the 
cpDNA cladogram to a species phylogeny, in 
the absence of a stable cladogram of Poa phy- 
logeny based on other data, is the congruence 
of RS synapomorphies with the existence of 
morphologically coherent groups and with pre- 
viously postulated relationships among groups. 
Examples of congruence between the cpDNA 
cladogram and morphological groups occurred 
for many sets of closely allied taxa: 1) Poa 
arachnifera of North America and P. iridifolia 
and P. lanigera of South America (Dioicopoa 
s.s.); 2) P. confinis, P. cusickii, P. fendleriana, 
P. piperi, (North American dioecious species 
of Madropoa); 3) P. tracyi, P. cuspidata, P. 
nervosa, P. rhizomata (North American gyn- 
odioecious species, Nervosae); 4) P. bigelovii 
and P. occidentalis (Diversipoa); 5) P. canda- 
moana and P. chamaeclinos (Punapoa); 6) P. 
arctica and P. pratensis {Poa); 7) P. secunda 
and its putative sexual counterparts P. napensis 
and P. tenerrima (Secundae); 8) P. alsodes, P. 
saltuensis, and P. autumnalis (Sylvestres); 9) P. 
palustris, P. nemoralis, and P. compressa (Sten- 
opod); 10) P. keckii and P. brachyanthera (Ab- 
breviatae); 11) P. laxa (unpublished data) and 
P. paludigena (Oreinos). 
Additional support for the congruence of the 
cpDNA phylogeny and morphology comes 
from the placement of some unique groups in 
positions consistent with previous hypotheses. 
1) Bellardiochloa has been placed within Poa 
as a sister group (Pseudofestuca Asch. & 
Graebner) to the remainder of the genus, or is 
treated as a distinct genus allied to Poa (Ed- 
mondson, 1980) or Festuca (Tzvelev, 1983). 
Cladistic analysis of morphological characters 
(Soreng, unpublished data) placed it outside of 
Poa, and of cpDNA RS (Soreng, Davis, and 
Doyle, in press) placed it near Festuca, sepa- 
rated from Poa by Dactylis and Arctagrostis. 
Bellardiochloa caryopses have soft to semi-liq- 
uid endosperm (as in Dactylis and Arctagrostis 
and many genera of the tribe Aveneae) (vs. 
solid endosperm in Poa and most other genera 
within the Poeae and other tribes of the sub- 
family) and round backed lemmas (character- 
istic of most genera in the subfamily) with a 
short awn (awns being common in the subfam- 
ily) (vs. keeled and unawned lemmas in Poa 
[two South American Poa have short awns]). 
2) Poa eminens and four relatives, recently re- 
moved to a new genus, Arctopoa, were sub- 
sequently reunited with Poa as subgenus Arc- 
topoa (Tzvelev, 1983). These species are 
postulated to be basally derived within Poa 
(Tzvelev, 1983), and indeed P. eminens comes 
out in a basal group in the cpDNA analysis. 
