Local differentiation and hybridization in wild rose populations in Western Ukraine

Abstract

Three species of wild roses, viz. Rosa spinosissima, R. gallica, and R. canina s.l., sympatrically co-occur in Western Ukraine. R. canina is the morphologically most diverse of them; its different morphotypes (or morphological species in narrow sense) are characterized by a considerable variability in both vegetative and generative characters. This variability, however, does not correlate with ISSR markers diversity. The latter shows an uneven geographical distribution indicating to a probable differentiation of local populations caused by restricted gene flow. On the contrary, R. gallica shows no sign of such a geographical differentiation. However, both species can rarely hybridize, R. gallica being always the pollen parent. Hybrids are morphologically diverse, but not strictly intermediate between the parental plants, usually deviating towards one or the other of them in their characters. Some of the hybrids or putative backcrosses are indistinguishable from R. gallica itself, what may indicate a probable introgression between the two species. R. spinosissima is a rare species in this area, and it is uniform in its characters and seems to be not involved into any hybridization with other species.

Full text

itteilungen des
ärntner otanikzentrums
lagenfurt

ulfenia 17  xxx–xxx
Local diff erentiation and hybridization in wild rose
populations in Western Ukraine
lina  edorova van  chanzer  lexander  agalo
Summary:  ree species of wild roses viz Rosa spinosissima R. gallica and R. canina sl sympatrically
cooccur in estern kraine R. canina is the morphologically most diverse of them its diff erent
morphotypes or morphological species in narrow sense are characterized by a considerable variability
in both vegetative and generative characters  is variability however does not correlate with 
markers diversity  e latter shows an uneven geographical distribution indicating to a probable
diff erentiation of local populations caused by restricted gene fl ow n the contrary R. gallica shows no
sign of such a geographical diff erentiation owever both species can rarely hybridize R. gallica being
always the pollen parent ybrids are morphologically diverse but not strictly intermediate between
the parental plants usually deviating towards one or the other of them in their characters ome of
the hybrids or putative backcrosses are indistinguishable from R. gallica itself what may indicate a
probable introgression between the two species R. spinosissima is a rare species in this area and it is
uniform in its characters and seems to be not involved into any hybridization with other species
Keywords: osaceae Rosa canina Rosa gallica Rosa spinosissima dogroses trnndh 
hybridization population
ild roses are common plants in the south of the ussian lain in general and in estern
kraine in particular  ey grow in diff erent habitats though most of them prefer remnants of
steppe vegetation and forest margins often on slopes of hills or eroded gullies  ey usually form
multispecies stands mostly composed of dogroses belonging to the section aninae  er
though members of other sections like Rosa majalis errm of innamomeae  er or R.
gallica  of allicanae  er sections may be present in such stands as well growing side by
side with the dogroses  e latter are notoriously diffi cult to be taxonomically identifi ed what is
usually attributed to their inherent hybrid nature and the ability to hybridize with each other and
members of other sections      e species of dogroses are commonly
delimited on the basis of a set of correlated morphological characters  eir composition is assumed
to be more or less established for entral and estern urope but opinions about astern
uropean species are much more controversial     
 up to a point of view that next to nothing is known about them  
estern kraine is a speciesrich astern uropean area where from many species of dogroses
were described by  esser in the th century and by  hrshanovsky in the th century see
 e nternational lant ames ndex httpwwwipniorgipniplantnamesearchpagedo
oth in herbaria and in the wild dogrose plants occur which are morphologically intermediate
between the described species commonly regarded as hybrids owever hybridization experiments
with dogroses         
     reveal that the  progeny from interspecifi c crosses as a
rule is not morphologically intermediate between the parents ften these hybrid plants are


