Alpine Botany

, Volume 121, Issue 1, pp 23–35

The stability of Quaternary speciation: a case study in Primula sect. Auricula

Authors

    • Institut für Spezielle Botanik und Botanischer GartenJohannes Gutenberg-Universität Mainz
  • H. Goldner
  • N. Holstein
    • Department Biologie I, Systematische BotanikLudwig-Maximilians-Universität München
  • G. Schorr
    • Institut für Spezielle Botanik und Botanischer GartenJohannes Gutenberg-Universität Mainz
  • L.-B. Zhang
    • Chengdu Institute of BiologyChinese Academy of Sciences
    • Missouri Botanical Garden
Original Paper

DOI: 10.1007/s00035-010-0084-y

Cite this article as:
Kadereit, J.W., Goldner, H., Holstein, N. et al. Alp Botany (2011) 121: 23. doi:10.1007/s00035-010-0084-y

Abstract

Primula sect. Auricula, a group of 25 species distributed in the European Alpine System, has been hypothesised to have diversified in the Quaternary through speciation in geographically isolated glacial refugia. We here examine whether the integrity of species is endangered through hybridisation upon contact in the Holocene. To do this, we (1) critically screened the literature for reported hybrids and supplemented this with our own knowledge of the group, (2) performed an admixture analysis of AFLP variation of two partly sympatric species pairs, P. hirsuta/P. daonensis and P. latifolia/P. marginata, and (3) analysed long-known hybrid populations of P. lutea × P. hirsuta in Wipptal/Austria to identify possible mechanisms of reproductive isolation. The literature survey revealed that populations of the 32 hybrid combinations known have been observed at 63 localities. In the admixture analysis, two admixed individuals per species were found among 524 individuals of P. latifolia and P. marginata, and 21 admixed individuals were found among 234 individuals of P. hirsuta. The analysis of P. lutea × P. hirsuta hybrids revealed that they show reduced pollen and seed fertility, and are limited to soils with intermediate pH values. We conclude that although species of P. sect. Auricula can readily be hybridised experimentally, hybridisation is rare in nature and species are stable. Mechanisms of reproductive isolation include geographical and ecogeographical isolation, ecological hybrid inviability and reduced hybrid fertility.

Keywords

AdmixtureAFLPsEdaphic differentiationHybrid fitnessQuaternary refugiaReproductive isolation

Introduction

An increasingly large number of studies from all parts of the world and from vastly different climates report extensive speciation in the Quaternary (e.g. Hagen and Kadereit 2001; Richardson et al. 2001, 2004; Zhang et al. 2007; Brochmann and Brysting 2008). Although there are many reasons why the absolute dating of molecular phylogenies with a molecular clock approach may be inaccurate (Bell and Donoghue 2005; Ho et al. 2005, 2007; Magallón and Sanderson 2005; Renner 2005; Pulquério and Nichols 2007), the existence of Quaternary speciation can hardly be doubted. Speciation rates in the Quaternary may not have been generally higher than in previous periods (Rull 2008), but it seems quite certain that the Quaternary was not a period of extinction only either (Willis and Niklas 2004). One major cause of speciation in the Quaternary, at least in regions strongly affected by Quaternary climatic oscillations, may have been the very dynamic and oscillating changes of distribution areas and population continuity in response to climatic changes, resulting, e.g. in isolated populations evolving into new species, and in new species of hybrid origin upon contact of previously isolated lineages.

When considering possible explanations for their finding (based on the fossil record) that ‘in mid-latitude regions, the Quaternary appears to have been a time of extinction rather than speciation’, Willis and Niklas (2004) reason (among other explanations) that ‘barriers to gene flow established during the comparatively brief duration of the Quaternary Ice Ages may have subsequently broken down during the intervening interglacials. If so, previously isolated populations may have interbred such that the fossil record gives the appearance of species stasis’. This describes well the problem we set out to investigate here. Will evolutionary products of, e.g. one glacial, namely species which originated in isolated glacial refugia, be erased in the following interglacial? More specifically, we ask whether the integrity of species of Primula sect. Auricula, which have been postulated to have originated in the Quaternary (Zhang et al. 2004) through geographical isolation in refugial areas (Kadereit et al. 2004), is endangered by interspecific hybridisation in the present interglacial.

Extinction through hybridisation, a concept first developed by Harper et al. (1961), has been discussed by, e.g. Arnold (1997) and Levin (2000). Mechanisms of extinction through hybridisation may be genetic assimilation, particularly of rare by more numerous species, and a lower realised reproductive potential of the hybridising species (Arnold 1997). The problem of ‘reverse speciation’ (Seehausen 2006) through hybridisation has recently also been discussed in the context of loss of environmental heterogeneity through human activity (Seehausen 2006; Seehausen et al. 2008). On the other hand, it has been claimed that contemporary hybridisation among species does not cause problems in species delimitation (Rieseberg et al. 2006), and that selection at a limited number of loci can maintain species differences irrespective of substantial levels of interspecific gene flow (Lexer et al. 2006).

Primula sect. Auricula is endemic to various high mountain ranges of Europe and contains 25 species (Zhang and Kadereit 2004), with the highest specific diversity (22 species) found in the Alps. Among these 25 species, range size differs tremendously. Whereas some species are very widespread (e.g. P. auricula, P. hirsuta, P. lutea), others are limited to very narrow areas (e.g. P. albenensis, P. apennina, P. carniolica, P. cottia, P. daonensis, P. deorum, P. palinuri, P. recubariensis, Table 1). The group has been subject to a phylogenetic and biogeographical analysis (Zhang et al. 2004) which showed that the section can be divided into two groups: a ‘western clade’ of 15 species (subsect. Euauricula; Zhang and Kadereit 2004, 2005) mainly distributed in the western part of the Alps, the Apennines, Pyrenees and Cordillera Cantabrica; and an ‘eastern clade’ of ten species (subsect. Cyanopsis; Zhang and Kadereit 2004, 2005) mainly distributed in the eastern part of the Alps, the Dinaric Alps, the Balkans, Carpathians and Sudetan Mountains. The split of sect. Auricula into these two groups was dated, using a molecular clock approach, to approximately 2.4 million years ago, and was hypothesised to have been caused by an early glaciation of the Alps forcing the last common ancestor of the two clades into western and eastern refugia, respectively (Zhang et al. 2004). A later study (Kadereit et al. 2004) used a simulation approach to investigate whether rates of diversification in sect. Auricula are correlated with the climatic oscillations of the Quaternary. It was found that in the ‘western clade’ diversification is negatively correlated with temperature, and this was interpreted to imply speciation in geographically isolated refugial areas in glacial periods. If this should be correct, species which originated in glacial refugia will have re-colonised unglaciated areas in interglacials, including the Holocene, and may have come into contact in areas of sympatry.
Table 1

