|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
EVOLUTIONARY PATTERNS IN THE DILATATA GROUP
(Paspalum, POACEAE): A POLYPLOID/AGAMIC COMPLEX
PABLO R. SPERANZA
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Pablo R. Speranza
This dissertation is dedicated to my son Mauricio who never forgot to give me a hug
when I left home for work at night.
I want to thank my advisor, Pamela Soltis, for having always been available and
willing to help and provide support during my whole program. I also wish to thank my
committee members (Pamela Soltis, Douglas Soltis, Kenneth Quesenberry, and Gloria
Moore) for having made of each meeting a constructive and worthy experience, each
providing a different point of view on my work. I am indebted to Enrique Estramil and
Marcos Malosetti for their help with data analysis, and to my laboratory mates,
particularly Joshua Clayton, Vaughan Symonds, and Jennifer Tate for their invaluable
help in writing this dissertation. I also want to thank Gabriel Rua, Rodrigo Vergara,
Jennifer Tate, Michelle McMahon, and Kevin Smith for providing fundamental plant
material for this study.
I wish to thank my colleagues of the Universidad de la Republica, Uruguay,
particularly Magdalena Vaio, Cristina Mazzella, and Ivan Grela for facilitating plant
material and staying constantly in contact with me. This work would not have been
possible without the training and inspiration that I received from Juan Carlos Millot
during my undergraduate education.
Finally, this dissertation was possible due to the support I received from my wife,
This work was supported in part by a credit-scholarship from the Programa de
Desarrollo Tecnol6gico, Ministerio de Educaci6n y Cultura, Uruguay.
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ................................................................................................. iv
LIST OF TA BLE S ......... ................... ... .......................... viii
LIST OF FIGURES ......... ......................... ...... ........ ............ ix
ABSTRACT ........ .............. ............. ...... ...................... xi
1 AN INTRODUCTION TO THE STUDY OF THE DILATATA GROUP OF
P a sp a lu m ............................................................ ................ 1
A Historical Perspective ................. ... ...... ......... ................
The Contributions of This Dissertation ............................................. ............... 7
Chapter 2: The Technique .............................................................................. 7
Chapter 3: The Germ plasm ............................................................................8
Chapter 4: The Complex ....................................................... 8
Chapter 5: The All-Important Pentaploid(s)........................................................9
2 NUCLEAR AND CYTOPLASMIC MICROSATELLITE MARKERS FOR THE
SPECIES OF THE DILATATA GROUP OF Paspalum (POACEAE).....................10
In tro d u ctio n ............................................. ........................ ................ 1 0
M materials and M methods ....................................................................... .................. 13
M icrosatellite C apture ......................................... ........ .... ........ .. ........ .... 13
Chloroplast M arkers .................. ...................... .... .. .. .. ................. 14
P lan t M material ...............................................................15
A m plification and Scoring ............................................................................ 17
R results .................. ..... ....... ................................ ......18
Capture and A m plification Success.......................................... ..................... 18
V ariab ility ...................................................................................... ............. 2 4
C hloroplast V ariability ............................................... ............................. 25
D isc u ssio n ............................................................................................................. 2 6
Capture Efficiency ............................................... .. .. .... ................. 26
Amplification Profile............... ............................. ............... ............... 26
N on-F local L oci .................................................................................. 27
N u clear L oci .......................................................................2 8
C hloroplast M icrosatellites............................................................. .... ........... 30
C o n c lu sio n s........................................................................................................... 3 0
3 BREEDING SYSTEM AND POPULATION GENETIC STRUCTURE OF
Paspalum dilatatum ssp.flavescens (POACEAE). .................................................32
In tro d u ctio n ........................................................................................................... 3 2
M materials and M methods. ..................................................................... ...................34
C collection Strategy ......................................... ................. ........ 34
M icrosatellite A m plification ........................................ .......................... 35
D ata A n a ly sis ................................................................................................. 3 6
R e su lts ................................................................................................................... 3 7
D iscu ssion .............................................................................................................. 43
M eating System ............................. .... .......................... ........ 43
Genetic Differentiation and Geographical Structure ................... ....... 44
4 EVOLUTIONARY RELATIONSHIPS AND MECHANISMS IN THE
DILATATA GROUP (Paspalum, POACEAE) ........................................ ...49
In tro d u ctio n ........................................................................................................... 4 9
M materials an d M eth od s ......................................................................................... 53
P lant M material ................................................................................................ 53
DNA Extraction and Microsatellite Analysis.............................................54
D ata A n aly sis................................................ ....................... .... 5 5
R esu lts ................................................. ..................... ......56
V ariability in the Tetraploids............................................ 56
V ariability in the A pom icts ....................................................... 60
Relationships among Apomicts .................................................................... 61
H eterozygosity of the A pom icts .................................................................. 63
D iscu ssion .................. .. ................... ....... ....................... ......64
Evolutionary relationships among the sexual tetraploid biotypes .....................64
G genetic structure of the sexual tetraploids .............. .......................... .............. 66
V ariability w within the apom icts ........................................... ...................... 67
The addition of the X genome, apomixis, and the origin of pentaploid P.
dilatatum ............................... ................................................ 67
Paspalum dilatatum Uruguaiana and 59B................... ............ ............ 68
Paspalum dilatatum clone 2........................................ ..... ............... 70
Paspalum dilatatum Chiru ................................................70
P aspalum pauciciliatum ................................................... ............................ 71
Paspalum dilatatum Torres ................ .................. ...... ...............72
C on clu sion s......................................................................................... .72
5 PENTAPLOID X TETRAPLOID HYBRIDIZATION CYCLES IN Paspalum
dilatatum (POACEAE): EXPLAINING THE CURRENT AND FUTURE
EVOLUTIONARY SUCCESS OF AN IMBALANCED POLYPLOID ...................74
In tro d u ctio n ............................ ....................................... ................ 7 4
M materials and M methods ....................................................................... ..................76
P lant M material .......................... ..................................................7 6
DNA Extraction and Microsatellite Analysis..............................................77
D ata A n a ly sis ................................................................................................. 8 0
R e su lts ......................................................................................................................... 8 1
D isc u ssio n ............................................................................................................. 8 7
The "Typical" Clone.................. .............. ......... 87
Distribution of Clonal Diversity in the Native Range ................... ...................89
Hybridization and its Genetic Consequences...................................................90
T h e M ech anism .............................. .......................... .... ........ .... ..... ...... 9 1
H ow M any T im es? ............................ ........................ ......... ........... 92
6 CONCLUDING REMARKS ....................................................................... 94
REFERENCES ................... ......... .. ...... ... ..................97
B IO G R A PH ICA L SK ETCH ......... ................. ...................................... .....................106
LIST OF TABLES
1-1 Paspalum species known to share genomes with P. dilatatum Poir ........................
2-1 Accession numbers and collection locations of the plant material used to test
microsatellite transferability among biotypes .................................. ............... 16
2-2 Primer sequences and structure for all the microsatellite loci reported in this
stu d y ............................................................................. 1 9
2-3 Estimated recombination frequency between pairs of loci amplified by the same
primer combination and LOD score for the test of independent segregation
b etw een th em ........... .......... ................ ................. ................................ 2 3
2-4 Test of the segregation ratios per microsatellite locus. ............................................23
3-1 Genetic diversity and heterozygosity for individual populations ofP. dilatatum
ssp.flavescens for 6 microsatellite loci. ................................................ ......... 38
3-2 AMOVA of a six-microsatellite-locus data matrix for 21 populations of P.
dilatatum ssp. flavescens. .................................. ........................................40
4-1 Genomic formulae and reproductive systems of the members of the Dilatata
group ....................... .................................... ........... ...... .. 50
4-2 Accession numbers of the materials retrieved from germplasm banks used to
analyze the relationships among the different biotypes .......................................55
4-3 Summary of the microsatellite data for the sexual tetraploid biotypes of
Paspalum group Dilatata and genotypes for the apomictic biotypes.....................58
5-1 Accession numbers, genotypes, and population of origin of the pentaploid P.
dilatatum material retrieved from germplasm banks. ............................................78
5-2 Allele frequency distributions in the tetraploid biotypes of P. dilatatum used to
estimate possible contributions to the recombinant pentaploid clones ..................79
5-3 Proportion of shared allele distances (Dps) of the recombinant pentaploids of P.
dilatatum to the tetraploid biotypes of the Dilatata group. ................ ..............85
LIST OF FIGURES
1-1 Strict consensus of 15264 most parsimonious trees based on four cpDNA non-
coding squ en ces.............................................................................. ............... .. 4
2-1 A graphical representation of all the microsatellite alleles amplified for a sample
of each biotype. .......................................................................20
2-2 Alignment of nucleotide sequences of representative alleles for the non-focal
loci compared to the originally cloned sequences........................................21
3-1 A spikelet of P. dilatatum ssp.flavescens and P. dilatatum ssp. dilatatum ...........34
3-2 Allele size distribution for six microsatellite loci in P. dilatatum ssp.flavescens...37
3-3 Genetic distances among 21 populations of P. dilatatum ssp.flavescens and their
g geographical distribution ............................................................... .....................4 1
3-4 Genetic distances among individual genotypes of P. dilatatum ssp.flavescens
and their geographical distribution............. ................... ......... ..................... 42
4-1 Geographical distribution of the accessions ....... ..................................... ....54
4-2 Population structure of a sample of the members of the Dilatata group estimated
by Structure under the admixture model based on microsatellite data for 13 loci ...59
4-3 UPGMA phenograms of the distances among the sexual tetraploid biotypes of
the Dilatata group based on 13 microsatellite loci obtained with different
distance m measures ........................................... .......... .......... ........ .... 60
4-4 Multilocus genotypes of the apomictic components of the Dilatata group..............62
5-1 Number of microsatellite allele differences between all genotypes of pentaploid
P. dilatatum and the nearest of the two most widespread genotypes (A and P). .....82
5-2 Genotypic relationships and geographical distribution of the clonal variants of
pentaploid P. dilatatum ................................. .........................................84
5-3 Estimated tetraploid biotype contributions to the pentaploid recombinant P.
dilatatum genotypes ........................................ .. ... .. ...... 85
5-4 Multilocus microsatellite genotypes for the recombinant genotypes of pentaploid
P dilatatum .. ...................................................................... .... 86
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
EVOLUTIONARY PATTERNS IN THE DILATATA GROUP
(Paspalum, POACEAE): A POLYPLOID/AGAMIC COMPLEX
Pablo R. Speranza
Chair: Pamela S. Soltis
Major Department: Botany
Paspalum dilatatum Poir. and its related species are warm-season grasses native to
temperate South America. The members of the Dilatata group include polyploid sexual
and apomictic components, some of which have reached worldwide distributions. The
common biotype of P. dilatatum is a complex apomictic pentaploid hybrid, and efforts to
identify its putative ancestors have led to the accumulation of a significant amount of
cytogenetic information about the relationships among biotypes within the Dilatata
group. In general, past work in this complex has suffered from the lack of representative
collections, and the low power of the techniques. In this study, I developed microsatellite
markers, analyzed their transferability within the Dilatata group, and applied them to
representative samples to analyze the evolutionary relationships within the group at
different levels. The markers developed here show great power to detect recent
hybridization and analyze genetic structure. The genetic structure of the sexual biotypes
was described for a collection of P. dilatatum ssp.flavescens. This biotype is highly
autogamous, and its genetic variability does not show significant geographical structure
probably due to continuous disturbance of the roadside environment it inhabits. The
relationships among the sexual and the apomictic components are analyzed, and the
origin of the apomictic biotypes is discussed. Genetic recombination was not detected in
the apomictic hexaploids and tetraploids. Among the pentaploids, a single clone and its
somatic variants were found on all the continents and in almost all the collection sites in
its native area. All the other apomicts in the group, including the recombinant pentaploids
are hypothesized to be derived from the pentaploid form. The probable mechanisms
involve either the production of unreduced female gametes or eu-triploid pollen grains by
the pentaploids. This is probably the most extensive study ever attempted in this group,
and it will undoubtedly change the direction of all future research in these species. The
new recombinant forms will have to be analyzed and their productive potential assessed,
while existing collections should be re-structured to reflect the unexpected distribution of
the genetic variability they contain.
AN INTRODUCTION TO THE STUDY OF THE DILATATA GROUP OF Paspalum
A Historical Perspective
The genus Paspalum contains ca. 350 species, most of them native to the tropical
and warm temperate New World (Chase 1929). Chase (1929) recognized about 20
informal taxomomic groups within the genus based on vegetative and reproductive
morphological characters, a classification that is still widely used.
The Dilatata group of Paspalum contains several species with great forage
potential, and several of them have been used as forage crops (Skerman and Riveros
1992). Paspalum dilatatum Poir. and its related species are warm-season grasses native to
the grasslands of temperate South America. Some members of the group, particularly P.
dilatatum ssp. dilatatum and P. urvillei Steud., have reached worldwide distributions
wherever a warm temperate climate combined with sufficient rainfall exists.
The members of the Dilatata group have been classified into several formal and
informal taxonomic categories which will be referred to as biotypes in this dissertation.
The common biotype of P. dilatatum (P. dilatatum ssp. dilatatum) is a complex
apomictic pentaploid hybrid, and efforts to identify its putative ancestors have led, over
several decades, to the accumulation of a significant body of cytogenetic information
about the relationships among all the species and biotypes within the Dilatata group and
between this group and the related Virgata group (Table 1-1). The fist comprehensive
treatment of the Dilatata species was done by Moraes Fernandes et al. (1968) based
solely on the meiotic behavior of the biotypes. Burson (1983) summarized the results of
the advancements achieved during the 1970s by several interspecific hybridizations and
assigned the genomes in the group to putative diploid donors. Several new tetraploid
members of the group have been identified since then, but no significant advances have
been made about the relationships among them and the apomictic components.
Table 1-1. Paspalum species known to share genomes with P. dilatatum Poir
Species or biotype 2n Genomic Formula Authority
P. panmculatum 20 JJ Burson (1979)
P. jurgensu 20 JJ Burson (1978)
P. dilatatum ssp. dilatatum 50 IIJJX Burson (1983)
P. dilatatum ssp. flavescens 40 IIJJ Burson et al. (1973)
P. dasypleurum 40 IIJJ Quarin and Capponio (1995)
P. urvillet 40 IIJJ Burson (1979)
P. dilatatum Virasoro 40 IIJJ Caponio and Quarin (1990)
P. dilatatum Vacaria 40 IIJJ Quarin unpub. res.
P. dilatatum "Chiri" 60 IIJJXX Burson (1991)a
P. dilatatum Uruguaiana 60 IIJJX1X2 Burson (1995)
P. conspersum 40 1212JJ Burson (1978)
P. virgatum 40 IIJ2J2 Burson and Quarin (1982)
P. rufum 20 II Quarin and Norrmann (1990)
P. haumann 20 II Quarin and Norrmann (1990)
P. brunneum 20 II Quarin and Norrmann (1990)
P. quadrifarum 20 II Quarin and Norrmann (1990)
P. intermedium 20 II Burson (1978)
P. densum 20 II Caponio and Quarin (1993)
P. durifolum 60 IIJ2J2X*X* Burson (1985)
unknown genome not related to other X genomes
Several possible donors have been suggested for the I genome of the Dilatata
group, most of which constitute species commonly included in the Quadrifaria group as
defined by Barreto (1966). This group has been seen as including several species
typically based on self-incompatible sexual diploids and their apomictic autopolyploid
cytotypes. Recently, an analysis of the relationships among the proposed sources of the I
genome was prepared in collaboration with M. Vaio. This study addresses the
relationships among the several proposed donors of the I genome to the Dilatata group
using two main approaches: the distribution of rDNA sites in the genomes and
phylogenetic analysis of the chloroplast sequences.
Our results suggest that the relationships among the species of the group are
complex, with several polyploids of interspecific origin. Remarkably, the pairing ability
of the chromosomes among the I genome species (Quarin and Norrmann, 1990) bears a
correlation with the phylogenetic distances among them inferred from chloroplast
The analysis published in Vaio et al. (2005) is part of a phylogenetic analysis of the
genus as a whole that has been undertaken by G. H. Rua and myself to provide a
framework in which the origin of polyploid species complexes of Paspalum can be
analyzed. This study is not yet completed as we still lack cytogenetic information on
most of the samples included in the study. A summary of the current progress of this
phylogenetic effort is shown in Fig. 1-1, and an outline of the methodology is provided in
Fig. 1-2. Concerning the origin of the Dilatata group, the relationships obtained so far
suggest that the definitive identification of the genomic sources for the group is far from
being achieved: the currently proposed sources of I genomes form a paraphyletic
assemblage within which the proposed J genome donor (P. juergensiilP. paniculatum) is
also found. Moreover, P. rufum, whose chromosomes also show a moderate degree of
pairing to the I genome, is located in a different clade which includes, among others,
most members of the Notata and Plicatula groups. If these relationships are confirmed in
the future, the J genome would represent a derived genome nested in a group in which the
plesiomorphic condition is the ability to pair with the I genome. The current
identifications of the I and J genomes are the result of the knowledge that was available
to the researchers when findings were made. Without a general understanding of the
94 P95 ---- bicdum
SPf-lct-m P polyphyllum
65 82 P commune
P ruergens( (JJ) Paniculata
P parculatum (JJ) Paniculata
-61 P quarnn (II) Quadrifara
Dilatata group (IIJJ)
P quadrnfanum (II) Quadrifaria
65 conspersm (IIJJ) Vlrgata
9 P mtermedum (II) Quadrifana
S 89 P denticulatum
Sl-- P lIvdum
99 89 P virgatum (IIJJ) Virgata
93 P exaltatum (II) Quadnfana
P haumanr (II) Quadrfana
-- P ovale
3 P cromyorrhizon
57 P onanthum
62 P lepton
8 P lhmbatum
0 P pa hstre
a P alcahmn m
97 P rum (II) Vlrgata
99 I P bertonn
-- P mtinus
53 -------- P glabrinode
96 P distchum
100 P mconstans
P b57 P repens
Fig. 1-1. Strict consensus of 15264 most parsimonious trees based on four cpDNA non-
coding sequences (see Box 1). Numbers above branches represent boostrap
support values, and numbers below the branches represent percent posterior
probabilities for the same branches in a Bayesian (see Box 1) tree of similar
topology when they were not 100%.
_______ Anthaenantiopsis rojasiana
The phylogeny shown in Fig. 1-1 represents the current status of an ongoing project
aimed at establishing a framework for the study of the relationships among species
and polyploid complexes within the genus Paspalum. This project is being carried out
in collaboration with G. H. Rua. The genus itself has not been analyzed using
molecular tools, but some partial analyses show that the genus is not monophyletic if
the species of Thrasya are excluded, and its nearest relatives are A4iih, lcini,1 \i\,
Thrasyopsis, and some species currently assigned to Panicum (G6mez-Martinez and
Culham 2000, Giussani et al. 2001, Duvall et al. 2001, Aliscioni et al. 2003).
