Skip to main content
NCBI home page
Annals of Botany logoLink to Annals of Botany
. 2015 Oct 31;117(1):121–131. doi: 10.1093/aob/mcv159

The evolution of dwarf shrubs in alpine environments: a case study of Alchemilla in Africa

Berit Gehrke 1,*, Martha Kandziora 1, Michael D Pirie 1,2
PMCID: PMC4701153  PMID: 26520565

Abstract

Background and Aims Alpine and arctic environments worldwide, including high mountains, are dominated by short-stature woody plants (dwarf shrubs). This conspicuous life form asserts considerable influence on local environmental conditions above the treeline, creating its own microhabitat. This study reconstructs the evolution of dwarf shrubs in Alchemilla in the African tropical alpine environment, where they represent one of the largest clades and are among the most common and abundant plants.

Methods Different phylogenetic inference methods were used with plastid and nuclear DNA sequence markers, molecular dating (BEAST and RelTime), analyses of diversification rate shifts (MEDUSA and BAMM) and ancestral character and area reconstructions (Mesquite).

Key Results It is inferred that African Alchemilla species originated following long-distance dispersal to tropical East Africa, but that the evolution of dwarf shrubs occurred in Ethiopia and in tropical East Africa independently. Establishing a timeframe is challenging given inconsistencies in age estimates, but it seems likely that they originated in the Pleistocene, or at the earliest in the late Miocene. The adaptation to alpine-like environments in the form of dwarf shrubs has apparently not led to enhanced diversification rates. Ancestral reconstructions indicate reversals in Alchemilla from plants with a woody base to entirely herbaceous forms, a transition that is rarely reported in angiosperms.

Conclusions Alchemilla is a clear example of in situ tropical alpine speciation. The dwarf shrub life form typical of African Alchemilla has evolved twice independently, further indicating its selective advantage in these harsh environments. However, it has not influenced diversification, which, although recent, was not rapid.

Keywords: Adaptation to alpine environments, Alchemilla, alpine speciation, dwarf shrubs, Fragariinae, secondary woodiness, species evolution, Rosaceae dating, Rosoideae

INTRODUCTION

Alpine and arctic environments are usually characterized by the absence of tall woody plants such as trees and large shrubs. However, many plants do develop woody rhizomes or stems. Woody-stemmed (dwarf) shrubs are abundant and conspicuous in alpine and arctic environments, to the extent that they have been used as a proxy for the investigation of climatic changes in these areas (Myers-Smith et al., 2011; Bokhorst et al., 2012; Kaarlejarvi et al., 2012; Rayback et al., 2012; Preece and Phoenix, 2013; and many more). They are often the tallest plants occupying tundra ecosystems upslope (subalpine and alpine environments) or northward (arctic environments) of the treeline ecotone (Lantz et al., 2010) and can form dense thickets with closed canopies in suitable habitats (Myers-Smith et al., 2011). Dwarf shrubs are often most abundant in wet tussock tundra (Gorsuch et al., 2001), including areas of tropical Africa with alpine-like climatic conditions, i.e. the Afroalpine (Hedberg, 1951; Gehrke and Linder, 2014; Fig. 1). These Afroalpine regions are found above the dense forests on the high mountains of tropical Africa and form a series of ‘sky islands’ with a flora sharply distinct from that of the surrounding tropical lowlands (Hedberg, 1951; White, 1983). Climatic conditions in the Afroalpine differ more between daytime and night-time conditions (‘summer every day and winter every night’; Hedberg, 1964) than between different times of the year, while generally being characterized by a typical equatorial climate that is modified by elevation and exposure (Lauer, 1975). Frost can occur during clear nights throughout the year (Hemp, 2006) and plants growing here must be able to withstand regular freezing and extreme, sudden temperature differences (Hedberg, 1964). In addition to dwarf shrubs, these regions also harbour giant rosette plants – giant senecios (Dendrosenecio, Asteraceae) and giant lobelias (Lobelia, Campanulaceae) – that show unusual ‘woody’ structures.

Fig. 1.

Fig. 1.

Open in a new tab

Distribution of Afrotemperate (Afromontane and Afroalpine) regions in Africa adapted from Gehrke and Linder (2014) based on White (1983). All Afrotemperate areas are indicated in black, with more inclusive named regions and subregions within black and dashed lines respectively. Elevation is indicated by greyscale shading. Tropical Afroalpine areas are restricted to the Ethiopian highlands, tropical East Africa and West Africa.

Part of the success of dwarf shrubs, especially those above the treeline, can be attributed to the way in which they effectively modify their own microhabitat (Körner, 2012). Their low, dense habit creates a microclimate that is warmer than that immediately below the treeline (Körner, 2012) and that is associated with an increase in leaf-litter accumulation and decomposition rates (Cornelissen et al., 2007), greater nitrogen (N) availability or faster N cycling (Chu and Grogan, 2010), greater seedling survival by acting as nurse plants (Klanderud and Totland, 2004) and moderating the effect of wind (Körner, 2012; see also Table 2 in Myers-Smith et al., 2011). Dwarf shrubs have a reported effect on a wide range of ecosystem processes, including snow depth and persistence (including associated hydrological dynamics; Myers-Smith et al., 2011) and soil cooling through shading (Blok et al., 2010). A further important advantage of the dwarf shrub habit seems to be that of permanence (Billings and Mooney, 1968). Hence, this life form can be seen as an important adaptation to alpine environments (Boucher et al., 2012), despite the fact that it is also common in other habitats and that other factors have recently been reported to be the ‘key angiosperm strategies’ successful in alpine environments (e.g. deciduous leaves, small conduits and herbaceous habit; Zanne et al., 2014).

As in many other places of the world with alpine-like environments, most species with a shrubby stature in the Afroalpine belong to the families Asteraceae, Ericaceae and Rosaceae (Hedberg, 1964; Gehrke and Linder, 2014). Here we focus on dwarf shrubs that belong to Alchemilla (Rosaceae; lady’s mantle). Alchemilla has been shown to consist of four clades (Gehrke et al., 2008): Alchemilla s.s., Aphanes, Lachemilla and the Afromilla clade, which is mostly confined to Africa (but which probably also includes two species found in Sri Lanka, southern India and Java). With the exception of mostly annual Aphanes, Alchemilla species are perennial, with basal leaves arising from woody rhizomes. Many Alchemilla species are long-lived and can form more or less dense clumps or mats, but, with the exception of Lachemilla polylepis and a number of Alchemilla species in Africa, they do not develop into entirely woody shrubs with erect branches. The flowers of Alchemilla are generally inconspicuous individually, being greenish to yellow, small and without petals, but inflorescences can be fairly showy. Members of the Alchemilla s.s. clade are usually highly polyploid or aneuploid [2n = 64–106(–224); Fröhner, 1995], regularly reproduce clonally and are mostly thought to be apomictic, i.e. agamospermous, in the form of obligate gametophytic diplospory, in which endosperm formation and embryo development are completely independent of fertilization (Fröhner, 1995). Aphanes species have been reported to have chromosome counts of 2n = 16, 32 or 48 (Izmailow, 1999) and to reproduce at least partially through sexual reproduction. However, little is known about ploidy and the form of reproduction in Lachemilla and the Afromilla clade.

