.rt. 1’; . a ‘ » ,
1‘ I Llu. .v...|. ‘ .
.t. .uflo. . KMALWH . . . us .1” u in. It ‘ .
. . , : , ‘ , ‘ I;
, ‘ . . L544

Ex?

I?)
~ I. .\ til.

.. Tin:
. .3. ti}
. .

3h!
:I.Z 3. . .0l. '1
9.0:»...rI-lv‘
A.

 

L. In I P
”00"." JLIV-‘l
1n.

 

I‘ 19):; ..
(E‘ .1.

 

b|. .I’L n. . lin.n.
‘l .5'05I‘A. .s...vr.l. Q‘
. I! . f . . .

 

.. . 3: f .
. 1. .I. .‘il

# . ’ .. , , . ‘
.dq..C\..u..3.............. 7...
‘. .. qutv. -‘ i u 5. V . Hal-1 ‘ | l .uh'u 13') a .

...§.... .flflvfi. a. roar... .} p

.oh.

 

 

{Hm

lll’iilflfiiil WI]Ifififlliliffliiflliififlilfiiflfil

3 1293 009101936

This is to certify that the

dissertation entitled

GENETIC DIVERSITY AND EVOLUTIONARY RELATIONSHIPS

WITHIN AND AMONG FIVE CYPRIPEDIUM (ORCHIDACEAE) SPECIES:

 

INFERENCE FROM ALLOZYME ELECTROPHORESIS
presented by

Martha Ann Case
has been accepted towards fulfillment
of the requirements for

Ph- D. degree in Botany and
Plant Pathology

 

/f/& 27%

Major professor
Dr. John H. Beaman

Date /{fbév~o7 /77:

MS U i: an Affirmative Action/Equal Opportunity Institution 0- 12771

 

 

 

_ ,fi—_-__ __ ._ +_ ——~—

 

 

LIBRARY
Michigan State
; University

L_

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

‘1

 

‘L J
.—v m “JT’I‘.‘
e Iiigji‘ p{ .
. ‘L :C“ "r:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

wit—i.

 

MSU Is An Affitmmive Action/Equal Opportunity Institution

emana-pd

 

 

GENETIC DIVERSITY AND EVOLUTIONARY RELATIONSHIPS
WITHIN AND AMONG FIVE CYPRIPEDIUM (ORCHIDACEAE) SPECIES:
INFERENCE FROM ALLOZYME ELECTROPHORESIS
BY

Martha Ann Case

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Botany and Plant Pathology

1993

ABSTRACT
GENETIC DIVERSITY AND EVOLUTIONARY RELATIONSHIPS
WITHIN AND AMONG FIVE CYPRIPEDIUM (ORCHIDACEAE) SPECIES:
INFERENCE FROM ALLOZYME ELECTROPHORESIS
BY
Martha Ann Case

The Orchidaceae are one of the most morphologically
diverse and species-rich families of flowering plants.
Although tremendous morphological diversity exists within
the family, very little is known about the extent of genetic
variation within natural populations of orchids, or the
degree of genetic differentiation among congeneric species.
This study is one of the first examinations of allozyme
variation in the Orchidaceae. It addresses current theories
of orchid evolution concerning the levels of genetic
diversity and taxonomic relationships within and among
Qypripedium candidum, Q. acaule, Q. reginae, Q. arietinum,
and three North American varieties of Q. calceolus.

Starch gel electrophoresis resolved the products of 14
isozyme loci which were used to quantify the degree of
allozyme divergence within and among these Qypripedium taxa.
Qypripedium calceolus and Q. candidum have a high average
genetic identity (Nei's genetic identity = 0.794) and appear
to be a progenitor-derived species pair whereas all other
species pairs have very low genetic identities ranging from

0.000 to 0.285. These low identity values are among the

lowest reported in the plant literature, and they indicate
that the level of genetic divergence among Cypripedium
species is substantially greater than most other congeneric
plant species. Qypripedium arietinum, which is sometimes
placed in a segregate genus Criosanthes, was more similar to
Q. calceolus and Q. candidum than were Q. reginae or Q.
acaule. This result indicates that Q. arietinum should be
retained in the genus Qypripedium. In contrast, varieties
within Q. calceolus show relatively little genetic
divergence at isozyme loci. Intravarietal Nei's identity
values ranged from 0.920 to 0.979, and they were not
consistently higher than intervarietal values. The genetic
data for these Q. calceolus varieties are consistent with
reports of genetic divergence among other sympatric
infraspecific taxa and do not support current hypotheses
that taxa within Q. calceolus are distinct species.

The species also show considerable differences in Nei's
diversity statistics. Average expected species-level
heterozygosities were 0.244 (Q. calceolus), 0.095 (Q.
agaule), 0.054 (Q. gandigum), 0.037 (Q. reginae), and 0.000
(Q. arietinum) with 19%, 16%, 7%, 35%, and 0% of the
variation partitioned among populations, respectively. The
relatively high level of heterozygosity in Q. calceolus is
also consistently high among each of its varieties. These
results are discussed in the context of current theories on
genetic variation in orchid populations and historical

events which could have influenced the levels of variation.

Dedicated to the loving memory of my father
Richard James Case

1921 - 1987

iv

ACKNOWLEDGMENTS

I am deeply grateful to my family, friends, and colleagues
who have contributed to this research project in ways too
numerous to elaborate. My parents, Rose and Richard Case,
were a continuous source of encouragement and support as
well as an extremely reliable source of Qypripedium
population information. During the early stages of this
project, it became part of their Sunday routine to explore
for new orchid populations in northern Michigan as well as
keep me informed on the status of my research populations
there. Fred and Roberta Case were also extremely helpful in
providing population location information and for allowing
me to obtain leaf samples from their own personal research
collection. Some of the most interesting populations in
this study would not have been discovered without their
expertise.

Numerous other people and organizations were very helpful
in providing access to populations, specimen loans, or
population information. These include The Nature
Conservancy, the Michigan Natural Features Inventory, the
state and federal park systems of the US and Canada, various
herbaria (AMES, BGSU, CAN, F, MICH, MO, MSC, MTMG, NFLD, NY,

OS, TRT, US, WIS), the many landowners who granted me

permission to sample populations on their land, and all
those who wrote or met with me to share their information.

I would especially like to thank Sigma Xi (both the

national and local organizations), the American Orchid
Society, and the Michigan Nongame Wildlife Research Fund for
providing grants which made this project possible.

Lastly, I am very grateful to my major professor, Dr. John
H. Beaman, and the other members of my committee, Drs. James
H. Asher, Karen Klomparens, and Stephen J. Tonsor for
providing inspiration, guidance, and critical expertise
throughout this project. In addition, Dr. Daniel J.
Crawford provided essential advice for me in the skill of
isozyme electrophoresis. His suggestions on electrophoretic
technique were invaluable for my project. This study also
benefitted greatly from numerous colleagues and friends whom
I have had the pleasure to know at Michigan State. I
especially thank Todd J. Bierbaum, Richard K. Rabeler,
Lucile McCook, Nick Stoynoff, and Jim and Beth Wee for
stimulating conversation, inspiration, critical comments,

and occasionally comic relief.

vi

TABLE OF CONTENTS

Page
LIST OF TABLES ......................................... viii
LIST OF FIGURES ........................................... x
GENERAL INTRODUCTION ...................................... 1
PART 1 EXTENSIVE VARIATION IN THE LEVELS OF GENETIC
DIVERSITY AND DEGREE OF RELATEDNESS AMONG
FIVE SPECIES Of CYPRIPEDIUM ..................... 3
Abstract .............................................. 3
Introduction .......................................... 5
Materials and Methods ................................ 10
Results .............................................. 19
Discussion ........................................... 29
Interspecific relationships ....................... 29
Intraspecific diversity ........................... 34
Literature cited ..................................... 51
PART 2 HIGH LEVELS OF ALLOZYME VARIATION WITHIN
CYPRIPEDIUM CALCEOLUS AND LOW LEVELS OF
DIVERGENCE AMONG ITS VARIETIES ................. 62
Abstract ............................................. 62
Introduction ......................................... 64
Materials and Methods ................................ 67
Results .............................................. 75
Discussion ........................................... 88
Literature cited ..................................... 99

vii

Table

LIST OF TABLES

Page

Location data, population collection numbers
(PCN), and numbers of individuals sampled (n)
for Qypripedium species analyzed .................... 11

Species-level allele frequencies across 14 loci
and total individuals sampled per taxon (n) ......... 20

Average Nei's identity values and ranges for all
intraspecific (along the diagonal) and inter-

specific (below the diagonal) comparisons among

taxa, and the frequency of loci in which no

alleles are shared (above the diagonal) ............. 22

Frequency of polymorphic loci (P), alleles per

locus (A), and effective alleles per locus (Ae)

at the population (p) and species (8) levels,

and the frequency of loci (Ua) which contain

unique alleles with frequencies greater than

0.25 for each taxon ................................. 25

Gene diversity statistics for polymorphic

Qypripgdium species ................................. 27

Significant (p < 0.05, 0.01 or 0.001) and near
significant (0.05 < p < 0.10) single-locus

deviations from Hardy-Weinberg equilibria and

the level of negative or positive disequilibria
expressed by the fixation index ..................... 28

Location information, number of individuals
sampled, and reference letter for populations
of C. calceolus var. pubescens (PUB), Q. calceolus

var. parviflorum (PARV), Q. calceolus var.
planipetalum (PLAN), and populations containing
putative hybrids or more than one morphological

entity .............................................. 68

Diversity measures for C. calceolus ................. 76

Status of shared alleles among varieties and
unique alleles within varieties of Qypripedium
calceolus ........................................... 78

viii

Table

LIST OF TABLES (continued)

Nei's genetic identity means and ranges for
intra- and intervarietal comparisons of

Q. calceolus .....................................

Allelic correlations with the first, second, and
third principal components, and the percent of
variation accounted for by each component

(percent trace) ..................................

Results from a chi-square contingency table
analysis for allele frequency heterogeneity
among varieties parviflorum (population P) and
pubescens (population Q) at a sympatric site on

Drummond Island ..................................

Allele-frequency means for polymorphic loci
(weighted according to population sizes),
population ranges, and number of individuals

sampled for varieties of Q. calceolus ............

ix

Page

...80

...86

...87

...92

LIST OF FIGURES

Figure Page

1.1. UPGMA dendrogram of Nei's (1978) genetic
identities for Q. calceolus (CALC), Q.
candidum (CAND), Q. arietinum (ARIE), Q.
acaule (ACAU), and Q. reginae (REGI) ............... 24

1.2. Average congeneric Nei’s identity values for
intercontinental species pairs within the genera
Agastache (AGA), Liguidambar (LIQ), Liriodendron
(LIR), and Datisga (DAT) compared to the average
congeneric identity values reported for plant
species in general (AV) and for Qypripedium
species in this study (CYP) ........................ 31

1.3. The frequency of expected heterozygotes, alleles
per locus, and the frequency of polymorphic loci
at the species level for Q. calceolus (CALC), Q.
candidum (CAND), Q. acaule (ACAU), Q. reginae
(REGI), and Q. arietinum (ARIE) compared to the
average species-level diversity measures
reported for plants in general (AV) ................ 35

2.1. UPGMA phenogram of Nei’s genetic identity values
for all pairwise population comparisons of pure
var. pubescens (PUB), var. planipetalum (PLAN),
var. parviflorum (PARV), and mixed populations
(indicated by two—taxon abbreviations separated
by a slash) ........................................ 81

2.2. Pure populations of vars. pubescens (triangles),
parviflorum (circles), and planipetalum (squares)
plotted in relationship to the first three
principal component axes (PCAl, PCA2, and PCA3)
of a principal components analysis of allele
frequencies ........................................ 84

GENERAL INTRODUCTION

The results from an electrophoretic analysis of seven
Qypripedium taxa are reported in two major subdivisions of
this dissertation, Part 1 and Part 2. Each part is a
complete paper with a separate abstract, introduction,
materials and methods, results, discussion, and literature
cited. Part 1 describes the extent of genetic divergence
among Q. calceolus, Q. candidum, Q. acaule, Q. reginae, and
Q. arietinum, particularly with respect to the generic
placement of Q. arietinum. This paper also quantifies the
amount of genetic variation found within Qypripedium
populations and also analyzes the distribution of genetic
variation among conspecific populations. These data are
discussed in the context of: (1) current theories regarding
the level of genetic variation in orchid populations, (2)
comparable isozyme studies reported for other plant taxa,
and (3) historical events that may have influenced the
observed genetic patterns.

Part 2 focuses on the problem of the taxonomic status of
Qypripedium calceolus var. pubescens, Q. calceolus var.
parviflorum, and Q. calceolus var. planipetalum. It
examines: (1) the degree of genetic divergence among these

taxa, (2) the distribution of genetic variation among

2
populations within each taxon, and (3) the allele frequency
differences among sympatric and allopatric populations of
vars. parviflorum and pubescens. Several methods of
analyses were employed including chi—square contingency
analysis, unweighted pair-group method analysis, and
principal components analysis. The results are discussed
with respect to other allozyme studies of sympatric and
allopatric infraspecific taxa reported in the plant

literature.

PART 1

EXTENSIVE VARIATION IN THE LEVELS OF
GENETIC DIVERSITY AND DEGREE OF RELATEDNESS AMONG

FIVE SPECIES OF CYPRIPEDIUM

ABSTRACT

An electrophoretic analysis of 14 enzyme loci was
conducted in order to describe the genetic variation
contained within and among Qypripedium calceolus L., Q.
candidum Muhlenberg ex Willdenow, Q. acaule Aiton, Q.
reginae Walter, and Q. arietinum R. Brown in Aiton
(Orchidaceae). While Q. calgeolus and Q. candidum have a
high average genetic identity (Nei's genetic identity =
0.794) and appear to be a progenitor-derived species pair,
all other species pairs have very low genetic identities
ranging from 0.000 to 0.285 which indicates extensive
phylogenetic divergence. The species also show considerable
differences in Nei's diversity statistics. Average expected
species—level heterozygosities were 0.244 (Q. calceolus),
0.095 (Q. acaule), 0.054 (Q. candidum), 0.037 (Q. reginae),
and 0.000 (Q. arietinum) with 35% to 0% of the variation
partitioned among populations. The low genetic variation

and uniformity of populations present in several taxa could

4
have resulted from prior genetic bottlenecks which affected
the ancestral populations. The results are discussed in the
context of historical events that could have influenced the

observed genetic patterns.

5
INTRODUCTION

The Orchidaceae are one of the most morphologically
diverse and species-rich families of flowering plants
(Dressler, 1981). Despite the family’s tremendous
diversity, very little is known about the level and
maintenance of genetic diversity within natural populations,
or the extent to which congeneric species differ at the
allozyme level. This scarcity of electrophoretic data on
the Orchidaceae may, to some extent, reflect the
difficulties in working with orchids. These include their
oftentimes sporadic occurrences (Case, 1987) in relatively
small populations, and the difficulties in rearing many
temperate orchids due to their long generation times
(Curtis, 1943) and fastidious germination and nutrient
requirements (Ballard, 1987; Stoutamire, 1990). Yet the
unique combination of life-history and ecological
characteristics which contribute to the difficulty in
working on orchids also make them an especially interesting
family to examine electrophoretically. For example, as a
result of one successful pollination, some species of
orchids can produce over a million dust-like, wind-dispersed
seeds (Withner, 1988). This potential for long-distance
dispersal in the family is thought to facilitate geographic
isolation and, in combination with selection pressure to
accommodate new pollinators, promote the extraordinary
levels of speciation (Dressler, 1981). Isolated populations

founded by one or a few individuals may also contain reduced

6
levels of genetic variation, especially if the population
growth after the bottleneck remains slow (Nei, Maruyama, and
Chakraborty, 1975). According to Gill (1989), the entire
process leading to orchid population establishment may be
viewed as a multiplicative series of improbable events.
This has led to speculation that orchid populations should
contain exceedingly low levels of genetic variation (Gill,
1989).

