‘ 3...
.227.

z ..
.. .tvhllflmlbr v.
£11.. a. dc.

 

47.1. p ‘ ..
:25. r. 2.“.
fiWffiumwéw.

. .... airshflv. '

, , .‘A.
i I. :V.‘
.. 11 n . ..
'3’...”

 

zcfluleha.l1. PItSHN‘... I
13.0HArinuMNDaulh’. %*O.l|lvlna
. . .3299 J I “A. .1“;

v37:
33!. ‘ v

i .... I.

[HM . .1.
.On’; X ‘
.1

‘4” a < ... .él. «I,
.|| .11:
In. 3“

ill;

fi

Ernet‘vivnl .

Kali...) inunnfi
. 1.1.

 

 

 

 

| . .13!!!
I (‘96.; ’53:.
rattan... 0|. (

E

 

(-11"): .
)1..i$ “all"
.iiltsr’tablfV. ,
I.'.’

I};
Stillman-5"}:

TAM" ~ I! v. - .fflafiIlff ”tun"

‘5? V on ‘1. ”I If \hc

‘ flan" f1:

)- ilLbII .r
E" I)!!!“-

.Il-ObI.. I

 

 

:;

lTYLlB

Illilllfllllllllflllflllllltl\llllllllll'llll

3 1293 02058 6461

’7
pk

This is to certify that the

dissertation entitled

Rarity and the Phylogeography of the
Large-Flowered Piptolobi of Astragalus L. (Fabaceae)

 

presented by

Jeffrey Wellington White

has been accepted towards fulfillment
of the requirements for

Ph.D. dualdegree in Botany and Plant Pathology,

 

 

and Ecology, Evolution, and
Behavioral Biology

 

 

. \1
Major professor

Date 33.5.14, WM

 

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

 

LlBRARY

Michigan State
_, University

 

 

 

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11/00 chlRC/DdeOupfiS-ou

RARITY AND THE PHYLOGEOGRAPHY OF THE LARGE-FLOWERED
PIPTOLOBI OF ASTRAGALUS L. (FABACEAE)

By
Jeffrey Wellington White

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

1999

of L'
lhiCI

area

 

 

ABSTRACT

RARITY AND THE PHYLOGEOGRAPHY OF THE LARGE-FLOWERED
PIPTOLOBI OF ASTRAGALUS L. (FABACEAE)

By
Jeffrey Wellington White

Phylogenetic patterns of rarity were explored using a case study group of 51 species
of the Large-flowered Piptolobi of Astragalus (Fabaceae). The study was divided into
three parts: elucidation of a species level phylogeny, quantification of range sizes using
areas of occupancy, and assessment of patterns of rarity in the phylogeny.

A morphological cladistic analysis uncovered a highly resolved phylogeny that
corresponds well with previous taxonomic work. The Argophyllean clade consisting of
47 species was identified and used in part three.

Areas of occupancy using nine measurement scales (10 to 2560 km) were quantified
using a procedure developed to eliminate sampling error associated with standard
methods. The program "Minimum Cell Count" searched for minimum areas of
occupancy among potential measurement grid placements. Categories of rarity using
generally accepted criteria were assigned to each species. Assessment of the fractal
properties of each species' distribution revealed that they are generally not fractal.
Measurement scale significantly affected rank ordering of species.

Phylogenetic patterns of rarity in the Argophyllean clade of Astragalus were assessed
using three lines of evidence aimed to address the main hypothesis: Namely, that high
rates of diversification in concert with local speciation is associated with rare species
clustering in the phylogeny and a preponderance of newly derived rare species.
Significant diversification rate variation within the clade was established based a tree
asymmetry tests. Two tests were conducted to assess patterns of rarity after range size

quantities and rare categories were mapped onto the phylogeny. Rarity is not a

phylu-

[CSIS ‘

differ

sped:
are rr.
amor-

distri‘

 

phylogenetic trait, thus standard comparative methods were not employed; instead, new
tests were developed. Phylogenetic clustering of rare was tested by employing a Mantel
randomization procedure using matrix values for topological distance among species, and
differences in their log area of occupancy. The null hypothesis, that there is no
association among matrix values (thus no clustering), was not rejected (p=.089) but
results point toward evidence of clustering. A Monte-Carlo procedure that randomized
species placement on the phylogeny was employed to test the hypothesis that rare species
are more newly derived than expected by chance. The count of 11 rare species found
among the 14 terminal pairs of species was not significantly greater than the reference

distribution's mean of 9.8 (p=.32).

Copyright
JEFFREY WELLINGTON WHITE
1 999

DEDICATION

in memory of
Richard B. Battelle

Eighth-grade Science Teacher

Mr. Battelle inspired enthusiasm about science and motivated many
to aim for high levels of personal achievement. Mr. B. introduced me to botany

and inspired my interest and commitment to teaching and to public schools.

I wish
contributii
good advi
Tom Gen)
during a ti
feedback 2
herbarium
Tonsor. pl.

Two 01
Murph} gr
and some :2
Challenged
him about (

Mike S,-
A5”0.galus_
Mike Pens};
PTOVided Sui

The Hei'ban'

M} COm
many days 21
gave 568mm
Thank you \\

Fundmg \
Research Tm

ACKNOWLEDGMENTS

I wish to first thank my guidance committee members for their support and
contributions. My advisor, Tao Sang, helped me distill my ideas and offered lots of
good advice and constructive criticism. He was flexible and patient from the beginning.
Tom Getty asked superb questions and took on a crucial role that supplied continuity
during a time of rapid change. Jim Hancock consistently gave upbeat and constructive
feedback and helped me understand subtleties within academe. Alan Prather furnished
herbarium work space, computational facilities, and advice. My former advisor, Steve
Tonsor, played a significant role during the early development of my dissertation work.

Two other professors significantly contributed to my program and work. Peter
Murphy graciously offered excellent advice on many occasions, much encouragement,
and some exceptional opportunities. His support is greatly appreciated. Don Hall
challenged me to think deeply and gave much needed support. I learned a great deal from
him about ecology, evolution, and life. Also, thank you Gus de Zoeten.

Mike Sanderson furnished data and shared his insights on the evolution of
Astragalus. Rupert Bameby offered his gracious help during my visit to New York.
Mike Penskar shared his rich understanding of rare plants. Marguerite Halversen
provided superb editorial assistance and David Wisner gave expert programming help.
The Herbarium staff at NY and RSA assisted during visits and with loans to MSC.

My companion in life, Doreen, gave unending support, love, and tolerance of the
many days and nights I spent on this work. Our two children, Nicolas and Anthony,
gave seemingly endless smiles and always reminded me of what is most important in life.
Thank you Wellington and Colleen White.

Funding was provided in part by the Department of Botany and Plant Pathology, the
Research Training Group at the Kellogg Biological Station, and the Ecology, Evolution,
and Behavior Program—all of Michigan State University.

vi

(A)

TABLE OF CONTENTS

List of Tables

List of Figures

1

A Brief Overview:
Age, Area, and the Evolution of Rare Plants

1.1 INTRODUCTION

1.2 A BRIEF HISTORY OF RARITY CONCEPTS

1.3 SPATIAL A'I'I'RIBUTES OF RARE SPECIES

1.4 RARITY, AGE, AND EVOLUTIONARY HISTORY

1.5 A CASE STUDY OF RARITY FROM A PHYLOGENETIC
PERSPECTIVE

A Morphological Cladistic Study of the

Large-Flowered Piptolobi of Astragalus L.

2.1 INTRODUCTION

2.2 METHODS

2.3 RESULTS

2.4 DISCUSSION

Rarity and the Biogeography of the

Large-Flowered Piptolobi of Astragalus L.

3.1 INTRODUCTION

3.2 STUDY SYSTEM AND METHODS
3.2.1 The Large-flowered Piptolobi of Astragalus
3.2.2 Data handling in the ArcView GIS environment

3.2.3 Minimum cell counts, area of occupancy, geographic extent,
and fractal dimensions

3.2.4 Species richness

3.2.5 Categories of rarity used in this study

vii

ix

10
17
20

2 4
24
26
26
27

33
36
36

(A)

Append;
136C163

AWendi
Grldpoll

Appencii;
Mini-mun

Littl’am TE

TABLE OF CONTENTS

continued

3.3 RESULTS 37
3.3.1 Fixed grid versus minimum cell counts 37

3.3.2 Geographic measures and resulting rare categories 37

3.3.3 Among scale comparisons 50

3.3.4 Fractal geometry of species distributions 50

3.3.5 Species richness 59

3.4 DISCUSSION 62

4 Phylogenetic Patterns of Rarity in the

Argophyllean Clade of Astragalus L. 73

4.1 INTRODUCTION 73

4.2 METHODS 74

4.3 RESULTS 78

4.4 DISCUSSION 82

4.4.1 Basic findings 87

4.4.2 Implications of findings 92

4.4.3 Summary 96
Appendix A

Species codes, Bameby numbers, A 10 rank, A rarity, and E rarity. 98
Appendix B

GridPoly Avenue Program--ArcView GIS 3.): 99
Appendix C

Minimum Cell Count Q-Basic Program 102

Literature Cited 104

viii

3.10

LIST OF TABLES

2 . 1 Species of Astragalus included in study 12
2 . 2 Characters and their states used in the cladistic analysis 13
2 . 3 Taxon by character data matrix 16
3 . 1 Scales, measurement areas, and number of grid cells used in this study 30
3 . 2 Comparison of fixed grid and minimized cell counts for Astragalus helleri 4O
3 . 3 Sites S, minimum cell counts C, and geographic extent E for each 41
specres
3 . 4 Spearman (Pearson) correlations among the three principal geographic 44
measures
3 . 5 Categories of rarity applied in this study 48
3 . 6 Species categorized by geographic extent E and area of occupancy A rarity 48
3 . 7 Spearman (Pearson) correlations among scales for areas of occupancy A 51
3 . 8 Sequential rank of species at each scale based on area of occupancy A 53
3 . 9 Fractal Dimensions D for each species between reference scales 56
3 . l 0 Lagilaction in species richness among cells for scales J = 10 km through 61
m

A . 1 Species codes, Bameby numbers, A 10 rank, andA and E rarity 98

ix

310
3.11
312

3.13

MWNN
Ni-‘N

emhamai»

«BMMMMUUUMWMMNUMM

LIST OF FIGURES

Strict consensus of 12 most parsimonious trees recovered
A randomly chosen tree among the 12 most parsimonious trees
Flow chart showing relations and sequences of data transformation

Distribution of the Large-flowered Piptolobi of Astragalus on an "equal-
area cylindric" map projection

320 km grid overlaying western North America

Grids overlaying Utah

Fixed grid cell count inflation compared to minimum cell counts
The three sites of Astragalus helleri in Mexico

Number of sites and minimum cell counts C10

Area of occupancy A 10 and Geographic extent E

Log area of occupancy A 10 versus Log geographic extent E
Location of rare and very rare species of the Lf-P Astragalus
Pairwise Spearman correlations among scales

Sequential rank changes across scale

Scale-area curves

Fractal dimensions across scales

Utah species richness--80 km scale

Utah species richness-40 km scale

High species richness cells--40 & 80 km scales

Phylogeny of the Argophyllean clade of Astragalus used for analysis in
Chapter 4

Phylogeny of the Argophyllean clade of Astragalus with an example of
one set of pairwise data used in the phylogenetic clustering test

Phylogeny of the Argophyllean clade of Astragalus with positions of the
14 terminal pairs and the 17 rarest species

18
19
28

29
31
32
38
39
45
46
47
49

55
58
60
63
64
68

76

77

79

 

4.4

4.6

4.7

4.8

4.9
4.10

A

Asymmetry tests of the 19 subclades within the Argophyllean clade of
Astragalus

Phylogeny of the Argophyllean clade of Astragalus with significantly
asymmetrical subclades indicated

Bivariate plot of the rarest species versus total species for each of the 46
subclades in the phylogeny

Results of the phylogenetic clustering test: Area of occupancy A and
intemode distance

Results of the phylogenetic clustering test: Geographic extent E and
intemode distance

Number of the rarest species among terminal pair species

Proportion of rare species within monophyletic groups related to the
Argophyllean clade.

xi

80

81

83

84

85
86
91

 

 

l.l

1
A Brief Overview:

Age, Area, and the Evolution of Rare Plants

1.1 INTRODUCTION

Humans are fascinated with rare things. In the case of organisms, an interest in rarity
often corresponds with concern about their persistence, particularly when human
activities are contributing to their decline. The majority of rare species, however, are
probably rare largely for reasons unrelated to human activity. Given the general lack of
understanding about natural causes of rarity, it is not surprising that biologists often have
difficulty elucidating the factors leading to rare status.

After more than a century of study, general theory regarding the causes and
consequences of rarity is just beginning to coalesce. A few general traits are emerging
that differentiate rare taxa from common species, including: 1) breeding systems that tend
away from outcrossing and sexual reproduction; 2) lower reproductive investment;

3) more limited dispersal; 4) higher levels of homozygosity; and 5) a narrower scope of
resource usage (Kunin and Gaston 1997). There are many exceptions to these traits,

however. New methods and data are leading to new insights about the nature of

biodiversity dynamics (McKinney and Drake 1998), a large umbrella under which studies
of rarity are found.

Rarity, here discussed as a biogeographic attribute, is most commonly measured in
terms of range size (area), abundance (number/area), or numbers of populations or
localities. Two general processes lead to rare status: demographic decline or incipient
local speciation (such as via peripatric speciation or chromosomal rearrangements).
Although these two processes are fundamentally different, they are nevertheless difficult
to differentiate and may operate simultaneously.

Rarity is not a genealogical trait like morphological traits, chromosome numbers, or
DNA sequences. Rather, it is a demographic trait—an aggregate property of the species
resulting from its genotype by environment interaction. During speciation events, rare
status is not inherited by the derived species; however, traits correlated with
biogeographic status may be heritable in some circumstances (Jablonski 1987).

In light of the large number of rare plant species and the increasing number threatened
with extinction, understanding the dynamics of their origins and their possible destinies
has become more urgent. Indeed, the International Union for the Conservation of
Nature’s (IUCN) recent publication of the 1997 Red List of Threatened Plants indicates
that more than 33,000 plant species, representing 12.5% of all plant species, are
threatened or extinct (Walter and Gillett 1998).

Threatened status is given to species that are more vulnerable to extinction, generally
because of declining numbers of individuals or populations through time. As a species
approaches extinction, its degree of rarity increases. However, not all rare species are
threatened; and although threatened and rare status are highly correlated, they are different
attributes. Yet, these two notions are often used interchangeably, especially vulnerability
or threat, as a proxy for rarity (Kunin and Gaston 1997). To understand the human role

in biodiversity loss and to provide better means for mitigating the effects of human

 

 

Biol
CVOil
Rafir
belie‘
to set:
new c
Specie
that ur
pan of
Plant 5;
The
Willis (_
CVolved
SignlfiC
(See Fig,
will;
in Which
also Will.
diminishe
the debate

Status {Or S

yOUng 0r 0

activity, we need a better understanding of the similarities and differences of natural and

human causes of rarity.

1.2 A BRIEF HISTORY OF RARITY CONCEPTS

Biologists have conjectured about the nature of rarity since before the advent of
evolutionary theory. The earliest scientific notions of rarity centered on species' age.
Rafinesque (1836) held the view that rare species are young while Lyell (1830-33)
believed that rare species are old. While then-current evolutionary theory provided little
to settle the debate, it actually led to more complex debate because of the introduction of
new concepts about speciation and extinction. Darwin (1872) settled on the view that rare
species are heading toward extinction and are therefore old, although he understood the
that underlying causes of rarity are complex and include many factors. During the first
part of the twentieth century, reiterations of Darwin's notions about rarity and the age of
plant species continued to prevail (Fiedler, 1986).

The view that rare plant species are old and going extinct was forcefully challenged by
Willis (1916) who, after studying the flora of Ceylon, argued that rare species are newly
evolved. Willis set off a vigorous debate about rarity, the age of taxa, and evolution.
Significant contributors to this debate included Ridley (1916), Femald (1918), and others
(see Fiedler 1986).

Willis (1922) expanded his theories about rarity in his treatise entitled Age and Area,
in which he argued that the age of a species corresponds with its area or range size (see
also Willis and Yule 1922). The controversy about Willis‘s concepts of "age and area"
diminished several years later following Gleason's (1924 and 1926) pluralistic solution to
the debate. Gleason argued that the area or range size of a species (and therefore rare
status for some species) is in part a function of its age, and that rare species may be either

young or old. Gleason's views on rarity and age have since prevailed.

 

 

By
new co
was thc
views 1
special
distribi
size an
the hyp
compet
of Specj

Othc‘

Cain (IS
(1956,) n
Durir
Classifica
age. Thei
adOpted b;
SPeCiation
acknowled‘
making Son
In the I!

Spuned an 1-

 
    
  
 

Net-em, e 165
had recehe
Among [iste
in 98 “move-

march. 84'

By the mid-twentieth century, the debate about rarity in plants broadened to include
new contributions from genetics, cytology, speciation, and ecology. Stebbins (1942)
was the most significant contributor during this period because of his synthesis of earlier
views and his hypotheses concerning rarity that utilized new concepts in genetics,
speciation theory and ecology. He also refined theoretical ideas about the geographical
distributions of rare plants and, in doing so, described three rarity types based on range
size and local abundance. Stebbins laid a groundwork of testable hypotheses, including
the hypothesis that rare species are genetically depauperate or that they are poor
competitors, which led to decades of research. Stebbins also acknowledged that the ages
of species were difficult to determine.

Others who contributed to discussion about rarity in plants during this period include
Cain (1940, 1944), Griggs (1940), and Wulff (1943). Simpson (1953) and Wright
(1956) made important contributions to general concepts of rarity.

During the early 1960's, Favarger and Contandriopoulos (1961) proposed a new
classification for causes of rarity based on cytological attributes, systematic position, and
age. Their scheme provided new testable hypotheses about the origins of rarity and was
adopted by Stebbins and Major (1965) in their important monograph on endemism and
speciation in the California Flora. As was the case with earlier contributors, they
acknowledged the limitations associated with determining the age of a species, thus
making some aspects of Favarger and Contandriopoulos' hypotheses difficult to test.

In the 1970's, the passage of the Endangered Species Act and the CITES Convention
spurred an increase in protection efforts and studies of rare plants in the United States.
Nevertheless, fewer than 370 threatened plant taxa, a small portion of the perceived total.
had received federal listing by 1992 (Schemske, et. al. 1994; Walter and Gillett 1998).
Among listed plants, Schemske, et. al. reviewed the research recommendations proposed
in 98 recovery plans and found that 96% of the recommendations called for ecological

research, 84% for demographic research, and 26% for population genetic research. For

 

 

the last 1
ecologic
hypothe
rare spe<
Exceptic
et. el. 19
except v.

Othe
Fiedler(

canbefo
1.3 S

Central to
measure a
Willis if 19
Categon'eS:
COmmon. ;
location of
Stebbm
OVErajj fang
forms of far;

panfrns 0f 1"

   
  
  

but 10cm), 3.
for two axis
Rablnou

together wm

the last several decades, studies of rarity have continued to focus almost entirely on
ecological and population genetics factors. Many of the findings have confirmed earlier
hypotheses about rarity while some refute them. For example, Stebbins's hypothesis that
rare species are genetically less diverse than common species has often been supported.
Exceptions have been found such as some members of the genus Astragalus (Karron
et. el. 1988). Studies of rare plants usually do not isolate simple causes for their status,
except when due to the direct effect of human activity.

