Content uploaded by Lorenzo Peruzzi
Author content
All content in this area was uploaded by Lorenzo Peruzzi on Oct 18, 2022
Content may be subject to copyright.
Different chromosome numbers but slight morphological
differentiation and genetic admixture among populations of the
Pulmonaria hirta complex (Boraginaceae)
Lijuan Liu,
1
Giovanni Astuti,
2
Andrea Coppi
3
& Lorenzo Peruzzi
1
1Dipartimento di Biologia, Universitàdi Pisa, Via Derna 1, 56126 Pisa, Italy
2Orto e Museo Botanico, Universitàdi Pisa, Via Ghini 13, 56126 Pisa, Italy
3Dipartimento di Biologia, Universitàdi Firenze, Via Micheli 1, 50121 Firenze, Italy
Address for correspondence: Giovanni Astuti, giovanni.astuti@unipi.it
DOI https://doi.org/10.1002/tax.12721
Abstract Hybridization and introgression have a significant impact on the taxonomically controversial genus Pulmonaria. Within
this genus, the P. hirta complex shows puzzling systematic relationships among P. hirta s.str. (2n= [22, 26] 28), P. apennina
(2n= 22 [26]), and P. vallarsae (2n= 22), showing range overlaps and mixed phenotypes in southern Europe. We carried out mor-
phometric analyses of basal leaves and flower features along with AFLP characterization of 236 plants belonging to 11 populations
within the complex and 1 population of P. officinalis. We also implemented an already available phylogeny with sequences from our
target populations and characterized their karyotype. For all the populations within the complex, we found molecular evidence of a
hybrid origin involving species belonging to different clades (angustifolia and officinalis clades). However, there is a certain morpho-
logical differentiation between some populations (“hirtoid”morph) and others (“vallarsoid”morph), albeit single individuals or entire
populations show intermediate features. According to our results, hybridization and/or backcrossing/introgression have occurred, and
gene flow is currently taking place among these “taxa”. Following the hybridization event(s), we can elaborate three possible evolu-
tionary scenarios: (1) one hybrid “vallarsoid”(2n= 22) species spread across the Italian peninsula, and from this originated the “hir-
toid”morph (2n= 28) through dysploidy; (2) two geographically distinct hybridization events produced both “vallarsoid”and
“hirtoid”morphs; (3) one “hirtoid”alloploid hybrid species originated and backcrossed with P. officinalis generating “vallarsoid”
plants. Under scenarios 1 and 2, the different morphs met again in central Italy, with massive current gene flow. Under scenario 3, “val-
larsoid”plants spread across the Italian peninsula, but further backcrossed with “hirtoid”plants in central Italy, leaving pure lineages
of “vallarsoid”plants only in the extreme north and south of their range. This latter scenario is supported by populations with 2n= 22,
26 chromosomes, having karyotype asymmetry indices intermediate between those of 2n= 16 and 2n= 28 cytotypes. Irrespective of
the evolutionary dynamics, today, a single lineage showing three cytotypes occurs throughout the Italian peninsula, supporting the
circumscription of a single polymorphic species, namely P. hirta.
Keywords AFLP; cpDNA; dysploidy; hybridization; introgression; ITS; morphometry
Supporting Information may be found online in the Supporting Information section at the end of the article.
■INTRODUCTION
Species delimitation can be extremely complicated in
plant genera that experience weak isolation and reproductive
barriers. The genus Pulmonaria L. (lungworts) can be of par-
ticular interest in this context since geographic and reproduc-
tive barriers among species are low (flowering synchrony and
same pollinators shared by species), so that several natural hy-
brids are reported, and artificial ones are easily produced as
well (Sauer, 1975; Bennett, 2003; Meeus & al., 2016). Pulmo-
naria is also well known for its peculiar distylous breeding sys-
tem (Darwin, 1877; Olesen, 1979; Richards & Mitchell, 1990;
Champluvier & Jacquemart, 1999;Meeus&al.,2012a,b)and
its controversial infrageneric systematics (Kerner, 1878; Lacaita,
1927; Sauer, 1975; Bolliger, 1982), in which a rather uniform
morphological asset is contrasted by a striking karyological
variability. The ancestral basic chromosome number of this ge-
nus is likely p= 7 (Bolliger, 1982; Vosa & Pistolesi, 2004), so
that it includes diploids (2n=2x= 14), tetraploids (2n=
4x= 28), and a series of dysploids with 2n= 16, 18, 20,
22, 24, 26, 30, 38 chromosomes (Tarnavschi, 1935;Merxmül-
ler & Grau, 1969; Merxmüller & Sauer, 1972; Sauer, 1975,
1979; Bolliger, 1982).
The whole genus represents thus a wide complex of re-
cently diversified and genetically different, tightly connected
lineages, spread throughout Europe, and encompassing a wide
Article history: Received: 8 Feb 2021 | returned for (first) revision: 14 Sep 2021 | (last) revision received: 26 Jan 2022 | accepted: 31 Jan 2022 | published online: 5
May 2022 | Associate Editor: Hans Peter Comes | © 2022 The Authors.
TAXON published by John Wiley & Sons Ltd on behalf of International Association for Plant Taxonomy.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in
any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
Version of Record 1025
TAXON 71 (5) •October 2022: 1025–1043 Liu & al. •Systematics of the Pulmonaria hirta complex
RESEARCH ARTICLE
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
range of karyological and morphological phenotypes. The first
phylogenetic study of Pulmonaria (Kirchner, 2004) found that
species are difficult to discriminate using molecular markers.
However, later phylogenetic analyses using different plastid
markers and tree reconstruction methods on the nine common-
est lungwort species in western Europe (Meeus & al., 2016)
showed that hybridization and introgression are common and
contributed to the evolution and species diversity in Pulmo-
naria, even though these phenomena did not lead to conver-
gence of lineages, as species recognized on morphological
grounds were equally distinct on genetic grounds. Nevertheless,
in the same study, it was shown that at least the lineage of
P. collina W.Sauer and P. mollis Hornem. was generated by a
hybridization event, anda similar pattern was also found ina re-
cent study on P. helvetica Bolliger (Grünig & al., 2021). The
northern Mediterranean area hosts a few cases of systematic
uncertainties involving hardly distinguishable taxa: the sub-
species of P. longifolia (Bastard) Boreau in the Iberian
peninsula, P. dacica (Simonk.) Simonk. and P. mollis in the
Balkans, and the species complex of P. h i r t a L. (= P. saccharata
Mill.) in Italy. Notably, mountain areas of southern European
peninsulas were claimed as major and deeply structured refu-
gia, where geoclimatic events may interact with plant distribu-
tion ranges favouring both speciation and secondary contacts,
producing thus a complex system of hybrid zones (Nieto-
Feliner, 2011).
The fuzzy assembly of lineages is markedly evident in the
group of Pulmonaria hirta, which includes P. hirtas.str.,endemic
to the Tyrrhenian area of NW Italy and SE France, P. v a ll a r s a e
A.Kern, restricted to a small area in NE Italy, and P. apennina
Cristof. & Puppi, another Italian endemic widely distributed
along the Apennines. The latter taxon is currently treated as a
subspecies of the allopatric P. vallarsae because of striking
morphological similarity (Cecchi & Selvi, 2015; Bartolucci &
al., 2018). All these taxa are putatively different in features of
summer basal leaves, such as shape, maculation, and hair pat-
tern (Kerner, 1878; Bolliger, 1982; Puppi & Cristofolini, 1996).
In addition, they also show differences in chromosome number:
2n=28inP. hirta, although 2n= 22 and 26 cytotypes have also
been reported (Puppi & Cristofolini, 1996); 2n=22in
P. a p e n n i n a , although 2n= 26 has also been reported (Astuti
& al., 2019);2n=22inP. vallarsae (Puppi & Crist ofolini, 1996).
Pulmonaria apennina overlaps with P. hirta in the northern
portion of its range (Tuscan-Aemilian Apennine), where a high
morphological convergence and intermediate chromosome
numbers have been observed (Puppi & Cristofolini, 1996;Vosa
& Pistolesi, 2004). It remains unknown whether these inter-
mediate traits represent a continuous variation within a single
polymorphic species or result from hybridization (and intro-
gression) between different species. Indeed, P. apennina and
P. h i r t a are interfertile (Puppi & Cristofolini, 1996), so that these
parapatric lungworts were also previously considered as sub-
species of the same species (Bernardo & al., 2010). Further-
more, the ranges of all these taxa partially overlap with that
of P. officinalis L., a species with 2n= 16 chromosomes that
is otherwise widely distributed in Europe.
Since the circumscription of the taxa within this com-
plex is uncertain, we aim at testing whether there are mor-
phological, karyological and genetic differences that allow
for safe identification all over their distribution range. To
clarify whether this complex represents an independent line-
age or is interspersed with other Pulmonaria lineages, we
also attempted a phylogenetic reconstruction. We expect to
confirm large morphological and karyological variability in
this group, opening two main alternative systematic scenar-
ios: (1) two/three morphologically and/or karyologicallydis-
tinct species exist, either closely related to each other or have
evolved from different ancestors; (2) a single morphologi-
cally and karyologically highly polymorphic species exists.