The inclusion of multiple polyploids and hybrids in a phylogenetic analysis would
confound the phylogenetic signal if we analyzed nuclear markers (including
morphology). As a consequence we were faced with the practical impossibility of
assembling a collection of only diploids or performing broad scale nuclear sequence
isolation and cloning of a big portion of the genus, while the ploidy levels of most of
our materials would be unknown. We decided to use a mostly living collection at the
Universidad de Buenos Aires, for which cytogenetic information could be obtained
and voucher specimens had been deposited. We undertook the phylogenetic effort
using chloroplast sequences. With this kind of marker, a congruent set of sequences
could be obtained, and our phylogenetic tree would represent the organismal history
of our diploid samples and the maternal progenitors of our polyploids. This approach
would circumvent the risk of insufficient taxon sampling by filling the gaps created by
unrepresented diploids with the chloroplast sequences of their derived polyploids.
Valuable information can also be obtained on the maternal origins of the polyploids.
Four non-coding chloroplast regions were amplified: the trnL(UAA) intron, the atpf,-
rbcL spacer, the trnG(UCC) intron and the trnL(UAA)-trnF(GAA) spacer (technical
details on amplification and sequencing are given in Vaio et al. 2005). A single
matrix was made of all the alignments for 72 species of Paspalum, one species of both
Thrasya and Aii/ihi,,iin,,ipi, and two species of Axonopus as outgroups. The
matrix was analyzed with parsimony using PAUP* (fully heuristic search with 20000
SAR) and Bayesian approaches using MrBayes (a model was selected with Modeltest
and the MCMC was run for 2,000,000 generations on MrBayes). The trees obtained
with both approaches, were fully congruent, and the tree obtained with MrBayes was
nearly identical to the majority-rule consensus obtained with parsimony. A slightly
more conservative parsimony strict consensus is shown in Fig. 1-1 which does not
show nine nodes that appeared in the Bayesian tree with posterior probabilities mostly
Fig 1-2. Main conclusions of a preliminary phylogeny of the genus Paspalum
In spite of the data being preliminary, a few well-supported hypotheses can be derived
from this tree concerning the placement and origin of the Dilatata group:
1. Paspalum is not monophyletic unless Thrasya is included in it and a)
Aiiihie iiuiniip \i\ is included or b) P. inaequivalve is excluded.
2. Paspalum racemosum, P. orbiculatum, P. repens, and P. conjugatum form a
basal grade to an otherwise poorly supported clade containing the rest of the
species of Paspalum and Thrasya.
3. Two major clades with high posterior probabilities but no parsimony bootstrap
support include most of the species analyzed.
4. The species of the Dilatata, Virgata, Paniculata, and Quadrifaria groups are
included in the same clade but in different subclades. The species of the
Quadrifaria group are found in both subclades.
5. The second major clade includes species of the Plicatula, Notata and Bertoniana
groups among others. This clade shows very little internal resolution. P. rufum
is included in this clade.
6. The chloroplast genome of the Dilatata group is included in a clade that includes
species that have been assigned both the I and J genomes.
7. The chloroplast genomes of the Virgata group are scattered within the clade that
contains most of the Quadrifaria species.
Fig. 1-2 (continued)
relationships among the main clades within the genus, the quest for the direct genome
donors of the polyploid groups (including the Dilatata group) is deemed to continue at
random, with occasional successes and many false positives.
The Contributions of This Dissertation
In general, past work in this complex has suffered from two main limitations: the
lack of representative collections, and the low power of the techniques. An effort was
made in this case to represent with more than one individual each component of the
complex. The main source of materials was a collection deposited by myself between
1992 and 1999 in the Germplasm Bank at the Facultad de Agronomia, Montevideo,
Uruguay, which was complemented with the USDA collection.
The main body of this dissertation is divided into four chapters dealing with
different levels of analysis of the Dilatata complex.
Chapter 2: The Technique
Two recent works (Speranza et al. 2003, Vaio et al. 2005) represent the first
examples of the use of cytogenetic information in this group beyond the sheer number of
chromosomes or their meitotic behavior; however, cytogenetics alone cannot answer
many of the questions that must be addressed in this group. It has been hypothesized
that, for example, the pentaploid biotype may be the direct derivative of Chiru and a
sexual tetraploid. Microsatellites, due to their co-dominant nature and high degree of
variability, appear as the ideal kind of markers to test such hypotheses. Furthermore,
other issues that need to be addressed, such as the breeding systems and genetic structure
of several biotypes, could also be effectively resolved using variable co-dominant
In Chapter 2, I will design and characterize several microsatellite loci for P.
dilatatum ssp.flavescens and test their transferability to all the other members of the
Dilatata group. I will also attempt to predict their potential to test different types of
hypotheses. These markers will be used as the main source of information in the
following three chapters.
Chapter 3: The Germplasm
In the context of the traditional hypothesis, P. dilatatum is the product of a
tetraploid (IIJJ) x hexaploid (IIJJXX) hybridization. The use of a variable, well-
characterized collection of tetraploids seems to be the most direct resynthesis route for
variable pentaploids. This collection could be hybridized to either of the two apomictic
and invariable hexaploids to produce new pentaploids. In spite of this, representative
collections oftetraploids have not been available. The collection of P. dilatatum ssp.
flavescens that I used here was the first one made with such an objective in mind, and a
molecular characterization of it will greatly increase its value. A comprehensive
populational-level study of the genetic variability and its structure will be presented in
Chapter 4: The Complex
Two previous attempts have been made to represent the Dilatata group in its
entirety. These attempts (Moraes Fernandes et al. 1968) and Burson (1983) were based
on the knowledge available at the time. In spite of the identification of new components
and new relationships mainly during the 1990s, the paucity of information, the lack of
comprehensive collections, or the limited interpretability of the markers used (see Casa et
al. 2002), no great advances have been made in the past two decades. With the power of
suitable techniques and a sufficient collection, Chapter 4 will surely become a landmark
in the understanding of the complex.
Chapter 5: The All-Important Pentaploid(s)
It was the interest in this biotype that initiated a long series of studies in the
Dilatata group. Since the establishment of its complex hybrid origin (Bashaw and Forbes
1958), the biotype has mostly been seen as a static, invariable entity. Chapter 5 is
probably the most extensive and intensive study of genetic variability ever attempted in
this biotype, and will undoubtedly change the direction of all future research on it
NUCLEAR AND CYTOPLASMIC MICROSATELLITE MARKERS FOR THE
SPECIES OF THE DILATATA GROUP OF Paspalum (POACEAE)
The Dilatata group of Paspalum is a polyploid complex native to the grasslands of
temperate South America. The complex contains several informal taxonomic entities that
will be generally referred to as biotypes in this paper. Paspalum dilatatum Poir. ssp.
dilatatum, a trihybrid pentaploid apomict (Bashaw and Forbes 1958, Bashaw and Holt
1958), is a widely recognized forage crop. This biotype has been assigned the IIJJX
genomic formula (Burson 1983). The Dilatata group also includes several sexual selfing
allotetraploids and several tetra- and hexaploid apomictic entities. The sexual
allotetraploids (P. urvillei Steud., P. dasypleurum Kunze ex Desv., P. dilatatum ssp.
flavescens Roseng. Arr. et Izag., and biotypes Virasoro and Vacaria of P. dilatatum) have
been shown either directly or indirectly to share the IIJJ genomic formula (Burson et al.
1973, Quarin and Caponio 1995, Burson 1979, Caponio and Quarin 1990), and their
interfertility has been either directly assessed by hybridization (Caponio and Quarin
1990, Quarin and Caponio 1995) or inferred from the occurrence of natural hybrids
(Valls and Pozzobon 1987). The group also includes two apomictic hexaploids (P.
dilatatum biotypes Uruguaiana and Chiru ) which have been assigned the IIJJXX2 and
IIJJXX genomic formulae, respectively (Burson 1991, 1992). Paspalum dilatatum
Torres, an asynaptic apomictic hexaploid (Moraes-Fernandes et al. 1968), and P.
pauciciliatum, an apomictic tetraploid (Bashaw and Forbes 1958) of unknown genomic
constitution, are also included in the group. Recently, Machado et al. (2005) have shown
that there may be several pentaploid apomictic entities in the group which have not yet
been described or named.
Interest in breeding the common pentaploid biotype has been the main motivation
for extensive interspecific hybridization and cytogenetic analyses in this group.
Pentaploid (IIJJX) resynthesis by hexaploid (IIJJXX) x tetraploid (IIJJ) hybridization has
been suggested as a possible breeding strategy for the group (Burson 1983). Vigorous
synthetic pentaploids have been successfully obtained by this kind of cross (Burson
1991b, 1992, Speranza 1994, unpubl. res.); however, the evaluation of available genetic
variability, particularly in the selfing tetraploids, has not been undertaken.
Sufficient knowledge has been accumulated not only to initiate the first breeding
attempts, but also to make this species complex an interesting model for the study of
apomixis and polyploidy. In spite of this, the study of the relationships among the
different entities that comprise this group has not advanced much in the last decade, most
probably due to the limitations of the tools available and the biological characteristics of
the organisms themselves. Only recently has it been possible to obtain further cytological
information on these biotypes through the modification of cytological techniques that
allow chromosome identification and karyotyping (Speranza et al. 2003, Vaio et al.
2005); however, cytogenetics alone cannot answer many of the questions that must be
addressed in this group.
As most of the apomicts in the complex have been hypothesized to be inter-
biotypic combinations, the markers required to address questions about their genetic
structure and origin must be transferable among all of the putative parents involved and
preferably co-dominant. On the other hand, assessment of parentage would be best
achieved with markers that are stable within biotypes, while the study of the genetic
structure of the sexual components of the complex requires high levels of variability.
Attempts have been made to use molecular markers for the study of these species,
but even when some degree of genetic differentiation between the biotypes and intra-
biotypic variability were confirmed (isozymes: Pereira et al. 2000, Chies unpublished
data, Hickenbick et al. 1992, AFLP: Speranza unpublished data, Casa et al. 2002), the
levels of variability have not been high enough or their interpretability has been very
limited. Allopolyploidy adds an additional level of complexity to the genetic
interpretation of molecular markers: the assessment of homology vs. homeology among
markers may not be straightforward, particularly in complex interbiotypic combinations.
Microsatellite markers, despite the greater technological investment required for their
development, provide the best tools for the study of the issues that have to be addressed:
genetic structure, relatedness among the entities of the complex, and parentage of the
apomictic components which are expected to be fixed hybrids. The generalized use of
simple sequence repeat (SSR) enrichment and PCR-based protocols has greatly facilitated
the development of new microsatellite loci (Fischer and Bachmann 1998, Kijas et al.
1994, Kandpal et al. 1994, JakSe and Javornik, 2001). Microsatellites not only provide
more powerful genetic data due to their co-dominant nature, they usually tend to be
extremely variable. Mutation rates of nearly 1x10-3 have been directly observed in maize
(Vigouroux et al. 2002). In well-studied selfing grass amphiploid systems like wheat,
microsatellites are capable of revealing great genetic variability (Roder et al. 1995) where
isozyme, RFLP, and AFLP markers show a high degree of marker conservation (Hazen et
al. 2002, Kim and Ward 2000).
Finally, determining the directionality of hybridizations within the species complex
may be crucial to understanding the mechanisms by which new genetic combinations are
being generated. Chloroplast microsatellites, typically (T/A)n, have been successfully
used to elucidate directional formation of allopolyploids in grasses (Ishii and McCouch
2000). Ishii and McCouch (2000) reported that despite successful cross-amplification, the
presence of variable (T/A)n tracts was not conserved among distantly related grass
genera; however, putatively universal primers for grass chloroplast microsatellites have
been reported in the literature (Provan et al. 2004).
Sixteen variable nuclear and one variable chloroplast microsatellite loci for P.
dilatatum ssp.flavescens were developed and characterized in this study. Their
transferability among all the taxonomic entities of the Dilatata group was assessed, and
their utility for addressing populational and phylogenetic studies at different levels is
Materials and Methods
A genomic DNA library consisting of Sau3AI fragments of P. dilatatum ssp.
flavescens was enriched for putative microsatellite-containing sequences following the
procedures of Ernst et al. (2004) with minor modifications. Briefly, genomic DNA was
extracted with Sigma GeneluteTM kit (Sigma-Aldrich, St. Louis, MO) and digested with
Sau3AI. Fragments smaller than approximately 400bp were removed by fractioning using
Chroma Spin columns (Clontech Laboratories). Sau3AI linkers were ligated to the
remaining fragments which were then amplified by PCR. The amplified fragment library
was enriched for (GT)n-containing sequences by binding to a Vectrex Avidin D matrix
(Vector Laboratories, Burlingame, CA) to which a biotinylated (CA)n oligonucleotide
probe had been previously bound. The eluted fragments were reamplified by PCR using
primers for the Sau3AI linkers, ligated into pCR II-TOPO plasmids (Invitrogen, Inc)
and transformed into ONE shot E. coli competent cells. Colonies were screened by
binding them to Magnacharge nylon transfer membranes (Osmonics, Inc.). The
membranes were probed with labeled (CA)n and positive colonies detected with Lumi-
Phos 480 (Lifecodes, Inc.). All probe labeling, hybridization, and detection was carried
out with Quick-Light TM system (Lifecodes, Stamford, CT). Positive colonies were
grown overnight in a liquid medium and plasmids purified with QIAprep Spin Miniprep
Kit (Qiagen, Inc.). Plasmid isolates were screened a second time by dot-blotting serial
dilutions on nylon membranes and hybridizing to a (CA)n probe. Isolates showing
consistent hybridization signal through the dilutions were sequenced and used for primer
Plasmid isolates were sequenced on a CEQ 8000 capillary sequencer (Beckman-
Coulter, Fullerton, CA) using /4 reaction volumes with the addition of 80mM Tris and 2
mM MgC12 (pH 9) to complete the volume of a full reaction. The sequences were edited
manually using SequencherTM (V4.1.4, Genecodes, Ann Arbor, MI).
Primers were designed for sequences containing repeats longer than (GT)io with
Primer 3 (Rozen and Skaletsky, 2000). Low-complexity regions were excluded for
primer design when possible.
Sequences for 6 chloroplast non-coding regions were obtained for all the
recognized entities in the Dilatata group (Table 2-1). Six regions were analyzed: the
trnT(UGU)-trnL(UAA) spacer, the trnL(UAA) intron, the PsbA-trnH spacer, the atpB-
rbcL spacer, the trnG(UCC) intron, and the trnL(UAA)-trnF(GAA) spacer. PCR and
sequencing conditions and primers were reported in Vaio et al. (2005) except for the
trnT(UGU)-trnL(UAA) spacer which was amplified and sequenced using primers A and
B (Taberlet 1991). Primers were designed flanking two poly-A repeats located in the
trnT(UGU)-trnL(UAA) spacer. All primers reported by Provan et al. (2004) were also
tested. A second poly A-tract not reported by Provan et al. (2004) was detected near the
trnL(UAA)3' exon but no length variability was observed among the available sequences,
and no further analysis was performed on it. Primer design, labeling, amplification, and
detection procedures were performed as for the nuclear SSR described above.
Potential variability of microsatellites was assessed by analyzing a total of 28
accessions representing different species of the Dilatata group. To assess intraspecific
variability, we analyzed ten accessions each of P. dilatatum ssp.flavescens and P.
urvillei. Accessions were chosen to represent as much of the native range of the species
as possible. To assess transferability, two accessions each of P. dasypleurum, P.
dilatatum ssp. dilatatum, and biotypes Virasoro and Vacaria of P. dilatatum were also
analyzed. Seed samples were obtained from the Southern Regional Plant Introduction
station, Griffin, GA, USA and the Germplasm Bank at the Facultad de Agronomia,
Montevideo, Uruguay. Collection localities and accession number information for all
materials are shown in Table 2-1.
DNA was also extracted from 43 F2 individuals derived from a hybrid between P.
dilatatum ssp. flavescens and P. dilatatum Virasoro to analyze segregation patterns of
loci and possible linkage relationships. Deviation from expected segregation ratios and
linkage between loci were assessed using Joinmap 3.0 (Van Ooijen and Voorrips, 2001).
When there were indications that more than one locus had been amplified by a primer
pair (see below), the loci were considered as putative homeologs and the absence of
linkage between them regarded as a test for their homeology.
Table 2-1. Accession numbers and collection locations of the plant material used to test
microsatellite transferability among biotypes
Species or blotype Individual Accession Location
P dilatatum ssp
2n=4x=40 (IIJJ) 2
P urvillet 1
2n=4x=40 (IIJJ) 2
P dasypleurum** 1
2n=4x=40 (IIJJ) 2
P dilatatum ssp
2n=550 (IIJJX) 2
2n= 5x50 (IIJJX) 2
N A 7355
N A 7363
N A 7434
N A 7439
N A 7468
N A 7470
N A 7476
N A 7492
N A 2957
N A 7392
N A 7389
N A 7390
Pila, Buenos Aires, Argentina
Route 41, 1 0 km W of General Belgrano, Buenos Aires, Argentina
Route 3 near Tnnidad, Flonda, Uruguay
Route 8 km 34 3, Canelones, Uruguay
Riachuelo, Coloma, Uruguay
Route 12, 600 m N of Route 9, Maldonado, Uruguay
Route 56 km 51 500, Honda, Uruguay
Route 11 km 65 600, San Jose, Uruguay
Route 9 km 227, Rocha, Uruguay
Route 6 km 189 600, Honda, Uruguay
Ivoti, Cascata de Sao Miguel, about 45 km N of Porto Alegre, Rio Grande do Sul, Brazil
Route BR 116, 6 kmn N of Pelotas River, Santa Catarma, Brazil
Route 8, 162 km W of Buenos Aires, Buenos Aires, Argentina
Villa Nueva, 3 km S of Villa Mana, Cordoba, Argentina
Floranopois, Santa Catarma, Brazil
Route 26, km 25, Paysandu, Uruguay
Road to Las Cumbres de La Ballena, Maldonado, Uruguay
Route 7, km 103, Florida, Uruguay
Route 29 near Mmas de Corrales, Rivera, Uruguay
Balneario Solis, Maldonado, Uruguay
N A 7207 Gobemador Virasoro, Comentes, Argentina
Garruchos, Comentes, Argentina Voucher BAA24352
PI 404370 Near Vacaria, 192 km on Route BR 116, N of Porto Alegre, Rio Grande do Sul, Brazil
PI 404382 On Route BR 285 10 km west of Vacaria Rio Grande do Sul Brazil
Botancal Garden, Valdivia, Chile
General Lagos, Valdivia, Chile
N A 7542 Quebrada de los Cuervos, Treinta y Tres, Uruguay
N A 7673 Masoller, Rivera, Uruguay
*Accession numbers preceded by PI correspond to the Southern Plant Introduction Station, Griffin, GA, USA. Numbers preceded by
N.A. correspond to the Germplasm Bank at the Facultad de Agronomia, Montevideo, Uruguay.