Other reviews of plant rarity can be found in Kruckeberg and Rabinowitz (1985),
Fiedler (1986), and Fiedler and Ahouse (1992). An excellent general discussion of rarity
can be found in Gaston (1994).

1.3 SPATIAL ATTRIBUTES OF RARE SPECES

Central to the notion of rarity is a taxon's spatial distribution. Depending on the scale of
measure and the criteria used, taxa that fall below a certain threshold are labeled "rare."
Willis (1916), in his study of the flora of Ceylon, divided species into six range size
categories: very rare; rare; somewhat rare; somewhat common; common; and very
common. His unit of measure was the area derived by drawing a boundary around the
location of collection sites. In effect he used range size for categorizing rarity.

Stebbins (1942) noted that spatial patterns of rarity exist at multiple scales, including
overall range size, number of populations, and local abundance, leading to different
forms of rarity. In a very similar way, Drury (1974 and 1980) described three spatial
patterns of rarity: geographically widespread but locally sparse; geographically restricted
but locally abundant; and geographically restricted and locally sparse, which are derived
for two axis of measure.

Rabinowitz (1981) brought Stebbins's and Drury's spatial categories of rarity

together with degree of habitat specificity in her "Seven forms of rarity," now the most

 

fiequend)
included i
Perhaps tl
biologists

Local
many of t
number oi
for measu
species dis
Maurer 19

Species be
1.4 R,

SICbbins. i
al-‘Iproacw
the genetic
COncemed.
and/0r 8CD:
imponifllCe

RISCUr
SPeCieg has
eXlSIjng at 3
Some be 'OIc
“Heme”, in
reproducn-Or

frequently cited classification system for rarity. Interestingly, the age of species was not
included in her classification system despite years of discussion about its importance.
Perhaps the frequency with which Rabinowitz's taxonomy of rarity is cited indicates the
biologist's preference for attributes that can be studied directly, unlike age.

Local abundance and range size are highly simplified categories that do not capture
many of the dimensions of species' distribution, such as number of populations or the
number of collecting localities, and are at two ends of a spectrum of spatial scales useful
for measuring geographic distributions. Scales of measure employed during studies of
species distribution great impact on the interpretation of data (Allen and Hoekstra 1992;
Maurer 1999) and thus need to be carefully considered. This is particularly true with rare

species because sampling errors are higher compared with more widespread species.

1.4 RARITY, AGE, AND EVOLUTIONARY HISTORY

Stebbins, in 1980, again proposed a major synthesis on causes of rarity. His "synthetic
approach" proposed that the study of rarity should take into account ecological factors,
the genetic structure of populations, and the evolutionary history of the lineage(s)
concerned. To date, nearly all empirical work on rarity has considered only ecological
and/or genetic factors (for an exception, see Linder 1995) in spite of the recognized
importance of species age toward understanding rare status.

It is curious that few have acknowledged or realized that the whole idea of the age of a
species has flaws. Indeed, in the words of G. G. Simpson (1953), . . all lineages
existing at any one time are of precisely the same age, so how can some be 'young' and
some be 'old'? Unless life has arisen in more than one period of Earth history, which is
extremely improbable, all must necessarily have undergone the same span of continuous

reproduction. "

 

 

Thi
species.
differen
number
fossil re
regardin
from uh
Phyl
origins a
history ir

of rarity.

l.5 A

The centra
diversificai
Within mgr
ClchmSIam

the We 5pc.

ramy may [1

 

This leads to important questions about previous authors' ideas about the age of a
species. I conjecture that references to "age" have really been references to several
different features of lineages, such as time since divergence from a common ancestor, the
number of derived versus ancestral traits, and taxonomic distinctness. In the absence of a
fossil record, phylogenetic estimation of evolutionary trees provides the best evidence
regarding these age-like features. Phylogenetic studies also result in testable hypotheses
from which future work can be gauged.

Phylogenetic studies provide new perspectives toward better understanding the
origins and destinies of rare plants. In heeding Stebbins call to include evolutionary
history in studies of rarity, I undertook a study aimed at assessing phylogenetic patterns

of rarity.

1.5 A CASE STUDY OF RARITY FROM A PHYLOGENETIC PERSPECTIVE

The central hypothesis addressed by the case study described here is that high rates of
diversification in concert with local speciation result in high proportions of rare species
within monophyletic groups of species. The causes of rarity in a group these
circumstances would be linked to the causes of diversification and thus would mean that
the rare species would have common causes for their rare status. Phylogenetic patterns of
rarity may thus result.

The genus Astragalus was chosen as a study group because it has a number of useful
attributes. More than 400 species have been described in North America, many of which
are rare. The taxonomy of the group has been extensively revised, most recently by
Bameby (1964), who spent nearly 25 years collecting and studying thousands of living
and pressed herbarium specimens. His species concepts are well accepted and have been

repeatedly upheld by more recent taxonomic and systematic work. His monograph

 

 

contair
study.
was sei
flouen
case 511
Thi
this dis,
cladistic
Astraga
sections
supponc-
the merr.
resolving
In ch
addresse<
methods.
3) Which
P 16011:? 4

Result
dunnglhe
Patterns 0f
(hiring the (

include. 1 ’

contains data in sufficient detail for both phylogenetic and biogeographic aspects of
study. The immense size of the genus precluded a comprehensive study, thus a subgroup
was selected. Recent phylogenetic work has unveiled the monophyly of the Large-
flowered Piptolobi (Lf-P) group with approximately 51 species and was used for this
case study.

This case study is presented in three parts corresponding to the next three chapters of
this dissertation. The first part, presented in chapter 2, focuses on the morphological
cladistic study that aimed to elucidate phylogenetic relations among the Lf-P of
Astragalus. Questions addressed during this study include: 1) Is the monophyly of
sections in Barneby's Large-flowered Piptolobi sub-phalanx (with a few modifications)
supported using morphological data? 2) What are the species level relationships among
the members of the group? and 3) What is the utility of morphological characters in
resolving interspecific relationships in Astragalus.?

In chapter 3, the biogeographic study of the UP group is presented. Questions
addressed include: 1) How do range size measures compare when based on different
methods of determination and different scales? 2 ) Are species distribution fractal?

3) Which species meet the rarity criteria used by the IUCN's 1997 Red List of Threatened
Plants? 4) How do values of species richness compare based on differing measurement
scales? and 5) Where are the most species rich areas?

Results from the phylogenetic and biogeographic studies were combined and analyzed
during the study presented in chapter 4. This third study aimed at assessing phylogenetic
patterns of rarity in the Argophyllean clade of Astragalus, a monophyletic group identified
during the cladistic study. Questions addressed in this third section of the case study
include: 1) Is there evidence of diversification rate variation among lineages in the group?
2) Do rare species cluster in the phylogeny? 3) Are newly derived rare species more
frequent than expected by chance? Finally, conclusions and a few speculations derived

from all the results and the literature are offered.

 

AM

Larg

The gfinus
”Ch ECnus
The Nonh
SlUdied b}.
STUdieg dun
as Well as ll
The firs
designed to
about relam
the monoph:
ge‘flowe
1W0 limitaiio

Size of the dr;

2
A Morphological Cladistic Study of the

Large-Flowered Piptolobi of Astragalus L. (Fabaceae)

2.1 INTRODUCTION

The genus Astragalus includes more than 2500 species and is known as the most species
rich genus within the Angiosperms (Sanderson and Wojciechowski 1996; Liston 1994).
The North American species number approximately 400 and have been extensively
studied by four monographers, most recently by Bameby (1964). Modern phylogenetic
studies during the last decade have shed significant light on relationships within the genus
as well as the position of Astragalus within the tribe Galegeae of the Fabaceae.

The first cladistic study of the genus, based mainly on morphological data, was
designed to elucidate relations among sections and resulted in preliminary conclusions
about relationships within the genus (Sanderson 1991). Results provided evidence for
the monophyly of a number of sections, including Argophylli, the largest among the
Large-flowered group within Barneby's Piptoloboid phalanx. Sanderson's study had
two limitations that should be noted. First, due to computational limitations and the large

size of the data set, 113 taxa, the resulting search did not uncover the most parsimonious

trees. Second, outgroup selection was based on taxonomic concepts rather than explicit
phylogenetic hypotheses.

Higher level cladistic relationships within Astragalus as well as relationships among
related genera were later studied using chloroplast DNA restriction site and nuclear
ribosomal DNA sequence data. Results from these studies have provided strong support
for several higher level clades within North American members. Monophyly of the genus
was strongly supported as well as the monophyly of the aneuploids within the
paraphyletic euploids (Sanderson and Doyle 1993; Wojciechowski et al 1993).
Additionally, the monophyly of Barneby's sub-phalanx, the Large-flowered Piptolobi,
which includes section Argophylli with a few modifications, was also well supported
(Sanderson and Doyle 1993).

Phylogenetic relationships below the sectional level have remained elusive due to the
lack of phylogenetically informative genetic markers. Indeed, in a review of phylogeny
of the tribe Galegeae and Astragalus, Sanderson and Liston (1995) emphasize that
. . in Astragalus, morphology appears to be the most useful at the level of section and
below." They go on to suggest that the quickest progress toward resolving relationships
at lower taxonomic levels will include both molecular and morphological data.

The objectives of the study presented here were twofold: l) to test the monophyly of
sections in Barneby's Large-flowered Piptolobi sub-phalanx (with a few modifications)
and to elucidate relationships among species; and 2) to test the utility of morphological

characters in resolving interspecific relationships in Astragalus.
2.2 METHODS

The present study consists of 51 ingroup taxa and are largely restricted to Barneby's
(1964) Large-flowered group of the Piptolobi phalanx (L-fP). Several modifications

have been made stemming from previous molecular studies. Three sections, Desperati,

10

Sam
were
to be
Sandi
specii
sectic
the L-
additi
earth
.4. ca:
A. car
Withir
Tl
Piptol
Pacifi.
based
two g,
D2
NY an
frOm tl
CharaC
Sigmf“
beCaus.
5pc
equiva]
111833, d

Sarcocarpi, and Tennesseenses (representing four, four, and one species, respectively),
were excluded because species from these sections have been found by molecular studies
to be outside of the otherwise monophyletic L-fP clade (W ojciechowski et. al. 1993;
Sanderson and Doyle 1993; Sanderson per. mm.) The section Diphysi, with three
species, was included because the widespread and diverse species A. lentiginosus of the
section has been twice shown to be nested within the L-fP species. Three species within
the L-fP have been revised since Barneby's (1964) treatment, resulting in two species
additions and one species deletion from the group as follows: A. tephrodes var.
eurylobus has been elevated to the rank of species as A. eurylobus (Bameby 1984);

A. castaneiformus var. consobrinus has been elevated to the rank of species as

A. consobrinus (Welsh 1978); and A. musimonum has been reduced to variety status
within A. amphioxys (Bameby 1989). Species in this study are summarized in table 2.1.

Three outgroup taxa were chosen for this study and are all from within Barneby's
Piptolobi phalanx. Astragalus pomonensis and A. trichopodus are members of the
Pacific Piptolobi and A. douglasii is a member of the Small—flowered Piptolobi. Results
based on chNA restriction site data (Sanderson and Doyle 1993) suggest that the above
two groups are basal to the Large-flowered Piptolobi.

Data were derived from examination of more than 500 herbarium sheets on loan from
NY and RSA as well as published data in Bameby (1964). In a few cases for which data
from these two sources were unavailable, Bameby (1989) was consulted. Among the 90
characters screened for use in this study, 37 were excluded due to insufficient variation,
significant overlapping variation and/or a high numbers of polymorphic species, or
because of difficulty in scoring the character from herbarium specimens.

Species were scored for 53 binary and multistate characters representing 137 binary
equivalents (see table 2.2). Thirteen species in the study have infraspecific taxa and for
these, data were combined for all varieties within a species resulting in higher average

levels of polymorphic characters compared to species without varieties. Among all

11

Table 2f

leuersli
WW

—-—-——
f

E“)

.MW

.1 ‘ ‘ 1 ‘ l I. ‘ ‘ 1‘ u‘ ‘

AAAAAAAAAAAAeAAeAeAAAAAmAn

N

mrnrleUri...”

p00

AAAAAAAWA AA

‘1

C

AAAAA.mA
A.

Table 2.1. Species of Astragalus included in study. Arrangement by section (all capital
letters), followed by subsections when applicable. Taxonomy after Bameby (1964,
1989).

 

ARGOPHYLLI LAYNEANI
Argophylli A. layneae Green
A. argophyllus Nutt. ex T. & G.
A. zionis M. E. Jones MOLLISSIMI
A. piutensis Bameby & Mabb. Mollissimi
A. desereticus Bameby A. mollissimus Torr.
A. callithrix Bameby Orthanthi
A. tephrodes A. Gray A. helleri Fenzl
A. eurylobus (Bameby) Bameby
A. iodopetalus (Rydb.) Bameby GIGANTEI
A. shortianus Nutt. ex Torr & Gray A. giganteus Wats.
A. cyaneus A. Gray
A. columbianus Bameby MEGACARPI
A. tidestromii (Rydb.) Clokey A. megacarpus (Nutt.) A. Gray
Pseudoargophylli A. oophorus S. Wats.
A. waterfallii Bameby A. beckwithii Torr. & Gray
A. feensis M. E. Jones
Neomexicanus LUTOSI
A. neomexicanus Woot. & Standl. A. lutosus M. E. Jones
Newberryani
A. uncialis Bameby PTEROCARPI
A. musiniensis M. E. Jones A. casei A. Gray ex Brewer & Wats
A. loanus Bameby A. pterocarpus S. Wats
A. newberryi A. Gray A. tetrapterus A. Gray
A. eurekensis M. E. Jones
Coccinei DIPHYSI
A. coccineus Brand. A. lentiginosus Douglas ex Hook
Eriocarpi A. iodanthus S. Wats.
A. purshii Douglas ex Hook. A. pseudiodanthus Bameby

A. leucolobus Wats. ex M. E. Jones

A. subvestitus (Jeps.) Bameby
A. funereus M. E. Jones

A. utahensis (Torr) Torr. & Gray OUTGROUP
A. nudisiliquus A. Nels.
A. inflexus Dougli. ex Hook. DENSIFOLII
Parryani A. pomonensis M. E. Jones
A. parryi A. Gray
Missourienses TRICHOPODI
A. castaneifonnis S. Wats. A. trichopodus (Nutt.) A. Gray
A. consobrinus (Bameby) Welsh
A. chamaeleuce A. Gray INFLATI
A. amphioxys A. Gray A. douglasii (Torr. & Gray) A. Gray
A. cymboides M. E. Jones
A. missouriensis Nutt.
A. accumbens Sheld.
Anisus
A. anisus M. E. Jones

 

12

Table 2.2. Characters and their states used in the cladistic analysis.

 

OO\IO\ MA LON—e

Caudex position: 0 = at or above soil surface; 1 = subterranean

Duration of plant: 0 = perennial; l = short-lived perennial, biennial, or annual
Stem growth pattern (ordered): 0 = strongly caulescent; l = moderately caulescent;
2 = acaulescent

Petiole: 0 = sub-sessile, very short; 1 = well developed

Leaflets per leaf (ordered): 0 = greater than 26; 1 = between 5 and 26;

2 = between 3 and 5; 3 = less than 3

Leaflet size: 0 = less than 3.9 mm; 1 = greater than 3.9 mm

Leaflet folding: 0 = present; 1 = absent

Leaflet veination: 0 = midrib visible only; 1 = secondary veins visible;

2 = conspicuously reticulate

Leaflet shape: 0 = orbicular-elliptic; 1 = linear

Terminal leaflet attachment: 0 = jointed; 1 = confluent

Leaflet apex: 0 = rounded; 1 = notched; 2 = acute

Stipule connation (ordered): 0 = all free; 1 = connate or amplexical at plant base
but free apically; 2 = all connate if only shortly so

Stipule texture (ordered): 0 = herbaceous; l = papery or membranous;

2 = scarious

Leaf pubescence type: 0 = basifixed and appressed; 1 = basifixed and spreading;
2 = medifixed

Leaf surface pubescence density: 0 = scanty; 1 = dense

Leaf abaxial pubescence density: 0 = scanty; l = dense

Calyx tube to diameter ratio (ordered): 0 = below 1.3; 1 = between 1.3 and 1.5;
2 = above 1.5

Petal color: 0 = white or whitish; l = ochroleucous to yellow; 2 = pink to purple;
3 = scarlet

Fruit orientation: 0 = erect-ascending; 1 = declined

Valve texture (ordered): 0 = membranous to papery, not stiff;

1 = stiffly papery to leathery; 2 = stiffly leathery to woody

Valve surface: 0 = smooth or faintly reticulate; l = coarsely reticulate

Pod drying dark or black: 0 = stramineous to brown;

1 = dark brown to purple, blackish

Pod color pattern: 0 = solid; 1 = mottled or speckled

Presence of spungy-pithy mesocarp: 0 = absent; 1 = present

Fruit septum (ordered): 0 = unilocular; l = bilocular but unilocular in the beak;
2 = fully bilocular

Fruit persistance: O = decidious; 1 = persistant

Number of ovules (ordered): 0 = below 14; l = between 14 and 18; 2 = above 18
Fruit dehiscence: 0 = apical ventral; 1 = apical ventral and through stipe;

2 = ventral and dorsal

Stipe: 0 = absent; 1 = present

Gynophore (ordered): 0 = absent; 1 = incipient and minute; 2 = present

Fruit curvature: 0 = straight, or nearly so; 1 = incurved; 2 = sigmoidal

Fruit beak: 0 = absent to cuspidate; 1 = strongly differentiated and upturned;

2 = weakly differentiated and decurved

Fruit inflation: 0 = present; 1 = absent

Fruit dorsal surface: 0 = smooth; 1 = dorsally grooved; 2 = carinate

 

Continued on next page

 

l3

Table 2.2 Continued

 

Fruit cross section: 0 = rounded, ob-, or dorsi-ventral compression;

1 = trigonous, triquetrous, or trianglular; 2 = laterally compressed

Black caylx hair: 0 = present; 1 = absent or nearly so

White calyx hair: 0 = present; 1 = absent or nearly so

Pod vestiture: 0 = pubescent; 1 = glabrous or nearly so

Flowers per inflorescence (ordered): 0 = below 3; 1 = between 3 and 22;

2 = above 22

Flower orientation: 0 = ascending; l = declined

Keel length (ordered): 0 = less than 22 mm; 1 = between 22 and 29 mm;

2 = above 29 mm

Keel incurvature (ordered): 0 = below 40 degrees; 1 = between 40 and 75 degrees;
2 = between 75 and 110 degrees; 3 = above 110 degrees

Banner recurvature (ordered): 0 = less than 15 degrees;

1 = between 15 and 32 degrees;

2 = between 32 and 60 degrees; 3 = greater than 60 degrees

Anther size (ordered): 0 = between 0.4 and 1.0 mm; 1 = less than 0.4 mm;

2 = greater than 1.0 mm

Fruit length to diameter ratio (ordered): 0 = less than 3.3; 1 = between 3.3 and 4.2;
2 = 4.2 and 6.5; 3 = above 6.5

Fruit ventral surface: 0 = smooth; 1 = dorsally grooved; 2 = carinate

Seed coat purple-speckled: O = absent; 1 = present

Seed coat texture: 0 = smooth; 1 = pitted and/or wrinkled

Seed coat lusture: 0 = dull; l = lustrus

Bracteoles, well developed: 0 = absent; 1 = present

Seed coat dark or black: 0 = absent; 1 = present

Wing to banner gradation (ordered): 0 = wing more than 5% longer;

1 = wing between 0% and 5% longer; 2 = wing up 12.5% shorter;

3 = wing more than 12.5% shorter

Keel to wing gradation (ordered): 0 = keel longer; 1 = keel up to 22% shorter;
2 = keel more than 22% shorter

 

14

 

 

 

species. a !
polymorp?
remains eq
varieties at
Eight q
controvers:
lhiele 199.”
than direct]
sampling 01
presents cor
order to mi
of measures
POl)’m0rphj¢
that overlap;
Quantita
(1993). Six
ordered bee a
qualitative as
2862 cells in
polFmOrphjc
Providing ph

A Starch

 

1998)0n alt

  
  

100 TandOm .'

30,000.

species, a significant correlation exists between the number of varieties and the number of
polymorphic characters for a species (Kendall's tau = 0.56, p < .001). This correlation
remains equally significant after the removal of the outlier A. lentiginosus with 36
varieties and 12 polymorphic characters.