Eight of the 14 species of Alchemilla that occur commonly in the upper parts of the Afroalpine are dwarf shrubs that are entirely woody (Graham, 2000; Gehrke and Linder, 2014). The remaining six species are prostrate herbs with a woody rootstock and without erect woody branches, or are prostrate subshrubs (A. subnivalis). Some species, e.g. A. johnstonii, are intermediate, but they usually possess herbaceous branches in addition to being partly woody. All dwarf shrub species occur either exclusively or at least predominantly in the Afroalpine and are endemic to single mountain ranges or regions, whereas those species predominantly occurring outside the Afroalpine are all at least partially herbaceous and generally more widespread in their distributions. One exception to this rule is the shrubby A. argyrophylla, which is distributed in Kenya (Mt Kenya, Aberdares), southern Sudan (Imatong), Tanzania (Kilimanjaro) and Uganda (Rwenzori).

In this paper, we use phylogenetic inference methods with plastid and nuclear ribosomal (nr) DNA sequence markers. We test: (1) how often a true dwarf shrub life form has evolved in Alchemilla; (2) when dwarf shrubs evolved and whether they did so in situ in Africa (i.e. potentially as an adaptation to newly formed alpine environments); and (3) whether the dwarf shrub life form could be a key innovation that has led to enhanced diversification rates.

METHODS

Taxon sampling and datasets

We sampled all African Alchemilla species available to us (Supplementary Data Table S1). We included all currently accepted species from East and West Africa, five of eight species from Madagascar and eight of 16 species from southern Africa, in total 30 species. The African Plant Database (2009) lists 40 species plus four subspecies and two hybrid taxa as being validly accepted for Alchemilla in sub-Saharan Africa and Madagascar. Species we failed to include are all herbaceous and morphologically similar to other species in their respective regions. We included multiple accessions of species that were geographically widespread and/or morphologically variable. As outgroups we included representatives of all clades in Fragariinae according to Dobes and Paule (2010) and Potentilla reptans as a representative of Potentillinae, the sister clade of Fragariinae (Supplementary Data Table S2). To enable us to include fossil calibrations for molecular dating, we included species from all major lineages of Rosaceae and related families (Eriksson et al., 2003; Potter et al., 2007; Töpel et al., 2011, 2012). From the Rosaceae dataset a reduced Rosoideae dataset was created by excluding Amygdaloideae (with the exception of Oemleria, Prunus laurocerasus and Lyonothamnus, required for the fossil assignment). Phylogenetic analyses were performed for each region of the Fragariinae subsets independently and combined nr and combined plastid datasets to identify any instances of supported conflict between individual markers or gene trees as described in Gehrke et al. (2010). For African Alchemilla, we additionally combined all plastid and nr data, but excluded conflict by removing plastid or nr data for A. elgonensis, A. hagenia and A. triphylla. The newly described monotypic genus Chamaecallis (Eriksson et al., 2015) was excluded from analyses due to the availability of only ITS1 and part of trnL–F in GenBank. These markers provided insufficient data to place the taxon with any degree of certainty and its inclusion led to decreased overall support (data not shown).

DNA extraction, sequencing and matrices

Leaf material from collections dried in silica and herbarium samples was pulverized using a Retsch MM301 ball mill (Retsch, Haan, Germany). Total genomic DNA was extracted using the Qiagen DNeasy Plant Mini Kit (Qiagen, Valencia, CA, USA) following the manufacturer’s protocols with minor modifications. To obtain any polymerase chain reaction (PCR) products, the extracted DNA had to be diluted 1:100. PCR was performed in 25-µl reactions [1 × PCR buffer with 2·5 mm MgCl2, 0·25 mm dNTPs, 1·6 μm primers and 1 unit of Taq polymerase (New England Biolabs Inc., Ipswich, USA)] in a GeneAmp PCR System 2700 thermal cycler (Applied Biosystems Inc., Foster City, USA).

After an initial screening of several plastid markers, the plastid regions, the trnL–trnF intergenic spacer plus the trnL intron (trnL–F), rpoB–trnC and the rpl32–trnL intron were employed using primers described in Shaw et al. (2005, 2007). The internal transcribed spacer (ITS, composed of ITS1-5.8S-ITS2) and external transcribed spacer (ETS) regions of nrDNA were amplified using ITS18S (Muir et al., 2001), ITSA and C (Blattner, 1999) or ITS4 (White et al., 1990) and ETS1 and IGS8 (Oh and Potter, 2005). The Rosaceae and Rosoideae matrices for molecular dating analyses were assembled based on plastid data only (matK, rbcL and trnL–F regions), including mostly published sequences from GenBank (Supplementary Data Table S2).

PCR products were purified with ExoSap-IT PCR Clean-Up (Affymetrix, Santa Clara, CA, USA) or the Macherey-Nagel PCR Purification Kit (Macherey-Nagel, Düren, Germany). Cycle sequencing was carried out with BigDye Terminator 3.1 (Applied Biosystems, Foster City, CA, USA) using the same primers as for the amplifications with the exception of ETS, for which a new internal primer was designed for this study (IGS-Alch: 5′ GTT GTG TGG GTT GGC GGG CT 3′). The fluorescently labelled samples were run on an ABI 3130xl Genetic Analyzer at Johannes Gutenberg University Mainz (Germany) for sequencing. Sequencher 4.10.1 (Gene Codes, Ann Arbor, MI, USA) was used for trace file editing and sequences were submitted to GenBank (Supplementary Data Table S2). Sequence alignment was performed using Mesquite version 3.03 (Maddison and Maddison, 2009) and was edited by hand. We excluded ambiguous areas in the alignment, such as poly-A regions. Gaps were coded following the simple indel coding method of Simmons and Ochoterena (2000) using SeqState version 1.32 (Müller, 2005).

Phylogenetic analyses

Selection of the best-fitting partitioning strategies and substitution models for all maximum likelihood (ML) and Bayesian analyses was performed using PartitionFinder v.1.1.1 (Lanfear et al., 2014), employing a heuristic search strategy and comparison by means of the Bayesian information criterion (Supplementary Data Table S3).

Parsimony analyses were performed using PAUP* 4b10 (Swofford, 2002) with a heuristic search of 500 replicates, random sequence addition, tree bisection–reconnection (TBR) branch swapping and MULTREE on (keeping multiple, equally parsimonious trees), saving a maximum of 50 trees per replicate. Support was assessed using 1000 replicates of non-parametric bootstrap analysis using the same settings as in the heuristic search.

The ML tree searches were performed using RAxML-VI-HPC 7.2.6 (Stamatakis et al., 2008) with 500 bootstrap replicates using the CIPRES Science Gateway (Miller et al., 2010).

Ancestral character and ancestral area reconstruction

We coded the geographical locality of individual samples instead of the recorded distribution of species because some species proved not to be monophyletic. To avoid bias in the ancestral area reconstruction, we included in as many cases as possible more than one accession per species to reflect the widespread distribution ranges of taxa. Character reconstructions were performed using Mesquite on the plastid and nuclear datasets separately. We did parsimony reconstructions (using the StochChar module), which we deemed appropriate because the number of inferred character transitions was low (Pirie et al., 2012). To take into account phylogenetic uncertainty, the character states were reconstructed using 100 randomly selected trees from the RAxML bootstrap analysis and mapped onto the best likelihood tree of the same analysis.