This study examines the genetic diversity within and among
five taxa of the genus Cypripedium, a terrestrial genus
containing 30 to 40 species of long-lived, herbaceous
perennials (Luer, 1975). The genus Qypripedium, along with
three other genera, comprise the subfamily Cypripedioideae,
commonly known as the Slipper Orchids. This subfamily
differs from the vast majority of orchids by having two
fertile lateral anthers whereas the other major subfamilies
contain one fertile median anther. This characteristic,
along with other primitive characteristics in the subfamily,
suggest that the lineage which has given rise to the
Cypripedioideae diverged early in the evolution of the
family (Correll, 1950; Dressler, 1981; Atwood, 1984). As
typical members of the Cypripedioideae, the taxa in the
present study have an inflated pouch-like labellum which
insects enter through an opening in the top and become
temporarily trapped inside. They pollinate the flower by
first passing under the stigmatic region and then under one

of two lateral anthers, each located near an exit hole in

7
the rear portion of the labellum. Self-pollination is
generally thought to be improbable due to a physical
separation of the reproductive structures (Luer, 1975).
Furthermore, Nilsson (1979) observed that insect behavior
which would lead to self-pollination in Q. calceolus occurs
infrequently. Therefore, it has been assumed that
Qypripedium taxa routinely outcross, although geitonogamy
may occur through the occasional production and cross
pollination of two or three flowers on the same peduncle or
from the cross-pollination of flowering shoots produced
through clonal reproduction. In an examination of
self-fertility in three Qypripedium taxa, Newhouse (1976)
found that Q. calceolus, Q. reginae, and Q. acaule produced
fruit with apparently normal embryos after artificial
self-pollination, and none of the taxa produced seeds after
bagged flowers were emasculated.

Cypripedium taxa range from arctic areas to the subtropics
of the Northern Hemisphere. Twelve species occur in North
America, five of which are found predominantly or entirely
within the northeastern United States and Canada (Luer,
1975). These five taxa are the subject of this paper and
include Qypripedium calceolus L. (Yellow Lady's Slipper), Q.
candidum Muhlenberg ex Willdenow (White Lady's Slipper), Q.
arietinum R. Brown in Aiton (Ram's Head), Q. acaule Aiton
(Moccasin Flower), and Q. reginae Walter (Showy Pink Lady's
Slipper). In addition, results from a preliminary analysis

of Q. kentuckiense C. F. Reed are reported. This taxon is

8
known primarily from E. Kentucky to Louisiana and Arkansas
(Atwood, 1985), and was only recently described by Reed
(1981). Because relatively little is known about Q.
kentuckiense, the following paragraphs do not pertain to
this species unless specifically mentioned. The initial
electrophoretic results for this taxon are included in the
discussion.

In North America, Qypripedium calceolus occupies the
largest range of any Qypripedium taxon. Its range extends
from Newfoundland to the Yukon and, in the eastern half of
the United States, as far south as Louisiana. It also can
be found in the greatest diversity of habitats including
northern cedar—fir fens, limestone barrens, mixed deciduous
forests, and roadside ditches. Similar to the ecological
diversity, the morphological diversity in this taxon is
likewise extensive and its relationship to species
delimitation has been the subject of numerous papers and
long-standing taxonomic debate (Sheviak, 1983). Due to this
morphological diversity, three North American varieties have
been recognized, vars. pubescens (Willdenow) Correll,
parviflorum (Salisbury) Fernald, and planipetalum (Fernald)
Victorin & Rousseau. Although populations of each of these
varieties have been included in the Q. calceolus populations
analyzed in the present study, the genetic differences among
these varieties will be the subject of a subsequent paper.
Among the five taxa discussed in this paper, C. acaule has

the second largest geographic range, which extends from

9
Newfoundland to the Great Bear Lake in the Northwest
Territories and down to northeast Alabama. All other taxa
occur predominantly east of North Dakota and occupy less
total geographic area (Luer, 1975). Because of extensive
geographic regions of sympatry and ecological requirements
which can be similar, these taxa can occasionally be found
in sympatric populations but only Q. calceolus and Q.
candidum are known to produce natural hybrids. When this
occurs, large hybrid swarms may develop which can be
accompanied by extensive introgression (Actor, 1984).
Qypripedium candidum and Q. calceolus are also most similar

morphologically whereas C.

acaule, Q. reginae, and Q.
arietinum are morphologically very distinct. They all have
a chromosome number of 2n = 20 (Love and Ritchie, 1966; Love
and Simon, 1968).

In a recent morphological analysis of evolutionary trends

in the Cypripediodeae, Q. acaule was considered to be

divergent from C.

candidum, Q. calceolus, and Q. reginae,
and it was therefore placed in a different clade (Atwood,
1984). Qypripedium arietinum was also placed in a separate
clade primarily because of three characteristics shared only
by Q. arietinum and Q. plectrochilon Franch. (a very similar
Asiatic taxon whose species distinction is questionable;
Atwood, 1984). These characteristics are: (1) separate
lateral sepals, (2) a spurred lip, and (3) a staminode which

strongly resembles a fertile stamen. Based on a numerical

analysis which incorporated these characters, Atwood (1984)

10
found Q. arietinum (and therefore also Q. plectrochilon) to
be sufficiently dissimilar from other Qypripedium species
that he proposed a transfer of these taxa to the segregate
genus Criosanthes.

This paper examines the degree of allozyme divergence
among Q. arietinum and the four other Cypripedium species
mentioned above. It also addresses the level of
intraspecific variation in these taxa, how this variation is
partitioned among populations, and how members of the
Orchidaceae that have been examined electrophoretically
compare to plants in other families with similar life

history and ecological characteristics.

MATERIALS AND METHODS

Thirty one populations representing five taxa were
analyzed by enzyme electrophoresis (Table 1.1). Since the
Great Lakes region is particularly suitable for the growth
of Qypripedium species (Case, 1987), most of the populations
were collected from this region in a wide variety of
habitats. For Q. calceolus, however, collections were made
from Newfoundland, Michigan, Indiana, Ohio, and Illinois in
order to encompass a similar proportion of the total species
range and morphological variation. Numbers of individuals
sampled per population ranged from 12 to 88 and in the
majority of cases, represented 25% to 100% of the estimated

population size observed. Samples of fewer than 20

individuals represent a sample from every individual in the

TABLE 1.1.

11

Location data, population collection numbers

and numbers of individuals sampled (n) for

Cypripedium species analyzed.

 

 

PCN n Location

Q. calceolus

301 20 Scioto Co., OH; 38°46’N 83°14’W, Shawnee
State Forest

B03 20 Defiance Co., OH; T4N RSE 823, Defiance

B06 20 Vermillion Co., IL; T20N R12W SZO,
Danville

BO? 20 Charlevoix CO., MI; T34N R7W 82, Bay
Shore

B08 20 Lake Co., IN; T37N R9W, Hammonda

B09 20 Charlevoix Co., MI; T34N R7W 811, Bay
Shore

B17 20 Baraga Co., MI; T50N R33W 821, Bovin

L01 18 Newfoundland, Canada; 19.3 km inland
from Eddie's Cove along highway 430

L02 20 Newfoundland, Canada; 16.1 km inland
from Eddie's Cove along highway 430

L03 20 Newfoundland, Canada; 3.2 km N of St.
Barbe Bay along highway 430

L06 20 Newfoundland, Canada; along Green Garden
Trail, seprentine area, Gros Morn
National Park

V01 88 Shiawassee Co., MI; T5N RlE 820, Rose
Lake Wildlife Research Area

V02 22 Presque Isle Co., MI; T33N R2E 818,
Canada Creek

V03 47 Livingston Co., MI; T2N R5Ea

V04 50 Livingston Co., MI; T2N R6Ea

Q. candidum
C01

C02

C03

C04

C05

12

20

20

15

40

Q. arietinum

R01

R02

R03

R04

Q. acaule
A01

A02

A03

A04

20

29

42

24

50

19

20

45

12

Table 1.1 (cont'd)

Kalamazoo Co., MI; T4S R11W S4, Hogset
Lake

Oakland Co., MI; T4N R8E 816, Davisburg

Jackson Co., MI; T4S R3W S9, Brail Lake
Creek

Livingston Co., MI; T4N R4E $11 $12,
Cohoctah

Kalamazoo Co., MI; T13 R9W S30,
Butterfield Lake

Emmet Co., MI; T39 R4W $30, Cecil Bay

Cheboygan Co., MI; T39N R3W $18 819,
Mackinaw City

Presque Isle Co., MI; T34N R7E 814,
Thompson’s Harbor

Leelanau Co., MI; Sleeping Bear Dunes
National Lakeshore8

Emmet Co., MI; T35N R5W $25, Conway

Chippewa Co., MI T51N R5W S32, Whitefish
Point

Washtenaw Co., MI; T2S R3E S4, Waterloo
Recreation Area

Presque Isle Co., MI; T33N R2E 818,
Canada Creek

13

Table 1.1 (cont’d)

Q. reginae

801 55 Clare Co., T17N R4W8

$02 19 Clinton Co., MI; T5N RIW 825, Rose Lake
Wildlife Research Area

803 23 Ingham Co., MI; T4N RlE 811, Coon Creek

 

aExact locations witheld upon request of the proprietor

14
population.

Each population was first surveyed for approximate size
and density of individuals and then tissue samples were
collected from a representative proportion of flowering and
non-flowering individuals. Care was taken to collect
uniformly throughout the populations and also to avoid
collecting from more than one ramet in a clump of leafy
stalks suspected of being clonally reproduced. In addition
to collecting plant material from naturally-occurring
Qypripedium populations, leaf samples were also collected
from three Q. kentuckiense individuals cultivated by
Frederick and Roberta Case in Saginaw, Michigan. Each plant
had originated from a different state, Kentucky, Alabama,
and Arkansas (F. Case, personal communication). For each
sampled plant, one leaf tip (ca. 3 cm long) was removed and
then divided into two equal portions which were each placed
in a 1.5 ml centrifuge tube. All samples were kept on water
ice in the field and then stored at -80 C in the lab until
they were electrophoresed.

The leaf tissue was ground with a chilled mortar and
pestle and an extraction buffer consisting of 0.1 M
Tris-HCl, pH 7.5, 14 mM 2-mercaptoethanol, 1 mM EDTA
(tetrasodium), 10 mM KCl, and 10mM MgC'l2 (Gottlieb, 1981a).
In addition, 10 mg of polyvinylpyrrolidone-lo was added to
the tissue at the time of grinding. The resulting slurry
was absorbed onto filter paper wicks made from Whatman #17

paper (for buffer systems I and II described below) and

15

Whatman #3 paper (for buffer system III described below).

Ten enzyme systems were employed on 12% starch gels using
three buffer systems. System I electrode buffer consisted
of 0.029 M monohydrate lithium hydroxide with 0.192 M boric
acid, pH 8.2, whereas the gel buffer consisted of one part
electrode buffer and nine parts 0.051 M Tris with 0.007 M
monohydrate citric acid, pH 8.4 (Crawford, 1982). This
system resolved glutamate oxaloacetate transaminase (GOT),
triose-phosphate isomerase (TPI), alcohol dehydrogenase
(ADH), glutamate dehydrogenase (GDH), and superoxide
dismutase (SOD). System II resolved phosphoglucomutase
(PGM). The electrode buffer of this system consisted of
0.060 M sodium hydroxide with 0.300 M boric acid, pH 8.2,
and the gel buffer consisted of 0.076 M Tris with 0.005 M
monohydrate citric acid, pH 8.7 (Crawford, 1982). Malate
dehydrogenase (MDH), phosphogluconate dehydrogenase (PGD),
isocitrate dehydrogenase (IDH), and shikimate dehydrogenase
(SKD) were resolved on a histidine system (system III) which
consisted of 0.400 M citric acid trisodium salt, pH 7.0, for
the electrode buffer, and 0.020 M histidine-HCl, pH 7.0, for
the gel buffer (Gottlieb, 1981a). Electrophoresis was
monitored by the position of a bromophenol blue marker band
and was stopped when the dye had migrated 10.5 cm across the
gel. Total time of electrophoresis varied from five to six
hours for systems I and II, to eight to nine hours for
system III. Agarose overlays, with minor modifications,

followed the procedures of Soltis et a1. (1983) and were

16
employed for all enzyme systems except GOT. This was a
stain bath which followed the protocol of Crawford (1982).
The preceding buffer and stain systems employed represent
the optimal three out of approximately ten systems tested
for enzyme separation, resolution and activity for the
proteins assayed.

Genotypes were inferred directly from enzyme phenotypes by
knowledge of the minimum number of isozymes present for
given enzymes together with the knowledge of the quaternary
structure of those enzymes (Gottlieb, 1981b, 1982; Weeden
and Wendel, 1989). Enzymes which did not conform to the
above expectations were dropped from the analyses. In cases
where it was difficult to distinguish allozymes from
isozymes, side by side comparison of leaf and pollen
electromorphs were conducted following the recommendations
of Weeden and Gottlieb (1979). Putative isozyme loci were
given sequentially higher numbers for decreasing anodal
migration of their protein products. Likewise, allele a was
assigned to the most anodally migrating protein, b to the
protein immediately following a, etc. Interpopulation
verification of enzyme mobilities was conducted through side
by side comparisons of allelic variants electrophoresed
under identical conditions (i.e., on the same gel and with
the same batch of extraction buffer).

Allele frequencies were calculated at the population levels
and weighted according to sample size to produce

species—level allele frequencies. From the populational

17
frequencies, Nei’s (1978) genetic identity values were
calculated for all pairwise population combinations using a
modified version of GENESTAT (Whitkus, 1985). Identity
values for each interspecific and intraspecific comparison
were then averaged from the pairwise population values. The
resulting average interspecific identity values were used to
produce a UPGMA dendrogram following the procedure of
Swofford and Olsen (1990).

For each population and species, the frequency of
polymorphic loci (Pp and Ps, respectively), alleles per
locus (Ap and As), and effective alleles per locus (Aep and
Ass) were calculated using the procedure reported by Hamrick
and Godt (1989). Within each species, the gene diversity
statistics of Nei (1973), unbiased for sample size (Nei and
Chesser, 1983), were calculated for all loci and averaged to
give species estimates. This was calculated with the
modified GENESTAT program cited above. The values reported
here include total genetic diversity or species-level
expected heterozygosity (Ht), average gene diversity within
populations or average population-level expected
heterozygosity (Hs), average gene diversity between
populations (Dst), and the proportion of total genetic
diversity attributable to among-population differentiation
(Gst). Standard errors among loci were calculated for Ht,
Hs, Dst, and Gst. For Gst, the jackknife method described
in Weir (1990) was employed. To determine whether or not

populations deviated significantly from Hardy-Weinberg

18
expectations, chi-square tests were performed for
each polymorphic locus within each population. For loci
with two alleles, Yates continuity correction was applied
(Weir, 1990). When expectations for any genotypic class in
any locus fell below one, the least common allele was
collapsed into the adjacent allelic class until all
genotypic classes had expectations greater than one. This
procedure helped eliminate spurious inflation of the
chi-square test statistic due to small numbers in the
denominator (Hernandez and Weir, 1989). For loci with
greater than two alleles present, each heterozygote class
was tested individually by comparing a squared estimated
disequilibrium coefficient to an estimate of its variance
(Hernandez and Weir, 1989). Critical values for the desired
alpha in this case were adjusted to account for correlated
comparisons (Sokal and Rohlf, 1981). Significant (p < 0.05)
and near significant (0.05 < p < 0.10) deviations are
reported as well as the level of disequilibrium which is
expressed by the fixation index, F = 1 - (H/2pq), where H is
the frequency of heterozygotes and p and q are the allele
frequencies in the population (Hedrick, 1985). Gene flow
estimates based upon the frequency of private alleles within
populations were calculated by the method of Slatkin (1985)

using a correction for sample size.

1 9
RESULTS

Ten enzyme systems clearly resolved the products of 14
putative loci used in this analysis. For all species
analyzed, one isozyme was observed for ADH, SKD, GDH and SOD
(except in Q. reginae in which two regions of staining were
observed for SOD). Two isozymes were observed for TPI, PGM,
IDH, and PGD, however, PGD-l and IDH-l were not used due to
low activity. Three isozymes were present for GOT. MDH had
a clearly-defined slow locus (polymorphic in Q. calceolus
populations not included in this analysis) and several other
clearly resolved but genetically equivocal faster
electromorphs which were not used. Except for SOD and MDH,
the number of isozymes and regions of staining were
consistent across all taxa, but the resolution and activity
of five enzymes varied greatly among species. Therefore,
these loci had to be omitted from certain interspecific
comparisons (see MDH-3, PGM-2, GOT-3, ADH, and GDH in Table
1.2).