Eight quantitative and three meristic characters were included in this study in spite of
controversy surrounding their usage in cladistic analyses (Baum 1988; Stevens 1991;
Thiele 1993). Data for these characters were taken largely from Bameby (1964) rather
than directly from herbarium sheets because Barneby's data represents a far greater
sampling of individuals than could possibly be achieved in this current study. Bameby
presents continuous measures as ranges rather than as means and standard deviations. In
order to make use of the gap-coding procedure of Archie (1985), the standard deviation
of measures for a species was conservatively estimated by dividing the range by two.
Polymorphic coding was used in the few cases when a species exhibited broad variation
that overlapped otherwise discrete patterns of variation among other species.

Quantitative and meristic characters were ordered following the suggestion of Thiele
(1993). Six qualitative characters, scored based on data from Bameby (1964), were also
ordered because variation among species for these characters was derived from the
qualitative assessment of underlying continuous variation along a single axis. Among
2862 cells in the data matrix, 20 (<O.7%) had missing data. Among the 135 cells with
polymorphic coding, 71 had fewer character states than were possible, potentially
providing phylogenetic information (see table 2.3).

A search for most parsimonious trees was undertaken using PAUP* 4 (Swofford
1998) on a Macintosh G-3 computer. A heuristic search strategy was employed using
100 random addition sequence replicates with TBR swapping with the following Options
in effect: steepest decent, multrees, zero length branches collapsed, and maxtrees set to

20,000.

15

 

TableZ 3
POl\ morl‘r'h
D: 2&3. i

T3 on

C
a:::;:y-...
-

zianrs
pic: tsis
desere:1:us
ca‘.1PhV'-.x
teehrcdes
e"*""b
ieispe -a--s
s::r:;anas
cyaneus

- mh~ “ c
C:-m.~ ~- .“

:jes: 655;;
t .‘ .'
.ata' .a:;..

‘93“ u

Lyme-
nmman;
I..- ".a;;s
TJSL.”.2‘3..".5-_‘
-aa:_s
IZEJZEV: n

ere-teas ; s

F‘H‘A‘

VJ» ‘ne“s

:reahaeleUCE
itsaln szfs

Citml jes

n§$$2uv

ie: .

5221:5815
3...;3-45
la‘vfleag
331118

- .
“=«.e:~

Jails

3‘si'lteus
mitigus
Ccp’lcb

‘w. .Q“b.‘
N‘*‘
‘

“Wu 4‘
p O"wit:
r3935~ 3
:‘rlak ‘
‘ lfi ‘
I “133*

 

Table 2.3. Taxon by character data matrix. Characters as in Table 2.2.
Polymorphic taxa are coded as follows: A = 0&1, B = 0&2, C = 1&2,
D = 2&3, E = 0&1&2. Unknown = ?. The last three taxa are outgroups.

argophyllus
zionis
piutensis
desereticus
callithrix
tephrodes
eurylobus
iodopetalus
shortianus
cyaneus
columbianus
tidestromii
waterfallii
feenis
neomexicanus
uncialis
musiniensis
loanus
newberryi
eurekensis
coccineus
purshii
leucolobus
subvestitus
funereus
utahensis
nudisiliquus
inflexus
parryi
castaneiformis
consobrinus
chamaeleuce
amphioxys
cymboides
missouriensis
accumbens
anisus
layneae
mollisimus
helleri
giganteus
megacarpus
oophorus
beckwithii
lutosus
casei
pterocarpus
tetrapterus
lentiginosus
iodanthus
pseudiodanthus
pomonensis
trichopodus
douglasii

11111111112222222222333333333344444444445555
12345678901234567890123456789012345678901234567890123

00C11100002010112202A00000200011100000100221020A01021
20111100002120112202101000200011100010100221020A01021
00C11100002010112201100000200211100000100121010AA1121
0011111000201111200110000010021110010010022101???l?22
01111000000011112202000000200011110100100121020011021
00C1110000812A1A220C11000020001110000A100221020AA0121
00C1111000002l1A2202110010200011100100C00221020AA0121
00111100002021102202110000200011100101100221020010022
0021110000001011220211000020001110010010022102011102l
00211110000010112202100000200011120000100221020001021
000111000020A000200211000020001110001110022lA20111012
01C111100000111120021000002000111200101002211211110C1
001111010030101022010010C0210000101000100221221111021
011111000000211022011000C0200010111100100121220110021
000101000020111122Al100010200010100000110221020001131
00212110002012112201000000200011120100001121020A11021
00212010002012112201000100200011100000000121010110021
0021111000001211200100000020011100000010022101???0?2l
0021111000802211220110000020021110000010A121000A11121
00211100002012112102100000200011100000100C21020010021
00211100008011112301000000200011100000102012080A00120
00211100002021112E01A100A02002lllA000010022lOElAOOlZl
0011110000202111220210001020011111A100100221020111021
0011110000301111200110001000001110010010022100010002l
00111l000OO011112201100000000011100100101121000001020
00011100000011112201100000200211100100100221110011121
00011100003011112202000000200011100100100221000101121
00001100008011112202100000200211110100100221010A00121
0000110A000011A0200210000020001111000010012lllOOOlODl
0021110000B0121120010000002000A0100000100221020A11121
002111000080121120010000002000A0100000100221020A11122
00C11110000012112C0111110020000010000A100221020111021
01111100008012112202llA0002200A1120000100221120111021
00211110008012112A01010100200000122000100221020110021
00C11llOOOBAC211220211000121000A12E000100221020111021
0011110000001211210111000121000012100010033002???0?01
00111100000012112201100120220000100100100121010001121
100111000000111120011010C020011111A010200221321111021
0lClAlOOOOOOllllZEOClOOOlOZOOOAAAlOOOOCAOZ11011101121
00110100000011112301100020200??01101001010010????0?Dl
0001A010002001112102000011200010100101210221010000121
00111111000010002800001000200200000000100221000101121
000llllOOOEOlOOOZClOOOlOOOZ00200000101110231000101021
00001111003010000C0200101020021AlA0101100231020001122
10011100000211102800000000210200000000A00221010000021
10001100A1E10011221111100021011012001A100221C20A01031
100030001121001122111100002101A1120001100221020A1103l
1000110010210000231111100021011112000A100221120111031
0A0111lAOOEO10002E0100lOlOZOOOAAOAOOOACOOZZlOAlAAOlZl
000011010000000028111010102100111AA00010012lZOOAAllZl
1000110000001110121All1010100011110000100221100100021
00000111000OOOOOOAOOOOOOOOZ00000000000210221010100021
0000A10A000010001A10000001212000A0200021022lOAOAOllDl
000111000OOOOOAAOAOOOO100020010A00000AC002310AOA0?021

16

2.3 RESI

The searcl
autapomor
and a reter
that these ‘
negativel}
other moq
strict cons
nodes witl
parsimonir

Contra
[3118. four I
lentiginosi
[3X3 Chose
Pacific Pi;
leading to
PomonenSi
OlllgrOupS

Seclior
among the
highly deri'
mOIIOph}.1 E
is paraphylr
me “”0 8pc.
"101101).ij 5

Clade.

2.3 RESULTS

The search uncovered 12 most parsimonious trees of 489 steps, including
autapomorphies with a consistency index (excluding uninforrnative characters) of 0.47
and a retention index of 0.51. Although these values seem low, it should be pointed out
that these two measures are highly correlated with each other and are significantly
negatively correlated with the number of study taxa. The consistency index is in line with
other morphological cladistic studies of plants (Sanderson and Donoghue 1996). The
strict consensus tree (figure 2.1) is well resolved and contains only three unresolved
nodes with three branches each. A randomly chosen tree among the 12 most
parsimonious trees is shown in figure 2.2.

Contrary to assumptions made about relationships among the ingroup and outgroup
taxa, four a priori chosen ingroup taxa, the three species of section Megacarpi, and A.
lentiginosus of section Diphysi, are nested among the outgroup taxa. Among the three
taxa chosen as outgroups, A. pomonensis and A. trichopodus, both members of the
Pacific Piptolobi, form a clade, whereas A. douglasii is several nodes from the branch
leading to the other two outgroups. For purposes of rooting, the branch leading to A.
pomonensis and A. trichopodus was used because it leads to two of the three chosen
outgroups (see figure 2.1).

Section Diphysi is polyphyletic with the widespread species A. lentiginosus placed
among the outgroup taxa and the remaining taxa, A. iodanthus and A. pseudiodanthus, as
highly derived taxa within the tree. Sections Mollissimi, Pterocarpi, and Megacarpi are
monophyletic (although Megacarpi is nested among outgroup taxa). Section Argophylli
is paraphyletic with four sections, Layneani, Gigantei, Pterocarpi, and Mollissimi, and
the two species of Diphysi nested within. This monphyletic assemblage, plus the
monotypic section Lutosi, are members of a lineage designated here as the Argophyllean

Clade.

17

 

 

 

 

Figure
fiction
enme‘

argophyllua

zlonls

cyaneus

calllthrlx

columblanus

parryl

waterfall"

teenla

lavneao - LAYNEANI
easel

pterocarpus PTEROCARPI
tetrapterus
lodanthus
pseudlodanthas
tldestromll
amphloxys
leucolobus
tephrodes
eurylobua
lodopatalus
ahortlanua
subvestltus
tunereua
nudlslllquus
neomexlcanua

alaanteus - GIGANTEI
molllslmua
I MOLLISSIMI

ll.

Clade

 

 

I DIPHYSI

Wain

 

 

 

fi

lln

 

 

Argophylli-Eriocarpean

 

hellerl

anlsus
mahensls
lnflexua
deseretlcua
plutensls
newberryl
purshll
unclalls
muslnlensls

. chamaeleuce
cymboldea
mlssourlensla
accumbene

n eurekensis
castanelformla
coneobflnua

 

 

 

 

 

r1 d1l—L7'il

 

ARGOPHYLLEAN CLADE

IL

 

rssourran
Clade

M

 

 

 

 

 

 

 

 

 

 

cocclneue _
loanus —.
lutosus * LUTOSI
douglasll INFLATI
I- megacarpus
oophorus MEGACARPI
beckwlthll
lentlglnoaus DIPHYSI
l' pomonensls DENSIFOLII
'- trlchopodua TRICHOPODI

Figure 2.1 Strict consensus of 12 most parsimonious trees recovered.
Sections names are in all capital letters. Monotypic sections are
denoted by (*).

18

argophyllua
zlonie
cyaneus
— callithrix
columbianus

parryi
caaei

waterfallii
- L: ......
l layneae
pterocarpus

tetrapterua

 
 
  
  

 

 

 

 

 

 

 

 
 
 
  

 

 

   
 
 
 
 
   
  

  

iodanthus
paeudiodanthaa
. ‘ tidestromii
amphioxys
leucolobus
tephrodes
_ eurylobua
iodopetalus
shortianus
‘— nudisiliquus
C subvestltus
F tunereua
neomaxicanus
gigantaus
mollisimua
hellerl
anlsua
utahensls
lnflexua
desentlcus
plutensis
newberryl
purshll
unclalla
musiniensis
_ chamaeleuce
cymboldea
mimurienaia
accumbena
.— eurekensis
castaneitormls

 

coneobrinua

   
  
   
  

 

 

coccineus
— loanus
ll—— lutoaua
douglaaii

megacarpua

oophorus

beckwlthii
lentiginoaua 5 steps

 
  

pomonensis
trichopodus

Figure 2.2 A randomly chosen tree among the 12 most parsimonious trees.
Branch lengths are proportional to inferred evolutionary changes.

19

The Argophyllean species are found principally in two sub-clades that correspond well
with Barneby's subsections in section Argophylli. The Missourian Clade includes 10 of
the 13 species of subsections Missourienses, Newberryani, and monotypic section
Coccinei. The Argophylli-Eriocarpean Clade, with 35 species, includes five sections
(discussed above), the species of subsections Eriocarpi and Argophylli (both of which are
paraphyletic), the monophyletic subsection Pseudoargophylli, plus the three remaining
monotypic subsections, Parryani, Neomexicani, and Anisi. Astragalus amphioxys and
A. newberryi are dispersed within the later clade while A. loanus is basal to both.
Assessment of support for phylogenetic findings presented here met with
computational limitations. Both bootstrap and decay analyses were attempted but were
truncated due to tree buffer overflow. This is not surprising given the large number of

taxa, the low character to taxon ratio, and the high level of homoplasy in the data set.

2.4 DISCUSSION

Results correspond well with Barneby's (1964) taxonomy, which can be regarded as a
coarse phylogenetic hypothesis, at least for higher levels. Nearly all sections and
subsections were found to be monophyletic or nearly so if a few modification are made.
The most notable exceptions are section Argophylli, and subsections Argophylli and
Eriocarpi, which are paraphyletic with numerous small sections and subsections nested
within. These findings are not surprising in light of Barneby's tendency to put the more
divergent species in their own sections or subsections.

Perhaps the most significant finding of this study is the placement of section
Megacarpi (three species) and A. lentiginosus between A. douglasii and the two outgroup
taxa used for rooting, A. pomonensis and A. trichopodus. The placement of Megacarpi
and A. lentiginosus is contrary to the results of Sanderson and Doyle (1993), and in the

case of A. lentiginosus, is contrary to the results of Wojciechowski et. al. (1993).

20

 

 

 

mi
pr:
To

c1835

”Um

Astragalus oophorus, a member of section Megacarpi, and A. lentiginosus were placed in
highly derived positions nested well within the L-fP clade, according to results reported
by Sanderson and Doyle (1993).

Explanations for these contradictory results include several possibilities. The
placement of A. lentiginosus, a species in section Diphysi, near the base of the tree may
be due to the high number of polymorphic characters--12 for the species. Eight of the
twelve characters are polymorphic for all character states, rendering them equivalent to
missing data. The two other species of section Diphysi, A. iodanthus and A.
pseudiodanthus, were placed in a highly derived position within the Argophyllean clade.
To the extent that Diphysi is a good group and that the number of polymorphic characters
for A. lentiginosus resulted in basal placement of the species, these findings are
consistent with those of the above molecular studies. Interestingly, Bameby placed A.
lentiginosus in the Small-flowered Piptolobi sub-phalanx where the designated outgroup
species A. douglasii is found.

Another explanation for these findings results from the choice of outgroup taxa, one
species from the Small-flowered Piptolobi, and two species from the Pacific Piptolobi.
The relationship among these two sub-phalanxes with respect to the Large-flowered
Piptolobi may be more complex than either Barneby's taxonomy or previous molecular
studies would suggest. More importantly, results presented here suggest that the Large-
flowered Piptolobi (as modified for this study) is not monophyletic, in spite of findings
from molecular studies. In his discussion of the Piptolobi phalanx, Bameby suggests
that the various species can be advantageously grouped into series and presumably
phylogenetic affinity, but he acknowledges the difficulty of placement of the Pacific
Piptolobi because they have attributes of both the Large- and Small-flowered Piptolobi.

Perhaps more telling than the placement of the above lineages is that fact that
assessment of support values (bootstrap and decay indices) failed to conclude due to the

number of trees generated in the analysis. Although it is common for large numbers of

21

trees to be produced when character/taxa ratios are low, it is also possible that levels of
homoplasy are sufficiently high such that assessment of the robustness of the
phylogenetic reconstruction is inhibited. Indeed, given the evidence that Astragalus is
part of a larger clade (together with the genus Oxytropis and other members of the tribe
Galegeae) that has undergone an increase in diversification rates relative to sister groups,
it is entirely possible that diversification has proceeded more rapidly than the evolution of
characters assessed in this study. This explanation is consistent with the findings that no
molecular markers have yet been found adequate for resolving among species
relationships within the genus (Sanderson, per. com.)

Future studies aimed at assessing monophyly of the sub-phalanxes and their
relationships to one another will be needed before more definitive conclusions can be
drawn about this group. Nevertheless, findings here and elsewhere lend considerable
support to the conclusion that a large majority of Large-flowered Piptolobi species form a
monophyletic clade, notwithstanding the few modifications mentioned above. The
relationship of the L-fP clade to other Piptolobi awaits future studies.

Results from this study provide evidence for the utility of morphological characters
for elucidating relationships among species in Astragalus. A few qualifications are
offered, however. First, the effect of a low character to taxa ratio, which leads to short
branch lengths, needs to be addressed to overcome the problem of assessing support.
Addressing this issue by analyzing a smaller study group is likely to help but undoubtedly
will result in some characters becoming uninforrnative. Second, as was concluded by
Sanderson (1991), high levels of homoplasy among morphological characters in
Astragalus are likely to frustrate efforts to resolve relationships among larger groups.
Perhaps the greatest future utility of morphological features will derive from the discovery
and assessment of new features hitherto not used in phylogenetic studies such as this.

An additional consideration regarding the use of morphology stems from the common

finding that phylogenetic hypotheses based on morphology frequently do not correspond

22

to those based on molecular data. The assessment of new morphological characters
combined with more refined choices of ingroup and outgroup taxa are likely to lead to
more robust hypotheses of relationships. The same can be said for the discovery of
molecular markers with sufficient variation for inferring among species relationships.
Progress developing both these lines of evidence and a combined analysis may yield
stronger hypotheses.

In summary, phylogenetic results correspond well with Barneby's (1964) taxonomic
concepts after a few modifications. Some results are contrary to previous molecular
findings. Morphological data have proven very useful in resolving among species
relationships. However, results should be considered preliminary because of the
difficulty in assessing support as well as issues concerning the relationship between the
ingroup and the outgroup taxa. More definitive conclusions about relationships within

and among the Large-flowered Piptolobi await better taxon sampling and additional data.

23

 

 

 

A)...

 

If}:
is}
U/n.

170/14

. ”an";
filming

”hie

 

3
Rarity and the Biogeography of the
Large-Flowered Piptolobi of Astragalus L.

3.1 INTRODUCTION

For centuries, biologists have catalogued and studied the spatial distribution of
organisms. As new concepts regarding distribution patterns have emerged and the
processes leading to these patterns have been elucidated, new spatial measures have also
been developed. At least a dozen biogeographic measures have been used to characterize
range size during the last two decades (Gaston 1994).