         
indistinguishable from the maternal plant or may display hip characters similar to those of the
pollen parent or even possess some novel characters depending on the character combinations of
the parental plants ostly matroclinal inheritance in dogroses is due to a very special breeding
system called balanced heterogamy       ll the dog
roses are allopolyploids with n=x x or x n= owever only two genomes of   or  in a
polyploid dogrose nucleus are pairing and forming bivalents during meiosis ese genomes are
transferred both via haploid pollen and polyploid egg cells e   or  unpairing genomes form
univalents during meiosis and are transferred exclusively via egg cells being lost during meiosis
in pollen mother cells uch an unequal meiosis results in a highly skewed mostly matroclinal
character inheritance in dogroses at the morphological     as well as
the molecular level  et al    
tudies of wild populations of dogroses with the use of  markers reveal that marker
polymorphism is more correlated with geographic origin of specimens than with their taxonomic
identity based on morphology   al  is fact may be due to introgression
or to hybridization resulting in the presence of certain hybrid morphotypes arising de novo in
geographically distinct localities where two or more rose species meet and hybridize imilar
patterns are found in sympatric populations of wild roses in ussia and kraine  
     
ence the aim of the present study is to test if morphological species found in this area can be
confirmed genetically and if any of them may spontaneously hybridize with the others
aterials and methods
Population sampling: amples were collected in une  from five localities in viv ernopil
and vanorankivsk administrative regions of kraine in the geographical province of the
odolian elevation e names and geographical coordinates of the localities together with
the names of species and specimen field numbers of each locality are listed in ppendix  e
localities are situated at different distances from each other  – ~  km  – ~  km
 – ~  km – ~  km  – ~  km  – ~  km  – ~  km so
that the most proximate can be grouped together as follows  –  –  eir relative
geographical positions are shown in ig 
 total of  individuals were sampled for our study e voucher specimens are deposited
in the erbarium of the ain otanical arden in oscow  e determined the
sampled plants with the key in the lora uropae rientalis   according to
the taxonomic treatment of the genus suggested there Rosa gallica  specimens R. canina 
 R. subcanina hrist  R. tomentosa m  R. caryophyllacea ess  R. spinosissima 
 R. glauca ourr  R. parviuscula hrshan  R. porrectidens hrshan  R. corymbifera
orkh  R. podolica hrshan  ne specimen morphologically deviated from a typical
R. gallica and was marked as a putative hybrid wo specimens of dogroses did not correspond
in their characters to any species in the key and remained undetermined ll the specimens were
examined for  morphological characters listed in ab  e choice of the characters for this
study depended largely on the characters traditionally used as diagnostic for the species in our
sample by other authors particularly by  


ocal differentiation and hybridization in wild rose populations in estern kraine
r haracter nit or states of qualitative characters
 ush height cm
 eaflet length mm
 eaflet width mm
 eaflet shape
 – narrow elliptic  – round  – round
with acute tip  – elliptic  – elliptic
with acute tip
 eaf texture  – soft  – medium density
 – leathery
 eaflet hairyness with simple hairs above  – glabrous  – sparse  – dense
 eaflet hairyness with simple hairs underneath  – glabrous  – sparse along nerves
 – sparse on surface  – dense on surface
 eaflet hairyness with glandulous hairs above  – glabrous  – sparse  – dense
eaflet hairyness with glandulous hairs
underneath
 – glabrous  – sparse along nerves
 – sparse on surface  – dense on surface
 etal colour  – white  – pale pink  – pink
 – bright pink to magenta
 landulous hairs on pedicel  – glabrous  – sparse  – dense
 landulous hairs on sepals  – glabrous  – sparse  – dense
 ypanthium length mm
 landulous hairs on hypanthium  – glabrous  – sparse  – dense
 edicel length mm
 rickle shape  – absent  – hooked  – sickle
shaped  – acicles
 epal shape  – entire  – slightly dissected
 – pinnate
 tyle head shape  – loose  – dense
 eaf margin dentations  – simple  – double to complex
glandulous
 eaf margin teeth  – without glands  – with one to few
glands  – with many glands
Table 1. orphological characters used in the study


