Colonization and diversification in the African ... - Wiley Online Library

Research

Colonization and diversification in the African ‘sky islands’: insights from fossil-calibrated molecular dating of Lychnis (Caryophyllaceae) Abel Gizaw1,2, Christian Brochmann1, Sileshi Nemomissa2, Tigist Wondimu1,2, Catherine Aloyce Masao1,3,4, Felly Mugizi Tusiime1,5, Ahmed Abdikadir Abdi1,6, Bengt Oxelman7, Magnus Popp1* and Dimitar Dimitrov1* 1

Natural History Museum, University of Oslo, PO Box 1172 Blindern, NO-0318 Oslo, Norway; 2Department of Plant Biology and Biodiversity Management, College of Natural Science,

Addis Ababa University, PO Box 3434, Addis Ababa, Ethiopia; 3Department of Forest Biology, Sokoine University of Agriculture, PO Box 3010, Morogoro, Tanzania; 4Institute of Resource Assessment, University of Dar es Salaam, Box 35097 Dar es Salaam, Tanzania; 5Department of Forestry Biodiversity and Tourism, School of Forestry Geographical and Environmental Sciences, Makerere University, PO Box 7062, Kampala, Uganda; 6Botany Department, National Museums of Kenya, PO Box 40658, 00100 Nairobi, Kenya; 7Department of Biology and Environmental Sciences, University of Gothenburg, PO Box 461, 405 30 Gothenburg, Sweden

Summary Author for correspondence: Abel Gizaw Tel: +47 48667183 Email: [email protected] Received: 15 January 2016 Accepted: 18 February 2016

New Phytologist (2016) 211: 719–734 doi: 10.1111/nph.13937

Key words: African mountains, diversification, Lychnis, molecular dating, origin of high-altitude lineages.

The flora on the isolated high African mountains or ‘sky islands’ is remarkable for its peculiar adaptations, local endemism and striking biogeographical connections to remote parts of the world. Ages of the plant lineages and the timing of their radiations have frequently been debated but remain contentious as there are few estimates based on explicit models and fossil-calibrated molecular clocks. We used the plastid region maturaseK (matK) and a Caryophylloflora paleogenica fossil to infer the age of the genus Lychnis, and constructed a data set of three plastid (matK; a ribosomal protein S16 (rps16); and an intergenic spacer (psbE-petL)) and two nuclear (internal transcribed spacer (ITS) and a region spanning exon 18–24 in the second largest subunit of RNA polymerase II (RPB2)) loci for joint estimation of the species tree and divergence time of the African representatives. The time of divergence of the African high-altitude Lychnis was placed in the late Miocene to early Pliocene. A single speciation event was inferred in the early Pliocene; subsequent speciation took place sporadically from the late Pliocene to the middle Pleistocene. We provide further support for a Eurasian origin of the African ‘sky islands’ floral elements, which seem to have been recruited via dispersals at different times: some old, as in Lychnis, and others very recent. We show that dispersal and diversification within Africa play an important role in shaping these isolated plant communities.

Introduction Understanding patterns of species diversity, distributions and the processes that govern the evolution of distinct regional biota has been a central theme for evolutionary biologists since the times of Darwin and Wallace. Tropical mountain biotas have been recognized as exceptionally diverse and have attracted much attention (e.g. Mittermeier et al., 2005; Orme et al., 2005; Merckx et al., 2015). Yet the diversity of the species assemblies that tropical mountains harbour has been notoriously difficult to explain on the basis of contemporary factors such as climate (e.g. Rahbek & Graves, 2001), indicating the importance of historical processes such as lineage diversification and dispersal (e.g. Wiens & Donoghue, 2004; Cadena et al., 2011; Madri~ nan et al., 2013; *These authors contributed equally to this work. Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

Dauby et al., 2014; Favre et al., 2015; Hughes & Atchison, 2015; Merckx et al., 2015). The mountains of eastern Africa (i.e. East Africa and Ethiopia) provide an excellent example of a system where the interplay between historical and contemporary processes has produced diverse floras and faunas rich in local endemics (Hedberg, 1969; Mittermeier et al., 2005). Based on their geological age and the processes that led to their formation, the eastern African mountains can be divided into two remarkably different groups: the young and volcanic East African Rift System (EARS) running across eastern Africa and the much older and mostly uplifted mountain blocks of the eastern Afromontane Biodiversity Region in Kenya and Tanzania (also known as the Eastern Arc or the Eastern Arc Mountains; Mittermeier et al., 2005). In the latter, the basal blocks of the mountains are at least 30 million years old (Myr), but are probably much older – perhaps > 100 Myr New Phytologist (2016) 211: 719–734 719 www.newphytologist.com

720 Research

(Griffiths, 1993). The two mountain groups have different geological histories, but both are located in eastern Africa and are separated only by short distances along the boundary between Kenya and Tanzania. Several recent biogeographical and phylogeographical studies have focused on the origins and evolution of the afro-montane flora and fauna (e.g. Kebede et al., 2007; Couvreur et al., 2008; Tolley et al., 2011; Dimitrov et al., 2012; Ceccarelli et al., 2014; Loader et al., 2014) and have stressed the importance of longterm climatic stability for the generation and maintenance of species diversity (Lovett et al., 1988; Fjelds a & Lovett, 1997). Clades that diversified in this region are often old and have their closest relatives in the African lowland tropical forests (e.g. Couvreur et al., 2008). In sharp contrast, the higher elevations of the EARS mountains are home to much younger afro-alpine habitats where night frost is common, and it has been suggested that several of its local plant lineages have immigrated from Eurasia via the Arabian Peninsula (Koch et al., 2006; Assefa et al., 2007; Ehrich et al., 2007; Popp et al., 2008; Gehrke & Linder, 2009). The actual time of colonization of these mountains is, however, poorly understood and remains contentious. The EARS is a prominent geological feature, extending from Syria in the north through East Africa to Mozambique in the south. Its formation began c. 45 million years ago (Ma) and divided the region into one western and one eastern branch of the Rift (Fig. 1; Ebinger et al., 2000; Trauth et al., 2005). Associated volcanism resulted in the formation of extensive mountain systems in the Ethiopian highlands and scattered high mountains in East Africa, the origins of which range from the Miocene to the late Pliocene (Griffiths, 1993). The mountains are famous for their unique ecosystem with distinct altitudinal zonation of the vegetation: the uppermost afro-alpine zone proper, the ericaceous transition zone, and the subalpine montane forest zone (Hedberg, 1951). Because of the high degree of isolation of the highest peaks in the EARS, these are often referred to as ‘sky islands’, as an analogy to isolated oceanic islands (Assefa et al., 2007; Popp et al., 2008; Gehrke & Linder, 2009). About 77% of the vascular plant species of the high-altitude frost belts and the proper afroalpine zones of these mountains are recognized as local or regional endemics (Hedberg, 1961, 1969; Gehrke & Linder, 2014), with a large proportion having their closest relatives in other temperate areas rather than in the adjacent tropical lowlands (Hedberg, 1965, 1970). This led Hedberg (1961) to state that ‘[the afro-alpine flora] seems to have existed in tropical East Africa before the formation of the mountains [Miocene to late Pleistocene] now harbouring it, and to have been strongly isolated from other high-mountain (or temperate) floras for a long time’ (but see Hedberg, 1970). By contrast, others have suggested that at least parts of the flora are rather young (Koch et al., 2006; Assefa et al., 2007; Ehrich et al., 2007; Gehrke & Linder, 2014; Linder, 2014), and that the colonization and current distribution of the mountain flora were mainly shaped by the climatic oscillations of the Plio-Pleistocene period (e.g. Livingstone, 1962; Hamilton, 1982; Mohammed & Bonnefille, 1998; Trauth et al., 2005). This view was also adopted in a recent study of the evolution of African plant New Phytologist (2016) 211: 719–734 www.newphytologist.com

New Phytologist diversity, where the afro-alpine flora was suggested to represent the youngest of all African floras (Linder, 2014). However, only a few studies have addressed the time of colonization of the alpine zones of the EARS by plants. Koch et al. (2006) used synonymous mutation rates and sequence distances calculated for the chloroplast gene matK and the nuclear alcohol dehydrogenases (adh) and chalcone synthase (chs) genes to infer a Pleistocene origin of the afro-alpine populations of the arctic-alpine plant Arabis alpina. A handful of other studies that included at least one African high-elevation endemic plant species have used mutation rates to infer ages of diversification events (e.g. Knox & Palmer, 1995). The use of fixed rates and sequence distances is, however, not optimal for molecular dating as it assumes a strict clock with no rate heterogeneity without testing the validity of those assumptions. More recently, Gehrke et al. (2015) used fossil information to date the phylogeny of African Alchemilla and its diversification in afro-alpine habitats, finding support for a Pleistocene evolution of the afro-alpine species. In addition, some broader scope studies using fossil-calibrated molecular clocks have included at least one high-elevation species found in the EARS region (Antonelli, 2009; Jabbour & Renner, 2012; N€ urk et al., 2015), finding a range of estimated ages for their origin (from Miocene to late Pleistocene). Here, we used explicit fossil-calibrated molecular clocks to estimate the minimum ages and the corresponding highest probability time intervals for the origin and colonization of the afroalpine/afro-montane habitats in eastern Africa by plants, exemplified by the high-altitude representatives of Lychnis L. (Caryophyllaceae), which are distributed in the ‘sky islands’ of Ethiopia, East/Central Africa (Kenya, Tanzania, Uganda, Rwanda, Burundi and the Democratic Republic of Congo) and West Africa (Cameroon and Nigeria). The genus Lychnis comprises some 30 species occurring mostly in temperate areas of the Northern Hemisphere (Oxelman et al., 2000). The six highaltitude African species of the genus (previously recognized as an endemic genus, Uebelinia Hochst.) may be referred to as afromontane rather than afro-alpine (e.g. Popp et al., 2008) because they often occur below the tree line (Ousted, 1985), but the afro-montane concept has been used as a loose framework applied to most African mountains south of the Sahara (Carbutt & Edwards, 2015). Occurrence restricted to areas above the tree line or above specific elevations provides an objective criterion with which to classify species as alpine (e.g. Linder, 2014; Carbutt & Edwards, 2015), although some species of the tropical alpine communities may also sporadically occur at lower elevations (Hedberg, 1957; Rundel et al., 1994; Leuschner, 1996 and references therein). In our previous biogeographical study (Popp et al., 2008), we found support for a single Eurasian origin of the high-elevation lineage of African Lychnis, which formed a monophyletic group and was inferred to be sister to the Eurasian Lychnis flos-cuculi L. The only other African representative of the genus, the Moroccan endemic Lychnis lagrangei Coss., thus seems to have immigrated independently to Africa. The African high-elevation clade comprised two distinct lineages: one diploid consisting of Lychnis rotundifolia (Oliver) M. Popp, Lychnis scottii (Turrill) M. Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