We investigated the following. (1) Ancestral areas (coding: Ethiopia, East Africa, Madagascar, southern Africa, West Africa, western Eurasia, eastern Eurasia, Australia, Central and South America and North America). We also tested the impact of splitting East Africa into western and eastern East Africa. (2) Life form (dwarf shrubs, perennial with a woody base, entirely herbaceous perennial or annual). (3) Elevational niche preference (alpine-like habitat, montane region and lower elevation). The life-form analyses were repeated coding herbaceous perennials with or without a woody base under the same state. This distinction is more meaningful in seasonal temperate regions, where a woody base is indicative of above-ground persistence during a distinct resting season. It is less meaningful in the Afroalpine, where growth occurs all year round. For example, A. andringitrensis, A. elongata and A. johnstonii have a woody base but also non-woody parts: these are not considered as dwarf-shrubs here (see also Soltis et al., 2013 for a discussion on coding herbaceousness versus woodiness). For the niche based on elevation we considered the areas with alpine-like climatic environments in tropical East Africa above 3500 m, in southern Africa above 2500 m and in Europe and Asia above 2000 m as alpine in order to take into consideration the effect of latitude on climatic conditions. We also only considered a species as alpine if it can be found commonly in these areas but not if it occurs there only rarely, i.e. if a taxon can be considered to be an alpine element. The coding can be found in Supplementary Data Table S4.

Molecular dating

We used 85·8 million years (Ma) as the most conservative estimate for the minimum age of the Rosaceae crown group based on Crataegites borealis, the oldest known Rosaceae fossil (Samylina, 1960, as described in Töpel et al., 2012). Crepet et al. (2004) considered the crown age of Rosaceae to be between 90 and 94  Ma and Wang et al. (2009) estimated the Rosaceae stem node as 103 Ma (maximum 115 Ma).

A strict molecular clock was rejected based on a likelihood ratio test, given the general time-reversible model and best tree with and without a molecular clock constraint. Two different relaxed-clock dating approaches were used: BEAST v1.8.0 (Drummond et al., 2012) and RelTime v.2.0 (Tamura et al., 2012).

In BEAST we attempted to apply a lognormal relaxed-clock model implemented with the full Rosaceae dataset, but abandoned the analyses due to consistent failure of runs to converge. This was potentially due to the stark difference in substitution rates between Rosoideae and Amygdaloideae (Supplementary Data Fig. S1), which violates the assumption of a lognormal rate distribution (Dornburg et al., 2012). We therefore constructed the Rosoideae dataset with only a few representatives of other Rosaceae (Supplementary Data Fig. S2). We applied a uniform prior of 115·0–85·8 Ma for root height (stem age of Rosaceae) and applied exponential priors to represent additional fossils used to constrain the (minimum) ages of internal nodes: 85·8  Ma (mean, 20) for the Rosaceae crown node; 26·5 Ma (mean, 20) for Cercocarpus, placed at the crown node of Dryadoideae (de León and Cevallos-Ferriz, 2000); 26·5 Ma (mean, 20) for the stem node of Potentillinae (Wolfe and Schorn, 1989) and 2·5 Ma (mean, 10) for the crown node of Fragaria (Matthews and Ovenden, 1990). We used unlinked substitution models for partitions, unlinked clock models with an uncorrelated lognormal relaxed-clock and a Yule model. Two separate independent runs with chain length of 10 000 000 generations sampling every 1000 with independent partitioning and substitution model according to the results of PartitionFinder were combined. The two independent runs were checked for convergence and burn-in using Tracer (Rambaut et al., 2014) and combined excluding the standard burn-in of 10 % using LogCombiner v. 1.7.2 and TreeAnnotater v. 1.7.2 (Drummond et al., 2012). Trees were displayed using FigTree v. 1.4.0 (Drummond et al., 2012).

RelTime analyses were based on the best trees of the RAxML analyses using the full Rosaceae dataset, the Rosoideae subset and the Fragariinae dataset on the plastid and nuclear data separately. We ran two separate analyses for each dataset to represent an approximation of the calibration uncertainty. For the Rosaceae and the Rosoideae datasets we used 103·0 and 85·8  Ma for the maximum and minimum stem ages of Rosaceae. For the Fragariinae dataset we constrained the split between Fragariinae and Potentillinae to an age of 50–25 Ma as a conservative result of the Rosaceae dating analyses.

Diversification rate shifts

To get time-calibrated trees we used BEAST as outlined above on the plastid dataset of Fragariinae and constrained the root height according to the Rosoideae BEAST dating (Table 1) with a normal distribution, giving a mean of 49·2 and standard deviation of 4 (Supplementary Data Fig. S3). Diversification rate constancy was tested under ML using LASER version 2.2 (Rabosky, 2006) in R version 3.2.0 (R Development Core Team, 2015) using birth–death and density-dependent speciation models. Shifts in diversification rates were calculated using MEDUSA (Alfaro et al., 2009) as implemented in the GEIGER package version 2 (Pennell et al., 2014). We also used 100 randomly chosen trees from the posterior distribution of trees to account for the phylogenetic uncertainty in the MEDUSA analysis. Additionally we used BAMM 2.0 (Rabosky, 2014) and BAMMtools 2.0 (Rabosky et al., 2014). We used setBAMMpriors to adjust the priors according to the scaling of the tree. The initial speciation rate was calculated according to Magallon and Sanderson (2001) and set to 0·122 (50 Ma and 900 species) and extinction rate was set to 0·005. The Markov-Chain-Monte-Carlo chain was run for 10 000 000 generations, sampling every 1000th generation. Convergence and effective sample size values were assessed in R (burn-in = 10 %). The credible shifts and reconstruction of the mean of the marginal posterior density of speciation, extinction and net diversification rates across the tree were computed using BAMMtools. Both analyses were run (1) on a combined, time-calibrated Fragariinae tree generated with BEAST with missing taxa assigned to terminals in clades randomly according to species numbers (as given in the Supplementary Data Table S5) and (2) on the same tree pruned to well-supported clades outside the Afromilla clade with total species numbers of the clade assigned to that terminal.

Table 1.

Results of dating analyses using different datasets and dating methods

Dobes and Paule, 2010 (BEAST plastid) Töpel et al., 2012 (BEAST plastid) BEAST Rosoideae (plastid) RelTime Rosaceae (plastid)
Rosaceae crown 75–76·9 108–93 92·2 (98·5–85·8) 103–85·8
Rosoideae crown 66·5–50 95–70 65·5 (75·8–55·2) 43·8–36·5
Potentilleae stem 55·6–49 86·2–61·2 67·5 (88·5–46·5) 34·4–28·6
Potentillinae crown 24·6–16·7 68·2–43·1 35·6 (45·9–25·3) 20·7–17·2
Ivesioid clade crown 2·9–0·6 31·6–15·9 7·9 (11–4·7) 2·9–2·5
Fragariinae stem 53·4–45·1 78·7–52·8 49·2 (59–39·4) 29·2–24·4
Fragariinae crown 32·2–22·3 – 26·1 (33·7–18·4) 14·1–11·7
Fragaria crown – – 7·2 (11·5–2·9) 4·2–3·5
Alchemilla s.l. stem 29·1–20·4 – 31·4 (39·2–23·6) 10·1–8·4
Alchemilla s.l. crown 23·3 (30–16·5) 5–4·1
Alchemilla s.s. crown – – 6·7 (10·1–3·3) 1·2–1·0
Lachemilla crown – – 17·2 (23·8–10·6) 3·5–2·9
Afromilla stem – – 17·2 (23·8–10·6) 3·5–2·9
Afromilla crown 10·8 (15·9–5·7) 1·5–1·2
Haumanii clade crown – – 2·0 (3·9–0·01) 1·4–0·1
Open in a new tab

RESULTS

Direct sequencing, phylogenetic conflict and phylogenetic analyses

Amplification and sequencing were generally challenging, with problems commonly encountered for Alchemilla and related species (own observations and Diego Morales-Briones, University of Idaho, USA, pers. comm.), which might be related to the presence of secondary compounds. Alignment was generally straightforward in the ingroup. Sequences of a number of species showed that accessions formed monophyletic or, given limitations in phylogenetic resolution, not demonstrably poly-/paraphyletic groups (e.g. A. abyssinica, A. andringitrensis, A. haumanii, A. rothii). Other species were clearly not monophyletic (e.g. A. fischeri, A. kiwuensis, A. rutenbergii and A. subnivalis; Fig. 3).

