Is speciation faster in the

Tropics? Comparison of diversification rates between temperate and tropical L. ()

Nicolás Lavandero López 19/08/2016

Thesis submitted in partial fulfillment for the MSc in the Biodiversity and of 1

ABSTRACT

The Latitudinal Diversity Gradient (LDG) is one of the most common observed and studied patterns in ecology. The study of the uneven distribution of across the globe has nourished a body of ecologic theories, with several hypotheses proposed in order to explain this pattern. LDG has been suggested to be caused by higher speciation rates in the tropics compared with temperate regions. In order to test this hypothesis, a phylogeny of the mostly Northern Temperate but also South American genus Berberis was constructed and dated. Nuclear ITS and chloroplast ndhF regions were used to build a molecular phylogeny representing most of the sections of Berberis found in the Northern Hemisphere and South America. Bayesian diversification analyses were done using the time-calibrated phylogeny. Our results suggest a significant rate shift in diversification near the base of Berberis, with no further increase in speciation rates towards the tropics. The evidence suggests no significant differences in speciation for Berberis between tropical and temperate zones. Further studies will be required to test whether higher rates detected in the Himalayan clade are linked to the mountain uplift.

2

ACKNOWLEDGEMENTS

First, I would like to thank to my great supervisors, James Richardson, Tiina Sarkinen and Bhaskar Adhikari. Without your help, suggestions, last minute corrections and their valuable support I would not have survived at the last three months.

Collecting samples and fieldwork is an exhausting task. I would like to thank Carlos and Andres for their collections in Colombia, the Jardin Botanico de Bogota Jose Celestino Mutis for the material provided. To Jim Solomon of the Missouri Botanical Garden, Federico Luebert, and Joao Stehmann for kindly providing samples from South America.

I wish to thank Laura Forrest and Michelle Hart for the training, the constant help in solving all my issues during lab work.

The last year would not have been the same without all my fellow MSc buddies. Thanks for making this year a memorable time. In particular, to Marita for bringing happiness and laughs, to Camila for sharing her Patagonian warmth and to Ludovica for helping me doing the acknowledgements and for saving me from eating instant couscous the all year.

To my adoptive parents Ricardo, Lucia and Gin for hosting me in the very first days of this experience abroad.

Last but not least, I wish to thank my family for encouraging me to come to Edinburgh and for their understanding throughout all these years of University.

3

Table of Contents

Abstract ...... 1 Acknowledgements ...... 2 List of Figures ...... 5 List of Tables ...... 6 1. Introduction ...... 7 1.1 Patterns of biodiversity: the latitudinal gradient ...... 7 1.1.1 Ecological Hypotheses ...... 8 1.1.2 Historical and Evolutionary hypotheses ...... 9 1.2 Diversity in the Neotropics and Andean Orogeny ...... 11 1.3 Himalayan uplift and Biodiversity ...... 13 1.4 Divergence time estimations using molecular and fossil data ...... 13 1.5 Inference of differential rates of species diversification ...... 16 1.6 Introduction to Berberis...... 16 1.7 Study Aims ...... 18 2. Materials and Methods ...... 19 2.1 Taxon sampling ...... 19 2.2 Laboratory Methods ...... 20 2.3 Sequence Alignment ...... 22 2.4 Phylogenetic analysis ...... 22 2.5 Fossil Calibration ...... 24 2.6 Divergence Age Estimation ...... 24 2.7 Calculation of diversification rates using BAMM software ...... 26 3. Results ...... 28 3.1 Phylogenetic analysis ...... 28 3.2 Divergence time estimation ...... 30 3.3 Rate Shifts And Diversification rates ...... 32 4. Discussion ...... 40 4.1 Diversification of Berberis ...... 40 4.2 Phylogenetic Relationships in Berberis and Divergence time Estimates ...... 45 4

5. Conclusion ...... 49 6. References ...... 50 7. Appendices ...... 61 7.1 Appendix 1 – accesion details ...... 61 7.2 Appendix 2 – ITS tree ...... 64 7.3 Appendix 3 – ndhF tree ...... 65 7.4 Appendix 6 – Bayesian tree with all South American species ...... 66

5

LIST OF FIGURES

Figure 1. Distribution of occurrence records for Neotropical Berberis from the Global Biodiversity Information Facility (GBIF)...... 20 Figure 2. Combined bayesian 50% majority rule consensus tree obtained from ndhF and ITS sequences of 102 accessions representing 94 species of Berberis...... 29 Figure 3. Combined Maximum clade credibility chronogram for combined ITS and ndhF obtained in Beast analysis...... 31 Figure 4. Prior and posterior probability of models with different number of shifts for Berberis dataset...... 33 Figure 5. Evidence of shifts occurring along Berberis branchs according to BAMM...... 35 Figure 6. phylorate plots for speciation showing the two distinct shift configurations within the 95% credible shift set...... 36 Figure 7. Macroevolutionary cohort matrix for speciation rates in Berberis Derived from BAMM analysis...... 37 Figure 8. Histograms of relative rates of speciation for Berberis clades...... 38 Figure 9. Speciation rates through time of Berberis compared with lineages of interest. black line indicates the mean speciation rate at any indicated time across Berberis...... 39 Figure 10. bayesian 50% majority rule consensus tree obtained from ITS sequences of Berberis...... 64 Figure 11. bayesian 50% majority rule consensus tree obtained from ndhF sequences of Berberis...... 65 Figure 12. Combined bayesian 50% majority rule consensus tree obtained from ndhF and ITS sequences of 116 accessions representing 108 species of Berberis...... 66

6

LIST OF TABLES

Table 1. Primers for Berberis in this study...... 22 Table 2.Substitution models for each partition of the Berberis Dataset...... 23 Table 3. Sampling for each Berberis clade considered for the BAMM analysis ...... 27 Table 4. mean divergence times, with 95% highest posterior densities (HPD) and posterior probability support for selected nodes for the MCC tree generated by BEAST ...... 32 Table 5. Bayes Factors for models with different number of shifts...... 33 Table 6. clade-specific speciation (λ), extinction (µ) and net diversification rates (Dr), with the 90% highest posterior density shown for selected clades in Berberis from the BAMM analysis...... 34 Table 7. relative speciation rates for clades of interest in Berberis...... 39 Table 8. List of accesions used in this study...... 61

7

1. INTRODUCTION

1.1 PATTERNS OF BIODIVERSITY: THE LATITUDINAL GRADIENT

Species richness is not evenly distributed across the globe. The higher species richness of the tropics compared with temperate regions is one of the oldest patterns known and studied in ecology (Hawkins and Porter, 2001) and remains a central topic in ecological studies (Gaston, 2000: Hillebrand et al., 2004). Understanding the processes responsible for generating the patterns of biodiversity on earth has been one of the most important objectives for naturalists since modern times (Gaston, 2000), and more recently by ecologists and biogeographers (Currie et al., 2004; Mittelbach et al., 2007). The increase in number of species towards the tropics appears to be common regardless of the taxa, with evidence for terrestrial and marine groups, animals, fungi and plants, including Angiosperms (Macpherson, 2002; Gaston, 2000; Davies et al., 2004). What is more, the Latitudinal Diversity Gradient (LDG) seems to be persistent over geological time scales, with evidence of a consistent latitudinal gradient since the Permian even when the thermal gradient was less steep towards Polar Regions (Stehli et al., 1969). The consistency of this pattern has been confirmed over time, with a constantly growing number of studies reinforcing the generality of the LDG (Gaston, 2000), with some few exceptions and irregularities (Willig et al., 2003; Hillebrand, 2004; Mateo et al., 2016). In fact, in the most recent review of 581 studies, Hillebrand (2004) has shown the consistency of this pattern across all studies regardless of the taxa, hemisphere or latitudinal range. From the perspective of conservation, this pattern has nourished the body of ecological studies, with concepts about local and regional diversity and conservation ecology (Chown & Gaston, 2000). Myers et al., (2000) indicated that 16 out of the 25 world’s biodiversity hotspots to occur in tropical latitudes. Moreover, Joppa et al. (2011) suggested that most of the species yet to be discovered would be located in tropical regions.