New Phytologist

Research 721

(a)

(b)

Fig. 1 Geographical distribution and sampling localities for the African highelevation representatives of Lychnis. (a) Sampling areas in eastern Africa. (b) Total range of each species and subspecies (solid or dashed lines) and sampling localities (symbols), modified from Popp et al. (2008) and Ousted (1985). SGS, small genome size; LGS, large genome size.

Popp, Lychnis abyssinica (Hochst.) Lide´n and Lychnis kigesiensis (R.D. Good) M. Popp, and one tetraploid consisting of Lychnis crassifolia (T. C. E. Fries) M. Popp and Lychnis kiwuensis (T. C. E. Fries) M. Popp (Popp et al., 2008). Because of its single origin and wide distribution in the eastern African mountains, Lychnis is an excellent system in which to study when and how these ‘sky islands’ were colonized. In addition, the occurrence of Lychnis in Eurasia and Africa provides a broader context that allows us to investigate the links of the African high-mountain flora to other regional floras (for example to infer the source area and how many times dispersal to Africa has occurred; see Popp et al., 2008). In the current study, we first used a fossil-calibrated Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

relaxed molecular clock for the plastid matK gene to infer the age of the genus Lychnis based on a minimum age constraint. Then we used this estimated age for joint estimation of the species tree and divergence time of the African high-elevation lineage of Lychnis based on sequences from three plastid (maturaseK (matK), a ribosomal protein S16 (rps16) and an intergenic spacer (psbE-petL)) and two nuclear (internal transcribed spacer (ITS) and a region spanning exon 18–24 in the second largest subunit of RNA polymerase II (RPB2)) DNA regions. We used our dated phylogeny to address the following questions. What is the timeframe for the colonization of the African high-elevation habitats by the genus Lychnis? What is the timeline of subsequent New Phytologist (2016) 211: 719–734 www.newphytologist.com

New Phytologist

722 Research

speciation events? Which factors may have contributed to speciation and diversification? Are, for example, colonization and diversification events correlated to past climatic and/or geological events? In particular, we aimed to investigate whether colonization of the African high mountains occurred before or during the Pleistocene, since this has been central in earlier discussions of the age of the afro-alpine flora (e.g. Hedberg, 1961, 1970; Koch et al., 2006; Gehrke & Linder, 2014; Linder, 2014). Answering these questions is central to our quest for better understanding of the origins and evolution of eastern African ‘sky island’ diversity in general and of Lychnis in particular.

Materials and Methods Plant material and DNA extraction Leaf tissue of Lychnis L. was collected in Ethiopia, Kenya, Uganda, Tanzania, and Cameroon in 2004–2009 (Table 1) and dried in silica gel. Voucher specimens were deposited in the National Herbarium of Ethiopia, Addis Ababa University, Ethiopia (ETH); with duplicate sets in the Natural History Museum, University of Oslo, Norway (O; full duplicate set of the 2007–2009 collections); the East African Herbarium, National Museum of Kenya, Kenya (EA; Kenyan collections); Sokoine University of Agriculture, Tanzania (SUA; Tanzanian collections); and Makerere University Herbarium, Uganda (MU; Ugandan collections). In addition, herbarium material of the Moroccan Lychnis lagrangei and the Eurasian Lychnis subintegra (Hayek) Turrill and silica-dried leaf samples of the Eurasian Lychnis flos-cuculi were used. Lychnis abyssinica consists of two morphologically indistinguishable lineages with significantly different genome sizes (Popp et al., 2008). We sampled large genome size (LGS) L. abyssinica, which is a putative hybrid (Popp et al., 2008), from its currently known geographical range in the Ethiopian highlands (Fig. 1b) and small genome size (SGS) L. abyssinica from across its Ethiopian range as well as from its strikingly disjunct area in West Africa (Bamenda, northwest Cameroon). Although we included samples of L. abyssinica for both genome size variants from most of its range, we were not able to obtain collections from its small East African range (Fig. 1). Samples of the narrow East African endemics were collected in Mt Kilimanjaro (Lychnis rotundifolia), Mt Aberdare/ Cherangani Hills (Lychnis crassifolia), and the Echuya Forest, Kanaba Swamp in Uganda (Lychnis kigesiensis (R.D. Good) M. Popp subsp. kigesiensis). We also collected representative samples of the narrow Ethiopian endemics Lychnis kiwuensis subsp. erlangeriana (Engl.) M. Popp, Lychnis kigesiensis (R.D. Good) subsp. ragazziana (S. Ousted) M. Popp, and Lychnis scottii. The widespread Lychnis kiwuensis subsp. kiwuensis was sampled from its disjunct areas in the Ethiopian highlands and from Mts Muhavura and Mgahinga in Uganda (Fig. 1). Total genomic DNA was extracted from leaf material using an automated GeneMole® robot (Qiagen Nordic, Oslo, Norway) or DNeasyTM Plant Mini Kit (Qiagen, Valencia, CA, USA). Leaf tissue was ground with two tungsten carbide beads for 2 min at

New Phytologist (2016) 211: 719–734 www.newphytologist.com

15 Hz in a mixer mill (MM301; Retsch GmbH & Co., Haan, Germany). PCR and DNA sequencing Amplification and DNA sequencing protocols as described in Popp et al. (2008) were used to generate sequences for two noncoding plastid DNA regions, the rps16 intron and the psbE-petL interspacer. Approximately 1.5 kb of the plastid matK region was amplified and sequenced in three overlapping fragments using primer pairs matK-F2b/matK-547.rc, matK-526/matK-R2, and matK-1066/matK-1463.rc (Mower et al., 2007; Sloan et al., 2009) using Ready-To-GoTM PCR Beads (Amersham Pharmacia Biotech, Amersham, UK), 0.5 ll of each 10 lM forward and reverse primer, and 1 ll of genomic DNA of 100 ng ll 1 concentration. The PCR products were purified using 2 ll of Exosap-IT (USB Corp., Cleveland, OH, USA), diluted 1 : 10 for 5-ll PCR products, and incubated for 45 min at 37°C and for 15 min at 80°C. Sequencing was performed using the PCR primers and the BigDye Terminator v.1.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturer’s manual (except for using 10-ll reaction volumes) and visualized with an ABI 3100 capillary sequencer (Applied Biosystems). The sequencing programme consisted of denaturing at 96°C for 1 min followed by 25–35 cycles of 96°C for 10 s, 50°C for 5 s, and 60°C for 4 min, ending with extension at 60°C for 10 min. The sequences were edited using CODONCODE ALIGNER v3.5.7 (CodonCode Corporation, Dedham, MA, USA), translated into amino acids for visual inspection, and manually aligned using MEGA v. 4 (Tamura et al., 2007). Substitution model selection was carried out in MRAIC v. 1.4.4 (Nylander, 2004) in conjunction with PHYML v. 3.0 (Guindon & Gascuel, 2003) using the Akaike information criterion. The phylogenetic analyses and dating were performed in two steps. The first analysis was carried out to obtain an estimate of the age for the entire genus Lychnis using matK DNA sequences representing the subfamilies Alsinoideae and Caryophylloideae and a single fossil as a minimum age constraint to calibrate the inferred phylogeny. The inferred age of the genus Lychnis from the matK phylogeny was then used as a secondary calibration point in a subsequent analysis including all available sequences of the African high-elevation lineage of Lychnis. Phylogenetic analyses and dating of the matK phylogeny A total of 33 matK sequences, mainly from Sloan et al. (2009), representing Alsinoideae and Caryophylloideae as well as one sequence of Polycarpon tetraphyllum L. from the subfamily Paronychioideae (for outgroup rooting) were downloaded from GenBank (Table 2). The resulting data set consisted of 46 sequences, including 13 additional matK sequences of African high-elevation Lychnis, the Moroccan L. lagrangei, and the Eurasian L. flos-cuculi and L. subintegra generated in this study. The input files for Bayesian Evolutionary Analysis Sampling Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

New Phytologist

Research 723

Table 1 Samples of high-elevation African Lychnis and other species of Lychnis used in this study

Database number

Population/ individual number

Taxon

Country

Locality

Altitude (m)