Nei’s average genetic identity values for intraspecific
population comparisons ranged from 0.937 in Q. calceolus to
1.000 in Q. arietinum (Table 1.3). For interspecific
comparisons, the highest average identity occurred for the
Q. calceolus-Q. candidum comparison with an average identity
of 0.794. This is in dramatic contrast to all other
interspecific comparisons which have values of 0.000 for the
Q. arietinum and Q. reginae pair to the second highest value

of 0.285 for Q. calceolus and Q. arietinum. A visualization

20
TABLE 1.2. Species-level allele frequencies across 14

loci? and total individuals sampled per taxon (n).

 

Qypripedium.species

CALC CAND ARIE ACAU REGI
Allele n a 425 n a 107 n a 115 n a 134 n a 97

 

 

TPIla - 0.028 - - —
b 0.344 0.967 — - —

c - - - - 1.000
d - - 1.000 0.974 -
e 0.656 - - 0.026 -
f - 0.005 - - -
TPI2a - - — 0.063 -
b - - - 0.015 -
c - - 1.000 - -

d - — — - 1.000
e 0.748 1.000 — 0.918 -
f 0.252 — - - -
g - - - 0.004 —
SODa - - - 1 000 -
b - - 1.000 - —

c - - - - 1.000
d 1 000 1 000 - - —
MDH3 1.000 1.000 X X X
PGMla 0.058 - - - -
b 0.074 0.009 - - -

c 0.846 0.991 - - 0.170

d 0.022 - — - 0.830
e - - 1.000 - -
f - - - 0.993 -
g - - - 0.007 -
PGM2a 0.245 0.981 - - X
b — - 1.000 1.000 X
c 0.755 0.019 - - X
GOTla 1.000 1.000 1.000 - -

GOT2a

HHDCLOIT

'U
C)
U
SD

D‘LQHHDQIOU'

SKD

U'ID

LTLQ Him CLO

GDHa

O

O

0000

.467
.031

.502

xxx

21

Table 1.2

.047
.934

.019 1.

xxx

xxx

(cont’d)

O
O
O

xxxxxx xxx

.086
.903
.011

.016
.974

.010

.000

xxxxxx

 

aAn X denotes unresolved loci.

22

 

A000.Himmm.ov Ammm.0i0HN.ov

050.0 bfim.0
Avmm.0ummm.ov
mmm.0 mbm.0
000.H 0H0.0
000.0 mom.0
000.0 bmb.0
HOMM DEU‘

A000.0i000.0v
000.0

AHHN.0IN0N.00
00N.0

A000.Hi000.HV

000.H

000.0

000.0

MHM4

Am>0.01m00.0v
bN0.0

Ammo.0imb0.0v
Hm0.0

Aamm.o-ama.oc
momé

A000.Hibmm.0v
mmm.0

000.0

A000.0IN00.0V
bN0.0

ANHH.0u0N0.00
mb0.0

Ammm.0nmmm.0v
mmm.0

Am¢0.0u00b.ov
00b.0

A000.Him¢b.ov
bmm.0

UAdU

HOHM

DdUd

HHM4

UAdU

 

.Aamnommflp mnu m>onmv Umumnm ohm mmHoHHm on £0H£3 GA flUoH up hocmdvonw

onu 0cm .mxmu maosm msomfiummsoo Aamcommfip mnu onmnv camaommmnoucw 0cm Aamsommfip

onu mnonv UHMAUQOMHucfi Ham HON mmmcmh paw mosam> xufiunmpfl m.wmz ommum>¢

.m.H mamfih

23
of these values through UPGMA revealed that Q. calceolus and
Q. candidum cluster first, followed by Q. arietinum which
joins the group at 0.247 (Figure 1.1). Cypripedium acaule
and Q. reginae form a separate cluster with an identity

value of 0.247 and join the C.

calceolus-Q. candidum-Q.
arietinum group at an average identity of 0.072. The large
genetic divergence among most taxa in the present study is
further indicated by the high frequency of loci per taxon
which contain moderate to high frequency unique alleles
(mean Ua = 0.38 in Table 1.4), and by an average of 71% of
the loci between species pairs which share no alleles (Table
1.3). This is in contrast to the high genetic identities
between Q. calceolus and Q. candidum. In Q. candidum, 79%
of its alleles are also found in Q. calceolus, and the
remaining 21% (i.e., five alleles) are unique to Q. candidum
(Table 1.2).

In general, percent polymorphic loci, alleles per locus,
and effective alleles per locus followed similar trends in
population and species comparisons (Table 1.4). At the
species level, 75% of the loci in Q. calceolus were found to
be polymorphic with 2.5 alleles per locus. Qypripedium
candidum is next with values of 66.7% and 2.00,
respectively, followed by Q. acaule (46% and 1.77), Q.
reginae (18% and 1.27), and lastly, Q. arietinum, which was
monomorphic at all loci examined. Differences among the
species were also found in the overall levels of genetic

diversity (as measured by Ht) as well as how that diversity

24

 

CALC
CAND
ARIE
ACAU
REGI
l l l
r I I
0.072 0. 247 0.794

GENETIC IDENTITY

FIGURE 1.1. UPGMA dendrogram of Nei's (1978) genetic

identities for Q. calceolus (CALC), C. candidum (CAND),

IO

arietinum (ARIE), Q. acaule (ACAU), and Q. reginae (REGI).

25

 

 

 

mm.o AHH.H0 mo.H .Hp.av mm.a .ma¢.ov mon.o nee:

mm.o loo.Hv oo.H Aoo.Hv oo.H Aooo.ov ooo.o ssqflmuflum u
sm.o Amo.H0 mo.H Asm.Hv mH.H Ammfl.ov mmH.o mmcflmmu m
64.0 ANH.H0 mo.H Ass.av ee.a Amm¢.ov sem.o masmom m
00.0 Amo.Hv mo.H Aoo.m0 me.a Asmm.ov mmm.o snoflccmo u
mm.o Amm.Hv em.a Aom.mv ms.a Aoms.ov mmm.o msflomoamo u
up Amodv mod Andy m4 Ammv mm mouoomm

 

.coxmu some 000 mm.o can» nmummnm mmflocmsvmnm nufi3 moHoHHm oswflcs samucoo zofln3 AmDV flooH
mo hocmsvoum can one .mHo>mH Amy mofloomm paw Amy coflumHsmom 030 um Andy msooH mom monHHm

m>fluommmm paw .Adv mzooH mom mmaoaam .Amv HUGH Uflnmuoexaom no hocmsvmhm .¢.H mqmdb

26
is partitioned among populations. Cypripedium calceolus has
an average Ht of 0.244 with 19% of this diversity
distributed among populations (Table 1.5). Qypripedium
acaule has the next largest level of diversity (Ht = 0.095)
with 16% of its variation residing among populations.
Qypripedium candidum and Q. reginae have similar levels of
diversity (Ht = 0.054 and 0.037, respectively) but may
partition that diversity differently. Whereas only 7% of
the total variation resides among populations in Q.
candidum, 35% of the total variation in Q. reginae is
distributed among populations. However, this relatively

high Gst in C.

reginae is due to a single polymorphic locus,
PGM-l, which had an unusually high frequency of allele c in
population 503.

Expected heterozygosity at the population level (i.e., Hs)
followed the same species-level trend; Q. calceolus
populations were most heterozygous with an average Hs of
0.197 followed by Q. acaule (0.080), Q. candidum (0.050),
and Q. reginae (0.024; Table 1.5). Most loci were in good
agreement with Hardy-Weinberg expectations, although
significant deviations (p < 0.05, p < 0.01, or p < 0.001)
occurred in eight cases distributed across five loci and six
populations (Table 1.6). Five of the cases involve
heterozygote excess. A deficiency of homozygotes was also
the most common form of deviation from Hardy-Weinberg

expectations in the nearly significant cases (0.05 < p <

0.10). Estimates of Nm (the average number of migrants

TABLE 1.5.

27

Gene diversity statistics for polymorphic

Qypripedium species; Ht is total diversity, Hs and Dst are

within- and among-population diversity, respectively, and Gst

is the proportion of variation attributable to among-

population differentiation; standard errors of loci are in

 

 

parentheses.

Species Ht Es Dst Gst

Q. calceolus 0.244 0.197 0.047 0.194
(0.054) (0.045) (0.012) (0.025)

Q. candidum 0.054 0.050 0.004 0.069
(0.020) (0.017) (0.002) (0.032)

Q. acaule 0.095 0.080 0.015 0.164
(0.046) (0.037) (0.011) (0.075)

Q. reginae 0.037 0.024 0.013 0.349
(0.032) (0.020) (0.013) (0.316)

 

28

Hoo.o v d u.s “Ho.o v d u . “mo.o v d u n “OH.o v d u .
nmmm.o- mom
nmmm.o- Hoe

.mam.o- 40¢

.mam.o mod

.mme.o- mom

.ooo.H Hoo

aomma Goa

.emm.o- amam.o usmmm.o- ammm.o- mo>

.eom.o- nmmm.o- mom

owmm.0 Hom

 

mum mnH mom «-900 «-20m H-20m H-Hme noeuuasmom

 

 

.xoocH uoflumxflm can >3 Uommohmxo mflHQflHfizvomflp o>fluflmom
Ho o>flummmc mo Hm>oH onu 0cm MHHQHHflsvo mumncflmzuxphmm Eonm mcowumfl>mp muooHuonnwm

Aoa.o v m v mo.ov unmowwwcmwm Ham: 0cm AH00.0 MO Ho.o .mo.o v my DGMUAMHamwm .m.H mammh

29
exchanged between local populations per generation) for
intraspecific populations were 3.62 (Q. reginae), 4.07 (Q.

acaule), 8.36 (Q. calceolus) and 11.07 (Q. candidum).

DISCUSSION

Interspecifig relationships—-The electrophoretic data
reveal an extremely wide range in the degree of relatedness
among the five taxa examined. While Nei’s intraspecific
average identities ranged from 0.937 in Q. calceolus to
1.000 in Q. arietinum, the average interspecific genetic
identity values ranged from 0.000 for the Q. reginae-Q.
arietinum comparison to 0.794 for the Q. calceolus-Q.
candidum comparison. Hence the difference in magnitude
between average intraspecific and interspecific values
indicate clearly demarcated and genetically distinct taxa.
This is further indicated by an average congeneric identity
value of 0.197 which is among the lowest yet reported in the
plant literature. This value is considerably lower than
0.670, the average congeneric identity reported for plant
species (Gottlieb, 1977, 1981b; Crawford, 1983).

Too few studies have examined electrophoretic variability
in the Orchidaceae to enable generalizations, but it is
interesting to note that other studies on the family also
have reported unusually low genetic identities. For
example, 11 species in the genus Orchis from Italy displayed
interspecific identity coefficients in the range of 0.000 to

0.767 with a mean congeneric identity of 0.194 (Scacchi, De

3O
Angelis, and Lanzara, 1990). Epipactis is another
orchidaceous genus which has relatively low identity values
(Scacchi, Lanzara, and De Angelis, 1987). These authors
found that congeneric identity among three Italian species
was 0.470; two of these species pairs had low interspecific
identities (0.289 and 0.336), whereas the third pair had a
higher genetic identity of 0.784. In non-orchidaceous
studies which have examined genetic identities among several
congeneric species, low genetic identities among species
pairs similar to that found in Qypripedium occasionally
occur but are usually confined to a small percentage of the
species comparisons or involve a relatively small number of
species (e.g., Manos and Fairbrothers, 1987; Olmstead, 1990;
but see also Nickrent, 1986). In part, this may reflect a
tendency for investigators to examine species which are
morphologically difficult to delineate because of recent
phylogenetic separation. Recently, however,
intercontinental disjunct species pairs which may represent
relic populations from ancient more widespread distributions
have been examined. Values reported for these species pairs
are oftentimes lower than the 0.67 mean reported for
congeneric species, but are still higher than many of the
Qypripedium or Orchis species pairs (Figure 1.2). For
example, Vogelmann and Gastony (1987) report congeneric
identity values among disjunct Agastache species ranging
from 0.51 to 0.75. Still, other reports of vicariad species

pairs are lower, such as 0.305 for a Ligpidambar species

31

 

0.7 .................................................
0.6 I” ________________________________________________
0.5 ” _____________________________
0.4 ” ...........................
0.3 ____________________

02

Congeneric Identity

OJ

 

 

AV AGA uo UH DAT CYP
Genera

FIGURE 1.2. Average congeneric Nei’s identity values for
intercontinental species pairs within the genera Agagtaghe
(AGA), Liguidambar (LIQ), Liriodendron (LIR), and Datisca
(DAT) compared to the average congeneric identity values
reported for plant species in general (AV) and for

Qypripegium species in this study (CYP).

 

32
pair (Hoey and Parks, 1991), and 0.434 for a Liriodendron
species pair (Parks and Wendel, 1990). Two of the lowest
non-orchid congeneric identity values yet reported are
0.142, among two intercontinental disjunct Datisca species
(Liston, Rieseberg, and Elias, 1989), and 0.087 among North
American Ceratophyllum demersum L. and Q. echinatum A. Gray
(Les, 1989).

Low genetic identities may reflect a relatively long
period of genetic isolation as well as an accelerated
evolutionary rate. One way orchids might evolve at a faster
rate would be through a tendency toward genetic bottlenecks
which may occur through population crashes, founder events
and genetic drift. These processes could potentially
promote population differentiation by causing chance
fixation of rare or uncommon alleles which otherwise might
remain in low frequency. Although these processes may have
played an important role in the evolution of Cypripedium
taxa at some point in time, several lines of evidence
suggest that at least a long period of genetic isolation has
contributed substantially to those species pairs with low
genetic identities. First (with the exception of the Q.
candidum-Q. galceolus pair), an average of 71% of the loci
among pairwise species comparisons share no alleles. This,
together with a high proportion of moderate to high
frequency unique alleles in most taxa (Table 1.4), suggest
that they have existed long enough in genetic isolation to

evolve very dissimilar genomes. This is a very different

33
pattern of genetic identities than is often found among
species that have undergone recent adaptive radiation. In
island endemics, for example, extensive morphological and
ecological diversity is often accompanied by little
divergence at isozyme loci (Helenurm and Ganders, 1985).

The occurrence of closely related taxa in the Old World is
another observation which suggests that most Qypripedium
taxa analyzed have experienced a long period of phylogenetic
separation. Qypripedium calceolus is represented in Europe
and Asia by Q. calceolus var. calceolus whereas Q. reginae
and Q. arietinum are morphologically and ecologically so
similar to the East Asian Q. flavum Hunt & Summerh. and Q.
plectrochilon Franch., respectively, that species
distinctions have not yet been resolved (Sing-Chi, 1983).

It has been suggested that these species pairs have
descended from a common stock which once had a continuous
range (Sing-Chi, 1983). Qypripedium acaule also shows
similar morphological affinities with a group of Old World
species (Atwood, 1984) whereas in North America it bears no
close resemblance to any extant species. The distribution
of Qypripedium in the Northern Hemisphere together with the
low genetic identities found is consistent with the
hypothesis that the Cypripediodeae is an old lineage, and
further suggests that the genus Qypripedium may also be very
old.

Although most taxa have evolved extensive genetic

differentiation, the allozyme data do not suggest that Q.

34
arietinum represents the most divergent taxon as the
morphology would indicate. On the contrary, Q. arietinum
and Q. calceolus have the second highest interspecific
identity value of all interspecific comparisons (Table 1.3).
Qypripedium acaule and Q. reginae have approximately the
same degree of genetic divergence from each other as does Q.
arietinum and Q. calceolus (or Q. candidum), but they have

an extremely low identity with the C. calceolus—Q.

candidum-Q. arietinum group (Figure 1.1). Therefore, a
transfer of Q. arietinum into a segregate genus based on
genetic identity values would also necessitate the creation
of two additional segregate genera for Q. reginae and Q.
acaule. Since relatively little electrophoretic analysis
has been conducted on closely related genera, little is
known about the degree of genetic divergence genera commonly
have. Likewise, until a larger body of electrophoretic
literature is available for the Orchidaceae from which to
compare interspecific values, the Cypripedium taxa in the
present study are best interpreted by the allozyme data as
highly divergent but congeneric.