3) Poa sieberiana and P. labillardierei, species 
of a distinctive group of tussock grasses en- 
demic to New Zealand and Australia, exhibited 
a cpDNA RS pattern little differentiated from 
the generalized group IVC type common in 
North and South America. Australopoa are un- 
likely to have arisen from hybridization with 
species from other continents and always 
formed a clade with the very similar P. stric- 
tiramea of North America in cladistic analyses 
of morphology. The relative ease of crossing 
between the tussock grass P. "cespitosa" (a 
problematic name, once widely applied to spe- 
cies of this group) from Australia and P. ar- 
achnifera from North America (Clausen, 1961), 
as compared to the difficultyof crosses between 
these and other species tested (Hiesey and Nobs, 
1982), supports the inclusion of Australopoa 
in group IVC. 4) Poa hybrida is a diploid mem- 
ber of the western Eurasian section Homalo- 
poa. It closely resembles certain North Amer- 
ican species, particularly P. occidentalis 
(Diversipoa) (one of the three known diploids 
in North America) and P. tracyi, with which 
November 1990] SORENG?POA CHLOROPLAST DNA 1395 
10   11   29   34   38  60   66   72   75   67   91   94   95 103 104 105 110 115 120 126 127130 
I?[>5H>OHDHDH>CH>OHDKl>-aH3H>H>CHD-?HDH^ 
30   33   40   76   61   62   92 102 109111 117 
?5HHD-<>D-D--D--D--n--D-[9--?--H-Bellardioohloa 
22   54   73   74 112 
i?D-O-D-8 B ominens Arctopoa 
98 
97  99 101 106 
HXHHH 
autumnalis Sylvestres 
66   68 
-alsodes Sylvestres 
artuensis 
13   17   37   41   63   93 108 109122 
I?|><H>CH>-C>-D--S-D-n--alpina Caespitosae 
16   44   79   60  83 116 
i?D-D-D-O-D-D-annua Oehlopoa 
71   84   96 l
?D-D-DH 
Poa Chloroplast Groups 
?     I 
I     HI 
50 107113 
15   28   51  112 
26   64  100 114 
?B-ibe/ica Macropoa 
-arctica   Poa 
L^-prateasls " 
39 
?O-sieberiana Australopoa 
23   49   56   57 
?D?D-CHS?conglomerata Dasypoa 
?bigelovii Diversipoa 
iccidentalis    " 
7 
?Q-orizabensis " 
112 |?hybrida Homalopoa 
'?tracyi   Nervosae 
35   57   61 
wspidata 
24   35 
?nervosa 
32   52 
?O-CHrhizomata   ? 
20 
?H?strictiramea 
43   53   91 
?D-CHS-candamoana Punapoa 
e 
?D-chamaeclinos    " 
48 
?D-wheeleri X 
129 
?D-cuslckii Madropoa 
albescens     " 
?f.fendleriana   " 
tonfinis 
?P?piper!        " 
IVA 
I      IVB 
IVC1 
19  21   25   42 Q-arachnifera Dioicopoa 
?iridifolia 
1
?lanigera      " 
?bulbosa Bulbosae 
1 
SkJ^J^L. ?E>-canbyi36 Secundae 
?canbyi37        " 
124 
?Q-napensis    " 
sandbergii      " 
'?tenerrima 
2    36   65   89   90 117128 
I?HK>Cl-a-CH5KIHrivialis Coenopoa 
?keckii Abbreviatae 
?brachyanthera   " 
?   ?   7ft   7, 12   67 106 
_Q-Q-Q_H- "~D-O-O-paludigena Oreinos 
IVC2 
VD 
VE 
?compressa Trtcopoa 
27 i?nemoralis Stenopoa 
L-OH   "a 1
?D-palustris    " 
Fig. 2. One of 11 most parsimonious trees of Poa chloroplast DNA restriction sites. Of the 11 most parsimonious 
trees, this tree most closely approximates the strict consensus tree. Excluding autapomorphies, the tree is 87 steps long 
with a consistency index of 0.84. Numbers over boxes refer to restriction site characters (Fig. 1; Table 2); open boxes 
represent unique site gains and losses. Putative gain-loss, loss-gain, and parallel loss events (polarized to Puccinellia) 
are indicated within boxes by / for gains and \ for losses; solid boxes represent parallel gains. Group designations 
specify chloroplast groups discussed in the text, and classification of taxa follows Table 1. 