Fig. 3.

Fig. 3.

Open in a new tab

Molecular phylogenetic reconstruction of the Afromilla clade based on combined plastid and nuclear ribosomal DNA data. Nuclear data for A. elgonensis, A. hagenia and A. triphylla have been removed due to conflicting positions in the separate nr and plastid analyses (see Supplementary Data Figs S3 and S4 for the separate gene trees). The best scoring maximum likelihood tree is shown here, with maximum likelihood bootstrap values ≥70 presented above the branches and maximum parsimony bootstrap values ≥70 below. Entirely woody dwarf shrubs with erect branches are indicated in bold and intermediate forms with non–woody tips of branches or only prostrate growth are underlined. Names are given for well supported clades of importance to the discussion.

The topologies and support values resulting from parsimony, ML and relaxed clock Bayesian inference were more or less consistent. The inferred relationships of taxa in Fragariinae (Fig. 2 and Supplementary Data Figs S4 and S5) corresponded to those presented in previously published phylogenetic trees (Eriksson et al., 2003; Potter et al., 2007; Dobes and Paule, 2010; Töpel et al., 2012). There was strongly supported incongruence between plastid and nuclear trees (Fig. 2), as reported previously in Fragariinae (Lundberg et al., 2009; Eriksson et al., 2015) with the exception of the placement of Aphanes arvensis in Alchemilla s.s., shown by Lundberg et al. (2009), which was not supported in our analyses. We did not combine the nr and plastid data of Fragariinae due to conflicting signals (Fig. 2), but did perform a combined analysis of the African Alchemilla material after removing the nuclear data of A. elgonensis, A. triphylla and A. hagenia (Fig. 3 and Supplementary Data Fig. S6). There was strong support for the monophyly of the Haumanii clade and two Madagascan clades and some support for the dwarf shrub clade and a southern African clade with the exception of widespread species such as A. kiwuensis and A. cryptantha (Fig. 3). The relative position of these clades was not supported.

Fig. 2.

Fig. 2.

Open in a new tab

Conflict between plastid and nuclear data in Fragariinae. Support values above the branches are parsimony bootstrap values; those below the branches on the left are maximum likelihood bootstrap values and Bayesian posterior probability are given below the branches on the right. Asterisks indicate support of 100 and 1·0 bootstrap and posterior probability, respectively, in all three analyses.

Ancestral area reconstruction

The origin of Fragariinae was most likely in eastern Asia (node 1 in Fig. 4), whereas that of Alchemilla s.l. (node 2) remains unclear, with the majority of reconstructions equivocal (Supplementary Data Table S6 and Fig. S7). Alchemilla is inferred to have colonized Africa by long-distance dispersal to East Africa from an unknown source, potentially Eurasia (Gehrke and Linder, 2009). From the East African mountains, Ethiopia was colonized at least three times independently. The clade of species endemic to southern Africa most likely represents a single colonization event from tropical East Africa. Widespread species (e.g. A. cryptantha and A. kiwuensis) represent independent colonization events. West Africa was colonized at least twice independently. Madagascar was colonized at least once, but potentially twice or more independently. The African accession of Aphanes is inferred to constitute an independent colonization of the Afroalpine from western Eurasia. The exact area of origin in tropical East Africa of the dwarf shrub clade was not reconstructed unequivocally.

Fig. 4.

Fig. 4.

Open in a new tab

Ancestral state reconstructions in Mesquite. Maximum parsimony reconstructions on 100 bootstrap trees from the maximum likelihood analyses of the combined plastid analysis are mapped onto the most likely tree. The scale bar indicates age in millions of years before present based on the RelTime analysis, and age ranges are given at the nodes. Pie charts from left to right represent origin, elevation and life form.

Ancestral character reconstruction

Parsimony optimization of elevation suggests that Fragariinae (node 1 in Fig. 4) evolved and diversified mostly at mid-elevations (see also Supplementary Data Table S7). The origin of Alchemilla s.l. (node 2) and of Afromilla (node 3) at a particular elevation range cannot be identified with certainty as most nodes are equivocal between alpine and mid-elevation.

The ancestral life form in Fragariinae (node 1) is most often reconstructed under parsimony to have been a perennial herb without a woody rootstock. The most recent common ancestor of Alchemilla s.l. probably had a woody base (node 2 in Fig. 4, Supplementary Data Table S8). Dwarf shrubs seem to have evolved in Fragariinae several times independently and at least twice in the Afromilla clade (nodes 4 and 5 in Fig. 4). Additionally, ancestral reconstructions indicate reversals in Fragariinae, especially in Alchemilla s.s. and in the Afromilla clade, from plants with a woody base to entirely herbaceous forms.

Molecular dating

Our age estimates in Rosaceae (Table 1 and Supplementary Data Table S9 and Figs S3 and S4) differed significantly from ages inferred by Töpel et al. (2012) and Dobes and Paule (2010). Using BEAST on the plastid Rosoideae dataset we estimated the 95 % highest posterior density of the crown age of Rosaceae as 98·5–85·8 Ma. The split between Alchemilla and Comarum/Farinopsis is estimated to have occurred 39·2–23·6 Ma ago. Lachemilla and African Alchemilla separated 23·8–10·6 Ma ago and the diversification of Alchemilla in Africa (crown age) started 15·9–5·7 Ma ago. RelTime dating of the same dataset shows the crown age of Rosaceae as 103·0–85·8 Ma, the split between Alchemilla and Comarum/Farinopsis as 10·1–8·4  Ma, Afromilla clade stem age as 3·5–2·9  Ma and the crown age of Alchemilla in Africa as 1·5–1·2 Ma.

Diversification rate shifts

A birth–death model was selected as the best-fitting model based on the corrected Akaike information criterion (Supplementary Data Table S10). MEDUSA using the 100 randomly sampled trees identified a total of three to five (to seven) diversification rate shifts in Fragariinae. Analyses of the most comprehensively sampled combined tree indicated four shifts (Supplementary Data Table S11A and Fig. S8): one at the branch leading to Alchemilla s.s., one at the crown node of Lachemilla, one at the crown of the Afromilla clade and one along the branch leading to the southern African clade. Similarly, analyses of the pruned tree (Supplementary Data Table S11B) revealed rate shifts at the crown node of the Alchemilla s.l. and related groups, one at the crown node of Alchemilla s.l. and one at the crown node of the southern African clade. These shifts do not obviously correspond with the evolution of the dwarf shrub habit (output details of all runs can be obtained from the authors upon request).