Intraspecific diversity—-Intraspecific measures of genetic
diversity also vary greatly among the taxa examined, again
reaching above and below the averages reported for
comparable flowering plants (Figure 1.3). Cypripedium
calceolus had the highest levels of genetic variation for

all of the genetic parameters examined. At the species

level, 75% of its loci were polymorphic with an average of

35

 

 

 

Diversity Measure

 

 

 

 

AV CALC CAND ACAU REGI ARIE
Species
DHeterozygosity -(Aileies per Iocus)/10 IPoiymorphic Loci

FIGURE 1.3. The frequency of expected heterozygotes,
alleles per locus, and the frequency of polymorphic loci at
the species level for Q. calceolus (CALC), Q. candidum
(CAND), Q. acaule (ACAU), Q. reginae (REGI), and Q.

arietinum (ARIE) compared to the average species-level

diversity measures reported for plants in general (AV).

36
2.5 alleles per locus, and its species-level heterozygosity
(Ht) was 24.4%. These values are considerably higher than
the averages reported for other plant taxa (50.5%, 1.96,
14.9%, respectively; Hamrick and Godt, 1989). They are also
consistently higher than the averages reported for plants
with life history characteristics similar to Q. calceolus
(i.e., herbaceous perennial plants that are long—lived,
animal-outcrossed, and have wind—dispersed seeds), as well
as plants that have a widespread geographic distribution
(Hamrick and Godt, 1989). Higher than average values of
genetic diversity were also observed at the population level
within Q. calceolus (alleles per locus equaled that reported
for plants with a widespread distribution). This high level
of diversity is consistent with the high level of
morphological and ecological diversity found within this
taxon. Moreover, each variety of Q. calceolus has generally
higher than average levels of genetic diversity as measured
by percent polymorphic loci, alleles per locus and
heterozygosity estimates (M. Case, Part 2). Therefore, the
high level of variation within Q. calceolus does not result
from the inclusion of divergent infraspecific taxa, rather
it is due to the high level of variation within each
variety. The distribution of genetic variation within this
species is also consistent with the observation that this
taxon is predominantly outcrossed. In an outcrossed
species, the majority of species-level variation, as

measured by expected heterozygosity, often resides within as

37
opposed to among populations (Hamrick and Godt, 1989).
Similarly, only 19% of the variation in Q. calceolus resides
among populations, which is equal to the average value I
reported for animal-outcrossed taxa (Hamrick and Godt,
1989). Moreover, the results of single-locus tests for
deviations from Hardy-Weinberg expectations (Table 1.6)
support the idea that self-pollination does not frequently
occur within this taxon. These tests indicate that there
were few significant deviations from the expected levels of
heterozygosity across the vast majority of single—locus
tests, and that when significant deviations did occur, they
were most often in the direction of excess heterozygosity.
In summary, the genetic diversity within Q. calceolus is
partitioned in a manner that is typical of a widespread,
outcrossed, long-lived herbaceous plant, but it contains
higher levels of diversity than average values reported for
species with similar characteristics.

In contrast to the high levels of variability in Q.
galgeglus, the remaining taxa have either reduced levels of
polymorphic loci or very low levels of heterozygosity.
While Q. candidum has the second highest percentage of
polymorphic loci at both the species and population levels
(66.7%, 38.3%, respectively), the levels of heterozygosity
at these loci are greatly reduced to 5.4% at the species
level and 5.0% at the population level. Although low
heterozygosities are typical for species with narrow

distributions (Hamrick and Godt, 1989) similar to Q.

38
candidum, the average values reported for such species are
still twice as large as those reported for Q. candidum.
Since the majority of differences in levels of
heterozygosity in plants has been found to be due to the
proportion of polymorphic loci rather than differences at
individual loci (Hamrick and Godt, 1989), Q. candidum is
somewhat atypical in this regard. In this species, low
heterozygosity values do not result from a low proportion of
polymorphic loci, but rather from very low heterozygosities
at a large number of polymorphic loci. This discrepancy may
result from past introgression of alleles from Q. calceolus
to Q. candidum, which may have increased the number of

polymorphic loci in C. candidum but perhaps not enough to

significantly affect overall levels of heterozygosity. More
extensive introgressive hybridization of these two taxa may
occur in Brandon, Manitoba (Actor, 1984). In an allozyme
survey of this population, Actor found very high levels of
heterozygosity in Q. candidum (34%), Q. calceolus (37%) and
in individuals that were morphologically classified as
hybrids (37%). Therefore, there is nearly a seven-fold
increase in heterozygosity in the Brandon population of Q.
candidum compared to Michigan populations when
heterozygosities are recalculated with the five loci that
were common to both studies. Either there are much higher
levels of introgression in the Brandon population, or strong
regional differences occur for the level of heterozygosity

in Q. candidum.

39

The relatively high genetic identity of Q. calceolus and
Q. candidum (0.794) and the reduced level of variation
present in all Michigan populations of Q. candidum as
compared to Q. calceolus may reflect a progenitor—derived
species relationship. In a progenitor-derived pair of
relatively recent origin, the derived taxon frequently
exhibits a subset of the variation present in the progenitor
as well as unique alleles which commonly exist in low
frequency (Gottlieb, 1977; Pleasants and Wendel, 1989).
This expectation is consistent with the variation found in
Q. candidum. Percent polymorphic loci, alleles per locus,
and heterozygosity at the species level are reduced in Q.

candidum compared to C.

calceolus by 11%, 20% and 78%,
respectively. Moreover, 79% of the alleles in Q. candidum
also occur in Q. calceolus, whereas the remaining 21% are
unique to Q. candidum and exist in low species—level
frequencies (0.005-0.028). If a Q. calceolus population
gave rise to Q. candidum, then the alleles present in
highest frequency in the progenitor population would also
have the highest probability of being present in the derived
taxon (e.g., see Stephanomeria in Gottlieb, 1973).

Consistent with this expectation, the highest-frequency

alleles present in C.

calceolus are to a large extent
present as the highest-frequency or second highest—frequency
alleles in Q. candidum. For example, six of the nine
mutually polymorphic loci share the same highest frequency

alleles, and the highest frequency alleles at two loci in Q.

40
candidum are the second highest frequency alleles in Q.

calceolus. Only one locus in C.

candidum has a high
frequency allele that is present at low frequency in Q.
calceolus.

Although only a very limited amount of material was
available for analysis, another progenitor-derived species
pair may occur. In the three Q. kentuckiense individuals
examined, no polymorphism occurred across the 12 loci
examined and all individuals were fixed for the same
alleles. These alleles were, at every locus, the highest
frequency alleles present in Qypripedium calceolus var.
pubescens, the North American taxon it most closely
resembles and with which it is sympatric (see Atwood, 1985,
who recognizes Q. calceolus var. pubescens as a distinct
species). One of the most notable features of Q.
kentuckiense is its relatively large size, which is rivaled
only by a Chinese species, Q. wilsonii Rolfe. It has been
suggested that the latter is a potential close relative of
Q. kentuckiense (Atwood, 1985). Although a larger genetic
sample of Q. kentuckiense is needed before any conclusions
can be drawn, it is doubtful that Q. wilsonii is more
closely related to Q. kentuckiense than is Q. calceolus.

Many other potential progenitor-derived species pairs have
been identified through enzyme electrophoresis (see reviews
by Gottlieb, 1981b; Crawford, 1983; Pleasants and Wendel,

1989) including the orchid pair, Epipactis microphylla and

E. hellebgrine (Scacchi, Lanzara, and De Angelis 1987).

41
Epipactis microphylla was found to exhibit no variability
within or between populations and, with the exception of one
unique allele, had the most common alleles of E.
helleborine. Founder events were implicated in the
depauperate nature of the E. microphylla populations. If
rapid speciation is common in the Orchidaceae as proposed by
Dressler (1981), then many such progenitor-derived species
pairs should occur. Therefore, protein electrophoresis has
great potential for testing this idea in genera like
Stanhopea whose species apparently undergo rapid speciation
due to strong selection for different pollinators (Dodson,
1962). Qypripedium acaule maintains the second highest
levels of species and population heterozygosities (0.095 and
0.080, respectively) which represent a 60% decrease from
values observed for Q. galceolus. Similarly, percent
polymorphic loci at the species and populational levels are
reduced by 38% and 45%, respectively in Q. acaule compared
to Q. calceolus. However, in comparison with other plant
taxa which have a regional to widespread distribution and
similar life history characteristics, these values are close
to the averages reported for polymorphic loci, but are
consistently smaller than the averages reported for
heterozygosity levels (Hamrick and Godt, 1989). Gill (1989)

has predicted that C.

acaule populations in particular
should contain extremely low levels of genetic variation
because the combined probability that a given seed will

reach a new site, survive the numerous highly vulnerable

42
stages to adulthood, and then be self—pollinated to
establish a population is thought to be extraordinarily
small. Although heterozygosities in Q. acaule are lower
than the averages reported for comparable species, this
taxon is not genetically depauperate. Furthermore, all
Qypripedium taxa in this study have comparable life history
characteristics but the levels and distribution of genetic
variation among them differ dramatically. This suggests
that factors other than their life—history characteristics,
at least in part, also affect the variation.

The remaining taxa, Q. reginae and Q. arietinum, have
remarkably low levels of genetic variation and could be
viewed as genetically depauperate. Qypripedium arietinum
contained no polymorphic loci whereas only two of 11 loci
were polymorphic in Q. reginae. Although overall
species-level heterozygosity in Q. reginae was 0.037, which
is only slightly smaller than the levels in Q. candidum, the
distribution of the variation in the two taxa is very
different. Whereas Q. candidum has 7% variation of its
variation distributed among populations, Q. reginae has on
average 35% of the variation distributed this way. However,
the two polymorphic loci in Q. reginae give very different
estimates (Gst for PGM-l = 45%; Gst for GOT-2 = 1%) because
a PGM-l allele which is rare in two populations is common in
population 803. When this population is omitted from the
analysis, average Ht and Hs each lower to approximately

0.012 and virtually no variation is spread among

43
populations. The two polymorphic loci also give nearly
identical estimates for Ht, Hs, and Gst when population 503
is removed.

The overall striking features of these five Cypripedium
species are the common occurrence of low genetic identities,
low intraspecific variation (with the exception of Q.
calceolus), and an average of 10% (omitting population 803)
of the total species-level variation in polymorphic taxa
distributed among populations. Likewise, only 7% of the
total species-level variation (Scacchi, De Angelis, and
Lanzara, 1990) in Orchis populations and 3.3% in Epipactis
helleborine (Scacchi, Lanzara, and De Angelis, 1987) are
distributed in such a manner. In species which are likely
to have isolated populations founded by one or a few
individuals, genetic drift should act to increase the
variance in allele frequencies among populations (Nei,
1987). This would cause a high proportion of the total
species—level variation to be distributed among populations
rather than within. This population structure has been
observed in Lisianthius skinneri (Hemsl.) Kuntze, a
tropical, perennial shrub found in moist, mid-elevational
forests from Guatemala to Panama (Sytsma and Schaal, 1985).
Lisianthius skinneri is very similar to Qypripedium taxa in
that populations are likely to be founded by few
individuals, and they are markedly patchy in distribution
and small in size. Consistent with the genetic expectations

of this population structure, no genetic variation was found

44
in Lisianthius skinneri at the populational level and
different populations were fixed for different allelic
variants across the polymorphic loci. Therefore, 100% of
the total species—level variation was found to reside among
populations. Unlike L. skinneri, Cypripedium species have
average to below average Gst values for perennial,
outcrossed taxa with wind-dispersed seeds, and they are only
one third as large as the average Gst value reported for
plants which routinely undergo inbreeding (e.g., see Gst
value reported by Hamrick and Godt, 1989, for
self—pollinating species). Therefore, genetic drift or
other evolutionary processes promoting differentiation among
populations do not seem to have contributed more to
among-population variation in Qypripedium taxa than in other
comparable plant taxa. This is a surprising result given
the apparent isolated nature of many orchid populations, and
further suggests that these orchid taxa, despite the
propensity for small, widely spaced populations, have or
have had average to relatively high levels of
interpopulational gene flow.

In these Qypripedium taxa, Slatkin's (1985) measure of
gene flow is also suggestive of high levels of gene flow.
This method uses the average frequency of alleles which are
confined to one population (i.e., "private alleles") for an
estimation of the average number of migrants exchanged
between local populations [(Nm)est]. In this analysis, the

Qypripedium taxa have values ranging from 3.62 in Q. reginae

45
to 11.07 in Q. candidum with an average (Nm)est of 6.8.
This is very similar to other averages reported for plants
which have either winged seeds [(Nm)est = 4.313] or
wind-pollination mechanisms [(Nm)est = 5.380], and is
considerably higher than average values reported for plants
which have animal-outcrossed pollination mechanisms [(Nm)est
= 0.801; Hamrick, 1987].

The general uniformity of populations and the frequent low
levels of heterozygosity in these Qypripedium taxa may have
several contributing factors. In Q. candidum, it is likely
that the low levels of heterozygosity and among-population
variation are, to some degree, related to the possible
origin of the taxon. However, the lack of any close
relatives of Q. arietinum, Q. acaule, and Q. reginae on this
continent and their likely occurrence in Asia, in addition
to the abundance of moderate to high frequency
species-specific unique alleles, suggest that these taxa are
phylogenetically very old. Therefore, it is unlikely that
the lowered diversity in these taxa is due to their recent
derivation.

Another factor which could potentially influence the
genetic variability may be the degree to which adaptation to
different ecological niches has maintained or reduced
variation within each taxon. One hypothesis regarding this
states that adaptation to different aspects of the
environment is responsible for the maintenance of species

variation (Van Valen, 1965). Therefore, species which can

46
exploit a variety of niches should, in theory, contain
larger amounts of variation than should species with more
restricted requirements. This has been termed the
"niche width-variation hypothesis" (Babbel and Selander,
1974). In a broad sense, a correlation of niche width and
genetic diversity can be seen with Qypripedium. The species
able to exploit the widest array of habitats (i.e., Q.
calceolus) is the most genetically diverse, while the most
ecologically restricted species (i.e., Q. candidum) is among
the lowest in species diversity as measured by expected
heterozygosity. Lastly, historical factors may also have
had substantial affects on the observed genetic variation
within these taxa. Since the majority of each species’
range now occurs north of the last Wisconsin continental ice
sheet border (Luer, 1975; Péwé, 1983), their current
northward distributions represent range expansions that have
occurred within the last 11,000 years (Webb, Cushing, and
Wright, 1983). For some species, glaciation may have
substantially reduced their geographic range and restricted
them to small, isolated populations. Under these
conditions, there would be loss of genetic variation within
populations due to genetic drift. Moreover, populations
which now occur in glaciated areas could have been derived
from a single genetically depauperate progenitor population.
In the extant populations, this would cause reduced genetic
variation within populations and produce a high level of

uniformity among populations. Species which could inhabit a

47

wider array of habitats or persist in a wider geographical
area during glaciation may not have been as affected by
glacial processes on a species level. In these species,
genetic variation could be lost in certain geographical
areas but the persistence of numerous populations across a
wide geographic range could, in theory, preserve
species-level variation (Templeton et al., 1990).

Among the North American taxa, Cypripedium calceolus
extends farthest south of the glacial boundary, occurring in
most eastern states south to Louisiana. If this relatively
large southern range and ecological diversity reflects an
equally extensive range during the late Quaternary, then
more species-level variation may have been maintained in Q.
calceolus than in the remaining taxa which are less
ecologically diverse and may also have had relatively less
available habitat during glaciation. The extreme of this is
seen in Q. arietinum, a less ecologically diverse species
which has a current range totally within the glacial
boundary. The corresponding complete lack of genetic
variation in this taxon at the species level is suggestive
of a genetic bottleneck which has eliminated variation in a
population ancestral to the current populations.

The low heterozygosities together with the uniformity of
,Q. reginae (excluding consideration of population 503) and
Q. gandidum populations are also suggestive of genetic
bottlenecks which occurred in ancestral populations. In the

latter species, only 7% of its variation is partitioned

48

among populations, whereas its close relative, Q. calceolus,
maintains 19% of the variation in this way. Although
Hamrick and Godt (1989) found that the breeding system and
the method of seed dispersal accounted for the largest
proportion of variation in Gst values, these two variables
are not likely to differ appreciably among Q. candidum and
Q. calceolus and thus account for the difference. However,
if the current populations of Q. candidum were derived from
a common ancestral population, then the allele frequencies
may reflect those of the ancestral population. A recent,
common evolutionary history of populations could also lead
to inaccurate estimates of gene flow as measured by the
frequency of private alleles. When genetically isolated
populations are founded from a common ancestral population,
and have not yet reached a genetic and demographic
equilibrium, then the frequencies of rare alleles in those
populations are determined by the time since radiation
rather than by current levels of gene flow (Slatkin, 1985).
This would lead to an overestimate of the actual gene flow.