O-DP-13596

AFR533-1‡

L. abyssinica SGS

Ethiopia

2700

7.1735

38.8746

O-DP-13600 O-DP-13604

AFR536-4‡ AFR540-1‡

L. abyssinica SGS L. abyssinica SGS

Ethiopia Ethiopia

3200 3100

6.9328 7.1056

39.9460 39.8015

O-DP-13608 O-DP-13616

AFR545-1‡ AFR548-4‡

L. abyssinica SGS L. abyssinica SGS

Ethiopia Ethiopia

3000 2300

9.0895 7.1492

38.7648 38.6990

O-DP-13619

AFR560-2‡

L. abyssinica SGS

Ethiopia

2300

5.8181

38.2650

O-DP-13621 O-DP-34660 O-DP-34661 O-DP-34662

O-DP-41977

AFR570-1‡ ET1526-1 ET1526-2 ET1526-3 AFR582-1 AFR582-2 AFR611-1 AFR611-2 21591 21571 21568 AFR522-4‡ AFR528-5‡ ET0098-2 ET271-2 ET391-2 ET509-2 ET1336-2 ET1351-2 ET1391-2 KN0443-2 KN0493-2 KN0563-2 21567 180610-1 UG2599-2

Ethiopia Ethiopia Ethiopia Ethiopia Cameroon Cameroon Cameroon Cameroon Cameroon Nigeria Nigeria Ethiopia Ethiopia Ethiopia Ethiopia Ethiopia Ethiopia Ethiopia Ethiopia Ethiopia Kenya Kenya Kenya Tanzania Sweden Uganda

2200 3140 3140 3140 na na na na 2400 1200–1500 na 3000 3400 3570 3650 3760 3600 3960 3940 3920 3180 3090 3090 2400 na 2300

7.7693 9.2195 9.2195 9.2195 na na na na na na na 7.9269 7.3663 13.2666 13.2697 13.3333 13.3203 10.6420 10.6560 10.6575 1.1770 0.5425 0.5425 na 58.4890 1.2553

35.4745 41.7867 41.7867 41.7867 na na na na na na na 39.1795 39.2020 38.1078 38.1059 38.2333 38.2351 37.8357 37.8257 37.8220 35.5184 36.7175 36.7175 na 11.5890 29.8093

O-DP-41982

UG2600-2

L. abyssinica SGS L. abyssinica SGS L. abyssinica SGS L. abyssinica SGS L. abyssinica SGS L. abyssinica SGS L. abyssinica SGS L. abyssinica SGS L. abyssinica SGS L. abyssinica SGS L. abyssinica SGS L. abyssinica LGS L. abyssinica LGS L. abyssinica LGS L. abyssinica LGS L. abyssinica LGS L. abyssinica LGS L. abyssinica LGS L. abyssinica LGS L. abyssinica LGS L. crassifolia L. crassifolia L. crassifolia L. crassifolia L. flos-cuculi L. kigesiensis subsp. kigesiensis L. kigesiensis subsp. kigesiensis L. kigesiensis subsp. kigesiensis L. kigesiensis subsp. ragazziana L. kigesiensis subsp. ragazziana L. kiwuensis subsp. erlangeriana L. kiwuensis subsp. erlangeriana L. kiwuensis subsp. erlangeriana

Arsi: Betw. Kofele and Asassa, 15 km from Kofele Bale Mts Bale Mts: Betw. Park Head Quarter and Goba, 2 km from Head Quarter Shoa, Entoto Arsi: Betw. Shashemene and Kofele, 12.5 km from Shashemene Betw. Agere Maryam and Dilla, 23 km from Agere Maryam, Sidamo Keffa, Masha Gara Muleta Gara Muleta Gara Muleta Bamenda Highlands Bamenda Highlands Bamenda Highlands Bamenda Highlands Bamenda, Ridge above Lake Oku Chappal Waddi Sardauuna, Gembu Arsi: Mt Chilalo Arsi: Mt Kaka Simen Mts: Close to Gich Camp Site Simen Mts: Gich Camp Site Simen Mts: Silki Simen Mts: Silki Mt Choke Mt Choke Mt Choke Cherangani Hills: Tululuwa Aberdare Mts: Mt Kinangop area Aberdare Mts: Mutumbio Gate Morogo Dist., Uluguru Mts, Lukwangulu plateau €taland Roadside, H aby, Munkedal, V€astra Go Echuya Forest: Kanaba Swamp

Uganda

Echuya Forest: Kanaba Swamp

2300

1.2553

29.8093

Rwanda

Gikongoro, Nyungwe forest

2300

Ethiopia

Bale Mts: Betw. Park Head Quarter and Goba, 2 km from Head Quarter

3100

7.1055

39.8015

Ethiopia

Shoa, Entoto

3000

9.0950

38.7662

Ethiopia

Arsi: Betw. Kofele and Asassa, 15 km from Kofele

2700

7.1735

38.8746

Ethiopia

Arsi: Betw. Shashemene and Goba, 26 km from Shashemene

2600

7.0985

38.7579

Ethiopia

Betw. Agereselam and Aleta Wondo, 17 km from Agereselam, Sidamo

2500

6.5749

38.4891

O-DP-13583 O-DP-13592 O-DP-29620 O-DP-30300 O-DP-30734 O-DP-31231 O-DP-33637 O-DP-33712 O-DP-33907 O-DP-27285 O-DP-27499 O-DP-27812

21595

O-DP-13628

AFR542-4‡

O-DP-13629

AFR546-1‡

O-DP-13637

AFR534-3‡

O-DP-13648

AFR550-4‡

O-DP-13654

AFR555-1‡

Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

Latitude

na

Longitude

na


New Phytologist

724 Research Table 1 (Continued)

Database number

Population/ individual number

O-DP-13657

AFR527-4‡

O-DP-13662

AFR538-4‡

O-DP-13664

AFR551-3‡

O-DP-13671

AFR561-2‡

O-DP-13674

AFR566-1‡

O-DP-39472

UG2001-2

O-DP-39477

UG2002-2

O-DP-40181

UG2167-2

O-DP-40186

UG2168-2

O-DP-27519 O-DP-36976 O-DP-37105 O-DP-13679 O-DP-13680 O-DP-13682

Taxon

Country

Locality

Altitude (m)

Latitude

Longitude

Ethiopia

Arsi: Mt Chilalo

3100

7.9206

39.1840

Ethiopia

Bale Mts: Rira Village

2800

6.7531

39.7204

Ethiopia

Gamu-Gofa, Chencha

2600

6.2296

37.5678

Ethiopia

Between Teppi and Gore, Illubabor

1800

7.3153

35.3692

Ethiopia

Keffa, Masha

2200

7.7633

35.4745

Uganda

Virunga Mts: Mt Mgahinga

2690

1.3600

29.6348

Uganda

Virunga Mts: Mt Mgahinga

2350

1.3528

29.6201

Uganda

Virunga Mts: Mt Muhavura, Kabaragnuma Swamp

3060

1.3675

29.6713

Uganda

Virunga Mts: Mt Muhavura, Kabaragnuma Swamp

3060

1.3675

29.6713

UP16925*

L. kiwuensis subsp. kiwuensis L. kiwuensis subsp. kiwuensis L. kiwuensis subsp. kiwuensis L. kiwuensis subsp. kiwuensis L. kiwuensis subsp. kiwuensis L. kiwuensis subsp. kiwuensis L. kiwuensis subsp. kiwuensis L. kiwuensis subsp. kiwuensis L. kiwuensis subsp. kiwuensis L. lagrangei

Morocco

na

35.4626

6.0317

UP16926*

L. lagrangei

Morocco

na

35.4700

5.8000

LAG01 LAG02 KN0497-2 TZ0016-2

L. lagrangei L. lagrangei L. rotundifolia L. rotundifolia

Morocco Morocco Kenya Tanzania

na na 3690 3640

na na 0.3218 3.0343

na na 36.6407 37.2430

TZ0043-2 21573 AFR554-4 AFR554-5 AFR554-1‡ UP16917*

L. rotundifolia L. rotundifolia L. scottii L. scottii L. scottii L. subintegra

Tanzania Tanzania Ethiopia Ethiopia Ethiopia Greece

Friedhof in Oulad-el-Arbi, s€ udlich Asilah, auf K€ ustenland Habib, near the road from Arba-Ayacha (Soko Arbaa) to Dar Chaoui Tanger-Tetouan Tanger-Tetouan Aberdare Mts: Mt Satima area Mt Kilimanjaro: Shira Plateau near Mt Simba Mt Kilimanjaro: Shira Plateau Mt Kilimanjaro: Mweka Gamu-Gofa, Mt Guge Gamu-Gofa, Mt Guge Gamu-Gofa, Mt Guge Ioanninon, above Metsovo, N of the road to the Katara pass

3540 3000 3000 3000 3000 na

3.0056 na 6.1992 6.1992 6.1992 39.7760

37.2416 na 37.3333 37.3333 37.3333 21.1750

Leaf material for Lychnis was collected in the field and dried in silica gel for most accessions; herbarium material was used in a few cases (indicated with *). Population number in bold represents newly added material in this study. The assignment of two genome size variants (LGS, large genome size; SGS, small genome size) to L. abyssinica samples was determined based on flow cytometry analyses (for samples from the Popp et al. (2008) study marked with ‡) and based on their phylogenetic position (for samples collected and analysed in this study). na, not available (lack of detailed geographical information).