** Seeds ofP. dasypleurum were kindly provided by Ing. For. Rodrigo Vergara from the Universidad Austral de Chile and the
University of Florida.
FI hybrids were obtained by manually emasculating a plant of P. dilatatum ssp.
flavescens and pollinating it with pollen from an individual of P. dilatatum Virasoro.
Emasculation and pollination were carried out about one hour after sunrise. The plants to
be emasculated were placed at approximately 20C and 100% RH to delay anther
dehiscense after anthesis was initiated each morning. Mature full seeds were counted and
germinated in Petri dishes on filter paper in an incubator with alternating temperatures
(16 h at 30C light, 8 h at 20C dark). Germinators were placed at 40C for 4 days prior to
incubation to break dormancy and homogenize germination. The resulting progeny were
grown and the hybrids were identified by the high number of nerves in glume II and
lemma I which characterize the pollen donor. Selfed seed of one F1 hybrid was collected
and treated as described above to establish the segregating F2 progeny used in this study.
Amplification and Scoring
For all plant materials DNA was extracted from fresh leaves or silica-gel-dried
leaves using Sigma GeneluteTM kit (Sigma-Aldrich, St. Louis, MO) according to the
Amplification, labeling, and separation conditions were adjusted for all primer
pairs following Boutin-Garnache (2001). Forward primers were extended by adding an
M13 tail (5'-CACGACGTTGTAAAAC-3'), and M13 primers were labeled with D4
(Beckman Coulter, Fullerton, CA). All PCR amplifications were carried out in 10 [tL
reactions containing 0.2 units of NEB Taq polymerase (New England Biolabs, Beverly,
MA), 1.5 mM MgCl2, 0.15 [tM of the reverse primer and labeled M13 primer, 0.01 [tM of
the extended forward primer, and 0.1 mM of each dNTP in the manufacturer's buffer.
Amplification was carried out in a Biometra T3 Thermoblock with the same two-step
program for all primer pairs. The PCR profile consisted of an initial denaturing step of 5
min at 940C, followed by 40 cycles of 15 sec at 940C and 3 min at 530C, and a final
extension step of 5 min at 720C. Labeled microsatellite products were separated in a
CEQ 8000 capillary sequencer (Beckman-Coulter, Fullerton, CA) by loading 0.75 pL of
the PCR product and 0.35 pL of CEQTM DNA Size Standard Kit-400 (Beckman
Coulter, Fullerton, CA) in 25[tL of formamide per well. Chromatograms were visualized
on CEQ TM Genetic Analysis system software (Beckman Coulter, Fullerton, CA) and
Alleles in different size ranges from different biotypes or subspecies were
sequenced to assess homology. For primer pairs amplifying a single locus, alleles were
amplified and sequenced directly from the PCR products of homozygous individuals.
For primer pairs amplifying more than one putative locus, PCR products were separated
in 2% agarose, and the bands were cut, purified with Wizard SV Gel and PCR Clean-up
System (Promega, Maddison, WI), and sequenced directly. When multiple bands could
not be separated in the gel, gel slices containing several bands were cut and combined
products cloned with a TOPO TA Cloning kit.
Capture and Amplification Success
A total of 24 clones containing (GT)n repeats were captured and sequenced. Four
clones were redundant. Primer pairs were designed for all clones, and all loci were
amplified with the same two-step PCR profile detailed above. Fifteen of these primer
pairs successfully amplified interpretable band patterns, and twelve of them were selected
for further analysis based on a preliminary assessment of amplification success under the
given conditions. Primer pairs for loci Pdfll, Pdfl2, and Pdfl26 are reported but not
further characterized (Table 2-2). Primer sequences, flanking sequence lengths, and
repeat units for all successfully amplified loci are shown in Table 2-2.
Table 2-2. Primer sequences and structure for all the microsatellite loci reported in this
study. When sequence information is available for more than one allele, the
variable repeat motif is reported. GenBank accession numbers for the
originally cloned sequence used to design each primer pair are given in
Primer sequences (5'-3')
(TA)2 AT (GT)6(GA)s
Primer pairs Pdfl6, Pdfll2, Pdfl 15, Pdfl20, and Pdfl22 amplified more than one
band in all tetraploid individuals, most of which appeared homozygous for all the other
loci (Fig lb, g, h, j, and k). In these cases, representative bands from both size ranges
were cloned, sequenced, and compared with the originally captured sequences (Fig. 2-2).
Two of the extra bands, a Pdfll2 190-bp band in P. dilatatum ssp.flavescens and a 140-
g) Pdfl12a Pdfl12b
* 2(6bp)- -- 4(12bpr- --
b) Pdfl6a Pdfl6b
h) Pdfl15a Pdfl15b -
i) Pdf118 -
j) Pdfl20a Pdfl20b Pdfl20c
- -- 245
-------------------- --, 1
Sk) Pdf122a Pdfl22b
I) Pdf128a Pdf128b
1 2 345 6 78 9 10 1 2 34 5 6 7 8 9 10 1 2 1 2 1 212 1 2 1 2 3 4 5 6 7 8 9 101 2 34 5 6 7 8 9 10 1 2 12 1 2 1 2
P dilatatum ssp flaM ecens P urwvle, P dlatatum p flasen P urwlle
Fig. 2-1. A graphical representation of all the microsatellite alleles amplified for a sample
of each biotype. Fragment lengths are plotted as scored including the M13
primer extension (19bp). Number of alleles within biotype are given for P.
dilatatum ssp. dilatatum and P. urvillei for those loci that showed variability
within biotypes. The corresponding maximum fragment size difference is
shown in parenthesis
GGT-TTGATACTAATTGTATATAAGTCGATTGCTTA-TTTTGCGTTA-ACTATTGCTGTGTCATTTTGGCGCGCTGTTTCTTTCCCTTCTTCATTTCGAGGTAGCT (GT) G
... ............ A ...G ..T ..T ........ ......... G .... C ...... ... .... A ............ T ...................... (G T )
............. A...G..T..T......... ........ G ..... .. ... .. A ......... T ........... ..... GT)
... ......... ..... ..G..T..T ........ .... .... ... ... ...... .. ..... ... ........ T .................T. GT
. ............A...G..T..T......... T ..........C.G.......... CAG.. ..----- -----..... CT GTGCG
S. ......... G .................... ... .... A ...
] ............. G ............................. A ...
S.. ............. ............. A...
1 . ..... ......... .................... A ...
AAGG (TA) (TG) 20--ATTCTTTTGATATGCTCCCTTGTAT
(TA G....... T ......... C .....
TA ....... T ......... C ..... (
.... TA ) T G ................... C ....
Pdfl22a Vir 1
Pdfl22b Vir 1
CAAGCCT (CT) 2 (GT) 3CGCGCGTGCACATGCATGGCGGCATGGCGCGCTTTGGCCGTTGGCCGATGCTTGCTGACTGAATCACTGAATGCTGGAGCTGAGCATGCA
.......( T) (GT .................................................... ...C ........ .. .
....... I(T)4 (GT) ............ .... ................................... A ............ T.C .........
. . .. . . . . . .. . . . . . .
...... ... ................. ........................T... I
S.... ................. .......... ... ........I..
... .. .. ... .. .. ...T-....A..A..... T.. =
Fig. 2-2. Alignment of nucleotide sequences of representative alleles for the non-focal loci compared to the originally cloned
sequences. Primer sequences are not included. All putatively homologous variable and non-variable repetitive sequences
near or around the main microsatellite are highlighted for comparison
a) Pdfl 6
I . .. .
. ..... .. ....... .
. ...... .C. ......... I
...... .c. ....... 7
.GT .............. 2P
.................. .C ............ T ....... ... A ..AC .
.................. .C ............ T ....... ... A ..A .
.................. .C ............ T ....... ... A ..A .
. .. .. .. .. .. .. .. T ... ... ... .. .
- (GT) 2GC (GT) 43
C (TA) 4 (GT)
c (TA) 4 (GT),5 C
-AGTTGCCACCT (TC) 6 (T) 3 GCTAAATGTATACTTTCCTTAA-CGCTGCTGAATATTTGTTAAGAACTAAACAATCTGCCGATATTATG
rT ........... [(TC )2 C T ) G (T ) G (T ) .... .. .. .. .. ..................... .... .. ...... T .... .....
r ........... (TC )2 C (T )2G (T )i .... G .... C ..G ......... -GT ................... .A ............ T .........
...... T... (TC)4 (T)3 ........ GC ............ A.T ........ ---- --- --- A ....... .............
G. ........ .......... ...CG.T... .G.. ........ ..... ...... .. ....
3G ......... T ............. CG.T .... G.T ....... G.CT --- --- .G .... C ....
bp band in P. urvillei did not contain the forward primer sequence. These bands were in
fact reproduced by using just the labeled M13 primer and the reverse primer for each
locus. These bands were not further scored and are not shown in Fig. 2-2. In all the other
cases, the extra bands showed extensive sequence similarity to the captured alleles (Fig.
2-2). The loci for which the primers were originally designed will be referred to as the
focal loci. When two different loci were detected, the focal locus was identified by
adding "a" to the primer pair name. For most non-focal loci, fragment sizes were smaller
than the captured bands, and the microsatellite repeat was either absent or showed a
lower number of repeats. The only exception was the Pdfll2a 140-bp allele in Vacaria
which was smaller than the non-focal locus Pdfll2b. Representative allele sequences are
shown in Fig. 2-2. For Pdfl20 and Pdfl28, an extra band was present in the apomictic
pentaploid relative to tetraploids (Fig. 2-2j and 1). These bands were sequenced, and they
showed again extensive similarity to the captured loci (Fig. 2-2).
When more than one locus was amplified by a primer pair, alleles were assigned to
either putative homeologous locus based on the sequences shown in Fig. 2-2.
Homeologous loci are expected to be completely unlinked because they necessarily lie on
different chromosomes. Linkage analysis based on the segregating F2 population showed
no evidence of linkage between loci amplified by the same primer pair with
recombination frequencies ranging from 0.3664 to 0.4371 and LOD scores from 0.03 to
1.09 (significance threshold LOD equal to 3) (Table 2-3).
In addition, no significant linkage was observed when considering the full set of 16
loci (including non-focal loci), even at a low LOD score threshold level of 2. Although
the test for independence between loci as implemented in Joinmap is robust against
segregation distortions, it is reassuring that all but one locus (PDfl 15a) showed
segregation ratios in agreement with the expectations (Table 2-4).
Most loci were amplified in all materials tested with a few exceptions (Fig. 2-1).
Paspalum urvillei could not be scored for Pdfl7 and Pdfll0. Pdfl7 was not amplified at all
in most P. urvillei samples, whereas for Pdfll0, weak peaks were observed but could not
be reliably scored probably due to interference during PCR from the M13-primed
fragment mentioned above. Paspalum dilatatum Virasoro showed null alleles for Pdfl 18
and Pdfll2a, and P. dasypleurum did not produce fragments for Pdfl 15. Locus Pdfl22b
amplified consistently only in P. dilatatum Virasoro and P. dilatatum ssp. dilatatum.
Table 2-3. Estimated recombination frequency between pairs of loci amplified by the
same primer combination, and LOD score for the test of independent
segregation between them. Significant threshold LOD=3
Locus pair Recombination LOD
Pdfll2a vs Pdfll2b 0.4326 1.09
Pdfll5a vs Pdfll5b 0.3664 0.38
Pdfl20a vs Pdfl20b 0.3665 0.21
Pdfl22a vs Pdfl22b 0.4371 0.03
Table 2-4. Test of the segregation ratios per microsatellite locus. Genotypes coded as: 'a'
homozygousflavescens allele, 'b' homozygous Virasoro allele, 'h'
heterozygous, 'c' dominant allele from Virasoro, and 'd' dominant from ssp.
.Locus a h b c d ratio chi-square df p
Pdfl4 13 22 8 0 0 1:2:1 1.2 2 0.549
Pdfl6 14 18 11 0 0 1:2:1 1.6 2 0.449
Pdfl7 16 17 10 0 0 1:2:1 3.6 2 0.165
Pdfl8 9 22 12 0 0 1:2:1 0.4 2 0.819
PdfllO 17 20 6 0 0 1:2:1 5.8 2 0.055
Pdflll 6 21 16 0 0 1:2:1 4.7 2 0.095
Pdfll2a 0 0 15 0 28 3:1 2.2 1 0.138
Pdfll2b 7 28 8 0 0 1:2:1 4.0 2 0.135
Pdfll5a 5 21 17 0 0 1:2:1 6.7 2 0.035
Pdfll5b 10 20 13 0 0 1:2:1 0.6 2 0.741
Pdfll8 0 0 10 0 33 3:1 0.1 1 0.752
Pdfl20a 8 25 10 0 0 1:2:1 1.3 2 0.522
Pdfl20b 14 20 9 0 0 1:2:1 1.4 2 0.497
Pdfl22a 7 25 11 0 0 1:2:1 1.9 2 0.387
Pdfl22b 8 0 0 35 0 3:1 0.9 1 0.343
Pcfl28 12 19 12 0 0 1:2:1 0.6 2 0.741
All biotypes showed polymorphisms for at least two loci. On the other hand,
several alleles were mostly fixed within biotypes but variable among biotypes. Notably
Pdfl7 provided excellent biotype-specific markers for this sample (Fig. 2-1c). Pdfl6b was
the only invariable locus scored in this sample.
At least two different alleles were cloned and partially sequenced from the focal
biotype for Pdfl4, Pdfl6a, Pdfll 1, Pdfll2a, Pdfl22a, Pdfll5a, and Pdfl28a. Despite the
effect of the long repetitive tract on the quality of the sequence, it can be clearly seen that
variation in the focal loci is due to expansion/contraction of the (GT/CA)n repeat (data
Despite being the source biotype, P. dilatatum ssp.flavescens does not display the
greatest variability nor allele sizes in this sample (Fig 2-1). The average number of
alleles for P. dilatatum ssp.flavescens is 2.5 with an average size range per locus of 9.9
bp. Paspalum urvillei showed a higher number of alleles per locus (3.9) and a higher size
range (28.9 bp); furthermore, P. urvillei showed polymorphisms for all the scored loci,
and all the individuals except 3, 7, and 8 were heterozygous for at least one locus whereas
in P. dilatatum ssp.flavescens only individual 7 appeared heterozygous for locus Pdfl6a.
The two individuals sampled from Virasoro and Vacaria were completely homozygous,
whereas heterozygosity was observed in both individuals of P. dasypleurum (Fig. 2-le, j,
h). The two individuals ofP. dilatatum ssp. dilatatum were heterozygous for all loci
except individual 2 at locus Pdfl22b (Fig. 2-1k).
In summary, 19 variable nuclear microsatellite loci were investigated, with one
locus monomorphic in all the accessions (Pdfl6b). Among the 18 variable loci, 11 were
successfully amplified and interpreted in all the biotypes in this sample. Of the original
12 focal loci, 4 could not be scored in one of the biotypes: Pdfl7 and PdfllO in P. urvillei,
Pdfll8 in Virasoro, and Pdfll5a in P. dasypleurum. Three of the non-focal variable loci
(Pdfll2b, Pdfll5b, and Pdfl20b) were successfully amplified and scored for all
individuals, while Pdfl22b was only amplified in Virasoro, the pentaploids, and one
individual ofP. urvillei, and finally, two loci (Pdfl20-3 and Pdfl28b) were only amplified
in the pentaploids.
All regions were successfully amplified and sequenced except for the trnT(UGU)-
trnL(UAA) spacer for which low-quality sequences were obtained due to the presence of
poly-A tracts near both ends (GenBank accession nos. DQ104273-DQ104323). No
further efforts were made to improve the quality of the sequences of the (UGU)-
trnL(UAA) spacer because they were considered appropriate for the purposes of this
study. Overall, no sequence variability was found among the biotypes of the Dilatata
group except for the length of the poly-A tract in the trnT(UGU)-trnL(UAA) spacer, and
a G-T transversion in the trnL(UAA) intron in the Vacaria individual. No repetitive
sequences were found except for poly-A/T tracts. Fragment sizes were not variable for
any of the loci reported by Provan et al. (2004) or the poly-A repeat located near the
trnL(UAA)5' exon in the trnT(UGU)-trnL(UAA) spacer. Only the poly-A repeat located
near trnT(UGU) (cpDilB) was variable as observed in the original sequences. Fragment
lengths (after subtracting the M13 tail) were 198 bp for P. dilatatum ssp.flavescens, ssp.
dilatatum, Virasoro, and Vacaria, 199 bp for P. urvillei and 197 bp for P. dasypleurum.
No intrabiotypic variability was observed except for individual 2 in P. urvillei, which
contained the 198-bp allele.
Among the loci reported here, Pdfl2, Pdfll2a, Pdfll5b, and Pdfl20a were found to
be compound (CA/GT)n (AT/TA)n repeats (Table 2-2). In a genome-wide survey in rice,
Temnykh et al. (2000, 2001) found that (CA)n repeats were frequently associated with
In genome-wide surveys of grasses, (CA)n repeats have been reported to be
relatively short compared to other dinucleotides (Temnykh et al. 2000). During the
enrichment phase of the capture protocol, this may have led to the retention of a limited
number of longer repeats, which may explain the high level of redundancy (4/24) of the
captured clones. A strong bias towards long repeats may be advantageous because the
length of the perfect repeats is expected to be associated with higher degrees of
variability (Symonds et al. 2003). Here, loci with more than 30 perfect repeats (Pdfl6,
Pdfl 5a, Pdfl22, and Pdfl28) revealed the highest number of alleles per locus in both P.
dilatatum ssp. flavescens and P. urvillei.
Low temperatures during the PCR extension step have been suggested to reduce the
generation of frameshift products (commonly known as "stutter"), particularly for
fragments containing (CA)n repeats (Hite et al. 1996). Under standard PCR conditions
like those used here, though, extension temperature cannot be reduced below the desired
annealing temperature. In preliminary amplifications a noticeable reduction in the
number of stutter peaks was observed, particularly for long alleles, when extension was
carried out at 530C rather than at 720C, while even lower temperatures resulted in the
production of locus-nonspecific products. However, the use of a single, robust
amplification profile does not require the adjustment of annealing and extension
temperatures for each primer pair individually, and greatly increases logistic efficiency
when simultaneously working with multiple primer pairs.