Geographical and altitudinal distribution, chromosome number, flowering time, ecological requirements, hybrids formed and absolute overlap scores (AOS) of the species in Primula sect. Auricula

Taxon number

Taxon name

Distribution

Chromosome number

Flowering time

Altitude (m)

Soil (substrate)

Number of hybrids (taxon number of second parent)

Absolute overlap score (AOS)

1

P. albenensis

S Alps (N Italy)

?

June

1,150–2,000

Basic (limestone)

0

1

2

P. allionii

Maritime Alps (France and Italy)

66

March–June

650–1,900

Basic (limestone)

1 (17)

2

3

P. apennina

N Apennines

62

May–June

1,400–2,000

Acid (sandstone)

0

1

4

P. auricula

S Alps, Apennines, Hungary, Balkans, SW Carpathians

62

May–July

250–2,900

Basic (limestone, etc.)

8 (5, 6, 8, 12, 21, 22, 23, 24)

9

5

P. carniolica

Julian Alps

62

April–June

500–1,100

Basic (limestone)

1 (4)

2

6

P. clusiana

NE Alps (Austria, Germany)

c. 198

May–July

(600–) 1,700–2,500

Basic (limestone)

2 (4, 18)

3

7

P. cottia

Cottian Alps (SE France, NW Italy)

62

May–July

(400–) 1,600–2,000

Acid (granite)

1 (20)

2

8

P. daonensis

S Austria, N Italy, E Switzerland

62

May–July

1,500–3,000

Acid (granite), rarely basic

5 (4, 12, 15, 18, 22)

6

9

P. deorum

Rila Mts.

66

June–July

1,900–2,900

Acid

0

1

10

P. glaucescens

S Alps (N Italy)

66

April–June

450–2,400

Basic (limestone)

1 (22)

2

11

P. glutinosa

Alps (Austria, Italy, Switzerland), Bosnia and Hercegovina

66

May–June

1,500–3,100

Acid (granite, schist, slate, syenite sand)

1 (18)

2

12

P. hirsuta

C and W Alps, C Pyrenees

62

May–July

(500–) 1,400–2,800

Acid (granite, schist, limestone/slate)

6 (4, 8, 13, 15, 16, 18)

7

13

P. integrifolia

C Alps (Austria, N Italy, C and E Switzerland), Pyrenees, Cantabria

66

June–July

1,500–3,000

Acid (granite, slate), rarely basic (calcareous slate)

3 (12, 15, 16)

4

14

P. kitaibeliana

W Balkans (Croatia, Bosnia and Hercegovina)

66

April–June

350–2,300

Basic (limestone)

0

1

15

P. latifolia

S and W/C Alps, E Pyrenees

62

May–July

650–3,050

Acid (granite), rarely basic

6 (8, 12, 13, 16, 17, 20)

7

16

P. lutea

N Alps, Jura Mts., Tatras

62

May–July

250–2,900

Basic (limestone, etc.)

4 (4, 12, 13, 15)

5

17

P. marginata

SW Alps (Maritime and Cottian Alps) (France and Italy)

62, c. 126

June–July

(500–) 1,000–3,000

Basic (limestone), rarely acid (slate, granite)

2 (2, 15)

3

18

P. minima

E Alps, Pirin/Rila Mts., S Carpathians, Tatras, Sudety Mts.

66

May–June

1,200–3,000

Acid (granite, schist)

8 (6, 8, 11, 12, 22, 23, 24, 25)

9

19

P. palinuri

Calabria (Italy)

66

February–April

50–300

Acid (sandstone)

0

1

20

P. pedemontana

Cottian and Graian Alps (France and Italy), Cantabria

62

May–July

1,400–3,000

Acid (granite)

2 (7, 15)

3

21

P. recubariensis

SE Carega Massif, N Italy (among Trento, Vicenza and Verona)

?

May–June

1,400–2,030

Basic (triassic dolomite)

1 (4)

2

22

P. spectabilis

S Alps (N Italy)

66

May–June

500–2,500

Basic (limestone)

4 (4, 8, 10, 18)

5

23

P. tyrolensis

SE Dolomites (Italy)

66

June–July

1,000–2,300

Basic (limestone)

3 (4, 18, 25)

4

24

P. villosa

S Austria, NE Italy, Slovenia (Karawanken Alps)

62

April–July

1,400–2,900

Acid (gneiss, schist, slate, trachyt, granite), rarely basic

2 (4, 18)

3

25

P. wulfeniana

SE Alps (Gailtal and Karawanken Alps, NE Italy), S Carpathians

66

June–July

1,700–2,100

Basic (limestone)

2 (18, 23)

3

Most data were extracted from Zhang and Kadereit (2004)

We here will examine whether the integrity of species of P. sect. Auricula is endangered through hybridisation upon contact in the Holocene. We will investigate the level of interspecific hybridisation in three different ways. First, we will examine the degree of geographical overlap and edaphic and altitudinal differentiation among species, and will critically screen the literature for the report of hybrids and will supplement this information with our own experience of the group from herbarium and field work. Second, we will present an admixture analysis of Amplified Fragment Length Polymorphism (AFLP) variation of two at least partly sympatric species pairs of the ‘western clade’ which had been identified as sister species by Zhang et al. (2004), i.e. P. hirsuta/P. daonensis and P. latifolia/P. marginata. Third, we will analyse hybrid populations of P. lutea × P. hirsuta (both ‘western clade’) to investigate whether these two species are reproductively isolated and what the mechanisms of reproductive isolation, if observed, might be.