It is also likely that patterns of gene flow among present
populations are different than those patterns which occurred
when populations were recolonizing newly deglaciated
territory. Since these species are infrequently found in
mature forest but are found most abundantly in more open
areas of transient successional stages (Case, 1987), it is
likely that open areas after glacial retreat may have been

particularly favorable to the establishment of these orchid

49

taxa. Their potential for long-distance dispersal would
also have aided their fast recolonization. The current
populations which are now confined to a patchy mosaic of
favorable habitats may therefore represent "relic"
populations from a much more contiguous range in a former
landscape that was more conducive to gene flow.
Glacial disturbances and other Pleistocene phenomena have
apparently reduced the genetic variation in many taxa such
as Pings resinosa Aiton (Fowler and Morris, 1977; Simon,
Bergeron, and Gagnon, 1986), Acacia mangium Willd. (Moran,
Muona, and Bell, 1989), Salicornia eurogea L. (Wolff and
Jefferies, 1987), Trillium nivale Riddell (Bayer et al.,
1987), Cirsium pitcheri (Torr.) Torr. & A. Gray (Loveless
and Hamrick, 1988), and Sullivantia species (Soltis, 1982).
In Sullivantia, three of the four taxa analyzed contained no
variation and the remaining taxon contained extremely low
levels. These populations currently are restricted to
unglaciated regions near Pleistocene glacial margins and
colonized these sites after the Wisconsin glaciation. It
was suggested that founder events occurred during
colonization. This, along with their narrow endemic nature,
may have contributed to the decreased levels of variation in
those populations (Soltis, 1982).

The levels of variation in Cypripedium taxa are complex,
ranging broadly across values reported for plant taxa.
Despite life histories that are similar, consideration of

them alone can be misleading in predicting levels of

50
variation contained within and among populations. Likewise,
evolutionary inference from current measurements of gene
flow and population structure may not be accurate because
current conditions may not be the same as those under which
the populations were originally established. In Qypripedium
it is apparent that each taxon has a unique evolutionary
history and that inference of genetic variation based upon
common life-history characteristics is not generalizable to

all taxa.

51
LITERATURE CITED
ACTOR, G. F. 1984. Natural hybridization of Cypripedium
candidum and Q. calceolus. M.S. thesis, University of

North Dakota.

ATWOOD, J. T. 1984. The relationships of the slipper
orchids (subfamily Cypripedioideae, Orchidaceae).

Selbyana 7: 129-247.

----------- . 1985. The range of Qypripedium kentuckiense.

American Orchid Society Bulletin 54: 1197-1199.

BABBEL, G. R., and R. K. SELANDER. 1974. Genetic
variability in edaphically restricted and widespread plant

species. Evolution 28: 619-630.

BALLARD, W. W. 1987. Sterile propagation of Cypripedium
reginae from seeds. American Orchid Society Bulletin 56:

935-946.

BAYER, R. J., J. C. LA DUKE, and D. J. CRAWFORD. 1987.
Isozyme variation in Trillium nivale (Liliaceae).

Canadian Journal of Botany 65: 2250-2254.

CASE, F. W. JR. 1987. Orchids of the western Great Lakes
region. Cranbrook Institute of Science Bulletin 48,

Cranbrook, Bloomfield Hills, Michigan.

52
CORRELL, D. S. 1950. Native orchids of North America north
of Mexico. Chronica Botanica Company, Waltham,

Massachusetts.

CRAWFORD, D. J. 1982. Electrophoresis. Lecture notes from
Botany 812, third edition, fall quarter, Ohio State

University.

------------- . 1983. Phylogenetic and systematic
inferences from electrophoretic studies. In S. D.

Tanksley and T. J. Orton [eds.], Isozymes in plant

genetics and breeding, Part A, 257-287.‘ Elsevier,

Amsterdam.

CURTIS, J. T. 1943. Germination and seedling development
in five species of Qypripedium L. American Journal of

Botany 30: 199-206.

DODSON, C. H. 1962. The importance of pollination in the
evolution of the orchids of tropical America. American

Orchid Society Bulletin 31: 525-234, 641-649, 731—735.

DRESSLER, R. L. 1981. The Orchids. Harvard University

Press, Cambridge, Massachusetts.

FOWLER, D. P., and R. W. MORRIS. 1977. Genetic diversity
in red pine: evidence for low genic heterozygosity.

Canadian Journal of Forest Research 7: 343-347.

53
GILL, D. E. 1989. Fruiting Failure, pollinator
inefficiency, and speciation in orchids. In D. Otte and

J. Endler [eds.], Speciation and its consegpences,

458-480. Sinauer Associates, Sunderland, Massachusetts.

GOTTLIEB, L. D. 1973. Genetic differentiation, sympatric
speciation, and the origin of a diploid species of

Stephanomeria. American Journal of Botany 60: 545-553.

------------- . 1977. Electrophoretic evidence and plant
systematics. Annals of the Missouri Botanical Garden 64:

161-180.

-------------- . 1981a. Gene number in species of Astereae
that have different chromosome numbers. Proceedings of
the National Academy of Sciences of the United States of

America 78: 3726-3729.

-------------- . 1981b. Electrophoretic evidence and plant
populations. In L. Reinhold, J. B. Harborne, and T. Swain,

[eds.], Progress in phytochemistpy, Vol. 7, 1-46.

Pergamon, New York.

_____________ , 1982. Conservation and duplication Of

isozymes in plants. Science 216: 373-380.

HAMRICK, J. L. 1987. Gene flow and distribution of genetic
variation in plant populations. In K. M. Urbanska [ed.],

Differentiation patterns in higher plants, 53-67.

Academic Press, Orlando.

54
------- , and M. J. W. GODT. 1989. Allozyme diversity in
plant species. In A. H. D. Brown, M. T. Clegg, A. L.
Kahler, and B. S. Weir [eds.], Plant population genetics,
breeding. and genetic resources, 43-63. Sinauer

Associates, Sunderland, Massachusetts.

HEDRICK, P. W. 1985. Genetics of populations, Jones and

Bartlett, Boston.

HELENURM, K., and F. R. GANDERS. 1985. Adaptive radiation
and genetic differentiation in Hawaiian Bidens. Evolution

39: 753-765.

HERNANDEZ, J. L., and B. S. WEIR. 1989. A disequilibrium
coefficient approach to Hardy-Weinberg testing.

Biometrics 45: 53-70.

HOEY, M. T., and C. R. PARKS. 1991. Isozyme divergence
between Eastern Asian, North American, and Turkish species
of Ligpidambar (Hamamelidaceae). American Journal of

Botany 78: 938-947.

LES, D. H. 1989. Population genetics of Ceratophyllum
demersum and Q. echinatum, two primitive hydrophilous
angiosperms. Supplement to American Journal of Botany

76(6): 254.

55
LISTON, A., L. H. RIESEBERG, and T. S. ELIAS. 1989.
Morphological stasis and molecular divergence in the
intercontinental disjunct genus Datisca (Datiscaceae).

Aliso 12: 525-542.

LOVE, A., and J. C. RITCHIE. 1966. Chromosome numbers from
central northern Canada. Canadian Journal of Botany 44:

429-439.

-------- , and W. SIMON. 1968. Cytotaxonomical notes on
some American orchids. The Southwest Naturalist 13:

335-342.

LOVELESS, M. D., and J. L. HAMRICK. 1988. Genetic
organization and evolutionary history in two North

American species of Cirsium. Evolution 42: 254-265.

LUER, C. A. 1975. The native orchids of the United States
and Canada excluding Florida. The New York Botanical

Garden.

MANOS, P. S., and D. E. FAIRBROTHERS. 1987. Allozyme
variation in populations of six northeastern American Red
Oaks (Fagaceae: Quercus subg. Erythrobalanus).

Systematic Botany 12: 365-373.

‘MORAN, G. F., O. MUONA, and J. C. BELL. 1989. Acacia
mangipm: a tropical forest tree of the coastal lowlands

with low genetic diversity. Evolution 43: 231-235.

56
NEI, M. 1973. Analysis of gene diversity in subdivided
populations. Proceedings of the National Academy of

Sciences of the United States of America 70: 3321-3323.

----- . 1978. Estimation of average heterozygosity and
genetic distance from a small number of individuals.

Genetics 89: 583-590.

------ . 1987. Molecular evolutionapy genetics. Columbia

University Press, New York.

------ , and R. K. CHESSER. 1983. Estimation of fixation
indices and gene diversities. Annals of Human Genetics

47: 253-259.

------ , T. MARUYAMA, and R. CHAKRABORTY. 1975. The
bottleneck effect and genetic variability in populations.

Evolution 29: 1—10.

NEWHOUSE, C. J. 1976. Pollination biology in seven taxa of
Michigan Orchidaceae and a study of Qypripedium calceolus
in Michigan based on living plants and herbarium

specimens. M.S. thesis, Michigan State University.

NICKRENT, D. L. 1986. Genetic polymorphism in the
morphologically reduced dwarf mistletoes (Arceuthobium,
Viscaceae): an electrophoretic study. American Journal of

Botany 73: 1492-1502.

57
NILSSON, L. A. 1979. Anthecological studies on the Lady’s

Slipper, Qypripedium calceolus (Orchidaceae). Botaniska

Notiser 132: 329-347.

OLMSTEAD, R. G. 1990. Biological and historical factors
influencing genetic diversity in the Scutellaria

angpstifolia complex (Labiatae). Evolution 44: 54-70.

PARKS, C. R. and J. F. WENDEL. 1990. Molecular divergence
between Asian and North American species of Liriodendron
(Magnoliaceae) with implications for interpretation of

fossil floras. American Journal of Botany 77: 1243-1256.

PEWE, T. L. 1983. The periglacial environment in North
America during Wisconsin time. In H. E. Wright Jr., and
S. C. Porter [eds.], Late- uaterna environments of the

United States, Vol. 1, 157-189. University of Minnesota

Press, Minneapolis.

PLEASANTS, J. M., and J. F. WENDEL. 1989. Genetic
diversity in a clonal narrow endemic, Erythronium

ppgppllgpg, and its widespread progenitor, Epythronium

albidum. American Journal of Botany 76: 1136-1151.

REED, C. F. 1981. Qypripedium kentuckiense Reed, a new

species of orchid in Kentucky. Phytologia 48: 426-428.

58
SCACCHI, R., G. DE ANGELIS, and P. LANZARA. 1990. Allozyme
variation among and within eleven Orchis species (fam.
Orchidaceae), with special reference to hybridizing

aptitude. Genetica 81: 143-150.

------ , P. LANZARA, and G. DE ANGELIS. 1987. Study of
electrophoretic variability in Epipactis helleborine (L.)
Crantz, E. palustris (L.) Crantz and E. microphylla

(Ehrh.) Swartz (fam. Orchidaceae). Genetica 72: 217-224.

SHEVIAK, C. J. 1983. United States terrestrial orchids:

patterns and problems. In E. H. Plaxton [ed.], North

Amerigan terrestrial grghids: symposium II proceedings and

lectures, 48-60. Michigan Orchid Society, Livonia.

SIMON, J., Y. BERGERON, and D. GAGNON. 1986. Isozyme

uniformity in populations of red pine (Pinus resinosa) in

 

the Abitibi Region, Quebec. Canadian Journal of Forest

Research 16: 1133-1135.

SING-CHI, C. 1983. Two pairs of vicarious species of
Qypripedium (Orchidaceae) from Eastern Asia and eastern

North America. Brittonia 35: 159-163.

SLATKIN, M. 1985. Rare alleles as indicators of gene flow.

Evolption 39: 53-65.

SOKAL, R. R., and F. J. ROHLF. 1981. Biometgy, W. H.

Freeman and Company, San Francisco, California.

59
SOLTIS, D. E. 1982. Allozyme variability in Sullivantia

(Saxifragaceae). Systematic Botany 7: 26-34.

------------ , C. H. HAUFLER, D. C. DARROW, and G. J.
GASTONY. 1983. Starch gel electrophoresis of ferns: a
compilation of grinding buffers, gel and electrode

buffers, and staining schedules. American Fern Journal

73: 9-27.

STOUTAMIRE, W. 1990. Eastern American Qypripedium species
and the biology of Qypripedium candidum. In C. E. Sawyers
[ed.], Norph American native terrestrial grchid
propagation and production, 40-48. Brandywine

Conservancy, Chadds Ford, Pennsylvania.

SWOFFORD, D. L., and G. J. OLSEN. 1990. Phylogeny
reconstruction. In D. M. Hillis and C. Moritz [eds.],

Molecular Systematics, 411—501. Sinauer Associates,

Sunderland.

SYTSMA, K. J., and B. A. SCHAAL. 1985. Genetic variation,
differentiation, and evolution in a species complex of

tropical shrubs based on isozymic data. Evolution 39:

582-593.

TEMPLETON, A. R., K. SHAW, E. ROUTMAN, and S. K. DAVIS.

1990. The genetic consequences of habitat fragmentation.

Annals of the Missouri Botanical Garden 77: 13-27.

60
VAN VALEN, L. 1965. Morphological variation and width of

ecological niche. The American Naturalist 99: 377-390.

VOGELMANN, J. E., and G. J. GASTONY. 1987. Electrophoretic
enzyme analysis of North American and Eastern Asian
populations of Agastache sect. Agastache (Labiatae).

American Journal of Botany 74: 385-393.

WEBB, T. III, E. J. CUSHING, and H. E. WRIGHT, JR. 1983.
Holocene changes in the vegetation of the Midwest. In H.
E. Wright, Jr. [ed.], Vol. 2, Late- uaternar environments
of the United States, 142-165. University of Minnesota

Press, Minneapolis.

WEEDEN, N. F., and L. D. GOTTLIEB. 1979. Distinguishing
allozymes and isozymes of phosphoglucoisomerases by
electrophoretic comparisons of pollen and somatic tissues.

Biochemical Genetics 17(3/4): 287-296.

------------- , and J. F. WENDEL. 1989. Genetics of plant
isozymes. In D. E. Soltis and P. S. Soltis [eds.],

Isozymes in plant biology, 46-72. Dioscorides Press,

Portland, Oregon.

WEIR, B. S. 1990. Genetic data analysis, Sinauer

Associates, Sunderland, Massachusetts.

WHITKUS, R. 1985. FORTRAN program for computing genetic
statistics from allelic frequency data. The Journal of

Heredity 76: 142.

61
WITHNER, C. L. 1988. Orchid physiology. In C. Withner
[ed.], The orchids: a scientific survey, 315-360. Krieger

Publishing Company, Melbourne, Florida.

WOLFF, S. L., and R. L. JEFFERIES. 1987. Morphological and
isozyme variation in Salicornia europea L.(Chenopodiaceae)
in northeastern North America. Canadian Journal of Botany

65: 1410-1419.

PART 2

HIGH LEVELS OF ALLOZYME VARIATION
WITHIN CYPRIPEDIUM CALCEOLUS AND LOW LEVELS

OF DIVERGENCE AMONG ITS VARIETIES

ABSTRACT
An electrophoretic analysis of 12 isozyme loci was
completed to determine the level of genetic variation within

and degree of genetic divergence among populations of

Qypripedium calceolps var. pubescens, Q. calceolus var.

parviflorum, and Q. calceglus var. planipetalum. In

addition, the genetic relationship between mixed and pure
populations of vars. pubescens and parviflorum was
determined. Differences were found among pure populations
of these taxa in the level of heterozygosity and
distribution of genetic variation. Variety-level
heterozygosity over all loci ranged from 0.180 in var.
pubescens to 0.274 in var. parviflorum, whereas the
proportion of variation distributed among populations ranged
from 5.3% in var. planipetalum to 17% in var. parviflorum.
Only slight levels of genetic divergence were found among
these taxa, which resulted from allele frequency differences

among the varieties rather than the presence of ubiquitous

62

63
unique alleles. Unweighted pair-group method analysis of
Nei’s genetic identities and a principal components analysis
of allele frequencies both revealed a tendency of the
populations to cluster with their respective varieties.
However, var. parviflorum displayed a large variance in
population differentiation. Intravarietal Nei’s identity
values, which ranged from 0.920 to 0.979, were not
consistently higher than intervarietal values (0.900 -
0.954). These results are comparable to other reports of
genetic divergence among sympatric infraspecific taxa and

suggest that vars. parviflorum, pubescens, and planipetalum

be considered members of a single species.

