EVIDENCE FOR HYBRIDIZATION BETWEEN TWO SYMPATRIC VIOLET SPECIES, VIOLA GRAHAMII AND V. HOOKERIANA (VIOLACEAE), IN CENTRAL MEXICO

2006·9.3 MB·English

by  Aurea C Cortés-Palomec

TWO
EVIDENCE FOR BETWEEN
HYBRIDIZATION
SYMPATRIC
AND
VIOLET VIOLA GRAHAMII V HOOKERIANA
SPECIES,
CENTRAL MEXICO
(VIOLACEAE),
IN
Aurea CCortes-Palomec Harvey
1 Ballardjr.
E.
Deportment of Environmental and Plant Biology Department of Environmental and Plant Biology
Ohio University Ohio University
Athens,Ohio4570lU.S.A.
Athens^Ohio
45701,
U.S.A.
[email protected]
[email protected]
ABSTRACT
Viola hooheriana and V.grahamii (Violaceae) are closely related species in section Viola, subsection Kkxicanac
occurring in mountainous habitats across much of central Mexico. In various locations, these species grow sym-
many
and
patrically the occurrence of individuals with intermediate morphological characters has suggested
hybridization between these species. Using a combination of morphological, ecological, and molecular evidence
we
evaluated two mixed populations of V.grahamii and hooker iana mountainssurroundingLal<e Patzcuaro
V. in
in the state of Michoacan to determine the potential of distinct hybrid morphologies, the strength of phenologi-
and among
cal difterences ecological preferences the species and putative hybrid, and to evaluate the level of
gene flow between the two species. Our results indicate that hybridization occurring and that hybrids are
is
morphologically distinct from the parental taxa. Pre-mating isolation weak and has favored extensive hybrid-
is
ization due to largely overlapping blooming times, weak ecological isolation and absence of spatial isolation.
Hybrids are morphologically and genetically closer to Viola grahamii, but do show unique alleles as well.
RESUMEN
Viola hooheriana y V.gyahamii (Violaceae) son dosespeciespertenecientes a la Subseccion Mexi can a c dcntro de
laSeccion Vio/tidelgenero Viola. Estasespeciescrecensimpatricamenteen variasregionesenelcentrodelMcxico
y la presencia de individuos con caracteristicas intermedias ha sugerido la existencia de hibridizacion entre las
Usando una combinacion
especies. de datos morfologicos, ecologicos moleculares evaluamos dos poblaciones
y
V
mixtas de V.grahamii y hooheriana en las montanas cercanas lago de Patzcuaro en estado de Michoacan
al el
en Mexico para determinar la presencia de morfologias caracteristicas a los hibridos, el grado de diferencias
ienologicas y ecologicas entre las especies y el nivel de flujo genico entre ellas. Nuestros resultados indican la
presencia de hibridos que estos son morfologicamente distintos a los padres. Los mecanismos de aislamiento
y
muy
entre las especies parentales no son fuertes y esto ha favorecido la hibridizacion extensa entre ellas debido a
muy
tiempos de floracion similares, poco aislamiento ecologico y la ausencia de separacion espacial entre las
mas
especies. Los hibridos son morfologicamente geneticamente similares a Viola grahamii, pero muestran
y
aleleos que son unicos a ellos.
common
phenomenon
Hybridization a in nature, as evidenced by an estimated 70,000
is
natural interspecific plant hybrids worldwide (Rieseberg &r Ellstrand 1993; Judd et
al.
and between and 37%
1999), the fact that 16 of the plant families reported in different
contain one hybrid taxon some
floras at least (Ellstrand et 1996). In cases, hybridiza-
al.
tion can lead to the creation of hybrid species and/or introgressants between hybrid de-
and
rivatives the parental taxa (Arnold 1992; Rieseberg 1995). The genus Viola
L.