The potential explanations for this pattern are numerous (Pianka, 1966; Gaston, 2000; Currie et al., 2004; Mittelbach, et al., 2007) and not mutually exclusive. More than 30 hypotheses have been suggested (Willig et al., 2003) Many of these look at the same pattern but offer different explanations across geographic scales (Gaston, 2000). Other problems derive from the narrow taxonomic groups used to test these hypotheses, incomplete diversity data and the scale at which they are tested (Rickleffs & He, 2016). 8

The proposed explanations can be categorized into three broad types (Mittlebach et al., 2007). Ecological hypotheses focus on abiotic conditions and their effect on the number of species, evolutionary hypotheses focus on speciation and extinction events and historical hypotheses focus on the duration and extent of tropical regions through time. I sum up here some of the most important hypotheses following Mittelbach’s (2007) classification, taking into account the revisions made by Pianka (1966) and more recent syntheses (Gaston, 2000; Mittelbach et al., 2007) and integrated approaches (Wiens & Donoghue, 2004; Dowle et al., 2013).

1.1.1 ECOLOGICAL HYPOTHESES One of the most basic explanations for the LDG is the area hypothesis (Rosenzweig, 1995). The earth, due to its geometry, differs in surface depending on the latitude. The surface present in a latitudinal band increases towards the equator. Therefore, the area of tropical latitudes is greater than any other climatic zone (Gaston, 2000). Larger biomes can sustain large populations, larger species ranges and lower chances of extinction (Terborgh, 1973). Species with large ranges are more prone to allopatric speciation, which will promote diversification. This hypothesis relies on two main assumptions: species number increases with area and that the Tropics are the largest biome. Evidence for the effect of area on diversity appears to be dependent on scale and other confounding effects (Hawkins & Porter, 2001). The latter assumption that the tropics are the largest biome may be wrong. While it is true that extratropical zones are disjunct, compared with the tropics, there is little support to claim that the area of tropical biomes is greater than temperate ones (Pianka, 1966; Hawkins & Porter, 2001).

The Climatic Zonation Hypothesis (Janzen, 1967), suggests that in high latitudes, the overlap of climatic conditions between different elevations is high. In contrast, at low latitudes, the overlap of climatic conditions at different elevations is almost absent. The idea that mountain passes are higher in the tropics (Janzen, 1967) has been continuously revisited and tested (Huey, 1978; Gill et al., 2006; Ruggiero & Hawkins, 2008; McCain, 2009). The different climatic zonation depending on latitude increases the possibilities of geographic zonation and isolation of closely related species towards tropical latitudes. Speciation may occur in two different ways (Kozak & Wiens, 2007). Firstly, montane 9 regions may serve as refugia during glacial-interglacial periods, isolating species in different regions while their distributions track their optimal climate. The case of the high Andean Páramo vegetation is an example of recent radiation, probably driven by allopatric speciation during glacial and interglacial periods (Madriñán et al., 2013). Another form of speciation promoted by the climatic zonation hypothesis is that particular lineages will adapt to higher elevations climates and posteriorly diverge from low elevation relatives (Bates & Zink, 1994).

The Mid-domain hypothesis (Colwell et al., 2004) is based on null models of random placement of species within a bounded domain. This model has been criticized for the limited importance of this effect in the case of species with a narrow range of distribution (Jetz & Rahbek., 2001), problems with the design of the null model (Hillebrand et al., 2004), and unrealistic assumptions about species ranges (Rahbek et al., 2007). The assumption of a two-dimensional space where species are distributed is also a huge fault in the model, considering the elevational variation in species distribution (Gaston, 2000).

A hypothesis that takes into account community interactions, the Predation hypothesis (Pianka, 1966), proposed that there are more predators at lower latitudes, and these are able to contain prey populations, decreasing the competition between and among them. This scenario of low competition allows the coexistence of a higher number of preys in the system. More recent works challenge this hypothesis as ecological interactions appears to act mainly at local scales, and the relevance for the LDG should be cast into doubt (Hillebrand et al., 2004). The findings by Lambers et al. (2002) on the density-dependent mortality of tree species revealed no difference between temperate zones and the tropics, overthrowing the importance of density-dependent mortality on maintaining high diversity in the tropics. Other studies have failed to confirm changes in community interactions such as - pollinator interactions (Ollerton & Cranmer, 2002) along the latitudinal gradient.

1.1.2 HISTORICAL AND EVOLUTIONARY HYPOTHESES The time hypothesis (Fischer, 1960) suggests that if all biomes diversify at the same rate over time, older biomes should have more species than younger ones. Fischer (1960) proposed that the Tropics have more species just because they have a longer and more 10 undisturbed evolutionary history compared to temperate zones that were impoverished due to glaciations during the Pleistocene. Fischer’s hypothesis assumes equal diversification rates at all latitudes and focuses mainly on the history of disturbances rather than evolutionary mechanisms.

Fischer’s hypothesis has many variations. The effective evolutionary time hypothesis (Rohde, 1992) proposes that the Tropics, besides having longer climatic stability, also has covered larger areas over time, which has increased speciation and higher mutation rates, resulting in higher species number. This hypothesis includes elements from both evolutionary and ecological mechanisms.

Stebbins’ (1974) hypotheses for tropical hyperdiversity can be categorized in three main ideas (Arita & Vásquez-Domínguez, 2008). The “Tropics as museum” idea suggests that the tropics are a region where old lineages have evolved and persisted through time. The museum theory assumes constant speciation rate and low extinction. Support for this hypothesis has been provided by the fossil record and molecular phylogenetic studies (Antonelli & Sanmartín, 2011). In contrast, the “cradle theory” suggests that most tropical lineages originated in recent times with high diversification rates. Climatic and tectonics events, such as the Andean uplift, may have caused these recent radiations (Hughes & Eastwood, 2006; Luebert & Wiegend, 2014). Finally, Jablonski et al. (2006) proposed the “Out of the tropics hypothesis”, where the tropics behave both as a cradle and a museum, with higher speciation rates, lower extinction rates, and taxa formerly originated in the tropics emigrating towards higher latitudes.

The idea that lineages originate in the tropics and then disperse towards higher latitudes comes from the Tropical Niche Conservatism (TNC) hypothesis, proposed by Wiens & Donoghue (2004) under the prepositions of Qian & Ricklefs (2004). This hypothesis suggests that species tend to retain their climatic niches in time (Antonelli & Sanmartín, 2011). In consequence, tropical regions will have more species because ancestral lineages originated and diversified in the stable climate of the tropics, underlying the tropical origin of most plant groups (Rolland et al., 2014). Hawkins et al. (2011) found a strong association between climate and species richness for trees. They also found that older lineages occur in warm places and temperate lineages are nested in tropical groups. This 11 work validates that phylogenetic niche conservatism has a major role in explaining the LDG. This hypothesis has been supported by phylogenetic evidence showing long distance dispersal to the tropics of pre-adapted lineages (Antonelli & Sanmartín, 2011).

The evolutionary speed hypothesis (Rhode, 1992) postulates that higher rates of mutations will cause higher rates of speciation in the tropics. Lower latitudes have been suggested to have higher evolutionary rates compared with temperate zones. The reasons for higher evolutionary rates in the tropics can be due to higher temperatures, high levels of UV radiation and shorter generation times compared with temperate zones (Dowle et al., 2013). This explanation is based on the considerably higher levels of solar radiation and energy found near the equator in comparison with higher latitudes. Higher levels of ultraviolet light have been linked with higher mutation rates and species diversity for Angiosperms (Davies et al., 2004; Willis et al., 2009). The exact mechanism by which faster mutation rates increases biodiversity is still unclear. Speciation might lead to faster evolutionary rates or higher evolutionary rates may increases speciation (Pagel et al., 2006; Dowle et al., 2013). Ultraviolet radiation has been associated with mutagenic properties and the intensity of radiation is higher towards lower latitudes (Flenley, 2011). UV radiation may be linked to an increase in speciation rate, affecting the early germline or early stages of development, such as pollen grains (Ahmad et al., 1991). The ultraviolet radiation increases as well with elevation. Montane tropics receive the highest UV radiation of all regions in the world (Sullivan et al., 1992). This could be a likely explanation for very fast evolving tropical montane lineages, such as the Páramo (Madriñán et al., 2013) and Lupinus radiation in the Andes (Hughes & Eastwood, 2006), or Rhododendron in South East Asia (Flenley, 2011).