64
INTRODUCTION

Qypripedium calgeolug L. is a diploid, outcrossed member
of the Cypripedioideae (Orchidaceae) with a vast geographic
range. In the Old World, it extends throughout much of
Europe and Asia, occurring as far north as northern Sweden
and as far south as south central China (Hultén and Fries
1986). In North America, the major portion of its range
extends from Newfoundland across Canada (including portions
of the northernmost United States), down to Louisiana (in
the eastern portion of the US), and west to New Mexico in
the southernmost states (Luer 1975). Over this extensive
geographic range, Q. calceolus displays extreme
morphological and ecological variability which has led to
the recognition of a myriad of poorly-defined taxa (see
review by Correll 1938). According to Sheviak (1983), this
species has "generated probably more thought and
contradictory pages of print than any other North American
orchid." One aspect of this controversy concerns the
delimitation of two out of the three currently recognized
North American taxa. These taxa have been viewed by
different authors as: (1) varieties of Q. calceolus [Q. 9.
var. parviflorum (Salisb.) Fern. and Q. g. var. pubescens
(Willd.) Correll; e.g., Luer 1975], (2) two distinct species
(Q. parviflgrum Salisb. and Q. pubescens Willd.; e.g.,
Atwood 1985), and (3) varieties of one species which is
86parate from the European Q. calceolus [Q. parviflorum

Salisb. var. parviflorum and Q. p. var. pubescens (Willd.)

65
Knight; e.g., Sheviak 1992]. For purposes of this paper,

they will be referred to as C.

calceolus var. parviflorum
and Q. calceolus var. pubescens.

This controversy, however, concerns more than the problem
of rank. It also includes the long-standing debate over
which characters can reliably be assigned to each entity
(see review by Newhouse 1976). This problem is further
complicated by regional variation in characters as well as
variation in the presence of isolating mechanisms. For
example, in some regions introgression or the presence of
plants with intermediate morphologies can be found. This
apparently occurs between varieties of Q. calceolus (Harms
1986; Sheviak 1992; Stoutamire 1967) as well as between Q.
calceolus and other closely-related species (Actor 1984;
Klier et a1. 1991; Sheviak 1992). In other geographic
regions, morphologically distinct varieties occur in
sympatry without the presence of plants with intermediate
morphologies (Atwood 1985; Newhouse 1976). It is the latter
situation that has led some researchers to view these
varieties as distinct species (e.g., Atwood 1985).

As commonly interpreted (e.g., Case 1987), var. pubescens
is the most widely distributed taxon and is found throughout
the range reported above for North American Q. calceolus.
It can occur in a variety of habitats (e.g., from swamp to
upland), and can be highly variable in height (10 - 80 cm).
The labellum is often large (i.e., more than 4 cm long) and

the lateral petals and sepals are greenish yellow, and

66
spotted or streaked with maroon. In contrast, var.
parviflorum tends to be smaller in overall size with a
labellum generally less than 4 cm long. It has deep
mahogany-maroon petals and sepals, and is most often found
in wet habitats. Other characters less commonly used to
distinguish the varieties are reviewed in Newhouse (1976).
Since the definition of var. parviflorum has been in debate,
its geographic range is also uncertain. Some authors feel
that it is largely restricted to the northeastern United
States (e.g., Luer 1975) while others maintain that its
range extends from the eastern United States throughout
Canada to Alaska (e.g., Harms 1986). Sheviak (1992), Atwood
(1985), and Case (1987) have discussed regional variation in
var. parviflorum and suggested the possibility that
additional taxa may be included within it.

Qypripedium calceolus var. planipetalum (Fern.) Vict. &
Rousseau is a third commonly recognized taxon which is also
of questionable rank. This taxon is shorter than the other
taxa (up to 20 cm tall) and is characterized by having
shorter, broader, and relatively untwisted lateral petals.
It was originally described from the limestone barrens of
Newfoundland, but plants with similar form have been found
to occur sporadically outside Newfoundland (Atwood 1985;
Luer 1975; Sheviak 1983) from as far away as British
Columbia (Correll 1950). Atwood (1985) recognizes this
taxon as a variety whereas Sheviak (1992) considers it to be

an extreme expression of var. pubescens which does not

67
warrant varietal status.

In this paper the results of an electrophoretic analysis
of vars. pubescens, parviflorum, and planipetalum from the
northeastern United States and southeastern Canada are
reported. This analysis was conducted to determine whether
vars. parviflorum, pubescens, and planipetalum show levels
of genetic differentiation that are most commonly associated
with congeneric species or conspecific taxa. Furthermore,
this study addresses the following additional genetic
questions: (1) Are there differences among these varieties
in their overall levels of genetic variation or the
proportion of genetic variation distributed among
populations? (2) How genetically similar are populations
which contain pure morphological forms of vars. pubescens
and parviflorum to populations that are intermediate in
morphology to vars. pubescens and parviflorum or contain
potential hybrids of these varieties? (3) Do the allele
frequencies found within some sympatric populations of vars.
parviflorum and pubescens indicate that these varieties

maintain distinct gene pools in sympatry?

MATERIALS AND METHODS
Allozyme electrophoresis was carried out on 554
individuals from 22 populations. The samples included seven
populations of var. pubescens (populations A through G,
Table 2.1), four populations each of var. parviflorum (H

through K), and var. planipetalum (L through 0), and seven

68

 

.mmm 2me KHz ..ou coummafl>sq be h ---
mam mmm 2mme “Hz ..oo mHmH msvmmum mm H ---
0mm mam 2me 1H2 ..oo mmmmmsmssm mm m >mmm
mmm mmm Zea Nmo ..ou communes om o ---
omm 3mam ZomB “AH ..00 coHHHflEHo> om m ---
z.eaomw z.m¢omm “mo ..00 cuoflom om m ---
Hmm 3mmm Zome “Hz ..oo mmmumm om . o ---
.3mm Zens 12H ..oo oxen om o ---
Ham Rem Z¢mB u iiiiiiiiiiiiiiiiii 0N m it:
mm sum Zame “Hz ..oo xao>oanmcu on a mom
noaucooq coamaum Heaven 00x09
mauava>fiuaH coconomom Huofimoaonmuoz
noaucaamom

 

.haao mmflmoHoanoe musecmEHousfl nua3

mamscw>flpcfi mo monomonm on» oumoflpnfl .m cm >9 cmkoHH00 .coumfla mxmu o3u suwz msoflumHsmom
.ucomonm mowmoHoanoe oumflcoshmucfl cam mxmu HMOflmoHonmuos noon nus3 mnofiumazmom mumoficcfl
.m 0 >9 meOHHoM .pmumfla mxmu 030 sues mcoflumasmom
mpfihnxn m>flumudm msflsfimuaoo mcofiumHDQOQ cam .Az¢qmv amflmmwmflmMAm .Hm> mmflmMMAww .U .A>mmmv

ammmdwwmmmm .Hm> mmdedew

.xuflucm HMUHmoHOQQHOE 0:0 awn» 0008 Ho

.0 .AmDmv mmwmmwmmm .HMb mmflmMMANM .u no mcoflumHsmom How

HOUUOH MUGGHUHGH UGM .UmHm—bmm mHMSUH>flUGH MO HOQESG .GOHUMEHOMGH COHUMUOQ .H.N mqmdh.

00Hm mans um c0>u0mno 0003 ammmdwflwmmm cam mmwwmwmmm .mnm> Cu 000flp0eu0ucfl
m0flmoHO£QHOE sufl3 mamdcfl>flpsfi on “muoflumHsmom oanummshm 0H0 O cum 0 mGOHumHsmom n

Hou0flnmonm 00 000300“ com: 0H0nuH3 msoflumooH somxm a

 

69

emm 3mm 2mme 1H2 ..00 cmmxonmno om z m-mom\>mmm
.mHm Zeb “Hz .00 smnmeH om D m-mpm\>m¢m

xmm mcflsmzoDHcmz mo cu0 m
ccmHmH :AHsOUHcmz «momcmo .oHnmuqo 0H s m-mom\>m¢m
mmm mmm 2mme NHz ..oo mamH 0580000 as m m-mom\>m¢m
cam mom ZNSB . ................. on m m-mom\>m¢m
............................... om no mom
.mem zHee 1H2 ..oo .msmddflno on an >m¢m

xmm 0Qumm .um
00 2 Ex m.m A uuuuuuuuuuuuuuuuuuuu on 0 it-

0suoz mono .0003 Hanna
G00H00 G00HU a iiiiiiiiiiiiiiiiiiii cm 2 in-

ccchH sx H.0H a .................... om z ---
0>OU m0flpcm Eoum

oceans ex m.mH “momemo .cemH005003mz ma A 2440

.mmm ZNB a .................. om x ---

Ac.ucoov H.N manna

 

70
"mixed" populations which could potentially consist of
introgressed or hybridized individuals. Determinations of
vars. pubescens and parviflorum followed the character
descriptions of Case (1987) whereas delimitation of var.
planipetalum followed the concept of Luer (1975).
Populations chosen to represent each variety were from
collection sites where all the individuals present in the
population could unambiguously be identified to the same
variety. Moreover, these sites were chosen in locations
where additional varieties of Q. calceolus were not commonly
found. These populations are referred to as "pure"
populations. If any population could not easily be
determined to variety or if more than one variety was
present, the population was either excluded from the study
or classified in one of the mixed categories. Mixed
populations were categorized on the basis of the
morphological forms and the degree of intermediacy present
at the site. Three different categories of mixed
populations were observed: (1) populations composed of
individuals with "pure" and intermediate morphologies
(populations R, T, and U; Table 2.1), (2) populations with
individuals having intermediate morphologies only (S and W),
and (3) sympatric populations with only pure forms of each
variety present and no intermediate morphologies (P and Q).
Voucher specimens were collected for the vast majority of
populations and are held at MSC.

The number of individuals sampled per population ranged

71
from 10 to 88. Larger sample sizes were obtained from
larger populations whereas sample sizes under 20 correspond
to smaller populations in which a sample was obtained from
every individual observed. Sampling occurred uniformly
throughout the populations and included collections from
flowering and non-flowering individuals in pure populations.
Since plants with only juvenile or adult vegetative
morphologies could not be determined to variety, samples
were taken only from flowering individuals in the mixed
populations. No more than one sample was taken from a
cluster of individuals suspected of being clonally
reproduced. For each plant sampled, one leaf tip
(approximately 3 cm long) was removed, divided in half, and
placed in two 1.5 ml centrifuge tubes. The samples were
kept on water ice in the field and subsequently frozen at
-80 C in the laboratory until electrophoresis.

Leaf tissue was ground with a Tris-HCl extraction buffer
following the protocol of Gottlieb (1981a). Three buffer
systems were used in combination with nine staining systems.
Glutamate oxaloacetate transaminase (GOT), triose-phosphate
isomerase (TPI), alcohol dehydrogenase (ADH), and superoxide
dismutase (SOD) were resolved on a lithium-borate system
(Crawford 1982), whereas malate dehydrogenase (MDH),
phosphogluconate dehydrogenase (PGD), isocitrate
dehydrogenase (IDH) and shikimate dehydrogenase (SKD) were
resolved on a histidine system (Gottlieb 1981a). A

sodium-borate system (Crawford 1982) was used for

72
phosphoglucomutase (PGM). The staining protocols for all
but GOT followed the agarose overlay procedures of Soltis et
al. (1983). A stain bath was used for GOT which followed
the procedure of Crawford (1982). Other details regarding
the electrophoretic procedure of Q. calceolus have been
described elsewhere (M. Case, Part 1).

Genotypes were inferred directly from enzyme phenotypes by
knowledge of the minimum number of isozymes present for
given enzymes together with a knowledge of the quaternary
structure of those enzymes (Gottlieb 1981b, 1982; Weeden and
Wendel 1989). Side-by-side comparison of leaf and pollen
electromorphs enabled a distinction of allozymes from
isozymes (Weeden and Gottlieb 1979). Enzymes which did not
conform to the above expectations were dropped from the
analyses. In cases where two isozymes were present, the
fastest anodally migrating isozyme was given the designation
1, and each successively slower isozyme was given a
sequentially higher number. Likewise, the allele
corresponding to the fastest migrating allozyme was
designated a, the second fastest was designated b, etc.
Interpopulation verification of enzyme mobilities was
conducted through side-by-side comparisons of allelic
variants electrophoresed on the same gel with the same batch
of extraction buffer.

Four measures of diversity were calculated: percent
polymorphic loci (P), alleles per locus (A), and the

diversity statistics of Nei (1973) and Wright (1984).

73

Percent polymorphic loci was calculated for each population
(Pp), variety (Pv), and species as a whole (Ps) by dividing
the number of loci which had at least two alleles present by
the number of loci analyzed. Likewise, alleles per locus
(A) was calculated for each population (Ap), variety (Av),
and the species as a whole (As) by dividing the total number
of alleles present by the total number of loci analyzed.
Nei’s (1973) diversity statistics, unbiased for sample size
(Nei 1978), were calculated over all loci with GENESTAT
(Whitkus 1985) for each variety and the species. Reported
here are Ht, the total genetic diversity or expected
heterozygosity, and Gst, the proportion of the total genetic
diversity attributable to among-population differentiation.
The standard error for Gst was calculated by the jackknife
method described in Weir (1990). In addition, a
hierarchical analysis of population differentiation (Wright
1984) was employed to determine the proportion of variation
distributed within populations, among populations within
varieties, and among varieties. These values were
calculated with BIOSYS-l (Swofford and Selander 1989), but
expressed in the form of Nei (1973) where each diversity
component represents a proportion of the total diversity.

Nei’s unbiased genetic identity (Nei 1978), Rogers
distance (Rogers 1972), and Cavalli-Sforza and Edwards arc
distance (Cavalli-Sforza and Edwards 1967) were calculated
for all populations (pure with mixed) with BIOSYS-l.

Several clustering techniques were calculated by the above

74
program and applied to the identity and distance measures.
These included unweighted pair-group method analysis
(UPGMA), weighted pair-group method analysis (WPGMA), single
linkage, and complete linkage. Nei’s genetic identity (I)
with UPGMA resulted in the lowest percent standard deviation
(Fitch and Margoliash 1967) and highest cophenetic
correlation coefficient (Sokal and Rohlf 1962) and are
reported herein.

In addition to UPGMA, a principal component analysis
calculated by NTSYS (Rohlf 1988) was applied to the pure
populations to evaluate the existence of potential
groupings. For this analysis, population allele frequencies
were standardized for populations (unit variance, zero mean)
and a variance-covariance matrix was produced from the
standardized allele frequencies. The first three principal
component axes were extracted and populations A through 0
were plotted.

Contingency table analysis was completed for each locus
with BIOSYS—l to determine the extent of allele frequency
heterogeneity among varieties and among populations within
varieties. For intervarietal comparisons the analysis was
conducted on pooled population allele frequencies and on the
allele frequencies of the sympatric populations P and Q.

All populations were tested for deviations from
Hardy-Weinberg equilibria by calculating exact probabilities
with BIOSYS-l. This procedure uses population allele

frequency estimates to calculate the probability that a

75
given genotypic array at a single locus will occur under the
assumption of random mating. Lastly, the possibility of a
correlation between Nei’s (1978) genetic distance and
geographic distance was tested for all intravarietal
populations using Spearman’s Rank correlation (Sokal and

Rohlf 1981).

RESULTS

Nine staining systems resolved the products of 17 putative
loci, 12 of which were used in this analysis. Three
isozymes were found for GOT and MDH although GOT-3 was not
used due to poor resolution and MDH-1 and MDH-2 were not
used because of genetically equivocal isozyme patterns. Two
isozymes were found for TPI, PGM, IDH, and PGD but low
activity in PGD-1 and IDH-l resulted in the omission of
these isozymes from the analysis. One isozyme was found for
SOD, SKD, and ADH. All individuals were scored for each of
the 12 reliable isozymes. Only two loci, GOT-1 and SOD,
were monomorphic in all populations examined. MDH-3 was
polymorphic (with a rare allele) only in populations
consisting of more than one variety.