Under the f irst scenario, we would expect to find a pattern
of genetic differentiation congruent with morphological and
karyological variation, whereas under the second scenario,
genetic interpopulational differentiation would be expected
to be low or not congruent with morphological and karyolog-
ical variation.
■MATERIALS AND METHODS
Sampling. —We sampled 236 plants belonging to
12 populations, 3 of Pulmonaria hirta,2ofP. apennina,
2ofP. vallarsae, 4 taxonomically critical (mixed hereafter),
and 1 of P. officinalis. Sampling localities are shown in
Fig. 1. For P. hirta, we sampled the two topotypical popu-
lations available (H1 and H2), corresponding to the locality
of the neotype designated for P. picta Rouy (a synonym
of P. hirta; Puppi & Cristofolini, 1996), and to the locality
of the epitype designated for P. hirta (Selvi in Cafferty
& Jarvis, 2004), respectively. While the former occurs in a
geographical area where no other Pulmonaria has been re-
ported, the latter occurs in an area where P. hirta is supposed
to overlap with P. apennina. Moreover, plants in this latter site
display a leaf phenotype not always concordant with that typ-
ical of P. hirta. For this reason, we sampled an additional pop-
ulation of P. hirta (H3), displaying a leaf phenotype typical of
this species. Similarly, also the topotypical population of
P. apennina (A1) occurs in the abovementioned overlapping
area. Hence, we sampled an additional population (A2) of this
taxon at the southern limit of its range, where no other Pulmo-
naria is recorded. For P. vallarsae, we sampled two popula-
tions: one in the topotypical area of Vallarsa (V1) and another
outside this area (V2). Mixed populations were selected among
those where co-occurrence of different chromosome numbers
(2n= 22 and 28; D1, D2) or intermediate chromosome numbers
(2n=22and25,D3;2n= 26, D4) have been reported (Vosa
& Pistolesi, 2004; Astuti & al., 2019,2020). Despite the large
sampling gaps, the choice of populations showing all the
different leaf phenotypes and occurring in different geograph-
ical contexts should provide a good representation of the
overall variability in the entire complex. Plants collected in
spring 2018 and 2019 (see Table 1for details on collection
sites) were cultivated in pots kept outdoor in the Botanical
1026 Version of Record
Liu & al. •Systematics of the Pulmonaria hirta complex TAXON 71 (5) •October 2022: 1025–1043
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Garden of Pisa (WGS84 43.719743N, 10.396097E; 4 m a.s.l.,
mean annual precipitation: 900 mm; mean annual tempera-
ture: 14.2°C), using the same soil and regular watering.
Morphological analysis. —Corolla tube and lobe length,
calyx tube and teeth length were measured on one flower cho-
sen randomly among those open in the inflorescence. Based
on the same flower, the individual was assigned to L (long-
styled) or S (short-styled) flower morph. Morph bias was
measured as (L –S) / (L + S) and varied between –1 (only
S-morphs) and 1 (only L-morphs), 0 representing the isoplethy.
For plants of Pulmonaria vallarsae from Lago di Cei (V2), we
could not evaluate flower features. Table 2summarizes the
morph distribution in the 11 populations available. The Chi-
square test for goodness of fit was used to assess statistically
significant differences.
Leaf features were measured on the same individual in
spring (at the time of collection) and summer (in cultivation)
to compare morphological variation. Shape, maculation, and
hairs were evaluated on one mature, undamaged basal leaf
chosen randomly. The leaf adaxial surface was captured with
a scanner and saved as 24-bit bitmap files.
Leaf shape was calculated by means of elliptic Fourier
analysis, reducing the outline to the coefficients of elliptic
Fourier descriptors (EFDs) (Kuhl & Giardina, 1982) by means
of SHAPE v.1.3 (Iwata & Ukai, 2002). The longest radius was
used as normalization method. To reduce dimensionality, sym-
metric coefficients were subjected to principal component
analysis (PCA). Multivariate analysis of variance (MANOVA)
was carried out on the effective principal components (PCs)
using PAST v.4.03 (Hammer & al., 2001;Hammer,2020)totest
differences among populations.
Maculated and total leaf area were calculated on pictures
by means of ImageJ v.1.47 (Rasband, 1997). The number of
spots was also counted on the same leaf.
Hair types were categorized as follows (Kerner, 1878;
Puppi & Cristofolini, 1996): normal long hairs (or bristles;
>0.5 mm), normal short hairs (or puberes; <0.5 mm), glandular
hairs (stipitate multicellular glands about as long as bristles),
microglands (shortly stipitate, unicellular glands) (suppl. Fig.
S1). Hair density was observed under a stereomicroscope in a
3 × 2 mm area, ca. 1 cm to the left of the middle vein, at the
leaf’s widest point. In a 1.5 × 1 mm sub-area at the top left, the
number of long normal hairs was counted. Comparison among
populations was made through a Kruskal-Wallis test, followed
by a post-hoc Mann-Whitney pairwise test with Bonferroni
correction. The analyses were carried out using R (R Core
Fig. 1. Geographic distribution of
Pulmonaria apennina (dotted),
P. hirta (diagonal lines) and
P. vallarsae (grid). Populations
sampled are indicated by circles for
P. apennina, diamonds for P. hirta,
triangles for P. vallarsae, squares
for P. officinalis and plus for mixed
populations.
Version of Record 1027
TAXON 71 (5) •October 2022: 1025–1043 Liu & al. •Systematics of the Pulmonaria hirta complex
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Team, 2020). Plots were generated with the R package ggplot2
v.3.3.5 (Wickham, 2010).
We also performed a linear discriminant analysis (LDA)
on the overall dataset, merging data of shape (the first three
PCs obtained from the EFD analysis, and the length-to-width
ratio), maculation, and hairs. First, we performed LDA without
mixed populations, a priori grouping the other according to:
(a) a taxonomic hypothesis of four different species, Pulmonaria
hirta (H1, H2, H3), P. apennina (A1, A2), P. v a l l a r s a e (V1, V2),
and P. officinalis (OV); (b) a taxonomic hypothesis in which
P. apennina and P. v a l l a r s a e belong to the same species. Then,
we included the mixed populations considering them either
(c) as a distinct group or, based on their morphological resem-
blance, (d) assigning D1, D3, and D4 to P. apennina +P.vallarsae
Table 1. Sampling localities of the populations of the Pulmonaria hirta complex and P. officinalis investigated.
Species Region, municipality (province), locality (code)
No. sampled
individuals Herbarium vouchers Coordinates
P. apennina Emilia-Romagna, Casalecchio di Reno (Bologna),
Parco Talon*(A1)
20 PI021283–021302 44.47278N,
11.28416E
P. apennina Calabria, San Fili (Cosenza), Foresta Luta (A2) 20 PI021323–021342 39.34085N,
16.08722E
P. hirta Tuscany, Santa Maria a Monte (Pisa), Valle Lupitana*
(H1, type locality of the heterotypic synonym P. p i c t a
Rouy)
20 PI021343–021357
PI036046–036050
43.71388N,
10.67218E
P. hirta Tuscany, Poppi (Arezzo), Camaldoli*(H2) 19 PI021358–021376 43.80902N,
11.82518E
P. hirta Emilia-Romagna, Grizzana Morandi (Bologna),
Favari (H3)
20 PI021377–021396 44.22385N,
11.09556E
Mixed Tuscany, Sambuca Pistoiese (Pistoia), Molino del
Pallone (D1)
20 PI034194–034202
PI035974–035984
44.100658N,
10.962573E
Mixed Emilia-Romagna, Bologna, Monte Paderno (D2) 17 PI034223–034228
PI035985–035995
44.452272N,
11.320769E
Mixed Tuscany, San Godenzo (Firenze), Passo del
Muraglione (D3)
20
**
PI034203–034214
PI035996–036002
43.935302N,
11.658088E
Mixed Abruzzo, Rocca di Mezzo (L’Aquila) (D4) 20 PI021303–021322 42.2125N,
13.51305E
P. vallarsae Trentino-Alto Adige, Vallarsa (Trento), Pian delle
Fugazze (V1)
20 PI034215–034216
PI034218–034219
PI034230–034234
PI036003–036013
45.760208N,
11.171882E
P. vallarsae Trentino-Alto Adige, Villa Lagarina (Trento), Lago
di Cei (V2)
20
**
PI034220–034222
PI036014–036029
45.96022N,
11.04163E
P. officinalis Friuli Venezia Giulia, Prepotto (Udine), Castelmonte
(OV)
20 PI034229
PI034235-034237
PI036030–036045
46.092828N,
13.516041E
*Topotypical population. ** Only 19 herbarium vouchers are available for the population.
Table 2. The Chi-square test for goodness of fit was used to assess flower morph bias.
Population code
A1 A2 H1 H2 H3 D1 D2 D3 D4 V1 OV
L-morphs 11 18 7 7 11 9 13 10 10 11 8
S-morphs 9 2 9 7 964610911
d.f. 1 1 1 1 1111111
χ
2
0.2 12.8 0.25 0 0.2 0.6 4.765 1 0 0.2 0.474
P-value 0.655 <0.001 0.617 1 0.655 0.439 0.029 0.317 1 0.655 0.491
d.f., degress of freedom. Significant deviation from isoplethy is indicated in bold.