Most studies focus on primer pairs that amplify highly variable single loci. In this
study, all bands were taken into account because stable, biotype-specific, co-dominant
markers could be extremely useful for hybrid analyses within the complex. Sourdille et
al. (2001) analyzed a set of wheat microsatellite primer pairs including primers that
amplified more than one locus taking into account their known chromosomal locations
and transferability. In that study, 54% of the primer pairs amplified more than one locus,
including cases in which the extra bands were monomorphic, independently segregating
variable loci or co-segregating linked markers. These results are very similar to the ones
obtained in this study, in which 50% of the primer pairs amplified more than one putative
locus. In this study, however, extra bands were either monomorphic or independently
segregating, but no putative tandem duplications were found. Definitive assessment of
homeology would require a genetic map showing that the loci are located in syntenic
homeologous chromosome segments.
The number of products amplified was always equal to or less than the number of
loci expected for a tetraploid amphiploid, always consistent with the interpretation that
primer pairs were amplifying products from either one or the two genomes. Remarkably,
primer pair Pdfl20 amplified a number of bands exactly corresponding to the ploidy level
in heterozygous individuals.
At least two variable loci were identified for each of the biotypes including the
three sexual tetraploids represented by only two individuals. It is likely then that if more
individuals were analyzed, this set of loci could contain useful markers for population
structure and breeding system assessment of all the biotypes in the group. Overall, the
focal species showed less variation than P. urvillei. Only in 25% of the loci (Pdfl4,
Pdfl6a, Pdfl20a, and Pdfl28) did P. dilatatum ssp.flavescens show consistently longer
repeats than P. urvillei. It is typically expected that due to selection for long repeats
during library enrichment, longer repeats and higher variability are more likely to be
found in the focal biotype (Ellegren et al. 1995), an artifact known as ascertainment bias.
However, P. urvillei consistently showed more variability and a much higher level of
heterozygosity for most loci. A higher number of alleles was captured in P. urvillei, even
at loci for which the average fragment length was clearly lower than in P. dilatatum ssp.
flavescens (Fig 2-2 g, j) which contradicts the accepted consensus that repeat number and
variability are associated regardless of the causes of this correlation (Schloterer 2000). A
statistical comparison of variability within different biotypes is beyond of the scope of
this paper; however, the clear differences in variability and heterozygosity between P.
dilatatum ssp.flavescens and P. urvillei are likely to be real despite the small sample size
presented here. Amos et al. (1996) claim that heterozygosity may lead to an increase in
mutation rates at microsatellite loci. This may seem to be the case when comparing P.
dilatatum ssp. flavescens and P. urvillei; however, the stochastic effects of the restricted
distribution of P. dilatatum ssp.flavescens and its apparently extreme selfing rate may
deserve further investigation as putative explanations for the observed "reverse
ascertainment bias". The relative distance between the species analyzed should also be
taken into account to interpret meaningfully cross-amplification and ascertainment bias.
The I, J, and X genomes within the Dilatata polyploids can be considered to represent
different species because they are implicitly assumed to have diverged independently as
different diploid species between their coalescence time and the polyploidization event
that brought them back together. When the putative homeologous sequences shown in
Fig.2-2 are compared, a strong ascertainment bias is evident in all of them. Similar
flanking sequences combined with shorter and imperfect repeats like those found in the
non-focal loci in this study were found by Chen et al. (2002) when they amplified
microsatellites developed for Oryza sativa in congeners containing different genomes.
Variation was also found between the two individuals of P. dilatatum ssp.
dilatatum. These two individuals share 13 heterozygous allele combinations, making it
very unlikely that the three differences found (Pdfll5a and Pdfl22b, Fig. 2-1 h and k,
respectively) are due to a sexual recombination event or independent origins. However,
enough mutations seem to have accumulated in this clonal biotype to observe variability
with this set of microsatellite loci.
Alleles found in P. dilatatum ssp. dilatatum in eight loci are also present in
Virasoro, suggesting that this tetraploid could have been involved in the origin of the
pentaploid biotype. The pentaploids could not have arisen directly from a cross involving
the Virasoro genotypes analyzed here because the pentaploids are heterozygous for loci
for which Virasoro shows null alleles (Pdfll2a and Pdfll8). More intra-biotypic
variability must be analyzed, but the markers developed in this study seem to have great
potential for assessing the relationships among the sexual and apomictic components of
the Dilatata group.
Clustering and uneven genomic distribution of (CA)n motifs has been reported in
several genomes (Elsik and Williams 2001, Schmidt and Heslop-Harrison 1996);
however, no close linkage was detected among the loci analyzed in this study. Any
subset of these loci can then be chosen for a specific application based on amplification
consistency and variability to provide independent characters.
Even though cpDilB was the only variable chloroplast microsatellite identified, it
could potentially be very informative for assessing hybridization among biotypes because
the chloroplast genome is inherited as a single cohesive group and different alleles are
fixed within biotypes. Paspalum urvillei is the most widespread of the sexual members
of the Dilatata group, and its current range overlaps with those of the rest of its members.
Putative hybrids can be confirmed or P. urvillei can be ruled out from being the female
progenitor by using this marker. In this sample, individual 2 of P. urvillei was the only
one that showed a chloroplast allele that is not typical of its biotype. This accession was
indeed collected near the area of co-occurrence with biotype Vacaria with which P.
urvillei has been reported to hybridize (Valls and Pozzobon 1987). Our ability to identify
this putative hybrid confirms the utility of this marker.
Nuclear and chloroplast markers are reported with potential applications in
population genetics and phylogenetic studies within the Dilatata group. Highly variable
nuclear markers can be used to address population structure and breeding system issues
for all the biotypes in the group. On the other hand, more stable biotype-specific loci
may be used as co-dominant markers to assess the relationships among biotypes and
particularly the origin of the apomictic components of the complex. The variable
chloroplast microsatellite locus reported may in turn provide valuable information about
the relationship between the most widespread sexual member of the complex (P. urvillei)
and the rest of the biotypes.
BREEDING SYSTEM AND POPULATION GENETIC STRUCTURE OF Paspalum
dilatatum ssp. flavescens (POACEAE).
Paspalum dilatatum Poir. is a warm-season grass native to the grasslands of
temperate and subtropical South America. The species includes several tetraploid,
pentaploid and hexaploid forms. The common pentaploid form (P. dilatatum ssp.
dilatatum) was one of the first warm-season grasses to be cultivated for pasture (Skerman
and Riveros 1992), however, its commercial use has been limited by poor seed
production and susceptibility to ergot (Clavicepspapali). The pentaploid biotype
reproduces by apomixis (Bashaw and Holt 1958) and it has been assigned the IIJJX
genomic formula (Burson 1983). Because no sexuality has been reported in this
subspecies and collections have not provided much variability for the characters of
interest, efforts were soon undertaken to elucidate its relationships with other Paspalum
species in the hope that a breeding strategy could be devised. As a consequence of
extensive collections and cytogenetic investigations in P. dilatatum and related species,
three sexual tetraploid biotypes with the IIJJ genomic formula have been identified and
are usually included in the species (P. dilatatum ssp.flavescens Roseng. Arr. et Izag., and
biotypes Virasoro and Vacaria). Burson (1983) proposed that recombinant IIJJX
pentaploids could be synthesized by crossing one of the sexual biotypes and an apomictic
hexaploid with the IIJJXX genomic formula. This genomic formula was later assigned to
the hexaploid biotypes Chiru and Uruguaiana (Burson 1991, 1995), and synthetic
pentaploids have been readily produced (Burson 1991b, 1995, Speranza 1994 unpubl.
res.) and show excellent forage potential. As only one population is known for Chiri, and
probably both Chiri and Uruguaiana are very likely a single clone, genetic variability for
the breeding scheme mentioned above must come from the tetraploid parent. Among the
tetraploids, P. dilatatum ssp.flavescens is morphologically the most similar to the
common pentaploid biotype. The forage potential of P. dilatatum ssp. flavescens itself
has been assessed and compared with that of other biotypes. Besides a lower production
potential, this biotype's production is more concentrated in the spring (Millot 1969).
Paspalum dilatatum ssp.flavescens produces more seed than the pentaploids and it is
thought to be more resistant to Claviceps. Paspalum dilatatum ssp. flavescens is
currently distributed in southern Uruguay and the eastern province of Buenos Aires,
Argentina (Rosengurtt et al. 1970). Paspalum dilatatum ssp. dilatatum is also found
throughout the range of P. dilatatum ssp.flavescens. The tetraploid subspecies can be
distinguished from the pentaploids by its more erect and usually taller culms, larger, more
rounded spikelets, and yellow rather than purple anthers (Fig. 3-1). Recently,
microsatellite data for a small sample of this biotype suggested a high level of
homozygosity (Chapter 2)
A collection of P. dilatatum ssp.flavescens was deposited at the Germplasm Bank
of the Facultad de Agronomia in Montevideo, Uruguay during the 1990s. An efficient
use of this collection will depend on knowledge about its genetic structure, the breeding
system of the species, and its geographical structure (Epperson 1990, Rao and Hodgkin
2002). No morphological qualitative markers have been identified in the biotype, and
there is no evaluation of the degree of genetic variability it contains or its structure. In
this study, a sample of this collection is analyzed using six microsatellite loci, and the
breeding system, genetic structure, and geographical structure of the genetic variability
Fig. 3-1. A spikelet of P. dilatatum ssp.flavescens (a) and P. dilatatum ssp. dilatatum (b).
Note the relatively shorter, less hairy stigmata in P. dilatatum ssp.flavescens
and trapped anthers (arrow). The bar represents 1 mm.
Materials and Methods.
Because tetraploid and pentaploid plants are practically indistinguishable in grazed
pastures, most collections were made on major roads in southern Uruguay. Based on
field and greenhouse observations that suggested autogamy, an emphasis was made on
collecting seed from a single panicle from each individual while sampling as many
individuals as possible. Paspalum dilatatum ssp.flavescens grows in dense patches on
roadsides and in hilly areas, and patches are typically restricted to lower slopes excluding
the bottom. Each population consisted of up to four patches including both sides of the
road and both sides of a waterway when one was present. Seeds were deposited in the
Germplasm Bank at the Facultad de Agronomia in Montevideo, Uruguay. For most
accessions, a voucher specimen was also deposited at the herbarium of the Facultad de
Seed samples of 21 accessions were later retrieved from the Germplasm Bank for
this study. Accession numbers are given in Table 3-1, and collection localities can be
seen in Figs. 3-3 and 4. Eight individuals per accession were grown in greenhouse
conditions. All individuals were observed during the reproductive stage to ensure that no
pentaploids were present.
DNA was extracted from fresh leaves using Sigma GeneluteTM kit (Sigma-Aldrich,
St. Louis, MO). PCR and separation were carried out as described in (Chapter 2) using
extended forward primers with an M13 tail (Boutin-Garnache et al. 2001). Briefly, 0.5-3
[tL of DNA were added to a PCR mix consisting of 2 units of NEB Taq polymerase (New
England Biolabs, Beverly, MA), 1.5 mM MgCl2, 0.15 pM of the reverse primer and
labeled M13 primer, 0.01 [tM of the extended forward primer, and 0.1 mM of each dNTP
in the manufacturer's buffer. Amplification was carried out with the same two-step
program for all primer pairs consisting of a denaturation step of 15 s at 940C and an
annealing/extension step of 3 m at 530C (Chapter 2). Primer pairs Pdfl4, Pdfl6, Pdfl 12,
Pdfl 15 and Pdfl20 (Chapter 2) were selected for this study. Three different fluorochrome
labels were used (D2, D3 and D4, Beckman Coulter, Fullerton, CA) for the M13 primers.
PCR products labeled with D2, D3 and D4 were combined 5:3:2, and 1 [tL of the
combined products was loaded on a CEQ 8000 capillary sequencer (Beckman-Coulter,
Fullerton, CA). Chromatograms were visualized on CEQTM Genetic Analysis system
software (Beckman Coulter, Fullerton, CA) and scored manually.
AMOVA (Excoffier et al. 1992) was implemented with Arlequin (Schneider,
Roessli, and Excoffier 2000) using both the number of different alleles (FST) and the sum
of squared size difference (RST) as a distance measure (Slatkin 1995). Average squared
distances (D1) (Goldstein et al. 1995, Slatkin 1995) and the proportion of shared alleles
(Dps) (Bowcock 1994) as implemented in MSAanalyzer (Dieringer and Shloterer 2002)
were used to measure population differentiation. D1 was chosen because it takes into
account the mutation history reflected by allele size similarities caused by the stepwise
mutation mechanism typical of microsatellite loci. Furthermore, D1 is expected to reflect
linearly the divergence time between populations even when averaged across loci with
different mutation rates (Goldstein et al. 1995). The calculation of D1 makes use of
intrapopulational allele size variance; when calculation is based only on average allele
size as is commonly implemented for (6S)2 (Goldstein et al. 1995), the distance between
populations including a combination of long and short alleles and populations with only
intermediate alleles is severely underestimated, and this situation was often approached in
the current data set.
Population parameters of inbreeding (Fis) and gene diversity (He) were calculated
with Genepop (http://wbiomed.curtin.edu.au/genepop/).
A Mantel test (Mantel 1967) was implemented using Passage (Rosenberg 2001)
between matrices of D1 and Dps genetic distances and geodesic distances between
populations to detect isolation by distance (Heywood 1991, Slatkin and Arter 1991).
Geodesic distances between populations were generated by Passage from the original
population geographical coordinate list. Cluster and principal component analyses were
performed with MVSP (KCS, Anglesey, Wales).
All primers amplified the expected products in all individuals. Lack of
amplification in a first attempt was overcome by varying the amount of DNA used in the
PCR reaction in all cases. Although all DNAs had been extracted with the same protocol,
ten-fold variations in the amount of DNA were sometimes necessary to achieve
satisfactory amplification. No homozygotes for null alleles were thus detected, and given
the levels of homozygosity found, it is not likely that they are present as heterozygous
combinations in this sample.
Pdfl4 Pdfl6 Pdfl12 Pdfl15 Pdfl20 Pdfl22
Fig. 3-2. Allele size distribution for six microsatellite loci in P. dilatatum ssp.flavescens.
Shaded areas represent relative allele frequencies.
Table 3-1. Genetic diversity and heterozygosity for individual populations ofP. dilatatum ssp.flavescens for 6 microsatellite loci.
Pop." Acc.b Gen' Hetd F,se He Ag Het
101 7494 2 0 1.00 0.04 1
102 7495 7 3 0.67 0.57 2 2
50 7363 4 1 0.92 0.30 2 1
57 7364 4 0 1.00 0.47 2 0
62 7429 7 0 1.00 0.52 2 0
64 7432 6 1 0.93 0.30 2 0
65 7435 7 2 0.87 0.32 2 0
66 7433 4 1 0.78 0.19 2 0
68 7434 5 1 0.96 0.53 2 0
7 7355 6 3 0.79 0.31 1
72 7438 7 0 1.00 0.40 1
76 7467 5 0 1.00 0.30 2 0
78 7469 4 0 1.00 0.19 2 0
79 7470 3 0 1.00 0.35 2 0
82 7473 3 0 1.00 0.23 2 0
87 7478 4 0 1.00 0.33 2 0
91 7481 7 3 0.73 0.39 2 1
92 7482 2 1 0.80 0.10 1
95 7485 3 0 1.00 0.21 1
98 7491 4 0 1.00 0.16 2 0
99 7492 1 0.00 1
Totals 80 16 0.92 0.59 1.7 0.49
Reph A Het
(37-39) 4 2
(37-39) 2 0
(37-39) 2 0
(37-39) 2 0
(37-39) 2 0
(37-39) 2 1
37 2 2
39 2 0
(37-39) 2 0
(37-39) 2 0
(37-39) 3 0
(37-39) 3 0
Rep A Het
(21-36) 2 2
30 2 0
(21-36) 2 0
(21-30) 2 0
30 2 1
(30-36) 2 1
(21-30) 2 0
(30-36) 2 1
(30-36) 3 0
30 2 0
(30-36) 2 0
(21-36) 2 0
(21-36) 2 0
30 3 2
Rep A Het
(40-43) 2 1
(43-44) 2 0
(40-43) 2 0
(40-43) 3 0
43 4 0
(43-44) 3 1
(40-43) 3 0
(40-43) 2 0
(40-43) 2 0
40 2 0
(40-43) 3 0
(38-43) 2 0
(38-43) 2 0
(40-44) 2 0
40 2 0
37 1 30 2 0 (43-45) 2 0 (62-65) 2 0
(37-39) 2 0 (21-32) 1 45 1 42 1 -
37 1 30 1 40 1 42 1 -
Ave He He Ave He Ave He
(37-39) 1.8 0.39 (21-36) 1.8 0.62 (38-45) 2.0 0.75 (32-66) 1.7 0.62
Rep A Het
46 2 0
(34-43) 5 1
(43-45) 2 0
(43-45) 2 0
(37-45) 2 0
43 3 0
43 2 1
(43-45) 3 0
43 2 0
(39-43) 3 0
(43-45) 3 0
45 2 0
(43-45) 3 1
(43-45) 2 1
(34-43) 2 0 (45-47)
45 2 0 (52-53)
45 1 45
(34-46) 2.1 0.66 (40-54)
Population label. b Accession number. c Number of genotypes per population. dNumber of heterozygous individuals. multilocus
fixation index. f Expected heterozygosity. g Number of alleles. number of repeat units in the microsatellite inferred from allele
All loci were variable, providing high expected heterozygosities (Table 3-1). Pdfl4
was the least variable locus (He= 0.39) and the only one to show a clearly unimodal allele
size distribution with one allele showing a frequency of 0.77 (Fig. 3-2).
Allele size ranges and their frequencies for the whole sample are shown in Fig. 3-1.
Allele size distributions are clearly bimodal for Pdfl6, Pdfll2, Pdfl 15, and Pdfl20.
Consequently, in spite of the high proportion of alleles with very low frequencies, He
values for all loci across populations are rather high (Table 3-1). Besides showing
bimodal distributions, not all possible allele sizes are represented. Three considerable
gaps were found: a 20-bp gap in Pdfl4 and two gaps of 20 and 24 bp, respectively, in
The sample contained a high level of overall homozygosity; only 16 individuals
(9.5%) showed any heterozygosity. However, only 4 populations contained more than a
single heterozygote, and notably only 4 individuals were heterozygous at more than one
locus, three of them from population 102. In turn, those three individuals share all the
segregating alleles, making it possible that they may belong to a single segregating
progeny (data not shown). Eleven populations (52%) contained no heterozygotes, and
fixation indices within individuals (Fis) for the rest of the populations are high, including
those for population 102 (0.673). The average Fis value is 0.916. In spite of the low
number of heterozygotes, intrapopulational variability is considerable, with about half of
the populations reaching He values above 0.35. Although one population (99) contained
a single genotype, the average number of genotypes is 3.8 per populational sample of
only 8 individuals. The average number of alleles per locus within populations is 1.8,
ranging between 1.7 and 2.1 (Table 3-1). Then, despite differences in variability, all loci
contributed allele size differences which would detect outcrossing if it were taking place.