BAMM analyses indicated acceleration in the diversification rate leading to Alchemilla s.l. from the low rates and slightly negative net diversification in other Fragariinae clades (Fig. 5, Supplementary Data Fig. S9). Extinction rates seem to be constant through time, but speciation rates vary. We inferred three distinct configurations within the 95 % credible shift sets using the pruned tree and five using the large tree. Distinct diversification regimes were associated with the Alchemilla s.l. clade (pruned) or the dissected clade of Alchemilla s.s., but never with the dwarf shrub clade (Supplementary Data Table S10).

Fig. 5.

Fig. 5.

Open in a new tab

BAMM analyses of diversification rate changes, showing net diversification rates through time and among lineages of Fragariinae. The phylogeny shows a time-calibrated BEAST tree based on plastid data, with branches coloured by reconstructed net diversification rates. Rates on each branch are means of the marginal densities of branch-specific rates.

DISCUSSION

Independent evolution of dwarf shrubs in Africa

Ancestral character reconstructions show clearly that dwarf shrubs have evolved in African Alchemilla at least twice independently (Fig. 3). The repeated evolution of dwarf shrubs as an in situ adaptation to alpine-like climate conditions seems to have occurred in other areas as well, e.g. numerous times in different groups in the Andean páramos (Sklenář et al., 2011). This is similar to the repeated parallel evolution of other characters that are key to survival in alpine environments, such as larger capitulum size in Artemisia (Tkach et al., 2008), a cushion life form (Roquet et al., 2013) or deciduousness (Zanne et al., 2014).

Our results suggest further that the ancestor of dwarf shrubs in African Alchemilla had a woody base (nodes 4 and 5 in Fig. 4; as seems to be the most common growth form in Fragariinae outside Fragaria and more closely related groups), as opposed to being strictly herbaceous. Both in situ origin and partial ancestral woodiness correspond well to the general concept of insular woodiness sensu Carlquist (1974) as a form of secondary woodiness (Lens et al., 2012). The fact that Alchemilla seems to have had pre-adapted wood structure that enabled it to establish itself in the Afroalpine also corresponds well to a recent meta-analysis showing that woody plants in freezing environments generally had already evolved necessary wood-vessel traits before the climate occupancy (Zanne et al., 2014). Moreover, our ancestral reconstructions indicate reversals in African Alchemilla outside the Afroalpine to an entirely herbaceous form, a transition rarely reported for angiosperms (Carlquist, 2001) but seemingly more common in some groups than previously believed (Soltis et al., 2013).

All dwarf shrub Alchemilla species are exclusively or at least mainly restricted to the Afroalpine, but not all Afroalpine Alchemilla have evolved into erect shrubs. The adaptive advantage of this habit remains to be tested. However, research on dwarf shrubs in other alpine-like environments suggests that woody and erect stature is advantageous due to its influence on the local microhabitat (Körner, 2012). This includes warmer conditions and lower wind chill (Körner and Paulsen, 2004), which might increase survival during night-time freezing, which can occur throughout the year in the Afroalpine. The increased leaf-litter accumulation and decomposition rates (Cornelissen et al., 2007) and greater seedling survival (Klanderud and Totland, 2004) shown in other dwarf shrubs might contribute to the ability of Alchemilla shrubs to become the dominant species in moist Afroalpine habitats just as much as it has helped other dwarf shrubs to dominate wet tundra ecosystems around the world (Gorsuch et al., 2001).

Böhle et al. (1996) suggested that diversification of Echium (Boraginaceae), a prominent example of a group with insular woodiness, was due to a longer reproduction lifetime in longer-lived woody plants, which leads to a greater ability to colonize new islands, rather than adaptation to the ecological environment. Similar effects could apply to the ability to reproduce asexually, i.e. to reproduce successfully in the absence of other individuals or pollinators. It is still unknown to what extent African Alchemilla species reproduce through agamospermy or sexually and, if they do reproduce sexually, to what extent they are reliant on biotic vectors. The absence of showy flowers and inflorescences in Alchemilla dwarf shrubs might suggest lesser reliance on pollinators. Several African species of hybrid origin have been described (e.g. A. dewildemanii, A. elgonensis × A. johnstonii, A. subnivalis × A. stuhlmannii), suggesting that gene flow does occur at least occasionally in African Alchemilla. This is further implied by the differences between plastid and nuclear gene trees for a number of accessions of the Afromilla clade (Supplementary Data Figs S3 and S4). It is thus conceivable that establishment of Alchemilla in the different disjunct sky islands of the Afroalpine was facilitated by longevity and/or agamospermy, but that the latter may then have been facultative.

Long-distance dispersal and establishment of tropical alpine dwarf shrubs

Alchemilla most likely established first in tropical East Africa after long-distance dispersal (LDD) from western Eurasia before spreading to other regions in Africa (Fig. 4). This is not consistent with Hedberg’s (1964) ‘stepping stone’ scenario, in which he postulated that most taxa colonized the Afroalpine via the Arabian Peninsula and initial colonization of the Ethiopian highlands. However, it must be noted that, irrespective of the route, the establishment of Alchemilla in Africa can only have been by LDD from another continent. Research into the assemblage of island floras has shown that the limiting factor in colonization of novel habitats is often establishment rather than dispersal (Alsos et al., 2007), even when there is a lack of obvious dispersal adaptations, as is the case with Alchemilla nutlets. The likelihood of LDD events is generally hard to model (Nathan, 2006) and dispersal from the mountains in Turkey to Simien in Ethiopia (∼3000 km) may be no more or less unlikely than to Mount Elgon on the border between Kenya and Uganda (∼4000 km). Hence, the most parsimonious reconstructed scenario, involving initial establishment in tropical East Africa with two or three subsequent colonizations of Ethiopia, may be more likely than initial establishment in Ethiopia with subsequent dispersal to tropical East Africa, which involves more LDD events (albeit over shorter distances). An aspect of LDD in the Afroalpine is that species with widespread species distributions often show occasional gene flow between populations on different mountains despite their separation by (for Alchemilla and other mountain plants) inhospitable lower elevations (Ehrich et al., 2007; Wondimu et al. 2014).

Recent evolution of dwarf shrubs

The evolution of dwarf shrubs in the tropical Afroalpine was relatively recent, regardless of the large differences in the age estimates derived from different dating methods (Table 1). The lognormal relaxed-clock model implemented in BEAST seems to perform particularly inconsistently with the matrices analysed here, especially the full Rosaceae dataset. Recent analyses of the family based on the same method (Töpel et al., 2012) date the stem age of the Rosaceae to be well over 100 Ma, compared with broader-scale analyses in which the stem age of the family is considered likely to be between 90 and 94 Ma (Crepet et al., 2004). Our own dating of Rosoideae shows large differences between the dates estimated using BEAST and RelTime. If we accept the older estimates (BEAST dates), then dwarf shrubs in the tropical Afroalpine environment would still not be older than 1·2  Ma (3·5–0·2 Ma: A. haumanii) and 6·0 Ma (9·8–2·5 Ma: stem age of the dwarf shrub clade). The more recent estimates (RelTime) suggest that the age could be as young as 0·9–0·5 and 1·7–0·9 Ma. RelTime, the seemingly more consistent dating method here and therefore potentially the more reliable one, places dwarf shrub evolution in the Pleistocene. In tropical Africa this period was characterized by repeated marked changes in vegetation cover with a trend in aridification due to the establishment of the Arctic ice sheet leading to a well-established downward shift in Afromontane forest (Bonnefille, 2010). These conditions probably resulted in an expansion of available niche space for Afroalpine plants (Harmsen et al., 1991).