Materials and methods

Geographical overlap, edaphic and altitudinal distribution, hybridisation and hybrid frequency

The geographical and altitudinal distribution, ecological requirements, chromosome number and flowering time of all species of P. sect. Auricula are compiled in Table 1. In this table, each species also is assigned an absolute overlap score (AOS). The AOS is based on the number of species with which a given species hybridises in nature. When a species produces no hybrids with other species, it is assigned an AOS of 1. Primula auricula received an AOS of 9 because it hybridises with 8 other species in nature (Jacquin 1778; Reichenbach 1830, 1855; Leybold 1855; Huter 1873; Kerner von Marilaun 1875; Pax 1889; Richards 1993; Prosser and Scrotegagna 1998). Based on the literature (Table 1), field observations, a herbarium survey, the close phylogenetic relationships among species (Zhang et al. 2004), and the fact that a number of hybrids have been raised in cultivation (Richards 1993), we here assume that species of sect. Auricula hybridise easily with one another when they encounter. Accordingly, the number of natural hybrid combinations a species forms in our opinion reflects the degree of geographical and ecological overlap with other species, i.e. strict sympatry.

To quantify the frequency of hybrids, each natural hybrid combination is assigned a hybrid frequency score (HFS) from 1 to 10 based on the number of different localities at a small geographical scale (e.g. individual mountains or valleys) where hybrids were reported. The most common hybrid, P. × floerkeana (P. glutinosa × P. minima), which occurs in the east Alps (Austria) and South Tyrol (Italy; Reichenbach 1855; Pax 1889; Lüdi 1927), received an HFS of 10 (Table 2).
Table 2

Natural hybrids in Primula sect. Auricula and their distribution and frequency scores

Number

Parents

Hybrid name

Distribution

Confirmed in field (F) or herbarium (H)

Hybrid frequency score

References

1

P. allionii × P. marginata

P. × meridionalis A. J. Richards

Maritime Alps, France/Italy

 

1

Richards (1993)

2

P. auricula × P. carniolica

P. × venusta Host.

Julian Alps

H

2

Reichenbach (1830, 1855); Pax (1889); Pax and Knuth (1905); Lüdi (1927)

3

P. auricula × P. clusiana

P. × lempergii F. Buxb.

NE Alps, Austria

 

1

Richards (1993)

4

P. auricula × P. daonensis

P. × discolor Leyb.

South Tyrol, Italy

H

3

Leybold (1855); Reichenbach (1855); Kerner von Marilaun (1875); Pax (1889); Lüdi (1927)

5

P. auricula × P. hirsuta

P. × pubescens Jacq.

Tyrol, Austria

H

5

Jacquin (1778); Reichenbach (1830, 1855); Widmer (1891); Pax (1889); Pax and Knuth (1905); Lüdi (1927)

6

P. auricula × P. lutea

P. × obristii Stein

Tyrol, Austria

 

1

Pax (1889)

7

P. auricula × P. recubariensis

P. × vallarsae Prosser and Scortegagna

SE Carega Massif, N Italy

 

1

Prosser and Scrotegagna (1998)

8

P. auricula × P. spectabilis

P. × weldeniana Reichenbach

Mt. Baldo, Italy

 

1

Reichenbach (1830)

9

P. auricula × P. tyrolensis

P. × obovata Huter

Mt. Cavallo, Belluno, Italy

 

1

Huter (1873); Pax (1889); Pax and Knuth (1905); Lüdi (1927)

10

P. auricula × P. villosa

P. × goebelii Kerner

Carinth and Styria, Austria

 

2

Kerner von Marilaun (1875); Lüdi (1927)

11

P. clusiana × P. minima

P. × intermedia Portenschlag.

Austrian and Eisenerz Limestone Alps

H

1

Schott, (1852); Reichenbach (1855); Neilreich (1859); Lüdi (1927)

12

P. cottia × P. pedemontana

P. × boni-auxilii A. Kreß

Mt. Orsiera, Italy

H

2

Kreß (1973, 1981)

13

P. daonensis × P. hirsuta

P. × seriana Widmer (P. × plantae Bruegger)

Val Seriano and Val di Sole, Italy and Val Muranza, E Switzerland

H

2

Pax (1889); Pax and Knuth (1905); Kreß (1973); Richards (1993); Prosser (2000); Prosser and Scrotegagna (1998)

14

P. daonensis × P. latifolia

P. × kolbiana Widmer

Bergamo, Italy

 

1

Pax (1889); Lüdi (1927)

15

P. daonensis × P. minima

P. × pumila Kerner

South Tyrol, Italy

 

1

Kerner von Marilaun (1875); Pax (1889); Lüdi (1927)

16

P. daonensis × P. spectabilis

P. × judicariensis Beyer

Tyrol, Austria

 

1

Lüdi (1927)

17

P. glaucescens × P. spectabilis

P. × caruelii Porta

Brescia, Italy

 

1

Pax and Knuth (1905); Lüdi (1927)

18

P. glutinosa × P. minima

P. × floerkeana Schrad.

East Alps, Austria; South Tyrol, Italy

F, H

10

Reichenbach (1855); Pax (1889); Lüdi (1927)

19

P. hirsuta × P. integrifolia

P. × heerii Bruegger

Voralberg, Austria and Graubünden, Switzerland

 

4

Pax (1889); Lüdi (1927)

20

P. hirsuta × P. latifolia

P. × berninae Kerner

Engadin, Switzerland

F, H

1

Kerner von Marilaun (1875); Pax (1889); Lüdi (1927)