1.2 DIVERSITY IN THE NEOTROPICS AND ANDEAN OROGENY

The Neotropics are one of the most species rich regions in the world, with an estimated 90,000 ~ 100,000 species (Myers et al., 2000; Hughes et al., 2013). These numbers may be underestimating the real diversity of the region, as many new species are described each year. The number of species in the Neotropics is greater than in tropical Africa (30.000~35,000) or tropical Asia and Oceania together (40,000~82,000 species) (Antonelli & Sanmartín, 2011). 12

Gentry (1982) proposed that the high diversity of species in the Neotropics is a consequence of high diversification rates. He hypothesized that the greater diversity of the neotropics was an “accident of the Andean orogeny” (Gentry, 1982). The Andean orogeny hypothesis by Gentry has recently been confronted with new molecular evidence on a number of different groups.

The Andes as a geographical region can be divided in three regions, which are not only of geographic importance, but also give insights about the complex biogeography of this orogeny event (Luebert & Weigend, 2014). The northern Andes comprise Venezuelan, Colombian, Ecuadorian, and Northern Peruvian montane regions. Mountain uplift in these regions started in the Paleogene, with a posterior intensification in the Early Miocene caused by the plate breakup in the Pacific and collision of Caribbean plates (Hoorn et al., 2010). montane building occurred in this time (Oligocene-Early Miocene), but posterior mountain build-up in the late middle Miocene and early Pliocene accounts for most of the modern elevations. This Northern region comprises plant diversity hotspots, like high elevation tropical Andean Páramo ecosystem (Madriñán et al. 2013). The Central Andes comprises Southern Ecuador, Northern Peru, Bolivia and Chile-Argentina montane regions (between 15° to 29° S) (Luebert & Weigend, 2014). This region had no more than half of its height by late Oligocene-Early Miocene. The most intense uplift events occurred in the late Miocene-Pliocene, with uplifts up to 2000-2500 m in the last 10 Ma. In this region, there is a change in the precipitation regimes, from a summer rainfall (Puna region) to winter-rainfall (Pliscoff & Luebert, 2006). Southern Andes comprise the rest of the Andes south to 29°S. Most of its uplift occurred in the Early Miocene, with late Miocene uplifts in Patagonian Andes (Ramos, 2014). This region is characterized by a decrease in elevation towards the south.

The Andes uplift changed the landscape dramatically influencing the rainfall regimes, sediment supply and atmospheric circulation (Antonelli & Sanmartín, 2011). The effect of the Andes orogeny on plant diversity can be grouped in four categories (Luebert & Wiegend, 2014). The Andes as a generator of new biomes outside the Andes, as the diversification of Inga in the Amazon basin (Richardson et al., 2001), or Cactaceae in the Atacama Desert (Guerrero et al. 2011). The Andes as a creator of new high elevation habitats, such as the Páramo (Madriñán et al., 2013). The Andes as a vicariant barrier between west and east slope (e.g. “dry diagonal”), or even within valleys (e.g., Seasonally Dry Tropical Forests). Finally, the Andes as a corridor, were temperate lineages could 13 move towards the north or tropical lineages to the south. Migration of Gondwanan elements from Southern South America to the tropics appears to be important input of new species to the Northen Andes (Segovia & Armesto, 2015). Pennington & Dick (2010) suggested that Weinmannia and many other members of Cunionaceae, which are an important element of the montane forests in the Northen Andes, may have traveled northwards through the Andes and diversified by allopatric speciation (Bradford, 2002).

1.3 HIMALAYAN UPLIFT AND BIODIVERSITY

The Himalaya is another montane biodiversity hotspot (Myers et al., 2000) located at temperate latitudes. Compared with the Andes, there is a lack of consensus about the events of uplift and time of these events. Until recent times, it was widely accepted that plate collision between the Indian and Eurasian plates in the cretaceous caused most of the uplift (Harrison & Copeland, 1992). Recent studies cast doubt about the dates, giving earlier dates and the order of plate collision events that give rise to the Himalayas (Aitchison et al., 2007; Van Hinsbergen et al., 2012).

This region has provided place for some of the most fantastic radiation and diversification among plants. The list include Meconopsis Vig. (Xie et al., 2014), Phyllolobium Fisch. (Zhang et al., 2012), Begonia L. (Thomas, 2012), Rhodiola L., Saussurea DC., Primula L. (Wen et al., 2014). Himalaya is also one of the hotspots of diversity of large groups like Rhododendron L. (Yan et al., 2014). The ranges of origin ages for Andean and Himalayan lineages are reported to be similar (Luebert & Muller, 2015; Wen et al. 2014), giving insights about common orogeny and diversification events occurring within the same periods for both montane regions.

Nevertheless, in comparison with the Andes, there are few phylogeographic studies in the Himalayas. Further studies might give support to hypothesis about diversification and orogenetic history. Many studies so far work only on the geological framework that suit their expectations and correlation among biologic and geologic events does not give a strong argument for causality (Miehe et al. 2015).

1.4 DIVERGENCE TIME ESTIMATIONS USING MOLECULAR AND FOSSIL DATA 14

The first dating methods go back to the constant rate molecular clock proposed by Zuckerkandl and Pauling (1965), where mutations rates amongst different lineages were assumed to be constant through time. This assumption has been dismantled by many studies on plants and animals demonstrating clear substitution rate variation between taxa (e.g. Andreasen & Baldwin, 2001). Many theories have been proposed, linking rate heterogeneity to UV mutations (Willis et al., 2009; Davies et al., 2004), population size (Martin and Palumbi, 1993; Ohta, 2002) and more recently, to life forms and generation time in plants (Andreasen & Baldwin, 2001, Smith & Donoghue, 2008). There is a correlation between the generation time and the amount of mutations, where lineages with short generation time accumulate more mutations through the same amount of time compared with those with long generation time (Smith & Donoghue, 2008). This has led to the development of alternative molecular dating methods that take into account heterogeneity in rates of molecular evolution.

The most important recent advance in the molecular dating methods is the development of relaxed-clock methods. These methods take into account the rate heterogeneity among clades, with the assumption that more closely related lineages will have more similar rates. Penalized likelihood (PL) methods (Sanderson, 2002), as the algorithm implemented in treePL (Smith & O’Meara, 2012) allows strict minimum and maximum age constrains for fossil data, with no explicit prior distribution. This method uses a semi-parametric approach that allows branches to follow different rates of autocorrelation and the final level will be determined by a smoothing parameter. This parameter is obtained using a cross-validation procedure before the final divergence time estimation (Sauquet, 2013). The penalized likelihood dating is then run with the chosen smoothing parameter (Smith & O’Meara, 2012; Sanderson, 2002). The most recent implantation of the PL algorithm is in the treePL software that allows relatively rapid divergence time estimation even in large datasets (Smith & O’Meara, 2012).

Bayesian relaxed-clock is another commonly used molecular dating method (dos Reis et al., 2016). These methods use Markov chain Monte Carlo algorithm sampling methods in order to obtain the posterior distribution for molecular rates. These methodologies are implemented in BEAST (Drummond & Rambaut, 2007) which is the most used software because of the ability to analyze many loci and use many fossil calibrations in the same analysis (dos Reis et al., 2016). BEAST offers an uncorrelated lognormal relaxed clock model (Drummond et al., 2006) which makes no prior assumption for rate correlation 15 among closely related taxa, and the rates are sampled from an a priori specified distribution. Compared with penalized likelihood methods, BEAST takes into account the phylogenetic uncertainty in divergence time estimation (Sauquet, 2013). A question has been raised about the reliability of these methods in the absence of complete data. The general rule has been to analyze only taxa with complete sampling of genes or regions under study to avoid the negative effects of missing data in such complex models (Jiang et al., 2014). The results found by Zheng & Wiens (2015) suggests, however, that using taxa with missing data does not lead to erroneous or misleading divergence time estimates and can even improve the estimates compared to excluding them.

In all divergence time estimations, one of the most important steps is the setting of calibration points. The molecular matrix only accounts for the number of substitutions and relations among taxa. In order to determine of how these substitutions accumulate over time, a minimum of one calibration point is necessary. A calibration point gives a relative or absolute timescale for divergence, depending on the calibration used. The most common method of calibration utilizes the fossil record. The palaeontological evidence can give minimum constraint ages for clades. The age assigned to fossils correlates with the dates of the rocks where they are found (Sauquet et al., 2012). Due to the nature of the fossil record, uncertainty should be considered when calibrating nodes with them. treePL can assign age intervals to nodes. In BEAST, a prior distribution for the age of the node can be assigned, like a lognormal distribution to model the temporal gap (Magallón, 2004). In studies where an absolute date is needed, the use of as many calibration points as possible is the general rule to obtain well supported divergence times for a phylogeny (Sauquet, 2013). The accuracy of the divergence time decreases dramatically when few calibrations points are used compared with multiple calibrated points (Zheng & Wiens, 2015). The inaccuracies imposed by any erroneous calibration points can be cross-validated by other calibrations, leading to better estimates of rates and divergence times (Ho & Duchene, 2014). This is not always possible because of the incompleteness and bias of the fossil (Quental & Marshall, 2010). When no fossils are present for the study group, secondary calibrations, based on ages obtained from previous molecular analyses can be used. The consequences of using secondary calibrations are not completely understood. One of the problems identified with secondary calibration points is the wider confidence intervals (Graur and Martin, 2004). Schenk (2016) found that using only secondary calibrations lead to the illusion of more precision, and decrease of variance, but greater width of credible 16 intervals. Schenk (2016) recommends the use of primary fossil calibrations whenever possible, and secondary age estimations should not be used as the only source of calibration until we understand how to model the uncertainties of this method.