Trees (BEAST) were prepared using BEAUTI v.1.6.1 (Drummond & Rambaut, 2007; Heled & Drummond, 2010). Monophyly was enforced for the Caryophylloideae and Alsinoideae sequences to ensure the tree was rooted with P. tetraphyllum from the subfamily Paronychioideae (Bittrich, 1993; Fior et al., 2006). A fossil inflorescence of Caryophylloflora paleogenica G. J. Jord. & Macphail (Caryophyllaceae), inferred as sister to either one or both of the subfamilies Alsinoideae and New Phytologist (2016) 211: 719–734 www.newphytologist.com

Caryophylloideae (Jordan & Macphail, 2003), was used to calibrate the tree. The inferred fossil age is middle to late Eocene, which corresponds to an age of 48.6–33.9 Myr (http://www.paleodb.org). A lognormal prior distribution with log mean = 0.0, SD = 1.37 and offset of 33.9 Myr was assigned to the Alsinoideae/Caryophylloideae crown node (Fig. 2). This approximates a 95% probability distribution spanning the inferred minimum and maximum ages of the fossil with a fixed minimum Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

New Phytologist

Research 725

Table 2 GenBank accession numbers of DNA sequence information for the study material retrieved mainly from the Popp et al. (2008) study (†) or generated in the present study (‡) GenBank accession number Taxon

Population number

matK

rps16

psbE-petL

ITS

RPB2

Lychnis abyssinica SGS

21568 21571 21591 AFR533-1 AFR536-4 AFR540-1 AFR545-1 AFR548-4 AFR560-2 AFR570-1 ET1526-1 ET1526-2 ET1526-3 AFR582-1 AFR582-2 AFR611-1 AFR611-2 AFR522-4 AFR528-5 ET0098-2 ET271-2 ET391-2 ET509-2 ET1336-2 ET1351-2 ET1391-2

– – – – – – – – KT581604‡ – – – – – – – – – – KT581605‡ – – – –

EF602350† EF602351† EF602352† EF602355† EF602356† EF602357† EF602358† – EF602360† EF602361† KU366236‡ KU366237‡ KU366238‡ KU366229‡ KU366230‡ KU366231‡ KU366232‡ EF602353† EF602354† KU366228‡ KU366239‡ KU366240‡ KU366241‡ KU366233‡ KU366234‡ KU366235‡ – EF674193 – – – EF602362† KU366251‡ KU366249‡ KU366250‡ Z83163 – – –

EF602323† EF602324† EF602325† EF602328† EF602329† EF602330† EF602331† – EF602333† EF602334† KU366208‡ KU366207‡ KU366209‡ KU366200‡ KU366199‡ KU366202‡ KU366201‡ EF602326† EF602327† KU366203‡ KU366210‡ KU366211‡ KU366212‡ KU366204‡ KU366205‡ KU366206‡ – FJ376841 – – – EF602335† KU366221‡ KU366220‡ KU366222‡ EF602320† – – –

EF602379† EF602380† EF602381† EF602382† EF602383† EF602384† EF602385† EF602386† EF602387† EF602388† – – – – – – – – – – – – – –

EF602406† EF602407† EF602408† EF602411† EF602412† EF602413† EF602414† EF602415† EF602416† EF602417† – – – – – – – EF602409† EF602410† – – – – –

X86894 X86891 AY857966 SCU30953 SCU30979 EF602389

AJ634068 AJ634069 – – – EF602418†

X86893 – SFU30957

AJ634070 AJ634071 FJ376910 –

Z83166 – – EF602366† KU366242‡ KU366243‡ EF602367† EF602368† EF602363† EF602364† EF602365† – EF602370† EF602371† EF602372† EF602373† EF602374† KU366255‡ KU366252‡

EF602321† – – EF602339† KU366213‡ KU366214‡ EF602340† EF602341† EF602336† EF602337† EF602338† – EF602342† EF602343† EF602344† EF602345† EF602346† KU366225‡ KU366223‡

X86892 AY936261 EF407940 EF602393†

– – – EF602422†

EF602394† EF602395† EF602390† EF602391† EF602392† EF602396† EF602397† EF602398† EF602399† EF602400† EF602401†

EF602423† EF602424† EF602419† EF602420† EF602421† – EF602425† – EF602426† EF602427† EF602428†

L. abyssinica LGS

L. chalcedonica L. coronaria

L. crassifolia

21567 KN0493-2 KN0443-2 KN0563-2

KT581609‡

L. flos-cuculi

180610-1

KT581613‡

L. flos-jovis

L. kigesiensis subsp. kigesiensis

L. kigesiensis subsp. ragazziana L. kiwuensis subsp. erlangeriana

L. kiwuensis subsp. kiwuensis

21595 UG2599-2 UG2600-2 AFR542-4 AFR546-1 AFR534-3 AFR550-4 AFR555-1 21598 AFR527-4 AFR538-4 AFR551-3 AFR561-2 AFR566-1 UG2168-2 UG2001-2


KT581606‡ KT581607‡ KT581611‡

KT581612‡


New Phytologist

726 Research Table 2 (Continued) GenBank accession number Taxon

L. lagrangei

L. rotundifolia

L. scottii

L. subintegra L. coronaria L. flos-jovis Polycarpon tetraphyllum Scleranthus perennis Arenaria serpylloides Moehringia macrophylla Stellaria media Lepyrodiclis holosteoides Minuartia nuttallii Bufonia paniculata Schiedea globosa Geocarpon minimum Gypsophila paniculata Petrorhagia saxifraga Dianthus armeria Arenaria bryophylla Arenaria gypsophiloides Agrostemma githago Petrocoptis pyrenaica Silene hawaiiensis Silene uniflora Silene douglasii Silene noctiflora Silene repens Silene schafta Silene ciliata Silene muscipula Silene antirrhina Silene armena Silene acaulis Silene campanula Silene yemensis Silene sordida Silene odontopetala

Population number UG2002-2 UG2167-2 LAG01 LAG02 UP16925 21573 TZ0043-2 KN0497-2 TZ0016-2 AFR554-1 AFR554-4 AFR554-5 UP16917

matK

KT581615‡ KT581610

‡

KT581608

‡

KT581614‡ FJ589507 AY936313 AY936287 AY514847 FJ404826 FJ404852 FJ404877 FJ404840 FJ404847 FJ404827 DQ907818 FJ404836 FJ404838 FJ404857 FJ404832 FJ404815 FJ404818 FJ589503 FJ589508 SHB0001 FJ589565 EF547238 EF547240 FJ589552 FJ404873 FJ589519 FJ589543 FJ589512 FJ589514 EF547235 AY936311 FJ589567 FJ589559 FJ589546

rps16

psbE-petL

KU366253‡ KU366254‡ – EF602349† KU366256‡ EF602375† KU366247‡ KU366246‡ KU366248‡ EF602376† KU366244‡ KU366245‡

KU366226‡ KU366224‡ – EF602322† KU366227‡ EF602347† KU366219‡ KU366217‡ KU366218‡ EF602348† KU366215‡ KU366216‡

– – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

– – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

ITS

RPB2

EF602377† EF602378†

EF602404† EF602405†

EF602402†

EF602429†

EF602403†

EF602430†

– – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

– – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

SGS, small genome size; LGS, large genome size. maturaseK (matK); a ribosomal protein S16 (rps16); an intergenic spacer (psbE-petL); internal transcribed spacer (ITS); a region spanning exon 18–24 in the second largest subunit of RNA polymerase II (RPB2).

age as suggested by Ho & Phillips (2009), and as applied, it constitutes a hard minimum boundary for the age of the group (see also Frajman et al., 2009). The long tail of the truncated lognormal prior distribution extends beyond the maximum age of the strata in which the fossil has been found, allowing for maximum ages that are significantly older. Initial analyses using a relaxed molecular clock in BEAST v.1.6.1 with uncorrelated lognormal distributed substitution rates New Phytologist (2016) 211: 719–734 www.newphytologist.com

for each branch showed that the posterior distribution of the SD of the clock rate variation did not include 0, and thus a strict molecular clock could be rejected (Drummond et al., 2007). All subsequent matK analyses were consequently modelled using a relaxed molecular clock. All other prior distributions used the defaults in BEAUTI v.1.6.1. The final analysis included four independent Markov chain Monte Carlo (MCMC) simulations run for 10 million generations with tree and parameter values Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

New Phytologist

Research 727

Fig. 2 Maximum clade credibility tree and 95% highest posterior density (HPD) age interval estimates of nodes inferred from 45 matK sequences representing the Caryophyllaceae subfamilies Alsinoideae and Caryophylloideae, using Polycarpon tetraphyllum from subfamily Paronychioideae as an outgroup. Internal node bars represent the 95% HPD age interval estimates for one fossil-calibrated node (node A) and the time to the most recent common ancestor (tMRCA) for the genus Lychnis (node B). Numbers associated with nodes are posterior probabilities. The scale is in million years. maturaseK (matK).

sampled every 1000 generations, resulting in a total of 10 000 trees, of which 2000 were discarded as the burn-in phase. To test the influence of the priors on the posterior estimates, one additional chain was run for 10 million generations without data, sampling only the priors. Chain convergence was confirmed by inspection of the MCMC samples from each run using TRACER v.1.5, and joint estimates were produced using LOGCOMBINER v. 1.6.1 (Drummond & Rambaut, 2007). In addition, topological convergence among pairs of MCMC runs was assessed using AWTY (Nylander et al., 2008). Phylogenetic dating and inference of the species tree A species tree and divergence times for the African high-elevation Lychnis and the closely related L. lagrangei, L. flos-cuculi, L. flosjovis and L. coronaria were co-estimated using the rps16 and psbE-petL sequences from this study and the nuclear ITS and RPB2 sequences generated using the same samples by Popp et al. (2008) in *BEAST (Heled & Drummond, 2010). In this analysis we were not able to include L. subintegra, which was found to be sister to the East African lineage in the matK Caryophyllaceae Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

analyses described above, because it had a high proportion of missing data. We do not consider this to be crucial for our conclusions, as L. subintegra is very closely related to our outgroup species, L. flos-cuculi. A total of 224 sequences were analysed (Table 3). Substitution models, clock models and tree models were unlinked among data sets, except for the rps16 and psbEpetL tree models, which were linked as these regions are inherited as a single linkage block. In addition, we ran an analysis that included the matK data set where tree models for all plastid gene partitions were linked. Because we were unable to produce matK sequences for all taxa (Tables 1, 2), this particular analysis included a large proportion of missing data and we focus further analyses and discussion on the species tree analyses that did not include matK. The posterior age estimate for the Lychnis crown group obtained from the fossil-calibrated matK BEAST analysis was used as prior age in the *BEAST species tree analysis. The XML input file was manually edited to include a prior for the species tree root age following a normal distribution approximating the posterior estimate of the age of Lychnis from the matK analysis. The normal distribution was set with a mean value of 7.87 and an New Phytologist (2016) 211: 719–734 www.newphytologist.com