(Violaceae), which comprises about 525-600 species (Clausen 1964; Ballard
et 1999),
al. is
known
well for its taxonomic problems due to hybridization and introgression, as well as
complex patterns of variation in individual traits (Brainerd 1924; Anderson 1954; Rus-
1 Current Address: Centro de Investigaciones en Ecosistemas, Universidad Nacional Autonoma de Mexico, Antigua Carretera
a
Patzcuaro No.8701,CoLEx-Hacienda de San Jose de Huerta,Moreiia,58190 Michoacan, MEXICO.
la
SiDA22{2):m9-1133.2006
1120 BRIT.ORG/SI0A 22(2)
sell 1955; Ballard 1994; Krahulcova cr al. 1996; Gil-ad 1998; Neuffer et al. 1999; Marcussen
&
Borgcn 2000; Caldcron de Rzcdowski 2001; Harniaja 2003).
Analysis of herbarium specimens from central Mexico has indicated that hybrid-
among Mcxicanac
ization hkely occurring species belonging to Viola subsection (Bal-
is
W
mono-
lard 1994). VioJa subsection Mcxicanac Becker (sensu Ballard et al. 1999) is a
phyletic group consisting of eight stolonifcrous or rosette-forming species: VioJa
hcmslcyana Vgrahamii humiJisU.B.&r
G. C'a.\dev6n, Bcnth., VoxyocIo/itisH.E. Ballard,
V.
K.,V.hookcrianaHB.&zK.,VAK'amaniiG.Cdldei'6ny.guatcma}cnsisW
&
Polak. (Ballard Sytsma 2000; Ballard et 1999), and an additional undescribed spe-
al.
have been from
cies (Ballard, per. obs.). Instances of hybridization identified several lo-
members
calities in central Mexico occurring between VioJa grahamii and four other of
V V
the group, VguatcmaJcn^is^ hcmslcyana, hoohcriana and humilis. Occurrences of
V.
these putative hybrids appear to be restricted however to the immediate areas where both
parents occur sympatrically.
m
Viola graha (known in Mexico as "hoja de pasmo" or "pensamiento del cerro") a
is
i
i
much
frequent locally abundant perennial distributed across of Mexico. Specifically,
to
occurs in mountainous regions from northern Mexico south to northern Guatemala,
it
members
and one of the most widely distributed of the subsection Mcxicanac (Fig.
is 1).
thrives in dry to mesic sandy loam under varied forest canopy and along stream banks
It
at elevations of 1950-3600 meters (Ballard 1994; Caldcron de Rzedowski 2001), and often
chasmogamous
forms large mats connected via aboveground stolons. produces flow-
It
begmnmg
ers at the of the rainy season (June) for approximately four weeks and cleisto-
gamous months
lowers several afterward (June-December) until the onset of the dry
for
1
season (Cortes-Palomec 2005).
As previously mentioned, one of the four species with which Viola grahamii appears
V
to hybridize is hoohcriana (known in Mexico as "violeta"). Viola hoohcriana is an infre-
common
quent to locally perennial generally growing in small isolated populations across
northern and Mexico grows mesic loam under mixed deciduous and
central (Fig. in
1). It
coniferous forest canopy at elevations between 1700-2500 meters (Ballard 1994; Caldcron
de Rzedowski 2001). Reproduction occurs via both chasmogamous and cleistogamous flow-
and Vgrahamii
ers follows a similar pattern to that seen in (Cortes-Palomec, pers. obs.).
Viola hoohcriana and Vgrahamii are divergent in several characters; growth habit,
flower and and can be separated without both herbarium
structure, foliage, difficulty in
specimens and living populations (Table (Ballard 1994). The putative hybrids have been
1)
and
identified in areas of overlapping distribution of the tw^o species (Fig. arc recog-
1),
nizable as exhibiting intermediate morphologies (Ballard 1994).
The was and between
goal ol our study to assess the presence extent of hybridization
V
grahamii and hoohcriana in natural populations, and identify intrinsic prc-mating
V.
We
may
isolation mechanisms that be functioning between the two species. exaniined
morphological and phcnological differences between the two species, characterized the
ecological factors governing where individual species and putative hybrids occur, and sur-
veyed genetic patterns to identify gene flow between species in two study sites in central
V
Mexico where Vgrahamii and hoohcriana intermingle extensively.