1.5 INFERENCE OF DIFFERENTIAL RATES OF SPECIES DIVERSIFICATION

There is a common agreement that the variation in species richness is the result of the variation in three main mechanisms: speciation rates, extinction rates and inmigration rates (Dowle et al., 2013). The interaction among these three processes gives the overall number of species and may be a way to understand the process behind the latitudinal diversity gradient. A key constraint in testing the effect of these processes in the species diversity is the impossibility to take direct measures of speciation and extinction through time (Arita & Vásquez-Domínguez, 2008). A novel approach is the development of models that allow scientists to simulate these events, with the use of dated phylogenies.

The Bayesian Analysis of Macroevolutionary Mixtures (BAMM) (Rabosky, 2014) is a program that models speciation and extinction dynamics between lineages through time using a reversible jump Markov chain Monte Carlo sampling. Alternative approaches, like the comparative method of Modelling Evolutionary Diversification Using Stepwise Akaike MEDUSA, developed by Alfaro et al. (2009) uses a stepwise approach that relies on the Akaike information criterion algorithm to add more rate shifts to the tree until the model fails to fit the data (Rabosky, 2014; Alfaro et al., 2009). The main difference with the two methods is that MEDUSA assumes constant diversification rates through time, (Rabosky, 2014), despite the fact that there is mounting evidence showing heterogeneity of diversification rates among lineages (Morlon, 2014). Models that do not take into account rate heterogeneity among clades often have strong deflation or inflation of rate estimates (Rabosky, 2010; Morlon et al., 2011).

1.6 INTRODUCTION TO BERBERIS

The family Berberidaceae Juss. is a member of the Juss. ex Bercht & J. Presl which belong to the Early Diverging . This order also contains the Circaeasteraceae Hutch., Eupteleaceae K. Wilh, Lardizabalaceae R. Br., Menispermaceae 17

Juss., Papaveraceae Juss. and Ranunculaceae Adans., according to the APG III (Bremer et al., 2009). Recent studies among the Ranunculales using molecular and morphological data suggest Ranunculaceae as a sister family of Berberidaceae (Wang et al., 2009). Berberidaceae contains 14 genera and c. 700 species. The distribution of this family is mostly northern hemisphere, North America, East Asia, Africa and South America, being totally absent in Australia. Most of the genera within Berberidaceae are either monotypic or have less than five species. The largest genus is Berberis Linn. This genus has been recently united with Mahonia Nutt. Despite their morphological differences, due to their close relationship suspected and confirmed by molecular evidence (Kim et al., 2004a, Adhikari et al., 2015). Phylogenetic studies confirm their close relationships, with simple- leaved Berberis (Berberis s. str.) nested within the paraphyletic compound-leaved Mahonia (Adhikari et al., 2015; Kim & Janzen, 1998). Berberis now comprises c. 500 simple- leaved species (Berberis s. str.) and nearly 100 compound-leaved species (Mahonia Nutt) (Ahrendt, 1961), although these numbers may vary due to new revisions. Berberis are mostly shrubs found in temperate, subtropical and tropical regions of the world. It is the only genus of Berberidaceae present in South America. Compound-leaved Berberis are mostly found in the Northern Hemisphere, especially in North America and Asia (Ahrendt, 1961).

Simple leaved Berberis (Berberis s.s.) have two major centers of diversity: section Septentrionales in Eurasia with c. 300 species and section Australes in South America, with c. 200 species (Ahrendt, 1961). The number of species in Asia has been steadily increasing due to recent discoveries (Adhikari, 2012; Harber, 2016; Rajbhandari et al. 2016). In South America, however, Landrum (1999) recognized only 20 out of the 60 species listed by Ahrendt in the most recent taxonomic revision of Berberis for Eaustrales group of southern South America. Within South America, the species found in the tropics belongs to two subgroups. Most of the tropical Andean species are from the subgroup Aequinoctales with c. 90 species found across Colombia, Ecuador, Peru, and Bolivia. Lesser number of tropical species area found in the subgroup Eaustrales, with the section Laurinae found in southeastern Brazil, Uruguay, Paraguay and in the central Andes of Peru and Bolivia. The firsts molecular phylogenies of the group were done by Kim et al (2004a) using ITS sequences, including 22 of the 33 sections of simple and compound-leaved Berberis proposed by Arhendt (1961). More recent studies have increased the number of species sampled and have found many of the morphological sections proposed by Ahrendt 18

(1961) to be not supported by molecular evidence (Adhikari et al., 2015). Considering the geographical approximation in Neotropical sections of Berberis in Ahrendt (1961) and recent molecular evidence suggesting that these sections might not be monophyletic, their actual legitimacy should be reconsidered. Nevertheless, for the purposes of this work I will follow Ahrendt’s classification unless phylogenetic results clearly show evidence against it. The definition of Berberis used in this study will encompass both Berberis s.str. and Mahonia.

1.7 STUDY AIMS

Ideally, in order to test theories on the reason for different number of species in tropical and temperate regions, one should choose a group that is distributed in those regions and within which the species vary little in generation time. The genus Berberis, with a relatively high number of species and diversity centers located both in tropical and temperate regions, is an excellent model group to test and compare diversification rates between the tropics and temperate regions. An estimate of the number of species for the genus is known (Ahrendt, 1961; Landrum, 1999) and phylogenetic studies have given adequate understanding about the relationships among different geographic groups (Kim et al., 2004a; Kim et al., 2004b; Adhikari et al., 2015).

In the present study, I aim to produce a dated phylogeny of Berberis to test the hypothesis for the LDG that there are faster diversification rates in the tropics compared to temperate regions. Specifically, we will test whether:

1. Tropical South American Berberis form a monophyletic clade;

2. All South American species form a monophyletic clade;

3. Net diversification rates of tropical Berberis are higher than in temperate clades;

4. Rates of speciation during mountain uplifts are higher in the Tropics (Andes) compared with similar events at temperate latitudes (Himalayas);

5. Rates of diversification for tropical Andean Berberis are higher compared with southern Andean (temperate) Berberis.

19

2. MATERIALS AND METHODS

2.1 TAXON SAMPLING

The dataset for this study comprised of 102 taxa, representing 94 species of Berberis (Appendix, Table 8). Of these, 65 accessions of Berberis were previously published in Adhikari et al. (2015) and 29 accessions of Berberis were newly sequenced for this study. These new taxa include species found in tropical South America, with 16 species from the tropical Andes of Colombia and Ecuador and one (B. laurina Thunb.) from subtropical southern Brazil (Figure 1). Due to the questioned validity of the sections and subsections proposed by Ahrendt (1961) for Berberis suggested by posterior taxonomic (Landrum 1999) and most recently phylogenetic studies (Adhikari et al., 2015), the target of this study was to sample Berberis species from across all the temperate and tropical regions within the distribution range of Berberis. The samples were kindly provided by Missouri Botanical Garden (MO), Jardin Botanico de Bogotá Jose Celestino Mutis and Dr. Federico Luebert from the University of Bonn Herbarium, provided the sample from Brazil. All samples were sequenced for both nuclear ITS and the coding plastid ndhF gene. Both regions were selected for their capability to resolve species-level relationships within the genus (Kim et al., 2004a, Kim et al., 2004b, Adhikari et al. 2015). To make sure sampling was even across South America, occurrence data was downloaded from the Global Biodiversity Information Facility (GBIF) for all species included in this study (Error! Reference source not found.A) and compared with all the records for Berberis in the region (Error! Reference source not found.B) with ArcMap 10.1 (ESRI, USA).