New Phytologist

728 Research Table 3 Data for the five DNA regions (total number of characters) used in the present study matK (1566 bp)

Number of terminals Conserved sites Variable sites Parsimony informative sites

Total data set 46 778 741 429

rps16 (666 bp)

Lychnis

African high-elevation Lychnis

Total data set

14 1211 85 44

9 1254 41 26

57 616 41 28

Lychnis


Total data set

Lychnis


57 616 41 28

52 632 25 22

57 1160 90 69

57 1160 90 69

52 1189 56 53

ITS (602 bp)

Number of terminals Conserved sites Variable sites Parsimony informative sites

psbE-petL (1253 bp)

RPB2 (1842 bp)

Total data set

Lychnis


35 500 100 63

35 500 100 63

26 539 58 37

Total data set

Lychnis


29 1567 231 134

29 1567 231 134

24 1686 117 83

maturaseK (matK); a ribosomal protein S16 (rps16); an intergenic spacer (psbE-petL); internal transcribed spacer (ITS); a region spanning exon 18–24 in the second largest subunit of RNA polymerase II (RPB2).

SD of 1.25, which corresponds to the 95% highest posterior densities (HPDs) of the age estimates from the four BEAST runs of the calibrated matK data set (5.34–10.55). Preliminary *BEAST analyses were run to test how well the data sets fitted a strict molecular clock as described above for the matK analysis. Strict clocks could not be rejected for ITS and rps16; the remaining data sets were analysed using uncorrelated lognormal clocks. The final *BEAST analysis was performed using constant population size coalescent models for each species with an autosomal nuclear ploidy type set for ITS and RPB2 and a mitochondrial type for the linked rps16 and psbE-petL. The species were linked with a Yule tree prior. Seven independent analyses were run for 150 million generations and trees and parameter values were sampled every 15 000th generation, resulting in a total of 10 000 trees of which 2000 were discarded as the burn-in phase. To test the influences of the prior on the posterior estimate, two additional chains were run for 300 million generations without data, sampling only the prior. Convergence of the chain to stationary distributions was confirmed by inspection of the MCMC samples in each analysis using the program TRACER 1.5 (Rambaut & Drummond, 2007) and joint estimates were produced using LOGCOMBINER v.1.6.1 (Rambaut & Drummond, 2007). Topological convergence among pairs of MCMC runs was also assessed using AWTY (Nylander et al., 2008). To avoid obviously violating the assumption of nonreticulate evolution in *BEAST, we included plastid data only for the LGS L. abyssinica, which together with L. scottii was inferred as a sister group to L. kigesiensis and the SGS L. abyssinica (cf Fig. 3).

Results Detailed information on each DNA region, including the number of characters and terminals, is presented in Table 3. New Phytologist (2016) 211: 719–734 www.newphytologist.com

Dating and phylogenetic inference based on matK GTR + G was selected as the best-fitting model for the matK data set including the additional Alsinoideae, Caryophylloideae and Paronychioideae representatives. Inspection of the MCMCs upon completion of the BEAST analyses showed that all individual runs had converged and the effective sample size (ESS) was > 200 for all parameters (> 1000 in most cases). Moreover, the results from AWTY indicated that pairs of MCMC runs converged on the same or highly similar tree topologies. The maximum clade credibility tree with estimated node ages based on the combined result from all runs is presented in Fig. 2. The time to the most recent common ancestor (tMRCA) of the genus Lychnis was estimated to be 8.0 Myr (95% HPD 5.5–10.7; node B in Fig. 2). The model analysed without data returned the prior distributions and we therefore conclude that the results were dominated by the data. Multigene Lychnis species tree and molecular dating GTR + G was again the best-fitting model for matK (including only Lychnis taxa), aK3Puf + G for rps16, aK3Puf + I for psbEpetL, SYM + G for ITS, and HKY + G for RPB2. The results from preliminary *BEAST analyses suggested strict clock models for the ITS, rps16 and matK regions. Uncorrelated relaxed lognormal clocks were used for the RPB2 and psbE-petL regions. Inspection of the *BEAST MCMCs showed convergence of model parameters in all individual analyses with ESSs > 200 for all parameters (> 1000 for most of them). Also in this case, the results from AWTY indicated that pairs of MCMC runs were converging on the same or highly similar tree topologies. Likewise, the model analysed without data resulted in recovering the prior distributions, indicating that the results were dominated by the data. Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

New Phytologist

Research 729

Fig. 3 Multilocus species tree for the genus Lychnis inferred from four DNA regions (rps16, psbE-petL, ITS and RPB2). Grey internal node bars represent the 95% highest posterior density (HPD) age intervals, co-estimated with the multilocus species tree. Nodes A–G and their estimated ages (see Table 4 for details) are discussed in the text. Numbers associated with nodes are posterior probabilities. The scale is in million years. White bars on terminal branches for the afromontane species represent the 95% HPD age interval ranges and the species mean age co-estimated for each DNA region (vertical lines inside the white bars) in the multilocus species tree (see Table 5 for details). Lychnis abyssinica LGS and L. abyssinica SGS denote the large-genome-size and small-genome-size variants of L. abyssinica, respectively. A ribosomal protein S16 (rps16); an intergenic spacer (psbE-petL); internal transcribed spacer (ITS); a region spanning exon 18–24 in the second largest subunit of RNA polymerase II (RPB2).

Table 4 Mean, median, 95% highest posterior density (HPD) interval in million years, and effective sample sizes (ESSs) for nodes A–G in the species tree (Fig. 3) Node

Mean

Median

95% HPD

ESS

A B C D E F G

5.16 4.12 2.49 1.50 0.48 1.71 4.32

5.11 4.09 2.43 1.46 0.42 1.67 4.28

2.92–7.68 2.18–6.53 1.21–3.95 0.63–2.61 0.14–0.96 0.70–2.92 2.27–6.67

2814 2492 2712 4325 5056 3154 3654

The species tree supported monophyly of an African highaltitude lineage of Lychnis including two well-supported clades: one diploid (L. abyssinica, L. kigesiensis, L. scottii, and L. rotundifolia) and one polyploid (L. crassifolia and L. kiwuensis), consistent with Popp et al. (2008). The analysis recovered two distantly related plastid lineages in L. abyssinica (Fig. 3; Supporting Information Figs S1a–c, S2). The Moroccan L. lagrangei and Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

the Eurasian L. flos-cuculi formed a sister group to the African high-elevation clade. Although consistently recovered in our analyses, the clade L. lagrangei + L. flos-cuculi was well supported only by the results from RPB2 (Fig. S1c). The tMRCA of the highaltitude African Lychnis and its sister lineage was estimated to be 5.2 Myr (95% HPD 2.92–7.68; node A in Fig. 3 and Table 4). The tMRCA for the high-altitude African Lychnis was estimated to be 4.1 Myr (95% HPD 2.18–6.53; node B in Fig. 3 and Table 4). The tMRCAs for the diploid and polyploid lineages were estimated to be 2.5 Myr (95% HPD 1.21–3.95; node C) and 1.5 Myr (95% HPD 0.63–2.61; node D), respectively. The two analyses with (Fig. 3; Table 3) and without (Fig. S2; Table S1) matK resulted in the same topology and very similar age estimates. The analysis that included matK showed higher uncertainty, probably as a result of the high proportion of missing data in the matK partition. An interesting pattern recovered in all analyses, particularly apparent in the results from the most densely sampled gene matrix (rps16/psbE-petL), was the very young ages for the crown group of each of the African high-altitude species ranging from New Phytologist (2016) 211: 719–734 www.newphytologist.com

730 Research

0.08 Myr (95% HPD 0.00–0.19 for L. kigesiensis) to 0.35 Myr (95% HPD 0.10–0.65 for L. abyssinica LGS; Fig. 4; Table 5). By contrast, the species were well defined and much older, as indicated by their long stem branches. For example, both L. kiwuensis and L. kigesiensis appeared to have colonized East Africa very recently (0.3 Ma or even later; Fig. 4; Table 5) and still share haplotypes with their distant putative source populations in Ethiopia. Similarly, the highly disjunct populations of L. abyssinica occurring in West Africa appeared to have originated from Ethiopian populations in the late Pleistocene.