1396 AMERICAN JOURNAL OF BOTANY [Vol. 77 
it shares the derived character states of strongly 
keeled sheaths closed over half their length, 
exceptionally tall stature, and elongate panicles 
(Soreng and Hatch, 1983). Poa hybrida was the 
only accession of Eurasian origin found to have 
a group IVC cpDNA type and it shared a nearly 
identical RS pattern with P. tracyi. (Inciden- 
tally, I have found no character support for 
maintaining Diversipoa as distinct from Hom- 
alopoa.) 5) The dioecious Madropoa and Dioi- 
copoa s.s. are postulated to be sister groups 
(Marsh, 1952; Clausen, 1961; Soreng, 1986), 
and this was not rejected by the cpDNA phy- 
logeny, nor was a putative relationship between 
these and the North American partially gyn- 
odioecious group, Nervosae (Soreng, 1986), re- 
jected. 6) Oreinos, Abbreviatae, and Stenopoa, 
considered to be closely allied by Edmondson 
(1980) and Tzvelev (1983) all fall into cpDNA 
group VE. 7) Poa trivialis is considered to rep- 
resent a distinct section allied to Stenopoa and 
Abbreviatae (Tzvelev, 1983), and also falls 
within group VE. 8) Laterally creeping rhi- 
zomes are present in many group IV species 
but rarely occur outside this group within Poa. 
The rhizomatous habit also coincides with 
sheaths closed over one-third their length or 
more within group IV. 
?Reticulate evolution?Reticulate relation- 
ships may have affected the positions of the 
following five taxa in the cpDNA cladogram 
(Fig. 2). 
1) The Bulbosae (including P. bulbosa) are 
usually considered to be closely related to the 
Caespitosae (including P. alpina). Edmondson 
(1978) includes these groups in their own sec- 
tion Bulbophorum, and Tzvelev (1983) treats 
them as two of four subsections within section 
Poa. Although the individual groups are de- 
fined by morphological autapomorphies (ex- 
cept for section Poa), characters used to infer 
relationships among them are widespread in 
the genus (smooth panicle branches, pilose pa- 
lea nerves, sheaths closed one-fourth to two- 
thirds their length, anthers 1.2-2.8 mm long). 
For these and the following reasons I tenta- 
tively consider each a separate section. The Poa 
alpina accession had a very distinct cpDNA 
RS pattern, which suggests that inclusion of 
the Caespitosae in or near section Poa should 
be reevaluated. The position of the P. bulbosa 
accession in Fig. 2 was also far removed from 
section Poa, and it shared three synapomorph- 
ic RS with the Secundae. Two plausible ex- 
planations for this are: P. bulbosa (a polyploid 
apomict) hybridized with the Secundae in Cal- 
ifornia (the origin of the accession, and where 
P. bulbosa is adventive), or an unidentified 
parent (Puccinellia has been suggested [Steb- 
bins, personal communication]) hybridized 
with some relative of Bulbosae in the distant 
past giving rise to the high polyploid, and pre- 
dominantly apomictic, Secundae. To resolve 
cpDNA relationships in these sections, diploid 
sexual species of both Bulbosae and Caespi- 
tosae, which occur in Europe and the Medi- 
terranean region, will have to be sampled. 
2) The possibility that the Secundae are of 
reticulate origin is postulated because of their 
possession of a mixture of'Puccinellia and Poa- 
like characteristics and occurrence of high pol- 
yploidy and extensive apomixis (Stebbins, 
1950, p. 404). Explicit support for this hy- 
pothesis comes from the combination of char- 
acters thought to be ancestral within Poa, or 
reticulately inherited in the Secundae (rounded 
lemmas, a crown of callus hair, and promi- 
nently papilliate epidermal cells in some spe- 
cies [Soreng, unpublished data]), coupled with 
a derived cpDNA type characteristic of species 
with advanced Poa morphology. 
3) Poa compressa is a polyploid apomict 
clearly allied on morphological grounds to sec- 
tion Stenopoa, in which it is placed as a sub- 
section by Tzvelev (1983). It, however, has one 
unique and one rare character within group V 
and section Stenopoa?strongly creeping rhi- 
zomes and highly compressed culm nodes. I 
suggest that these characters might be derived 
from hybridization between P. rhemannii 
(Asch. & Graebner) Woloszcz., a diploid Sten- 
opoa species with compressed culms (as a chlo- 
roplast donor), and P. pratensis, a group IV 
species with similarly strong creeping rhi- 
zomes. 
4) The placement of P. iberica (section Mac- 
ropoa), in group IVA with section Poa, requires 
additional confirmation. Macropoa is usually 
aligned near section Homalopoa (group IVC1). 