At the species level, Qypripedium calceolus had 75% of its
loci polymorphic with an average of 63% of the loci
polymorphic in populations (Table 2.2). Similar values were
found in each of the three varieties (i.e., 75% of the loci
were polymorphic in each variety, and population-level

polymorphism ranged from 60% in var. pubescens to 69% in

76

 

 

1000.00 000.0 1000.00 000.0 100.00 «0.0 100.00 00.0 000
0040 04 000. mm

Am00.00 mm0.0 1000.00 000.0 100.00 00.0 100.00 00.0 2000

1000.00 000.0 1000.00 000.0 100.00 00.0 100.00 00.0 >000

1000.00 000.0 1000.00 000.0 100.00 00.0 100.00 00.0 000

1.0.0. 000 ..0.00 00 .>00 04 .>00 00 00000

 

>00000>00 0.002 000 0.0.00 000000 00000000 0000x0000

.0000000000m 00 000 0000000000

.00000 00000000> 0000005000

00080 00 000005000000 00000000> 00000 000 00 0000000000 000 000 .A0mv >00000>uo0000£

00000QX0 00>0HI0000000 00 -0000000> 00000 000 0000 000000000

.10000 00003 0 00

0000000 000 000 0:0 .Az¢dmv amflmwwmmeflm .A>m<mv ammmdwwfimmm .Ambmv mmwmmwmmm .000> 000

0000 000 00>0 0000050000 0003 0000000000 >00000>00 0.002

.000 00>00 0000003000 00000>0

000 000 .000>00 A00 0000000 000 A>V 0000000> 000 000 00>0m 000 Adv 05000 000 0000000

000 Amy

0000 00£Q0osx0om 0000000

.mmflmMWAMw .0 000 00000008 >00000>00

.N.N mqmflfi

I-‘hu

 

77
var. parviflorum). An average of 2.5 alleles per locus was
observed at the species level (Table 2.2). Among the
varieties, var. parviflorum had the highest number of
alleles per locus (Av = 2.33), followed by vars. pubescens
and planipetalum (Av = 2.25 for each). One unique allele
was found in var. pubescens, two were found in var.
parviflorum, and one was found in var. planipetalum (Table
2.3). The average frequencies of these rare unique alleles
were 0.004, 0.004, and 0.006 in each of the taxa,
respectively. Furthermore, a large percentage of
high-frequency alleles was shared among the varieties. The
percentage of shared alleles ranged from 69% among vars.
parviflorum and var. planipetalum to 81% among vars.
parviflorum and var. pubescens, with the average allelic
frequency of unshared alleles ranging from 0.003 to 0.086
(Table 2.3).

The highest varietal-level expected heterozygosity was
found in var. parviflorum (0.274) followed by var.
planipetalum (0.233), and var. pubescens (0.180; Table 2.2).
The majority of this variation resides within populations
for each variety. Variety planipetalum is least structured
with only 5.3% of its variation distributed among
populations, and var. pubescens has a similar value of 8.4%.
Variety parviflorum, however, has a two-fold increase in
structure with 17% of its variation distributed among
populations (Table 2.2). ‘The hierarchical analysis of g.

calceolus revealed that 81% of the total species-level

78

 

 

. . mmo.o 0b 2400\mD0
- . mmo.o m0 2000\>000
. . moo.o 0m mD0\>m00
000.0 0 u 1 2400
woo.o m u . >040
000.0 0 u u 080
0000004 0000004 0000004 00000000 0000004 0000000500
00000D 000002 00 000000000 000000 00 0000B
«0 000000000 00 000802 0000004 0000o>4 0000000
00000>4

 

00000000> 000003 0000000 000000 000

00000000> 00080 0000000

.3333

000000 00 000000

.m.N m0m€B

79
variation is found within populations, 12% of the variation
is within varieties, and only 7% is distributed among
varieties.

Inter- and intravarietal values of Nei's genetic
identities were similar. The means for intravarietal
population comparisons ranged from 0.920 to 0.979, and the
intervarietal comparisons ranged from 0.900 to 0.954 (Table
2.4). The lowest value occurred between a
parviflorum-pubescens comparison (I = 0.747) whereas the
highest value was among two populations of var. pubescens
(I = 0.999). The UPGMA revealed that populations of var.
pubescens and var. planipetalum tended to cluster on
separate branches of the phenogram (Figure 2.1). Population
D was the only pubescens population that did not cluster
with the others. Although parviflorum populations also
tended to group together, these populations have a large
variance in their genetic identity values. For example,
population J and the remaining populations have an average
identity value of only 83% compared to the next highest
clustering value of approximately 93%. Despite this
relatively low genetic identity with other populations, no
apparent morphological or ecological aberrations were
observed in population J. In fact, it was morphologically
and ecologically very similar as well as geographically
close to parviflorum population K (Table 2.1). Similarly,
no significant geographic association with genetic distance

was found for populations of var. parviflorum (rs = -0.31; p

80

TABLE 2.4. Nei’s genetic identity means and ranges for

 

 

intra- and intervarietal comparisons of Q. calceolus.
Comparison Mean Range

Intravarietal:

PUB 0.978 .952 - 0.999

PARV 0.920 .839 - 0.988

PLAN 0.979 .971 - 0.995

MEAN 0.959 .839 - 0.999
Intervarietal:

PARV/PUB 0.900 .747 - 0.980

PARV/PLAN 0.900 .842 - 0.963

PUB/PLAN 0.954 .887 - 0.984

MEAN 0.918 .747 - 0.984

 

81

FIGURE 2.1. UPGMA phenogram of Nei’s genetic
identity values for all pairwise population comparisons
of pure var. pubescens (PUB), var. planipetalum (PLAN),
var. parviflorum (PARV), and mixed populations
(indicated by two-taxon abbreviations separated by a
slash). In the mixed populations, an H indicates
populations that are only composed of morphologies
intermediate to the extreme manifestations of the
varieties, whereas a B represents populations with both
varieties and intermediates present. The last letter

is a key to each population referenced in Table 2.1.

 

. . Do. PI .‘rflaflu EU
... §

82

908 Q

PUB c

903 P
PARV/PUB-H w
903 3

PUB A

PUB c

PUB a
PARV/PUB-H 3
pm: 1.

PL»! u

Putt 11

PLAN 0

Pm a

PARV I
PARV/PUB-B u
PARV x

PUB D

Pm P
PARV/PUB-B '1'
PARV/PUB-B R
Pm J

«in

A A A A J
V Y t V T T

.83 .87 .90 .93 .97 1.00

1b

L A
‘

4»

GENETIC IDENTITY

83
> 0.1), or populations of var. planipetalum (rs = 0.23; p >
0.1). A significant positive correlation was found for var.
pubescens (rs = 0.50; p < 0.025) but this result should be
interpreted with caution. The removal of populations B, C,
D, F, or G cause this correlation to become nonsignificant.
Similar to var. parviflorum populations, geographically
close pubescens populations (e.g., populations A and B;
Table 2.1; Figure 2.1) do not necessarily have the closest
genetic identities.

The mixed populations are scattered throughout the
phenogram (Figure 2.1). Populations with only hybrid
morphologies present (i.e., populations W and S) clustered
with var._pubescens whereas populations with both varietal
forms and intermediates present (i.e., populations U, T, and
R) clustered with either pure parviflorum or mixed
populations. Population Q (a pubescens population which is
sympatric with P) clustered with var. pubescens. Population
P (var. parviflorum) clustered in a group of mixed
populations which also included the one pubescens outlier
population. The phenogram had a cophenetic correlation
coefficient of 0.868 and a percent standard deviation of
2.39.

Principal components analysis revealed a similar
dispersion of pure populations (Figure 2.2). Populations of
var. pubescens formed the tightest cluster and population D
was the moSt divergent of all populations within this

variety. Likewise, populations of var. planipetalum tended

84

>
0
+3”
a:
4;-

 

 

 

PCA 3
~’"n
:—>

 

 

 

 

 

 

 

Z'I'T'us ‘/¢

PCA I 7 <29}

FIGURE 2.2. Pure populations of vars. pubescens
(triangles), parviflorum (circles), and planipetalum
(squares) plotted in relationship to the first three
principal component axes (PCAl, PCA2, and PCA3) of a
principal components analysis of allele frequencies. The
first three axes together account for 66% of the total
variation in allele frequencies. Population letters refer

to populations listed in Table 2.1.

85
to cluster together whereas populations of var. parviflorum
showed the highest level of dispersion. Population J was
the most divergent population as it was with UPGMA. The
principal component axis 1 (PCA 1) is most highly correlated
with variation at TPI-2, PGD-2, and IDH-2 and represents
33.28% of the total variation (Table 2.5). Axes 2 and 3
represent an additional 21.31% and 11.75% of the total
variation, respectively. Axis 2 is most highly correlated
with variation at PGM-la, PGM-ld, and SKD-1a, whereas axis 3
has the highest loadings from PGM-lb, GOT-2a, PGD-2a,
SKD-1c, and SKD-d. The first seven eigenvalues were greater
than one and represent 90.3% of the total variation.

For the vast majority of single-locus contingency table
analyses, significant allele-frequency heterogeneity (p <
0.05) was found for comparisons within and among the
varieties (PGD-2 was the only locus that did not show a
significant mean difference among the varieties). Within
var. planipetalum, four loci did not show significant
differences among populations in allele frequencies (i.e.,
TPI-2, PGM-2, ADH, and PGD-2). The only other locus that
did not display significant heterogeneity was ADH in var.
parviflorum. Four loci showed significant allele frequency
heterogeneity among the sympatric populations P and Q (Table
2.6). Most loci in most populations did not show
significant deviations (p < 0.05) from Hardy-Weinberg
expectations. In pure populations, significant deviations

occurred only in five cases distributed across three

86

TABLE 2.5. Allelic correlations with the first, second,
and third principal components, and the percent of variation

accounted for by each component (percent trace).

 

Principal Component

 

 

Allele l 2, 3
TPI-lb 0.592 -0.447 -0.237
TPI-lc -0.592 0.447 0.237
TPI-2a -0.860 0.284 0.016
TPI-2b 0.860 -0.284 -0.016
PGM—la 0.094 0.766 -0.305
PGM-lb 0.647 -0.221 0.529
PGM-lc -0.486 -0.674 -0.080
PGM-ld 0.082 0.835 —0.062
PGM-2a 0.727 0.286 -0.012
PGM-2b -0.727 -0.286 0.012
GOT-2a 0.659 0.012 -0.542
GOT—2b 0.705 -0.261 0.477
GOT-2c -0.784 0.050 0.396
ADH-a -0.656 -0.401 0.300
ADH-b 0.668 0.368 -0.318
ADH-c 0.329 -0.293 —0.431
ADH-d -0.214 0.002 0.142
ADH-e 0.082 0.602 0.184
PGD-2a 0.065 0.348 -0.519
PGD-2b 0.445 0.544 0.182
PGD-2c 0.829 -0.347 0.048
PGD-2d -0.464 -0.545 —O.155
IDH-2a -0.831 0.122 —0.476
IDH-2c 0.832 -0.124 0.476
SKD-a 0.003 -0.899 —0.262
SKD-c -0.045 0.718 0.602
SKD-d 0.080 0.490 -0.595

Percent Trace 33.28 21.31 11.75

 

87

TABLE 2.6. Results from a chi-square contingency table
analysis for allele frequency heterogeneity among varieties
parviflorum (population P) and pubescens (population Q) at a
sympatric site on Drummond Island. Reported here are the
four loci that show significant heterogeneity, their allele

frequencies, and associated p values for each significant

 

 

locus.

Locus and Allele PARV PUB p
TPI-2a 0.475 0.925
TPI-2b 0.525 0.075 < 0.000
GOT—2a - 0.350
GOT-2b - 0.100
GOT-2c 1.000 0.550 < 0.000
PGD-2b 0.525 0.800
PGD—2d 0.475 0.200 0.009
SKD-a 0.525 0.275
SKD-c 0.350 0.600
SKD-d 0.125 -
SKD-e - - 0.125 0.001

All Loci Tested < 0.000

 

88
populations and three loci. These have been reported
elsewhere (M. Case, Part 1). Only one additional deviation

was found (PGM-2 in population R).

DISCUSSION

As seen in Table 2.4, only a slight difference was found
between the average intravarietal (I = 0.959) and average
intervarietal (I = 0.918) genetic identity values. This
indicates that, on average, populations within a variety are
only slightly more genetically similar than populations
belonging to different varieties. There is, however,
considerable overlap in identity values. For example, the
mean genetic identity of pubescens-planipetalum comparisons
is higher (I = 0.954) than the mean genetic identity of
populations within var. parviflorum (I = 0.920). The
relatively large variance in genetic identities among
parviflorum populations can be visualized in the UPGMA
phenogram (Figure 2.1) as well as in a plot of these
populations on the first three principal component axes
(Figure 2.2). For both techniques, pure populations of var.
parviflorum (populations H through K) show a high level of
dispersion. Most striking is population J which was
geographically close to and morphologically and ecologically
very similar to population K. This large difference was
due, in part, to IDH-2 which had an uncommon allele in high
frequency in population J, and also due to TPI-2, PGM-lb,

and PGM-2, all of which displayed aberrant allele

 

 

89
frequencies in this population (tables of populational
allele frequencies are available from the author upon
request). In contrast, there was substantially less
intravarietal differentiation among populations of vars.
pubescens and planipetalum. Population D is a notable
exception to this trend since it contained frequencies that
were unlike other pubescens populations at GOT-2c, ADH-b,
and PGD-2b.

The overall genetic identity values among the varieties
are high in comparison to the mean value of 0.67 commonly
reported for congeneric taxa (Crawford 1983; Gottlieb 1977,
1981b). Furthermore, these identity values are very high
when compared to identity values reported for other
Cypripedium taxa (I = 0.00 - 0.79; M. Case, Part 1).
However, these values are typical of those reported for
conspecific populations (Gottlieb 1977) including population
comparisons among infraspecific taxa (Crawford and Smith
1984). For example, Crawford and Smith (1984) found that
four morphologically variable varieties of Coreopsis
grandiflgra Hogg ex Sweet exhibited intravarietal values
which ranged from 0.81 to 0.99 and intervarietal values
which similarly ranged from 0.79 to 0.99. A similar lack of
allozyme divergence among infraspecific taxa has also been
documented for taxa within Ipomopsis aggregata (Pursh) V.
Grant (Wolf et a1. 1991), Gaillardia pulchella Foug.
(Heywood and Levin 1984), and others reviewed in Crawford

and Smith (1984). In contrast, relatively low genetic

90

identities have previously been documented for some other
infraspecific taxa. In varieties of Allium douqlasii Hook.
(Rieseberg et al. 1987) and Coreopsis cvclocarpa Blake
(Crawford and Bayer 1981), mean intervarietal genetic
identities were 0.84 and 0.75, respectively, compared to
mean intravarietal values of over 0.95 for both species.
Even lower values were found among ecotypes of the orchid
Gymnaedenia conopsea (L.) R. Br. (Scacchi and De Angelis
1989). Typical conspecific identity values of 0.98 and 0.96
were found within a dry and wet ecotype of this species,
respectively, whereas the average interecotype genetic
identity was only 0.65. Although infraspecific taxa can
exhibit a wide range of genetic identity values, low values
are most commonly associated with either allopatric or
ecologically isolated taxa (Heywood and Levin 1984;
Rieseberg et al. 1987) as in the above examples. This
result suggests, as Heywood and Levin (1984) have discussed,
that large divergence in allozyme frequencies occurs
predominantly after the severance of gene flow among taxa
(i.e., after "speciation" as defined by Heywood and Levin
1984). Therefore, the lack of large allozyme differences
among Qypripedium varieties is typical of sympatric
infraspecific taxa which experience gene flow.

The lack of ubiquitous, moderate- to high-frequency unique
alleles within these varieties of Qypripedium (Table 2.3) is
another indication that they are genetically similar and

have recently experienced gene flow. The differences that

91
contributed to the structure in the phenogram were allele-
frequency differences among the varieties. Although there
is often great overlap in frequencies, the alleles at TPI-l,
and GOT-2c showed the greatest mean differences among
parviflorum and pubescens populations. Other moderate
differences were found between vars. parviflorum and
planipetalum at SKD-a and TPI-l, and at PGM-lc for vars.
pubescens and planipetalum (Table 2.7). The overall
similarities of these taxa is further reflected by the
hierarchical F-statistics which revealed that the largest
amount of variation within the species is contained within
individual populations (81%). Of the variation that is
distributed among populations, 12% was found within
varieties and the smallest portion (7%) was distributed
among varieties.

Although the general lack of large allozymic differences
among the varieties creates some ambiguity with the
interpretation of the mixed populations, the phenogram
(Figure 2.1) suggests several trends. The first involves
the status of populations W and S. These populations
consisted entirely of plants with morphologies intermediate
to the extreme manifestations of vars. parviflorum and
pubescens. According to the concept of Case (1987) these
plants fit well within the variation described for var.
pubescens, which is generally considered to be more
morphologically variable than var. parviflorum. The

allozyme data also suggest that these populations with

92
TABLE 2.7. Allele-frequency means for polymorphic loci
(weighted according to population sizes), population ranges,

and number of individuals sampled for varieties of Q.