Discussion Origin and diversification of Lychnis in the African ‘sky islands’ The multilocus species tree presented here (Fig. 3) largely corroborates the single and concatenated gene tree analyses in our

New Phytologist previous study (Popp et al. (2008), including the rejection of a direct sister group relationship between the African highelevation Lychnis and the only other African species of Lychnis, the Moroccan endemic L. lagrangei. We have thus corroborated the hypothesis of two independent colonizations of Africa by Lychnis (Popp et al., 2008). However, the inferred position of L. lagrangei differs between the present study and that of Popp et al. (2008). The concatenated analysis of Popp et al. (2008) resolved the Eurasian L. flos-cuculi as immediate sister to the African ‘sky island’ Lychnis, with L. lagrangei as sister to this group, while in our results L. lagrangei and L. flos-cuculi form a clade that is the sister to the ‘sky island’ lineage. We consider our new result to be more reliable because it is based on a larger data set, in terms of both number of specimens and number of loci. In the current study, we increased the sample size with the inclusion of two additional samples of L. lagrangei from herbarium specimens (Table 1). In addition, phylogenetic analysis of concatenated sequence data may be misleading, in particular if the

Fig. 4 Single-locus trees for the genus Lychnis, with 95% highest posterior density (HPD) age interval estimates for nodes inferred from sequences of the rps16/psbE-petL combined data matrix. Node bars represent 95% HPD age intervals. Nodes A–G and their estimated ages (see Table 5 for details) are discussed in the text. Numbers associated with nodes are posterior probabilities. The scale bar indicates million years. Lychnis abyssinica LGS and L. abyssinica SGS denote the large-genome-size and small-genome-size variants of L. abyssinica, respectively. A ribosomal protein S16 (rps16); an intergenic spacer (psbE-petL). New Phytologist (2016) 211: 719–734 www.newphytologist.com


New Phytologist

Research 731

Table 5 Mean, median, and 95% highest posterior density (HPD) interval in million years for the time to the most recent common ancestor (tMRCA) of each species inferred from ITS, RPB2, matK, and the combined psbE-petL/rps16 co-estimated in the multilocus species tree ITS

RPB2

matK

psbE-petL/rps16

Species

Mean

Median

95% HPD

Mean

Median

95% HPD

Mean

Median

95% HPD

Mean

Median

95% HPD

Lychnis abyssinica SGS L. kigesiensis L. scottii L. abyssinica LGS L. rotundifolia L. kiwuensis L. crassifolia

0.28 0.24 na na na 0.84 na

0.25 0.21 na na na 0.79 na

0.07–0.54 0.01–0.53 na na na 0.30–1.44 na

0.43 0.15 na na na 0.73 na

0.40 0.13 na na na 0.68 na

0.13–0.80 0.01–0.37 na na na 0.23–1.33 na

na 0.09 na 0.10 0.08 0.33 0.09

na 0.07 na 0.06 0.05 0.29 0.06

na 0.00–0.23 na 0.00–0.29 0.00–0.26 0.03–0.71 0.00–0.30

0.13 0.08 0.28 0.35 0.16 0.24 0.21

0.11 0.06 0.24 0.32 0.13 0.21 0.18

0.02–0.30 0.00–0.19 0.03–0.64 0.10–0.65 0.01–0.39 0.05–0.48 0.03–0.46

na, not available (species was represented by fewer than two sequences for a specific DNA region and no tMRCA was inferred). The inferred gene trees are presented in Fig. 4 and Supporting Information Fig. S1(a–c). Lychnis abyssinica LGS and L. abyssinica SGS denote the large-genome-size and smallgenome-size variants of L. abyssinica, respectively. maturaseK (matK); a ribosomal protein S16 (rps16); an intergenic spacer (psbE-petL); internal transcribed spacer (ITS); a region spanning exon 18–24 in the second largest subunit of RNA polymerase II (RPB2).

internal branches are short (Kubatko & Degnan, 2007). Popp et al. (2008) used maximum parsimony to analyse a concatenated data set of both plastid and nuclear DNA regions, whereas the present study used a coalescent-based approach that may be more robust to incomplete lineage sorting. Both hypotheses, however, reject a single colonization of Africa. Our molecular dating analysis resulted in an estimated mean age for the stem of the African high-elevation lineage of Lychnis of 5.2 Myr (95% HPD 2.92–7.68 Myr; node A in Fig. 3 and Table 4), thus clearly rejecting a hypothesis of a Pleistocene origin. Rather, our results suggest an origin in the late Miocene to early Pliocene, coinciding with a period of increased aridity in East Africa at or after 6–7 Ma (Cerling et al., 1993, 1997). Aridification happened again in the late Pliocene c. 3 Ma (Bobe, 2006). With minor variations, warm climate has been predominant during the last 5 Myr (Wara, 2005) and several studies have shown that trees and shrubs dominated East African plant communities during this period (e.g. Corlett, 2014). These climatic conditions also had an effect on the EARS, where a warm and humid climate towards the end of the Miocene has been inferred based on the paucity of grass pollen in Gonder in the northwestern Ethiopian highlands (Yemane et al., 1985). Evidence from the Turkana Basin of Kenya indicates that there was a humid closed rainforest environment c. 4 Ma (Bonnefille, 1995; Bobe, 2006). The rainforest expanded in eastern Africa (Williamson, 1985; Pickford et al., 2004) and lasted until the late Pliocene (c. 3.4 Ma) in the Ethiopian highlands (Woldegabriel et al., 1994; Bonnefille, 1995). Most of the 95% HPD of the estimated time of origin and initial diversification of the high-elevation African lineage of Lychnis corresponds to this warm and moist period. This also fits well with the preferences for moist habitats of the extant species; most of them grow in swampy areas and in bogs. It is, therefore, possible that the ancestor of these cold-adapted species dispersed from Eurasia to eastern Africa and diversified into two lineages at the time when the first high mountains started to rise. The capability of some of them (Table 1) to grow at lower elevations (as low as 1500 m under current conditions; Ousted, 1985; A. Gizaw, pers. obs.) shows that they can tolerate Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

a wide range of climatic conditions. Such environmental flexibility may have played an important role in the early establishment of these plants in the EARS mountains. It was also during this period that most of the eastern African mountains that today are high enough to sustain subalpine and alpine vegetation started to rise (Wichura et al., 2010). Interspecific divergence within the two main lineages of ‘sky island’ Lychnis commenced c. 2.5 Ma (95% HPD 1.21–3.95 Ma; diploids) and 1.5 Ma (95% HPD 0.63–2.61 Ma; polyploids). These time estimates roughly fit with renowned periods of aridification in East Africa peaking at 2.8, 1.7, and 1.0 Ma (Cane & Molnar, 2001; deMenocal, 2004; Bobe, 2006; Sepulchre et al., 2006). Thus, habitat fragmentation triggered by aridification and subsequent allopatric speciation in widely distributed ancestors may explain the diversification pattern in both diploids and tetraploids. However, the 95% HPD divergence time estimates of both lineages span times beyond these arid intervals and overlap with periods known to have been more humid. Such more humid periods in Africa occurred c. 2.7–2.5, 1.9–1.7 and 1.1– 0.9 Ma (Trauth et al., 2005). Thus, the uncertainty in time estimates does not allow us to distinguish between the two alternative hypotheses of diversification during arid periods and diversification under humid periods. The distributions of divergence times inferred from our analyses indicate that early Pliocene diversification may be as likely as middle Pleistocene diversification, suggesting that speciation within the African Lychnis may not be triggered solely by the global climatic oscillations of the Plio-Pleistocene. The rise of high mountains that provide suitable, but highly fragmented habitats with cooler climate may have been one of the main drivers of diversification and range dynamics in African Lychnis. Our previous results (Popp et al., 2008) also favoured multiple dispersal hypotheses including long-distance dispersal from eastern to western Africa. The current study suggests that such dispersal events might have occurred very recently in the late Pleistocene (within the last 0.3 Myr), when for example L. kiwuensis and L. kigesiensis appear to have expanded southwards from Ethiopia into the distant high mountains of Uganda. In each case, a single Ethiopian New Phytologist (2016) 211: 719–734 www.newphytologist.com

New Phytologist

732 Research

population may have served as a source for dispersal (e.g. Fig. 4). However, because of the use of a single minimum age constraint in dating the phylogeny, our results might be underestimating the ages of evolutionary events. Further dating refinement aimed at resolving uncertainties in age estimates, particularly close to the tips of the phylogeny, is needed to gain deeper insight into the timing of these recent dispersals. While the ‘sky island’ species of Lychnis are distinctly differentiated from each other, their intraspecific variation is often very low and results in short terminal branches within species (Fig. 4). This pattern is most apparent in younger lineages such as L. kiwuensis and L. kigesiensis, while older lineages (e.g. L. rotundifolia and L. scottii) show higher intraspecific variation (Fig. 4; Table 5). This may be indicative of small effective population sizes at the initial stages of speciation (as typically resulting from long-distance colonization of a mountain by few propagules) and lack of sudden population expansions afterwards, but additional population-level data would be necessary to explicitly test this hypothesis. Origins and evolution of the eastern African ‘sky island’ flora – the Lychnis perspective The results presented here are consistent with Merckx et al ’s. (2015) findings, which underline the importance of dispersal of lineages pre-adapted to cold conditions and local speciation in the process of the generation of high-altitude species communities on isolated tropical mountains. Our findings also largely corrborate Linder’s (2014) suggestion that the afro-alpine flora is among the youngest in Africa and contains many lineages of Eurasian ancestry. However, we found strong support for a late Miocene to early Pliocene arrival and initial diversification of Lychnis, challenging the idea that, as a whole, the afro-alpine flora is a product of recent colonization events and the climatic oscillations of the Plio-Pleistocene period (Livingstone, 1962; Hamilton, 1982; Mohammed & Bonnefille, 1998; Gottelli et al., 2004; Koch et al., 2006). We have shown that at least some alpine plants, such as Lychnis, may have colonized eastern Africa as soon as the first high mountainous habitats where frost occurred were formed, or even some time before that. Widespread species that are capable of surviving under a wide range of conditions, such as present-day L. kiwuensis (found from 3100 to 1800 m), may have established in eastern Africa first and then diversified in the region as soon as the cooler high-elevation habitat expanded with the formation of higher mountains. The age inferred for the stem of the clade including L. kiwuensis in our phylogeny is consistent with such a hypothesis (see also Popp et al., 2008). Pre-Pleistocene origins have also been suggested for several East African plant genera such as Isolona and Monodora (Couvreur et al., 2008) and Saintpaulia (Dimitrov et al., 2012). However, Saintpaulia has its origin in Africa and much of the afro-montane area that it currently occupies (e.g. the Eastern Arc Mountains) has been under direct climatic influence from the Indian Ocean and remained stable despite the climatic oscillations of the Pleistocene (Fjelds a & Lovett, 1997; Hewitt, 2000). Because the range of the ‘sky island’ Lychnis in eastern Africa New Phytologist (2016) 211: 719–734 www.newphytologist.com