         
DNA extraction: oung leaves were collected from the same plants as the corresponding
herbarium specimens and dried in silica gel  was extracted with the ucleopin lant 
 extraction kit achereyagel ermany following the manufacturers instructions
Marker selection: e used  nter imple equence epeat markers to study 
polymorphisms within and between species and to detect putative interspecific hybrids since
they proved to be adequate and useful for these purposes in our previous studies of wild roses
           
rimers used for  were synthesized and purified in  by yntol td oscow ussia
ey are listed in ab 
or this study we used the trnndh intergenic spacer of chloroplast  which was shown to
be one of the most variable regions of cp  in different groups of flowering plants  et
al  ough this region was not previously sequenced from any member of the genus Rosa
by other authors it proved to be informative in one of our recent studies in wild rose populations
 et al  e primer formulas were taken from   al  and synthesized
by yntol td oscow ussia
ISSR PCR conditions: or amplification of  markers polymerase chain reactions 
were conducted in  l aliquots containing  l of eadytose  aix   of
each d  m gl   hotstart maraq olymerase and reaction buffer
ialat td oscow ussia  l deionized water  p primer and  – ng of template
 in a  esearch   ngine yad ermal ycler ioad aboratories
 under the following conditions ° –  min pretreatment ° –  s annealing
temperature –  s ° –  s +  s for each cycle  cycles with a final extension step for  min
at ° e annealing temperature for all  primers used was °  control containing
all components except genomic  was included in each set of reactions to prove that no
contamination occurred
  reactions were characterized on  agarose gels in  ×  els were stained with
ethidium bromide and documented digitally using a eloct maging ystem  
Chloroplast DNA PCR conditions and sequencing: e  protocol for trnndh region
amplification slightly differed from that for  markers ° –  min pretreatment
° –  s ° –  s ° –  s  cycles with final  cycles ° for  s ° for
rimer equence
        
a     
     
      
       
       
Table 2.  primers used for 


ocal differentiation and hybridization in wild rose populations in estern kraine
 min  s e lower elongation temperature was used because of the high  content in the
target sequence is improved the work of the polymerase and strongly increased the yield of
the  product oublestranded  products were then purified using centrifugation with
a solution of ammonium acetate in ethanol urified  products were cycle sequenced using
the   igye™ erminator v  kit pplied iosystems and further analyzed on
an    automated sequencer pplied iosystems at the facilities of yntol td
oscow ussia
Analyses of morphological characters: o analyze morphological features we performed rincipal
oordinates nalysis o as implemented in the  v  program  et al 
e same data were analyzed also by cluster analysis using nweighted air roup ethod based
on rithmetic mean  ince both qualitative and quantitative characters were analyzed
together ower similarity measure was used
Analyses of molecular data: e digital image files of  marker electrophoresis results were
analyzed using the ross hecker  software   ach fragment that was
amplified using  primers and visualized as a band in an electrophoretic gel was treated
as a unit character and scored in terms of a binary code  = +– e resulting matrix was
analyzed using o and cluster  analyses as implemented in the  v  program
 et al  accard coefficient was used as the measure of genetic similarity
 sequences were aligned manually using iodit    and manually edited
afterwards e alignment was collapsed into haplotypes using   software  
al 
opulation structure and probability of hybrid origin of particular specimens was analyzed
using ayesian inference with the programs tructure    al  
 al  and ewybrids      e program tructure
assesses probability of subdivision of a sample into  populations basing on calculation of allele
frequencies in each of these hypothetical populations using arkov chain onte arlo method
e used the admixture model with correlated allele frequencies and the no admixture model with
independent allele frequencies for the analyses e first model implies genetic relatedness of the
populations compared ardyeinberg equilibrium and linkage equilibrium for the markers
being analyzed e second model implies low genetic relatedness of the populations compared
e analyses using both models were applied to the whole sample and to dogroses separately
using the admixture model e numbers of =– were tested with  replicates per  and 
million arkov chain onte arlo repetitions
e ewybrids program uses a similar algorithm of analysis but implies a different model trying
to assess probability of subdividing the sample into a priori classes of genotypes e used the
default model of hybridization between two diploid species implying six possible genotype classes
sp – first pure species sp – second pure species  – first generation hybrids  – second
generation hybrids x and x – backcrosses e model implies the following distribution
of allele frequencies between the genotype classes sp –  homozygous of the first parent
speciesdiagnostic markers sp –  homozygous of the second parent speciesdiagnostic
markers  –  heterozygous  –  heterozygous + homozygous of both
parent speciesdiagnostic markers x –  heterozygous  homozygous of the first parent