1028 Version of Record
Liu & al. •Systematics of the Pulmonaria hirta complex TAXON 71 (5) •October 2022: 1025–1043
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
and D2 to P. hirta. We also tested the hypotheses that mixed
populations altogether belong either to (e) P. apennina +
P. vallarsae or to (f ) P. hirta. Lastly, based on AFLP results
(see further), which see D4 somehow distinct from D1, D2
and D3, we tested the hypothesis that D4 belongs to P. apen-
nina +P. vallarsae and D1, D2, and D3 to P. hirta (g) and vice
versa (h). The different grouping criteria are summarized in
suppl. Table S1. A jackknifed a priori correct classification
was calculated for each analysis. LDA was carried out by means
of PAST v.4.03.
Karyotype asymmetry analysis. —We chose the best
metaphasic plate available for each population, already ob-
tained from our previous analyses (Astuti & al., 2019,2020)
(Appendix 1). On these plates, we built monoploid idiograms
and calculated karyotype asymmetry indices using Karyo-
Type v.3.0 software (Altinordu & al., 2016). We used CV
CL
(Coefficient of Variation of Chromosome Length) and M
CA
(Mean Centromeric Asymmetry) for interchromosomal and
intrachromosomal asymmetry, respectively (see Peruzzi &
Eroğlu, 2013). CV
CL
measures the variation among chromo-
some length within a given karyotype (Paszko, 2006): karyo-
types showing a large variation of chromosome length are
asymmetric (high CV
CL
), while karyotypes showing chromo-
somes of the same length are perfectly symmetric (CV
CL
= 0).
M
CA
expresses the contribution to inner asymmetry given by
each chromosome, where a metacentric chromosome is per-
fectly symmetric, while a telocentric one is perfectly asymmetric
(Levitsky, 1931). In the calculation of M
CA
, the centromeric
asymmetry of a single chromosome is expressed as (l –s) /
(l + s) × 100, where l and s are the long and short arms, re-
spectively: karyotypes constituted by only telocentric chro-
mosomes are asymmetric (M
CA
= 100), while karyotypes
constituted by only metacentric chromosomes are perfectly
symmetric (M
CA
=0).
AFLP analysis. —Genomic DNA was extracted from
silica dried leaves using a modified version of the CTAB
protocol (Doyle & Doyle, 1990).
AFLP analysis was attempted on all 236 individuals, but
was successful in only 226 individuals. The quality of AFLP
products was preliminarily tested on 16 samples randomly se-
lected from the 12 populations (ca. 7% of the total dataset);
these preliminary products were used as replicates in the fol-
lowing steps. DNA restriction, ligation, PCR amplification
with selective primers, and selection of amplif ication products
were carried out according to the AFLP protocol (Vos & al.,
1995). For each sample, 200 ng of genomic DNA was used
for restriction digestion with the endonucleases EcoRI and
MseI. Digested DNA was ligated with 10 pmol of double-
strand oligonucleotide adapters for 2 h at 20°C. For amplifica-
tion, 5 μl of 1 : 2-diluted ligation mixture was added to a final
volume of 20 μl of the reaction mixture, which contained
2.5 μl of 10× reaction buffer, 4 μl dNTPs, 0.25 μl Taq DNA po-
lymerase, 10 pmol of (6-carboxyfluorescein)-labelled EcoRI
TAC primer and 10 pmol MseI ATG primer. The PCR pro-
gramme was 13 cycles of 30 s at 94°C, 30 s at 65°C and 1 min
at 72°C, followed by 24 cycles of 30 s at 94°C, 30 s at 56°C
and 1 min at 72°C, with a final step of 2 min at 72°C. Subse-
quently, 5 μl samples of PCR products were preliminarily ex-
amined by agarose gel (1.2% w/v) electrophoresis in TAE
buffer containing ethidium bromide and visualized using a
UV-transilluminator. The visual inspection of PCR products
allowed to better organize the preparation of samples for the
subsequent sizing analysis. The AFLP electropherograms
were obtained by capillary electrophoresis using the Applied
Biosystems 3130xl Genetic Analyzer. The analysis of molecular
profiles was performed with GeneMarker v.1.5 (SoftGenetics
LLC, Pennsylvania, U.S.A.). Following the programme man-
ual, we set up the software to remove stutter peaks within
2.5 bp of each detected allele peak. After running, the data
with the standard size were analysed along with the replicates,
and no mismatch was detected. The presence, absence, and
questionable presence of alleles are shown for each sample
during this step. In case of a complete mismatch of the peaks,
the programme automatically removes the sample to eliminate
errors. AFLP analysis yielded 200 unambiguously scorable
fragments from 547 profiles of the combined selected primers.
Fragment length ranges from 40 to 600 bp. Within-population
genetic variation was assessed in terms of polymorphic loci,
number of private fragments at single locus and expected het-
erozygosity (H
E
). Analysis of molecular variance (AMOVA;
Excoffier & al., 1992) as implemented in Arlequin v.2.000
(Schneider & al., 2000) was used to analyze the partition
of total genetic variation at two different hierarchical levels:
within populations and among populations. Genetic struc-
ture among populations within each species was also investi-
gated using the Bayesian-based clustering as implemented in
STRUCTURE v.2.3.3 (Pritchard & al., 2000). Five runs per
each Kwere performed by setting the number of clusters (K)
from 1 to 12. Each run consisted of a burn-in period of
200,000 steps, followed by 10
6
MCMC (Markov chain Monte
Carlo) replicates, assuming admixture model and correlated
allele frequencies. The optimal number of clusters (K)was
assessed by calculating an ad hoc statistic ΔK(Evanno &
al., 2005) as implemented in STRUCTURE HARVESTER
v.0.6.94 (Earl & von Holdt, 2012). To assign samples to clus-
ters, a membership coefficient q> 0.1 was used, while coeffi-
cients ≤0.1 were discarded. In order to test the hypothesis
that our populations belong to four (Pulmonaria apennina,
P. hirt a ,P. officinalis,P. vallarsae) or two different taxa
(P. h i r t a s.l., P. o f f i c i n a l i s ), the population structure was sur-
veyed also at K=2andK=4.
To test the relationship between morphological and
genetic variation at population level, we obtained pairwise dis-
tance matrices from both datasets: for morphological distance,
the Euclidean method was applied to z-transformed average
values per population of each variable; for genetic distance,
Slatkin linearized pairwise F
ST
values were used. Matrices were
compared using the multiple regression method (MRM)
(Legendre & al., 1994; Lichstein, 2007) as implemented in eco-
dist v.2.0.1 (Goslee & Urban, 2007). The statistical significance
of the coefficient of regression was evaluated with 10,000
permutations.
Version of Record 1029
TAXON 71 (5) •October 2022: 1025–1043 Liu & al. •Systematics of the Pulmonaria hirta complex
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Phylogenetic analyses. —Phylogenetic analyses were
performed on five individuals for each population, except
for two populations (H2, H3) where only four individuals
were used, using the rpl16,trnH-psbA, and rps16 chloroplast
markers, and the nuclear ITS region (Appendix 2). Primers
and temperature programmes used for the amplification
followed Meeus & al. (2016) for rpl16, Sang & al. (1997)
for trnH-psbA, Oxelman & al. (1997) for rps16, and Cheng
& al. (2016) and White & al. (1990) for ITS. Excess salts
and primers were removed from the PCR reactions with the
PCR Purification Kit (Roche, Mannheim, Germany). Auto-
mated DNA sequencing was carried out directly from the pu-
rified PCR products using BigDye Terminator v.2 chemistry
and an ABI310 sequencer (PE-Applied Biosystems, Norwalk,
Connecticut, U.S.A.).
Raw sequences were checked for homology with BLAST
(http://blast.ncbi.nlm.nih.gov/Blast.cgi). Alignments (suppl.