The choice of geographic measures and measurement scales depends on both the aims
of the study and practical limitations. An unfortunate consequence of the wide variety of
measures and scales that have been used is that comparisons among studies are often
difficult. This condition is improving, however, due to recent trends in collecting and
storing biogeographic data.

Distributional point data are now emerging as one of the most useful forms for
quantifying spatial distributions for a number of important reasons. Point data can

represent a variety of attributes about an organism, depending on the goals of the

24

 

 

my
La}.
1113.

P0}

qu.

13C

bio

ben
nut

Ole

his a
Plate
hall

”POI

investigator. For example, points can be used to represent individuals or collections of
individuals for studies focused at local scales. Points can also be used to represent
populations or collection localities for larger scale studies. Furthermore, other parameters
can be linked to the points, such as dates, measurement scales, and error estimates.

Equally important is the convertibility of point data into other biogeographic
quantities, including abundance and range size measures such as area of occupancy and
area of extent. Recent advances in Geographical Information Systems (GIS) software
technology have significantly reduced the difficulty of converting point data into these
biogeographic parameters.

Another important trend in studies of geographic distributions is the increasing use of
grids for summarizing the range size of organisms and for calculating species richness.
Range size in the form of area of occupancy is generally based on the presence or absence
of organisms within grid cells and is simply the number of cells occupied multiplied by
the cell area. For convenience, many use grids that correspond to latitude and longitude
lines. This methods works well for small and local areas, but at larger scales, grids need
to be based on equal area cells to control for diminishing cell sizes as the poles are
approached.

The use of area of occupancy measures to summarize point data has two important
benefits: 1) the area can be summarized at any scale of interest as long as the cell size is
not smaller than the scale in which the data was collected or substantially larger that the
overall geographic extent; and 2) the use of multiple scales can be examined to determine
the fractal properties of an organism's distribution.

In spite of these benefits, summarizing point data in the form of area of occupancy
has a drawback that is commonly overlooked. The measure, when based on arbitrarily
placed grids, can result in high levels of sampling error among organisms found at few
localities and with small range sizes. This problem is particularly worrisome in studies

purporting to compare rare and widespread species. For example, if an organism is

25

found at four localities that are less distant from each other than the distance across a grid
cell, the calculated area of occupancy can vary by a factor of four, depending on where
the grids are placed. This problem can be eliminated by testing different placements of
grids to find the minimum area of occupancy measures.

In this chapter, I explore how grid size and placement affects the area of occupancy
measures based on distributional-dot data for 51 species of the Large-flowered
Piptoloboid (Lf-P) group of Astragalus (discussed in Chapter 2). Results are used to
quantify the fractal geometry of the species distributions as well as for classifying their
range-size rarity. Seven specific questions are addresses for the study group: 1) How do
area of occupancy measures based on the arbitrary placement of grids compare with grids
placed to rninirrrize the number of grid cells occupied? 2) How does this latter measure
compare with area of extent measures? 3) Which of the species meet the rarity criteria
used by the International Union for the Conservation of N ature's (IUCN) 1997 Red List
of Threatened Plants? 4) How do area of occupancy measures compare when different
measurement scales are used? 5) Are the species' distributions fractal and how do they
compare? 6) How do values of species richness compare based on different measurement

scales? and 7) Where are the most species rich areas located?

3.2 STUDY SYSTEM AND METHODS

3.2.1 The Large-flowered Piptolobi of Astragalus

Barneby's (1964) revision of the North American members of Astragalus provides the
data for this study. Bameby presents distributional point maps of nearly every species
covered in the revision based on more than 25 years of study that included thousands of

specimens from the field and from herbarium collections.

26

dis:

11hr

{Oil

Art

ha:

 

3.2.2 Data handling in the ArcView GIS environment

Geographic data were digitized and a large proportion of data manipulations were done
within the ArcView 3.x GIS software environment (Environmental Systems Research
Institute 1998), as shown in figure 3.1. Approximately 3000 points representing
localities derived from Barneby's distributional point maps were utilized. Point data,
hereafter called sites and denoted by S, were entered as latitude-longitude positions.
ArcView maps were manipulated using an equal-area cylindric map projection as shown
in figure 3.2 . The "equal-area cylindric" projection was chosen because meridians are
perpendicular to latitude circles throughout the map area.

A series of nine fixed position grids were created using an Avenue script (the
programming language used within the ArcView environment) created specifically for this
procedure by the programming staff at Environmental Systems Research Institute (ESRI).
This script, "GridPoly," appears in Appendix B. Beginning with 10 km x 10 km square
cells (100 kmz), a sequence of grids was created, with each cell larger in area than the
preceding one by a factor of four. A total of nine grids were created with the largest cell
size equaling 6,553,600 kmz. Linear dimensions of these cells are referred to as the scale
of measure and are symbolized by J. Scales began at 10 km and continued to 2,560 km.

The location of the initial 10 km grid was chosen randomly. From this initial
position, additional grids were created such that cells at smaller scales nested perfectly
within the cells at higher scales. Table 3.1 provides details about the scales, areas of
measure, and number of cells created for this study. Figure 3.3 shows the overall
distribution of the group together with the 320 km grid (cell area = 102,400 kmz),
whereas figure 3.4 shows the 40 and 80 km scale grids overlaying Utah. Point data,
corresponding to sites S, were linked to the individual cells in the grids using the
ArcView "spatial join" feature. This feature creates database linkages among attributes

that overlap grid cells. The number of cells occupied for each of 51 species at 9 spatial

27

 

 

   
    

 

           

 

 

 

 

 

 

 

 

 

 

 

       
    

 

 

, one > echo
_ @435 s. a m
i . III-IIIIIIII-IIIIIIIIIIIIIIIIII-IIIIIIIIIIIIIIIIIIIIIIII "
, «enema "
mmmcnoE ...
meadow main... use mans. m
2650? u
- 232.8%. "
3:300 :00 m ”Emu Re 28320 m
258...". m u
m «39.5300 >x EoE—S..$:H m
, . 0.83 e 32 o.— .
«my «coaxm u a m 96 .> < u
umF—Quhmomo IIIIIIIIIIIIIIIIIIII IIII-IIII-IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII-

 

 

 

 

1:an ~39 53.5.53: Saemé "Suggest =§~§=§<

 

‘ «cc Ecuaaooo
yo «no.3

 

G» £58 :8
£38.52

 

 

- «Q 20.3655
.508“.

 

:euafiueumuab 52. he 3:258 can macaw—on “c.5326 9:39 32m
5. 2&3

28

 

 

 

, ..... a.

.1. .

 

=e=uo_.e.a n5: zeta—Emu sonar—«=3: :a :c miawabm<
Me 32395 negate—noun“: 2: he .852:me
«a use...

 

29

 

 

 

Table 3.1 Scales, measurement areas, and number of grid cells used in this study.

 

Scale (J) in km Measurement area (f) in km’ Number of grid cells

 

10 100 102,400

20 400 25,600
40 1,600 6,400

80 6,400 1,600
160 25,600 400
320 102,400 100
640 409,600 25
1,280 1,638,400 6+
2,560 6,553,600 1+

 

30

 

 

auto—:44. 5.82 588.5 Maia—.82. Saw .5. can QM PEER

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

31

 

 

 

 

 

 

 

32

scales was tabulated. Species richness values for each cell were also tabulated. Equal—
area normalized coordinates for each site were exported for use in determining minimum

cell counts and geographic extents.

3.2.3 Minimum cell counts, area of occupancy, geographic extent,

and fractal dimensions

Minimum cell counts C were determined by the following procedure. Grids were
numerically moved in a 2 km step search procedure in both the X and the Y directions
until all possible unique positions were assessed. The smallest cell counts were retained.
At the 10 km scale, 25 different positions were assessed. Each larger scale was also
searched using 2 km steps resulting in four times the number of positions assessed
compared to the previous scale. This procedure was carried out with "Minimum Cell
Count," written in Q-Basic specifically to accomplish this search (see Appendix C).
Minimum cell counts are referred to simply as cell counts and are denoted by Cj for scale
of measure J.

A species area of occupancy at each scale J, denoted by A j, was calculated by
multiplying the cell counts by J2 (referred to as the measurement area). Each species had
up to nine measures for area of occupancy depending on the geographic extent of the
species.

The geographic extent of each species, denoted by E, was calculated by multiplying
the coordinate range in both the X and the Y directions. Range values were rounded up
to the nearest 10 km because the smallest cells were created with a measurement scale of

10 km. Measurement scale bias is thus kept constant among area measures.

33

A summary of measures is as follows:

0 S = Sites: number of localities recorded by Bameby (1964).
0 = Scale: linear dimension of grid cell
[10, 20, 40, 80, 160, 320, 640, 1280, & 2560 km]
Little j used as a scale identifier.
0 J2 = Measurement area at scale J .
0 C j = Cell count or cells occupied at scale J

0 A j = Area of occupancy at scale J. A = C12.

0 E = Geographic extent.

A comparison of fixed grid versus minimized area of occupancy measures was carried
out using cell counts directly for scales J = 10 and 160 kms. Correlations, both
Spearman and Pearson, were run to assess relations among sites S, cell counts C10, and
geographic extent E. In order to assess at which scales the largest changes occurred in
the rank order of species by area of occupancy A, each scale was compared with the next
larger scale using a sequential ranking procedure. Ordinary ranks, which give ties the
same average rank, introduce a large amount of variation at higher scales not pertinent to
these comparisons. Sequential ranks provide a unique rank for each species and, in the
case of ties, do not change the ranks used at the previous scale. The sequential rank
procedure was accomplished by first sorting species by A 10 and then assigning ranks.
Next, species were sorted first on A20 and second on A10, followed by ranking. This
series of steps continued until ranks were assigned at all scales. A count of rank changes

and the magnitude of changes were calculated for each pair of scales compared.

34

Area of occupancy A was used to calculate the fractal dimensions D of each species

using the following equation (Peigten et. al. 1992):

_ alogA

_ (3.1)
810g J2

D

 

2 . . . . .
where J is the measurement area. This equation can be reorganized into:

 

where 112 and 122 are the scales of measurement.
Another frequently used fractal dimension is the box—counting dimension DB, used
specifically for count data The dimension D is numerically related to DB in the following

way (Peigten et. al. 1992) and is presented here for comparisons with other studies:

D3=2—2D (3.3)

Fractal dimensions were calculated for each pair of sequential scales using equation
3.2 and resulted in eight comparisons for each species. Log 12 was plotted versus log
A10 to express the fractal dimension as the slopes of the lines connecting the individual
points for a species.

The degree to which a species distribution is fractal was assessed by calculating the

standard deviation among a sequential triplet of dimensions (four sequential scales).

35

Standard deviations of zero are considered perfectly fractal, whereas standard deviations

between zero and 0.05 are considered nearly fractal.

3.2.4 Species richness

Species richness values, that is, the number of species represented in a grid cell, were
tabulated for each fixed grid cell at each scale. Means, variances, and coefficients of
dispersion (CD) were calculated among all cells (when occupied) within a given scale.
Kolmogorov-Smimov tests for goodness of fit with respect to a Poisson distribution

were carried out.

3.2.5 Categories of rarity used in this study

Rare categories were defined in terms of two measures: area of occupancy A 10 and
geographic extent E. Using both the IUCN Red List guidelines as well as guidelines

adopted by the Joint Nature Conservation Committee, the criteria are defined as follows:

0 Very Rare, R] A < 100 km2 E < 100 km2

° Rare, R2 100 km2 < A (500 km2 100 km2 < E <5000 km2
0 Somewhat Rare, R3 500 km2 < A < 2000 km2 5000 km2 < E < 20000 km2
0 Common 2000 km2 < A 20000 km2 < E

36

3.3 RESULTS

3.3.1 Fixed grid versus minimum cell counts

Nearly all species in the study showed significantly higher cell counts based on fixed
grids compared to rrrinimized cell counts. Figure 3.5 shows levels of inflation at scales
10 and 160 km for all species in the study. Average inflation error at the smallest scale of
10 km is 14%, whereas at the intermediate scale of 160 km is 49%. Overall, inflation
rates are largest for species with the smallest distributions and inflation rates rise as the
measurement scale increases.

The cell count measures for the species Astragalus helleri found in Vera Cruz,
Mexico, provide a good illustration of the error introduced by using arbitrarily placed
fixed grids. Figure 3.6 shows the locations of the three sites and the position of the fixed
grids for scales J = 160 and 320 km. As chance would have it in this case, the three sites
for the species fall in three different 160 km cells, even though they are at most 36 km
apart. The intersection of grid lines happens to fall within the triangle created by the three
points. A comparison of fixed grid and minimized cell counts show sizable inflation

through the 160 km scale (see table 3.2).

3.3.2 Geographic measures and resulting rare categories

Species values for number of sites S, minimum cell counts C, and geographic extents E,

are given in table 3.3. Cell counts for areas larger than a species' geographic extent are

by definition equal to 1 and are indicated with shading in table 3.3. The four rarest

species (A. desereticus, A. columbianus, A. eurylobus, A. accumbens) have geographic

37

82 a 2 u 2 3.58 :8 8..

 

 

 

QM m.N o.N m. w 6... m6 :6
ling—flu.- 9 £6
D
‘ fi 3 a
E”— 48 4 U U
4 4 E . $3
4 4
M
4 4 a 4.8. m.
o
u
amwuquoa canon 5. 034
man—“I805 Oddom Eu— cHD
. #63.
ES»: afieuuhuvc
85.8 :8 £5.55:— 3 3.39:8 1 $oou

Sauce :58 =3 2..» Be...
an 23E

38

 

8:32 E
22.2. $3.54
he 8%. 8...: 2i.

3 as»;

 

 

 

M

/

N30 80>

583‘

E: can

 

E
a

 

 

 

 

 

 

39

Table 3.2 Comparison of fixed grid and minimized cell counts for Astragalus helleri.

 

 

 

Cell Scale of measure (J)

Counts 10 20 40 80 160 320 640 1280 2560
Fixed grid 3 3 3 3 3 l l 1 1
Minimized 2 2 l l 1 1 1 l 1

Inflation 50% 50% 200% 200% 200% -- -- -- --

 

40

Table 3.3 Sites S, minimum cell counts C, and geographic extent E for each species.
Sorted by C10 values. Area of occupancy A (kmz) is equal to CJZ. Shaded cells are scales
that are larger than the species' area of extent.

 

 

 

Species Sites Cells counts (C) at scale (J), in kilometers ‘ G§$:?:2;

Code (S) 10 20 40 80 160 320 640 1280 2560 (km)2

dm 1 1 1 1 1 1 ' 1- 1 _ 1 1 100

cola 1 1 1 1 1 ' 1 1 1 1 1 100

“1'? 2 1 1 1 1 1 1 1 1 1 100

"c“ 2 1 1 1 1 1 1 1 1 l 100

“3‘31 2 2 1 1 1 1 1 l 1 1 400

9“" 3 2 1 1 1 1 l l 1 1 200

1°“ 3 2 2 1 , 1 1 1 1 1 1 400

13.11 3 2 2 l 1 1 1 1 1 l 400

“11 3 2 2 2 2 1 1 1 1 1 4,800
“a" 3 3 2 2 2 1 1 1 1 1 2,700
pt“ 4 3 2 2 2 2 1 1 1 1 5,400
a.“ 5 4 3 3 1 1 1 l 1 1 3,500
1““ 5 5 4 2 2 1 1 1 1 1 4,200
cm 5 5 5 4 2 l 1 1 1 1 9,900
“1' 6 5 4 2 1 1 1 l 1 1 700
Plan 6 5 4 3 3 2 l l 1 1 22,000
1"“ 7 5 5 4 3 1 l 1 1 1 9,900
CY“ 7 6 3 3 2 1 1 1 1 1 6,000
1“" 8 6 4 3 2 1 1 1 1 1 12,800
1040 11 10 8 7 6 2 1 1 1 1 47,500
1119': 12 10 8 7 6 3 3 1 1 1 294,000
m“- 12 10 9 7 5 3 1 1 1 1 46,800
W 13 10 7 5 3 2 1 1 1 1 13,300
at. 14 11 9 8 6 4 2 1 l 1 203,500
8103 16 12 9 7 5 3 2 1 1 1 40,000
nudi 14 13 8 5 4 2 1 1 1 1 29,700
cfilt 15 l 3 10 8 4 2 l 1 1 1 55,100
0‘11“ 17 15 14 10 5 2 1 1 1 1 34,500
=14. 20 1 6 10 6 4 2 1 1 1 1 50,400
t0": 2] 18 15 13 10 6 3 2 1 1 374,400
pint 22 20 16 12 7 4 2 1 1 1 77,700
ell-0 25 22 16 12 8 4 2 1 1 1 107,300
new. 25 24 20 16 12 7 4 2 1 1 588,000

Continued

41

Table 3 .3 continued

 

ccumc
:pmumr
chum:
chum:
luau:
iaxll
town:
*ntaht
loan

urge
buck
anyh
nofib
n13.
noll

lent

30
39
43
41
46
60
64
64
68
72
87
97
114
119
201
320

426
739

26
35
37
39
40
43
56
59
62
69
80

88
105

H1

194
mm
MW
6”

22
30
28
32
31
33
45
50
51
59
71
76
92
98
179
flfl
In
flfi

14
19
23
22
23
23
35
39
41
48
57
57
69
82
151
210
241
345

10
10
12
13
13
14
22
21
26
31
38
35
38
59
117
142
132
188

OOO\OOO\J>O\

12
16
21
18
15
28
66
66
59
72

42

\lOOOQ-bAUI-bw4>UJNN

NNNNH
ht-‘UJUI—t

\JOO\O\O U1634>.n uthth¢lQ —tna~‘..._

wwmANu—‘t—‘NNu—av—au—an—u—su—ay—Iy—spu

flfl—lp—l—‘HHI—IflHHHHHflfiHH

225,000
79,200
144,400
434,000
148,500
199,500
496,800
318,200
374,400
634500
1,092,000
1,379,400
792,000
1,626,200
3,569,600
4,275,000
3,376,800
3,013,200

extents equal to 100 kmz, the smallest measurement area and smallest scale used in this
study.

Spearman and Pearson correlations among sites S, cell counts C10, and geographic
extent E, are shown in table 3.4. Among the three pairwise comparisons, a nearly perfect
correlation was found between sites S and minimum cell counts C10. Figure 3.7 shows
histograms of sites and cell counts. Figure 3.8 shows histograms of areas of occupancy
A 10 and geographic extent E. Figure 3.9 is a bivariate plot of the latter. Kolomogorov-
Smirnov tests for normality returned the following probabilities: log S = 0.978;
log C10 = log and A10 = 0.944; log E = nil. Assuming 95% confidence levels,
normality for the log number of sites S is not rejected while log cell counts C10 and log
areas of occupancy A 10 are just marginally rejected. Normality for log extent E is
rejected.

Species in this study were categorized for rarity by area of occupancy A10 and
geographic extent E. A summary of results appear in table 3.5, with dual-categorical
tabulation presented in table 3.6, which has been formatted to correspond with figure 3.9.
Rare categories for individual species appear in Appendix A. The locations of the 17
species categorized as "rare" or "very rare" for at least one type of measure are mapped in
figure 3.10. The four "very rare" species, Rl-Rl categories, are coded black on the map.
These species are very rare by both measures. Nine species fall in the "rare" category,
R2-R2 for both measures, and are coded light gray. Four additional species are in the R2
category for area of occupancy but are either "somewhat rare," R3, or "common" in
terms of geographic extent. These species are thus considered "rare" by only one

category and are also color-coded light gray.