Outgroups were selected based on previous phylogenetic analysis of Berberis (Kim et al., 2004b), and included representatives of four of the seven families of the order Ranunculales. The sister genus of Berberis, Ranzania T. Ito, was also sampled (Wang et al., 2009). The outgroup accessions included Akebia quinata (Houtt.) Decne. (Lardizabalaceae), Cocculus carolinus (L.) DC. (Menispermaceae), Glaucidium palmatum Siebold & Zucc., Hydrastis Canadensis L., Clematis drummondii Torr. & A. Gray (Ranunculaceae), Nandina domestica Thunb., Caulophyllum robustum Maxim. and Ranzania japonica T. Ito (Berberidaceae). All these genera belong to Ranunculales and are hence suitable as outgroups of Berberis, following Kim et al. (2004b). 20

FIGURE 1 DISTRIBUTION OF OCCURRENCE RECORDS FOR NEOTROPICAL BERBERIS FROM THE GLOBAL BIODIVERSITY INFORMATION FACILITY (GBIF). (A) SPECIES INCLUDED IN THE ANALYSIS. RED INDICATES B. LAURINA; ORANGE, TEMPERATES SPECIES FROM CHILE AND ARGENTINA; GREEN, TROPICAL ANDEAN SPECIES. (B) OCCURRENCE DATA FOR ALL BERBERIS SPECIES ACROSS SOUTH AMERICA.

2.2 LABORATORY METHODS

In order to prepare the samples for the DNA extractions, material from Silica gel-dried material and herbarium specimens were grinded using a mixer mill (TissueLyser II, QIAGEN Inc.) to disrupt leaf tissue. DNA extractions were performed using Qiagen DNeasy Plant Mini kit (QIAGEN Inc.) following the manufacturer’s protocol. All DNA samples were placed in the Edinburgh DNA bank. Information about the primers used on this study is provided in Table 1.

The nuclear ribosomal Internal Transcribed Spacer (ITS) region was amplified using the ITS1 (Kim et al., 2004a) and ITS8P (Moller & Cronk, 1997) primer pairs. Polymerase Chain Reaction (PCR) was carried out in the DNA Engine Tetrad 2 Peltier Thermal Cycler by Bio-RAD in 25 uL volume reactions containing 11.05 uL of H2O, 2.5 uL of 0.2 mM dNTPs, 2,5 uL of 10x NH4 reaction Buffer, 1.25 uL of 50mM MgCl2, 0.75 uL of each primer, 5 uL of 5x Combinatorial Enhancer Solution (CES), 0.2 uL of 5U/uL taq Polymerase and 1-2 uL of DNA template. The PCR programme had an initial denaturation at 95°C for 3 minutes, followed by 30 cycles of denaturation at 95°C for 1 minute, primer annealing at 52°C for 1 minute, 1-minute extension at 72°C and a final primer extension at 72°C for 10 minutes. 21

The plastid gene ndhF was amplified using the primers ndhF20F and ndhF2155R (Adhikari et al., 2015). Polymerase Chain Reaction (PCR) was carried out in the DNA Engine tetrad 2 Peltier Thermal Cycler by Bio-RAD in 25 uL volume reactions containing

14.65 uL of H2O, 2,5 uL of 0.2 mM dNTPs, 2,5 uL of 10x NH4 reaction Buffer, 1.25 uL of

50mM MgCl2, 1 uL of each primer, 0.8 uL 1 g/mL of Bovin Serum Albumin (BSA), 0.3 uL of 5U/uL taq Polymerase and 1-2,5 uL of DNA template. PCR programme started with an initial denaturation at 95°C for 3 minutes, followed by 30 cycles of denaturation at 95°C for 1 minute, primer annealing at 50°C for 1 minute, 2 minutes of extension at 72°C and a final primer extension at 72°C for 10 minutes.

Following amplification, PCR products were visualized by electrophoresis on a 1% agarose gel stained with SYBER Safe (Invitrogen Inc.) under Ultra Bright-blue light on the G BOX F3 (Syngene Inc.) in order to assess quality and quantity of product for posterior purification and sequencing reactions. PCR products were purified using ExoSAP-IT (Applied Biosystems, Foster City, California). Sequencing PCR of ITS was performed with the same primer that were used for amplification and was done in 10 uL volume reactions containing 6.68 uL of H2O, 2 uL of 5x BigDye buffer, 0.32 uL of 10 mM primer, 0.5 uL of bigDye (Applied Biosystems, Foster City, California) and 0.5-1 uL of purified PCR product. The sequencing PCR was composed of 30 cycles of 30 seconds denaturation at 95°C, primer annealing at 50°C for 20 seconds and primer extension at 60°C for 4 minutes, followed by incubation at 7°C forever.

For the ndhF gene, the forward and reverse primers, plus an internal primer, ndhF 727F (Kim et al., 2004b) were used in the sequencing reactions. Sequencing PCR was carried out in 10 uL volume reactions containing 6.68 uL of H2O, 2 uL of 5x BigDye buffer, 0.32 uL of 10 mM primer, 0.5 uL of bigDye (Applied Biosystems, Foster City, California) and 0.5-1 uL of purified PCR product. The sequencing PCR had 30 cycles of 30 seconds denaturation at 95°C, primer annealing at 50°C for 20 seconds and primer extension at 60°C for 4 minutes, followed by incubation at 7°C forever. Sequencing PCR products were sent to the GenePool Sequencing Facility at the University of Edinburgh for analysis.

22

TABLE 1. PRIMERS FOR BERBERIS USED IN THIS STUDY.

DNA region Primer Primer sequence Reference ITS1 5′-GGA AGG AGA AGT CGT AAC AAG Kim & al., 2004a G-3′ ITS ITS8.p 5′-CAC GCT TCT CCA GAC TAC A-3′ Moller & Cronk, 1997 ndhF ndhF20F 5’-TGC ATG AT CAT ACC TTG G-3’ Adhikari et al., 2015 ndhF727F 5’-CAA TTC CCC CTT CAT GTA TGG Kim & al., 2004b TTA CC-3’ ndhF2155R 5’-TCC GCG CCC TAT ATA TTT T-3’ Adhikari et al., 2015

2.3 SEQUENCE ALIGNMENT

Sequencher 5.2 software (Gene Codes Corp., Ann Arbor, Michigan, USA) was used to read, edit and assemble contigs using sequences obtained from forward, reverse and internal sequencing PCR reactions. Sequences were aligned using clustalW multiple alignment algorithm (Thompson et al., 1994) as implemented in BioEdit v7.2.5 (Hall, 1999). A later manual adjustment of the alignment was performed to assure high quality of alignment. For the ndhF gene coding region, the alignment was done at codon level by hand on Mesquite (Maddison & Maddison, 2016). Posteriorly, the newly generated sequences were aligned by hand to Berberis alignment matrix used in Adhikari et al. (2015). ITS and ndhF sequences were subsequently concatenated using Geneious v7.18 software.

2.4 PHYLOGENETIC ANALYSIS

A final data matrix of the combined ITS and ndhF data consisted of sequences for 92 species, with ndhF only for the eight taxa used as outgroups. Taxa with only ITS data available were included in the final combined matrix, leading to some missing data, considering that better taxon sampling overcomes the negative impacts of missing data (Zheng & Wiens, 2015). Phylogenetic analyses were carried out using both Maximum- likelihood and Bayesian methods implemented in RAxML-HPC v.8 (Stamatakis, 2014) on XSEDE, implemented in the CyberInfrastructure for Phylogenetic Resarch (CIPRES) Gateway v3.3 (Miller et al., 2010) and MrBayes x64 v3.2.5 (Ronquist et al., 2012) respectively. The best-supported model of nucleotide sequence evolution for each partition 23 was determined using MrModeltest v2 (Nylander, 2004) for ITS and each ndhF codon position separately based on the Akaike Information criterion (AIC). The model selected for each partition can be found in Table 2. Gaps and indels were treated as missing data in both RAxML and MrBayes.

TABLE 2.SUBSTITUTION MODELS FOR EACH PARTITION OF THE BERBERIS DATASET. EACH MODEL WAS SELECTED BASED ON THE AKAIKE INFORMATION (AIC) IN MRMODELTEST V.2 (NYLANDER, 2004).