overlaps to some degree with these stable areas, their importance in the process of speciation in Lychnis cannot be ruled out and needs to be further investigated. Our results are consistent with the hypothesis that Lychnis colonized Africa via the Arabian Peninsula and later dispersed southwards and westwards within Africa (Popp et al., 2008). Dispersal over the Arabian Peninsula (either as long-distance or steppingstone dispersal) has been suggested for other lineages that have colonized Africa from Eurasia, such as Alchemilla (Gehrke et al., 2015), Canarina (Mairal et al., 2015) and A. alpina (Koch et al., 2006). However, Canarina is restricted to the montane forest zone and is not adapted to night frost. That plants with different natural histories and timings of colonization of Africa apparently have used the same dispersal route underlines the importance of the Arabian Peninsula and its high mountains as one of the main ‘gates’ to eastern Africa from Eurasia. Our results provide further support for a Eurasian origin of at least parts of the high-altitude African flora and suggest that such elements were recruited at different times. The afro-alpine and arctic-alpine A. alpina seems to have dispersed twice to eastern Africa during the Pleistocene, the second time very recently (Koch et al., 2006), although this has so far not been corroborated using fossil-calibrated molecular data. Here, we have shown that dispersal of cold-adapted plants to eastern African mountains from Eurasia happened long before the onset of the Pleistocene glaciations. Our results also demonstrate the importance of dispersals and diversification within Africa, even very recently during the late Pleistocene, in shaping the present African ‘sky island’ plant communities, where local diversification in several genera has resulted in high numbers of endemics restricted to higher altitudes (e.g. Gehrke & Linder, 2009). As a result, although less diverse than the afro-montane flora of the Eastern Afromontane Biodiversity Region, the contemporary highaltitude African flora is unique in its origins and species composition. The combination of unique evolutionary history and high degree of endemism underlines the importance of the African ‘sky islands’ as a reservoir of unique yet fragile diversity.

Acknowledgements The study was funded by The Norwegian Programme for Development, Research and Higher Education (NUFU; project 2007/ 1058 grants to S.N. and C.B., entitled ‘AFROALP-II – Afroalpine ‘sky islands’: genetic vs taxonomic biodiversity, climate change, and conservation’). We thank Birgit Frosch for providing material of Lychnis lagrangei and Peter Linder for his valuable comments on an earlier version of the manuscript. We also thank the Editor, J. Vamosi, and three anonymous reviewers for their helpful comments and swift handling of the manuscript.

Author contributions A.G., C.B., S.N. and M.P. planned and designed the research. A.G. and M.P. performed the experiments and analysed data. A.G., C.B., S.N., T.W., C.A.M., F.M.T., A.A.A. and M.P. conducted fieldwork. A.G., C.B., M.P. and D.D. wrote the Ó 2016 The Authors New Phytologist Ó 2016 New Phytologist Trust

New Phytologist manuscript. B.O. provided material and commented on the manuscript.

References Antonelli A. 2009. Have giant lobelias evolved several times independently? Life form shifts and historical biogeography of the cosmopolitan and highly diverse subfamily Lobelioideae (Campanulaceae). BMC Biology 7: 82. Assefa A, Ehrich D, Taberlet P, Nemomissa S, Brochmann C. 2007. Pleistocene colonization of afro-alpine ‘sky islands’ by the arctic-alpine Arabis alpina. Heredity 99: 133–142. Bittrich V. 1993. Caryophyllaceae. In: Kubitzki K, Rohwer I, Bittrich V, eds. The families and genera of vascular plants. Berlin, Germany: Springer Verlag, 206– 236. Bobe R. 2006. The evolution of arid ecosystems in eastern Africa. Journal of Arid Environments 66: 564–584. Bonnefille R. 1995. A reassessment of the Plio-Pleistocene pollen record of East Africa. In: Vrba ES, Denton GH, Partridge TC, Burckle LH, eds. Paleoclimate and evolution with emphasis on human origins. New Haven, CT, USA: Yale University Press, 299–310. Cadena CD, Kozak KH, Gomez JP, Parra JL, McCain CM, Bowie RCK, Carnaval AC, Moritz C, Rahbek C, Roberts TE et al. 2011. Latitude, elevational climatic zonation and speciation in New World vertebrates. Proceedings of the Royal Society of London, Series B: Biological Sciences 279: 194– 201. Cane MA, Molnar P. 2001. Closing of the Indonesian seaway as a precursor to east African aridification around 3–4 million years ago. Nature 411: 157–162. Carbutt C, Edwards TJ. 2015. Reconciling ecological and phytogeographical spatial boundaries to clarify the limits of the montane and alpine regions of sub-Sahelian Africa. South African Journal of Botany 98: 64–75. Ceccarelli FS, Menegon M, Tolley KA, Tilbury CR, Gower DJ, Laserna MT, Kasahun R, Rodriguez-Prieto A, Hagman R, Loader SP. 2014. Evolutionary relationships, species delimitation and biogeography of Eastern Afromontane horned chameleons (Chamaeleonidae: Trioceros). Molecular Phylogenetics and Evolution 80: 125–136. Cerling TE, Harris JM, MacFadden BJ, Leakey MG, Quade J, Eisenmann V, Erleringer JR. 1997. Global vegetation change through the Miocene/Pliocene boundary. Nature 389: 153–158. Cerling TE, Wang Y, Quade J. 1993. Expansion of C4 ecosystems as an indicator of global ecological change in the Late Miocene. Nature 361: 344–345. Corlett RT. 2014. The ecology of tropical East Asia. Oxford, UK: Oxford University Press. Couvreur TLP, Chatrou LW, Sosef MSM, Richardson JE. 2008. Molecular phylogenetics reveal multiple tertiary vicariance origins of the African rain forest trees. BMC Evolutionary Biology 6: 54. Dauby G, Hardy OJ, Leal M, Breteler F, Stevart T. 2014. Drivers of tree diversity in tropical rain forests: new insights from a comparison between littoral and hilly landscapes of Central Africa. Journal of Biogeography 41: 574– 586. Dimitrov D, Nogues-Bravo D, Scharff N. 2012. Why do tropical mountains support exceptionally high biodiversity? The Eastern Arc mountains and the drivers of Saintpaulia diversity. PLoS ONE 7: e48908. Drummond AJ, Ho SYW, Rawlence N, Rambaut A. 2007. A rough guide to BEAST 1.4. Auckland, New Zealand: University of Auckland. Drummond AJ, Rambaut A. 2007. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evolutionary Biology 7: 214. Ebinger C, Yemane T, Harding D, Tesfaye S, Kelley S, Rex D. 2000. Rift deflection, migration, and propagation: linkage of the Ethiopian and Eastern rifts, Africa. Geological Society of America Bulletin 112: 163–176. Ehrich D, Gaudeul M, Assefa A, Koch MA, Mummenhoff K, Nemomissa S, Brochmann C. 2007. Genetic consequences of Pleistocene range shifts: contrast between the Arctic, the Alps and the East African mountains. Molecular Ecology 16: 2542–2559. Favre A, Packert M, Pauls SU, Jahnig SC, Uhl D, Michalak I, Muellner-Riehl AN. 2015. The role of the uplift of the Qinghai-Tibetan Plateau for the