43

Table 3.4 Spearman (Pearson) correlations among the three principal geographic
measures.

 

Log sites S Log minimum cell count C1o

 

Log minimum cell count C10 .997 (.996)
Log geogrphic extent E .956 (.934) .958 (.941)

 

Iii—“q— " ‘

 

«so» :58 :8 £255: .2.“ a» 85
:2 emu 3 2

 

 

 

 

iI‘I‘I . . I I‘ll a..lm. X IIIIII . . I‘I‘I J.
I I I . I I I . I I I r I I I
. . U . I I I . . . . I
I I I I I I I I I I I I
I I I I I I I I I I O I
I I I I I I I I I I I I
I I I I I I I I I. I I I
I I I I I I I I I I I I
I I I t I I I , I I I I I I .
IIIIII ., IIIIII . III II . J. IIIIII
I I I I I I I III _ I I I
I I I . I I I I I I . I I I
I I I I I I ., I I I .. I I I
IIIIII IIIIII IIIIII v. IIIIII,
I I I . I I I I I I I I I
I I I I I I. I I I I I I
I I I I I I I I I t I I I
I I I , I I I I I I I I I
I I I I I I I I I
I I I
I , . . .. I I I .
1 l . I I I
A . I I v
I I I
I I I
. I I I
I . , I I
. I I
,. I I .
I . I I .
. I
A , U
I I
I
I I
I I
I
0 I
I I
1
I
I
I I
I I
I

 

 

I. lfiL‘.‘ ~~.r .7.
l:‘\."t ,. \~'\r\

 

 

5- .

er

 

 

 

Sauefiseeouc. t
as. u an M2 “82:52.8.—
bm—IEIo: I8 38 SEE—miseweE—ev—

2 0 3:58 :3 5555:. as mega a: Ian—~52
hm «Sufi

 

 

 

 

8.": 2.8.“.
,6ng

 

 

sagoads to sequin"

45

AmhouoEo=¥2maamv no.3

 

mN 2.033 ems—Qahwcanv tau 3

QM ohflwmra

 

 

 

 

 

 

“swim E
Iguana-500 Io «Qin—

< 5.5558 .8 §<

 

 

 

 

 

 

 

 

 

 

 

 

 

 

46

 

Eat.» oEnEmomm mo;
ad «.m 9m o.m ed ad «.m QN c.~ I;
D b n n D I I r P Q. P
v
u r ed
m
n. n_ n.
1 6
n. D D D D D m
N a
U D an U
D n? 1U l N.” m.-
n o
D D D m
D D m U .m
n on a.» m
D 1 o
D D .A
D D
D
u .. I I
n.
U
W 2.08.5 emu—Q9300» mo:— msma?» S «x hugazooc a: 8.5 as
an 8&5 .. 9m

 

47

Table 3.5 Categories of rarity applied in this study.

 

 

 

 

Area of Occupancy (A '0) Geographic Extent (E)
Category Criteria In Study Criteria In Study
R1 5100 km’ 4 (3%) 5100 km2 4 (8%)
R2 100-500 km2 13 (25%) 100-5000 km’ 9 (18%)
R3 500-2000 km2 14 (27%) 5000-20000 km’ 6 (12%)
Common >2000 km2 20 (39%) >20000 km2 32 (63%)

 

Table 3.6 Species categorized by area of occupancy A and geographic extent E rarity.

 

 

 

 

 

Common 20
Area
of R3 3 1 1
Occupancy R2 9 3 1
(A...) R1 4
R1 R2 R3 Common
Geographic Extent (E)

 

48

32x22 .NEU Eo>v .233: .v. nfln

mfiuuuhnax .—

.: 2: .8 860% 0.5.. an: ecu 0.5.. no .5539— eud Paar—

 

a==u~§w5om= .<. ..

can.
..._u

  

 

 

82.23 .<

 

 

3.9.amvwwuw .«u . .

 

 

.5

 

    
 

.3392» . . . .
815.58 .<

 

 

 

 

 

 

33m :chfimué §=S¢E=~3 .< I

49

3.3.3 Among scale comparisons

Pairwise Spearman and Pearson correlations among scales, based on areas of occupancy
A, are shown in table 3.7. The highest correlations were found at the smallest scales of
measure. Further, correlations were always highest for scales nearest in size. Figure
3.11 is a chart of the values in table 3.7.

The sequential ranks of species at each scale are given in table 3.8. Shaded cells in
the table indicate rank changes for a species. The number of species with changes in rank
varied from as low as 2 at the largest scales to a high of 26 (or 51%) between scales
[J = 40 and 80 km]. The magnitude of rank changes, taking into account the size of each

rank change, varied similarly. Figure 3.12 provides a summary of these changes.

3.3.4 Fractal geometry of species distributions

Table 3.9 gives each species' fractal dimensions D for the eight pairs of scales compared.
Note that a significant portion of the cells in the table are shaded, indicating D when it is
equal to 1. As previously mentioned, this occurs when the scales of measure exceed the
geographic extent of a species. Visual inspection of this table reveals that species'
distribution are generally not fractal across scales. Indeed, no species is perfectly fractal
across the full range of scales examined in this study. Scale-area curves on a log-log plot
are shown in figure 3.13. The slope between each pair of points is equal to D. The bold
diagonal line graphically represents the position where geographic extent equals
measurement scale, and D, therefore, equals 1.

The four rarest species (A. desereticus, A. columbianus, A. eurylobus,
A. accumbens) do not have fractal dimensions because they are too rare. Four additional
species, A. uncialis, A. funereus, A. loanus, and A. helleri, have sufficiently small range

sizes such that they do not have a sequence of three dimensions for comparison. Among

50

 

 

$.93 35. 9359 5mm. Gmbg boo. Avg—5V Sm. ESQ cwm. €©©Q vwm. 3.89 gm. cwflu
Agog ohm. A259 mmw. Amcwg 2w. ammo mow. Ammwo was. 8;.V was. 3%

$39 30. A309 mam. Ammoo 2o. Cbmg Goa. Abowg mam. can

23¢ m3. Avcog ova. Ammag Vmo. Amvoo N3. :3

33¢ cwo. Swag Eb. Avgg #3. ca

Amomo 5a. awao gm. 3.

Avmmo mmo. ca

:3 cmm :3 an 3. an A: Q.» 23m

 

.< 3:338 mo £83 .8 838 muons macaw—9:8 385?: 55.8on Em 033.

51

3 28m
owe cmn cap on ow cu o.

suonelauoo

 

camp U 93 fl own I oo F I oo I 2‘ E ON D

816% uEE—w 95:29.39 525.3% omEmam
5a 2:5

52

Table 3.8 Sequential rank of species at each scale based on area of occupancy A. Ranks
were not determined for the largest scale [J=2560] because all species have the same
area. Shaded cells indicate rank changes. See text for further details.

 

Sequential rank for each scale (J), sorted at [=10

 

 

 

Sluxjes
Code 10 20 40 80 160 320 640 1280
4". 1 1 1 1 1 1 1 1
°°1u 2 2 2 2 2 2 2 2
'“rY 3 3 3 3 3 3 3 3
new 4 4 4 4 4 4 4 4
““91 5 5 5 5 5 5 5 5
fun. 6 6 6 6 6 6 6 6
1°‘” 7 7 7 7 7 7 7 7
h'11 8 8 8 787 8 8 8 8
“‘11 9 9 9 1 1 l 1 1 1 1 1 l 1
'“bv 10 10 10 12 12 12 12 12
Pt“ 11 11 11 13 18 18 18 18
“'0“ 12 12 14 10 10 10 10 10
1“t° 13 14 12 14 13 13 13 13
°°n' 14 18 18 17 16 16 16 16
‘31. 15 15 13 9 9 9 9 9
P"“ l6 l6 16 18 19 19 19 19
‘9'“ 17 19 19 19 17 17 17 17
“Y'3 l8 13 15 15 14 14 14 14
1'“° 19 17 17 16 15 15 15 15
1°40 20 21 23 27 25 25 25 25
gig: 21 22 24 28 28 33 33 33
"u'1 22 24 25 24 26 26 26 26
cymb 23 20 2O 2O 20 20 20 20
"t0 24 25 27 29 29 28 28 28
”1°“ 25 26 26 25 27 27 27 27
nudi 26 23 21 21 21 2 1 21 21
9"t 27 27 28 23 23 23 23 23
'“r' 28 29 29 26 24 24 24 24
“id. 29 28 22 22 22 22 22 22
t°tr 30 3O 32 32 33 34 36 36
Piut 31 31 30 3O 3O 29 29 29
°"° 32 32 31 31 31 3O 30 3O
“‘9' 33 33 34 35 37 37 37 37
Continued

53

Table 3.8 continued

 

 

 

°°¢° 34 34 33 33 34 32 32 32
9‘" 35 36 35 34 32 31 31 31
cm 36 35 37 36 35 35 34 34
“1°” 37 38 36 37 38 38 38 38
1W“ 38 37 38 38 36 36 35 35
in“ 39 39 39 39 39 39 39 39
“9h 40 40 40 41 41 42 42 42
““11 41 41 41 40 40 40 40 40
1°“ 42 42 42 42 42 41 41 41
00911 43 43 43 43 44 44 44 45
“‘90 44 44 44 45 46 46 46 46
he" 45 45 45 44 45 45 45 44
mph 46 46 46 46 43 43 43 43
a.” 47 47 47 47 47 47 47 47
mi" 48 48 48 48 49 51 51 51
“011 49 49 49 50 50 49 49 49
9“" 50 50 50 49 48 48 48 48
1'“ 51 51 51 51 51 50 50 50

Count of

changes 18 20 26 23 11

Magnitude

of changes 32 35 50 38 18

 

54

coca—=00 «E mofiom
camp m> SB 30 m> can can m> cow 8— m> 8 on m> ow ow w> on em m> 3 3 m> mozm

 

 

 

 

 

omcuzo 05.03“ 3 8.6 E
«5.50 nu

 

23m 833 8935 :5.— 155.5%
«3. 2:3...

55

Table 3.9 Fractal Dimensions D for each species between reference scales. Shaded cells
indicate D values which are 1 by definition. List sorted as in table 3.3. Doubly
underlined values represent scales for which the species' distribution is perfectly fractal;
singly underlined represents nearly fractal distributions.

 

 

Sp eci es Fractal Dimensions D between reference scales J
Code 10-20 20-40 40-80 80-160 160-320 320-640 640-1280 1280-2560

 

dos. 1.00 1.001 "1.00 1:00 ‘ 2100"}; 41.00 1.00 1.00
colu 1.00 1.00 1.00 .g';;1.oo* .3100. [1.00 1.00 1.00
cm 1.00 . 1.00 1.001 100 ‘ 100 3 1.00 1.00 1.00

 

30°“ 0.79 1 00 0.21 1.00 100 1 00 1 00 1 00
lut° 0 84 0 50 1 00 0.50 100 100 1 00 l 00
CO!“ 100 084 050 050 ' 100 100 100 100

 

 

fun 1.00 0 84 0 79 0.21 1.00 ; 1.00 1.00 1.00
cyan 0.50 1.00 0.71 0.50 ‘ 1.00 1.00 1.00 1.00
low 0.71 0.79 0.71 0.50 1.00 1.00 1.00 1.00
iodo 0.84 0.90 0.89 0.21 0.50 1.00 1.00 1.00
stic- 0.84 0.90 0.89 0.50 1.00 0.21 1.00 1.00

 

““1 0.92 0.82 0.76 0.63 0.21 1.00 1.00 1.00
cm 0.74 0.76 0.63 0.71 0.50 1.00 1.00 1.00
“t. 0.86 0.92 0.79 0.71 0.50 0.50 1.00 1.00

 

 

 

 

 

 

 

 

11°“ 0.79 0.82 0.76 0.63 0.71 0.50 1.00 1.00
““41 0.65 0.66 0.84 0.50 0.50 1.00 1.00 1.00
“'15 0.81 0.84 0.50 0.50 0.50 1.00 1.00 1.00
"I" 0.95 0.76 0.50 0.34 0.50 1.00 1.00 1.00
‘11“ 0.66 0.63 0.71 0.50 0.50 1.00 1.00 1.00
“tr 0.87 0.90 0.81 0.63 0.50 0.71 0.50 1.00
91111: 0.84 0.79 0.61 0.60 0.50 0.50 100 1.00
c“. 0.77 0.79 0.71 0.50 0.50 0.50 1.00 1.00
mg: 0.87 0.84 0.79 0.61 0.60 0.50 0.50 1.00
Continued

56

Table 3.9 continued

 

coca
par:
chm
char

layn
11:11

£091:
utah
10d.

back
much
new}:
nin-
noll
purl
lent:

0.88
0.89
0.80
0.86
0.82
0.81
0.84
0.88
0.86
0.89
0.91
0.89
0.90
0.91
0.94
0.90
0.88
0.83

 

 

 

 

 

 

 

 

0.67 0.76 0.63 0.21 0.50 1.00 1.00
0.67 0.54 0.34 0.50 0.50 1.00 1.00
0.86 0.53 0.50 0.50 0.21 1.00 1.00
0.73 0.62 0.65 0.50 0.50 0.50 1.00
0.78 0.59 0.44 0.50 0.21 1.00 1.00
0.74 0.64 0.60 0.50 0.50 0.50 1.00
0.82 0.67 0.50 0.43 0.34 0.50 1.00
0.82 0.55 0.30 0.50 0.50 0.50 1.00
0.84 0.67 0.44 0.21 0.50 0.50 1.00
0.85 0.68 0.52 0.40 0.39 0.71 0.50
0.84 0.71 0.57 0.39 0.42 0.50 0.50
0.79 0.65 0.52 0.42 0.50 0.00 1.00
0.79 0.57 0.33 0.45 0.39 0.21 1.00
0.87 0.76 0.46 0.33 0.43 0.34 0.50
0.88 0.82 0.59 0.30 0.26 0.42 0.00
0.84 0.72 0.45 0.24 0.32 0.21 0.21
0.76 0.57 0.42 0.25 0.30 0.29 0.21
0.73 0.56 0.31 0.21 0.11 0.39 0.21

 

57

the remaining 43 species, 7 are perfectly fractal across four scales of measure (three
measures of D), and 14 additional species have nearly fractal dimensions across four or
more scales. By this criterion, nearly 50% of species show a high level of fractalness
across at least four scales of measure. On the other hand, among 166 sequences of four
scales (three measures of D), only 29 (17%) of them are fractal by this criterion. These
sequences are indicated by underlining in table 3.9. Perfectly fractal sequences are
doubly underlined, whereas nearly fractal sequences are singly underlined.

As a group, D was highest at the smallest scales and declined with the larger scales.
Within scale variance increased with increasing scale owing to the smaller number of

species with range sizes at large scales (see figure 3.14).

3.3.5 Species richness

Species richness among individual cells varied substantially with scale. At the smallest
scale (J = 10 km), species richness averages just greater than one species per cell (when
occupied) with a maximum of three species per cell. The largest scale analde (J = 640
km) had an average of more than seven species per cell and a maximum of 23 species per
cell. Table 3.10 shows species richness counts among the six scales analyzed.
Kolmogorov-Smirnov tests for a goodness of fit with a Poisson distribution returned
very low probabilities (data not shown). Coefficients of dispersion, CD (variance/mean,
expressed as a decimal), are shown in table 3.10. CDs were less than 1.0 for scales
J = 10, 20, and 40 km, indicating that at these scales, species distributions are repulsed
(i.e., contrary to clumped). Species richness counts are clumped at scales J = 80, 160,
320, and 640 km. A plot of mean and variance values among scales (not shown)
indicates values cross between scales 40 and 80 km. This suggests that, with respect to

species richness, cell counts may have a Poisson distribution at a scale near 60 km.

59

Ooh—3:30 moiom
2: 2 8

OOMN on OONF OONF 0. Ovm Ovm 8 ONO ONO 2 Our

OO 3 cc

OV 3 ON ON 3 O_.

 

n O _. mN an

8.3% 896: "1,535.56 .5931.—
38 2:5

we

39v.—

no

 

_.O.

we

 

he":

«$8 a
.cotoucom
.3530me

1 NO

1 MO

I
‘1
C

1 m6

1 9O

1 Nd

I
“2
O

1 ad

 

(lo %96) a 0°!909w!al91331:l

60

 

 

N? «Na o: 8 N _ _ _ _ _ N _ N N :3

m3 2: 0N1 3 _ N _ N N _ _ N m _ _ w c 2 8 can
SN a: 42 a: _ 1 v o m 4 o_ 2 3 MN 9. mm :3

NZ 3N 3N 9:. _ _ o N 2 nN 8 K 9: m2 8

we. 02 $2 82 m m N t. NE SN m3 3.

ON. 33 oN._ E: N _ t 2 2m mm: ON

:. a. _.o :._ NSN .. om 8N SNN :—

8 B. .62.. a NN 2 2 2 2 3 2 N_ 2 3 a a b e m w m N _ 3 23m

 

awe—=35— 832mm

 

28% some .8 :33 Ba 56336
ho E20508 can 625:? .532 .Ex owe smack”: Ex 2 u H 3.3m c8 m=oo 35:8 30:20: 860% E cougar—3’ o_ .m 035.

61

The largest numbers of species rich cells fall in Utah and Nevada. At the smaller scales
they are predominantly found in Utah, whereas at the larger scales, Nevada has the
greater number of species rich cells. Figures 3.15 and 3.16 show species richness in
Utah at scales J = 40 and 80 km. Tabulation of species richness by state boundary,
rather than grid cells, found 24 species located in Utah and 23 species in Nevada.
Among the two states, 14 species are common to both, and together they contain 33
species, or 65% of the total number of species in the study group. Additionally, these
two states have the majority of the 17 rare species (as defined above) in the group: 6 in
Nevada, and 4 in Utah (see figure 3.10).

The region with the largest cluster of species rich cells is found in southern Utah,
especially on and adjacent to the Utah Plateaus and in the area south extending toward the
north rim of the Grand Canyon and extending cast into the Pine and Bull Valley
Mountains. A second area of high species richness lies on the eastern flanks of the Sierra
Nevada Range in California, extending across the White Mountains and including the
southern portion of the Toiyabe Range. As expected, the center points of maximum
species diversity depend on the measurement area. They are nevertheless very near the

most species rich cells.

3.4 DISCUSSION

Perhaps the most significant result from this study was the discovery of high levels of
variation in the area of occupancy measure A due to variation in the placement of grids.
The "Minimum Cell Count" procedure was developed to eliminate this arbitrary variation.
A comparison of values derived from an arbitrarily placed grid versus placements that
minimized A revealed large differences for species with small range size and/or a small

number of localities. Other procedures might be employed to eliminate this variation,

62

24'

1.8.
3D

5033925

23... Exam
luuoaaot
228%
:3:

m3 2:2...

 

 

 

 

63

 

 

E

"S m
2 .2
5138
12:5-

richness-

40km scale

Species/Cell

 

 

such as taking the mean of 10 random grid placements; however, sampling error due to
estimating the mean will be influenced by the spatial pattern of a particular species and
will thus vary among species.