AND MRTHODS
N'lATHRIAI.S
and Sampling
Sile Eslablishinent
Two
study sites were established in the mountains to the north of Lake Patzcuaro in the
C0RTES-PAL0MECAN0BALLARD,HYBRIDI2ATI0NBETWEENTW0VI0LA SPECIES IN MEXICO 1121
A
grahamii
Viola
hookeriana
Viola
©
x
grahamii hookeriana
V.
Study
sites
Michoacan
Northern
y
h(jA.D\sXr\but\onoiViolagrahamii,V.hook€nana,dn6V,grahamii hookeriana acrossM^
from Ballard (unpublished data).
municipality of Quiroga, Michoacan, Mexico (Fig. The sites were established in early
L).
June 2000 and observations were made weekly over a period of five weeks during
tlie
summer rainy season when chasmogamous flower production was most extensive. The
two specific sites were chosen due to the presence of abundant populations of the two
species and their putative hybrids as identified through a survey of the area conducted
W]
earlier in the spring. The first site (Site A) [19°4r57" N; 10P32'27" was situated on the
town The
lower southern slopes of Mt. Zirate, above the of Santa Fe de la Laguna. second
km
W]
[W42' was Laguna
site (Site B) N; 10P35' located 8 to the west of Santa Fe de la
(Km
48 on the road from Quiroga Zacapu (MX-15)). While both were located in
to sites
superficially similar environments, they differed in the level of disturbance and abun-
dance of violet species. Viola grahamii was abundant in both but areas of visually
sites,
pure hookeriana separated from grahamii were found only in Site A.
V. V.
m
In Site A, four 20 long transects were established. Two transects were arbitrarily placed
A-TIVG V
through the middle of a visually "pure" population of the parental species in a
(
A-T4VH V
grahamii population and in a hookeriana population). Two additional transects
(A-T2HY A-T3HY)
and were placed through areas where putative hybrid plants inter-
mingled with the parental taxa. Ten quadrats (0.5 m^) were randomly established along
each transect with position determined using a random number table. In Site B the same
was
procedure followed except that only three transects were positioned resulting in a to-
V
30 quadrats (B-TIVG grahamii, B-TSHY mostly hookeriana and mixture
tal of for of V. a
B-T2HY
of Vgrahamii and hybrids, and consisting mostly of putative hybrids).
1122 BRIT.ORG/S10A 22(2)
Table Comparison of morphological characters used to distinguish between Viola grahamii and V.hookeriana
1.
(on the basis of Ballard 1994). The characters of hybrids are not included since they are not consistent
(i.e.,
show combinations making them
hybrids varying of traits, difficult to characterize).
Character Viola grahamii Viola hookeriana
above ground underground
Stolon position
Stolon nodes present absent
adnation adnate
Stipule free or free
Pubescence on petioles and present [abundant] absent
abaxial surface of leaves
shape ovate-oblong broadly ovate reniform
Leaf to elliptic to
3-12 5-21
Length of pedicels (cm)
Calyx pubesence glabrous
ciliate
sometimes
Corolla color white white, purplisl'
m m
1950-3600 700-2500
Elevation
1
For transects representing "pure" parental taxa at each study three plants that
site,
One
were not connected via stolons were randomly selected in each the 10 quadrats,
o(
DNA
leaf was removed from each of the three plants and dried in silica gel for later ex-
A
second each was and
traction. leaf, the largest of individual, collected labeled to corre-
DNA
spond the sample. This was pressed and used subsequent morphological
to leaf ior
analysis. The same methodology was used for plants in the putative hybrid/mixed-taxon
sampled
transects, except that 5 individuals w^ere per quadrat.
Phenology and
Pollinator Visitation
number
In each ol the 70 quadrats (40 from Site A, 30 from Site B), the of individuals
present per quadrat and the number of individuals with open chasmogamous flowers
were recorded weekly during the five weeks of field observations to interpret phenology.
chasmogamous some
Tlie percentage o( (lowering individuals, those bearing lowers at
I
pomt durmg the study, was calculated for each visit for the two species and for the puta-
among
tive hybrids to allow for a comparison of phenology the parental taxa and puta-
hybrids both Quadrats with more than one taxon were recorded separately
tive at sites.