Allele size distribution between populations varied among loci; however, for some
loci, particularly Pdfl6 and Pdfl 15, several variable populations showed non-overlapping
ranges (Table 3-1). This observation suggests that there may be structure in the allele
size range within populations, which could shown by using a distance method that takes
allele size into account.
Table 3-2. AMOVA of a six-microsatellite-locus data matrix for 21 populations of P.
dilatatum ssp. flavescens.
Distance measure RST
Source of Sum of Variance Percentage
variation d.f. squares components of variation
populations 20 28961.05 86.62 Va 58.27
populations 315 19537.88 62.03 Vb 41.73
Total 335 48498.93 148.65
Fixation Index FST : 0.583
Significance test P(rand. value >= obs. value) 0.00000+-0.00000
Distance measure FST
Source of Sum of Variance Percentage
variation d.f. squares components of variation
populations 20 330.14 0.98Va 54.04
populations 315 262.44 0.83 Vb 45.96
Total 335 592.58 1.813
Fixation Index FST : 0.54
Significance test P(rand. value = obs. value) = 0.00000+-0.00000
Significance test P(rand. value = obs. value) = 0.00000+-0.00000
Va and FST : P(rand. value >= obs. value) =- 0.00000+-0.00000
240 D1 200 160 120 80 40 0
b -580 -570 -560 -550 -540
S \ 99A
82 1 78 9 7 r 8 -4 u
68 61-I 62 92, 57 /
--- < 79' 91 1 \ 02 ; '/
Montevideo -. roads -35-
0 0 00
0 1 S *
0 10 20 30 40 50
Fig. 3-3. Genetic distances among 21 populations ofP. dilatatum ssp.flavescens and
their geographical distribution, a. Complete linkage phenogram of D1
distances, b. Geographical distribution of the clusters shown in a (see
symbols). Only roads on which the samples were collected are shown, c. a
Mantel correlogram of D1 genetic distances and geographical distance
between populations. Open circles show significant correlations.
b 58 55
87360 300 240 180 13420 60 0
S 76 87 -340 u
68 66 62 92 57
S91 95 102
S72 Atlantic Ocean
Rio de la Plata 64 50
Fig. 3-4. Genetic distances among individual genotypes of P. dilatatum ssp.flavescens
and their geographical distribution, a. Average linkage (UPGMA) phenogram
of D1 distances. b. Geographical distribution of the clusters shown in a (see
symbols in a).
Population subdivision as analyzed by AMOVA produced FST values of 0.58275
when using the average allele size difference as a distance measure, and 0.54041 when
using the number of shared alleles. Both estimates were highly significant (Table 3-2).
D1 genetic distances between populations were calculated and the distance matrices
used for cluster analysis. Several linkage methods were used and all resulted in similar
clusters (results not shown). One of the best resolved phenograms is shown in Fig. 3-3a.
The resulting cluster memberships were plotted on a map (Fig. 3-3b) to visualize any
possible geographical structure. Mantel tests were also performed comparing the D1 and
Dps distance matrices to the geographical distance matrix. None of these tests was
significant. Mantel correlograms were also produced which showed no association
between geographical and genetic distance with either distance measure. One of those
correlograms is shown in Fig. 3-3c.
A similar approach was followed to analyze possible geographical structure in the
distribution of individual genotypes. A UPGMA phenogram of D1 genetic distance
between pairs of individual genotypes is shown in Fig. 3-4a, and the geographical
distribution of the two main clusters is shown in Fig. 3-4b. Dps produced similar results
(data not shown). The two main clusters plotted in Fig. 3-4b not only show a complete
lack of geographical segregation, but they also show extensive overlap.
The observed inbreeding rate as reflected by the fixation index (Fis) depends not
only on autogamy but also on pollinations between flowers from the same inflorescence
(geitonogamy), flowers on different ramets of the same genet (tillers), and, finally, the
genetic substructuring of the populations which increases biparental inbreeding (Ritland
1984, Brown 1989). While all these effects are confounded in populational data, as in
our case, there are indications that the selfing rate itself is very high.
The high intrapopulational genotypic differentiation observed in P. dilatatum ssp.
flavescens, coupled with extremely high fixation indices (Fis=0.97), suggests a lower
contribution of biparental inbreeding to the observed homozygosity in relation to the
actual selfing rate. Extensive observation of florets of this species during manual
hybridization suggests two mechanisms for such a high selfing rate. First, P. dilatatum
ssp.flavescens shows several characteristics that are typical of cleistogamous grasses
(Cambpell et al. 1983): stigmata and stamen filaments are relatively short in reference to
floret dimensions when compared to other biotypes of the species, and anther dehiscence
occurs immediately after or during anthesis. Furthermore, one, two or all three anthers
have been often observed to remain within the floret during and after anthesis, while
stigmata barely protrude outside the spikelet. Frankel and Galun (1977) classified this
mechanism as functional autogamy, and it has frequently been observed in grasses. The
second mechanism involves anthesis prior to emergence of the panicle from the sheath;
this has been more frequently observed for the first one or two racemes of the panicle.
Campbell et al. (1983) listed this mechanism as Type Ia and considers that cleistogamy in
these cases may be mostly an environmental response. In agreement with field and
greenhouse observations in P. dilatatum, factors such as light, temperature and moisture
have been reported to affect anthesis and panicle emergence (Campbell et al. 1983). For
a warm-season grass with an extended reproductive phase like P. dilatatum,
environmental responses could cause wide seasonal variations in selfing rates. Delayed
anthesis and faster culm elongation in the early spring under cool temperatures and no
moisture restrictions may increase the probability of outcrossing.
Genetic Differentiation and Geographical Structure
Genetic differentiation overall is very high in P. dilatatum ssp. flavescens. FST
estimates for continuous populations in which autogamy is thought to be the main factor
contributing to structuring have been reported to vary between 0.29 and 0.78 (Heywood
1981). The current sample of P. dilatatum ssp. flavescens was mostly collected in
disturbed environments such as roadsides, and seed morphology suggests dispersion by
gravity. These two factors, together with the observed selfing rates, contribute to a high
expectation for population subdivision (Loveless and Hamrick 1984). Hamrick and Godt
(1990) estimated a proportion of total genetic diversity among populations of 0.51 for 78
selfing species analyzed with isozymes, whereas their estimates for outcrossing species
vary between 0.1 and 0.2 depending on the mode of seed dispersion. Our FST estimate of
0.54 is therefore high and reflects the biological characteristics of the species. In our data
set, non-overlapping allele size distributions in variable populations suggest the presence
of informative non-random distribution of allele sizes. Regardless of the exact
evolutionary model each microsatellite locus follows, the process seems to retain some
mutational memory that can be captured by RST much more efficiently than FST with the
resulting underestimation of overall genetic differentiation by the latter (Slatkin 1995). In
this case, however, the use of RST instead than FST as a distance measure in the AMOVA
did not significantly increase the estimate as expected. There are several possible reasons
for this. FST and RST estimates of genetic subdivision would converge in situations where
mutation plays a minor role in population differentiation in relation to drift (Slatkin
1995). Population and mating system dynamics leading to high levels of genetic drift and
recent long-distance dispersal of individuals or population expansion will then improve
the performance of FST relative to RST (Slatkin 1995, Estoup et al. 1998).
Under a strict stepwise mutation model, gaps in the allele-length distributions are
expected after severe drift, like that caused by small local effective population sizes or
founder effects followed by a rapid expansion of the population. With time, mutation is
expected to fill the gaps between the sampled alleles (Cornuet and Luikart 1996). Not
only were some large gaps detected in this sample, but also, in ranges in which allele
sizes are rather continuous, the distributions were multimodal, with very low frequencies
for the intervening length categories. It could be argued that microsatellites have not
been observed to follow a strict stepwise mutation model. It seems reasonable to assume
that a mixed model (Di Rienzo 1994) is most realistic. Direct observation of
microsatellite mutations in maize has shown a majority of single-step length increase
events and fewer greater downward mutations (Vigouroux 2002). Observed allele-size
frequency distributions in self-pollinated Arabidopsis thaliana show continuous, mostly
unimodal distributions, especially for loci with a high number of perfect repeats
(Symonds and Lloyd 2003). In P. dilatatum ssp.flavescens, loci with very long perfect
repeat tracts like Pdfll5 and Pdfl6 (Chapter 2) show clearly discontinuous multimodal
distributions. It follows then that observed mutation patterns alone would not fully
explain the allele-size distributions observed in this sample.
The lack of correlation between geographical and genetic distance also supports a
scenario in which drift or long-range dispersal dominates the genetic dynamics of the
population. No correlation or clear geographical pattern was observed with population or
individual genotype data. Furthermore, the geographical distribution of identical
genotypes shared by several populations did not show geographical clustering (data not
It has been shown, particularly in selfing grasses, that molecular marker diversity,
either allozymes or microsatellites, could reflect the effects of selection (Allard 1972,
Nevo 1998, Li 2000). Rare hybridization events or long-distance dispersal can trigger
rapid genotype turnover at a particular site or microsite. Seemingly continuous
populations may then harbor widely divergent locally adapted genotypes whose
relatedness would not be reflected by neutral molecular markers. Green et al. (2001),
working at a more detailed spatial scale than in this study, found admixtures of highly
divergent microsatellite genotypes in Anisantha sterilis, a selfing weedy grass. Selection,
migration and temporal variation in the mating system were invoked as possible causes
for the lack of spatial genetic structure and persistence of divergent genotypes within
The great majority of our accessions were collected on roadsides; for two main
reasons: first, no dense stands of this subspecies are commonly found in other
environments, and second, P. dilatatum ssp. dilatatum is found almost continuously in
nearly all areas in Uruguay and almost always where P. dilatatum ssp.flavescens is
found. Distinguishing between subspecies in the field is not easy unless panicles are
present. Except for vacant lots in urban areas, most of the country is under grazing by
cattle, and panicles of P. dilatatum ssp.flavescens are seldom observed under grazing. In
fact, P. dilatatum ssp.flavescens may be largely excluded by grazing. Road construction,
shoulder leveling and periodical mowing of the roadsides are three types of events that
can strongly affect population dynamics. Construction may provide an opportunity for
long-distance dispersal or severe local bottlenecks. The plant cover of a road
construction site is completely cleared, creating open environments for colonization. The
seeds of the future colonizers may be brought from variable distances depending on the
origin of the gravel that is used or the application of turf patches to the barren slopes that
are created. Particularly in secondary roads, many stands of this subspecies extend into
the gravel shoulders and even into minor cracks in the asphalt itself. Shoulder leveling in
these areas necessarily involves uprooting and dragging whole or big portions of P.
dilatatum ssp.flavescens stands over variable distances. On main roads, the roadsides
also are periodically mowed with inevitable movement of panicles and seeds. From a
long-term perspective, species composition in Uruguayan grasslands has been greatly
influenced by the introduction of cattle in the 17th century. Rodriguez et al. (2003) have
shown that during a ten-year grazing exclusion, dominant grass types in a Uruguayan
grassland rapidly shifted from prostrate warm-season species to cool-season erect types
with narrow leaves and bigger seeds. Grazing patterns must have been changing
continuously due to the gradual replacement of Pampas deer by cattle. Pampas deer
(Ozotoceros bezoarticus) were still abundant at the beginning of the 19th century (Darwin
1839), while they are currently restricted to only two small populations in the whole
country (Gonzalez et al. 1998). Altesor et al. (1998) have in turn reported dramatic
floristic change for the same site during 55 years under continuous grazing. It is clear
that the impact of grazing in Uruguay during the last four centuries must have been and
still is strong, particularly on erect grasses like P. dilatatum ssp.flavescens. It can be
hypothesized that the subspecies may have suffered a strong bottleneck after the
introduction of cattle followed by relatively recent recolonization of the fenced roadsides.
This event, coupled with the present short-term effects of roadside habitats on population
dynamics and dispersal, sufficiently accounts for the lack of geographical structuring,
lack of mutational memory evidenced by the low RST values, and discontinuous allele
EVOLUTIONARY RELATIONSHIPS AND MECHANISMS IN THE DILATATA
GROUP (PASPALUM, POACEAE)
The genus Paspalum contains approximately 350 to 400 species (Clayton and
Reinvoize 1986) and has traditionally been divided into informal groups (Chase 1929).
The Dilatata group of Paspalum contains several species with great forage potential, and
several of them have been used as forage crops (Skerman and Riveros 1992). Paspalum
dilatatum Poir. and its related species are warm-season grasses native to the grasslands of
temperate South America and they are well adapted to resist frequent frosts during the
winter (Burson et al. 1991, da Costa and Scheffer-Basso 2003). This environmental
adaptability has allowed some members of the group, particularly P. dilatatum ssp.
dilatatum and P. urvillei Steud., to reach worldwide distributions wherever a warm-
temperate climate combined with sufficient rainfall exist.
The members of the Dilatata group have been classified into several formal and
informal taxonomic categories that will be referred to as biotypes in this study. The
common biotype of P. dilatatum (P. dilatatum ssp. dilatatum) is a complex apomictic
pentaploid hybrid, and efforts to identify its putative ancestors have led, over several
decades, to the accumulation of abundant cytogenetic information about the relationships
among all the species and biotypes within the Dilatata group. The conclusions of a whole
era of cytogenetic analysis based on meiotic studies in interspecific hybrids are
summarized in Table 4-1.
Table 4-1. Genomic formulae and reproductive systems of the members of the Dilatata
Species or biotype 2n Genomic Author* Reproductive system**
(x 10) Formula
dilatatum ssp. 50 IIJJX Burson (1983) Apomictic (Bashaw and Holt 1958)
P. dlatatumssp. 40 IIJJ Burson et al. (1973) Sexual
P. dasypleurum 40 IIJJ Quarin et al. (1995) Sexual
P. urvillet 40 IIJJ Burson (1979) Sexual
P. dilatatum 40 IIJJ Caponio et al. (1990) Sexual
P. dilatat40 IIJJ Quarin et al. (1993 unpub). Sexual
P. dilatatum 60 I X Burson (1991) Facultative apomict (Burson et al. 1991,
"Chirui" 60 IIJJXX Burson (1991) Millot 1977)
"Chinru" Millot 1977)
P. dilatatum 60 IIJJXX2 Burson (1992) Apomictic (Burson et al. 1991)
P. dilatatum Torres 60 ? Apomictic (Burson et al. 1991)
P. paucicilatum 40 AA1BC Moraes Fernandes et al. (1968) Apomictic
* Authorities are given for the publication of the genomic formula.
**Authorities are given for works that specifically addressed the reproductive system. All
sexual biotypes have been crossed, and sexuality is well established.
At least two attempts have been made previously to present the Dilatata complex in
a comprehensive way, including hypotheses on the relationships among its members.
Moraes Fernandes et al. (1968) did not provide genomic formulae but deduced genomic
architectures from meiotic behavior. They represented the genomic architectures in terms
of the number of copies of entire genomes which were assigned letters that do not
necessarily signify homology across biotypes. A number of biotypes were described in
this work, including the pentaploid apomictic common form (P. dilatatum ssp.
dilatatum), the Uruguaiana and Torres hexaploid biotypes, P. pauciciliatum and the
sexual tetraploid biotype of P. dilatatum now known as Vacaria. The latter was not
differentiated from the "yellow anthered" form (P. dilatatum ssp. flavescens) which had
previously been described by Bashaw and Forbes (1958). The common pentaploid
biotype (AABBC) was hypothesized to have arisen as a 2n+n hybrid involving a
tetraploid form and a diploid genome donor, or alternatively, as the product of an
allotetraploid and an allohexaploid sharing the A and B genomes. The hexaploid
Uruguaiana (AAAiA1BB) was proposed to be a 2n+n hybrid between two tetraploids,
and P. pauciciliatum (AAiBC) was hypothesized to be a hybrid between two tetraploids
with one partially homologous genome. Finally, the completely asynaptic Torres
(2n=6x=60) was hypothesized to be a putative hybrid between a tetraploid P. dilatatum
and an octoploid cytotype of P. virgatum with which it is sympatric and may share
morphological similarities. This hypothesis, however, does not explain the lack of
chromosome pairing in Torres given that variations of the IIJJ genomic formula have
later been assigned to tetraploid P. dilatatum sexual biotypes and tetraploid P. virgatum
(Burson et al. 1982).
An extensive program of interspecific hybridizations undertaken during the 1970s
provided the foundations for the current assignment of genomic homologies within the
group. Burson (1983) summarized this information including genomic formulae and
putative diploid donors for the component genomes. The sexual tetraploids were
assigned the IIJJ genomic formula, and they were hypothesized to have originated
independently from the diploid sources for these genomes. The diploid genome donors
were thought to be P. intermedium and P. juergensii, respectively (Burson 1978, 1979).
Several other putative I genome donors have been identified since then (Quarin and
Normann 1990, Caponio and Quarin 1993). The phylogenetic relationships among these
putative genome donors have recently been shown to be complex, spanning a
polyphyletic array of species (Vaio et al. 2005), and the identity of the direct donor of the
I genome has not been clarified yet. It is likely that new additional sources of the I
genome will be identified in the future.
In the arrangement presented by Burson (1983), the pentaploid biotype was thought
to be derived from a cross between P. dilatatum ssp.flavescens (IIJJ) and an unknown
hexaploid with the IIJJXX genomic formula. A hexaploid with the appropriate meiotic
behavior was described in Uruguay (Albicette 1980), and later, this hexaploid was found
to possess the predicted IIJJXX genomic formula (Burson 1991). Because Paspalum
species in general have very small genomes (Jarret et al. 1995), chromosomes are
generally small and relatively featureless; as a consequence, all of the cytogenetic work
relied solely on chromosome numbers and pairing in interspecific hybrids. Only recently
have prometaphase karyotypes been used as a source of phylogenetic information in the
genus (Speranza et al. 2003, Vaio et al. 2005). However, assessing relationships between
specific fixed apomicts requires techniques that can identify individual clones as putative
parents. For this approach multiple accessions of each biotype must be used to account
for variability. Complete collections of all the biotypes are not currently available.