These finding have important implications for studies that include species with highly
divergent range sizes. Indeed, when area of occupancy is used for assigning rare status
to species leading to listing for special protection, it is possible that some species may not
attain status and listing by fixed grid methods whereas they would if a minimization
procedure were employed. Given the importance of a sound and repeatable process for
assigning special status to species for conservation protection, the arbitrary nature of
fixed grid methods should be avoided. Furthermore, fixed grids should be avoided to
reduce experimental error in studies of biogeography.

The effect of scale with respect to the area of occupancy measure is readily apparent
when the scale of measure is large compared with the areas occupied. Large
measurement scales, in effect, summarize variation at smaller scales and thus reduce
information content. This highlights the need to use scales that correspond with species
with the smallest area, especially for studies that focus on rare and more restricted
species.

Rank abundances are also strongly influenced by changes in scale due to the
interaction between a species' distribution, the scale in which the data was collected, and
the scale in which the data is summarized. Nearly all species in the study changed rank
position as scales changed, the extreme example being Astragalus giganteus which shifted
from position 20 at the smallest scales to 33 at the largest scales. This suggests the need
to choose scales carefully when employing ranks.

The use of fractal properties to mitigate the among scale effects discussed above
appears to be problematic. Based on results in this study, species are not strictly fractal
across the scales employed in accordance with the findings of others (Krummell, et. al.

1987; Allen and Hoekstra 1992).

65

The majority of species in the current study have fractal dimensions near the value of
1 between scales J = 10 and 20 km. Two possible explanations are offered: 1) The
numbers at these scales may simply be a sampling artifact. If the average distance
between sites is substantially greater than 10 km and sampling (collecting) was done
systematically leading to over—dispersion, this would lead to high values of D at the
lowest scales; and 2) the organisms themselves may be more widely distant than the
sampling measure and/or the species may be systematically distributed due to ecological
factors. Unfortunately, both of these effects can lead to higher values. Untangling the
extent to which these factors affect the data set is difficult to elucidate.

Recent work by Kunin (1998) made use of fractal properties to predict areas of
occupancy of scarce plant species at scales finer than the scale at which data was
collected. Kunin did not quantify the fractal properties of the species he studied,
however; rather, he assumed the distributions were fractal and used this assumed
property to derive predicted-values. Predicted values were generally greater than a second
set of observed data collected separately at the fine scale. He concluded that either
undersampling occurred at the fine scale or that species are not strictly fractal over the
range of scales employed. Given the results from the current study, Kunin's later
suggestion is clearly supported. His first suggestion may also be simultaneously
influencing the outcome.

These findings raise concerns about using fractal dimensions for estimating the
abundance of a species at scales smaller than the scale for which data has been collected.
However, given the effect of variation introduced by using fixed grids, there may yet be
hope for refining a procedure similar to that of Kunin (1998).

Regarding findings on species richness, because values were derived from fixed
grids, the largest values found were not the maximum possible. Indeed, an informal
search for higher maximum species richness counts at scales 160 and 320 km netted

higher values than with the fixed grids—17 species versus 13 for the former, and 21

66

species versus 18 for the later. Procedures to optimize species richness values were not
undertaken due to the computational difficulty and the lack of a rationale for doing so.

One of the more interesting aspects of the findings on species richness is that high
values correspond with the boundaries of the Great Basin Floristic Province (Bameby
1989) (see figure 3.17). Indeed, a ridge of species richness contours (not shown) runs
parallel east of the boundary from central Utah, continuing south and curving to the west
in southern Utah. The high level contours coincide with the Utah Plateaus.

Two explanations are offered for these patterns. The first rests with the logical
consequences of how floristic regional boundaries have been determined. Boundaries are
defined by areas with high rates of change in species assemblies and/or areas of regional
endemism. Floristic traits such as these lead to higher species richness values near
boundaries.

Perhaps a more encompassing explanation for these correspondences derives from the
underlying topographic heterogeneity along the boundaries. Not only do the topographic
features act as barriers to dispersal, but the topographic complexity may lead to higher
rates of diversification (Cracraft 1985). Topographic discontinuities also have large
impacts on local climate patterns, which further create environmental heterogeneity.
Thus, for both ecological and evolutionary reasons, it is not surprising that high levels of
species richness are found in south-central Utah and east-central California.

During the search for the best possible source(s) of biogeographic data for this study,
the large databases maintained by state governments and nonprofit organizations were
seriously considered. Indeed, I held lengthy discussions about the attributes and
availability of these data with a number of individuals involved in rare plant monitoring.

Several matters became clear following these discussions and are worthy of mention.
First, because each state or regional unit maintains its own data, the effort required to
retrieve and organize the data would be enormous; further there would be no guarantee of

obtaining all the data desired due to varying policies of dissemination. Further

67

 

 

....ZNM. <m...~. «161.6...

 

.... 1.
8:2 :u. 8 e s. . ...
lm=ou $233.. ..
“38% :3: . 1.
SN 2:5 1 .....

    

complicating matters is the fact that each unit stores data in different formats and uses
different criteria for measuring and categorizing the species they monitor. Thus, there is
great variation within and among the data sets attributable to the various individuals and
the specific agencies and organizations involved. Sorting through this quagmire would
have required more effort than a single individual could accomplish in a reasonable
amount of time. For these reasons, use of data from these sources was dropped early in
this study.

The possibility of compiling data directly off herbarium sheets was also entertained
but eventually rejected. It would be impractical, and probably inadvisable, to have
thousands of herbarium sheets sent to one’s home herbarium for this type of work.
Visitations at the herbaria with the principal Astragalus collections would have been a
better alternative, although that would have been costly and very time consuming.
Another contributing problem is that many Astragalus collections have not been curated
for some time. Sorting out synonymy, etc. would have added even more time to the
effort. These limitations left only one viable alternative—the use of Barneby's (1964)
monographic work as the source of biogeographic data.

Bameby spent more than 25 years compiling data on biogeographic and other
attributes of Astragalus in N. America. His distribution maps are exceptional because:
1) he provided dot maps for nearly every species in his monograph, and 2) he provided a
great deal of data, estimated at more than 20,000 points, for the whole monograph.

In spite of the above strengths, his data contain a few weaknesses. First, the data are
now 30 years old. New sites and new species have been discovered since publication of
his monograph. Second, Barneby's distribution maps were constructed using different
map projections and scales. This required careful digitizing to overcome. Third, error
rates for plotting points are difficult, if not impossible, to fully ascertain because specific
plotting rules were probably not followed. If they were, they are not described and

remain unknown. Even with these weaknesses, however, the data set is exceptional.

69

The number and location of sites S were taken directly from Barneby's monograph;
thus, the quality of his distribution maps affects both the area of occupancy A and
geographic extent E measures. For both these measures, the accumulation of error
resulting from plotted maps and from digitizing is mitigated by the summary procedure of
converting points to areas. Furthermore, the summarization procedure diminishes any
underrepresentation of the true number of sites. The difference between actual versus
measured occupancy diminishes as the area of measure increases to the overall extent of
the species.

Collecting bias also affects some of Barneby's data. For example, southern Nevada,
which includes the inaccessible areas of the Nevada Test Site and Nellis Air Force Base,
has a dearth of sites. This is not surprising considering most of Barneby's work
corresponds with some of the most intense periods of nuclear testing in these areas.

Regardless of these reservations, it is satisfying that there is close correspondence
between the categories of rarity applied to data presented in this study and the listing of
rare species in the IUCN Red list. Among the 17 "rare" species categorized in this study,
14 appear on the Red List. Astragalus helleri is not on the Red List, quite possibly
because data from Mexico did not include it—in spite of its obvious rarity. Astragalus
subvertitus did not make the Red List either, although it is on the California Native Plant
Society's (CNPS) Inventory of Rare and Endangered Vascular Plants of California
(Skinner and Pavlik 1994). A California endemic, this species is not vulnerable at this
time but is rare according to CNPS. The classification used in this study places this
species in the R3 and R3 categories—somewhat rare for both A and E. Astragalus
lutosus also did not make the Red List for unknown reasons; Bameby (1989) suggests
that this species should be closely monitored because of its potential endangerment.

Four species in this study group made the Red List but did not meet the R2
requirement in at least one category for this study. These include A. cymboides and

A. leucolobus, both classified as an R3-R3 in this study; A. musiniensis, classified as an

70

R3 for A but common for E; and A. shortianus, classified as "common" for both
measures. This latter species' native range lies east of the Rockies and corresponds to a
large extent with the Denver metropolitan area. This species is probably becoming rare

due to human activity.

The principal aim of the work presented in this chapter was to quantify the
biogeographic distributions of the 51 species of the Large-flowered Piptoloboid group of
Astragalus. Data from this study were also used in the study discussed in Chapter 4.

In the interest of obtaining the most robust results possible, an exploration of area of
occupancy measure A with respect to two methods of determination (fixed grid and
minimization) and the effect of measurement scale on resulting quantities were
undertaken. As predicted, these factors had large effects on the resulting quantities.
Results led to the use of the minimization procedure to reduce arbitrary variation in the
area of occupancy measure A and to the use of the 10 km scale to reduce information
loss. The geographic extent measure E, which is far less subject to the vagaries of grid
placement and scale, was also determined because it is typically used for rarity
classification. Using these results, species were categorized into rarity classes as per
IUCN's criteria.

Additional analysis, secondary to the principal aims of this study, included
assessment of the fractal properties of the species distribution to test the possibility of
using fractal dimensions as a scale-independent distributional attribute. Results show that
distributions are generally not fractal. Further analyses are required to elucidate the
underlying mechanism generating the curvilinear fractal relationships uncovered for most
species.

Finally, a brief analysis of patterns of species richness of the 51 study species

revealed high levels of diversity in the Great Basin and the Utah Plateaus to the east.

71

Specific regions of high diversity depended on the scale of measure and underscore the

need for choosing measurement scales with reason when undertaking diversity studies.

72

4
Phylogenetic Patterns of Rarity
in the Argophyllean Clade of Astragalus L.

4.1 INTRODUCTION

In the introduction of this dissertation, several questions were raised regarding the
relationship between rare status and the processes of evolution. Among the many vantage
points for addressing the nature of this relationship, work presented in this chapter
focuses on the role evolution may play in explaining the origins of rare species.

Considerable evidence suggests that the majority of speciation in higher plants occurs
locally—either via peripatric speciation or via chromosomal reorganization (Levin 1993).
Indeed, it has also been suggested that peripatric speciation is the most common mode
among animals (Brooks and McLennan 1991). Diversification by local processes leads to
rarity of at least one sister lineage during the early stages following a cladogenic event,
unlike allopatric speciation which takes place within large regional or continental areas
without the necessity of rare status.

To the extent that a lineage has undergone a shift toward higher rates of
diversification, we might expect to see a larger proportion of rare species in that lineage

compared to its sister lineage or compared to a more inclusive and larger group within

73

in

which it is found. If this scenario is true, macroevolutionary patterns of rarity are likely
and may be discernible with a well resolved species level phylogeny.

The aim of the work presented here was to explore patterns of rarity among species
within a monophyletic assembly. Specifically, the following hypothesis was tested:
Lineages undergoing higher net rates of diversification, when coupled with local
speciation, generate a disproportionate number of species with small range sizes
compared to lineages with lower net rates of diversification. If this hypothesis is true, the
following features are expected within the group: 1) evidence of diversification rate
variation; 2) phylogenetic clustering of rare species in the region of high diversification;
and 3) a preponderance of newly derived rare species. Testing the generality of this
hypothesis requires study of many groups and is beyond the scope of this study. Here, a
single group will be used as a case study.

Using the Argophyllean clade of Astragalus, identified during the phylogenetic study
presented in chapter 2, and results from the biogeographic study of the group presented in
chapter 3, the following questions are addressed: 1) Is there evidence of diversification
rate variation among lineages? 2) Do rare species cluster in the phylogeny? 3) Are newly

derived rare species more frequent than expected by chance?

4.2 METHODS

The Argophyllean clade of 47 species was identified during a morphological cladistic
study of the Large-flowered Piptolobi of Astragalus (presented in chapter 2) and was
used as the study system. The 12 most parsimonious trees recovered were very similar
and resulted in a highly resolved strict consensus tree. Furthermore, all branches leading

to single rare species were fully resolved. A randomly chosen tree among the 12 most

74

parsimonious trees was used for this study and is shown in figure 4.1. Extensive
preliminary testing using alternative trees among the 12 revealed nearly identical results.

Range size and classification of rarity for the 47 Argophyllean species were taken
from results presented in chapter 3. Log values for the minimum area of occupancy A
based on the number of 100 km2 grid cells occupied and log values for geographic extent
E were used. Rare species were defined following criteria used for the I 997 Red List of
Threatened Plants (Walter and Gillett 1998) and are indicated on the phylogeny in
figure 4.1.

Evidence of diversification rate variation within the clade was assessed using
Kirkpatrick and Slatkin's (1993) phylogenetic tree asymmeU'y test. Average internal node
distances from the inferred common ancestors to terminal taxa were tabulated for each
subclade with eight or more species—l9 in all. Results were compared with confidence
intervals based on data provided by the authors. Average internal node distances above
confidence intervals provide evidence to reject the null hypothesis that the subclade is not
asymmetrical.

A preliminary assessment of rare species clustering was attained by plotting the
number of rarest species versus the total number of species for each of the 46 subclades
within the study group. For purposes of this plot, both "Very Rare" and "Rare" species,
as defined in section 3.2.5, were combined into a single category comprising the rarest
species. Among the 47 species in the group, 17 (36%) met this criteria.

The Mantel procedure (Sokal 1979; Manly 1997) was adapted to statistically test for
the clustering of rare species within the phylogeny. Specifically, the association between
topological distance as measured by internal node distances and dissimilarity in log range
size was tested. Two tests were conducted, one using log area of occupancy A and one
using log geographic extent E. The two half-matrices used in each procedure contained
the following data: 1) pairwise internal node distances among species and 2) pairwise

differences of log A and again of log E. Figure 4.2 shows how these data were

75

Figure 4.1

Phylogeny of the Argophyllean clade
of Astragalus used for analysis in
chapter 4 with area of occupancy A
and geographic extent E categories
indicated for each species. Values
correspond with those in table 3.5.

. Very Rare - R1
Q Rare - R2
0 Somewhat Rare - R3

Common (no symbol)

 

J.

.600 b

 

 

_,.|

O 06 6 60

O O

 

 

O $6 06660 O.

O 66 $66

 

 

 

 

 

m

argophyllus
zionis
cyaneus
callithrix
columbianus
parryi
waterfallii
feenis

O .60

layneae

casei
pterocarpus
tetrapterus
iodanthus
pseudiodanthas

O

tidestromii
amphioxys
leucolobus
tephrodes
eurylobus
iodopetalus
shortianus
nudisiliquus
subvosfitus
funereus
neomexicanus
giganteus
mollisimus
helleri
anisus
utahensis
inflexus
desereticus
piutensis
newberryi
purshii
uncialis
musiniensis
charnaeleuce
cymboides
missouriensis

0 O6

66 O O O 6

accumbens
eurekensis
castaneiformis
consobrinus
coccineus
loanus

$6 600.

lutosus
OUTGROUP

Figure 4.2

Phylogeny of the Argophyllean clade
of Astragalus with an example of one
set of pairwise data used for the
phylogenetic clustering test: (a) The
dissimilarity of log area of occupancy
A is 0.2 (data from chapter 3); and (b)
the topological distance is 7.

(b)

 

l.

 

 

A
l‘ v
V 1

 

 

 

ll.

 

 

 

 

T

zionis
cyaneus
callithrix
columbianus
parryi
waterfallii
feenis
layneae
casei
pterocarpus
tetrapterus (a)
iodanthus
pseudiodanthas
tidestromii
amphioxys
leucolobus

te nhmdes
eurylobus
iodopetalus
shortianus
nudisiliquus
subvestitus
funereus

 

neomexicanus
giganteus
mollisimus
helleri

anisus
utahensis
inflexus
desereticus
piutensis
newberryi
purshii
uncialis
musiniensis
chamaeleuce
cymboides
missouriensis
accumbens
eurekensis
castaneifomiis
consobrinus
coccineus
loanus
lutosus
OUTGROUP

determined for this procedure. Corresponding matrix values were multiplied and
summed to determine the Mantel coefficient observed values. The null hypothesis, that
there is no association between topological distance and dissimilarity in log range size,
was tested by comparing the observed value with a reference distribution of 1000
coefficients derived from the randomization of the first matrix with respect to the second.

A test for the preponderance of newly derived rare species was conducted by first
counting the number of the rarest 17 species among the 14 terminal pairs in the
phylogeny. Figure 4.3 shows the position of the terminal pairs and the rarest species
used for this test. Next, the null hypothesis, that the number of rare species among
terminal pairs is not different than expected by chance, was tested by comparing the
observed value of 11 with a reference distribution of 200 values obtained by

randomization of species placement in the phylogeny.

4.3 RESULTS

Among the 19 subclades within the Argophyllean clade tested for asymmetry, 11
were found to be significantly asymmetrical. Results are charted in figure 4.4.
Significantly asymmetrical subclades are indicated on the phylogeny in figure 4.5. An
unbroken internal node link with significant asymmetry begins at the base of the
phylogeny and extends to the node which includes A. nudisiliquus. Continuing up the
phylogeny, the node with A. callithrix is also significantly asymmetrical while the two
nodes below it are marginally not symmetrical. Results provide strong evidence, given
the phylogeny, that net diversification rates vary significantly within the group. The
region of the phylogeny with the highest net rate of diversification is indicated by an

asterisk (*) in figure 4.5.

78

Figure 4.3
Phylogeny of the Argophyllean clade of
Astragalus with positions of the 14

terminal pairs and the 17 rarest species.

Terminal pair

0 Rarest species

 

 

 

 

 

 

 

 

 

 

 

 

 

79

argophyllus
zionis
cyaneus
callithrix
columbianus
parryi
waterfallii
feenis
layneae
casei
pterocarpus
tetrapterus
iodanthus

pseudiodanthas

tidestromii
amphioxys
leucolobus
tephrodes
eurylobus
iodopetalus
shortianus
nudisiliquus
subvestitus
funereus
neomexicanus
giganteus
mollisimus
helleri
anisus
utahensis
inflexus
dwereticus
piutensis
newberryi
purshii
uncialis
musiniensis
chamaeleuce
cymboides
missouriensis
accumbens
eurekensis

, castaneiformis
, consobrinus

coccineus
loanus
lutosus
OUTGROUP

8
(D

O

O
O
O
G

O

O

O

ommocfi c_ avian—w no con—:32

geese—Gaza
89.779089”:

817
917
W
37
OZ

Irlrlrlrlr
89.7308

09
0V

 

82 .ceaam :5 caresses. 52. 85o:

 

B~oEE>m

   

    

.5335 00:25:00 .xbm

ESE—Em"

O .m—EOEEO ma O88:— o...“ Egon—Emma ESE—a?
NE? nova—2.5m . usuuehufi we ova—u :aflisaefla

2: 5.53 mots—2.3 a 2: he 3mm: ESE—nag.

3. 2:5

 

O_.

5.

NF

aouersga apoN leuraiul aBeJaAv

80

Figure 4.5
Phylogeny of the Argophyllean clade of

Astragalus with the positions of significantly

asymmetrical subclades indicated with a

black diamond O . The asterisk (*)

indicates the region of the phylogeny with

the highest net rate of diversification.