Appendix S1, S2) including sequences from our accessions
together with sequences published by Meeus & al. (2016),
downloaded from NCBI GenBank, were built with MAFFT
v.7 (Katoh & al., 2002) and then manually edited with BioEdit
v.7.0 (Hall, 1999). We avoided the inclusion of sequences
from Meeus & al. (2016) that showed incongruences between
cpDNA and nrDNA phylogenies, except those corresponding
to entire lineages, such as Pulmonaria affinis Jord., P. collina,
and P. mollis. Sequences of Borago officinalis L. and Symphy-
tum asperum Lepech. were chosen as outgroup. Polymorphic
sites were coded with ambiguous marks only when the L/
(L+l) ratio was between 0.50 and 0.55 (peaks of similar
height), where Lis the highest peak in the electropherogram
and lthe lowest one. Gaps were coded as separate characters
according to the modified simplex coding and simple coding
(Simmons & Ochoterena, 2000) for cpDNA and nrDNA,
respectively, using SeqState v.1.4.1 (Müller, 2005). The best-
fit nucleotide substitution model for each chloroplast and nu-
clear dataset was determined using jModelTest v.2.1.4 (Posada,
2008) under the Akaike information criterion (AIC). A Bayesian
inference (BI) using 2 runs of 4 chains (3 heated, 1 cold) from a
starting random tree and 25% of burn-in was conducted with
MrBayes v.3.1.2 (Ronquist & Huelsenbeck, 2003) on ITS
and a combined cpDNA data matrix (combining all chloro-
plast markers) separately. Nucleotide sites and gaps were trea-
ted as different partitions using GTR + Γ+ I and Jukes-Cantor
substitution models, respectively. Calculation was stopped when
the average standard deviation of split frequencies was below
0.01. Trees below the burn-in (25%) were discarded from the
analysis, and a majority-rule consensus cladogram was built
with the remaining trees. In addition, we built phylogenetic
trees using BI adding one population at a time for both com-
bined cpDNA and nrDNA alignments. Phylogenetic trees were
drawn using TreeGraph2 v.2.14 (Stöver & Müller, 2010). A
cpDNA haplotype network reconstruction was made using
the function haploNet available in pegas package v.1.0-1 in R
environment (Paradis & Schliep, 2019), whereas an ITS Neigh-
borNet split network with uncorrected P distances was built
using SplitsTree v.5 (Huson & Bryant, 2006). For the survey
of polymorphic sites in the ITS alignments, we only considered
sites with L/(L+l) between 0.50 and 0.90.
■RESULTS
Morphological analysis. —Following the LDA (Fig. 2)
performed on the dataset of basal leaf characters –combining
features evaluated in both spring and summer leaves –the best
jackknifed correct classification (87.93%) is obtained when
mixed populations are excluded and populations of Pulmo-
naria apennina and P. vallarsae are merged into a single
group (Fig. 2B). However, when mixed populations are in-
cluded, the best jackknifed correct classification (82%) is
obtained by assigning D1, D3, and D4 to P. apennina/P. val-
larsae (“vallarsoid”morph hereafter) and D2 to P. hirta s.str.
(“hirtoid”morph hereafter) (Fig. 2D). Flower features (suppl.
Fig. S2) and leaf maculation (suppl. Fig. S3) show a morpho-
logical continuum and differences among populations are
only found for few of them and/or for few pairwise compari-
sons (not shown). We found a balanced flower morph ratio
in almost all populations, with the exception of A2 and D2
(Table 2; suppl. Fig. S4). Basal leaf shape is highly polymor-
phic within the complex (suppl. Fig. S5), albeit significantly
different between “hirtoid”and “vallarsoid”morphs in both
spring and summer (not shown); nevertheless, significant dif-
ferences are also found between populations within the same
morph (suppl. Tables S2, S3). The number of short and long
hairs is significantly different between morphs in both spring
and summer (not shown), and, also in this case, a few popula-
tions within the same morph are significantly different (suppl.
Fig. S6; suppl. Tables S4, S5). In summary, the “hirtoid”
morph is characterized by lanceolate leaves, with a lamina ex-
tending across a more or less winged petiole and an indumen-
tum showing a higher number of long hairs with respect to
short hairs. On the contrary, the “vallarsoid”morph shows
rounded leaves with a more or less truncate base, and an indu-
mentum characterized by a higher number of short hairs with
respect to long hairs.
Karyotype asymmetry analysis. —The variation of in-
terchromosomal (CV
CL
) and intrachromosomal (M
CA
) asym-
metry indices across populations is shown in Table 3, while
comparisons between values of all populations and monoploid
idiograms for each cytotype (2n= 16, 22, 26, 28) are shown in
Fig. 3. Cytotypes with 2n= 22, 26 show M
CA
values interme-
diate between those shown by cytotypes with 2n= 16 (lower)
and 2n= 28 chromosomes (higher).
AFLP analysis. —Within-population heterozygosity (H
E
)
ranges from 0.168 (H2, OV) to 0.305 (D4), and the number of
polymorphic loci from 89 (D2) to 167 (D4) (suppl. Table S6).
AMOVA (suppl. Table S7) shows that genetic differentiation
among populations is relatively high. However, the greatest
part of the total variation in our sample (65.74%) is due to
intra-population differences. The structure analysis supports
the recognition of seven genetic clusters (Fig. 4A), as evidenced
by ΔKscores (suppl. Fig. S7). Populations of Pulmonaria
1030 Version of Record
Liu & al. •Systematics of the Pulmonaria hirta complex TAXON 71 (5) •October 2022: 1025–1043
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
vallarsae (V1, V2) and P. officinalis (OV) are distinct from all
the others in having a homogeneous profile characterized by a
peculiar genetic group (in green, blue and orange for V1, V2,
and OV, respectively) that are rarely, or not, shared with
any other population. Populations of P. apennina,P. h i r t a
s.str. and those mixed are instead characterized by an ad-
mixture of different genetic groups. Interestingly, all the
mixed populations show a prevailing group (in light blue),
which is absent in P. apennina and P. h i r t a s.str. popula-
tions, and some individuals within the mixed populations,
especially in D1, show the peculiar genetic group of the topoty-
pical P. vallarsae (V1). Considering a lower number of clusters,
such as K= 2 and K= 4, we can see that there is still no corre-
spondence between genetic groups and taxonomy, since popu-
lations of P. vallarsae (V1, V2), P. officinalis (OV) and mixed
populations D1, D2, and D3 share the same genetic group
(Fig. 4B,C, in orange), and other genetic groups are shared
among populations of P. apennina,P. hirta,andmixed popula-
tion D4. Populations of P. a p e n n i n a ,P. h i r t a , and the mixed
population D4 also show admixtures of different genetic groups
(Fig. 4B,C).
Morphological and genetic distances are not related, as
shown by the statistical test (R
2
= 0.030, P= 0.233).
Phylogenetic analyses. —The alignment is 1926 bp
long (rpl16 868 bp, trnH-psbA 246 bp, rps16 812 bp), with
93 gap positions (indel maximum length of 21), for combined
cpDNA, and 675 bp long (indel maximum length of 2), with
12 gap positions, for nrDNA. We found 59 parsimony-infor-
mative sites and 93 singletons for the cpDNA alignment,
and 50 parsimony-informative sites and 118 singletons for the
Fig. 2. Linear discriminant analysis of the whole dataset of basal leaf characters (in spring and in summer) of the Pulmonaria hirta complex and
P. officinalis based on different groupings: A,Nomixed populations; B,P. apennina and P. vallarsae merged together; C,Mixed populations added
as a single distinct group; D, D1, D3 and D4 along with A1, A2, V1 and V2 assigned to “vallarsoid”morph, and D2 along with H1, H2 and H3 to
“hirtoid”morph.
Version of Record 1031
TAXON 71 (5) •October 2022: 1025–1043 Liu & al. •Systematics of the Pulmonaria hirta complex
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
nrDNA alignment. The occurrence of topological differences
between cpDNA and nrDNA cladograms prevented us to use
a single combined dataset.
In the cpDNA tree (Fig. 5), we found that all the acces-
sions of the Pulmonaria hirta complex fall in a clade with
P. affinis,P. angustifolia,P. collina,P. longifolia,P. mollis,
P. montana, and ‘P. saccharata’accessions (angustifolia
clade), except for two populations (A2, V1), which instead fall
in the same clade with P. obscura and P. officinalis (officinalis
clade). Within the angustifolia clade, H1 is sister to the rest of
the accessions that are grouped together with a low support
(PP = 0.60). A strongly supported subclade (PP = 0.99) in-
cludes P. affinis,P. angustifolia,P. collina,P. longifolia,
P. mollis,P. montana,‘P. saccharata’accessions, three mixed
populations (D1, D3, D4), and one individual of H2. In the
cpDNA haplotype network (Fig. 6), 49 haplotypes were de-
tected, differing by 1 to 16 mutation steps. They can be di-
vided into two main clusters: one includes the haplotypes of
the officinalis clade, A2 and V1, the other one includes the
haplotypes of the angustifolia clade and the rest of the
P. hirta complex sequences, and all the accessions from cen-
tral Italy form a unique lineage, separated from V2.
In the ITS tree, the inclusion of sequences of the Pulmo-
naria hirta complex heavily destabilizes the phylogenetic
reconstruction, yielding numerous polytomies and weaken-
ing the support of nodes (Fig. 5). When added one at a time
(suppl. Figs. S8–S11), all the sequences cluster in the angu-
stifolia clade, except for few sequences of H3 and D2. Popu-
lation V2 is sister to P. montana. The ITS split network
(Fig. 7) shows that all the accessions of the P. h i r t a complex
are in between the two extremes of the network, i.e., P. mon-
tana on one side and the officinalis clade on the other side.
Most of the populations within the P. hirta complex fall close
to the centre of the network, where all edges meet. On the
other hand, A2 and V1 diverge distinctly from the centre and
occupy tip edges, whereas other populations occupy an inter-
mediate position linking the centre of the net to tip edge.
There are 18 polymorphic sites (Table 4) in the ITS sequences
of the P. hirta complex showing nucleotide additivity if com-
pared to sequences of angustifolia and officinalis clades. The
number of polymorphic sites varies among sequences from
2 to 16.