Research 733 evolution of Tibetan biotas. Biological Reviews of the Cambridge Philosophical Society 90: 236–253. Fior S, Karis PO, Casazza G, Minuto L, Sala F. 2006. Molecular phylogeny of the Caryophyllaceae (Caryophyllales) inferred from chloroplast matK and nuclear rDNA its sequences. American Journal of Botany 93: 399–411. Fjelds a J, Lovett JC. 1997. Geographical patterns of old and young species in African forest biota: the significance of specific montane areas as evolutionary centres. Biodiversity and Conservation 6: 325–346. Frajman B, Eggens F, Oxelman B. 2009. Hybrid origins and homoploid reticulate evolution within Heliosperma (Sileneae, Caryophyllaceae) – a multigene phylogenetic approach with relative dating. Systematic Biology 58: 328–345. Gehrke B, Kandziora M, Pirie MD. 2015. The evolution of dwarf shrubs in alpine environments: a case study of Alchemilla in Africa. Annals of Botany 117: 121–131. Gehrke B, Linder HP. 2009. The scramble for Africa: pan-temperate elements on the African high mountains. Proceedings of the Royal Society of London, Series B: Biological Sciences 276: 2657–2665. Gehrke B, Linder HP. 2014. Species richness, endemism and species composition in the tropical Afroalpine flora. Alpine Botany 124: 165–177. Gottelli D, Marino J, Sillero-Zubiri C, Funk SM. 2004. The effect of the last glacial age on speciation and population genetic structure of the endangered Ethiopian wolf (Canis simensis). Molecular Ecology 13: 2275–2286. Griffiths CJ. 1993. The geological evolution of East Africa. In: Lovett JC, Wasser SK, eds. Biogeography and ecology of the rain forests of Eastern Africa. Cambridge, UK: Cambridge University Press, 19–23. Guindon S, Gascuel O. 2003. A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Systematic Biology 52: 696–704. Hamilton AC. 1982. Environmental history of East Africa: a study of the quaternary. London, UK: Academic Press. Hedberg O. 1951. Vegetation belts of the East African mountains. Svensk Botanisk Tidskrift 45: 140–202. Hedberg O. 1957. Afroalpine vascular plants. A taxonomic revision. Symbolae Botanicae Upsalienses 15: 1–411. Hedberg O. 1961. The phytogeographical position of the afroalpine flora. Recent Advances in Botany 1: 914–919. Hedberg O. 1965. Afroalpine flora elements. Webbia 19: 519–529. Hedberg O. 1969. Evolution and speciation in a tropical high mountain flora. Botanical Journal of the Linnean Society 1: 135–148. Hedberg O. 1970. Evolution of the afroalpine flora. Biotropica 2: 16–23. Heled J, Drummond AJ. 2010. Bayesian inference of species trees from multilocus data. Molecular Biology and Evolution 27: 570–580. Hewitt G. 2000. The genetic legacy of the Quaternary ice ages. Nature 405: 907– 913. Ho SY, Phillips MJ. 2009. Accounting for calibration uncertainty in phylogenetic estimation of evolutionary divergence times. Systematic Biology 58: 367–380. Hughes CE, Atchison GW. 2015. The ubiquity of alpine plant radiations: from the Andes to the Hengduan Mountains. New Phytologist 207: 275–282. Jabbour F, Renner SS. 2012. A phylogeny of Delphinieae (Ranunculaceae) shows that Aconitum is nested within Delphinium and that Late Miocene transitions to long life cycles in the Himalayas and Southwest China coincide with bursts in diversification. Molecular Phylogenetics and Evolution 62: 928–942. Jordan GJ, Macphail MK. 2003. A Middle-Late Eocene inflorescence of Caryophyllaceae from Tasmania, Australia. American Journal of Botany 90: 761–768. Kebede M, Ehrich D, Taberlet P, Nemomissa S, Brochmann C. 2007. Phylogeography and conservation genetics of a giant lobelia (Lobelia giberroa) in Ethiopian and Tropical East African mountains. Molecular Ecology 16: 1233–1243. Knox E, Palmer J. 1995. Chloroplast DNA variation and the recent radiation of the giant senecios (Asteraceae) on the tall mountains of eastern Africa. Proceedings of the National Academy of Sciences, USA 92: 10349–10353. Koch MA, Kiefer C, Ehrich D, Vogel J, Brochmann C, Mummenhoff K. 2006. Three times out of Asia Minor: the phylogeography of Arabis alpina L. (Brassicaceae). Molecular Ecology 15: 825–839.


New Phytologist

734 Research Kubatko LA, Degnan JH. 2007. Inconsistency of phylogenetic estimates from concatenated data under coalescence. Systematic Biology 56: 17–24. Leuschner C. 1996. Timberline and alpine vegetation on the tropical and warmtemperate oceanic islands of the world: elevation structure and floristics. Vegetation 123: 193–206. Linder HP. 2014. The evolution of African plant diversity. Frontiers in Ecology and Evolution 2: 1–14. Livingstone DA. 1962. Age of deglaciation in the Ruwenzori range, Uganda. Nature 194: 859–860. Loader SP, Sara Ceccarelli F, Menegon M, Howell KM, Kassahun R, Mengistu AA, Saber SA, Gebresenbet F, de Sa R, Davenport TRB et al. 2014. Persistence and stability of Eastern Afromontane forests: evidence from brevicipitid frogs. Journal of Biogeography 41: 1781–1792. Lovett JC, Bridson DM, Thomas DW. 1988. A preliminary list of the moist forest angiosperm flora of the Mwanihana Forest Reserve, Tanzania. Annals of the Missouri Botanical Garden 75: 874–888. Madri~ nan S, Cortes AJ, Richardson JE. 2013. Paramo is the world’s fastest evolving and coolest biodiversity hotspot. Frontiers in Genetics 4: 1–7. Mairal M, Pokorny L, Aldasoro JJ, Alarcon M, Sanmartın I. 2015. Ancient vicariance and climate-driven extinction continental-wide disjunctions in Africa: the case of the Rand Flora genus Canarina (Campanulaceae). Molecular Ecology 24: 1335–1354. deMenocal PB. 2004. African climate change and faunal evolution during the Pliocene-Pleistocene. Earth and Planetary Science Letters 220: 3–24. Merckx VSFT, Hendriks KP, Beentjes KK, Mennes CB, Becking LE, Peijnenburg KTCA, Afendy A, Arumugam N, de Boer H, Biun A et al. 2015. Evolution of endemism on a young tropical mountain. Nature 524: 347–350. Mittermeier RA, Gil PR, Hoffman M, Pilgrim J, Brooks T, Mittermeier CG, Lamoreux J, Da Fonseca GAB. 2005. Hotspots revisited: earth’s biologically richest and most endangered terrestrial ecoregions. Washington, DC, USA: Conservation International. Mohammed MU, Bonnefille R. 1998. A late Holocene pollen record from a highland peak at Tamsaa, Bale Mountains, South Ethiopia. Global and Planetary Change 16–17: 121–129. Mower JP, Touzet P, Gummow JS, Delph LF, Palmer JD. 2007. Extensive variation in synonymous substitution rates in mitochondrial genes of seed plants. BMC Evolutionary Biology 7: 135. N€ urk NM, Uribe-Convers S, Gehrke B, Tank DC, Blattner FR. 2015. Oligocene niche shift, Miocene diversification – cold tolerance and accelerated speciation rates in the St. John’s Worts (Hypericum, Hypericaceae). BMC Evolutionary Biology 15: 80. Nylander JAA 2004. MrModeltest version 2. Program distributed by the author. Uppsala, Sweden: Evolutionary Biology Centre, Uppsala University. Nylander JAA, Wilgenbusch JC, Warren DL, Swofford DL. 2008. AWTY (are we there yet?): a system for graphical exploration of MCMC convergence in Bayesian phylogenetics. Bioinformatics 24: 581–583. Orme CD, Davies RG, Burgess M, Eigenbrod F, Pickup N, Olson VA, Webster AJ, Ding T, Rasmussen PC, Ridgely RS et al. 2005. Global hotspots of species richness are not congruent with endemism or threat. Nature 436: 1016–1019. Ousted S. 1985. A taxonomic revision of the genus Uebelinia Hochst. (Caryophyllaceae). Bulletin du Jardin Botanique National de Belgique/Bulletin van de National Plantentuin van Belgi€e 55: 421–459. Oxelman B, Liden M, Rabeler RK, Popp M. 2000. A revised generic classification of the tribe Sileneae (Caryophyllaceae). Nordic Journal of Botany 20: 743–748. Pickford M, Senut B, Mourer-Chauvire C. 2004. Early Pliocene Tragulidae and peafowls in the Rift Valley, Kenya: evidence for rainforest in East Africa. Comptes Rendus Palevol 3: 179–189. Popp M, Gizaw A, Nemomissa S, Suda J, Brochmann C. 2008. Colonization and diversification in the African ‘sky islands’ by Eurasian Lychnis L. (Caryophyllaceae). Journal of Biogeography 35: 1016–1029.


Rahbek C, Graves GR. 2001. Multiscale assessment of patterns of avian species richness. Proceedings of the National Academy of Sciences, USA 98: 4534–4539. Rambaut A, Drummond AJ. 2007. Tracer, version 1.5. [WWW document] URL http://beast.bio.ed.ac.uk/Tracer [accessed 15 October 2014]. Rundel PW, Smith AP, Meinzer FC. 1994. Tropical alpine environments: plant form and function. Cambridge, UK: Cambridge University Press. Sepulchre P, Ramstein G, Fluteau F, Schuster M, Tiercelin JJ, Brunet M. 2006. Tectonic uplift and eastern Africa aridification. Science 313: 1419–1423. Sloan DB, Oxelman B, Rautenberg A, Taylor DR. 2009. Phylogenetic analysis of mitochondrial substitution rate variation in the angiosperm tribe Sileneae. BMC Evolutionary Biology 9: 1–16. Tamura K, Dudley J, Nei M, Kumar S. 2007. MEGA4: Molecular Evolutionary Genetics Analysis (MEGA) software version 4.0. Molecular Biology and Evolution 24: 1596–1599. Tolley KA, Tilbury CR, Measey GJ, Menegon M, Branch WR, Matthee CA. 2011. Ancient forest fragmentation or recent radiation? Testing refugial speciation models in chameleons within an African biodiversity hotspot. Journal of Biogeography 38: 1–13. Trauth MH, Maslin MA, Deino A, Strecker MR. 2005. Late cenozoic moisture history of East Africa. Science 309: 2051–2053. Wara MW. 2005. Permanent El Nino-like conditions during the Pliocene warm period. Science 313: 1739. Wichura H, Bousquet R, Oberh€a nsli R, Strecker MR, Trauth MH. 2010. Evidence for middle Miocene uplift of the East African plateau. Geology 38: 543–546. Wiens JJ, Donoghue MJ. 2004. Historical biogeography, ecology and species richness. Trends in Ecology and Evolution 19: 639–644. Williamson P. 1985. Evidence for an early Plio-Pleistocene rainforest expansion in East Africa. Nature 315: 487–489. Woldegabriel G, White TD, Suwa G, Renne P, Heinzelin JD, Hart WK, Heiken G. 1994. Ecological and temporal placement of early Pliocene hominids at Aramis, Ethiopia. Nature 371: 330–333. Yemane K, Bonnefille R, Faure H. 1985. Paleoclimatic and tectonic implications of Neogene microflora from the Northwestern Ethiopian highlands. Nature 318: 653–656.

Supporting Information Additional Supporting Information may be found online in the supporting information tab for this article: Fig. S1 Single-locus trees for the genus Lychnis from sequences of the matK, ITS, and RPB2 genes. Fig. S2 Multilocus species tree for the genus Lychnis inferred from five DNA regions. Table S1 Divergence time estimates for nodes in the species tree of Fig. S2 Please note: Wiley Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.