21

P. hirsuta × P. lutea

P. × helvetica Donn

Voralberg, Austria and Switzerland

H

2

Pax and Knuth (1905); Lüdi (1927)

22

P. hirsuta × P. minima

P. × forsteri Stein

Central Tyrol, Austria

H

2

Pax (1889); Lüdi (1927)

23

P. integrifolia × P. latifolia

P. × muretiana Kerner

Bernina, Engadin and Graubünden, Switzerland

F, H

3

Kerner von Marilaun (1875); Pax (1889); Lüdi (1927)

24

P. integrifolia × P. lutea

P. × escheri Bruegger

St. Gallen, Switzerland

 

1

Pax (1889); Pax and Knuth (1905); Lüdi (1927)

25

P. latifolia × P. lutea

P. × rhaetica Reichenbach

Mt. Javernaz, Switzerland

 

1

Reichenbach (1855); Pax (1889)

26

P. latifolia × P. marginata

P. × crucis Bowles

Maritimes, France/Italy

 

2

Richards (1993)

27

P. latifolia × P. pedemontana

P. × bowlesiana Farrer

Mt. Cenis, Italy

 

1

Richards (1993)

28

P. minima × P. spectabilis

P. × facchinii Schott

Tyrol, Austria

H

4

Schott (1852); Reichenbach (1855); Kerner von Marilaun (1875); Lüdi (1927)

29

P. minima × P. tyrolensis

P. × juribella Suenderm.

Mt. Castellazzo, Paneveggio, Italy

 

2

Pax (1889); Widmer, (1891); Lüdi (1927)

30

P. minima × P. villosa

P. × truncta Lehm.

Styria, Austria

 

1

Lehmann (1817); Kerner von Marilaun (1875); Pax (1889); Widmer, (1891); Lüdi (1927)

31

P. minima × P. wulfeniana

P. × vochinensis Gusm.

Carinth, Austria

 

1

Pax (1889); Widmer, (1891); Lüdi (1927)

32

P. tyrolensis × P. wulfeniana

P. × venzoi Huter

Cimolais, Italy

 

1

Kerner von Marilaun (1875); Pax (1889); Widmer, (1891); Lüdi (1927), Richards (1993)

Hybrid populations P. lutea × P. hirsuta

The analysis of hybrid populations between P. lutea × P. hirsuta was conducted in the Wipptal (North Tyrol/Austria) area where such populations have long been known and documented (Ernst and Moser 1925). Primula lutea in this area is mostly referred to as P. auricula. For the nomenclature of this species see Zhang and Kadereit (2005). Plants and soil samples from altogether eight localities (Fig. 1) in this area were analysed. Hybrids were encountered in three of these localities (I, V and VI). In I and VI both parental species were found, but only P. lutea in V.
https://static-content.springer.com/image/art%3A10.1007%2Fs00035-010-0084-y/MediaObjects/35_2010_84_Fig1_HTML.gif
Fig. 1

Distribution of populations sampled of P. hirsuta, P. daonensis, P. latifolia and P. marginata used in the admixture analysis, and of populations sampled of P. lutea, P. hirsuta and their hybrid in Wipptal (inset, black squares). Hybrids were found in the presence of both parents in populations I and VI and in the presence of P. lutea only in population V. Primula lutea was found in populations I, II, V and VI, and P. hirsuta in populations I, III, VI, VII, VIII and IX. Distribution areas are based on data from IntraBioDiv Floristic Consortium—database Update 25.09.2006, Alberto Selvaggi and Thomas Wilhalm (unpublished data), Swiss Web Flora © WSL 2000, Prosser and Scrotegagna (1998), Moser (1998, 1999), Pigniatti (1982), Hess et al. (1970) and Hegi (1927)

Flowering time

The number of flowering individuals of P. lutea, P. hirsuta and their hybrid was recorded in six of the eight localities (I, II, III, V, VI and VIII) between 2 May and 14 July 2006.

Experimental crosses

For experimental crosses between P. lutea and P. hirsuta, flowers were emasculated between 3 and 4 days before opening by removing the corolla and the attached stamens. The emasculated flowers were isolated in bags of garden fleece. Between 5 and 15 days after isolation, pollen was manually applied to the stigmata. Seed parents originated from populations which contained only one parental species. Pollen of P. lutea was collected in populations II and V, pollen of P. hirsuta in population VIII, and crosses were made in populations I, II, III and VIII. Altogether 55 individuals were used for interspecific hybridisations of which 26 could be recovered at the end of the experiment. Crosses were distributed as follows among these 26 individuals (female × male): P. lutea-pin × P. hirsuta-thrum: 3; P. lutea-thrum × P. hirsuta-pin: 6; P. hirsuta-pin × P. lutea-thrum: 5; P. hirsuta-thrum × P. lutea-pin: 12. Female individuals of P. lutea originated from populations I (11 individuals) and II (12 individuals), and male individuals of P. hirsuta from population VIII. Female individuals of P. hirsuta originated from populations III (10 individuals) and VIII (22 individuals), and male individuals of P. lutea from populations II and V. Crosses between P. lutea-thrum and P. hirsuta-thrum and between P. lutea-pin and P. hirsuta-pin were not performed based on the assumption that intramorph crosses are self-incompatible across species boundaries.

Hybrid fertility

Pollen fertility

Pollen fertility was determined in 21 individuals of P. lutea, 30 individuals of P. hirsuta and 45 hybrid individuals by MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] staining (Gahan and Kalina 1968; Rodriguez-Riano and Dafni 2000). For each individual 100 pollen grains from one flower were counted. Pollen grains either stained purplish or having irregular purplish lines on their surfaces were classified as fertile.