■DISCUSSION
Lineages assembly within the Pulmonaria hirta com-
plex. —Morphological features of summer basal leaves have
been considered diagnostic for discriminating species of
the Pulmonaria hirta complex (Merxmüller & Sauer, 1972;
Bolliger, 1982; Puppi & Cristofolini, 1996). However, from
our results, morphological patterns are generally similar in
spring and summer leaves, and we did not detect a homoge-
nizing effect of common garden conditions on population
features.
Generally, a safe recognition of species is nearly impos-
sible in central Italy, and a patchy distribution of morpho-
logical and karyological phenotypes is instead present.
Nevertheless, this morphological pattern may be inter-
preted as either a single lineage in course of differentiation,
or as multiple lineages converging after secondary contacts.
At first glance, phylogenies built using cpDNA and nrDNA
seem to support a multiple origin of taxa within the
complex, as sequences are dispersed over the gene trees
(Fig. 5). However, in the cpDNA network, we can see that
most of the sequences of the P. h i r t a complex form a single
cluster, very close to the rest of the species within the angu-
stifolia clade (Fig. 6). Only three populations, located at the
northern (V1, V2) and southern (A2) extremes of the distri-
bution range of the complex, fall outside this cluster. From
the nrDNA network, however, all the sequences of the
P. h i r t a complex occupy the same position in the middle
of the network (Fig. 7). The topologies shown by networks
are more consistent with a single origin of the P. h i r t a
complex, and the patterns shown by phylogenetic trees
may be due to reticulate evolution, rather than to multiple
lineages branching in different clades, as further discussed
below.
Table 3. Karyotype features of sampled populations of the Pulmonaria
hirta complex and P. officinalis.
Species Population Chromosome number CV
CL
M
CA
P. apennina A1 2n=22
a
19.49 20.07
P. apennina A2 2n=22
a
19.88 17.33
P. hirta H1 2n=28
a
17.72 23.46
P. hirta H2 2n=28
a
19.31 23.66
P. hirta H3 2n=28
a
22.10 26.32
Mixed D1 2n=22
b
14.16 21.05
Mixed D1 2n=28
b
19.21 27.30
Mixed D2 2n=22
b
//
Mixed D2 2n=28
b
13.42 20.01
Mixed D3 2n=22
b
,25
c
15.62 17.35
Mixed D4 2n=26
a
17.95 17.60
P. vallarsae V1 2n=22
b
20.07 18.55
P. vallarsae V2 2n=22
b
23.19 19.82
P. officinalis OV 2n=16
b
15.22 16.75
P. officinalis /2n=16
d
12.42 13.36
P. officinalis /2n=16
d
10.71 14.68
P. officinalis /2n=16
d
14.09 12.29
P. officinalis /2n=16
d
14.96 14.29
P. officinalis /2n=16
d
13.64 11.78
P. officinalis /2n=16
d
17.58 13.41
Chromosome numbers and metaphasic plates taken from:
a
Astuti &
al. (2019);
b
Astuti & al. (2020);
c
Vosa & Pistolesi (2004);
d
Astuti &
al. (2013). CV
CL
= Coefficient of Variation of Chromosome Length;
M
CA
= Mean Centromeric Asymmetry.
1032 Version of Record
Liu & al. •Systematics of the Pulmonaria hirta complex TAXON 71 (5) •October 2022: 1025–1043
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Phenotypic and genetic differentiation. —Our results
clearly show that phenotypic and genetic features in the inves-
tigated Pulmonaria hirta complex populations are not related.
Indeed, taken altogether, our results highlight a certain
degree of morphological variation occurring within the com-
plex (Fig. 2, suppl. Fig. S5), with two extreme morphs grossly
corresponding to the traditional circumscriptions of Pulmo-
naria hirta s.str. and P. vallarsae/P. apennina, that is a “hir-
toid”morph showing more elongate leaves with decurrent
base and few short hairs intermingled between longer hairs,
and a “vallarsoid”morph with more rounded leaves with trun-
cate base and more short hairs mixed to longer hairs. These
two different morphs usually correspond to different chromo-
some numbers and karyotypes (Fig. 3), with the “hirtoid”
morph showing 2n= 28 chromosomes and the “vallarsoid”
one showing 2n= 22, 26 chromosomes. However, the mor-
phological variation between these two extremes is rather con-
tinuous and could attest for some gene flow between the two
morphs. This is also supported by the fact that all the popula-
tions showing intermediate features occur in those areas
where P. h i r t a s.str. and P. apennina are reported to overlap
(Puppi & Cristofolini, 1996; Vosa & Pistolesi, 2004; Cecchi
&Selvi,2015). Accordingly, we expected to find a similar
pattern in AFLP data, reflecting a genetic differentiation be-
tween P. h i r t a s.str. and P. apennina/P. vallarsae, and possi-
bly signals of admixture in all populations, but especially in
Fig. 3. Karyotype asymmetry indi-
ces (CV
CL
,M
CA
) of populations
within the Pulmonaria hirta com-
plex and P. officinalis (see Table 3).
Values for the latter species (violet
squares) were also taken from Astuti
& al. (2013). Idiograms of each
chromosome number are reported
on the right.
Fig. 4. Population structure analysis of 226 accessions of the Pulmo-
naria hirta complex and P. officinalis inferred using STRUCTURE.
The different colours distinguish the genetic groups found for each K.
Version of Record 1033
TAXON 71 (5) •October 2022: 1025–1043 Liu & al. •Systematics of the Pulmonaria hirta complex
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1
A1.03
A1.14
A1.15
A1.18
A1.20
0.99
0.99
D4.05
D4.08
D4.11
D4.16
0.99
A2.03
A2.04
A2.10
A2.11
A2.18
D4.06
D1.02
D1.03
D1.09
D1.11
D1.18
D2.07
D2.16
D3.01
D3.03
D3.04
D3.06
D3.09
H1.06
H1.15
H2.01
H2.04
H2.06
H2.08
H3.12
H3.15
P. saccharata ITA-I
P. saccharata SLL-F
V1.04
V1.07
V1.12
V1.13
V2.11
0.50
V2.07
0.89
V2.14
V2.18
V2.19
1
0.99
P. montana AVI-F
P. montana BF-F
P. montana GR-F
P. montana JAN-F
P. montana VES-F
P. montana RB-G
0.99
0.72
P. angustifolia BRZa-C
P. angustifolia HS-G
P. angustifolia KAR-E
P. angustifolia SS-G
1
P. longifolia BW-GB
P. longifolia FC-GB
P. longifolia FP-F
P. longifolia GY-GB
P. longifolia LG-F
P. longifolia PH-GB
P. longifolia RY-GB
P. angustifolia PAN-E
0.70
V1.02
0.85
H3.20
0.75
D2.08
D2.14
1
H1.08
H1.12
H3.05
0.52
0.58
D2.06
0.94
0.95
P. affinis CF-F
P. affinis ML-F
P. officinalis DOM-B
P. officinalis EUL-G
P. officinalis FD1-B
P. officinalis PV-G
P. officinalis RT-G
P. officinalis WA-G
OV.01
OV.05
OV.06
OV.12
OV.13
P. obscura BH-G
P. obscura BRZ-C
P. obscura G-GB
P. obscura KON-E
P. obscura KRU-E
P. obscura LUKE-E
P. obscura PD-C
P. obscura S-GB
P. obscura TOR-B
P. obscura ULM-G
P. obscura VIL-F
P. obscura VnVo-C
P. collina EN-G
P. collina WH-G
P. mollis ALB-G
P. mollis VOS-F
P. mollis CHB-G
P. mollis CHW-G
1
0.77
Symphytum asperum
Borago officinalis
1
0.60
A1.03
A1.14
A1.15
A1.18
A1.20
D3.04
H3.20
0.99
H2.01
H2.04
H2.08
0.99
V1.02
1
V2.07
V2.11
V2.14
V2.18
V2.19
0.85
0.73
1
D2.06
D2.07
D2.08
D2.14
D2.16
H3.05
0.60
H3.12
H3.15
0.99
1
D4.05
D4.06
D4.08
D4.11
D4.16
0.98
D1.02
D1.03
1
D1.09
D1.11
D1.18
0.99
1
D3.01
D3.06
D3.03
D3.09
H2.06
0.88
0.98
P. affinis CF-F
P. saccharata ITA-I
P. saccharata SLL-F
P. affinis ML-F
0.70
P. angustifolia BRZ-C
P. angustifolia HS-G
P. angustifolia KAR-E
P. angustifolia PA N-E
P. angustifolia SS-G
0.64
P. collina EN-G
P. collina WH-G
P. mollis ALB-G
P. mollis CHB-G
P. mollis CHW-G
P. mollis VOS-F
0.76
P. longifolia BW-GB
P. longifolia FP-F
P. longifolia RY-GB
P. longifolia GY-GB
P. longifolia LG-F
0.94
P. longifolia PH-GB
P. montana JAN-F
P. montana AVI-F
P. montana BF-F
P. montana GR-F
P. montana RB-G
P. montana VES-F
H1.04
H1.06
H1.08
H1.12
H1.15
1
A2.03
1
A2.04
A2.10
A2.11
A2.18
0.98
1
OV.01
OV.05
OV.06
OV.12
OV.13
1
V1.04
V1.07
V1.12
0.95
0.92
P. obscura BH-G
P. obscura BRZ-C
P. obscura G-GB
P. obscura KON-E
P. obscura KRA-E
P. obscura LUKE-E
P. obscura PD-C
P. obscura S-GB
P. obscura TOR-B
P. obscura ULM-G
P. obscura VIL-F
P. obscura VnVo-C
1
P. officinalis DOM-B
P. officinalis EUL-G
P. officinalis FD1-B
P. officinalis PV-C
P. officinalis RT-G
P. officinalis WA-G
1
Symphytum asperum
Borago officinalis
cpDNA
nrDNA
angustifolia clade officinalis clade
Fig. 5. Bayesian inference (BI) trees of Pulmonaria, including accessions published by Meeus & al. (2016), based on a nuclear marker (ITS; left)
and three chloroplast markers (rpl16,trnH-psbA,rps16; right). Values on branches are posterior probabilities. For population codes of the Pulmo-
naria hirta complex, see Table 1and Fig. 1; for population codes of the remaining accessions, see Meeus & al. (2016).