Region Model selected ITS1 & ITS2 SYM+I+G ITS 5,8S SYM ndhF - 1st Codon Position GTR+G ndhF - 2nd Codon Position GTR+I+G ndhF - 3rd Codon Position GTR+G

Phylogenetic analyses were carried out using both Maximum-Likelihood (ML) and Bayesian methods available on XSEDE CyberInfrastructure for Phylogenetic Research (CIPRES) Gateway v3.3 (Miller et al. 2010). Maximum-likelihood analyses were run using RAxML-HPC v.8 (Stamatakis, 2014) with GTRCAT model of nucleotide substitution, which approximates to a GTR+I+G model. Substitution parameters were estimated independently for the ITS and ndhF partitions. The rapid hill-climbing algorithm was selected. Taxa with exactly the same sequence alignment were discarded before starting the analysis. The species Akebia quinata was specified as the outgroup. The analysis included 10 starting trees and 1,000 ML bootstrap replicates. In order to identify rogue taxa, that is, the taxa that cause loss of Bootstrap support and lack of resolution due to missing data or other reason, the RogueNaRoK algorithm (Aberer et al., 2013) was used. Four taxa with a high value were discarded from the matrix and the software was run again.

Bayesian analyses were run using MrBayes x64 v3.2.5 (Ronquist et al., 2012). Two parallel independent runs were run for 15 million generations sampling every 10,000 generations, incorporating one cold and three heated chains per run. Time series plots and Effective sample size (ESS) were analyzed using TRACER v1.6 (Rambaut & Drummond, 2013) to ensure convergence between runs and ESS values above 200. Ten percent (1500) of the sampled trees were removed as burn-in. The trees were summarized in a 50% majority rule tree with posterior probability values for each clade. In order to see evidence of incongruence between nuclear and chloroplast datasets, independent analyzes were run 24 using only one of the sets. Not mayor incongruences were found between ITS and ndhF tree topologies for both ML and Bayesian methods.

2.5 FOSSIL CALIBRATION

Our study used two fossil calibrations. The fossil of Prototinomiscoum belonging to Menispermaceae found in Central Europe, dates back to the Turonian c. 91 Ma (Wang et al., 2007). The latter fossil was used to calibrate the stem node between Menispermaceae and Ranunculaceae. The Second fossil calibration was used to constrain the stem node of Berberis. Fossils of Berberis have been found throughout the present distribution range of the genus. Fossils from the Pleistocene have been found from the Northern Hemisphere and South America. A fossil of a compound-leaved Berberis dating back to the Middle Eocene (47.8 – 38 Ma) is represented in the Western North American Terciary floras (Manchester, 1999). Considering that in order for a structure to appear in the fossil record, it must be abundant in that time. In that sense, the fossils used here can only provide information about when these groups become abundant (Magallón, 2004). Using a conservative approximation (Magallón, 2004) these fossils were used as a constraint age for the crown group of Berberis-Ranzania (i.e, stem node of compound-leaved Berberis) to give a minimum age for the split between these two groups.

The presence of Berberis in South America dates back to the Oligocene (Selles & Hinojosa, 1997), Late-Oligocene (Cesari et al., 2015) and early Neogene (Torres et al. 2013) in the Southern Andes and Patagonia. In the Central Andes, the oldest record is found in the Jakokkota flora from west-central Bolivia, with an estimate age of 10.66 Ma (10.72-10.60) (Gregory-Wodzicki et al. 1998). Due to the polyphyly of the South American clade, it was not possible to use these fossils to calibrate the phylogeny

2.6 DIVERGENCE AGE ESTIMATION Dating analyses were conducted using two different methods. The penalized likelihood age estimation method (Sanderson, 2002) and the Bayesian method with an uncorrelated lognormal relaxed molecular clock (UCLN). 25

The Penalized Likelihood method was conducted in treePL (Smith & O’Meara, 2012) using as input the best tree found in the ML analysis in RAxML. In order to find the appropriate smoothing parameter, a full cross-validation analysis was carried out in treePL. The cross-validation tested 13 smoothing values (λ) from 100,000 to 1x10-8 with an order of magnitude of difference between them. The smoothing value was selected based on the cross validation test was 10,000. Node calibrations were done with minimum and maximum dates for the outgroups and with a minimum age constraint for Berberis s.l.

The Bayesian dating analysis was performed using Bayesian Evolutionary Analysis by Sampling Trees (BEAST) software (Drummond and Rambaut, 2007, Drummond et al 2012). An input file was generated using BEAUTi v.1.8.2. Tree topologies and clock models were linked for both partitions but substitution models were unlinked for the ITS and ndhF partitions. The MrModeltest v2 (Nylander, 2004) software was used to identify the best model for each partition. The SYM+I+G model was applied to the ITS partition, modifying the GTR model with base frequencies all equal. In order to avoid make the model excessively complex and obtain spurious results, we decide to treat the ndhF gene as a single partition. The GTR+I+G model was applied to the ndhF gene, with estimated base frequencies and four gamma categories. The uncorrelated lognormal relaxed molecular clock (Drummond et al., 2006) was selected in order to allow for changes in rates between ancestral and descendant lineages (Forest, 2009). Fossil calibrations were modeled with a 95% highest posterior density constrained to a lognormal distribution, with an upper limit equal to the end of the geologic period. The model selected to describe how speciation events are distributed over time was the mechanistic Birth-Death Incomplete- Sampling (Stadler, 2009). The starting tree was chosen randomly. Two independent analyses were run for 50 million generations, with trees sampled every 10,000 generations, for a total of 5,000 samples per run. The logfiles were analyzed with Tracer v1.6 (Rambaut et al., 2014) in order to check convergence between runs and confirm estimated Effective Sample Size (ESS) parameters were above the required value of 200. The tree files from both runs were combined using LogCombiner v1.8.2 (Drummond & Rambaut, 2007) using a burnin of 500 trees (10%). The combined tree file was then the input file of TreeAnnotator v.1.8.2 (Rambaut & Drummond, 2013), from which the Maximum Clade Credibility (MCC) tree with mean node heights was calculated. The minimum posterior probability limit was set to 0.05, as diversification analyses in BAMM require as an input an ultrametric, fully branching tree. 26

2.7 CALCULATION OF DIVERSIFICATION RATES USING BAMM SOFTWARE

Fossil evidence suggests that Berberis dates back to at least the Paleocene (60 Ma), and fossil evidence suggests large extinction of lineages, such as simple-leaved Berberis in North America after the Miocene (Ramı́rez & Cevallos-Ferriz, 2000; Adhikari et al., 2015). In addition, there is a link between mountain ranges and diversity centers for the genus. This should be considered when choosing the methodological framework used to test diversification analyses. MEDUSA has been shown to not be as robust as BAMM (Rabosky, 2014) when the data violates the assumption of invariant evolutionary rates through time (Rabosky, 2014). For the present dataset, the assumption of a constant rate of diversification through time appears to be very unlikely. In order to identify distinct rate- shifts across Berberis phylogeny and estimate diversification rates, we decided to use the algorithm implemented in Bayesian Analysis of Macroevolutionary Mixtures (BAMM) (Rabosky, 2014). The program is able to model speciation and extinction rates over time and between different lineages using a reversible-jump Markov chain Monte Carlo (jMCMC). This model was selected because it does not assume constant rates of speciation and extinction through time within lineages and does not require prior information about the location and number of rate shifts. As the topology for both BEAST and treePL chronograms was the same, the analysis only used the MCC tree obtained from BEAST. Because outgroups were poorly sampled in the phylogeny presented here, the outgroup was removed prior to rate shift analysis in order to focus on rate shifts within the ingroup using the package ape (Paradis et al., 2004) in R (R development team, 2013). Four MCMC simulations were run for 10 million generations, sampling every 1,000 and discarding the 10% of the sampled trees as a burnin. In order to evaluate convergence between MCMC runs we checked the log-likelihood of the ESS and the number of shifts events on each sample using the package CODA, expecting minimum values of 200 (Plummer et al., 2006). The R package BAMMtools (Rabosky et al., 2014b) was used to obtain the prior values for the control file with the evolutionary rate parameters. The function setBAMMpriors was used to generate a prior block that scales the distributions based on the root depth of the tree (expectedNumberOfShifts = 1, lambdaInitPrior = 0.38, lambdaShiftPrior = 0.11, muInitPrior = 0.38). The expected number of shifts selected was changed from one to two, and finally to 100, in order to assess whether the results were robust in relation to the selected prior. Because of the incomplete and uneven sampling 27 across the ingroup, it was necessary to specify the sampling probabilities for each clade in the model. To check if the missing taxa had a significant effect on the estimations presented here, one analysis was ran assuming complete taxon sampling and another taking into consideration the sampling percentages for each clade (Table 3). While the taxon sampling in the phylogeny presented here was generally low, (nearly 16% of the species considered by Ahrendt (1961)), studies suggests that BAMM is able to identify rate shifts accurately even with incomplete taxon sampling (Spriggs et al., 2015)