Hybrid seed set and germination rate

To determine seed set in hybrid individuals, inflorescences of 33 individuals were isolated in bags of garden fleece after flowering and the number of seeds formed was determined at the fruiting stage. Seeds obtained from hybrids in nature as well as seeds obtained from the parental species were used to determine the percentage of germination. For this purpose, a total of 505 seeds (P. lutea: 150, P. hirsuta: 175, hybrids: 180) seeds were treated for 3 days with a 1.5 mM solution of gibberellic acid (GA3; Duchesfa Biochemicals, Netherlands) and sown in pots on standard garden substrate (TS1 Standard; Klasmann-Deilmann, Germany). Germination was recorded for a period of 11 weeks.

Soil analysis

Soil samples were collected near the roots of 15 individuals of P. lutea, 15 individuals of P. hirsuta and 14 hybrid individuals. Before taking soil samples, litter and/or moss were removed from the soil surface. Soil samples were placed in the freezer not later than 6 h after collecting to prevent biogenic changes.

To determine soil pH, 5 g of air-dried soil were suspended in a 12.5 ml 0.01 M CaCl2 solution and left for 2 h. This solution was resuspended and left for another 10 min before measuring the pH with a pH meter (MultiCal® pH 526; WTW, Weilheim, Germany). The pH was read after 30 s of stability. The protocol followed Emde and Szöcs (2001).

AFLP analysis

Plant material

In the P. lutea and P. hirsuta hybrid populations, 34 hybrid individuals (4 individuals from population I and 30 from VI), 12 individuals of P. hirsuta (2 individuals each from populations I, III, VI, VII, VIII and IX), and 6 individuals of P. lutea (2 individuals each form populations I, II and V) (Fig. 1) were subjected to an AFLP analysis.

The distribution ranges of P. marginata and P. latifolia to some extent overlap, and the two species can be found growing sympatrically at some sites in the Maritime Alps. Of P. marginata, 28 populations with altogether 352 individuals and of P. latifolia, 15 populations with altogether 172 individuals were sampled (Fig. 1). At two sites (CFRE, LAC) the two species grew sympatrically, and at four sites (BRI, BAR, VACH and BFRE, FON and CAS) the two species were sampled at less than 3 km apart from each another.

Primula hirsuta and P. daonensis have a small but not clearly circumscribed zone of overlap in the Alpi Orobie and the Ortler in the southern Central Alps, where the species are known to hybridise (Table 2; Prosser 2000; Prosser and Scrotegagna 1998). More widespread hybridisation between the two species has been claimed by Kreß (1998), and hybrids have been reported from the Bergamask Alps by Widmer (1891). Of P. daonensis eight populations with altogether 113 individuals and of P. hirsuta 21 populations with 234 individuals were sampled (Fig. 1). In P. marginata and P. daonensis the sampled populations were well-distributed across the whole distribution area. In P. latifolia and P. hirsuta sampling was limited to the Alps and did not include the Pyrenees. P. hirsuta was sampled with an emphasis on the geographical neighbourhood of P. daonensis.

DNA extraction

Total genomic DNA was extracted from silica gel-dried leaf material (c. 20 mg dry weight) using the Qiagen DNeasy Plant Mini Kit following the manufacturer’s protocol except for the addition of sodium metabisulfite (S9000; Sigma-Aldrich) to the AP1-buffer to a 10 mM final concentration (Horne et al. 2004), but not in the P. hirsuta × P. lutea analysis. DNA was eluded twice using 40 μl AE-buffer.

AFLP protocol

The AFLP protocol followed Vos et al. (1995) and Kropf et al. (2003) with modifications. For P. latifolia and P. marginata (lat, mar), all master mixes were prepared at once and aliquots for subsequent use were frozen at −20°C to ensure comparability. All reactions were performed three times for a third of all individuals, respectively. For P. hirsuta and P. daonensis (hir, dao), the restriction-ligation was performed simultaneously for all individuals of one species, respectively, and for the P. lutea × P. hirsuta analysis (lut × hir), all reactions were performed simultaneously.

Approximately 100 ng total genomic DNA was simultaneously digested and ligated to adaptors (EcoRI, 5′-CTCGTAGACTGCGTACC-3′/5′-AATTGGTACGCAGTC-3′; MseI, 5′-GACGATGAGTCCTGAG-3′/5′-TACTCAGGACTCT-3′). Reactions were incubated at 23°C for 14 h (lat, mar) or 37°C for 2 h and 15°C for 8 h (hir, dao, lut × hir). Products of the restriction-ligation reaction were diluted tenfold (lat, mar) or fourfold (hir, dao, lut × hir).

Pre-selective amplification was done using primers E01 (5′-GACTGCGTACCAATTCA-3′) and M02 (5′-GATGAGTCCTGAGTAAC-3′). For P. latifolia and P. marginata 2.5 μl were used as template in the pre-selective PCR in a reaction volume of 10 μl containing 5 μl 2 × PCR Master Mix (Promega), 26 ng MseI + 1 primer, 26.5 ng EcoRI + 1 primer and DEPC treated water. The thermocycling profile started with 5 min at 65°C followed by 30 cycles of 30 s at 94°C, 30 s at 56°C, 1 min at 72°C.

For all species and hybrids products of the pre-selective PCR were diluted 20-fold and used as template in selective amplifications using the primers E38 (5′-GACTGCGTACCAATTCACT-3′) combined with M52 (5′-GATGAGTCCTGAGTAACCC-3′), E38 with M55 (5′ -GATGAGTCCTGAGTAACGA-3′) and E39 (5′-GACTGCGTACCAATTCAGA-3′).

For the selective amplifications thermocycling conditions were 10 min at 95°C followed by 36 cycles of 30 s at 94°C, 1 min at X°C and 2 min at 72°C. In the first 13 cycles the annealing temperature was reduced by 1°C at each cycle starting at 65°C with the exception that 64, 62 and 58°C were repeated twice. The remaining 23 cycles were continued at 56°C, followed by 10 min at 72°C.