The accession used in this study might be of 
hybrid origin with P. pratensis, which forms 
fertile hybrids with P. longifolia Trin. also of 
section Macropoa (Vandijk and Winkelhorst, 
1982). 
5) Poa wheeled, a high polyploid apomict 
occurring widely in western North America, is 
usually submerged within P. nervosa as a va- 
riety. If the fairly narrow endemic P. nervosa 
s.s. was one of P. wheelerVs parents, some other 
species (I have suggested P. cusickii, and one 
I/D found was shared between P. wheeled and 
one of the P. cusickii accessions) may have been 
the chloroplast donor. Poa nervosa and P. cus- 
pidata, two sexual gynodidecious species dis- 
junct between temperate deciduous forests of 
eastern and western North America, shared a 
RS loss. Poa wheeled could have received its 
November 1990] SORENG?POA CHLOROPLAST DNA 1397 
chloroplast from P. nervosa, but its origin would 
then predate the vicariance of P. nervosa and 
P. cuspidata. Alternatively, P. wheeleri could 
have derived its different chloroplast through 
hybridization, introgression, or lineage sorting. 
The examples of possible reticulation events 
cited above suggest caution when interpreting 
the cpDNA phylogeny in Poa. Stebbins may 
have been right in suggesting Poa might rep- 
resent one huge polyploid complex, yet such 
complexes often contain structure. Further 
testing will be required to confirm putative 
reticulation events. On the other hand, the 
presence of clearly definable cpDNA groups 
that correspond to morphologically defined 
groups and sections agrees with the prediction 
that there were and are independent lines of 
evolution in Poa, and that the cpDNA clado- 
gram can be used, cautiously, and along with 
other data, for developing a classification of 
the genus. 
Biogeography ?The cpDNA cladogram, 
coupled with other information, can generate 
hypotheses of origin and radiation of sectional 
groups. The center of diversity of Poa in num- 
ber of sections or groups, diversity within 
groups, and range of cpDNA types, is clearly 
Eurasia. This agrees with chorological studies 
that demonstrate Eurasia to be the center of 
generic and species diversity within the inclu- 
sive subfamily Pooideae and tribe Poeae (Hart- 
ley, 1961, 1973; Cross, 1980). Of the 11 sec- 
tions in which diploids are known, seven are 
centered in, or restricted to, western Eurasia. 
Of the 19 groups of species studied for RS, nine 
are restricted to Eurasia, or their primary cen- 
ters of diversity are in Eurasia with few taxa 
indigenous to North America. The latter taxa 
have circumpolar distributions (e.g., P. alpina, 
P. arctica, P. eminens, P. laxa (Oreinos), P. 
nemoralis s.l. (including P. glauca Vahl), P. 
pratensis), and all except P. eminens and P. 
laxa are known apomicts. The Abbreviatae are 
centered in Berringia where there are several 
diploid species. The sister group relationship 
of section Poa and cpDNA group IVC, which 
is supported by morphological similarity and 
cladistic character analysis, suggests a vicari- 
ance event, because section Poa has its center 
of species richness in northeastern Asia and 
Berringia, whereas group IVC is principally 
New World. 
The cpDNA connection between Homalo- 
poa, a group centered in Europe with several 
diploid species (the accession studied is a dip- 
loid collected from Greece) and group IVC is 
more difficult to explain. The morphology of 
the group is, as discussed above, strikingly sim- 
ilar to that of some North American species. 
It may represent a primary link to North Amer- 
ica or secondary radiation back to Eurasia. The 
latter hypothesis seems plausible in view of the 
fact that RS patterns of P. tracyi and P. hybrida 
are nearly identical. 
The distribution of the remaining subgeneric 
groups is predominantly New World or Ant- 
arctic. Only the Sylvestres, Nervosae, and pos- 
sibly Madropoa can be considered to have 
cpDNA types principally or entirely confined 
to North America. Dioicopoa s.s., Dasypoa, 
and the gynomonecious Punapoa of group IVC 
are principally South American, the Secundae 
occur in both North and South America, and 
Australopoa are endemic to New Zealand and 
Australia. 