A
each taxon and aiialyzcd as separate observations. Kolmogorov-Smirnov two-sample
for
test was performed in NCSS (Hintze 1999) to compare phonological patterns among taxa
each two
at of the
sites.
Pollmator visitation was assessed during weekly A, wduch was the only
visits to Site
A
site to have clearly differentiated populations of V.grahamW and hoo]ier\ana. total of
V.
40 hours of direct observations of the flowering individuals were recorded with special
made among
emphasis to observe instances of pollmator moveiuent between or differ-
ing taxa. The hours oi pollmator observation were equally divided between two daily
when
periods relating to potential insect activity based on the times previous observa-
would more between
tions in the area suggested the potential pollinators be active, 8:00
and am, and between and pm.
11:00 12:00 3:00
Ecological Characterization
To determine grow modally whether
the taxa in different microhabitats, infer ecologi-
if
cal isolation might reduce gene flow between the species and the putative hybrids, and
understand the local distribution of putative hybrids relative to the parents, an ecologi-
K
was
cal characterization of microhabitat in the 70 cjuadrats conducted. Soil N, con-
P,
CORTES-PALOMEC AND BALLARD, HYBRIDIZATION BETWEEN TWO VIOLA SPECIES MEXICO 1123
IN
pH
centrations and were evaluated using a LaMotte™ Combination Soil Test Kit (LaMotte,
pH K
Chestertown, MD). Values were given in exact units for while values for N, and
P,
were determined via colorimetric procedure from which concentration could be
a test
ppm ppm ppm
estimated within range, 5-75 N, 5-100 H and 50-200 Due
a set for for for K.
to the inability of obtaining exact concentration values for N, and K, these values were
P,
range standardized and analyzed rank form.
in
was method
Percent moisture recorded using the gravimetric described by Hadley
soil
and Levin (1967), with the modification that soil mass was determined after one week of
means canopy was measured
air drying. Light availability as a direct of inferring closure
ModehC
using spherical densiometer (Vora Quadrats with more than one taxon
a 1988).
were recorded separately each taxon and analyzed as separate observations. Data were
for
standardized and an analysis of variance was performed in NCSS to test for differences
among
and
hookeriana, V.grahamii putative hybrids.
V.
Morphological Analysis
From
the pressed leaves, measurements were taken of the 0°, 30"", 60'', 90"", 120°, 150"" and
180"" radii, following the procedure outlined by Ballard and Wujek (1994). The center of
was
the leaf considered to be the point along the midrib opposite the widest part of the
leaf blade. Presence or absence of leaf pubescence on the abaxial surface of the leaf, a
characteristic considered diagnostic separating hookcriana (glabrous) and
for
\^ V.
grahamu (pubescent), was also recorded. Measurements were standardized and a dis-
criminant analysis (Manly 1994) was performed, both including and excluding visually
among
identified putative hybrids to determine overlap in leaf characters the two taxa
m
NCSS
and putative hybrids. The analysis was performed using an automatic variable
selection procedure. Measurements 30"" and 120"" were excluded from the analysis due
at
to colinearity of the data. The canonical scores generated from these analyses were used
(ANOVA) LSD
in an analysis of variance with a Fisher's pairwise comparison analysis
(Zar
1996).
Genetic Analysis
among and
To the presence of genetic differences the parental taxa identify potential
test
gene flow within the putative hybrids, biparentally inherited nuclear markers, inter-
simple sequence repeats (ISSRs) were used. ISSR data have proven to be highly effective
among
in detecting hybridization and/or gene flow closely related species, and to test
&
the hybrid speciation hypothesis (Wolfe 1998a, Wolfe Randle 2001; Archibald
et al. b;
&
et 2004; James Abbott 2005). Additionally these markers are useful in studies of
al.