AFLP products were separated as a multiplex of the three primer combinations labelled with fluorescent dyes (6-FAM®, NED® and HEX®; ABI) on an ABI 3130xl Genetic Analyzer (ABI) (lat, mar) or ABI 3730 DNA Analyzer (hir, dao, aur × hir) with ROX500 (ABI) as internal size standard.

AFLP data analysis

Scoring of AFLP fragments was performed with GeneMarker 1.75 (Softgenetics) except for the P. lutea × P. hirsuta trial where GeneMapper 3.5 (ABI) was used. We calculated mismatch error rates to evaluate the quality of our analyses as the ‘number of genotype mismatches’/(‘number of replicate pairs’ × ‘number of loci’).

PCO

The AFLP dataset of P. lutea, P. hirsuta and their hybrid was subjected to a principal coordinates analysis (PCO; Gower 1966). A PCO condenses information contained in a multi-dimensional distance matrix to a few dimensions. This is helpful in investigating the diversity of hybrid and parental populations, and the genetic relationships of hybrid individuals to their parents, respectively (e.g. Hansen et al. 1999). The binary AFLP data matrix of P. lutea, P. hirsuta and the hybrid individuals was used for calculating squared Euclidean distances for all possible pairs of individuals with SPSS for Windows 14.0.2 (SPSS, Munich, Germany). Based on this distance matrix, the PCO was performed using PAST 1.6 (Hammer et al. 2001). The graphic output for the first and second principal coordinates was created with SPSS.

Admixture

For the two sister species pairs P. marginata/P. latifolia and P. hirsuta/P. daonensis admixture analyses based on pre-defined clusters were done with BAPS 5.1 (Corander and Marttinen 2006; Corander et al. 2008). Admixture analyses aim at identifying ancestral sources of alleles observed in different individuals. Hybrid individuals are assumed to have both parental species as source populations. Admixture coefficients indicate the proportional representation of different clusters in the genome of each individual. For analyses of P. marginata and P. latifolia as well as P. hirsuta and P. daonensis, two clusters corresponding to the two species, respectively, were pre-defined. Admixture analyses were run using 100 iterations to estimate the admixture coefficients for the individuals, 1,000 reference individuals from each population and 20 iterations to estimate the admixture coefficients for the reference individuals. The minimum size of a cluster considered was one individual. Evidence for admixture was considered significant for individuals with a Bayesian p value < 0.05.

Results

The geographical and altitudinal distribution, ecological requirements, chromosome number and flowering time as well as absolute overlap scores (AOS) of all species of P. sect. Auricula are shown in Table 1. Most species are found on either basic or acidic substrates, and only few can be found on both substrates. Flowering time of species is largely overlapping with the exception of the early flowering P. palinuri from south Italy. Except for P. clusiana, all species have a hexaploid (2n = 66) or hypohexaploid (2n = 62) chromosome number. Intraspecific polyploidy has been reported for P. marginata. Species hybridise with between none and eight other species irrespective of chromosome number. Thus, hybrids between species with 2n = 66 and 2n = 62 chromosomes are common, and P. clusiana (2n = 198) hybridises with P. auricula (2n = 62) and P. minima (2n = 66). The identity of hybrids between species of P. sect. Auricula found in nature and their distribution and frequency are shown in Table 2. Of the 32 hybrid combinations known, 18 show a hybrid frequency score of one. The most common hybrid is that between P. glutinosa and P. minima which has been recorded ten times.

Flowering time

Flowering time of P. lutea, P. hirsuta and their hybrids overlap to a large degree. Time of onset of flowering of the two species is essentially identical, the flowering peak is delayed in P. lutea in comparison to P. hirsuta, and flowering lasts longer in P. lutea. This is evident from Fig. 2, which shows the percentage of flowering individuals through time at Gschnitz.
https://static-content.springer.com/image/art%3A10.1007%2Fs00035-010-0084-y/MediaObjects/35_2010_84_Fig2_HTML.gif
Fig. 2

Flowering time of P. lutea, P. hirsuta and their hybrid in population I at Gschnitz (see Fig. 1 for locality)

Experimental crosses

No fruiting individuals were obtained from the cross P. lutea-pin × P. hirsuta-thrum. The other crosses resulted in three (of six; P. lutea-thrum × P. hirsuta-pin), four (of five; P. hirsuta-pin × P. lutea-thrum) and nine (of 12; P. hirsuta-thrum × P. lutea-pin) individuals with seed set. The mean percentage of capsules with seeds was 92.1%, and the average number of seeds per capsule was 30.5.

Hybrid fertility

Pollen fertility

The mean (±SD) pollen fertility of hybrid individuals was 76.4% (±17.6) compared to 92.2% (±9.3) in P. lutea and 92.4% (±7.9) in P. hirsuta.

Hybrid seed set and germination rate

Of the 33 individuals bagged, 13 were found to have produced seeds. The average number of seeds per inflorescence was 41.8. The strong disintegration of capsules at the time of fruiting did not allow determining the number of fertile capsules, and accordingly, did not allow determining the number of seeds per capsule in comparison to seed set in the parental species. Of the seeds obtained, germination rate was 34% for hybrid seeds compared to 72% in P. lutea and 40% in P. hirsuta.