The Sylvestres, so far as is known, are en- 
demic to the temperate deciduous forests of 
eastern North America. They appear to be bas- 
al in the cpDNA tree and could represent a 
group derived from a time when eastern North 
America was in close floristic contact with the 
temperate flora of Europe. However, Old World 
counterparts of the Sylvestres have not been 
identified. 
The Secundae are a small group in which I 
include P. curtifolia Scribner, P. napensis, P. 
secunda s.l., P. stenantha Trin., P. tenerrima, 
and P. unilateralis Scribner. Several forms of 
the facultative apomict, P. secunda s.l., are dis- 
junct to South America. Because the Secundae 
cpDNA type is relatively advanced and related 
to the Stenopoa group VE type (principally 
Palearctic), and the sexual members of the group 
and species diversity occur in North America, 
I postulate that they arose here and secondarily 
spread to South America. 
Group IVC species contain the most wide- 
spread cpDNA type in North America. This 
type is clearly derived within the genus and 
presumably marks the major radiation event 
among New World Poa. By extrapolation to 
related species I estimate that this type is pres- 
ent in about 50 (ca. 50%) North American and 
90 (ca. 90%) South American species. Within 
group IVC, only Dioicopoa s.s., and possibly 
Dasypoa, are well marked by cpDNA RS ad- 
vancement. 
Dioicopoa s.s. includes some 40 species, of 
which only one occurs north of the equator. 
The cpDNA type present in the three repre- 
sentatives sampled evidently is derived from 
the group IVC type. Poa arachnifera is endemic 
to the southern Great Plains where it displays 
6M, In, 8M, and 12M ploidy levels. As there was 
a polychotomous rather than a dichotomous 
relationship between this species and two South 
American representatives, and it is morpho- 
1398 AMERICAN JOURNAL OF BOTANY [Vol. 77 
logically nearly indistinguishable from two 
other lower polyploid (4?) species from South 
America, it seems probable that P. arachnif- 
era's occurrence in North America represents 
an amphitropical dispersal north as is sug- 
gested by Marsh (1952) rather than a relic dis- 
tribution of a stem or sister species involved 
in the radiation of Dioicopoa. 
One last geographical connection supports a 
floristic link between New World species of 
group IV and those of Australia/New Zealand. 
Poa sieberiana and P. labillardierei are rep- 
resentatives of the widespread cytotype in New 
Zealand (Hair, 1968). If the group IVC cpDNA 
type is endemic to the New World and most 
closely related to the circumpolar section Poa, 
then it is reasonable to postulate that the major 
radiation of Poa in the Holoantarctic Floristic 
Kingdom stemmed from dispersal from North 
America to South America to the Neozeylandic 
Region (as is suggested by Takhtajan, 1986) 
rather than from Asia. The presence of dioecy 
and rhizomes in some New Zealand species, 
morphologically close to Dioicopoa s.s., also 
supports this connection, as do crossing rela- 
tionships (discussed above). (Evidently there 
are Asian Poa connections as well [e.g., be- 
tween Poa drummondiana Nees of Australia 
and P. tuberifera Faurie ex Hack, of far East 
Asia], but they are minor constituents of the 
genus in Australia.) Poa sieberiana and P. la- 
billardierei retain a RS that was lost among the 
remainder of group IVC species tested. This 
RS may mark an alternative path of radiation, 
possibly from Southeast Asia, and raises the 
possibility that group IV Poa arrived in the 
New World from the antarctic region. How- 
ever, because not all group IVC species have 
been screened for this single RS, support for 
speculation is minimal. 