         
speciesdiagnostic markers x –  heterozygous  homozygous of the second parent
speciesdiagnostic markers ike the model used by the tructure program this model also implies
ardyeinberg equilibrium and linkage equilibrium for the markers being analyzed e
analysis was run for   repetitions in several replicates to assess the stability of the results
esults
rincipal coordinate analysis o gives a fairly well resolved picture of groups corresponding
to all the species determined with the key in lora uropae rientalis ig  e first principal
coordinate explains  of distances the second one explains  s anticipated the most
distant groups correspond to R. spinosissima R. gallica and R. canina while all the other groups
gradually fill the gap between R. canina and R. gallica e specimen of R. porrectidens falls within
the cloud of specimens of R. canina pecimens determined as R. subcanina form a cloud just
Figure 1. esults of rincipal oordinates nalysis of  morphological characters for  specimens ower similarity
measure  – R. canina  – R. subcanina  – R. sp indet  – R. glauca  – R. jundzillii  – R. tomentosa
 – R. porrectidens  – R. corymbifera  – R. caryophyllacea  – R. gallica  – R. parviuscula  – R. spinosissima
 – R. podolica


ocal differentiation and hybridization in wild rose populations in estern kraine
next to that of R. canina pecimens of R. parviuscula and R. tomentosa are placed closer to R.
gallica than to R. canina ll the other specimens are placed more or less between R. gallica and
R. canina somewhat closer to the latter f the two undetermined specimens one groups with
a specimen of R. podolica the other one groups with R. jundzillii
e results of cluster analysis  of morphological data not shown are similar to those of
o ll the morphological species are clearly distinguished by the set of selected characters and
form separate more or less distanced clusters some with medium to high bootstrap support  
replicates n ig  solid line circles surround groups receiving high  – bootstrap support
haplotype
aa--agaaactaaaattctatttct--------tatttctataccattagactatacaattgg-----a
aa--agaaacgaaaattctatttct--------tatttctataccattagactatacaatttg-----a
ac--agaaactaaaattctatttcttatttctatatttctataccattagactatacaattgg-----a
-c--agaaactaaaattctatttcttatttctatatttctataccattagactatacaattgg-----a
aata---------------------------------------------------------ttggtgc-
Table 3. trnndh haplotypes nly variable positions are shown
Figure 2. esults of rincipal oordinates nalysis of   markers for  specimens accard similarity measure
pecies designation is the same as in ig 


         
in cluster analysis ey are R. spinosissima  R. caryophyllacea  and R. porrectidens
 lusters of R. tomentosa  and R. gallica  receive medium support shown by dashed
line circles everal terminal small clusters uniting couples of specimens from the same locality
are rather highly supported too not shown ll the other clusters receive low to no bootstrap
support what is not surprising given the small number of characters in the matrix evertheless
all the clusters resolve the same groups of specimens as they were determined with the key
hloroplast intergene spacer trnndh was partially sequenced from  specimens of total 
in the sample e length of the sequence varied between  to  bp enank accession
numbers are given in parentheses after specimen numbers in ppendix  e sequences were
manually aligned and the alignment length after editing and introducing gaps was  e
alignment was converted into haplotypes using  software   al  gaps were
treated as the th state ive haplotypes are recognized their differences are shown in ab 
and their distribution among the specimens in igs  and  ost of the specimens possess
haplotype  e specimen of R. glauca bears haplotype  which differs from  by the only
deletion in the th position of the alignment ost specimens of R. gallica possess haplotype 
Figure 3. esults of ayesian analysis in tructure  program posterior probabilities of clusterization of  Rosa
specimens into  groups by  marker composition pecimen numbers are shown below the diagram trnndh
chloroplast haplotypes are designated above the diagram