among some
differentiation closely related species for they are able to utilize of the vari-
ability present at microsatellitc loci without the need to develop species-specific
microsatellite primers, an advantage when working with species which have not been
well-studied genetically
DN
A
Genomic was extracted from the silica gel -dried leaves using a Wizard Genomic
&
DN A
purification kit (Promega, Madison, WI). Five different ISSR primers (McCauley
which would
Ballard 2002) were initially screened to identify primers produce consis-
among Two
and polymorphic bands primers were Wolfe
tent scorable the taxa. selected:
#99B [(CA)6GG] and HB#15 [(GTG)3GC]. These were used amplification of ISSR prod-
for
DNA,
genomic
ucts in replicated 25 reactions consisting of diluted 19 auto-
\xg \A~^
\x\ 1 jil
MgCh
mM,
PCR
claved distilled water, 2.4 ^1 lOx Buffer (Gibco BRL), 2 (50 Gibco BRL),
|il
dNTP mix mM, BSA (Bovine Serum Albumin, 4
2 of (10 Fisher), 0.5 of |ig jil"\ Fisher),
|al |il
1124 BRIT.ORG/S1DA 22(2)
U
The
primer and Taq polymerase Gibco BRL). polymerase chain
0.25 0.25 of (5
id |il
).il''
,
reaction was performed in a Stratagenc RoboCyclcr 96 with liot-top (Stratagene Inc, La
CA) and programmed 2 min. 94°C; 40 (primer Wolfe #99b) 44 (primer IB
JoUa. for at
I
mm
mm.
#15) X 30 94°C 45 44"C, 30 72"C; 20 72°C.
sec. at sec. at 1 sec. at at
PCR x TBE with
products were electrophoresed in a 1.3% agarose gel in 0.5 buffer
ethidmm
flanking 250 bp ladders (Gibco BRL). Gels were stained with a solution of bro-
UV
mide X TBE 20 minutes and imaged under Gel images were
in 0.5 buffer for light.
Com-
analyzed with BioMax ID miage analysis software (Version Eastman Kodak
2.0.3,
pany, Rochester, NY) to identify and size fragments- Fragments were scored as present (1)
or absent with fragments comigrating at identical rates (± 10 base pairs, the general
(0),
limit of resolution for agarose gels) considered equivalent. Statistics concerning fragment
fragment number, polymorphic and
occurrence, including total distribution across taxa,
number among
fragments, and fragments shared taxa per were calculated
fixed of site
TFPGA
using (Miller 1997).
1.3
among
methods and
Analysis used three discrete to evaluate relationships similarity
UPGMA
the taxa and putative hybrids. cluster analysis and Principal Coordinates Analy-
NTSYS
(PCoA) were performed both within and between the two using
ses sites ver. 2.02j
An (AMOVA)
was
(Applied Analysis Molecular Variance addition-
Biostatistics of
Inc.).
WINAMOVA
among
performed hierarchical fashion species and using L55
ally in a sites
(Excoffier 1993). All analyses were performed according to the methods described in
McCauley and
Ballard (2002).
RESULTS
Phenology and Pollinator Visitation
The blooming
parental species expressed divergent but non-significant tendencies in time;
m
A
bloomed whereas B bloomed
in fact, in Site hooJzcriana first, Site later (Fig. 2).
it
V^.
among
The Kolmogorov-Smirnov two-sample shov/ed no significant differences the
test
blooming times of the hybrids and the parental taxa in either site values - in some
(p 1
No
any
pairwisc comparisons) (Fig. pollinators were observed visiting violets at time
2).
during
this study.
Ecology
All three taxa inhabited modally different microhabitat conditions influenced by light
and where were between
certain nutrients, in all cases there significant diflerences the
ANOVA
grew The
taxa, the hybrids in intermediate conditions to the parental species. of
K
and
the ecological factors revealed significant differences in light (p = 0.02), P (p - 0.04)
among
= 0.03) hooJzcriana, V.grahamii and the putative hybrids; moisture (p- 0.32),
(p V.
N
pLI = 0.86) and (p- 0.62) did not differ significantly Viola hookeriana grew in higher
(p
K
concentrations lower concentrations and more shaded environments than
of of
V.
P,
grahamii
(Table
2).
Morphology
Leaf
Two hundred and eighty eight individuals were classified into three groups correspond-
ing to the two parental taxa and the putative hybrid individuals (Fig. The discrimi-
3).