No evidence for increased diversification rates in tropical Afroalpine dwarf shrubs

Of two dwarf shrub lineages in African Alchemilla, one did not diversify, i.e. A. haumanii (the dominant Alchemilla species in Bale and adjacent mountain areas of south-eastern Ethiopia). The other, the dwarf shrub clade, radiated into at least seven species, mostly with distribution confined to single mountain regions in East Africa.

We did not recover evidence for an increase in diversification rates in tropical Afroalpine dwarf shrubs in Alchemilla. It is a relatively species-rich clade in the tropical Afroalpine, as on average genera are represented by only three species in the area (Hedberg, 1957), compared with 12 species of Alchemilla (nine dwarf shrubs and three other species with a different growth form). However, it is likely that these numbers are simply too small to reveal diversification rate shifts with statistical significance. Some other plant groups in alpine environments appear to show a slowdown in speciation rates (Kadereit et al., 2004). Such a scenario is also not apparent in the Afromilla clade. Dwarf shrub species of Alchemilla can dominate habitats in the Afroalpine, forming ‘Alchemilla woodlands’ (Hedberg, 1951). The trait may contribute to local species abundance and facilitate colonization of alpine habitats in different areas. Our results suggest that occasional LDD events, followed by successful establishment and allopatric speciation, may explain the modest diversification in this clade and the pattern of regional endemism observed today.

SUPPLEMENTARY DATA

Supplementary data are available online at www.aob.oxfordjournals.org and consist of the following. Table S1: collection details. Table S2: NCBI GenBank accession numbers. Table S3: PartitionFinder results. Table S4: characters coded for the Mesquite reconstruction. Table S5: number of species assigned to genera/clades. Table S6: summary of the ancestral area reconstruction in Mesquite. Table S7: summary of the ancestral character reconstruction of elevation in Mesquite. Table S8: summary of the ancestral character reconstruction of lifeform in Mesquite. Table S9: extended overview of the results from the dating analyses. Table S10: identification of diversification model using LASER. Table S11: results of the BAMM analyses. Figure S1: phylogenetic reconstruction of Rosaceae using RaxML. Figure S2: dating of Rosoideae using BEAST. Figure S3: time-calibrated Fragariinae tree constructed using BEAST on plastid data. Figure S4: phylogenetic reconstructions of Fragariinae using plastid data (RAxML). Figure S5: phylogenetic reconstructions of Fragariinae using nuclear data (RAxML). Figure S6: phylogenetic reconstructions of African Alchemilla using the combined plastid and nuclear ribosomal DNA data. Figure S7: details of ancestral area reconstruction. Figure S8: result of the Medusa analyses of diversification rate changes. Figure S9: result of the BAMM analyses of diversification rate changes. Figure S10: example of Alchemilla dwarf shrubs.

Supplementary Data
supp_117_1_121__index.html (790B, html)

ACKNOWLEDGEMENTS

This work was supported by the German Research Foundation (GE2347/1-1 to B.G.). We thank national authorities of Ethiopia, Kenya, Madagascar, South Africa, Tanzania and Uganda for plant collection and export permits. Joachim Kadereit is gratefully acknowledged for in-depth discussions of the research. We thank two anonymous reviewers for their helpful comments.