Despite the lack of a comprehensive sample of all biotypes, an effort was made to
assemble a collection representing, at least partly, some of the variability within all of the
biotypes. A set of microsatellite markers developed for P. dilatatum ssp. flavescens
(Chapter 2) was used to describe the relationships among the apomictic biotypes of the
Dilatata group, find evidence of multiple origins or variability within them, and assess
their putative relationships with the sexual tetraploids. The variability and putative
evolutionary patterns within the common pentaploid biotype P. dilatatum ssp. dilatatum
will be discussed in depth elsewhere; here this biotype will be represented by a few
Materials and Methods
Seeds were retrieved from the USDA Plant Introduction Station (Griffin, Georgia,
USA) and the Germplasm Bank at the Facultad de Agronomia, University of the
Republic, Uruguay (Table 2). Seeds of P. dasypleurum were kindly provided by Ing.
For. R. Vergara from the Universidad Austral de Chile, and leaf material of an additional
accession of P. dilatatum Virasoro was provided by Dr. G. H. Rua from the Universidad
de Buenos Aires, Argentina. All four accessions ofP. dilatatum Chiru come from the
location indicated in Figure 4-1 and represent 4 different clusters of plants up to 2 km
apart. The heptaploid clone 59B ofPaspalum dilatatum was collected near Villa Serrana
(Lavalleja, Uruguay) in 1997. The site was revisited in 2000, and the same individual
was identified and recollected. Samples of both P. dilatatum ssp. dilatatum and P.
dilatatum ssp.flavescens from the same site were included in this study. Silica-gel dried
leaves of two triploid clones (N.A. 7663 and 7608) and a diploid individual (N.A. 7623)
were used to represent P. quadrifarium.
Accessions were classified by biotype at the Facultad de Agronomia, Uruguay, but
a preliminary biotype assignment for the USDA material was made based on field notes
except for previously described accessions (mostly Burson et al. 1991). Plants were
grown to the reproductive stage, and biotype assignments were corrected or confirmed by
chromosome counting whenever morphology did not agree with field notes.
Chromosome counts were made as described in Speranza et al. (2003). Briefly, root tips
were treated with 2mM 8-hydroxyquinoline for 4 h, fixed in 3:1 (ethanol: acetic acid) for
at least 24 h and stored in 70% ethanol at 40C. After fixation root tips were digested in
4% (w/v) cellulase (Calbiochem) and 4% (v/v) pectinase (Sigma), squashed and stained
with lacto-propionic (1:1) orcein.
-75o -65 -545o
S + Atlantic Ocean
M \ A A
0 / P dilatatum Vacana
0 P dasypleurum -350
Sbo O P Dilatatum Virasoro
o /e p 0 P dilatatum ssp flavescens
SP dilatatum ssp dilatatum
SP dilatatum clone 2
X P urville/
+ P pauciciliatum
A P dilatatum Urugualana
A P dilatatum Chin -40o
S dilatatum Torres
P dilatatum 59B
Fig. 4-1. Geographical distribution of the accessions
DNA Extraction and Microsatellite Analysis
Fresh leaves were collected and DNA extracted with Sigma GeneluteTM kit (Sigma-
Aldrich, St. Louis, MO). Microsatellite amplification and separation were carried out as
described in Chapter 2. Eleven primer pairs designed for nuclear microsatellites (Table
4-3) and chloroplast microsatellite primer pair cpDilB were chosen from those reported in
Chapter 2. Microsatellite amplification, separation, and detection were performed as
described in Chapter 3. Extended forward primers with an M13 tail were used in
combination with labeled M13 forward primers. Three different fluorochrome labels
were used (D2, D3, and D4, Beckman Coulter, Fullerton, CA). Chromatograms were
visualized on CEQ TM Genetic Analysis system software (Beckman Coulter, Fullerton,
CA) and scored manually.
Alleles in complex patterns were assigned to loci based on the approach given in
Chapter 2 for all biotypes that are known to contain a basic IIJJ genome assemblage. For
P. pauciciliatum, assignment was preliminarily made the same way. In the case of P.
dilatatum Torres, homology assessment was not attempted, and the data are presented
only graphically by primer pair. For P. dilatatum Uruguaiana and P. dilatatum 59B,
assignment was made a posteriori.
Table 4-2. Accession numbers of the materials retrieved from germplasm banks used to
analyze the relationships among the different biotypes
P. dilatatum Chini P. dilatatum ssp.flavescens P. urvillel
N.A.7537* N.A.7493 PI 462305
N.A.7662* N.A.7494 PI 462306
N.A.7672* N.A.7495 PI 509008
N.A.7359* N.A.7363 PI 509009
P. dilatatum ssp. dilatatum N.A.7364 PI 509010
N.A.7673* N.A.7433 PI 509012
N.A.7542* N.A.7434 PI 509013
N.A.7365* N.A.7355 PI 164065
P. dilatatum clone 2 N.A.7438 N.A.2957
PI310044* N.A.7441 N.A.7392
P. dilatatum clone 59B N.A.7468 NA 7390
N.A.7686* N.A.7470 N.A.7389
P. dilatatum Torres N.A.7474 PI 203752
N.A.7196* N.A.7475 N.A.7199
PI 404439** N.A.7476 P. dilatatum Vacaria
P. pauciciliatum N.A.7486 PI 404398
NA 7533* N.A.7492 PI404388*
PI310222* PI508723 PI404370
N.A.2500* PI 508720 PI 404382
PI310214* PI 508716 PI404372
P. dilatatum Uruguaiana PI 508722* PI 404434
N.A.7527* P. dilatatum Virasoro PI 508689*
PI404444** N.A.7207 PI404436*
Microsatellite data were summarized using MSAnalyzer (Dieringer and Shloterer
2002). Multilocus fixation indices (Fis), genetic diversity (He), and population
differentiation tests (Goudet et al. 1996) for the tetraploid biotypes were performed with
The distribution of microsatellite variance between and within sexual biotypes was
assessed with AMOVA (Excoffier et al. 1992) as performed by Arlequin (Schneider,
Roessli, and Excoffier 2000). The same program was used to compute Slatkin's
linearized distances (Slatkin 1995) to analyze the similarities among the tetraploids. For
Vacaria and P. urvillei, Mantel tests (Mantel 1967) were performed to test isolation by
distance between the geographical distance matrices and both average squared distance
(6[t)2 (Goldstein et al. 1995) and proportion of shared alleles (Dps) (Bowcock 1994).
Distance matrices were produced by MSAnalyzer. Both the Mantel tests and the
geographical distance matrices were obtained with Passage (Rosenberg 2001) from the
geographical coordinates of accessions.
The ancestry and possible admixture of apomicts and their contributions to hybrids
were analyzed using Structure (Pritchard et al. 2000). The ancestry model was set to
admixture, and population information was only used as a starting point. All further
simulations were performed based solely on the genotype of each individual. The
number of clusters (K) was set to 5 to represent the five known tetraploids in the group.
Otherwise all the default options were left in effect. Simulations were carried out for
50000 burn-in runs followed by 100,000 MCMC generations.
Variability in the Tetraploids
Among the three biotypes that were best represented in this data set, P. urvillei was
the most variable for all loci except Pdfl4 and Pdfl7, and Vacaria had the lowest Fis in
spite of having the lowest gene diversity (Table 4-3). These two biotypes show rather
high fixation indices (0.793 and 0.734, respectively). To analyze whether at least part of
the Fis values found for P. urvillei and Vacaria can be attributed to isolation by distance,
Mantel tests were performed between genetic and geographical distance matrices for both
biotypes, but no significant correlations were found (data not shown). The lowest Fis
over all the tetraploids was estimated for P. dasypleurum, although this value could be
highly biased because two individuals were available. The two individuals of Virasoro
were also completely homozygous. The breeding system and population structure of P.
dilatatum ssp.flavescens was studied in detail elsewhere (Chapter 3) and will not be
Samples of P. dilatatum ssp.flavescens from both Argentina and Uruguay were
included in this study; however, all the alleles found in the Argentinean accessions were
also present in the Uruguayan populations, and all individuals within the biotype were
assigned to the same population cluster by Structure (Fig. 4-2). The five tetraploid
biotypes were well differentiated based on the AMOVA, with a highly significant 89% of
the variability found among biotypes (Table 4-4). The exact test of population
differentiation also showed highly significant differentiation for all population pairs (not
The results obtained with Structure generally assigned all individuals within each
biotype to the same population cluster, showing very clear differentiation among the five
tetraploid biotypes and significant admixture only for one individual in Vacaria and some
components of Vacaria within P. urvillei (Fig. 4-2). Some admixture of P. dasypleurum
within P. urvillei in one individual was also estimated. Finally, no variability for the
chloroplast marker cpDilB was found within biotypes except for one individual of P.
urvillei that showed the chloroplast haplotype found in P. dilatatum (Table 4-3).
Table 4-3. Summary of the microsatellite data for the sexual tetraploid biotypes of
Paspalum group Dilatata and genotypes for the apomictic biotypes.
Pd Pd ssp
cilhatum Chi ddatatum
172 176 180
P d ssp
27 223 233
224 221 235
201 201 201
210 210 210
187 204 204
FIs 0 571
a Sample size.
0988 0 793 0 734 1 000
0424 0641 0302 0167
b Observed number of heterozygotes.
c Gene diversity.
d Number of alleles.
e Only the alleles not present in common
P. dilatatum ssp. dilatatum are shown here. The
alleles shared with P. dilatatum ssp. dilatatum are shown in Fig. 4-2.
F m ssp flavescens P urvillel Vacarla 2
Fig. 4-2. Population structure of a sample of the members of the Dilatata group estimated
by Structure under the admixture model based on microsatellite data for 13
loci. Each color represents the contribution to each genotype of each one of
the five clusters generated
If the five tetraploid biotypes were derived from a single ancestral population, and
given that a strong ascertainment bias could not be detected within this group (Chapter
3), Slatkin's linearized RST would represent relative divergence times between pairs of
biotypes. RST distances are expected to be linearly related to evolutionary divergence
time and could be used to estimate phylogenetic relationships among the biotypes. A
UPGMA tree based on linearized RST is shown in Fig. 4-3a. Under this hypothesis,
biotypes Vacaria and Virasoro are very similar, while P. dasypleurum and P. urvillei
form a tight cluster to which P. dilatatum ssp.flavescens attaches. In this result, P.
dilatatum as currently delimited can still be considered monophyletic with P. dilatatum
ssp. flavescens very close to a sister clade formed by P. urvillei and P. dasypleurum.
To circumvent the reliance of the RST-based measures on a stepwise mutational
model of microsatellites, the proportion of shared alleles (Dps) was also used as a distance
measure. The UPGMA phenogram based on Dps shows high distance estimates for all
population pairs with the nearest pair (P. dasypleurum-Vacaria) joined at a distance of
0.77 (Fig. 4-3b). The arrangement of the biotypes is completely different from that
obtained with RST. In this case all the possible rooting options would nest either P.
dasypleurum or P. urvillei within P. dilatatum.
30 24 18 12
0.96 0.8 0.64 0.48
0.32 0.16 0
Fig. 4-3. UPGMA phenograms of the distances among the sexual tetraploid biotypes of
the Dilatata group based on 13 microsatellite loci obtained with different
distance measures a. Linearized RST b. Proportion of shared alleles
Variability in the Apomicts
P. pauciciliatum and P. dilatatum Chiru, Uruguaiana (Table 4-3), and Torres (not
shown) appeared as single clones; individuals within each biotype share a number of
heterozygous allele combinations ranging from a minimum of 4 in P. pauciciliatum to 13
among the different accessions ofP. dilatatum ssp. dilatatum. Within Torres and Chiri,
no differences were detected among accessions, and the single genotypes found are
shown in Fig. 4-3. In the case of pentaploid P. dilatatum, two clearly different clones
were identified among the individuals reported here, a typical, widespread clone and a
second clone referred to here as P. dilatatum clone 2 (Fig. 4-4). On the other hand,
mostly in typical P. dilatatum ssp. dilatatum (Table 4-3) and Uruguaiana (Fig. 4-4),
individuals differ by a small number of allele length differences attributable to mutation.
Most of this variability is restricted to the longer alleles of loci PdfllO and Pdfll5 (Table
4-3 and Fig. 4-3). The highly variable longer allele in locus Pdfll5 showed a different
length in each of the 3 individuals of typical pentaploid P. dilatatum presented here
(Table 4-3). Due to its extreme instability, this locus was not used in any of the
comparisons among biotypes.
Relationships among Apomicts
Paspalum. dilatatum clone 59B and Uruguaiana share 11 or 12 heterozygote allele
combinations with the typical clone of P. dilatatum ssp. dilatatum, and differences are
again restricted to the longer alleles of loci PdfllO (Fig. 4-3) and Pdfll5 (not shown). If
these shared bands are interpreted to be homologous, the "extra" bands (Table 4-3) can
be assumed to be located on the extra 10 chromosomes (Burson's (1995) X2 genome).
All these bands were also found in the two triploid P. quadrifarium clones (Table 4-3)
except for the 180-bp band in locus Pdfll 1 shared by both Uruguaiana and 59B.
285 Pdfl4 (+100bp)
265 1 1 Pdfl7
1 Pdfl20-1 (5Obp)
1 m Pdfll8
Fig. 4-4. Multilocus genotypes of the apomictic components of the Dilatata group
Chiri, P. pauciciliatum, and P. dilatatum clone 2 clearly share a significant
proportion of bands with Virasoro; however, Chiri contains alleles that cannot be directly
attributed to its putative Virasoro ancestry. The typical P. dilatatum ssp. dilatatum, on
the other hand, appears to contain alleles that can be assigned to P. dilatatum ssp.
flavescens (Fig. 4-2). The putative contributions of Virasoro, Chiri, and P. dilatatum
clone 2 to other biotypes are confounded because they all share a significant proportion
of alleles themselves (Fig. 4-2). Chiri and P. dilatatum clone 2 share 6 out of 8
heterozygous allele combinations making the likelihood of a sexual event linking the two
All the apomicts except Torres shared a 170-bp allele in locus Pdfl28b and a 177-
bp allele in locus Pdfl20c which are thought to be located on the X genome (Chapter 2).
Torres does not show the 177-bp band in locus Pdfl20c and it shows a 218-bp allele in
the chloroplast locus cpDilB instead of the 217-bp allele found in the other biotypes of P.
dilatatum. Torres frequently showed more alleles in the nuclear loci than the biotypes
known to be built on the basic IIJJ combination of the group. Alleles were scored to
maximize its similarity to other biotypes in the data set. This approach makes sense if the
asynaptic behavior of Torres is due to lack of chromosome homology and is not under
genetic control. In this case, if Torres is part of the Dilatata group, it can only contain
one copy of each of the I and J genomes. Even following this strategy, it is not possible
to derive all of its alleles from any other apomict in the group. In spite of this, at least
one allele from each of nine loci can be traced to one of the two clones of pentaploid P.
dilatatum (Fig. 4-3).
Heterozygosity of the Apomicts
The degree of heterozygosity in the apomicts is variable. Typical P. dilatatum ssp.
dilatatum is heterozygous for every locus, while Chirf and P. dilatatum clone 2 show
lower levels of heterozygosity (Fig. 4-3). In the case of P. pauciciliatum, half of the loci
located on an I or J genome are expected to be hemizygous because its genomic
architecture does not allow for two copies of both genomes, so the observed presence of
only one band at several loci cannot be reported as homozygosity as was done for Chirf
and P. dilatatum clone 2. A similar situation is found in Torres in which only PdfllO
amplified a single band (Fig. 4-3). In fact, based solely on its asynaptic meiotic behavior,
all the bands amplified by a single primer pair should be reported as homeologous and
the genotype as completely hemizygous.
Evolutionary relationships among the sexual tetraploid biotypes
The use of microsatellite data to assess phylogenetic relationships is questionable
(Garza et al. 1995, Doyle et al. 1998). Size homoplasy, stochastic effects of past
bottlenecks, allele size asymmetry, and the possibility that at least some of the biotypes
had independent origins (making a hierarchical hypothesis of relatedness meaningless)
are some of the concerns.
The effect of size homoplasy on population parameter estimates has been
frequently discussed, and its consequences have been modeled and predicted at the
populational level (Estoup 2002). Size homoplasy at the interspecific level has also long
been a concern (Doyle et al. 1998). On the other hand, the use of microsatellite data for
phylogeny reconstruction within species and between species has been tested against
other sources of information with relative success (Alvarez et al. 2001, Richard and
Thorpe 2001). In the present data set, however, it is rather the strong asymmetry in allele
sizes observed among the biotypes that raises the major concerns. This problem is in
some way related to the artifact known as ascertainment bias (Ellegren et al. 1995) where
the non-focal species appear as less variable due to the original selection of loci to show
long allele sizes in the focal species. It has been well established that the length of a
microsatellite allele is related to its variability and mutation rate (Symonds and Lloyd
2000). Once a microsatellite increases its length, it also increases its upward mutation
rate. One peculiarity of this microsatellite set as applied to the current group of species is
that different loci seem to have been amplified in different biotypes reaching higher allele
sizes associated with higher mutation rates (Pdfl 28 in P. dilatatum ssp. flavescens,
Pdfl22 in P. urvillei, and Pdfl20 in Vacaria) (Table 4-3). As a consequence, when
applying distance measures based on allele length, the biotypes not showing long allele
sizes at a particular locus necessarily appear as more similar to one another. However,
this artifactual similarity is in reality a symplesiomorphy in a phylogenetic context.
When applying a distance measure based only on the presence of identical alleles (Dps),
the larger alleles are not matched with those of other biotypes; these biotypes appear
similar to one another again based shared ancestral character states. This problem is
further evidenced by the high distance values among pair of biotypes obtained with Dps.
Alvarez et al. (2001) circumvented similar problems in Lycopersicon by using only loci
showing very low diversity indices (<0.25). If such a criterion were applied to this data
set, most information would be eliminated because the only loci showing diversity
indices near 0.25 are Pdfll5b, Pdfl7 and Pdfll2b, of which Pdfl7 is null in most P.
The tetraploid components of the group have been previously proposed to have
originated independently (Burson 1983), a possibility that is somewhat supported by the
presence of fixed, non-shared chloroplast haplotypes among the three named species:
Paspalum urvillei, P. dasypleurum and P. dilatatum. This scenario is further supported
by the clear genetic differentiation in their nuclear microsatellite loci. With abundant
evidence accumulating for the recurrent formation of polyploids in the last years (D.
Soltis and P. Soltis 1993, P. Soltis and D. Soltis 2000), the independent formation of at
least some of these biotypes remains a likely possibility. Regardless of mode of origin of
the five sexual tetraploids, the hypotheses of relatedness presented in Figure 4-2 should
be interpreted with extreme caution.