Chloroplast DNA RS patterns demonstrate 
that dispersal events from North to South 
America, in different groups of Poa, occurred 
at least twice; once by group V species, and 
one to several times by group IV species. That 
amphi-neotropical disjunctions have occurred 
more frequently than this is indicated by the 
number of disjunct species or species pairs: 
(group V) 1) P. glauca Vahl {Stenopoa), 2) P. 
secunda (three forms), and 3) P. stenantha Trin. 
(Secundae); (group IV), 4) P. arachnifera-P. 
lanuginosa Poir. {Dioicopoa), 5) P. douglasii 
Nees-i5. cummingii Nees (Madropoa). 6) Poa 
acinaciphylla Desv. (= P. villaroelei Phil.) (An- 
dinae or PunapoaT), and 7) P. chamaeclinos 
Pilger (Punapod), and 8) P, conglomerata-P. 
scaberula Hook. f. (Dasypoa), represent South 
American group IVC species that occur in North 
America only on Mexican volcanoes. Thus, the 
cpDNA cladogram yields a conservative es- 
timate of the number of such dispersal events. 
Reflections on the mode ofcpDNA radiation 
in Poa?The limited amount of variation in 
cpDNA RS in group IVC1 contrasts with the 
high degree of morphological and physiological 
differentiation in this group. Breeding system 
variation is extensive, including self-compat- 
ible and self-incompatible hermaphroditism, 
gynomoecism, gynodioecism, and dioecism. 
Longevity ranges from annual to long-lived 
perennial. The species have adapted to many 
habitats (e.g., temperate rain, coniferous and 
deciduous forests, riparian habitats, arid- 
steppe, Mediterranean coastal sand-dunes, 
warm-temperate grasslands, margins of warm 
deserts, and temperate and tropical alpine hab- 
itats). Laterally creeping rhizomes range from 
highly developed to absent. Sheaths range from 
closed to the collar to open nearly to the base. 
Blades range from broad, thin, and flat, to nar- 
row, thick, and involute. Panicles range from 
open and averaging 20 cm long with several 
hundred spikelets, to condensed and averaging 
1 cm long with fewer than ten spikelets. In fact, 
nearly all morphological and anatomical char- 
acter states found to vary in Poa vary within 
this group. 
It is worth noting that excluding the Eurasian 
section Homalopoa, P. occidentalis is the only 
diploid species known in group IVC. Of some 
70 species investigated, the majority are tet- 
raploid or have tetraploid races. 
There are several plausible explanations for 
the lack of hierarchical structure among cpDNA 
types within this diverse group: 1) There are 
species groups, but morphological and ecolog- 
ical evolution proceeded more rapidly so that 
few cpDNA RS markers appeared; 2) There 
are no groups; radiation proceeded rapidly and 
independently from one common ancestor; 3) 
There are groups but they were not exposed 
because closely related species were not sam- 
pled, RS were undersampled, or RS sampling 
was uneven among species; 4) Some equili- 
brating process is involved: Possibly common 
chloroplast genomes circulate through exten- 
sive hybridization and introgression along with 
some paternal leakage of chloroplasts (e.g., Se- 
cundae could have arisen from hybridization 
between species with an uncommon chloro- 
plast type and a more widespread type, by 
chance retaining the more common chloroplast 
type). Such a system may be operating in Quer- 
cus (Whittemore, unpublished data). In other 
words, the most common chloroplasts in a re- 
ticulating group are the ones most likely to end 
up in new hybrid taxa that may subsequently 
replace their predecessors. 
November 1990] SORENG?POA CHLOROPLAST DNA 1399 
Hypotheses supposing a lack of species groups 
among the samples may be rejected. There is 
good morphological evidence for derived 
groups of closely related species. For example: 
condensed-panicles, thick-involute leaf-blades 
with hairy upper surfaces, and reduced upper 
leaf-blades, are all synapomorphies for Mad- 
ropoa, of which P. confinis, P. cusickii, P.fend- 
leriana (two subspecies), and P. piperi were 
sampled, yet no RS synapomorphies (and few 
autapomorphies), were found. The fourth hy- 
pothesis, that of reticulation, is rejected as a 
primary process, in this case on grounds that 
reticulation does not explain the limited hi- 
erarchical variation between cpDNAs derived 
from diverse species of Europe, North Amer- 
ica, South America, and Australia. 