ocal differentiation and hybridization in wild rose populations in estern kraine
which differs from the  haplotype by a – transition in th position of the alignment and
two indels in positions – and – aplotype  is characteristic of the two specimens
of R. tomentosa and is the closest to the haplotype  differing from it in two transitions – 
in th position and – in th position aplotype  is the most distanced from them and
characteristic of the three specimens of R. spinosissima rom the closest haplotype  it differs by
a –  transition in the th position and by four indels one of which is quite large positions
 –
 reactions with six  primers resulted in total  reproducible bands n  specimens
        failed with at least one primer so they were
excluded from further analyses ll the bands appeared to be informative ie no one was present
in all the specimens or in a single specimen o analysis separates specimens of R. spinosissima
R. gallica and R. canina ig  e first principal coordinate explains  of distances the
second one explains  owever specimens determined from their morphology as other
species of sect aninae appear to be either not separable from R. canina itself or are placed in
the scatterplot between R. canina and R. gallica everal R. gallica specimens are deviating toward
R. canina too
ayesian analyses of the total sample with the tructure software reveal that the highest n
value is always achieved for = both under admixture and no admixture models ab 
e diagram in ig  shows posterior probabilities of assigning particular specimens to one
of the groups  for =– or both models used the program divides the sample into two
similar groups under = e first group consists of specimens of R. spinosissima –
and R. gallica – the second group includes all the specimens initially assigned to
R. canina and other species of the aninae section even specimens of R. corymbifera 
R. parviuscula   and R. gallica     show admixed nature as
well as two specimens of R. spinosissima   under the admixture model nder the no
admixture model all the specimens are assigned to one of the two groups with  posterior
probability or the number of groups = the program gives a picture nearly identical to =
odel  n
dmixture allele frequencies correlated
 
 
4 -3020.8
 
 
o admixture allele frequencies independent
 
 
4 -3030.0
 
 
Table 4. e results of the  data analyses in tructure  n values for different 


         
with only the three specimens of R. spinosissima recognized as a separate group e analyses
for = return the same pattern with the major exception that the group representing the sect
aninae species disentangles into two parts however not corresponding to any morphological
species e distribution of trnndh chloroplast haplotypes among the specimens is shown at
the top of the diagram in ig  nder = the specimens of the first group R. spinosissima
bear haplotype  those of the second group most of R. gallica bear haplotype  the specimens
of groups  and  including the admixed ones mostly bear haplotype  wo specimens of
R. tomentosa  and  bearing haplotype  appear to belong to different groups  and 
respectively e specimen of R. glauca with its  haplotype belongs to the third group together
with other haplotype  bearing specimens e analyses for = and further do not change this
pattern though an intermediate group never represented by any specimen belonging to it with
high probability appears under the admixture model
ince ayesian analyses in the program tructure revealed some admixed specimens combining
markers from R. gallica group and Rosa sect aninae group we further analyzed the sample
using another model implemented in the ewybrids program e specimens of R. spinosissima
   were excluded from these analyses e results of the analysis are shown in
ig  with the chloroplast haplotypes indicated for each specimen at the top of the diagram e
program assigns most of the specimens to two parental species corresponding to R. gallica and Rosa
sect aninae respectively nly seven specimens determined as R. corymbifera R. parviuscula
and R. gallica corresponding to the admixed specimens in the tructure analyses appear to be
 hybrids or backcrosses with high posterior probability lso several specimens of the section
aninae group have small posterior probability of being backcrosses ll the plants of the second
parental species R. gallica bear chloroplast haplotype  while all the putative hybrids bear
haplotype  as most of the first parental species specimens e plants bearing haplotypes  and
 do not otherwise differ from the rest of the first parental species specimens
iscussion
t first glance all morphologically determined species with a few exceptions are more or less
clearly distinct from each other on the basis of a set of morphological characters owever
molecular data are in contradiction with the morphology oth chloroplast and  markers
clearly discriminate between species belonging to different sections of the genus ie between
Figure 4. esults of ayesian analysis in ewybrids program posterior probabilities of clusterization of  Rosa
specimens into genotype classes by  marker composition pecimen numbers are shown below the diagram
trnndh chloroplast haplotypes are designated above the diagram sp – first parental species sp – second parental
species  – first generation hybrids  – second generation hybrids x x – backcrosses