On the other hand, evidence for ongoing gene flow was only found between
Vacaria and P. urvillei. Valls and Pozzobon (1987) reported, based on field observations,
that Vacaria formed natural hybrids with P. urvillei where their areas of distribution
overlapped, while natural hybrids have not been recorded between Virasoro or P.
dilatatum ssp.flavescens and P. urvillei. This gene flow apparently has not had strong
effects on the genetic identity of the involved biotypes which remain morphologically
and genetically distinct. Although more data are required to quantify any degree of gene
flow among the tetraploid biotypes ofP. dilatatum and analyze their morphological
differentiation in a systematic way, it seems clear that they form distinct units and should
all deserve at least formal subspecific status.
Genetic structure of the sexual tetraploids
The genetic structure of P. dilatatum ssp.flavescens has been discussed in detail
elsewhere (Chapter 3). For the other biotypes, the topic has been specifically addressed
only in Virasoro (Hickenbick et al. 1992). Hickenbick et al. (1992) concluded that both
selfing and cross-pollination must occur in this biotype based on segregating progenies
detected using isozyme markers. They also observed developing embryos in Virasoro
spikelets dissected prior to anthesis. In this study, the two individuals analyzed were
fully homozygous for the 13 loci, indicating a high level of homozygosity; however, very
few differences were detected between the two accessions scored, making it difficult to
detect the occurrence of allogamy. Paspalum dilatatum ssp.flavescens (Chapter 3) and
Virasoro show relatively shorter anthers and stigmata than the other biotypes,
characteristics that are usually considered morphological evidence of functional
autogamy (Frankel and Galun 1977, Campbell 1999). Paspalum urvillei, P. dasypleurum
and Vacaria showed lower levels of inbreeding. Even though isolation by distance could
not be demonstrated in this data set, it is likely that at least part of the observed
homozygosity may be due to crossing between individuals carrying the same alleles.
Variability within the apomicts
Paspalum dilatatum ssp. dilatatum was the only apomictic biotype that showed
evidence of containing more than a single clonal genotype. In spite of this, some
microsatellite variability was found within typical P. dilatatum ssp. dilatatum,
Uruguaiana and P. pauciciliatum. Such variability could be useful in assessing genetic
diversity and phylogeographic patterns in these rather widespread clones, especially in P.
dilatatum ssp. dilatatum.
The addition of the X genome, apomixis, and the origin of pentaploid P. dilatatum
Two different pathways have been suggested for the origin of the pentaploid IIJJX
biotype. The X genome could have been added to the group by means of a hexaploid
(IIJJXX) and then transferred to the pentaploids when this hexaploid crossed to a
tetraploid (IIJJ) (Burson 1983, Moraes Fernandes et al. 1968) or it could have been
directly added to form a pentaploid by a 2n+n hybridization between a tetraploid (IIJJ)
and an unrelated diploid (XX) (Moraes Fernandes et al. 1968). The data discussed above
show that the two hexaploid candidates identified so far are more likely explained as
derivatives than progenitors of the pentaploids. The second pathway invokes the
production of unreduced gametes by P. dilatatum tetraploids. Chloroplast sequence data
support the derivation of the maternal genome from within P. dilatatum rather than from
the donor of the X genome (Chapter 2). Production of unreduced gametes by tetraploids
has been observed (Speranza unpub. res.). One P. dilatatum ssp.flavescens individual,
when pollinated by Chiri produced several 2n+n (2n=7x=70) hybrids. The hybrids of the
same P. dilatatum ssp.flavescens individual by Virasoro (Chapter 2), when crossed with
Chiri also produced several 2n+n hybrids with 2n=7x=70. This time, the maternal plant
was expected to be highly heterozygous and its genotype can be inferred from the
segregation data shown in Chapter 2 for its progeny. Analysis of its 2n+n hybrids with
Chirn, however, did not show the transmission of any heterozygosity from its maternal
parent for several microsatellite loci (data not shown). This indicates that the unreduced
gametes were produced by some second division restitution (SDR)-like mechanism.
Even if the possibility cannot be rejected, no mechanism has been observed that can
explain the formation of a primary pentaploid by the contribution of a heterozygous
unreduced gamete by a heterozygous tetraploid. There are, however, in the present data
set, relatively homozygous pentaploids, namely P. dilatatum Clone 2. The level of
heterozygosity in the pentaploids is then variable. How did a heterozygous pentaploid
form if hexaploids are excluded as intermediaries? Existing pentaploid individuals have
the ability to transmit euploid IJX gametes and produce new IIJJX pentaploids when
crossed to an IIJJ tetraploid (Mazzella and Speranza 1997). The observed extreme
heterozygosity could have been gained by existing pentaploids by successive cycles of
crossing to other tetraploid members of the group. An extensive survey of the
intrabiotypic variability within P. dilatatum could provide further evidence of this
Paspalum dilatatum Uruguaiana and 59B
Both Uruguaiana and 59B share almost all of the heterozygous allele combinations
found in the widespread typical P. dilatatum ssp. dilatatum. As a consequence, their
mutual relationship and the relationship of both clones to the pentaploid biotype cannot
be explained by sexual events. Early pollination of pentaploid florets has been shown
lead to the fertilization of aposporic embryo sacs and the production of 2n+n hybrids
(Bennett et al. 1969, Espinoza and Quarin 2000, Burson 1992). Furthermore, a putative
donor of the alleles not present in the pentaploid parent has been identified here as P.
quadrifarium. This suggests that these two clones are 2n+n hybrids of pentaploid P.
dilatatum and a diploid or tetraploid individual of P. quadrifarium. The close
morphological and physiological similarity between Uruguaiana and pentaploid P.
dilatatum has been previously noted (Burson 1991). Furthermore, the spikelets of
Uruguaiana (and clone 59B) differ from those of the pentaploid biotype in that their
maximum width is located near the middle of the spikelet like those found in species of
the Quadrifaria group (Barreto 1966) rather than near the base as seen in P. dilatatum.
The two accessions of Uruguaiana analyzed here were collected more than 220 km apart,
and they differ at some microsatellite loci. These differences were interpreted as
mutations given that the alleles hypothesized to represent the X2 genome were identical
for the two locations. Its geographical range and the presence of mutations suggest that
Uruguaiana is not a recent derivative of pentaploid P. dilatatum and that several more
such hybrids could have been formed in a similar way. The finding of 59B, a single
hybrid of the same kind shows that the complex is active and still generating new
combinations at higher ploidy levels. Clone 59B is a single, highly sterile individual
found near Villa Serrana, Uruguay and it is likely to be the product of a contemporary
hybridization event. This plant was collected at the edge of a water stream that crosses a
secondary road. Clone 59B was growing among plants of P. dilatatum ssp.flavescens and
pentaploid P. dilatatum. Immediately next to 59B were several plants of P. exaltatum
which shares genomes with P. quadrifarium (Vaio et al. 2005) and several dispersed
individuals of P. quadrifarium. Spikelet morphology and the presence of P.
quadrifarium-specific bands in 59B strongly suggest that a species from the Quadrifaria
group could have contributed the extra genomes in 59B.
Paspalum dilatatum clone 2
This accession of P. dilatatum can be partially explained by a cross between
Virasoro and Chiru. Morphologically this accession seems to correspond to the putative
Virasoro x Uruguaiana hybrids reported by Machado et al. (2005). Even though
Uruguaiana is presently known to occur near the collection site of P. dilatatum clone 2,
its genotype does not support such parentage. On the other hand, Chiru is only known
from a relict population in Paysandu, Uruguay and Clone 2 was found in Brazil,
approximately 240 km NE of this location. A greater distribution range of Chiru in the
past or the still undetected occurrence of genotypes closely related to it in southern Brazil
must be invoked to explain this hybrid as the product of a tetraploid x hexaploid hybrid.
The reverse hypothesis where Chiru is a derivative of P. dilatatum clone 2 is considered
Paspalum dilatatum Chiru
Chiru is closely related to Virasoro and less heterozygous than P. dilatatum ssp.
dilatatum and Uruguaiana. The involvement of Virasoro as the donor of the basic IIJJ
genomes seems evident; however, Chiru also shows alleles that cannot be explained by
the genotypes found in Virasoro, and given that Virasoro has been found to be highly
homozygous, it is not likely that the relationship between the two biotypes is direct. The
reverse pathway, i.e., the addition of an extra copy of the X genome to a P. dilatatum
clone 2 type individual in a fashion similar to that described for Uruguaiana, cannot be
excluded as the origin of Chirn. In fact, only the shorter allele in Pdfl 18 and the absence
of a second allele in locus Pdfll2 could not be explained this way. Under this scenario, P.
dilatatum clone 2 would be more widespread than detected here and Chiri would not be a
relict population of a formerly widespread biotype but a possibly recent 2n+n hybrid of
P. dilatatum clone 2. The only extra putative X genome allele found in Chiri is a 187 bp
allele at locus Pdfl20. This allele was also found in clone 59B, and it was attributed to its
P. quadrifarium parent. In the case of Chiri, this constitutes very weak evidence of the
origin of the X genome from within the Quadrifaria group.
P. pauciciliatum appears to be very closely related to and is sympatric with both
Chiri and P. dilatatum clone 2. The contribution of a reduced IJX gamete from either P.
dilatatum biotype as a pollen donor could explain most of the alleles found in P.
pauciciliatum and even more so if one-step mutations are allowed. Its chloroplast
haplotype, however, is different from that of the other members of the Dilatata group.
This difference was detected only at the cpDilB locus whereas the chloroplast haplotype
of P. pauciciliatum is identical to that of the rest of the Dilatata group five non-coding
regions (Chapter 2). Its cytogenetic architecture (Moraes Fernandes et al.1968) implies
that one of its genomes is present in two copies. If the difference found in the chloroplast
genome is taken as evidence of an independent origin of its chloroplast genome, then its
maternal progenitor must be closely related to the diploid donor of either the I or J
genome to the Dilatata group but not the same individual or population. The maternal
progenitor should be responsible for the contribution of extra bands at loci Pdfl 18,
Pdfll2b and Pdfl7. Under this scenario, its genomic formula should be either IIJX or
Paspalum dilatatum Torres
Chloroplast DNA data strongly support the contribution of the maternal genome of
Torres from within the Dilatata group (Chapter 2). However, its chloroplast haplotype
corresponds to P. urvillei instead of that shared by all the other biotypes of P. dilatatum.
For 7 out of 13 loci analyzed Torres shows alleles also found within either the P.
dilatatum ssp. dilatatum-Uruguaiana complex or the Chiru-P. dilatatum clone 2 complex;
however, it lacks one of the markers of the X genome that is shared by all the other
apomicts including P. pauciciliatum. Its link to the rest of the group may be some
genotype of the highly variable P. urvillei. If Torres contains a copy of the I and the J
genomes, as suggested by its similarities to P. urvillei, then its other genomes must not
come from another member of the group based on the lack of pairing of its chromosomes.
The relationship of this biotype to the rest of the group cannot be completely clarified
based on this data set and the task would probably require analyzing species belonging to
other taxonomic groups.
This study has established a series of evolutionary hypotheses within the Dilatata
group which will dramatically change the future direction of both basic and applied
research within this group.
1. Both nuclear and chloroplast markers show that the five sexual tetraploids studied
here are well differentiated and do not show significant gene flow except for P.
urvillei and Vacaria.
2. The Uruguayan and Argentinean accessions of P. dilatatum ssp.flavescens form a
single coherent unit.
3. P. dilatatum ssp.flavescens and Virasoro appear highly homozygous, in agreement
with floret morphology and previous reports. P. urvillei, Vacaria and P.
dasypleurum probably show higher degrees of allogamy.
4. No clear evidence was found to support the monophyly of the group.
5. The relationships among the sexual tetraploid biotypes could not be reliably
explained by the current data set.
6. Chloroplast haplotype differentiation is in agreement with current species
7. Uruguaiana, Chiru, Torres and P. pauciciliatum each appears to comprise a single
8. Pentaploid P. dilatatum comprises an assemblage of more than one clone.
9. Variability attributed to somatic mutations was found within Uruguaiana, P.
pauciciliatum and the typical clone of P. dilatatum for at least two loci, which can
be used for analyzing genetic structure within these clonal biotypes.
10. All of the apomicts in the group show the same alleles in the loci thought to reside
on the X genome, suggesting a single origin for this genome in the group.
11. All of the apomicts seem to contain at least one copy of each of the I, J and X
12. The hexaploids and heptaploids are better explained as derivatives of the pentaploid
P. dilatatum rather than its ancestors.
13. Virasoro is the sexual tetraploid that shows the greatest degree of similarity to all of
14. A genotype of pentaploid P. dilatatum derived from Virasoro might have been the
primary apomict in the group. Further pentaploid-tetraploid crosses may have
created the variability and the heterozygosity found in the apomicts.
15. The suggested mechanisms for the formation of new apomicts involve either an
unreduced female gamete or a euploid IJX pollen grain of pentaploid P. dilatatum.
PENTAPLOID X TETRAPLOID HYBRIDIZATION CYCLES IN Paspalum dilatatum
(POACEAE): EXPLAINING THE CURRENT AND FUTURE EVOLUTIONARY
SUCCESS OF AN IMBALANCED POLYPLOID
Paspalum dilatatum Poir. is a warm-season grass native to the grasslands of
temperate South America. This species is part of an informal taxonomic group within the
genus known as the Dilatata group (Chase 1929). Included in this group are several
sexual and apomictic biotypes and species whose evolutionary relationships have recently
been discussed in detail based on data from 13 microsatellite loci (Chapter 4). The sexual
members of the group (P. urvillei Steud., P. dasypleurum Kunze ex Desv., P. dilatatum
ssp.flavescens Roseng. Arr. et Izag., and biotypes Virasoro and Vacaria of P. dilatatum)
are all tetraploid and share the same genomic formula (IIJJ). The relationships among the
sexual tetraploids and between them and the apomicts have not been completely clarified;
however, microsatellite data suggest that Virasoro has greatly contributed to the genetic
makeup of the apomictic components (Chapter 4). Except for pentaploid P. dilatatum, the
rest of the apomicts in the group (P. dilatatum biotypes Uruguaiana, Chinr and Torres
(6x) and P. pauciciliatum (Parodi) Herter (4x)) seem to each comprise a single clone and
its mutational derivatives, and all of them have been hypothesized to be derivatives of the
pentaploid form (Chapter 4). In this context, the apomicts would all include at least one
copy of the I and J genomes and at least one copy of the third unassigned genome (X).
The pentaploid form (IIJJX) had previously been proposed to be the product of a cross
between a hexaploid (IIJJXX) and a tetraploid (IIJJ) (Burson 1983); however, the
transmission of an unaltered heterozygous multilocus genotype between pentaploids and
the hexaploids Chiru and Uruguaiana suggests that the latter may be more likely
derivatives of the pentaploid form by means of a 2n+n hybridization in which the
pentaploid contributed an unreduced gamete and the second X genome was contributed
by a diploid. Chloroplast sequence data suggest that the pentaploid probably acted as the
maternal progenitor in such crosses (Chapter 2, Chapter 4), a mechanism that had already
been reported experimentally (Bennett et al. 1969, Burson 1997). Genotypic information
for thirteen microsatellite loci suggests yet another mechanism by which the pentaploid
may be involved in the origin of the remaining apomicts, i.e., P. pauciciliatum, and at
least one recombinant pentaploid clone. The pentaploids are able to produce euploid IJX
pollen grains (Mazzella and Speranza 1995) which can fertilize a sexual tetraploid to
yield a recombinant pentaploid. Pentaploid P. dilatatum is not only the most widespread
biotype, but it also seems to have been the basis of the entire apomictic complex in the
group. It may have been involved in the origin of the other apomicts either by means of
unreduced female gametes or euploid IJX pollen grains, and it could be the original
carrier of the X genome. If the transmission of euploid IJX gametes is a frequent event,
then several more recombinant pentaploids like the one reported in Chapter 4 should be
found in the wild. Variability has been detected among pentaploid P. dilatatum
accessions with dominant markers (Casa et al. 2002); however, with this kind of markers,
and particularly for a small sample, clonal variants cannot be distinguished from
recombinants. Microsatellites have proven efficient in detecting recent hybridization
events and if recombinants are found it should be possible to infer their mode of origin
from their genotypes.
Regardless of its mode of origin or evolutionary role, pentaploid P. dilatatum is
currently distributed worldwide in warm-temperate regions of the north and south
hemispheres where it has become an important forage grass (Burson 1983, Skerman and
Riveros 1992). Because of this, there is great applied potential for the detection of genetic
variability in the available germplasm collections. Microsatellite markers can reveal
mutational variation that could be useful for assessing the extent and distribution of
mutational genetic variability within the typical form of this biotype (Chapter 2, Chapter
In this study, an extensive collection of pentaploid P. dilatatum representing its
worldwide distribution has been retrieved from existing germplasm banks and analyzed
using microsatellite markers developed for P. dilatatum ssp. flavescens (Chapter 2) to
assess the level of variability.
Materials and Methods
Seeds of pentaploid P. dilatatum were retrieved from the USDA Plant Introduction
Station (Griffin, Georgia) and the Germplasm Bank at the Facultad de Agronomia,
University of the Republic, Uruguay. Accession numbers are given in Table 5-1.
Additional dry leaves from California, Texas, and Australia were kindly provided by M.
McMahon, J. Tate, and K. Smith, respectively.
Material retrieved from Georgia mostly contained populational samples. The
Uruguayan collection used in this study was mostly developed by myself between 1992
and 1999. This collection is composed primarily of single-plant progenies produced in
the greenhouse from off-type individuals collected in the field. Populations 1 and 3 had
been previously reported by Prof J.C. Millot (1997 pers. comm.) as variable based on
morphological evidence. These populations were collected as single individuals, and each
apomictic progeny kept under a different accession number (Table 5-1). Chromosome
numbers were determined for all the Uruguayan single-plant accessions by P. Speranza,
M. Vaio and C. Mazzella following the technique described in Speranza et al. (2003).
For the current study, full seeds were germinated in Petri dishes on filter paper.
Germinators were placed at 40C for 4 days prior to incubation to break dormancy and
homogenize germination and were then transferred to an incubator with alternating
temperatures (16 h at 300C light, 8 h at 200C dark). For eight populational samples with
good seed quality, eight individuals were grown per accession, otherwise only one
individual was kept in each case (Table 5-1). Plants were cultivated in greenhouse
conditions for at least one complete growth season and screened for contaminants or
biotype assignment errors at the reproductive stage. Chromosome numbers were
determined when the originally reported biotype assignment and morphological
appearance of a plant were in disagreement. Chromosome numbers were also determined
aposteriori for all putatively recombinant individuals (Table 5-1).