GLM ANOVA
nant analysis indicated three marginally distinct groups with the
(Fig.
4),
and the Fishers LSD Multiple Comparison analysis indicating that they were well-sup-
and
ported significant < 0.05) groups.
(p
CORTES-PALOMEC AND BALLARD, HYBRIDIZATION BETWEEN TWO VIOLA SPECIES MEXICO 1125
IN
70
A hookeriana AT4
V-
Site
^^^
AT2
hybrids
AT3
60 hybrids
R^^^
ATI
grahamii
V.
AT1 AT2 AT3
50
cu
3
AT2
.77
ATS
1 1
AT4 22
,77 .77
E
40
H
CO
a;
30 -
o
I
o
20 -
o
t
I
I
10 -
^
I
a
ZA
'A ti zi
June 25 July 4 July 6 July 12
Visits
70
V
hookeriana BT1
SiteB
BT2
hybids
^^^^^^^
Vgrahamii BJ3
F^^^^
60 -
BT1 BT2
CO
i
BT2
.873
50
03
BT3 873
^A 1
E
40
-
133
<D 30 - ^
O
o
20
o
10
:^
^
i
i
z
<: F 21
ti
I
June 25 July 4 July 6 July 12 July 18
Visits
x
Fig. 2. Histogram of phenology for flowering individuals in two sites of Viola grahamii, V. hookeriana and !/. grahamii hookeriana.
among
Abundance of flowering individuals summarized by transects. P-values depicting strength of similarities or differences taxa
is
= =
using a Kolmogorov-Smirnov two-sample test are shown in the box, VH l/.//0(?fcer/flna,VG=l<5rfl/?(7m//,HY hybrids Jotali
als sampled are as follows (number of reproductive individuals): Site A: VH(AT1)=24(12); HY(AT2)=151(76); HY(AT3)=232(134);
=142
VH
VG(AT4)=146(77).Site B V6=35(19); HY=95(55); (33).
1126 BRIT.ORG/SIDA
22(2)
among
Means and SEfor selected ecological parameters V.hookeriana iyi\)y.grahomii {VG) and hybrids
Tml>li 2.
among
(HY). Superscript letters indicate a significant difference for a particular parameter the taxa. Moisture
and light are expressed as percents. N and K values are along a scale of to 7 (l--^very low, 7= high).
P,
1
Taxon Moisture Light pH P N K
VH
26(±0.1) 16(±0.4)' 6.40(±0.]4) 5.32(±0.49)' 1.28(±0.10) 6.44(±0.16)
VG
24(±I,1) 29(±0,3)" 6.51 (±0.1 3,82(±0.37)" 1.14(±0.09) 7.0(±0.13)'^
2)
HY
25(±0,1) 29(±0,3) 6.43(±0.]0) 4.40(±0.33) 1.18{±0.08) (±0.1
6.81
2)
Data
Genetic
A
total of 242 Viola individuals were examined genetically with ISSRs: 33 Viola
m m
A
A
hoohcriaua and Vgrahamii and
in Site 10 Site B, 65 individuals in Site 35 Site
B,
A
and 48 putative hybrid individuals in Site and 42 in Site B. InSitc A, 70 ISSR fragments
The
were resolved vs. 53 in Site B. majority of these were identical between the two sites,
resulting in a total of 75 unique and scorable fragments for the study sites together. The
relative frequency of individual fragments ranged from less than 10% of individuals (rare
100%
to infrec[uent) to (ubic[uitous). Distribution of fragments across taxa (Table showed
3)
some
that were specific to one of the parental taxa five fragments were species-spe-
(e.g.,
m
Vgraha and
cific for three were specific for \^ hoohcriana) whereas others were shared
ii,
among
taxa. Most of the shared fragments occurred between Vgrahamii and the puta-
hybrids fragments Nine fragments were
tive (e.g., 17 in Site B). specific to the putative
hybrids.
UPGMA
showed
cluster analysis (Fig. 5) strong separation of the two study All
sites.