ocal differentiation and hybridization in wild rose populations in estern kraine
R. spinosissima R. gallica and dogroses of the section aninae owever they fail to discriminate
between morphological species of the dogroses hloroplast markers discriminate R. tomentosa
from the rest of the aninae section but  markers do not ll the three specimens of
R. tomentosa are assessed with the ewybrids program as pure members of the same parental
species as R. canina with posterior probability  –  e same relates to the single sampled
clone of R. glauca included in our study ts chloroplast trnndh haplotype  differs from the
rest of the aninae haplotypes by a single mononucleotide indel however  markers analyzed
with the ewybrids program bring this specimen to the same parental species as the rest of the
dogroses with posterior probability of  o analyses conducted using the tucture program
discriminate R. tomentosa and R. glauca from the rest of the dogroses e other specimens of
dogroses determined as R. subcanina R. caryophyllacea R. podolica R. jundzillii and one of the
two specimens of R. porrectidens share the same chloroplast haplotype  and they are assigned
to the same parental species by the ewybrids with posterior probabilities higher than 
o ordination of the ewybrids analysis results shows a clear pattern of distribution of
Figure 5. rdination of the ewybrids analysis results with o analysis by   markers accard similarity
measure  – sp0 – the first parental species  – sp1 – the second parental species  – F2 – second generation hybrid
 – sp0 to less than  probability backcross Bx0 to the first parental species  – Bx0 F2 or sp0 with comparable
probabilities  – sp0 or Bx0 with less than  probability of being F2  – F2 or Bx0 with nearly equal probabilities
 – F2 or Bx1 backcross to the second parental species  – sp1 to less than  probability backcross Bx1 to the
second parental species  – R. spinosissima


         
genotype classes ig  pecimens of R. spinosissima not involved in hybridization are equally
distanced from both putative parental species R. gallica and R. canina in a broad sense ybrids
 and x either  or less probably backcrosses to R. gallica are placed in the middle
between the parental species while putative backcrosses are strongly shifted to their corresponding
parental species
t is worth mentioning that all putative hybrids and backcrosses to R. canina x possess
the same chloroplast haplotype  as R. canina while all R. gallica including backcrosses x
possess haplotype  is may be interpreted as R. gallica being exclusively the pollen parent in
hybridizations with the exception of the three specimens having –  posterior probability of
being backcrosses to R. gallica and sharing its  haplotype e specimen  assigned  with
 posterior probability is morphologically determined as R. corymbifera is corresponds well
to mostly matroclinal inheritance of morphological characters in the section aninae owever
the specimens assigned as probable  hybrids or backcrosses x are morphologically either
dwarf shrublets R. parviuscula specimens  and  or rather a typical R. gallica specimens
   e specimen  initially determined as an atypical and probably hybrid
R. gallica falls into this category as well  possible interpretation for this observation is that
morphological type of R. gallica may reappear through segregation from hybrid progeny
e results achieved via ayesian analyses in tructure and ewybrids we should however
treat with major caution n both cases the underlying models assume populations of diploid
outcrossing species e have not studied chromosome numbers of plants in our sample but
basing on published data from adjacent areas we can reasonably assume that R. gallica and
R. spinosissima in our study are tetraploids with normal meiosis while members of the aninae
section may be tetra to hexaploids with heterogamous meiosis    
       owever the results of the analyses under
all the above mentioned models look quite reasonable with the majority of the specimens being
assigned to separate groups with high posterior probabilities oreover at least for R. spinosissima
and R. gallica these results are strongly correlated with the morphological data e suppose that
deviations of the actual data from the models are not that considerable to render these results as
completely erroneous
e contradiction between morphological and based on  markers subdivisions of the dog
rose group in our analyses may rise a suspicion that the result achieved is artefactual in its nature
due to lack of statistical power in the molecular  data set to discriminate between the dog
rose species e one thing that may lead to such a suspicion is that the R. canina sl group is
subdivided by the tructure program into two parts which neither correspond to morphological
species nor coincide with the distribution of chloroplast haplotypes  and  characteristic of
R. glauca and R. tomentosa respectively owever placing these two groups onto a geographical
map of the area ig  shows that they are not arbitrary e first group marked black in the
pie diagrams is concentrated in  and  parts of the area while the specimens of the second
group grey are mostly concentrated in the  e differences in group membership are roughly
proportional to the distance between the localities is observation may serve as an argument in
favor of interpretation of morphological variability of dogroses in the area under consideration
as mostly intraspecific while the variability in  markers reflects the restricted gene flow
between the geographically distanced localities e complex and yet unclear nature of species in