1034 Version of Record
Liu & al. •Systematics of the Pulmonaria hirta complex TAXON 71 (5) •October 2022: 1025–1043
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 6. Network connecting 49 plastid haplotypes, including accessions of the Pulmonaria hirta complex studied here and those published by Meeus
& al. (2016). For population codes of the Pulmonaria hirta complex, see Table 1and Fig. 1.
P. officinalis
P. obscura
P. affinis
P. collina
P. mollis
P. mollis
P. mollis
P. longifolia
P. angustifolia
P. montana
P. montana
P. saccharata COR-F
P. saccharata SLL-F
P. saccharata ITA-I
A1 H2 D1 D2 D3
A2
OV D4
D4
D2
D2 D2
D2
H1
H3
H3
H3
H1
H1
H1
V2 V2
V2
A1
V2
V1
V1
H2
H3
Fig. 7. NeighborNet split network of ITS sequences of all populations of the Pulmonaria hirta complex (enclosed in the dashed line) and accessions
published by Meeus & al. (2016). Population code size in the Pulmonaria hirta complex proportional to the number of haplotypes represented. For
population codes of the Pulmonaria hirta complex, see Table 1and Fig. 1.
Version of Record 1035
TAXON 71 (5) •October 2022: 1025–1043 Liu & al. •Systematics of the Pulmonaria hirta complex
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 4. ITS polymorphic nucleotide sites of populations of the Pulmonaria hirta complex and comparison with Pulmonaria sequences taken from Meeus & al. (2016).
74 77 91 97 106 107 115 139 143 210 211 255 488 489 490 499 622 648
A1 R G G W(A)
1
M R(A)
1
YW YYWWKMYYR S
A2 A K R+ A C A C A T T A T G C C C A G
H1 R K*(G)
2
GWMRYW YYWWKMYYAG
H2 A G G A C(M)
1
A C(Y)
2
W(A)
1
Y°° T(Y)
2
A W K(G)
1
CCYRS
H3 A G G A C A Y W Y°°° T A W
&
K C C(Y)
1
YRS
D1 R G G W(A)
2
M R Y W Y Y W W K M Y Y(C)
1
RS
D2 R G G W M(C)
1
R(A)
2
YW Y°YWW
&
K M(C)
2
YYRS
D3 R(A)
2
G G A M(C)
2
A Y W Y Y(T)
2
W(A)
2
W K(G)
1
M(C)
2
Y(C)
2
Y(C)
1
R(A)
2
G
D4 AK RA C AYW TTA TKC CYRG
V1 R(A)
2
G G A C A Y W Y°°°° T A W
&&
K C C Y R G(S)
1
V2 R G G W M R Y W Y Y W W K M Y Y R S(G)
2
angustifolia clade
P. affinis AG GA C ATT TTA TT C CT GC
P. angustifolia GA GA C ACA TTA TGC CCAG
P. collina AG GA C ACT TTA TT C CTAC
P. longifolia GG GA CACA TTA TGC CCAG
P. mollis A G G A C A Y(C)
4
A(T)
2
(W)
1
TT A TT C CT AC
P. montana G G G T(W)
1
A(C)
1
G C A T C(Y)
1
T T G A(M)
1
T(-)
1
CAG
P. saccharata AG GA CACA TTA TGC CCAG
officinalis clade
P. obscura AG GA C ATT TTA TT C CT GC
P. officinalis AG GA C ATT TTA TT C CTGC
The number of accessions showing the nucleotide code in brackets is indicated in superscript.
*= T peak higher than G peak; + = A peak higher than G peak; ° = one sequence with C peak higher than T peak; °° = two sequences with C peak higher than T peak; °°° = three sequences with
C peak higher than T peak; °°°° = all sequences with C peak higher than T peak;
&
= one sequence with A peak higher than T peak;
&&
= all sequences with A peak higher than T peak.
1036 Version of Record
Liu & al. •Systematics of the Pulmonaria hirta complex TAXON 71 (5) •October 2022: 1025–1043
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
the mixed ones. On the contrary, mixed populations D1, D2,
and D3 show only limited admixture, and involving only ge-
netic groups of P. h irt a s.str. and P. vallarsae (i.e., yellow and
green groups in Fig. 4A), without any trace of P. apennina.
Furthermore, these mixed populations are all characterized
by a peculiar genetic group (in light blue; Fig. 4A), not
shared with any other population. We could interpret this un-
expected pattern as the consequence of retention of alleles
otherwise lost in other populations, a pattern that has been
claimed to involve hybrid zones as reservoir of genetic diversity
(Brennan & al., 2012).
Moreover, Pulmonaria apennina and P. vallarsae are
morphologically very similar, but do not share any genetic
group, whereas P. apennina and P. hirta s.str. do (grey, yellow,
and dark blue groups in Fig. 4A). This is of course consistent
with geographical proximity, albeit geography had a limited
effect on the genetic pattern when considering mixed popu-
lations. Actually, the only mixed population that shares ge-
netic groups (grey and dark blue groups in Fig. 4A) with
P. apennina and P. h i r t a s.str. is D4, which is quite geo-
graphically distant from all the others. On the other hand,
the peculiar group largely occurring in the topotypical pop-
ulation of P. vallarsae (V1) may have been lost in popula-
tions of P. h i r t a s.str. and P. apennina, whereas traces in
mixed populations may be seen as remnants of the past pres-
ence of this genetic group in central Italy. It may even have
originated in central Italy, assuming that plants with 2n=22
chromosomes spread from the Apennines to the Alps as
hypothesized by Bolliger (1982). Despite the smallest dis-
tribution range among these taxa, P. vallarsae is quite vari-
able as shown by molecular markers. Indeed, its two
populations show different genetic groups in the STRUC-
TURE analysis, where V2 is characterized by a peculiar
group not shared with any other population. This differenti-
ation may be congruent with the phylogenetic data, where
V2 is closely related to P. montana in the nrDNA tree and
to other species of the angustifolia clade in the cpDNA net-
work. Although being morphologically and karyologically
differentiated from the P. h i r t a complex, P. officinalis (OV)
shows a strong genetic affinity with many populations of this
complex (V1, D1, D2, D3) at lower Ks (Fig. 4B,C), and this
could be due to hybridization events at different evolutionary
scales, as discussed later.
Finally, populations showing an excess of long-styled
morphs show either low (D2) or high (A2) genetic diversity,
so that, in our case, deviation from isoplethy had a negligible ef-
fect on genetic diversity. On the contrary, Meeus & al. (2012b)
showed that genetic diversity decreased in populations of Pul-
monaria officinalis showing an excess of either short-styled
or long-styled individuals.
Systematic relationships of the Pulmonaria hirta com-
plex. —Considering the relationships with other species within
the genus, we found hybridization signals in our phylogenetic
reconstruction for all the populations of the Pulmonaria hirta
complex, and particularly in ITS polymorphic sites (Table 4).
As cpDNA and nrDNA data may support a single origin of
the P. h i r t a complex, a unique hybridization event may have oc-
curred through crosses between species of the officinalis clade
and species of the angustifolia clade. The in-between position
of the P. h i r t a complex accessions in the split network is similar
to that found by Grünig & al. (2021)–using ddRAD loci –for
P. helvetica, another hybridogenic species of Pulmonaria,orig-
inated from a cross between P. officinalis and a species among
P. collina,P. mollis,andP. montana. Hypothesizing a hybrid or-
igin for the whole P. h i r t a complex lineage, we should assume
that a parental species from the angustifolia clade acted as the
ovule donor and a parental species from the officinalis clade
as the pollen donor, since most of the accessions cluster within
the angustifolia clade incpDNA phylogeny. However, two “val-
larsoid”populations (A2, V1) cluster in the officinalis clade in
the cpDNA. The topological incongruence between cpDNA
and ITS phylogenies shown by these two “vallarsoid”popu-
lations may be due to their distance from the core area of the
complex (central Italy) that prevents them from crossing with
“hirtoid”plants. However, some AFLP genetic groups are
shared between A2 and P. h i r t a s.str. (Fig. 4), probably be-
cause of the connection ensured by the rather continuous
range occupied by the P. h i r t a complex along the Apennines.