LITERATURE CITED

  1. African Plants Database (version 3.4.0). Conservatoire et Jardin botaniques de la Ville de Genève and South African National Biodiversity Institute, Pretoria. http://www.ville-ge.ch/musinfo/bd/cjb/africa/ (February 2014). [Google Scholar]
  2. Alfaro ME, Santini F, Brock C, et al. 2009. Nine exceptional radiations plus high turnover explain species diversity in jawed vertebrates. Proceedings of the National Academy of Sciences of the USA 106: 13410–13414. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Alsos IG, Eidesen PB, Ehrich D, et al. 2007. Frequent long-distance plant colonization in the changing Arctic. Sciences 316: 1606–1609. [DOI] [PubMed] [Google Scholar]
  4. Billings WD, Mooney HA. 1968. Ecology of Arctic and alpine plants. Biological Reviews of the Cambridge Philosophical Society 43: 481–529. [Google Scholar]
  5. Blattner FR. 1999. Direct amplification of the entire ITS region from poorly preserved plant material using recombinant PCR. Biotechniques 27: 1180–1186. [DOI] [PubMed] [Google Scholar]
  6. Blok D, Heijmans MMPD, Schaepman-Strub G, Kononov AV, Maximov TC, Berendse F. 2010. Shrub expansion may reduce summer permafrost thaw in Siberian tundra. Global Change Biology 16: 1296–1305. [Google Scholar]
  7. Böhle UR, Hilger HH, Martin WF. 1996. Island colonization and evolution of the insular woody habit in Echium L. (Boraginaceae). Proceedings of the National Academy of Sciences of the USA 93: 11740–11745. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bokhorst S, Bjerke JW, Tommervik H, Preece C, Phoenix GK. 2012. Ecosystem response to climatic change: the importance of the cold season. Ambio 41: 246–255. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bonnefille R. 2010. Cenozoic vegetation, climate changes and hominid evolution in tropical Africa. Global and Planetary Change 72: 390–411. [Google Scholar]
  10. Boucher FC, Thuiller W, Roquet C, et al. 2012. Reconstructing the origins of high-alpine niches and cushion life form in the genus Androsace sl (Primulaceae). Evolution 66: 1255–1268. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Carlquist SJ. 1974. Island biology. New York: Columbia University Press. [Google Scholar]
  12. Carlquist SJ. 2001. Comparative wood anatomy: systematic, ecological, and evolutionary aspects of dicotyledon wood. Berlin: Springer. [Google Scholar]
  13. Chu HY, Grogan P. 2010. Soil microbial biomass, nutrient availability and nitrogen mineralization potential among vegetation-types in a low arctic tundra landscape. Plant and Soil 329: 411–420. [Google Scholar]
  14. Cornelissen JHC, van Bodegom PM, Aerts R, et al. 2007. Global negative vegetation feedback to climate warming responses of leaf litter decomposition rates in cold biomes. Ecology Letters 10: 619–627. [DOI] [PubMed] [Google Scholar]
  15. Crepet WL, Nixon KC, Gandolfo MA. 2004. Fossil evidence and phylogeny: the age of major angiosperm clades based on mesofossil and macrofossil evidence from Cretaceous deposits. American Journal of Botany 91: 1666–1682. [DOI] [PubMed] [Google Scholar]
  16. de León PV, Cevallos-Ferriz SRS. 2000. Leaves of Cercocarpus mixteca (Rosaceae) from Oligocene sediments, near Tepexi de Rodriguez, Puebla. Review of Palaeobotany and Palynology 111: 285–294. [DOI] [PubMed] [Google Scholar]
  17. Dobes C, Paule J. 2010. A comprehensive chloroplast DNA-based phylogeny of the genus Potentilla (Rosaceae): implications for its geographic origin, phylogeography and generic circumscription. Molecular Phylogenetics and Evolution 56: 156–175. [DOI] [PubMed] [Google Scholar]
  18. Dornburg A, Brandley MC, McGowen MR, Near TJ. 2012. Relaxed clocks and inferences of heterogeneous patterns of nucleotide substitution and divergence time estimates across whales and dolphins (Mammalia: Cetacea). Molecular Biology and Evolution 29: 721–736. [DOI] [PubMed] [Google Scholar]
  19. Drummond AJ, Suchard MA, Xie D, Rambaut A. 2012. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Molecular Biology and Evolution 29: 1969–1973. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Ehrich D, Assefa A, Taberlet P, Nemomissa S, Brochmann C. 2007. Pleistocene colonization of Afro-alpine ‘sky islands' by the arctic-alpine Arabis alpina. Heredity 99: 133–142. [DOI] [PubMed] [Google Scholar]
  21. Eriksson T, Hibbs MS, Yoder AD, Delwiche CF, Donoghue MJ. 2003. The phylogeny of Rosoideae (Rosaceae) based on sequences of the internal transcribed spacers (ITS) of nuclear ribosomal DNA and the trnL/F region of chloroplast DNA. International Journal of Plant Sciences 164: 197–211. [Google Scholar]
  22. Eriksson T, Lundberg M, Topel M, Ostensson P, Smedmark JEE. 2015. Sibbaldia: a molecular phylogenetic study of a remarkably polyphyletic genus in Rosaceae. Plant Systematics and Evolution 301: 171–184. [Google Scholar]
  23. Fröhner S. 1995. Alchemilla. In: Conert HJ. ed. Gustav Hegi. Illustrierte Flora von Mitteleuropa: Band 4 Teil 2B. Berlin: Parey, 13–248. [Google Scholar]
  24. 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. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Gehrke B, Linder HP. 2014. Species richness, endemism and species composition in the tropical Afroalpine flora. Alpine Botany 124: 165–177. [Google Scholar]
  26. Gehrke B, Brauchler C, Romoleroux K, Lundberg M, Heubl G, Eriksson T. 2008. Molecular phylogenetics of Alchemilla, Aphanes and Lachemilla (Rosaceae) inferred from plastid and nuclear intron and spacer DNA sequences, with comments on generic classification. Molecular Phylogenetics and Evolution 47: 1030–1044. [DOI] [PubMed] [Google Scholar]
  27. Gehrke B, Martin-Bravo S, Muasya M, Luceno M. 2010. Monophyly, phylogenetic position and the role of hybridization in Schoenoxiphium Nees (Cariceae, Cyperaceae). Molecular Phylogenetics and Evolution 56: 380–392. [DOI] [PubMed] [Google Scholar]
  28. Gorsuch DM, Oberbauer SF, Fisher JB. 2001. Comparative vessel anatomy of arctic deciduous and evergreen dicots. American Journal of Botany 88: 1643–1649. [PubMed] [Google Scholar]
  29. Graham RA. 2000. Alchemilla. In: Graham RA. ed. Rosaceae. Flora of tropical East Africa. Kew: Royal Botanic Gardens, 2–20. [Google Scholar]
  30. Harmsen R, Spence JR, Mahaney WC. 1991. Glacial interglacial cycles and development of the Afroalpine ecosystem on east-African mountains. 2. Origins and development of the biotic component. Journal of African Earth Sciences 12: 513–523. [Google Scholar]
  31. Hedberg O. 1951. Vegetation belts of the East African mountains. Svensk Botanisk Tidskrift 45: 140–202. [Google Scholar]
  32. Hedberg O. 1957. Afroalpine vascular plants. Uppsala: Lundequistska Bokhandeln. [Google Scholar]
  33. Hedberg O. 1964. Features of Afroalpine plant ecology; avec un résumé en française. Uppsala: Almqvist & Wiksells. [Google Scholar]
  34. Hemp A. 2006. Continuum or zonation? Altitudinal gradients in the forest vegetation of Mt. Kilimanjaro. Plant Ecology 184: 27–42. [Google Scholar]
  35. Izmailow R. 1999. Embryological analysis of Aphanes arvensis L. (Rosaceae). Acta Biologica Cracoviensia Series Botanica 41: 197–203. [Google Scholar]
  36. Kaarlejarvi E, Baxter R, Hofgaard A, et al. 2012. Effects of warming on shrub abundance and chemistry drive ecosystem-level changes in a forest-tundra ecotone. Ecosystems 15: 1219–1233. [Google Scholar]
  37. Kadereit JW, Griebeler EM, Comes HP. 2004. Quaternary diversification in European alpine plants: pattern and process. Philosophical Transactions of the Royal Society of London Series B. Biological Sciences 359: 265–274. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Klanderud K, Totland O. 2004. Habitat dependent nurse effects of the dwarf-shrub Dryas octopetala on alpine and arctic plant community structure. Ecoscience 11: 410–420. [Google Scholar]
  39. Körner C. 2012. Alpine treelines. New York: Springer. [Google Scholar]
  40. Körner C, Paulsen J. 2004. A world-wide study of high altitude treeline temperatures. Journal of Biogeography 31: 713–723. [Google Scholar]
  41. Lanfear R, Calcott B, Kainer D, Mayer C, Stamatakis A. 2014. Selecting optimal partitioning schemes for phylogenomic datasets. BMC Evolutionary Biology 14: 1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Lantz TC, Gergel SE, Kokelj SV. 2010. Spatial heterogeneity in the shrub tundra ecotone in the Mackenzie Delta region, Northwest Territories: implications for arctic environmental change. Ecosystems 13: 194–204. [Google Scholar]