TABLE 3. SAMPLING FOR EACH BERBERIS CLADE CONSIDERED FOR THE BAMM ANALYSIS

Clade Node Total number Number % sampled of species of species sampled Compound-leaved 5 100 9 9 Berberis Septentrionales 10 300 59 19 Southern South 8 24 13 54.2 America Tropical South 7 110 11 10 America

The package BAMMtools was used to identify the credible set of distinct rate-shift configurations that accounts 95% of the probability of the data and to compute the Bayes factors. These rate shifts with a marginal probability greater than expected under the prior distribution were considered “Core-shifts” (Rabosky et al., 2014b). In order to infer clade- specific rate heterogeneity, rates of speciation (λ), extinction (µ) and net diversification (Dr = λ - µ) and their 90% highest posterior density (HPD) were estimated for each clade of interest with the function getCladeRates. In this way, an average across all sampled trees was obtained, so every clade could have potentially had significantly distinct rates of speciation that would be ignored if the only focus were the best shift configuration (Rabosky et al., 2014a). In order to compare the speciation rates of the clades of interest with the background rate of Berberis, a relative rate was calculated, as the mean value of each specific clade, divided by the mean of all the other clades. Ratios between different clade rates of speciation were also computed. Rate-through-time curves were plotted for Berberis lineage and for specific clades of interest. A macroevolutionary cohort matrix was done in order to show pairwise probabilities for any two lineages to share the same rate dynamics (Rabosky et al., 2014b). 28

3. RESULTS

3.1 PHYLOGENETIC ANALYSIS

The aligned ITS sequence, comprising the spacers ITS1 and ITS2 plus the 5.8S and 26S regions, varied between 640-722 base pairs in length. The aligned matrix for the ndhF gene was between 1745-2106 base pairs long. The combined ITS and ndhF matrix was 2852 bp long including gaps.

While ITS topology shows better resolution between the Septentrionales and Australes clades (Figure 10), the ndhF topology shows higher resolution and posterior probabilities within the Septentrionales clade (Figure 11). Because topologies across both ITS and ndhF datasets were consistent in both ML and Bayesian analyses, the description of the results is based on the 50% majority rule consensus tree obtained in the Bayesian analyses for the combined analyses (Figure 2).

The results from the combined data set are highly consistent with the findings by Adhikari et al. (2015). Both combined analyses show strong support for the basal position and the paraphyly of compound-leaved Berberis. Berberis higginsae and B. nevinii diverge early from the rest of the compound and simple-leaved Berberis with high branch support (PP= 1, BS= 100; Figure 2). The rest of the compound-leaved Berberis form a well-supported clade (PP= 0.98) as a sister clade of simple-leaved Berberis. The topology within this compound-leaved Berberis gives support for the group Orientales (PP= 0.98), comprised for species found in Central and East Asia. The only species of this group found in North America is B. nervosa, which appears as a sister clade of Occidentales, but with low support (PP= 0.57). The rest of Occidentales group is well supported (PP=1.0).

Within the simple-leaved Berberis, two distinct groups are well defined: Septentrionales (PP= 0.97), with mostly Northern hemisphere species plus three tropical South American species: B. beauverdiana, B. quindiuensis and B. glauca. The position of these three taxa in Septentrionales is consistent for both ITS (Figure 10) and ndhF (Figure 11), although the relative position within the clade is uncertain. The other group consists of the remaining South American species (PP=0.81, BS=37), including Southern temperate species, eastern tropical species from Brazil (B. laurina) and Tropical Andean species (Figure 2).

leaved 29

-

leaved

leaved

leaved

-

- -

hern

Compound Berberis

Simple TropicalAndes

Outgroup

Simple Sout temperate Andes Brazil +

Septentrionales + Septentrionales SouthAmerica Simple

FIGURE 2. COMBINED BAYESIAN 50% MAJORITY RULE CONSENSUS TREE OBTAINED FROM NDHF AND ITS SEQUENCES OF 102 ACCESSIONS REPRESENTING 94 SPECIES OF BERBERIS. BAYESIAN POSTERIOR PROBABILITIES (PP) AND MAXIMUM LIKELIHOOD BOOTSTRAP VALUES (BS) ARE SHOWN ON EACH BRANCH (PP/BS). SPECIES ENCLOSED IN BLACK BOXES CORRESPOND TO TROPICAL SOUTH AMERICAN SPECIES FOUND NESTED WITHIN THE SEPTENTRIONALES CLADE. 30

Two species from the group Eaustrales (Southern South America) appear at the base of this group (B. congestiflora and B. empetrifolia). The monophyly of the remaining Eaustrales is poorly support (PP= 0.57) and includes B. laurina (Section Laurinae). The sister clade has strong support (PP= 0.99) comprised by the remaining eleven tropical South American species from the sections Goudotiae, Quindiuenses, Confertae and Virgatae, all belonging to Aequinoctiales subgroup.

3.2 DIVERGENCE TIME ESTIMATION

The maximum clade credibility phylogeny produced by BEAST, slightly differs in topology from the 50% majority rule consensus tree produced in MrBayes. The tropical South American clade is well supported (PP= 0.99) after B. ovalifolia. The temperate South American clade is also well supported (P= 0.96). The main difference is in the Septentrionales group. A clade consisting of mostly Himalayan and Chinese species appears well-supported (PP= 0.92), while a group consisting mainly in Asian and two North American species, B. canadensis and B. fendleri has good support as well (PP= 0.94).

The maximum clade credibility tree with all divergence age estimates and 95% highest posterior density (HPD) intervals is shown in Figure 3. The crown node Berberis is estimated to have diverged from Ranzania in Middle-Early Eocene (43.81 Ma, HPD= 38.59-49.67). Simple-leaved Berberis diverged from the compound-leaved Berberis during the late Miocene (5.69 Ma, HPD= 3.55-8.23). The split between Septentrionales and South American clade occurred in the early late Miocene – early Pliocene (4.39 Ma, HPD= 2.66- 6.24). The South American crown clade dates back from the Pliocene (3.61 Ma, HPD= 2.21-5.29). Age estimates from the analysis for nodes of interest are shown in Table 4.

31

leaved

leaved

- -

leaved

leaved

-

- clade

Himalayan

Outgroup Outgroup

Septentrionales Simple

Southern temperate + Andes Brazil Simple

Tropical Tropical Andes Simple

Berberis Compound

8

11

12

6

7

5

10

9 4

3

2

MIOCENE

OLIGO

1

EOCENE

O

PALE

CRETACEOUS

FIGURE 3. COMBINED MAXIMUM CLADE CREDIBILITY CHRONOGRAM FOR COMBINED ITS AND NDHF OBTAINED IN BEAST ANALYSIS. NODE HEIGHTS INDICATE MEAN AGES. BLUE NODE BARS INDICATE 95% HPD AGE RANGES. POSTERIOR PROBABILITY VALUES ABOVE 0.90 ARE SHOWN FOR EACH BRANCH. NUMBERS ON BLACK CIRCLES INDICATE CLADES IN TABLE 4. 32

TABLE 4. MEAN DIVERGENCE TIMES, WITH 95% HIGEST POSTERIOR DENSITIES (HPD) AND POSTERIOR PROBABILITY SUPPORT FOR SELECTED NODES FOR THE MCC TREE GENERATED BY BEAST

Node Bayesian Posterior Mean Divergence Age (Ma) 95% HPD number probabilities Ma 1 1.00 43.81 38.59 - 49.67 2 1.00 10.74 5.64 - 16.64 3 0.94 7.91 4.70 - 11.55 4 0.94 5.69 3.55-8.23 5 1.00 3.52 1.76 -5.52 6 0.85 3.61 2.21-5.29 7 0.99 3.07 1.79 -4.68 8 0.96 2.44 1.38-3.69 9 1.00 4.39 2.66-6.24 10 0.98 3.83 2.43-5.55 11 0.92 3.09 1.82-4.46 12 0.94 2.82 1.64-4.20

3.3 RATE SHIFTS AND DIVERSIFICATION RATES

BAMMtools indicate a conservative value of one expected rate shift for the prior modeling. Before analyzing the output from BAMM, we assessed convergence among the four MCMC chains and ESS above 200 for both number of shifts and log-likelihoods. The ESS value for the number of shift and for the log-likelihood was 1,538.39 and 2,699.98, respectively. The runs converged after 5,000 generations, so a conventional burnin of 10% was applied. The BAMM analysis ran under the MCC tree sampled models containing up to seven macroevolutionary regimes.