Undersampling of RS is always a possibility, 
yet several cases have been discovered wherein 
little or no synapomorphic cpDNA RS vari- 
ation has been detected among morphologi- 
cally diverse species (e.g., subsets within genera 
of trees of the Juglandaceae [Smith and Doyle, 
personal communication] and Arecaceae [Wil- 
son and Clegg, personal communication] and 
herbs such as Antennaria [Michaels, personal 
communication]). Morphological evidence in- 
dicates that the radiation of group IVC1 re- 
sulted in distinct groups but little structure 
among them. Whether, or to what degree, re- 
ticulation is responsible in part for the lack of 
hierarchical structure in the cladistic analysis 
of cpDNA RS in group IVC requires further 
investigation. 
Cladistic analysis of cpDNA RS demonstrat- 
ed the presence of significant phylogenetic 
structure within the genus and agreed fairly 
well with some traditionally postulated sec- 
tional level species groupings, but refuted oth- 
ers. In addition, the study provided informa- 
tion on putative parentage for further analyses 
of the origin of polyploid and apomictic com- 
plexes. Limited sampling of species in cpDNA 
group I?III (particularly of diploids), and a lack 
of samples from Southeast Asian groups, al- 
lows only tentative conclusions to be drawn 
about phylogenetic relationships among cer- 
tain basal species and groups within Poa. How- 
ever, what has been discovered provides a sub- 
stantial basis for revision of the subgeneric 
classification of the genus in the New World, 
and interpretation of character evolution and 
biogeographic events. 
LITERATURE CITED 
BOR, N. L.   1952.   The genus Poa L. in India. Part I. 
Journal of the Bombay Natural History Society 50: 
787-838. 
CLAUSEN, J.   1961.  Introgression facilitated by apomixis 
in polyploid poas. Euphytica 10: 87-94. 
CLAYTON, W. D., AND S. A. RENVOIZE. 1986. Genera 
graminum, grasses of the World. Kew Bulletin Addi- 
tional Series 13: 1-389. 
CROSS, R. A. 1980. Distribution of subfamilies of Gra- 
mineae in the Old World. Kew Bulletin 35: 279-289. 
DALLY, A. M., AND G. SECOND. 1989. Chloroplast DNA 
isolation from higher plants: an improved non-aque- 
ous method. Plant Molecular Biology Reporter!: 135? 
143. 
DEBRY, R. W.,ANDN. A. SLADE. 1985. Cladistic analysis 
of restriction endonuclease cleavage maps within a 
maximum-likelihood framework. Systematic Zoolo- 
gy 34: 21-34. 
DOYLE, J. J., J. L. DOYLE, AND A. H. D. BROWN. In press. 
Multiple origins of polyploids in the Glycine tabacina 
complex inferred from chloroplast DNA polymor- 
phism. Proceedings of the National Academy of Sci- 
ences (USA). 
EDMONDSON, J. R. 1978. Infrageneric taxa in European 
Poa L. Botanical Journal of the Linnaean Society 76: 
329-334. 
 .   1980.  Poa L. In T. G. Tutin, V. H. Heywood, 
N. A. Burges, D. M. Moore, D. H. Valentine, S. M. 
Walters, and D. A. Webb, [eds.], Flora Europea, vol. 
5, 159-167. Cambridge University Press, Cambridge, 
U.K. 
FARRIS, J. S. 1988. Hennig-86. ver. 1.5 [computer soft- 
ware and manual] (published by author). 
FELSENSTEIN, J. 1987. PHYLIP. ver. 3.0 [computer soft- 
ware and manual] (published by author). 
HAIR, J. B. 1968. Contributions to a chromosome atlas 
of the New Zealand flora. New Zealand Journal of 
Botany 6: 267-276. 
HARTLEY, W. 1961. Studies on the origin, evolution, and 
distribution of the Gramineae 4: the genus Poa. Aus- 
tralian Journal of Botany 9: 152-161. 