 

 

calceolus.
Taxon and
Number of Individuals Sampled
Locus and PARV PUB PLAN
Allele n a 207 n . 140 n s 78
TPI-lb 0.572 0.093 0.186
(0.295 - 0.676) (0.000 - 0.350) (0.025 - 0.306)
TPI-lc 0.428 0.907 0.814
(0.324 - 0.705) (0.650 - 1.000) (0.694 - 0.975)
TPI-2a 0.614 0.907 0.821
(0.298 - 0.835) (0.675 - 1.000) (0.750 - 0.875)
TPI-2b 0.386 0.093 0.179
(0.165 - 0.702) (0.000 - 0.325) (0.125 - 0.250)
PGM-la 0.002 — 0.308
(0.000 - 0.010) - (0.200 - 0.500)
PGM-lb 0.114 0.057 -

(0.000 - 0.489) (0.000 - 0.175) -

PGM-lc 0.884 0.940 0.577
(0.511 - 1.000) (0.825 - 1.000) (0.389 - 0.750)

PGM-ld - 0.003 0.115
- (0.000 - 0.025) (0.050 - 0.225)

PGM-2a 0.266 0.186 0.295
(0.108 - 0.511) (0.050 - 0.450) (0.222 - 0.350)

PGM-2b 0.734 0.814 0.705
(0.489 - 0.892) (0.550 - 0.950) (0.650 - 0.778)

GOT-2a 0.582 0.261 0.532
(0.375 - 0.850) (0.025 - 0.475) (0.333 - 0.700)

GOT-2b 0.053 0.014 -
(0.000 - 0.202) (0.000 - 0.100) -

93
Table 2.7 (cont'd)

GOT-2C 0.365 0.725 0.468
(0.150 - 0.625) (0.525 - 0.975) (0.300 - 0.667)

ADH-a 0.022 0 114 -
(0.000 — 0.045) (0.025 - 0.250) —

ADH-b 0.976 0.882 0.994
(0.955 - 1.000) (0.750 - 0.975) (0.975 - 1.000)
(0.000 — 0 010) - -

- (0.000 - 0.025) -

- - (0.000 - 0.025)

PGD-2a - - 0.006
- - (0.000 - 0.025)

PGD—2b 0.841 0.861 0.917
(0.682 - 0.979) (0.550 - 1.0000 (0.875 - 1.000)

PGD-2c 0.005 - -
(0.000 - 0.011) - —

PGD-2d 0.154 0.139 0.077
(0.011 - 0.318) (0.000 - 0.450) (0.000 - 0.125)

IDH-2a 0.780 0.964 0.942
(0.245 - 0.983) (0.775 - 1.000) (0.800 - 1.000)

IDH-2C 0.220 0.036 0.058
(0.017 - 0.755) (0.000 - 0.225) (0.000 - 0.200)

SKD-a 0.606 0.368 0.109
(0.372 - 0.773) (0.175 - 0.625) (0.000 - 0.200)

SKD-C 0.394 0.632 0.712
(0.227 - 0.628) (0.375 - 0.825) (0.425 - 0.975)

SKD-d - - 0.179
- - (0.025 - 0.425)

 

94
intermediate-looking plants have the genetic constitution of
var. pubescens. It should be noted, however, that
individuals in these populations were morphologically
uniform as compared to other "mixed" populations where both
varieties were present in addition to plants with
intermediate morphologies. In the latter case, the
intermediates often display much morphological variation.
This is the situation in populations U, T, and R, which
consisted of extreme morphologies of var. parviflorum and
var. pubescens in addition to a continuum of forms between
the extremes. All of these populations clustered away from
the majority of pubescens and were most like other
parviflorum populations or other mixed populations of the
same type. This is suggestive of a genetic influence from
var. parviflorum. Some of the most interesting results in
this study concern the sympatric populations P and Q.
Individuals from these populations appeared to be randomly
intermixed in the field with no indication of morphological
intermediacy. In the phenogram, they each cluster in the
vicinity of their respective morphological varieties (Figure
2.1), and they display significant allele-frequency
differences at four loci (Table 2.6). This result is highly
suggestive of restricted gene flow between populations P and
Q and supports the observations of Newhouse (1976), Atwood
(1985), Case (F. Case, pers. comm.), and others that, in
some geographic areas, these two varieties display

characteristics typical of species. Investigation of

95
population structure and evolutionary processes in these and
other populations of vars. pubescens and parviflorum are
ongoing.

All varieties of Q. calceolus contain species—level
expected heterozygosity values which are higher than the
average reported for other plant species (Ht = 0.149;
Hamrick and Godt 1989), and generally higher or very near
the average values reported for plants with either a
widespread distribution (Ht = 0.20), long-lived herbaceous
perennial habit (Ht = 0.20), or an animal-outcrossed mating
system (Ht = 0.16; Hamrick and Godt 1989). Although a high
level of genetic variation accompanies a large level of
morphological variation in this species, the taxon often
cited to have the most morphological variation (i.e., var.
pubescens) is the taxon with the lowest overall level of
genetic variation (Ht = 0.18; Table 2.2). Another
difference among the taxa is the proportion of variation
that is distributed among populations. Varieties pubescens
and planipetalum populations have similar Gst values of
0.084 and 0.053, but var. parviflorum populations are
approximately twice as structured with a Gst value of 0.171
(Table 2.2). Hamrick and Godt (1989) found that a
long-lived perennial habit and an animal-outcrossed
pollination system were the two traits which accounted for
the largest proportion of variation in Gst values in plant
species. However, since varieties of g. calceolus do not

differ qualitatively for these traits, this is an unlikely

96
explanation for the observed intervarietal variation in Gst
values. One hypothesis which could account for this
variation is that there are differences among the taxa in
the amount of intravarietal gene flow. Since var.
parviflorum is most often associated with wet environments
(Case 1987) and is more ecologically restricted than var.
pubescens, it is likely that parviflorum populations are
more geographically isolated and therefore occur in more of
a mosaic pattern than populations of var. pubescens, which
occupy a diverse array of habitats. Genetic drift could act
to increase genetic variance among parviflorum populations
relative to pubescens populations which would be reflected
in higher Gst values for var. parviflorum. Likewise, the
relatively high degree of population similarity among
planipetalum populations could be due to higher levels of
intravarietal gene flow for this taxon. The occurrence of
neighboring populations of var. planipetalum in close
proximity on treeless, windswept, limestone barrens in
Newfoundland could produce higher levels of intravarietal
gene flow relative to the more geographically-isolated
populations of var. parviflorum.

Although var. parviflorum has a higher Gst value than
either var. pubescens or var. planipetalum, var. parviflorum
is most similar to the averages reported for either long-
lived herbaceous perennials or animal-outcrossed plants (Gst
= 0.213 and 0.197, respectively; Hamrick and Godt 1989).

Moreover, the Gst values for var. planipetalum and var.

97
pubescens are substantially lower than these mean values.
This could be due to a common origin of the study
populations within each of these varieties. Since all of
the populations in the present study are either north of or
very close to the southernmost border of the Wisconsin
continental ice sheet (Péwé 1983), plants have probably
colonized these areas within the last 11,000 years (Webb et
al. 1983). It is possible that a rapid radiation from a
single source population into the geographic regions
examined in the present study could produce populations
which are genetically similar. This has been proposed for
other Qypripedium species in addition to taxa in other
genera (see review in M. Case, Part 1).

It is evident that Cypripedium calceolus is composed of
populations which have a great potential to evolve with high
levels of heterozygosity and regional variation in the
tendency to form hybrids. In the northeastern United States
and southeastern Canada, small levels of genetic divergence
were found among vars. pubescens, parviflorum, and
planipetalum, and slight differences were observed in levels
of heterozygosity and distribution of genetic variation
among populations within these taxa. These differences
suggest that multiple taxa exist. However, the lack of
high-frequency novel alleles in any of the varieties, the
large overlap in allele frequencies among the varieties, and
the large variance in the allele frequencies found in var.

parviflorum prevent definitive classification of these taxa

98
through allozyme data. Because of this and the general
agreement of these results with other previously-reported
allozyme surveys of sympatric infraspecific taxa, it is
suggested that vars. parviflorum, pubescens, and

planipetalum be considered members of a single species.

99
LITERATURE CITED
ACTOR, G. F. 1984. Natural hybridization of Cypripedium
candidum and C. calceolus. M.S. thesis, University of

North Dakota.

ATWOOD, J. T. JR. 1985. The Cypripedium calceolus L.
complex in North America. Pp. 106—110 in Proceedings of
the eleventh world orchid conference, ed. K. W. Tan.

Singapore: International Press Co.

CASE, F. W. JR. 1987. Orchids of the western Great Lakes
region. Cranbrook Institute of Science Bulletin 48.

Bloomfield Hills: Cranbrook.

CAVALLI-SFORZA, L. L. and A. W. F. EDWARDS. 1967.
Phylogenetic analysis: models and estimation procedures.

Evolution 21: 550-570.

CORRELL, D. S. 1938. Cypripedium calceolus var.

pubescens. Botanical Museum Leaflets Harvard University

............. . 1950. Native orchids of North America north

of Mexico. Waltham: Chronica Botanica.

CRAWFORD D. J. 1982. Electrophoresis. Lecture notes from
Botany 812, third edition, fall quarter. Ohio State

University.

100
-------------- . 1983. Phylogenetic and systematic
inferences from electrophoretic studies. Pp. 257-287 in

Isozymes in plant genetics and breeding, Part A, eds. S. D

Tanksley and T. J. Orton. Amsterdam: Elsevier.

------------- and R. J. BAYER. 1981. Allozyme divergence
in Coreopsis cyclocarpa (Compositae). Systematic Botany

6: 373-379.

-------------- and E. B. SMITH. 1984. Allozyme divergence
and intraspecific variation in Coreopsis grandiflora

(Compositae). Systematic Botany 9: 219-225.

FITCH, W. M. and E. MARGOLIASH. 1967. Construction of

phylogenetic trees. Science 155: 279-284.

GOTTLIEB, L. D. 1977. Electrophoretic evidence and plant
systematics. Annals of the Missouri Botanical Garden

64: 161-180.

-------------- . 1981a. Gene number in species of Astereae
that have different chromosome numbers. Proceedings of
the National Academy of Sciences of the United States of

America 78: 3726-3729.

.............. , 1981b. Electrophoretic evidence and plant

populations. Pp. 1-46 in Progress in phytochemistry, eds.

L. Reinhold, J. B. Harborne, and T. Swain. New York:

Pergamon.

101
______________ , 1982. Conservation and duplication of

isozymes in plants. Science 216: 373—380.

HAMRICK, J. L. and M. J. W. GODT. 1989. Allozyme diversity
in plant species. Pp. 43-63 in Plant population genetics,

breeding, and genetic resource, eds. A. H. D. Brown,
M. T. Clegg, A. L. Kahler, and B. S. Weir. Sunderland:

Sinauer.

HARMS, V. L. 1986. The yellow lady's-slipper varieties in

Saskatchewan and western Canada. Blue Jay 44: 87-95.

HEYWOOD, J. S. and D. A. LEVIN. 1984. Allozyme variation
in Gaillardia pulchella and G. amblyodon (Compositae):
Relation to morphological and chromosomal variation and to

geographical isolation. Systematic Botany 9: 448-457.

HULTEN, E. and M. FRIES. 1986. Pp. 269 in Atlas of North

European vascular plants north of the Tropic of Cancer,

Vol. 1. Kénigstein: Koeltz Scientific Books.

KLIER, K., M. J. LEOSCHKE, and J. F. WENDEL. 1991.
Hybridization and introgression in white and yellow
ladyslipper orchids (Qypripedium candidum and g.

pubescens). Journal of Heredity 82: 305-318.

LUER, C. A. 1975. The napive orchids of the United States
and Canada excluding Florida. New York: The New York

Botanical Garden.

102
NEI, M. 1973. Analysis of gene diversity in subdivided
populations. Proceedings of the National Academy of

Sciences of the United States of America 70: 3321-3323.

------ . 1978. Estimation of average heterozygosity and
genetic distance from a small number of individuals.

Genetics 89: 583-590.

NEWHOUSE, C. J. 1976. Pollination biology in seven taxa of
Michigan Orchidaceae and a study of Qypripedium calceolus
in Michigan based on living plants and herbarium

specimens. M.S. thesis, Michigan State University.

PEWE, T. L. 1983. The periglacial environment in North
America during Wisconsin time. Pp. 157-189 in
Late-Quaternary environments pf the United States, Vol. 1,
eds. H. E. Wright Jr., and S. C. Porter. Minneapolis:

University of Minnesota Press.

RIESEBERG, L. H., P. M. PETERSON, D. E. SOLTIS, and C. R.
ANNABLE. 1987. Genetic divergence and isozyme number
variation among four varieties of Allium douglasi

(Alliaceae). American Journal of Botany 74: 1614-1624.

ROGERS, J. S. 1972. Measures of genetic similarity and
genetic distance. Studies in genetics, University of

Texas Publications 7213: 145-153.

ROHLF, J. E. 1988. NTSYS-pc, numerical taxonomy and

multivariate analysis system. Setauket: Exeter software.

103
SCACCHI, R. and G. DE ANGELIS. 1989. Isoenzyme
polymorphisms in vanaedenia conopsea and its inferences
for systematics with this species. Biochemical

Systematics and Ecology 17: 25-33.

SHEVIAK, C. J. 1983. United States terrestrial orchids
patterns and problems. Pp. 49-60 in North American
terrestrial orchids: symposium II proceedings and
legtures, ed. E. H. Plaxton. Livonia: Michigan Orchid

Society.

------------- . 1992. Natural hybridization between
Qypripedium montanum and its yellow-lipped relatives.

American Orchid Society Bulletin 61: 547-559.

SOKAL, R. R. and F. J. ROHLF. 1962. The comparison of

dendrograms by objective methods. Taxon 11: 33-40.

............................ . 1981. Biometry. San

Francisco: W. H. Freeman and Company.

SOLTIS, D. E., C. H. HAUFLER, D. C. DARROW, and G. J.
GASTONY. 1983. Starch gel electrophoresis of ferns: a
compilation of grinding buffers, gel and electrode
buffers, and staining schedules. American Fern Journal

73: 9-27.

STOUTAMIRE, W. P. 1967. Flower biology of the lady’s-

slippers (Orchidaceae: Qypripedium). Michigan Botanist

6: 159-175.

104
SWOFFORD, D. L. and R. B. SELANDER. 1989. BIOSYS-l, a
computer program for the analysis of allelic variation in
population genetics and biochemical systematics, release

1.7. Champaign: Illinois Natural History Survey.

WEBB, T. III, E. J. CUSHING, and H. E. WRIGHT, Jr. 1983.

Holocene changes in the vegetation of the Midwest. Pp.

142-165 in Late-gpaternary environments of the United
States, Vol. 2, ed. H. E. Wright Jr. Minneapolis:

University of Minnesota Press.

WEEDEN, N. F. and L. D. GOTTLIEB. 1979. Distinguishing
allozymes and isozymes of phosphoglucoisomerases by
electrophoretic comparisons of pollen and somatic tissues.

Biochemical Genetics 17: 287-296.

............. and J. F. WENDEL. 1989. Genetics of plant

isozymes. Pp. 46-72 in Isozymes in plant biology, eds. D.

E. Soltis and P. E. Soltis. Portland: Dioscorides.

WEIR, B. S. 1990. Genetic data analysis. Sunderland:

Sinauer Associates.

WHITKUS, R. 1985. FORTRAN program for computing genetic
statistics from allelic frequency data. Journal of

Heredity 76: 142.

105
WOLF, P. G., P. S. SOLTIS, and D. E. SOLTIS. 1991. Genetic
relationships and patterns of allozymic divergence in the
Ipomopsis aggregata complex and related species

(Polemoniaceae). American Journal of Botany 78: 515-526.

WRIGHT, S. 1984. Genetic variability in natural
populations: methods. Pp. 79-103 in Evolution and the
genetics of populations, Vol. 4. Chicago: University of

Chicago Press.

nICHIan STATE UNIV. LIBRARIES
(“(1)1")(WWI)Ill"(INN)“(1)111)“
31293009101936