 

 

 

 

 

 

 

 

 

 

 

 

81

argophyllus
zionis
cyaneus
callithrix
columbianus
parryi
waterfallii
feenis
Iayneae
casei
pterocarpus
tetrapterus
iodanthus
pseudiodanthas
tidestromii
amphioxys
leucolobus
tephrodes
eurylobus
iodopetalus
shortianus
nudisiliquus
subthitus
funereus
neomexicanus

inflexus
dosereticus
piutensis
newberryi
purshii
uncialis
musiniensis
chamaeleuce
cymboides
missouriensis
accumbens
eurekensis
castaneiformis
consobrinus
coccineus
loanus
lutosm
OUTGROUP

The preliminary assessment of rare species clustering is shown in figure 4.6. Each point
on the bivariate plot represents the relative proportion of rare species in each of the 46
subclades with respect to the expected proportion of 0.36, indicated by a line. Points
above the line represent subclades with higher proportions of the rarest species compared
to the expected value. Inspection of this plot reveals points that are not far from the
expected values. The statistical properties of this assessment are not known.

Results from the phylogenetic clustering test using the log area of occupancy A,
shown in figure 4.7, returned a probability (p = .089) and points toward an association
between topological distance and area of occupancy A but does not meet standard
significance levels. Results using log geographic extent E returned a probability
(p = 0.25) and does not indicate an association. This latter result is shown in figure 4.8.

Of the 17 rarest species in the study group, 11 were found among the 14 terminal
pairs. The 200 trees with randomized placement of species averaged 9.8 rare species
among the pairs. The distribution of these values is shown in figure 4.9. Although the
observed number of rare species is greater than the expected value, it is not significantly

different (p = .32) from that expected by chance.

4.4 DISCUSSION

The principle aim of the research undertaken for this dissertation has been to explore
patterns of rarity from a phylogenetic viewpoint The discussion of this undertaking will
first focus on the basic findings presented in this chapter. Next, implications of these
findings will be discussed and will include some results presented in earlier chapters.
Potential limitations of this study will also be addressed. Finally, a summary, including a

few speculations about the nature of rarity in the Argophyllean clade, will be offered.

82

non—209.6 :. convene .0 .383: Each
we 09 Ne cw mm an em an an on ma um um ON 2. or cw NF 2.

D D D D D D D D D D D D D D D D D D P

 

on... u 8.2m
62; “.2398 $5352 on...

 

..Eoweisn 2: 5 meta—2.5m ea 2: he :93
.8.— mfiucam .53 25.5, 830% $98.— .ue «a... 3.3.3.»:—

o m V N c
n L - IIIIIO
I I

I
O
P

I
N
P

 

I I I
Q 0 Q
1- 1- v-

3. 2&3

40 .IaqiunN

saloads are:

83

0l96
0696

0LS6

0996
0896

0L96

m
m

 

«5:50.32 82. *0 cozaebflu 3:232 E30508 .255.

wwmwmummmnmumnmmmmmmmmmma
«wommuwommuwmmmuwmmmummmm

23. u 3:32.05
E 5 u o:_u> autumn... 20.05000 .0222

8:859 ace—E85 can 4.. 552—38 be 8.2
38. 9583.9 ougoweina 95 a: :33—
: «Sufi

0L66

    

 

c w
ON
an
ce
cm
cm
on
am
am
cc _.

84

Ir
6
9
9
0

0976 l

0986 l-

mcozmgao. 83 no cos-523:9 «05.80.. «230530 .0222

OSZGL
09l6l
0906l
0968L
0988i
OSLBL
0998L
OSSBL
0978i
0988i
OSZBL
05L8L
0908l

mu. u 3.332..
8mg u 2:: 32030
5205000 .2222

 

853%.. 25:32: 33 m «:35 92:5.»an
33. «5.3.3.9 o=o=cwe_.3a 2: Eat ...—mom
, a... 2:5

096i l-

 

    

c _.
cN
on
oe
on
em
on
am
cm

2:

85

23.30.32 can .. 20:2:sz 3:222 2.2. 35.52 measu «£02.» 2mm
3. 2 up 3 or o a b

  
 

o:_m> 32030

ANQHE ad we 2:: 6292—5 2: Eat 320.56 22.35:»? 8: am who.— .afiEh: 2: 95:3
832% “3.8.. = ..e :58 2; 623% .53 .2:an mac—:5 832W «8.5.. 2: ac hug—E: Z
3. 2&3

86

4.4.1 Basic findings

The clearest finding from the work presented in this chapter is that the Argophyllean clade
of Astragalus is significantly asymmetrical. Eleven of the 19 tested subclades were found
to be asymmetrical, clearly indicating diversification rate variation within the
Argophyllean clade. Estimates of the direction of change were not possible, however,
without data in which to estimate the timing of events in the phylogeny. Furthermore,
because diversification rate is the net result of the speciation rate minus the extinction rate,
inferences regarding changes in these two rates are also not possible (Sanderson and
Donoghue 1996).

Two scenarios might explain the diversification rate variation as illustrated within the

example below. Consider the following phylogeny:

|/--- High diversity (I!)
/ _____
\--- Low diversity (1.)

\ ---------- Outgroup (0)

With respect to the outgroup (O) lineage, the above pattern might be explained by:

SCENARIO 1: Diversification rate increase in clade H is due to either:
(a) an increase in the speciation rate in clade H or

(b) a decrease in the extinction rate in clade H.

SCENARIO 2: Diversification rate decrease in clade L is due to either:
(a) a decrease in the speciation rate in clade L or

(b) an increase in the extinction rate in clade L.

Of course, more than one of these rate changes may occur simultaneously.

87

 

The Argophyllean clade is obviously more complex than the above three-taxon
system. As such, it is reasonable to suppose that its diversification rates may have varied
in more complex ways as well. Evidence from the asymmetry tests supports this.
Furthermore, Wojciechowski et.al. (1993) speculate that the rate of diversification within
Neo-Astragalus (also called the aneuploid groups), a well supported monophyletic
assembly of nearly 500 species and the clade in which the Argophyllean clade is nested,
may also have been heterogeneous based on their results. They suggest that "Groups that
diverged early may have remained depauperate while those splitting off later radiated
rapidly generating the bulk of diversity now evident within the aneuploid groups."
Indeed, among the 93 sections that constitute what is now called Neo-Astragalus, one
section, Argophylli, accounts for 10% of the species (Bameby 1964).

Regarding the potential clustering of rare species in the phylogeny, the bivariate plot
of the number of rarest species versus the total number of species (figure 4.6) and results
from the phylogenetic clustering test suggest that rare species are not significantly
clustered within the phylogeny. A few qualifications should be noted, however. In the
case of the test based on area of occupancy A, the p-value of 0.89 is low, which may
indicate a tendency toward clustering in the phylogeny. In the absence of a power
analysis, it is not known how large a sample size is required to detect a significant
difference. Therefore, failure to reject the null hypothesis should not mean acceptance of
the null hypothesis.

Another issue affecting the interpretation of these results is that this test included all
taxa, not just the rarest. The clustering of some common or widespread species could
also lead to lower probabilities. Of course, if all the common and widespread species
were clustered, the rare ones would necessarily be clustered as well.

Other findings are consistent with the conclusion of "no significant clustering,"

namely from results presented in chapter 3. The locations of the rarest 17 species are not

88

 

clustered geographically, as can be seen in figure 3.10 (chapter 3). Indeed, the two
geographical pairs that overlap in their geographic extent E are not closely related species.
Astragalus callithrix and A. uncialis of Nevada are very distantly related, and A. loanus
and A. consobrinus of Utah are interrnediately related.

One small clade of five species, A. neomexicanus, A. giganteus, A. mollisimus,

A. helleri, and A. anisus, does appear to have a cluster of rarity (see figure 4.1). Three
of the five species are rare and one is somewhat rare. Astragalus mollisimus is
widespread. All five species are either regionally sympatn'c or are in close proximity to
one another. Having offered this observation, it is also unlikely that this clade has a
significant level of clustering or had much influence on the overall clustering test for the
whole group owing to the small sample size of five.

Another interesting finding is that Argophyllean rare species do not appear to be
disproportionately newly derived. They appear both ancestral and newly derived, based
on the sister group relationships. No direct estimation of time since divergence, from
findings presented here, was attempted because this requires: 1) estimates of absolute
time for at least one node in the tree, preferably at the base, 2) estimates of branch
lengths, and 3) estimates of evolutionary rates among branches.

In the case of the Argophyllean clade, the assumption of homogeneous evolutionary
rates throughout the tree would be contrary to the evidence (significant asymmetry).
Moreover, the use of branch lengths to assess evolutionary rates along the branches is
difficult to support with morphological data that are inherently homoplasious. For these
reasons, no assumptions were made with respect to evolutionary rate homogeneity among
branches.

An approximation of the relative timing of cladogenic events was assessed by
considering the size of sister lineages. Among the 17 rare species, 11 are considered

newly derived because their sister lineage size is one. They are thus members of terminal

89

pairs on the phylogeny. The ancestral rare species have sister lineage sizes of 46, 45, 31,
13, 4, and 4; the latter two are probably best considered somewhat newly or
intermediately derived. Although 65% of the rarest species are members of the 14
terminal pairs, this is not significantly more than would be expected by chance according
to the randomization test (p=.32). This is because 28 of the 47 terminal positions (60%)
are part of the terminal pairs.

Another important finding is that the Argophyllean clade of Astragalus has a
disproportionate number of rare species, at least when viewed from the perspective of the
genus as a whole. Among the 47 species in the lineage, 17 (36%) are rare by
internationally accepted criteria This percentage is substantially larger than Astragalus as
a whole ( 14%), its sister lineage Oxytropis + the Coluteoid clade (8%), and the Fabaceae
(17%). More importantly, however, is that Neo-Astragalus, the clade in which the
Argophyllean clade is nested, is not substantially different in its proportion of rare species
(30%). These conclusions are drawn from comparisons of current findings with results
and data from Sanderson and Wojciechowski (1996) and Walter and Gillett (1998) and
are presented in figure 4.10.

Sanderson and Wojciechowski (1996) also found evidence of a shift toward higher
diversification rates at the base of the Astragalus + Oxyrropis + the Coluteoid clade
(known collectively as the Astragalean clade) but did not find statistically significantly
higher rates for Astragalus compared to Oxytropis or the Coluteoid clade, both of which
are also quite diverse. They did not test for diversification rate variation within
Astragalus.

In a separate study, Wojciechowski, et. al. (1999) applied a simplistic molecular
clock model to nuclear ribosomal ITS sequence data and crudely estimated the age of

Astragalus at around 11 million years and Neo-Astragalus at around 4-5 million years.

90

Species richness in respective
monophyletic assembledge

2496

 

 

 

Diversification rate increase

13,100

Figure 4.10

Percentage of
rare species

Astragalus 14%

Nee-Astragalus 30%

Argophyllean 36%
Clade

Oxytropis 5%
Coluteoid 1 1%
Clade

FABACEAE 17%

Proportion of rare species within monophyletic groups related to the Argophyllean
clade. Diagram after Sanderson and Wojciechowski (1996). Rarity data from

Walter and Gillett (1998).

91

With these estimates, net diversification rates can be estimated using a standard

exponential model that assumes homogeneous rates:
N(t) = e" (4.1)

where N(t) represents standing diversity at time t, and r is net diversification (speciation

minus extinction). This equation can be rearranged to give:
1' = ln N/t (4.2)

Given the estimated age of Astragalus and Neo-Astragalus, the following rates are

obtained (W ojciechowski, et. al. 1999):

Astragalus 0.71 species/Myr
Neo-Astragalus 1.48 species/Myr

These finding suggest that Neo-Astragalus is diversifying at least twice as fast as
Astragalus as a whole and perhaps faster, given that Neo-Astragalus is nested within
Astragalus and thus adds a boost to the rate for the genus. Furthermore, this would put
an age for the Argophyllean clade at 2.5 million years. This latter estimate is also a rough
approximation especially in light of the evidence of heterogeneous evolutionary rates in

the clade.
4.4.2 Implications of findings

What are the implications of this work regarding the causes of rarity in the Argophyllean

Clade? First, there is no suggestion that genealogical traits are correlated with rare status.

92

Correlations are used to estimate the phylogeny, and any phylogenetic correlations with
rare status should result in the clustering of rare species in the phylogeny. Even if
clustering were apparent, assessment of a possible correlation would require testing with
an appropriate comparative method (Harvey and Page] 1991) to establish the significance
of the correlation.

The phylogeny was reconstructed using morphological traits and, to the extent that
they have tracked diversification events of the group, we would expect to see clustering
of rarity, if those traits were correlated with rare status. There is no suggestion of this.
Even if other genealogical traits, such as root biochemical or physiological traits, are
correlated with rare status, we should see clustering if the phylogenetic hypothesis is
robust. Given the high levels of homoplasy in the group (as well in Astragalus
generally), it is also possible that correlated traits with rare status are homoplasious and
therefore do not cluster.

Another implication of this work is that the genesis of rare species probably occurs
more commonly via peripatric speciation. This conclusion rests on two lines of evidence.
First, the Argophyllean clade is significantly asymmetrical. Recent investigations by
Chan and Moore (1999) on the effects of mode of speciation on tree symmetry led them
to conclude that asymmetry increases when significant time lags are built into modeling
the branching process. The rationale for time lags under the peripatric speciation mode
rests on the assumption that the likelihood of speciation increases with range size. Newly
derived species are less likely to speciate because they are inherently small in range size.
Time is required for their ranges to expand and lead to increased probability of speciation.

Second, many biogeographic patterns among species pairs with rare species are
consistent with a peripatric mode of speciation. Most of the recently derived rare species
are adjacent to or nested within the range of their sister species (data not shown). Among
the 11 rare species in terminal pairs, 7 are adjacent to their sister species. Two species,

A. funereus and A. subvestitus, are each other's sister species and are in close proximity

93

to each other, similar to the seven species. The rare A. uncialus of Nevada is somewhat
distant from its sister species, A. musiniensis, found in eastern Utah. The most extreme
disjunction is the rare species A. columbianus of Washington, whose sister species,

A. parryi, is found in Colorado and southeast Wyoming. Taken together, most of the
rare species are geographically located consistent with peripatric speciation.

Because peripatric speciation leads to the emergence of new species that are rare,
shifts toward higher speciation rates may lead to a proportional increase in the number of
rare species. Speciation rate variation may result from either intrinsic or extrinsic causes
or both. Intrinsic causes include genealogical traits that either forestall extinction and thus
provide greater opportunities for speciation to occur or result in higher absolute rates of
speciation. Speciation rates might also increase for extrinsic reasons, such as climate
change or radiation of lineages into regions with greater topographic complexity and/or
habitat heterogeneity (Cracraft 1986). Speciation rate variation can thus be lineage
specific or episode specific.

We know from geological and climate studies that North America has undergone great
climate changes during the last 100,000 years, plenty of time for new species to evolve
via peripatric speciation. Indeed, during the Pleistocene, large areas of the Great Basin—
the center of Argophyllean diversity—were inundated with inland lakes and its mountain
ranges were covered with glaciers (Tierney 1995). Many of these areas are now
inhabited by Argophyllean species. It remains to be seen whether new species occupy
areas that were unavailable in recent times. Nevertheless, climate change has been an
influential factor for the Argophyllean species. Although it is likely that significant
episodes of climate change can cause speciation rate variation across time, it is difficult to
assess the degree to which episodic speciation has occurred in the group because of the
problems of timing.

In spite of some evidence suggesting increased diversification as a possible

explanation for the rarity in this study group, the effect of extinction due to demographic

94

decline cannot be ignored. Indeed, extinction processes are likely to affect a number of
rare species in the group. Unfortunately, little is know about extinction events because
no fossil record exists for the group.

Among the four ancestral rare species in the group, as measured by the size of the
sister lineages, two species, A. loanus and A. desereticus are found in habitats common
for other Astragalus species and species of other genera. Given that they are rare in a
common habitat and are relatively old based on time since divergence, these two species
may be going extinct and may be the last descendants of a more widespread species or a
more diverse lineage from the past.

The two other species, A. lutosus and A. callithrix, are found in unusual habitats and
are not, therefore, likely to be rare because they are going extinct, although they are
probably very vulnerable to extinction because of their small range size. These two
species probably adapted to their unique habitats far in the past and have continued to
persist as rare species. They were probably never widespread and are less likely to have
been part of a more species rich lineage of the past that is now going extinct.

Finally, results presented in this chapter, taken with the work of others, strongly
suggest that elucidating the reason for the high proportion of rare species in the
Argophyllean clade is best studied from within the Neo—Astragalus clade. This simple
conclusion rests on the fact that the proportion of Neo-Astragalus species that are rare is
very similar to that of the Argophyllean clade and because of the evidence of a
diversification rate shift toward higher rates in Neo-Astragalus. If genealogical traits do
exist that are correlated with rare status, they may have evolved before the evolution of
the Argophyllean clade. Furthermore, many other questions will be easier to test
statistically with a larger study group. Indeed, if the whole Neo-Astragalus clade were

used, the sample size would increase approximately tenfold.

95

 

4.4.3 Summary

In summary, evidence for the hypothesis that high rates of diversification and local
speciation have led to high proportions of rare species in the Argophyllean clade is
equivocal. Significant clustering of rare species and a significant preponderance of newly
derived rare species were not found. However, these results are not sufficient to reject
the hypothesis because alternative explanations may account for these findings. The
existence of specific attributes leading to rarity in the Argophyllean clade also remains an
open question. Nevertheless, several conclusions can be drawn from the present work
and are offered below.

First, some lineages of the Argophyllean clade have experienced higher net rates of
diversification than other lineages in the group. Moreover, the clade is nested within two
larger assemblies, the Neo-Astragalus and Astragalean clades, which have experienced
rate shifts toward higher diversification (Sanderson and Wojciechowski 1996;
Wojciechowski et. al. 1999)

Second, peripatric speciation is probably the most common mode of speciation in the
group, given the highly asymmetrical topology of the phylogeny. Asymmetrical
phylogenies are consistent with peripatric speciation tree branching models. Also,
biogeographic evidence presented in chapter 3 is consistent with peripatric speciation.

Third, the Argophyllean clade clearly has a higher proportion of rare species
compared to Astragalus and the Fabaceae; however, the clade does not have a higher
proportion of rare species compared to Neo-Astragalus. Clade specific attributes of
Neo-Astragalus may account for the high proportion of rare species in the group. The
Argophyllean clade of 47 species (10% of Neo-Astragalus) is most likely too small to
uncover the presumed patterns in the phylogeny. At the phylogenetic scale of the

Argophyllean clade, the position of rare species is likely to appear random, which might

96

explain the lack of significant clustering of rarity in the group and that newly derived rare
species are not more frequent than expected by chance.

Forth, extinction is most probably contributing to rarity in concert with rapid local
diversification. The effect may be somewhat smaller in the Argophyllean clade compared
to Neo-Astragalus.

Finally, although it is evidently "not necessary to seek explanations for the
'exceptional' diversity of Astragalus" (Sanderson and Wojciechowski 1996),
explanations for the high proportion of rare species within the Neo-Astragalus and

Argophyllean clades remain a mystery and are worthy of further investigation.