  43. Lauer W. 1975. Vom Wesen der Tropen. Mainz: Akademie der Wissenschaften und der Literatur. [Google Scholar]
  44. Lens F, Smets E, Melzer S. 2012. Stem anatomy supports Arabidopsis thaliana as a model for insular woodiness. New Phytologist 193: 12–17. [DOI] [PubMed] [Google Scholar]
  45. Lundberg M, Topel M, Eriksen B, Nylander JAA, Eriksson T. 2009. Allopolyploidy in Fragariinae (Rosaceae): comparing four DNA sequence regions, with comments on classification. Molecular Phylogenetics and Evolution 51: 269–280. [DOI] [PubMed] [Google Scholar]
  46. Maddison WP, Maddison DR. 2009. Mesquite: a modular system for evolutionary analysis. Version 3.03. http://mesquiteproject.org (29 April 2015). [Google Scholar]
  47. Magallon S, Sanderson MJ. 2001. Absolute diversification rates in angiosperm clades. Evolution 55: 1762–1780. [DOI] [PubMed] [Google Scholar]
  48. Matthews JV, Ovenden LE. 1990. Late Tertiary plant macrofossils from localities in arctic sub-arctic North-America - a review of the data. Arctic 43: 364–392. [Google Scholar]
  49. Miller MA, Pfeiffer W, Schwartz T. 2010. Creating the CIPRES Science Gateway for inference of large phylogenetic trees. Proceedings of the Gateway Computing Environments Workshop (GCE), New Orleans, 1–8. [Google Scholar]
  50. Müller KF. 2005. SeqState - primer design and sequence statistics for phylogenetic DNA data sets. Applied Bioinformatics 4: 65–69. [DOI] [PubMed] [Google Scholar]
  51. Muir G, Fleming CC, Schlotterer C. 2001. Three divergent rDNA clusters predate the species divergence in Quercus petraea (Matt.) Liebl. and Quercus robur L. Molecular Biology and Evolution 18: 112–119. [DOI] [PubMed] [Google Scholar]
  52. Myers-Smith IH, Forbes BC, Wilmking M, et al. 2011. Shrub expansion in tundra ecosystems: dynamics, impacts and research priorities. Environmental Research Letters 6 doi: 10.1088/1748-9326/6/4/045509. [Google Scholar]
  53. Nathan R. 2006. Long-distance dispersal of plants. Science 313: 786–788. [DOI] [PubMed] [Google Scholar]
  54. Oh SH, Potter D. 2005. Molecular phylogenetic systematics and biogeography of tribe Neillieae (Rosaceae) using DNA sequences of cpDNA, rDNA, and LEAFY. American Journal of Botany 92: 179–192. [DOI] [PubMed] [Google Scholar]
  55. Pennell MW, Eastman JM, Slater GJ, et al. 2014. Geiger v2.0: an expanded suite of methods for fitting macroevolutionary models to phylogenetic trees. Bioinformatics 30: 2216–2218. [DOI] [PubMed] [Google Scholar]
  56. Pirie MD, Humphreys AM, Antonelli A, Galley C, Linder HP. 2012. Model uncertainty in ancestral area reconstruction: a parsimonious solution? Taxon 61: 652–664. [Google Scholar]
  57. Potter D, Eriksson T, Evans RC, et al. 2007. Phylogeny and classification of Rosaceae. Plant Systematics and Evolution 266: 5–43. [Google Scholar]
  58. Preece C, Phoenix GK. 2013. Responses of sub-arctic dwarf shrubs to low oxygen and high carbon dioxide conditions. Environmental and Experimental Botany 85: 7–15. [Google Scholar]
  59. R Development Core Team. 2015. R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; http://www.R–project.org/. [Google Scholar]
  60. Rabosky DL. 2006. LASER: a maximum likelihood toolkit for detecting temporal shifts in diversification rates from molecular phylogenies. Evolutionary Bioinformatics 2: 247–250. [PMC free article] [PubMed] [Google Scholar]
  61. Rabosky DL. 2014. Automatic detection of key innovations, rate shifts, and diversity-dependence on phylogenetic trees. PLoS One 9: e89543. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Rabosky DL, Grundler M, Anderson C, et al. 2014. BAMMtools: an R package for the analysis of evolutionary dynamics on phylogenetic trees. Methods in Ecology and Evolution 5: 701–707. [Google Scholar]
  63. Rambaut A, Suchard MA, Xie D, Drummond AJ. 2014. Tracer v1.6. http://beast.bio.ed.ac.uk/Tracer. [Google Scholar]
  64. Rayback SA, Henry GHR, Lini A. 2012. Multiproxy reconstructions of climate for three sites in the Canadian high Arctic using Cassiope tetragona. Climatic Change 114: 593–619. [Google Scholar]
  65. Roquet C, Boucher FC, Thuiller W, Lavergne S. 2013. Replicated radiations of the alpine genus Androsace (Primulaceae) driven by range expansion and convergent key innovations. Journal of Biogeography 40: 1874–1886. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Samylina VA. 1960. Angiosperm plants from the Lower Cretaceous deposits of the Kolyma River. Botanicheskii Zhurnal 45: 335–352. [Google Scholar]
  67. Shaw J, Lickey EB, Beck JT, et al. 2005. The tortoise and the hare II: relative utility of 21 noncoding chloroplast DNA sequences for phylogenetic analysis. American Journal of Botany 92: 142–166. [DOI] [PubMed] [Google Scholar]
  68. Shaw J, Lickey EB, Schilling EE, Small RL. 2007. Comparison of whole chloroplast genome sequences to choose noncoding regions for phylogenetic studies in angiosperms: the tortoise and the hare III. American Journal of Botany 94: 275–288. [DOI] [PubMed] [Google Scholar]
  69. Simmons MP, Ochoterena H. 2000. Gaps as characters in sequence-based phylogenetic analyses. Systematic Biology 49: 369–381. [PubMed] [Google Scholar]
  70. Sklenář P, Dusková E, Balslev H. 2011. Tropical and temperate: evolutionary history of Páramo flora. Botanical Review 77: 71–108. [Google Scholar]
  71. Soltis DE, Mort ME, Latvis M, et al. 2013. Phylogenetic relationships and character evolution analysis of Saxifragales using a supermatrix approach. American Journal of Botany 100: 916–929. [DOI] [PubMed] [Google Scholar]
  72. Stamatakis A, Hoover P, Rougemont J. 2008. A rapid bootstrap algorithm for the RAxML web servers. Systematic Biology 57: 758–771. [DOI] [PubMed] [Google Scholar]
  73. Swofford DL. 2002. PAUP*. Phylogenetic analysis using parsimony (*and other methods), 4th edn Sunderland: Sinauer Associates. [Google Scholar]
  74. Tamura K, Battistuzzi FU, Billing-Ross P, Murillo O, Filipski A, Kumar S. 2012. Estimating divergence times in large molecular phylogenies. Proceedings of the National Academy of Sciences of the USA 109: 19333–19338. [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Tkach NV, Hoffmann MH, Roser M, Korobkov AA, von Hagen KB. 2008. Parallel evolutionary patterns in multiple lineages of arctic Artemisia L. (Asteraceae). Evolution 62: 184–198. [DOI] [PubMed] [Google Scholar]
  76. Töpel M, Lundberg M, Eriksson T, Eriksen B. 2011. Molecular data and ploidal levels indicate several putative allopolyploidization events in the genus Potentilla (Rosaceae). PLoS Currents: Tree of Life. Edition 1. doi: 10.1371/currents.RRN1237 [DOI] [PMC free article] [PubMed] [Google Scholar]
  77. Töpel M, Antonelli A, Yesson C, Eriksen B. 2012. Past climate change and plant evolution in western North America: a case study in Rosaceae. PLoS One 7: e50358. [DOI] [PMC free article] [PubMed] [Google Scholar]
  78. Wang HC, Moore MJ, Soltis PS, et al. 2009. Rosid radiation and the rapid rise of angiosperm-dominated forests. Proceedings of the National Academy of Sciences of the USA 106: 3853–3858. [DOI] [PMC free article] [PubMed] [Google Scholar]
  79. White F. 1983. Vegetation of Africa: a descriptive memoir to accompany the UNESCO/AETFAT/UNSO vegetation map of Africa. Paris: UNESCO. [Google Scholar]
  80. White TJ, Bruns T, Lee S, Taylor J. 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In: Innis MA, Gelfand DH, Sninsky JJ, White TJ. eds. PCR protocols: a guide to methods and applications. New York: Academic Press. [Google Scholar]
  81. Wolfe JA, Schorn HE. 1989. Paleoecologic, paleoclimatic, and evolutionary significance of the Oligocene Creede flora, Colorado. Paleobiology 15: 180–198. [Google Scholar]
  82. Wondimu T, Gizaw A, Tusiime FM, et al. 2014. Crossing barriers in an extremely fragmented system: two case studies in the Afro-alpine sky island flora. Plant Systematics and Evolution 300: 415–430. [Google Scholar]
  83. Zanne AE, Tank DC, Cornwell WK, et al. 2014. Three keys to the radiation of angiosperms into freezing environments. Nature 506: 89–92. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Data
supp_117_1_121__index.html (790B, html)
supp_mcv159_mcv159_supplementary_material_proof.doc (6.3MB, doc)

Articles from Annals of Botany are provided here courtesy of Oxford University Press

RESOURCES