A
individuals diverged substantially between Site and Site regardless of taxon. The
B,
PCoA
confu-med PCoA
these discrete site differences (Fig. 6A). Additional analyses were
A V
carried out separately on the two 6B) and B both
sites [Site (Fig. Site (Fig. 6C)l In sites,
grahamii and the putative hybrids showed the most extensive genetic overlap. In Site A,
V
hookcriaua mostly segregated from the other taxa, although some individuals clus-
tered with putative hybrid derivatives, and others w4th Vgrahamii. B 6C)
In Site (Fig.
V
two distinct groups hooker were evident, with some putative hybrids occurring
of ia /u/
within each of them. In all cases, putative hybrid individuals were widely placed and
substantially overlapping in distribution w^ith the parental taxa, togetlicr expressing a
greater level of genetic diversity
AMOVA
The showed most was
(Table 4) that of the variation within morphologi-
cally defined taxa although significant differences were detected between
(69%^), sites
among
and Variation between accounted 27%
taxa. sites for of the total variance. Differ-
among
ences parental taxa and hybrids accounted only 3.72% The
for of the variance.
same pattern was observed when the two were analyzed 94.5%
sites separately. In Site A,
was
of the variance due to variation within the taxa; 95.3% in Site the hybrids were
B. If
was
eliminated, variation slightly reduced within taxa (90.32%o and 89.17%); however, lev-
els of variation within the species themselves remained high.
DISCUSSION
Our show
results conclusively that hybridization occurring between Vgrahamii and
is
V. hoohcria na. Phenetic analysis of leaf shape characters in the species, more exploratory
than exhaustive, showed separation among hoohcriana, Vgrahamii and
the hybrids.
V.
Hybrid individuals formed group from
a distinct the parental taxa suggesting that leaf
CORTES-PALOMEC AND BALLARD, HYBRIDIZATION BETWEEN TWO VIOLA SPECIES MEXICO 1127
IN
Representative leaves of Viola grabam!iV,grahamii\j^ hookeriana, and V.hookeriana.
Fig. 3.
4
-
3
-
2
-
1
CM
-
X
<
-
-1
2
-
-3
-4
-4 -2 2 4
Axis
1
Fig. 4. Results of discriminant analysis of leaf shape from individuals of Viola grahamii (white dots), V. hookeriana (black dots) and V.
x
grabamii hookeriana (gray dots).
1128 BRIT.ORG/SIDA
22(2)
Number
Table of ISSR bands found per site per taxon (V.hookeriano,V.grahomii and hybrids), as well as nunn-
3.
among
ber of bands shared the taxa.
Number
of bands SITE A SITEB
hookeriana
Viola 3
1
grahamii
Viola 5 5
hybrids 3 9
and
V.liookeriana hvbrids 2
1
Vgrohamii
hybrids 17
a\']d 11
individuals 46 20
All
VGA
HYA
VHA
VGB
HYB
VHB
"T"
I" "T"
"I
020
-0.01 0.09 0.30 0.40
Genetic distance
Fig. 5. UPGMA dendrogram depicting relationships among taxa and transects. Similarity compared using the AMOVA-derived PhiST
= = =
distance matrix. (VHA Violo hookeriana in Site A, HYA hybrids in Site A, VGA V. grahamii in Site A; the same format for Site B).
traits are useful in distinguishing the hybrids from the pure species, sniiilar to the find-
ings seen in other Viola (Russell 1954;Jonsell 2000). While these differences appear
ct
al.
may
to follow a clear pattern, our potential discovery of mtrogressant individuals actu-
ally blur the distinction to some degree. Extensive gene flov/ was confirmed, and has
proceeded to the extent that the two sites harbored substantially different genotypic com-
numerous
binations. At both morphologically "pure" parental individuals of the
sites,
two species were demonstrated to be cryptic hybrid derivatives, with a preponderance
of these resembling V.grahamii. High levels of genetic variability within taxa may be the
AMOVA
product long-continued gene The from around 95%
of flow. results suggest that
of the genetic variation is due to variability within each taxon. While this high level of
variation could be partly artifactual, owing to rapid evolution of ISSR primer
sites, is
it