Soil analysis

The soils in the root area of the parental species had a clearly different pH, ranging from 7.09 to 7.73 (mean 7.08 ± 0.6; one outlier of 5.1) in P. lutea and from 3.51 to 6.32 (mean 4.89 ± 0.89) in P. hirsuta (Fig. 3). Soil pH of hybrid individuals ranged from 3.94 to 7.73 (mean 5.71 ± 1.38) and was significantly (p < 0.001; two-tailed Mann–Whitney U test) different from soil pH of the parental species.
https://static-content.springer.com/image/art%3A10.1007%2Fs00035-010-0084-y/MediaObjects/35_2010_84_Fig3_HTML.gif
Fig. 3

Soil pH for 15 individuals of P. lutea, 15 individuals of P. hirsuta and 13 individuals of their hybrid. Box plots show the median, boxes mark the interquartile range, whiskers extend to 1.5× the length of the box or to the minimum and maximum values, respectively; outliers are more than 3× larger than the extend of the box

AFLP

P. lutea × P. hirsuta

In the analysis of P. lutea × P. hirsuta hybrid populations 266 scorable AFLP fragments were generated. We found a mismatch error rate of 8.2% considering two of three primer-pairs (HEX- and NED-labled) and using five replicate pairs. Except for two hybrid individuals all individuals genotyped showed different fragment patterns. The PCO analysis showed (Fig. 4) that the two parental species can be clearly separated along the first two axes, and that the hybrid individuals are widely scattered between the two parental species. One individual identified as P. hirsuta clearly grouped with the hybrid individuals.
https://static-content.springer.com/image/art%3A10.1007%2Fs00035-010-0084-y/MediaObjects/35_2010_84_Fig4_HTML.gif
Fig. 4

Principal coordinate analysis of AFLP variation in P. hirsuta (12 individuals), P. lutea (6 individuals) and their hybrid (34 individuals) using squared Euclidean genetic distances

P. latifolia/P. marginata

A total of 247 polymorphic AFLP fragments were detected in P. marginata and P. latifolia, 223 in P. marginata and 185 in P. latifolia. The mismatch error rate was 3.8% within 19 replicate pairs. Of all fragments, 62 (25.1%) were private to P. marginata and 24 (9.7%) to P. latifolia. Negligible admixture was found between species. Of the 524 individuals analysed, only two individuals per species were significantly admixed with the other species (Fig. 5). They belong to populations ALB and CFRE in P. latifolia (with admixture coefficients of 0.16 and 0.28, respectively), and to GUI and TEN in P. marginata (with admixture coefficients of 0.11 and 0.23, respectively; see Fig. 1 for location of populations). CFRE and TEN are located in the Alpi Marittime where the distribution of the species overlaps. The other two populations are located in areas where the two species do not grow together.
https://static-content.springer.com/image/art%3A10.1007%2Fs00035-010-0084-y/MediaObjects/35_2010_84_Fig5_HTML.gif
Fig. 5

Estimated admixture coefficients for the species pairs aP. latifolia/P. marginata and bP. hirsuta/P. daonensis. Each column corresponds to one individual, and populations are separated by black vertical lines. Pre-defined groups corresponding to the species are shown in light and dark grey. Each column is shown in light or dark grey in proportions corresponding to estimated admixture coefficients of the corresponding individual. Populations are sorted alphabetically within species

P. hirsuta/P. daonensis

In P. hirsuta and P. daonensis a total of 133 fragments were detected, 97 in P. daonensis and 130 in P. hirsuta. A mismatch error rate of 4.5% comparing 150 replicate pairs within and between three different AFLP analyses was found. Three of these fragments (2.3%) were private to P. daonensis and 36 (27.1%) were private to P. hirsuta. Individuals that are significantly admixed with the other species were detected in two P. hirsuta populations. In SUL, 18 of 22 individuals are significantly admixed, and three of 11 individuals in GRI (Fig. 5). Whereas SUL is geographically close to the distribution range of P. daonensis and both species have been collected in this valley (W. Gutermann and L. Schratt-Ehrendorfer, 29/07/1982/WU; Arnold, 23/06/1913/M), GRI is situated far from the distribution range of P. daonensis (see Fig. 1).

Discussion

Frequency of hybridisation

Irrespective of the ease of hybridisation in P. sect. Auricula as evident from the observation of 32 hybrid combinations in nature (Table 2), our own crosses between P. hirsuta and P. lutea, and easy crossability in cultivation (Richards 1993), all evidence we have suggests that hybrids are rare in nature. First, the literature, herbarium and field survey revealed that hybrid populations of the 32 hybrid combinations known have only been observed at 63 localities. Considering the very close morphological similarity among several species of P. sect. Auricula (Zhang and Kadereit 2004), however, the possible existence of undetected hybrid populations should be considered. Our survey would then be an underestimation of hybrid frequency. Second, our admixture analysis of P. latifolia and P. marginata and P. hirsuta and P. daonensis revealed the existence of only very few individuals with significant evidence for interspecific admixture even in areas of sympatry (25 of 871 individuals examined). However, in P. hirsuta and P. daonensis sampling in the region of distributional overlap was not very dense, and only four populations from the Ortler and one from the Alpi Orobie were sampled.

Mechanisms of reproductive isolation

Considering the ease of crossability of species of P. sect. Auricula in cultivation (Richards 1993) on the one hand, and the rarity of interspecific hybrids in nature on the other hand, the mechanisms of reproductive isolation between species of this group require analysis. In the following discussion, we adopted the categories of reproductive isolating barriers recently used by Lowry et al. (2008), supplemented by ecological hybrid inviability as an additional barrier as discussed by Nosil et al. (2005).