Nevertheless, the close phylogenetic relationship of the “val-
larsoid”populations A2 and V1 with species in the officina-
lis clade, and particularly with the only population of
P. officinalis sampled in Italy (OV), may be the result of a
backcross (see further under “Evolutionary hypotheses”)or
simple introgression at population level, even though this
seems more plausible for V1 than for A2, since P. officinalis
populations occur close to V1, while no species in the offici-
nalis clade is reported from southern Italy (Bartolucci & al.,
2018). AFLP supports the close relationship between P. offici-
nalis and V1 at both K= 2 and K= 4, whereas the affinity be-
tween P. officinalis and A2 is found only at K=2.
As discussed above, in the ITS phylogeny (Figs. 5A,7), a
single population (V2) shows a particular affinity with Pulmo-
naria montana Lej., another morphologically polymorphic
species, showing an equally large variation of chromosome
numbers (2n= 22 [20, 24, 25, 26, 27, 28]) and occurring
in Belgium, S Germany, E France, and W Switzerland
(Merxmüller & Sauer, 1972; Bolliger, 1982). Although this
species has never been reported from the Italian peninsula,
there are records very close to the Italian border in the
database of Swiss flora (https://www.infoflora.ch/it/flora/pul
monaria-montana.html#map). The affinity of P. hirta (under
the name P. saccharata) with P. montana in the ITS phylogeny
was already documented by Meeus & al. (2016), who inter-
preted it as an introgressive relationship at population level.
On the other hand, it is possible that the affinity of the
P. hirta complex with P. montana may reside in the putative
hybrid origin of the former, possibly involving the latter as
one of the parents.
In their phylogenetic reconstruction, Meeus & al. (2016)
included a total of three accessions belonging to the Pulmo-
naria hirta complex (under the name P. saccharata). Al-
though nodes are not highly supported, our accessions
Version of Record 1037
TAXON 71 (5) •October 2022: 1025–1043 Liu & al. •Systematics of the Pulmonaria hirta complex
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
cluster close to these ‘P. saccharata’populations in nrDNA
tree, confirming the synonymy of P. saccharata with P. hirta
(Puppi & Cristofolini, 1991; Selvi & Cristofolini in Cafferty
& Jarvis, 2004).
Evolutionary hypotheses. —Because of its widespread
distribution in the northern part of the Italian peninsula and
the relationships found in the STRUCTURE analysis, Pulmo-
naria officinalis may be the best candidate as one of the paren-
tal species involved in the origin of the P. hirta complex. We
can hence elaborate three possible scenarios starting from a
hybridization event between species of the angustifolia clade
and P. officinalis: (1) a unique resulting hybrid species, mor-
phologically “vallarsoid”with 2n= 22 chromosomes, spread
across the Italian peninsula and gave rise to the “hirtoid”
morph (2n= 28) through ascending dysploidy; (2) two geo-
graphically distinct hybridization events produced both “val-
larsoid”(2n= 22) and “hirtoid”(2n= 28) morphs; (3) a
unique “hirtoid”(2n= 28) alloploid hybrid species originated,
which then produced the “vallarsoid”(2n= 22) plants through
descending dysploidy or backcrossing with P. officinalis.In
all scenarios, dysploidy played an important role in producing
new cytotypes. Dysploidy is an important driver for specia-
tion, capable of originating species-rich lineages at macroevo-
lutionary level (Escudero & al., 2014). This role has been
deeply documented in Brassicaceae (e.g., Mandáková&
al., 2015,2018,2020), but chromosome number changes
through fission and fusion have been claimed as a major cause
for species differentiation in Pulmonaria as well (Merxmüller
& Grau, 1969; Sauer, 1975,1987; Sauer & Gruber, 1979;
Bolliger, 1982;Vosa&Pistolesi,2004). Nevertheless, hy-
bridization event(s) between species with different chromo-
some numbers may equally induce speciation through the
production of different cytotypes, as hypothesized here
concerning the origin of the P. h i r t a complex, and as simi-
larly evidenced by Grünig & al. (2021) for the origin of
P. helvetica.
Under the first scenario, “vallarsoid”plants originated
“hirtoid”morphs, maintaining connections in central Italy,
where a massive gene flow caused morphological and molec-
ular convergence, as well as co-occurrence of different chro-
mosome numbers, sometimes within the same population
(e.g., mixed populations). However, ascending dysploidy seems
less common than descending dysploidy in angiosperms (Carta
&al.,2020), and the latter played a major role for speciation in
Nonea Medik. (Selvi & al., 2002), a genus closely related to
Pulmonaria.
Under the second scenario, the two independently origi-
nated morphs established a gene flow in central Italy, where
they co-occur. In this case, the plants with 2n= 28 chromo-
somes would have originated through a cross between an un-
reduced gamete of Pulmonaria officinalis (n= 16) and a
cytotype with 2n= 24 chromosomes, or through a cross be-
tween normal gametes of P. officinalis and unreduced gametes
of a cytotype with 2n= 20 chromosomes. The cross between
unreduced gametes and normally reduced gametes has been
reported by Bolliger (1982)forP. officinalis and P. helvetica.
Nevertheless, the hypothesis of two different hybrid events is
not fully supported by our data, given that polymorphisms in
the ITS are mostly found in the same position of all the se-
quences analyzed.
Under the third scenario, the “vallarsoid”morph origi-
nated through descending dysploidy from “hirtoid”plants or
through backcross of these latter plants with Pulmonaria offi-
cinalis. Once originated, “vallarsoid”plants spread across the
Italian peninsula, maintaining connection (gene flow) with
“hirtoid”plants in central Italy, so that pure “vallarsoid”
morphs remained only at the southern and northern extremes
of the range, where “hirtoid”plants have never been reported
(Cecchi & Selvi, 2015; Bartolucci & al., 2018). The STRUC-
TURE analysis shows that the genetic group dominant in the
populations of P. hirta s.str. (Fig. 4, in yellow) is also found
in almost all the other populations, so that it may represent
the most ancient genetic group among those found. The hy-
pothesis of descending dysploid origin of plants with
2n= 22 chromosomes from plants with 2n= 28 chromosomes
has been already postulated by Bolliger (1982). In addition,
the same author interpreted plants with 2n= 28 chromosomes
from the Apennines as tetraploid relic populations over-
whelmed by dysploid swarms with 2n= 22 chromosomes, that
subsequently spread towards the NW and NE Alps. This sce-
nario may be consistent with the overrepresentation of some
genetic groups prevailing in plants with 2n= 22 (e.g., grey
group in A1; Fig. 4) in other populations with 2n= 26,
28 chromosomes (H3, D4). A possible origin of “vallarsoid”
plants from a backcross of “hirtoid”plants with P. officinalis
is instead supported by phylogenetic data, especially by
cpDNA: the original “hirtoid”lineage inherited its plastid
genome from a species in the angustifolia clade, whereas “val-
larsoid”morphs (A2, V1) inherited it from P. officinalis. All
other “vallarsoid”accessions may have inherited a cpDNA
genome through introgression with “hirtoid”plants (A1, D1,
D3, D4) or with other species of the angustifolia clade (V2).
It is possible that in some cases (D4) introgression produced
a stable intermediate chromosome number (2n= 26, Astuti
& al., 2019). In addition, populations with 2n= 22, 26 chromo-
somes show karyotype asymmetry values intermediate be-
tween those of P. hirta s.str. and P. officinalis, corroborating
the hypothesis that “vallarsoid”morphs originated from cross-
ing events between these latter taxa (Fig. 3).
Irrespective of the exact evolutionary dynamics, today,
a unique lineage showing three cytotypes (2n= 22, 26, 28)
is widely distributed throughout the Italian peninsula. On
taxonomic grounds, we propose to consider these plants as be-
longing to a single polymorphic species, namely Pulmonaria
hirta.
In order to fully understand the possible evolutionary tra-
jectories of this group, further studies on mechanisms of re-
productive isolation are needed. More detailed cytogenetic
analyses may clarify the role of chromosome rearrangements
occurred and their contribution to karyotype differentiation.
Karyotype differences may indeed cause reproductive isola-
tion, but Bolliger (1982) provided evidence of extreme
1038 Version of Record
Liu & al. •Systematics of the Pulmonaria hirta complex TAXON 71 (5) •October 2022: 1025–1043
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
versatility displayed by lungwort species at meiotic stage,
where normalization of pairings in otherwise unbalanced situ-
ations has been found in artificially produced intraspecific
(crossing different cytotypes) and interspecific hybrids.