97

APPENDICES

 

 

U m N N. _ N «van 333233:

u u S NNN 23 exegesis o N N me no: seen
u u 8 N_N 9:5 $2.3 6 N «N SN 3.: assets?
0 o as NNN :8 2.53:3 m N N «N as? 8332.9
O U wv mNN and: 336.2838 0 m NN EN .255 33352.5
6 o t. 8N as." 323%: o N _N mmN «use seesaw
u u 8. NNN ass. Ensues u N 8 we 83 3888s..
0 u 9. MN 53 63:38 N m 2 EN 23 338:3
6 u a. NE on». 33%»; N N 2 SN 5% 326%
u u a. NVN .38 223%.. N N : 3N noon 65$
0 0 Ne EN 33 35:33.. o N 2 SN :23 Sensesssa
o o 3. EN :3: asses: N N 2 NNN .53 sass
u u 9. :2 .33 883? m N 3 SN 38 .. 3:828
o u on EN 3.: 3%... N N 2 SN 33 325
U U m m NMN Hid." NGNQAS N N N — WON a.“ gtaumkwsbwt
u o 2 oz .8... gasses} N N _ _ 9N 8% 2.582%
u 6 cm _NN 5% 83383... N N 2 EN 32. 3583:...
u 0 mm 2N an...“ .328 N N a we :8 arenas
O U 3” _ _N once “.5588 N N w va .30: .262»:
u 0 mm :N 33 3588...... N N N MEN 5.3 3:5
o 6 NM SN :8 .38 N N 6 EN 33 goes:
0 m z a. 33 Eases N N m SN «23 eases.
u m om SN 33 stages _ _ v cNN 32. 23588
6 m N SN .3» seesaw: _ _ N :2 sun. .. 333»
O m N SN 9.30 32323 _ _ N SN 300 3:33:58
U m N CNN undo hfiaohwacfico _ _ _ m3 cnov gomamaoaow
has. 5.5. is. .252 2.8 .233 5.5. 5.52 :3. .25.... 88 :53»
M < ._< .3952— §~cu§u< m «N ..< .3055: 3395;

 

5:8 M was t Sons 25% c8 m.N.m .558 com .2< 5:838 Co was :65 .3 3532: mm 528 ..EB NEE—Eon 3:8
86on .mzflow L8 N Basso com Avg: 325m 605m mama bows; Set @838 366% 3.865 datum/w Avon: Sofiam :8:
89:5: @3255 dd 28 .w.m .m.m 838 E vow: .8600 SEE m was <23 15: 2v. ion—Es: 365mm .368 86on _.< 635.

.< “£52er

98

A=wsmne 3wz= .=Qnm.*= ~wmwvusm.mofimflamafim n www
A=Q£m= .ngwnu:vwEmszflmmxmz.uowwoumumw.>m H “mm
0cm
6cm
AAHZVmfimnewanmuflomumm.3wfi>w£u
wmam

Hflc Cucumu
Ag; .=coHqu m4 mufiwm w>mm wnu mm: wmmwam ..
+ mEMZuwo.E£BuNUm +
. ou muflom m>mm 0» wNnmcD=VONcH.xommmz
awn» Auoz.Aw>mmovvchuNummoum.Eneuflvm. MN
0cm
Ha: chauwn
cwnu Add: n w>mm0©v ma
Away»..chUNom
QOum=.gm:+mEm2umo.E£BuflUm+= ou muHUw m>mmgvamuCMUOme>.xommmz u wbmmov
cwnu ANN: Av Eneuflvwv ma
mEmSBmHQMuNUMum0.3wN>wsu n Eneuflom
mEmSu 3m: mcflummuo wHONwQ ufl mcfiufiwm QOum .fimuflwm ocflwn ma 3wfl> mnu Cw wEwnu m NH.

uoaw>fluu<uww.>m u 3wfi>m£u uuuuuuuu .

.UCH .wusuwumcH noumwmmm meumxm Hmucwficoufi>cm wo mucum mGNEEmHmOHQ msu >3 cmuUNHB 6cm wuflnz .3 >muwwmh
>3 ©w>flmucou mm3 Emumoum mane .wEmnu w>fluum may Hem mmonu mm wEMm wnu mum mafia: .xon monflU ms» Ga
menm>wu mum mvawflm uamcfl GESHOU can 30» m3» umnu muoz .wflnm wzu mo EOuuon cam umwa mnu mo coflumooH may

cam .AmHHwo QESHOU mo quESG way can mHHmU 30H «0 Hwnfidcv wuflm >munm .mNHm Hamo mcwummsvmn mummamm XOQ
monNU a .3mfl> m>fiuom wnu Ou wEwnu 0 mm Umvcm mum £0N£3 mcomxaom mumsvm mo Ufinm m mwumwno Emumonm mane

2:». 3% 38.5 un<.-E§wo.£ u::u>< 2.5th

m 32.3.?

99

 

.mawnmfi

A

 

cwnu Auoz.NN22mN.Nwazuv MN
coNuomflonmuwo.3mN>m£u n humwnu
Amfixowx s>®x .>®xo .>0®xovvvwme.som>HOQ n Q
U + >0 H >
U + X0 H X
NmQESCmm.Avvuwm.NwUNo..H CN n 30mm New
NmQESCmm.Amvuwm.NwUNO..N SN N nomm MOM

mCOHU.>o u Nxo
umnESCmm.ANvuwm.Nw©Ho >0
NwQEdem.ANvuwm.quNo xo
NwQESCmm.onuwm.um©Ho u o

A muNsmme

.gwwdam> uwucm= .=wusuoduum UNHw mumwuug VuSQCHNuH32.xommmz n vauo
m :OH: ~:OH: .:0: ‘..O: ~=M= v H WUHDMMNU

.mCESHOU ESE: .=w3ou Eda. .=EOuqu= .=uwwag .szNm Hamog v u mawan

AwsuuvwHQmquwumm.QMuMm£u
A,uN.VwNwNMncNM.ananM n oNMcNmnu
AgwmmnmgvanNchNM.nmuMmzu u vNanmmnu
Na» u amuMmsu

6cm

AwEwannuvmamneoo<.3wN>mnu
ANnuvwxmz.mswgem u mamnemnu
ANcNMMVchmNmov¢.Nnu
A mama VmNnNmN>uwm.vNM
No .m .432NumonoqumN .goNngxmz.vaNm u nNM
6cm
HNG Cudumu
vcw
Ag. .wEmzwwmmqu.wa + g mumwuu ou mHQMCD=VNONHm.XommmZ
mmaw
Ag. .mEMmemmuww.Mww
+ g wNNM MOM xuoq wufluz wNNSUum Ou manwcsgvuouum.xommmz
cwnu AHOHmeuoqmmm.Hnuv MN
:wnu Auouummmm.Nnuv MN
AnomNNom .vavzmzwxmz.nmem n Nnu
cmnu ANN: Av anv MN

100

AmmHvamagnuflomumm.nmuMw2u

0cm

0 + x0 u x0

N>o n >0
flaw

o + >0 u >0
AM My .Uauvflmnuvwzam>uwm.QMumwnu
AQ .H ‘Uauusmwfiuvwsam>uwm.nmuuwnu
cucumuvom.nwuwwnu n u
Ucw

AflnmwnuvwmuomwouQCDCHSuwm.Q I Q

101

Ahv>QHmO
AWVXDHmO
H 09 H u h mom
N mwfim Am I mNHm. 09 o u D mom
N mmEm Am I mNHmv 09 o u E mom
OOOH H BZDOUZHZ
meB UZH u mNHm mH
OH QMBm own 09 OH H mNHm mom
OH H UZH

AMNHm \ AD + Ahvwvvam
AmNHm \ A8 + Ahvxvvah

mOOQ
H + H u H
AHVW .AHvx .AvamHummm .H* BDmZH
AAHVmomv AHBZD CO
H u H
mAU

mt m¢ BDQBDO mom =BDO.mmHummm/0Hm<mO/NU: ZMQO
Ht m< BDQZH mom gBXB.mMH0mmm/0Hm¢mO/HU. ZEQO

AOOOHVZBmOm .AOOOHVXUWmU .AOOOHVQmOOU 2H0
AOOOHVHQHmO .AOOOHVNDHmO .AOOOHVN .AOOOHVN .AOOOHVWMHummm 2H0

.mcoHuowhoum memIHmsvw.

EOHM Uwumuwcwm cm£3 mmum ou HmcoHuHOQOHQ mum mucsoo HHmU.

.wcmHQ m CH mucHom hmHuw>o £UH£3 Uflum m CH UmHmsooo mHHmo.

mumsvm Mo uwQEdc EDEHCHE may wCHEHmqu ou UmeHmww mH Emumoum mHSB.

.HowHB UH>MQ EouM mocmMmHmmm mcHEmumoua nuHB manz .3 zwummmh hm.
unsou Hku ESEHGHZ.

=3..ch 395.0 2:50 :00 52552

D 51:25“.

102

02m
Na mmOdU
Ht mmOAU

mNHm me2
EH 02m

N * mNHm u UZH

BZDOUZHE .mNHm .Nw NBHMS

B BNWZ

D mez
mH 02m
92300 n BZDOUZHZ

ZMEB BZDOUZHE v BZDOU mH

h exmz
mH 02m
H + 92000 n 92:00
a exmz
mH 02m
H u Axvxummu
zame “cvnmooo u .xvomooo mH
H 09 H + n u x mom
zmme o u Amvxummo mH
H 09 H u n mom
0 n 92:00
n exmz
o n Anvxommu
thonmo + MoooHI I Anvonmwv u Anvamoou

H 09 H H h mom

w Exmz

103

LITERATURE CITED

LITERATURE CITED

Allen, T. F. H. and T. W. Hoekstra. 1992. Toward a Unified Ecology. New York,
Columbia University Press.

Archie, J. W. 1985. Methods for Coding Variable Morphological Features for Numerical
Taxonomic Analysis. Systematic Zoology 34:326-345.

Bameby, R. C. 1964. Atlas of North American Astragalus. Memoirs of the New York
Botanical Garden 13: 1-1 188.

Bameby, R. C. 1989. Fabales. New York, New York Botanical Garden.

Baum, B. R. 1988. A Simple Proceedure for Establishing Discrete Characters from
Measurement Data, applicable to Cladistics. Taxon 37:63-70.

Brooks, D. R. and D. A. McLennan. 1991. Phylogeny, Ecology, and Behavior: a
Research Program in Comparative Biology. Chicago, University of Chicago Press.

Cain, S. A. 1940. Some Observations on the Concept of Species Senescence. Ecology
21:213-215.

Cain, S. A. 1944. Foundations of Plant Geography. New York, Harper and Brothers.

Chan, K. M. A. and B. R. Moore. 1999. Accounting for Mode of Speciation Increases
Power and Realism of Tests of Phylogenetic Asymmetry. American Naturalist
153:332-246.

Cracraft, J. 1985. Biological Diversification and it's Causes. Annals of the Missouri
Botanical Garden 72:794-822.

Darwin, C. 1872. The Origin of Species. Toronto, Random House.
Drury, W. H. 1974. Rare Species. Biological Conservation 6: 162-169.
Drury, W. H. 1980. Rare Species of Plants. Rhodora 82:3-48.

Environmental Systems Research Institute. .1998. Arc View 615 3.x. Redlands,
California,

Favarger, C. and J. Contandriopoulus. 1961. Essai sur l'endemisme. Bulletin de la
Societe Botanique Suisse 384-408.

Femald, M. L. 1918. The Geographic Affinities of the Vascular Floras of New England,
the Maritime Provences, and Newfoundland. American Journal of Botany 52219-247.

Fiedler, P. L. 1986. Concepts of Rarity in Vascular Plant Species, with Special
Reference to the Genus Chalochortus Prush (Liliaceae). Taxon 35:502-518.

104

 

 

Fiedler, P. L. and J. J. Ahouse. 1992. Hierarchies of Cause: Toward an Understanding
of Rarity in Vascular Plant Species. Pages 23-47 in Conservation Biology: The
Thoery and Practice of Nature Conservation, Preservation, and Management (P. L.
Fiedler and S. K. Jain, ed.). New York, Chapman and Hall.

Gaston, K. J. 1994. Rarity. New York, Chapman & Hall.
Gleason, H. A. 1923. Review: Age and Area. Ecology 4:281-312.

Gleason, H. A. 1924. Age and area from the viewpoint of phytogeography. American
Journal Of Botany 41541-546.

Griggs, R. F. 1940. The Ecology of Rare Plants. Bulletin of the Torry Botanical Club
67:575-594.

Harvey, P. H. and M. D. Page]. 1991. The Comparative Methods in Evolutionary
Biology. Oxford, Oxford University Press.

 

Jablonski, D. 1987. Heritability at the Species Level: Analysis of Geographic Ranges of
Cretaceous Mollusks. Science 238:360-363.

Karron, J. D., Y. B. Linhart, C. A. Chaulk and C. A. Robertson. 1988. Genetic
structure of populations of geographically restricted and widespread species of
Astragalus (Fabaceae). American Journal of Botany 75:1114-11 19.

Kirkpatrick, M. and M. Slatkin. 1993. Searching for Evolutionary Patterns in the Shape
of a Phylogenetic Tree. Evolution 47:1171-1181. ‘

Kruckeberg, A. R. and D. Rabinowitz. 1985. Biological Aspects of Endemism on
Higher Plants. Annual Review of Ecology and Systematics 16:447-179.

Krummell, J. R., R. H. Gardner, G. Sugihara, R. V. O'Neill and P. R. Coleman. 1987.
Landscape patterns in a disturbed environment. Oikos 48:321-324.

Kunin, W. E. 1998. Extrapolating Species Abundance Across Spatial Scales. Science
281:1513-1515.

Kunin, W. E. and K. J. Gaston, Ed. 1997. The Biology of Rarity: Causes and
consequences of rare-common difierences. New York, Chapman & Hall.

Levin, D. A. 1993. Local Speciation in Plants: The Rule Not the Exception. Systematic
Botany 18:197-208.

Linder, H. P. 1995. Setting Conservation Priorities: The Importance of Endemisim and

Phylogeny in the Southern African Orchid Genus Herschelia. Conservation Biology
92585-595.

Liston, A. and J. A. Wheeler. 1994. The Phylogenetic Position of the Genus Astragalus
(Fabaceae): Evidence from the Chloroplast Genes rpoCI and rpoC2. Biochemical
Systematics and Ecology 22:377-388.

 

Lyell, C. 1830-33. Principles of Geology, 1 st edition. London, Murray.

 

105

Manly, B. 1997. Randomization, Bootstrap and Monte Carlo Methods in Biology, 2nd
Edition. New York, Chapman & Hall.

Maurer, B. A. 1999. Untangling Ecological Complexity. Chicago, University of Chicago
Press.

McKinney, M. L. and J. A. Drake, Ed. 1998. Biodiversity Dynamics. New York,
Columbia University Press.

Peitgen, H., H. Jurgens and D. Saupe. 1992. Chaos and Fractals: New Frontiers of
Scoence. New York, Springer-Verlag.

Rabinowitz, D. 1981. Seven Forms of Rarity. Pages 205-217 in The biological aspects
of rare plant conservation (H. Synge, ed.). New York, John Wiley & Sons.

Ridley, H. N. 1916. On Endemism and the Mutation Theory. Annals of Botany 30:551-
574.

Sanderson, M. J. 1991. Phylogenetic Relationships within North American Astragalus L.
(Fabaceae). Systematic Botany 16:414-430.

Sanderson, M. J. and M. J. Donoghue. 1996. Reconstructing Shifts in Diversification on
Phlogenetic Trees. Trends in Ecology and Evolution 11:15-20.

Sanderson, M. J. and M. J. Donoghue. 1996. The Relationship Between Homoplasy and
Confidence in a Phylogenetic Tree. Pages 67-89 in Homoplasy: The Recurrence of
Similarity in Evolution. (M. J. Sanderson and L. Hufford, ed.). New York,
Academic Press.

Sanderson, M. J. and J. J. Doyle. 1993. Phylogenetic Relationships in North American
Astragalus (Fabaceae) Based on Chloroplast DNA Restriction Site Variation.
Systematic Botany 18:395-408.

Sanderson, M. J. and A. Liston. 1995. Molecular Phylogenetic Systematics of Galegeae,
with Special Reference to Astragalus. Pages 331-350 in Advances in Legume
Systematics 7: Phylogeny (Crisp, M. and J. J. Doyle, ed.). Kew, Royal Botanic
Gardens.

Sanderson, M. J. and M. F. Wojciechowski. 1996. Diversification Rates in a Temperate
Legume Clade: Are There "So Many Species" of Astragalus (Fabaceae)? American
Journal of Botany 83: 1488-1502.

Schemske, D., B. Husband, M. Ruckelshaus, C. Goodwillie, 1. Parker and J. BishOp.
1994. Evaluating Approaches to the Conservation of Rare and Endangered Plants.
Ecology 75:584-606.

Simpson, G. G. 1953. The Major Features of Evolution. New York, Columbia
University Press.

Skinner, M. W. and B. M. Pavlik, Ed. 1994. California Native Plant Society's Inventory
of Rare and Endangered Vascular Plants of California. Sacramento, California Native
Plant Society.

106

Sokal, R. R. 1979. Testing Statistical Significance of Geogrphic Variation Patterns.
Systematic Zoology 28:227-232.

Stebbins, G. L. 1942. The Genetic Approach to Problems of Are and Endemic Plants.
Madrono 62241-272.

Stebbins, G. L. and J. Majors. 1965. Endemism and Speciation in the California Flora.
Ecological Monographs 31: 1-35.

Stevens, P. F. 1991. Character States, Morphological Variation, and Phylogenetic
Analysis: A Review. Systematic Biology 16:553-583.

Swofford, D. 1998. PA UP* 4.01)]: Phylogenetic Analysis Using Parsimony (and Other
Methods). Sunderland, MA, Sinauer Associates

Thiele, K. 1993. The Holy Grail of the Perfect Character: The Cladistic Treatment of
Morphometric Data. Cladistics 9:275-304.

Tierney, T. 1995. Geology of the Mono Basin. Lee Vining, California, Kutsavi Press.

Walter, K. S. and H. J. Gillett, Ed. 1998. 1997 I UCN Red List of Threatened Plants.
Gland, Switzerland, and Cambridge, UK., Compiled by the World Conservation
Monitoring Certre. IUCN - The World Conservation Union.

Welsh, S. 1978. Utah Flora: Fabaceae (Leguminosae). Great Basin Naturalist 38:225-
367.

Willis. 1922. Age and Area: A study of Geographical Distributions and Origin of
Species. Cambridge, Cambrodge University Press.

Willis, J. C. 1916. The evolution of species in Ceylon, with reference to [1: -.ying out of
species. Annals of Botany 30: 1-23.

Wojciechowski, M. F., M. J. Sanderson, B. G. Baldwin and M. J. Donoghue. 1993.
Monophyly of Aneuploid Astragalus (Fabaceae): Evidence from Nuclear Ribosomal
DNA Internal Transcribed Spacer Sequences. American Journal Of Botany 80:71 1-
722.

Wojciechowski, M. F., M. J. Sanderson and J .-M. Hu. 1999. Evidence on the
Monophyly of Astragalus (Fabaceae) and its Major Subgroups Based on Nuclear
Ribosomal DNA ITS and Chloroplast DNA trnL Intron Data. Systematic Botany
24:409-437.

Wright, S. 1956. Modes of Selection. American Naturalist 90:5-24.

Wulff, E. V. 1943. An Introduction to Historical Plant Geography. Waltham, MA,
Chronica Botanica.

107

MICHIGAN STRTE UNIV. LIBRQRIES
IIIIIWNIlllllllllllllIlllllHllllllllllllHI
31293020586461