ocal differentiation and hybridization in wild rose populations in estern kraine
Rosa section aninae makes impossible to draw any final conclusion from our data evertheless
at the adopted level of approximation all the dogrose plants behave as a single species is
partly may explain the fact why some morphotypes of the dogroses are common like R. canina
while others are rare like R. caryophyllacea or R. podolica despite they grow together in the same
habitat bsolutely the same pattern is observed in diploid outcrossing populations of R. majalis
    where morphotypes with glabrous R. glabrifolia   ey
or glandulous R. gorinkensis illd leaves may occur in different proportions in populations of
otherwise morphologically typical R. majalis
onclusions
 Rosa canina is the morphologically most diverse species in the area under consideration ts
different morphotypes are characterized by variously pubescent and glandulous leaves peduncles
hypanthia and loose to dense heads of styles as well as the presence of two rare chloroplast trn
ndh haplotypes characteristic of R. tomentosa and R. glauca. ese species however show no
clear differentiation from the rest of the dogrose specimens studied regarding their  marker
compositions ayesian analyses include them in R. canina with high posterior probabilities
us circumscribed R. canina sl shows clear geographical differentiation between its eastern
and western local populations distanced ca  km from each other probably due to restricted
gene flow
 Rosa gallica shows no sign of geographical differentiation in this area regarding its morphology
or  marker composition owever it rarely hybridizes with dogroses being always the
pollen parent e presence of putative backcrosses indicates to a probable introgression between
R. gallica and the dogroses
Figure 6. eographical distribution of dogrose specimens assigned to groups  and  in the tructure analysis =
e size of circles approximately reflects the number of specimens sampled from each locality black sectors – group
 grey sectors – group 


         
 Rosa spinosissima is the rarest species in this area which is quite uniform in its characters t
does not seem to be involved into any hybridization with other species
cknowledgements
e study was financially supported by  grant no a for the first two
authors
eferences
A E. C. & T E. A. (2002):  model based method for identifying species hybrids using
multilocus genetic data – enetics 160 –
B J. B. (2000): ross hecker computer assisted scoring of genetic  data – lant nimal
enome th onf  iego  anuary –  – httpwheatpwusdagovjag
paperspaperindexphtml 
B I. O. (2001): Rosa  – n    ed lora uropae rientalis ol 10  –
– t etersburg ir i semya n ussian
C M., P D. & C K. A. (2000):  a computer program to estimate gene
genealogies – ol col 9 –
DC K., V M K., B P., V B E. & V S J. (2008):
orphological and based differentiation within the taxonomical complex section aninae
subgenus Rosa – nn ot 102  – 
F F. (1951): nfluence of the pollen giver on the production of hips achenes and seeds in the
anina oses – cta orti ergiani 16 –
F D., S M. & P J. (2007): nference of population structure using multilocus
genotype data dominant markers and null alleles – ol col otes 7  –
G A. (1944): e constitution of the Rosa canina complex – ereditas 30  – 
H T. A. (1999): iodit a userfriendly biological sequence alignment editor and analysis program for
indows  – ucl cids ymp er 41  –
H Ø., H D. A. T. & R P. D. (2001):  alaeontological tatistics software package
for education and data analysis – alaeontologia lectronica 4  pp – httppalaeo
electronicaorg_pastissue_htm 
H H. (2003):  Rosa – n eber   rsg ustav egi  llustrierte lora von itteleuropa
nd ed and  eil  permatophyta ngiospermae icotyledones  osaceae
osengewächse – – erlin arey uchverlag
K I. (1969): ytology and some chromosome numbers of zechoslovak oses  – olia geobot
phytotax 4  –
K I. & N A. T. (1974): ytological studies of the genus Rosa with special reference
to the section aninae – ereditas 76 –
K I. (1968): Rosa  – n   et al eds lora uropaea ol 2  – – ambridge
ambridge niversity ress
M J. & P R. (1982): aryological studies in the olish representatives of the genus Rosa 
 – cta biol racoviensia er ot 24  –
M J. & P R. (1984): aryological studies in the olish representatives of the genus Rosa 
 – cta biol racoviensia er ot 26  –
P J. K., S M. & D P. (2000): nference of population structure using multilocus
genotype data – enetics 155  –