Description and diagnosis. —In order to clarify the cir-
cumscription of Pulmonaria hirta, we provide here an updated
diagnosis and description. Pulmonaria hirta is distinct from
P. officinalis by never cordate basal leaves, without aculeoli
(conical hairs less than 0.1 mm long) on the upper surface;
also, chromosome number is 2n= 22, 26, 28 and not 2n=
16. It is distinguished from P. montana by basal leaves with
short hairs on the upper surface (not showing only long bris-
tles). It is also distinct from the allopatric, but morphologically
very similar, P. affinis (from central and SW France and
Spain) by short hairs showing a narrower base and by the indu-
mentum being generally softer, albeit these subtle differences
should be investigated more in detail.
Pulmonaria hirta has lanceolate basal leaves with base
decurrent on indistinct petiole (“hirtoid”morph) to ovate-
rounded with truncate-attenuate base distinct from petiole
(“vallarsoid”morph); leaves (including petiole) up to 33 cm
long and 13 cm wide, 1.5–5 times longer than wide; upper sur-
face with evident pale blotches, or with numerous to sparse,
white to greenish, spots, or no spots at all (in “vallarsoid”
morph); trichomes on upper surface consisting of long bristles
mixed with short hairs (up to 40 per mm
2
in summer leaves),
glandular hairs, and sessile microglands; short hairs ranging
from ≤10 per mm
2
(in “hirtoid”morph) to 50 per mm
2
(in “vallarsoid”morph), long hairs 3–4permm
2
up to 12 per
mm
2
(in “hirtoid”morph); inflorescence a terminal scorpioid
cyme; flowers actinomorphic, bisexual, distylous; calyx lobe
2–7 mm, tube 5–20 mm; corolla colour varying from purple,
or blue to shades of pink and red, petals 5, connate, lobe
3–20 mm, tube 2–8mm.
■CONCLUSION
Populations within the Pulmonaria hirta complex show a
variety of morphological and karyological phenotypes with a
certain degree of intrapopulational variability. A continuous
morphological variation between two extreme morphs, here
called “hirtoid”and “vallarsoid”and grossly corresponding
to P. hirta s.str. and P. apennina/P. vallarsae, is highlighted
by our analyses. The chromosome number 2n= 28 is only
found in “hirtoid”plants, which occasionally may also show
2n= 22 cytotypes. On the contrary, “vallarsoid”plants are
usually characterized by 2n= 22 chromosomes, although also
2n= 26 can be found.
Morphological similarities are not paralleled by genetic
data: the reconstruction of genetic structure revealed that mor-
phologically and/or karyologically ambiguous populations,
occurring in areas where the ranges of Pulmonaria hirta
s.str. and P. apennina overlap, show a distinctive genetic group
and only limited admixture signals. Conversely, P. apennina
and P. hirta s.str. populations show extensive admixture
signals, attesting for a massive gene flow between the two
taxa. The two studied populations of P. vallarsae are quite dis-
tinct from each other and from all the other populations in the
complex, without any relationship with P. apennina, which is
otherwise morphologically very similar.
Phylogenetic data suggest a hybrid origin for the whole
complex: a cross between species belonging to the angustifo-
lia clade and officinalis clade may have produced a novel lin-
eage with a novel karyotype. This lineage spread throughout
the Italian peninsula, originating other cytotypes through dys-
ploidy and/or backcrosses with parental species. Introgression
at population level may also have contributed in shaping the
evolution of this complex.
■AUTHOR CONTRIBUTIONS
GA and LP conceived the idea and designed the experiments. LL
and GA conducted the field sampling and collected the plant material.
LL and GA performed the morphometric and karyological analyses.
LL and AC performed the molecular experiments. LL, GA, AC, and
LP analyzed the data and drafted this manuscript. —GA, https://
orcid.org/0000-0001-5790-3516; AC, https://orcid.org/0000-0003-
4760-8403; LP, https://orcid.org/0000-0001-9008-273X
■ACKNOWLEDGEMENTS
The authors gratefully acknowledge the financial support of
“Progetto di Ricerca di Ateneo”(PRA) of the University of Pisa, grant
number PRA_2018_15. This work was also supported by the “Progetto
di Ricerca di Rilevante Interesse Nazionale”(PRIN) “PLAN.T.S. 2.0 –
towards a renaissance of PLANt Taxonomy and Systematics”lead by
the University of Pisa, under the grant number 2017JW4HZK (Principal
Investigator: Lorenzo Peruzzi). The authors wish to thank the staff of
the Orto e Museo Botanico of Pisa: Francesco Roma-Marzio, Marco
D’Antraccoli, Luca Ciampi, Piero Micheletti, Silvia Zublena, and Andrea
Giannotti for helpful assistance in the cultivation of plants. We also thank
Ilaria Colzi, Elisabetta Bianchi, and Cristina Gonnelli for the support in
molecular lab work. We are deeply grateful to Paolo Pupillo, Fabrizio
Bartolucci, Giancarlo Marconi, and Filippo Prosser for their help in field
activities. Open Access Funding provided by Universita degli Studi di
Pisa within the CRUI-CARE Agreement.
■LITERATURE CITED
Altinordu, F., Peruzzi, L., Yu, Y. & He, X. 2016. A tool for the anal-
ysis of chromosomes: KaryoType. Taxon 65: 586–592. https://
doi.org/10.12705/653.9
Astuti, G., Cristofolini, G., Peruzzi, L. & Pupillo, P. 2013. A new
subspecies of Pulmonaria officinalis (Boraginaceae) from the
southern Alps. Phytotaxa 186(3): 148–157. https://doi.org/10.
11646/phytotaxa.186.3.3
Astuti, G., Liu, L. & Peruzzi, L. 2019. Chromosome numbers for the
Italian flora 7. Ital. Bot. 7: 183–187. https://doi.org/10.3897/
italianbotanist.7.36004
Astuti, G., Bedini, G., Ciccarelli, D., Liu, L., Tiburtini, M. &
Peruzzi, L. 2020. Chromosome numbers for the Italian flora 9.
Ital. Bot. 9: 101–110. https://doi.org/10.3897/italianbotanist.
9.54973
Version of Record 1039
TAXON 71 (5) •October 2022: 1025–1043 Liu & al. •Systematics of the Pulmonaria hirta complex
19968175, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tax.12721 by Universita' Di Pisa, Wiley Online Library on [18/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Bartolucci, F., Peruzzi, L., Galasso, G., Albano, A., Alessandrini,
A., Ardenghi, N.M.G., Astuti, G., Bacchetta, G., Ballelli, S.,
Banfi, E., Barberis, G., Bernardo, L., Bouvet, D., Bovio, M.,
Cecchi, L., Di Pietro, R., Domina, G., Fascetti, S., Fenu, G.,
Festi, F., Foggi, B., Gallo, L., Gottschlich, G., Gubellini, L.,
Iamonico, D., Iberite, M., Jiménez-Mejías, P., Lattanzi, E.,
Marchetti, D., Martinetto, E., Masin, R.R., Medagli, P.,
Passalacqua, N.G., Peccenini, S., Pennesi, R., Pierini, B.,
Poldini, L., Prosser, F., Raimondo, F.M., Roma-Marzio, F.,
Rosati, L., Santangelo, A., Scoppola, A., Scortegagna, S.,
Selvaggi, A., Selvi, F., Soldano, A., Stinca, A., Wagensommer,
R.P., Wilhalm, T. & Conti, F. 2018. An updated checklist of
the vascular flora native to Italy. Pl. Biosyst. 152: 179–303.
https://doi.org/10.1080/11263504.2017.1419996
Bennett, M. 2003. Pulmonarias and the Borage family. London: BT
Batsford.
Bernardo, L., Passalacqua, N.G. & Peruzzi, L. 2010. Notulae 1736–
1748 in: Nepi, C., Peccenini, S. & Peruzzi, L (eds.), Notulae alla
Checklist della flora vascolare italiana 10. Inform. Bot. Ital. 42:
529–532.
Bolliger, M. 1982. Die Gattung Pulmonaria in Westeuropa. Vaduz:
J. Kramer.
Brennan,A.C.,Barker,D.,Hiscock,S.J.&Abbott,R.J.2012. Molecular
genetic and quantitative trait divergence associated with recent homo-
ploid hybrid speciation: A study of Senecio squalidus (Asteraceae).
Heredity 108: 87–95. https://doi.org/10.1038/hdy.2011.46
Cafferty, S. & Jarvis, C.E. 2004. Typification of Linnaean plant
names in Boraginaceae. Taxon 53: 799–805. https://doi.org/10.
2307/4135454
Carta, A., Bedini, G. & Peruzzi, L. 2020. A deep dive into the ances-
tral chromosome number and genome size of flowering plants.
New. Phytol. 228: 1097–1106. https://doi.org/10.1111/nph.16668
Cecchi, L. & Selvi, F. 2015. Boraginaceae in Italy –II. Pl. Biosyst. 149:
630–677. https://doi.org/10.1080/11263504.2015.1057261
Champluvier, D. & Jacquemart, A.L. 1999. Quelques observations
sur deux populations gaumaises (Lorraine belge) de Pulmonaria
montana (Boraginaceae). Belg. J. Bot. 132: 13–25.
Cheng, T., Xu, C., Lei, L., Li, C., Zhang, Y. & Zhou, S. 2016. Barcod-
ing the kingdom Plantae: New PCR primers for IT