When assuming that species of P. sect. Auricula hybridise easily with one another when they encounter, strict sympatry seems to be comparatively rare and the number of locations in which species co-occur in close neighbourhood is very limited as evident from our absolute overlap scores (Table 1). Also, the number of localities in which hybrids have been observed is low (Table 2). Species distribution modelling for two species pairs examined here, P. latifolia/P. marginata and P. daonensis/P. hirsuta, which have neighbouring and partly overlapping distribution ranges, has shown that P. latifolia and P. marginata occupy distinct climatic niches irrespective of their broadly similar distribution (Schorr 2009). Sympatry of species also is limited by edaphic differentiation between species, and limitation to either basic or acid soils is strict in almost all species (Zhang et al. 2004; Table 1). In consequence, geographical and ecogeographical isolation is an important reproductive barrier among species of P. sect. Auricula. Considering the edaphic differentiation between species, immigrant inviability is highly likely to contribute to reproductive isolation. Although no experimental data have been collected to directly investigate this aspect of reproductive isolation, the limitation of P. hirsuta and P. lutea to acid and basic soils, respectively, and the growth of their hybrids on intermediate soils only, implies immigrant inviability. As regards reproductive isolation by phenology, i.e. flowering time, both the general survey of flowering time in the section (Table 1) and our detailed investigation of flowering time in P. hirsuta and P. lutea (Fig. 2) show that divergent flowering time between species is not likely to contribute to reproductive isolation. The only exception to this is the early flowering P. palinuri from south Italy which, however, is also geographically isolated from all other species of the section (Zhang and Kadereit 2004). No species-specific data are available on pollinators. However, the report of 32 different hybrid combinations in the section, and the observation of hybridisation between P. lutea and P. hirsuta, of which the former has yellow and the latter violet flowers, may imply that pollinator specificity plays no role in prezygotic isolation. Apart from the differentiation of flower colour as exemplified by P. lutea and P. hirsuta, flowers of the different species of the section are very similar in morphology and size. Also, all species of the section have heterostylous flowers as found in most species of Primula (Mast et al. 2006). As all species thus are likely to be self-incompatible, differences in breeding system are not likely to contribute to reproductive isolation between species. No data are available on competition between conspecific and non-conspecific pollen. Hybrids have been reported between species of different chromosome number and ploidy level. Accordingly, karyological differentiation among species does not seem to affect hybridisation. The mean percentage of capsules with seed resulting from controlled crosses in the field was 92.1%, and the mean number of seeds per capsule was 30.5. As no intraspecific crosses were performed in P. hirsuta and P. lutea, we do not know whether F1-seed formation was reduced in comparison to parental seed formation. The germination rate of hybrid seeds was 34 compared to 72% in P. lutea and 40% in P. hirsuta. Accordingly, the germination rate of hybrid seeds is reduced at least in comparison with P. lutea. The hybrids between P. hirsuta and P. lutea observed in the field were vegetatively viable and no difference between them and parental individuals was evident (H. Goldner, personal observation). Considering the restriction of hybrid individuals between P. hirsuta and P. lutea in nature to soils with pH values intermediate between the pH values of the soils of the parental species, it seems likely that these hybrids show ecological inviability and cannot grow in the parental habitats. This is supported by the finding that these hybrids have long been known to exist and persist (Jacquin 1778; Reichenbach 1830, 1855; Widmer 1891; Pax 1889; Pax and Knuth 1905; Ernst and Moser 1925; Lüdi 1927) without expansion of the area they occupy. However, no experimental transplant data are available to support the hypothesis of ecological hybrid inviability. In the Alps, the limitation of hybrid individuals between taxa of divergent edaphic preference to soils of intermediate pH has also been found in Rhododendron ferrugineum L. × R. hirsutum L. (Milne and Abbott 2008), and in introgressants of two subspecies of Carex curvula All. (Choler et al. 2004). As found in these two genera, the hybrid population of P. hirsuta × P. lutea investigated by us is likely to represent a segregating and/or backcrossing population (Fig. 4). This is supported by the observation of a broad spectrum of flower colours in this population (H. Goldner, personal observation). Mean male fertility of hybrid individuals was 76.4% and thus reduced in comparison with the parental species P. lutea (92.2%) and P. hirsuta (92.4%). Seed set with an average number of 41.8 seeds per inflorescence was observed in hybrids between P. hirsuta and P. lutea in the field. Field conditions, resulting in disintegration of mature capsules inside bagged inflorescences, did not allow the determination of the number of seeds per capsule. Also, seed set in the parental species was not quantified. Accordingly, we cannot compare seed set in hybrid and parental individuals.

Conclusions

From all evidence we have, hybridisation clearly does take place in P. sect. Auricula, but hybrids appear to be rare. From the data we have, three mechanisms of reproductive isolation can be identified. The first is geographical and/or ecogeographical isolation. This we concluded from the rarity of hybrids in nature assuming that hybridisation takes place when species encounter. Second, ecological hybrid inviability seems to be important at least in hybrids between edaphically differentiated species as suggested by our data on hybrids between P. hirsuta and P. lutea. Habitat selection against ecologically intermediate hybrid individuals is a well-known phenomenon (Schluter 2000, 2001; Nosil et al. 2005). Third, the hybrid population examined by us in detail showed reduced male fertility in comparison with both parental species and hybrid seeds obtained from experimental crosses showed reduced germination in comparison with one parent. This reduced reproductive fitness is likely to contribute to the lack of success of hybrid plants. Following Arnold (1997), replacement of parental species by hybrids requires that hybrids are as fit as or fitter than the parental species. Although we did not quantify the relative contribution of these three barriers to reproductive isolation among species of P. sect. Auricula, prezygotic isolation through geographical and ecogeographical isolation may be most important. All evidence we have suggests that species of P. sect. Auricula at present are stable and well isolated from each other, and that neither is species delimitation aggravated by hybridisation (Rieseberg et al. 2006), nor is there any evidence for genetic assimilation of rare by widespread and more numerous species (Rieseberg 1991; Ellstrand 1992; Levin et al. 1996; Rhymer and Simberloff 1996; Arnold 1997; Levin 2000). Should future changes in geographical distribution of species in response to either natural or anthropogenic climatic or other environmental changes result in increased sympatry of species, interspecific hybridisation is likely to become more frequent and, at least when considering hybrids between species of similar ecology, interspecific hybrids are likely to become more common. We cannot predict whether reduced hybrid reproductive fitness, if also found in other hybrid combinations than P. lutea × P. hirsuta, will be sufficient to prevent the replacement of parental species and thus their extinction.

Acknowledgments

Christian Parisod (Associate Editor/Neuchâtel) and two anonymous reviewers provided helpful comments on an earlier version of this paper. The Deutsche Forschungsgemeinschaft (DFG) is gratefully acknowledged for continued financial support of the Primula project to J.W.K., and we thank Doris Franke (Mainz) for help with the figures.

Copyright information

© Swiss Botanical Society 2011