ocal differentiation and hybridization in wild rose populations in estern kraine
R C. M. & W V. (2003): ale correlated nonmatroclinal character inheritance in reciprocal
hybrids of Rosa section aninae  er osaceae – lant yst vol 241  –
S I. A. & K N. A. (2010): nterspecific hybridization in wild roses Rosa  sect aninae
 – iol ull 37  – 
S I. A. & V A. V. (2007):  nter imple equence epeat markers reveal natural
intersectional hybridization in wild roses Rosa  sect aninae  er and sect
innamomeae  er – ulfenia 14 –
S I. A., V A. V. & O V. M. (2011):  critical study of wild roses Rosa  from the
homutovskaya steppe reserve – ull osc oc at 116 n press n ussian
S I. A. & V V. N. (2008): ow many species related to Rosa majalis grow in the
uropean part of ussia – ot hurn 93  – n ussian
S J., L E. B., S E. E. & S R. L. (2007): omparison of whole chloroplast genome
sequences to choose noncoding regions for phylogenetic studies in angiosperms the tortoise and
the hare  – mer  ot 94  –
W G. & N H. (2001): kewed distribution of morphological character scores and molecular
markers in three interspecific crosses in Rosa section Caninae – ereditas 134 –
W G., U M. & N H. (1999): orphological and  markers show a highly
skewed distribution in a pair of reciprocal crosses between hemisexual dogrose species Rosa sect
aninae – eor ppl enet 98 –
W V. (2003): onventional taxonomy wild roses – n       
eds ncyclopedia of rose science – – ondon cademic ress
W V. & H F. H. (1997): eproduction and hybridisation in the genus Rosa section
aninae er ehd – ot cta 110 –
W V. & R C. M. (2007): volutionary patterns and processes in the genus Rosa osaceae
and their implications for hostparasite coevolution – l yst vol 266  –
ddresses of the authors
van  chanzer
lina  edorova
erbarium 
sitsin ain otanical arden ussian cademy of ciences
 otanicheskaya str
 oscow
ussia
mail ischanzer@mailru
lexander  agalo
nstitute of cology of the arpathians ational cademy of ciences of kraine
 ozelnitzka str
 viv
kraine


         
Appendix 1. ampled localities and specimens composition enank accession numbers of trnndh sequences
are given in parentheses after specimen numbers
ocality abel eographical
coordinates ample composition
viv region olochiv district near the
hulychi village ature reserve ora
ysoka pine wood margin on a slope
    
   
R. gallica
 
R. podolica

R. subcanina




R. tomentosa
 
viv region olochiv district near the
hervone village ature reserve ora
ysa and ora ypukha mixed wood
margin on a steppe slope
    
   
R. canina
 
R. gallica
 
R. subcanina

 
R. tomentosa

R. sp indet

ernopil region erezhany district near
the utysko village
ature reserve ora olytsya steppe
slopes
    
   
R. canina


 
 
 
 
 

R. corymbifera

 
R. gallica
 
 
 
R. gallica hybr
 
R. jundzillii
 
R. parviuscula
 
 
 
R. porrectidens
 

R. subcanina
 
R. spinosissima
 
 


ocal differentiation and hybridization in wild rose populations in estern kraine
ocality abel eographical
coordinates ample composition
ernopil region idvolochysk district
near the stapie village ature reserve
edobory stony steppe with limestone
outcrops
     – 
    –
R. gallica
 
 
 
 
 
 
 
R. subcanina
 
 
 


 
R. spinosissima
 
R. canina

 


 
 
 
 
 
 
 
 
R. caryophyllacea

 
R. sp indet
 
 
R. tomentosa
 
R. glauca
 
R. porrectidens

vanorankivsk region alych district
 of ovshev village left bank of
urshtyn reservoir at nyla ypa iv
asova ora hill ational ature ark
alytskyy
    
   
R. canina
 
 
 
 
R. gallica
 
 
 
 