 . 1973. Studies on the origin, evolution, and dis- 
tribution of the Gramineae 5: the subfamily Festu- 
coideae. Australian Journal of Botany 21: 201-234. 
HIESEY, W. M., AND M. A. NOBS. 1982. Experimental 
studies on the nature of species VI. Interspecific hybrid 
derivatives between facultatively apomictic species of 
bluegrasses and their responses to contrasting envi- 
ronments. Carnegie Institution of Washington Pub- 
lication 636, Washington, D.C. 
HILU, K. W. 1988. Identification of the "A" genome of 
finger millet using chloroplast DNA. Genetics 118: 
163-167. 
KNOBLOCH, I. W. 1968. A checklist of crosses in the 
Gramineae. Privately published. 
MARSH, V. L. 1952. A taxonomic revision of the genus 
Poa of the United States and Canada. American Mid- 
land Naturalist 47: 202-250. 
NICORA, E. G. 1978. Gramineae. In M. N Correa [ed.], 
Flora Patagonia, part 3. Instituto Nacional de Tec- 
nologia Agropecuria, Buenos Aires. 
PALMER, J. D. 1987. Chloroplast DNA evolution and 
biosystematic uses of chloroplast DNA variation. 
American Naturalist 30: S6-S29. 
 , G. P. SHIELDS, D. B. COHEN, AND T. J. ORTEN. 
1983. Chloroplast DNA evolution and the origin of 
amphidiploid Brassica species. Theoretical and Ap- 
plied Genetics 65: 181-189. 
REED, K. C, AND D. A. MANN. 1985. Rapid transfer of 
DNA from agarose gels to nylon membranes. Nucleic 
Acids Research 13: 7207-7221. 
SALTZ, Y., AND J. BECKMAN. 1981. Chloroplast DNA 
preparation for Petunia and Nicotiana. Plant Molec- 
ular Biology Newsletter 2: 73-74. 
1400 AMERICAN JOURNAL OF BOTANY [Vol. 77 
SORENG, R. J. 1985. Poa in New Mexico, with a key to 
middle and southern Rocky Mountain species (Po- 
aceae). Great Basin Naturalist 45: 395-422. 
 .  1986. Distribution and evolutionary significance 
of apomixis in diclinous Poa of Western North Amer- 
ica. Ph.D. dissertation, New Mexico State University. 
Las Cruces. 
 , J. I DAVIS, AND J. J. DOYLE.  In press.  A phy- 
logenetic analysis of chloroplast DNA restriction site 
variation in Poaceae subfamily Pooideae. Plant Sys- 
tematics and Evolution. 
 , AND S. L. HATCH.   1983. A comparison of Poa 
tracyi and Poa occidentalis. Sida 10: 123-127. 
SOUTHERN, W. M. 1975. Detection of specific sequences 
among DNA fragments separated by gel electropho- 
resis. Journal of Molecular Biology 98: 503-517. 
STEBBINS, G. L. 1950. Variation and evolution in plants. 
Columbia University Press, New York. 
TAKHTAJAN, A. L. 1986. Floristic regions of the world. 
University of California Press, Berkeley. 
THOMAS, K., B. J. WOOD, C. L. BASSET, AND J. R. Y. 
RAWSON. 1984. A restriction endonuclease map of 
the chloroplast genome of pearl millet. Current Ge- 
netics 8: 291-297. 
TUTIN, T. G. 1957. A contribution to the experimental 
taxonomy of Poa annua L. Watsonia 4: 1-10. 
TZVELEV, N. N. 1983. [English translation of 1976 Rus- 
sian ed.]. New Delhi: for Smithsonian Institution by 
Amerind Publishing Company. 
VANDUK, G. E., AND G. D. WINKELHORST. 1982. Intra- 
specific crosses as a tool in breeding Poa pratensis L., 
1 Poa longifolia Trin. X P. pratensis. Euphytica 31: 
215-223.