DNA Extraction and Microsatellite Analysis
Fresh leaves were collected in the greenhouse and DNA extracted with Sigma
GeneluteTM kit (Sigma-Aldrich, St. Louis, MO). Microsatellite amplification and
separation were carried out as described in Chapter 2. Eight primer pairs reported in
Chapter 2 were used in this study (Pdfl4, Pdfl7, Pdfl8, Pdfll0, Pdfll 1, Pdfll2, Pdfll5,
and Pdfl20). Fragment amplification was obtained for all individuals by varying the
amount of DNA added to the PCR mix between 0.5-3 tL. Extended forward primers with
an M13 tail were used in combination with labeled M13 forward primers. Three different
fluorochrome labels were used (D2, D3 and D4, Beckman Coulter, Fullerton, CA). PCR
products labeled with different dyes were combined, and 1 aL of the combined products
was loaded on a CEQ 8000 capillary sequencer (Beckman-Coulter, Fullerton, CA).
Chromatograms were visualized on CEQ TM Genetic Analysis system software (Beckman
Coulter, Fullerton, CA) and scored manually.
Table 5-1. Accession numbers, genotypes, and population of origin of the pentaploid P.
dilatatum material retrieved from germplasm banks. The number of
individuals analyzed per accession is indicated in parenthesis if it is more than
one. The genotypes/individuals for which the chromosome number was
confirmed or is first reported in this study are marked by an asterisk. A
population number is given in those cases in which more than one individual
from the same location was analyzed.
Accession Genotypes Population Accession Genotypes Accession Genotypes Population
N.A.7346 (8) A/RecC2*/RecC7* 7 PI 173004 M1 PI 410284 P1
N.A.7368 (8) P/V 2 PI 202298 G PI 410286 A
N.A.7404 A* PI 202300 A PI 410287 A
N.A.7416 L* PI 217623 A PI 462248 P
N.A.7430 Recbl* PI 222812 Al PI 462254 (8) A/Q 9
N.A.7440 (8) J/P/RecB4* 6 PI 235068 A PI462256 (8) P/T 4
N.A.7465 (4) I/P/RecB3* 1 PI 271592 A PI 462258 A
N.A.7471 O* PI 273255 A PI 462261 (8) RecC2* 10
N.A.7524 A* PI 274081 A PI 462262 P
N.A.7525 W* PI 283015 RecA2* PI 462264 A
N.A.7528 D* PI 285302 A PI 508671 U
N.A.7529 Y* 3 PI 300076 A PI 508676 P
N.A.7540 G* PI 300077 A PI 508682 P
N.A.7541 P* PI 304015 E* PI 508692 F
N.A.7542 A* 7 PI 310044 RecAl* PI 508694 D
N.A.7543 RecB2* PI 310076 RecC4* PI 508701 A
N.A.7544 A* PI 310077 N PI 508703 D
N.A.7545 P* PI 310078 P PI 508705 A
N.A.7562 (8) H* 8 PI 310083 D PI 508706 A
N.A.7563 (8) C/RecC1* 5 PI 310088 A PI508707 M
N.A.7588 I* PI 310091 M PI 508708 A
N.A.7606 P* PI 331112 A PI 508712 B
N.A.7609 RecC3* 3 PI 338660 A PI 508715 A*
N.A.7613 A* PI 404394 P PI 508719 G
N.A.7618 RecC6* PI 404410 T PI 508725 P
N.A.7619 H* PI 404412 P PI 508857 T*
N.A.7657 X* 1 PI 404415 RecC5* PI 576135 M
N.A.7658 P* 1 PI 404431 I
N.A.7661 A* 1 PI 404432 P
N.A.7665 P* 3 PI 404823 P
N.A.7673 H* PI 409854 A
N.A.7674 RecA3* PI 410281 A
N.A.7688 S* 1 PI 410283 A
N.A.7690 A* NSL 28721 D1
Table 5-2. Allele frequency distributions in the tetraploid biotypes of P. dilatatum used to estimate possible contributions to the
recombinant pentaploid clones.
P P dilatatum ssp P P dilatatum P dilatatum P P dilatatum ssp P P dlatatum P dlatatum P P dilatatum ssp P P dilatatum P dilatatum
dasypleurum flavescens urville, Vacara Virasoro dasypleurum flavescens urville, Vacaria Virasoro dasypleurum flavescens urville/ Vacaria Virasoro
Pdfll Pdfl15 Pdfl20
172 018 100 216 004 207 008
174 1 00 218 209 062
176 004 018 100 222 211 027 013
178 050 224 008 215 1 00
180 096 014 226 217
Pdfl4 228 050 221 100
196 1 00 232 223
200 236 018 233
202 013 021 238 034 235 050
204 079 006 240 237 004
206 242 002 013 050 239 046
208 244 056 263 004 031
210 094 246 004 019 271 019
212 1 00 250 013 273 038
220 077 252 Pdf12
226 254 012 174 100
232 011 264 212 1 00
238 276 029 221 064
240 278 0 07 223
242 282 004 225 009
Pdfl8 286 004 229 009
198 004 292 008 233 018
218 011 044 294 008 235 039
220 006 296 012 239
224 038 298 015 241 061
226 007 006 302 008 245
230 1 00 1 00 312 008 251
232 326 008 253
234 332 008 Pdfl12b
236 011 0 32 0 06 354 0 04 200 1 00 1 00 0 86 1 00
238 082 039 PdfllO 210 014 100
240 007 175 Pdfl7
244 007 179 233 100
266 181 018 100 235
272 187 005 241 100
274 191 045 249 081
276 195 009 006 251 100 019
Pdfl15b 207 253
202 092 209 005 257 100
210 025 211 025 259
212 075 100 008 100 213 050 Pdfl20b
214 1 00 221 009 196 036
225 004 197 025 057
227 050 068 019 198 057
229 029 009 199 075 018 100
237 050 200 025
245 201 100
Alleles in complex patterns were assigned to loci based on Chapter 2. MSAnalyzer
(Dieringer and Shloterer 2002) was used to summarize microsatellite data and calculate
distance matrices based on the proportion of shared alleles (Dps) (Bowcock 1994).
The number of mutational steps of each genotype from the most common
genotypes was estimated based on a Dps matrix and used to decide which genotypes
should be further analyzed as recombinants.
For the genotypes identified as mutational variants of the major clone, a genotype
network was created using Network (www.fluxus-engineering.com ) by the Median
Joining method (Bandelt et al. 1999). Alleles of loci interpreted to be located on the I or J
genome according to Chapter 2 were assigned to either homologous locus in each
individual based on size and/or conservation of the other allele, i.e. the shorter band could
be considered a mutant of the "long" allele if the common "short" allele was conserved.
The entire genotype was further treated as a single haplotype because it was assumed that
recombination did not take place.
Information about allele frequencies in the tetraploid biotypes of the Dilatata group
was obtained from Chapter 4 and is here summarized and shown in Table 5-2 for
comparison. Two exploratory approaches were followed to identify the putative donor of
alleles to each recombinant pentaploid. First, a matrix of Dps was generated with
MSAnalyzer and the distance of each recombinant to each tetraploid biotype was taken as
an indication of putative ancestry. Second, allele admixture proportions were explored
using Structure (Pritchard et al. 2000). To circumvent the confounding effect of the high
number of alleles shared by the typical pentaploid genotype and Virasoro, a population of
a hypothetical ancestor of the typical clone was simulated by removing the alleles clearly
attributable to a Virasoro-like parent. The ancestry model was based on prior population
assignments for the tetraploids and admixture for the recombinant genotypes. The
simulation was re-run with the number of clusters set to represent the actual number of
putative ancestral populations present in the matrix. Otherwise, all of the default options
were left in effect. The final simulations were carried out for 50000 bum in runs
followed by 100000 MCMC replicates.
A total of 29 multilocus genotypes for 12 loci were identified using the eight
microsatellite primer pairs. Primer pairs Pdfll5 and Pdfll2 amplify two and Pdfl20 three
independent loci (Chapter 2). The 177-bp band which is thought to reside on the putative
X genome locus Pdfl20c was present in all the individuals and is not shown in the
figures. The data were first displayed graphically (as in Fig. 5-4) and visually analyzed.
A single most common genotype with a few mutations was evident while a few
individuals were clearly recombinants. For exploratory purposes the most frequent
genotypes (A and P) were treated as ancestral and mutational steps counted from them.
To assess the extent of the major clone, the number of allele size differences of each
genotype to the closest of either of the two versions of the major clone was calculated
(Fig. 5-1). All genotypes with 5 or fewer allele differences to the common genotype
were tentatively considered its mutational variants. A highly contrasting pattern was
observed between those genotypes with 7 or more allele differences with the common
genotypes and those with 5 or fewer. A progression in the number of allele-size changes
from 0 to 5 was clearly interpretable as somatic mutations because alleles accumulate in
the most unstable loci (Pdfll5 and Pdfll10) and the rest of the loci typically remain
unchanged or show single-step mutations in the longer alleles. These genotypes maintain
at least 6 original heterozygous combinations of the common genotype. This pattern is
not observed in genotypes showing 7 or more allele differences from the common
genotype; changes affect either allele and their size ranges are drastically different.
0 K--- ----
A P P1Q S T U W X B D I A1 CL YE H J FD1 G V M1M N C6B3B2B4B1C5C1C4A2C3A1A3C
Fig. 5-1. Number of microsatellite allele differences between all genotypes of pentaploid
P. dilatatum and the nearest of the two most widespread genotypes (A and P).
The total number of alleles considered is 22.
A genotype network was built for the major clone and its inferred somatic variants
(Fig. 5-2a). The network represents a completely hierarchical arrangement with a total of
44 changes in 12 variable alleles of which 13 changes in the same 4 alleles are
homoplasious (long alleles in Pdfll0, Pdfll5 and Pdfl8 and the short allele in Pdfl7) (Fig.
5-2a). Two frequent genotypes (A and P) include most of the individuals. Several
genotypes are connected to these two common ones by one or a few mutations. Of the
total 44 changes required to build the network, six were allele losses where the actual size
of the resulting allele cannot be observed. For the remaining 38 changes, 20 or 21
represent size increases and 18 or 17 size decreases depending on whether genotype A or
P is considered ancestral. These changes are not distributed evenly along the network or
across loci. A single branch connecting genotypes G, H, and J to A accounts for 11 size
increases. The ratio of increases to decreases for the rest of the tree is 9:16. Of the 44
total changes scored for 22 alleles, 10 are accounted for by the longer allele of Pdfl 15
alone and 7 by the longer allele of Pdfll. A significant size difference of 7 steps
separates a completely inactive Pdfl 15 long allele of approximately 28 GT repeats
(branches departing from P) from a hypermutable Pdfl 15 whose size increased
independently twice: once leading to genotypes M and Ml and a second time on the
branch ending in H where it reaches a maximum length of approximately 40 repeats.
The geographical locations of the genotypes shown in Fig. 5-2a are displayed in
Fig. 5-2b. The genotypes show a high degree of admixture in the native range of the
species. A meaningful statistical treatment of the geographical structure is not possible
because the sampling strategy was deliberately biased in Uruguay where most of the
samples and genotypes were found. Despite this, the most remarkable geographical
patterns are the absence of genotype P towards the southwest and the high degree of
admixture east of the Uruguay River. This extreme admixture of genotypes is best
exemplified by the presence of highly divergent genotypes within a single population (see
legend of Fig. 5-2b and Table 5-1).
In contrast to the high diversity and admixture found within the native range, the
samples retrieved from the rest of the world all show the A genotype except for one
accession from South Africa which shows the P genotype, and the samples from Florida,
Greece, and Turkey, which show the M or a related genotype (Fig. 5-2c).
As a first approach to analyze the origin of the recombinant genotypes, their allelic
constitutions were compared to those of the known sexual biotypes in the group. The
estimation of the putative contributions of Virasoro, Vacaria, and P. dilatatum ssp.
flavescens to the recombinants suggests the same pattern when estimated by Structure
Fig. 5-2. Genotypic relationships and geographical distribution of the clonal variants of
pentaploid P. dilatatum A. A haplotype network of the multilocus
microsatellite genotypes. Allele losses are indicated by crosses and size
variations as boxes on the branches assuming either genotype A or P is
ancestral. Box sizes are proportional to allele-size increase (black boxes) or
decrease (white boxes) in dinucleotide repeats. B and C.. Distribution of
clonal variants of typical P. dilatatum B. within its native range and C. outside
its native range. Letters refer to the genotypes shown in Figure. 2A. Circled
numbers in Fig. 2B correspond to locations for which more than one
individual was analyzed. The genotypes found in each population are
indicated in the legend.
0.5 V dilatatum-Vmrasoro
>> > > ; ; U ;U ;U ;U ;O ;O ;O ;O ;O ;O
Genotype C-O -I oI oC
Fig. 5-3. Estimated tetraploid biotype contributions to the pentaploid recombinant P.
dilatatum genotypes. The clusters were generated by Structure based on prior
population assignments and the number of clusters (K) set to 4 to represent
each of the indicated putative tetraploid sources: Virasoro, Vacaria, ssp.
flavescens, and a fourth sample inferred from the alleles present in the most
common genotypes of typical P. dilatatum not shared with Virasoro. The
simulation resulted in no admixture estimated for the tetraploid biotypes each
of which was completely assigned to its own cluster (not shown).
Table 5-3. Proportion of shared allele distances (Dps) of the recombinant pentaploids of P.
dilatatum to the tetraploid biotypes of the Dilatata group. The distance of each
recombinant to the nearest biotype is indicated in bold italics.
Ge e Pm P. dilatatum ssp. P. P. dilatatum P. dilatatum
fGenotype P .da flavescens urvillel Vacaria Virasoro
RecAl 0.90 0.94 0.85 0.95 0.35
RecA2 0.95 0.89 0.82 0.95 0.50
RecA3 0.90 0.92 0.86 0.95 0.35
RecB1 0.78 0.51 0.79 0.80 0.80
RecB2 0.83 0.48 0.79 0.86 0.75
RecB3 0.85 0.55 0.78 0.86 0.80
RecB4 0.78 0.55 0.78 0.80 0.85
RecC1 0.85 0.86 0.78 0.85 0.65
RecC3 0.85 0.86 0.77 0.80 0.65
RecC4 0.80 0.89 0.83 0.80 0.60
RecC5 0.85 0.86 0.78 0.86 0.65
RecC6 0.80 0.84 0.71 0.78 0.55
RecC7 0.64 0.78 0.80 0.69 0.83
(Fig. 5-3) or represented by Dps (Table 5-3). Three putative groups of recombinants are
apparent in both analyses. Recombinant group A shows greater similarity to Virasoro
than the typical clone, recombinant group B shows a clear contribution from P. dilatatum
ssp.flavescens, and recombinant group C is more heterogeneous and appears to share
more alleles with Virasoro than the typical clone. The multilocus genotypes shown in
Fig. 5-4 are in complete agreement with this. In agreement with the estimated allele
admixture proportions shown in Fig. 5-3, the genotypes in groups A and B can be almost
fully accounted for by a cross between typical clones A or P and the individuals of
Virasoro and P. dilatatum ssp.flavescens included in Fig. 5-4.
.-- m Pdfl20b
mm C m
0 0 ;; ;30<< 0 ;0 ;0 ;0 ; ;; ; ; 0 ; ;0 0
Fig. 5-4. Multilocus microsatellite genotypes for the recombinant genotypes of pentaploid
P. dilatatum. The nearest genotype from the typical clone (A or P) is shown
next to each recombinant group. Two individuals of P. dilatatum ssp.
flavescens and Virasoro are shown next to each of the clusters they appear to
be related to for comparison. The highly variable locus Pdfll5 is not shown.
i- 1 1 i 1 1i -
Furthermore, recombinants of group B are found in the area of co-occurrence of
pentaploids and P. dilatatum ssp. flavescens, and the recombinants of group A are found
either in the area of distribution of Virasoro or near it (Fig. 5-5).
The putative recombinants of group C present a different situation. First, a clear
contribution of an existing tetraploid to their genotypes was not found. Second, their
geographical distribution does not overlap with that of any currently known tetraploid,
except for genotype C5, and third, they show allele sizes that are beyond the range so far
detected in the tetraploids analyzed (longer allele of Pdfll5 in Cl, 2 and 7; longer allele
of Pdfl8 in C1-5; shorter allele ofPdfllO in Cl, 2 and 5) (Table 5-2). Genotype C5, in
spite of its geographical location, does not show a clear contribution of Vacaria (see Fig.
5-4 and Table 5-2).
The "Typical" Clone
One of the difficulties encountered in this data set was the discrimination of the
extent of the variability caused by mutation from that due to hybridization. Although it
was evident from the data that a single common genotype comprises the majority of the
accessions analyzed, hybridization to a homozygous source could be confounded with a
high degree of mutagenic activity. The genotypic matrix contained 22 different alleles
because the common genotype is heterozygous at all the loci assigned to the I or J
genomes. If this clone crossed to any unrelated genotype, a maximum of 11 allele-size
differences would be expected. However, the two tetraploids with which hybridization
was hypothesized (Virasoro and P. dilatatum ssp.flavescens) share a number of alleles
with pentaploid P. dilatatum; hybridization to either of them would then produce fewer
than the maximum of 11 differences. With a low number of hybrids as detected here,
sexual events would not necessarily cause conflict in the data matrix and could maintain a
hierarchical structure in spite of recombination. More systematic character compatibility
approaches to detecting recombination in apomicts rely on the generation of this conflict
(Mes 1998) and only provide an approximation to the quantification of sexual
reproduction (Mes et al. 2002). In spite of the above, our case by case analysis has
probably resulted in a very realistic distinction of clonal variants and recombinants.
In well-studied apomictic plant systems like Taraxacum, apomicts tend to behave
as good overseas colonizers (van Dijk 2003); however, unlike P. dilatatum, native ranges
usually contain an assemblage of independently generated clones regardless of how
widespread they may be (van der Hulst et al. 2000). Regardless of the detection of a
certain number of recombinants, most of pentaploid P. dilatatum comprises a single clone
with its somatic variants; moreover, most of the individuals collected outside the native
range of the species have shown the same genotype when analyzed for this set of loci.
The actual extension of this clone around the globe is not known but it has been shown
here to be present on every continent. The current collection is not suitable for estimating
the proportion of typical genotypes within the native range because most of the samples
were collected as individual off-types or as different types within a morphologically
variable population. It is likely then, that recombinants are less frequent than detected
here. It is remarkable that the typical clone was present in all populational samples
containing recombinants except population 10. This population, however, was collected
and annotated as P. pauciciliatum, so it is likely that if typical individuals were present at
the site, they may have been avoided.