Results show the highest posterior probability assigned to a model with only one rate shift with 75% of the sampled configurations accounting for this model (Figure 4). Posterior probability for models with two shifts is lower (20%), while models with zero or more than two shifts is extremely low (>0.03%). In order to assess the sensitivity of the analysis to the prior, several runs were done with higher number of expected number of shifts up to 100. In all cases, even if the sampled configurations included up to 10 rate shifts, the highest posterior probability was for the model with one rate shift. 33

FIGURE 4. PRIOR AND POSTERIOR PROBABILITY OF MODELS WITH DIFFERENT NUMBER OF SHIFTS FOR BERBERIS DATASET. A HIGHER POSTERIOR PROBABILITY COMPARED WITH THE PRIOR SUGGESTS A SIGNIFICANT CONFIGURATION. THE NUMBER OF SAMPLED MODELS CONSIDERED UP TO SEVEN RATE SHIFTS ACROSS THE PHYLOGENY (EXCLUDING OUTGROUPS).

Bayes factor evidence in favor of a model with one rate shift in comparison to the null model with zero shifts is 6720. Values over 12 should be taken into consideration as significant evidence (Rabosky, 2014), so the results here indicate strong support for a one- shift model as the overall best model. Bayes Factor values can be seen in Table 5.

TABLE 5. BAYES FACTORS FOR MODELS WITH DIFFERENT NUMBER OF SHIFTS.

Number Bayes of Shifts Factor 0 1 1 6720 2 3652 3 1400 4 672 5 192 6 192 7 128

The branch-specific marginal shift probabilities were computed in order to localize the occurrence of shifts along the branches of the MCC tree (Figure 3). These values show evidence in favor for a model with a shift at the base of Berberis after the divergence of B. 34 nevinii (MP = 0.95) and lower support for a shift at the base of the Southern Temperate clade (MP = 0.06). The set with distinct shift configurations that explains 95% of the posterior distribution contains only two different configurations. All configurations within the 95% credible shift set contain a core shift in the Berberis crown node after the divergence of B. nevinii. The single most frequent shift configuration, represented by 4,052 out of 4,310 (90%) samples, recovers a model with a single rate shift after the divergence of B. nevinii (Figure 6). The second most sampled configuration represented by 258 configurations (5.7%) shows two shifts: one after the divergence of B. nevinii, the same as the first set, and another in the crown node of Southern South American Berberis. None of the configurations within the 95% credible shift set presented a rate shift towards the Tropical clade. The branch-specific marginal probabilities show a high marginal probability (MP = 0.95) for a shift in the branch after B. nevinii (Figure 5). In the BAMM analysis assuming complete taxon sampling, the same branch is identified to show a diversification rate shift. The credible shift set accounting for 95% of the probability of the data show a single configuration with a rate shift on the same branch as the model considering sampling fractions.

TABLE 6. CLADE-SPECIFIC SPECIATION (Λ), EXTINCTION (µ) AND NET DIVERSIFICATION RATES (DR), WITH THE 90% HIGHEST POSTERIOR DENSITY SHOWN FOR SELECTED CLADES IN BERBERIS FROM THE BAMM ANALYSIS. RATES ARE SHOWN IN UNITS OF SPECIES PER MILLION OF YEARS.

Clade λ 90% HPD µ 90% HPD Dr 90% HPD

Berberis s.l. 2.20 1.48-3.11 1.25 0.32-2.32 0.96 0.69-1.24 Tropical 2.59 1.67-3.69 1.37 0.24-2.66 1.21 0.89-1.55 Temperate 2.56 1.72-3.60 1.37 0.29-2.60 1.19 0.88-1.52 temperate South 2.44 1.42-3.54 1.27 0.18-2.50 1.17 0.80-1.52 America Septentrionales 2.61 1.73-3.70 1.40 0.29-2.68 1.21 0.89-1.54 Himalaya 2.62 1.74-3.73 1.41 0.30-2.68 1.22 0.90-1.56 Andes 2.49 1.64-3.53 1.32 0.25-2.51 1.18 0.85-1.51

35

FIGURE 5. EVIDENCE OF SHIFTS OCCURRING ALONG BERBERIS BRANCHS ACCORDING TO BAMM. (A) ORGINAL CHRONOGRAM OF BERBERIS FROM BAYESIAN RELAXED CLOCK OBTAINED IN BEAST. (B) SAME PHYLOGENY, SHOWING BRANCHES PROPORTIONAL TO THE MARGINAL PROBABILITY OF A SHIFT OCCURRING IN THE PARTICULAR BRANCH. LONGEST BRANCHES HAVE BEEN LABELED. THESE BRANCHES ARE THE CROWN NODE OF BERBERIS AFTER THE DIVERGENCE OF B. NEVINII (MP= 0.95) AND THE BRANCH LEADING TO THE SOUTHERN TEMPERATE CLADE (MP= 0.06). BOTH BRANCHES HAVE BEEN LABELED FOR REFERENCE.

The mean speciation and net diversification rate for Berberis is 2.02 and 0.96 speciation events per million years (Ma-1), respectively (Table 6). The speciation rates for all clades studied are slightly higher than the background Berberis rate (Table 7, Figure 8). The relative rate of speciation for Septentrionales and Himalayan clade are the highest, with 1.39 (HPD = 1.19-1.72) and 1.29 (HPD= 1.14-1.51), respectively. Tropical, temperate and Andean clades presented show relative rates of speciation. The lowest relative rate is found in the Southern South America clade, with 1.14 (HPD= 0.62-1.25). Speciation rate-ratios between clades show little difference among clades. The highest speciation rate-ratios are shown between the tropical (1.11 HPD= 0.68-1.26), Septentrionales (1.13 HPD= 0.86- 1.27) and the Himalayan (1.14 HPD= 0.71-1.35) clade relative to the Southern South American clade. Rate through time (RTT) plots in Figure 9 also show evidence for increasing speciation rates for Berberis towards the present. With the exception of the Southern Temperate clade, all clades show slightly higher speciation rates than the background rate of Berberis (Figure 9). 36

F = 0.9 F = 0.057

Compound- leaved Berberis

Tropical Andes

Southern Temperate Andes + Brazil

Mean speciation rate (species Ma-1) Septentrionales

FIGURE 6 PHYLORATE PLOTS FOR SPECIATION SHOWING THE TWO DISTINCT SHIFT CONFIGURATIONS WITHIN THE 95% CREDIBLE SHIFT SET. MODELS ARE LISTED IN ORDEROF FREQUENCY (F). COLORS MAPPED IN THE PHYLOGENY SHOW MEAN SPECIATION RATE. RED INDICATING HIGH SPECIATION AND BLUE LOW SPECIATION RATES. CIRCLES DENOTE THE LOCATION OF THE CORE RATE SHIFTS, INDICATED IN BAMMTOOLS. RED CIRCLES INDICATE BRANCHES WITH SIGNIFICANT INCREASE IN SPECIATION RATES AND BLUE CIRCLES INDICATE NODES IDENTIFIED WITH SIGNIFICANT DECREASE IN SPECIATION RATE.

37

The macroevolutionary cohort matrix shows a shared macroevolutionary rate dynamic across Berberis, except for B. higginsae and B. nevinii, which share their own dynamics (Figure 7). There is a clear difference in the probability (0%) of both B. nevinii and B. higginsae belonging to the same macroevolutionary cohort as the rest of Berberis (Figure 7). There is a weaker support for the Southern Temperate clade to share a distinct macroevolutionary dynamic and belong to a different cohort (Figure 7).

Similarity index A

B

C

FIGURE 7. MACROEVOLUTIONARY COHORT MATRIX FOR SPECIATION RATES IN BERBERIS DERIVED FROM BAMM ANALYSIS. EACH CELL IN THE MATRIX DEPICTS THE PAIRWISE PROBABILITY THAT TWO SPECIES SHARE THE SAME MACROEVOLUTIONARY RATE REGIME. A VALUE OF 1 IS ASSIGNED TO A PAIR OF SPECIES THAT SHARE A SAME MACROEVOLUTIONARY REGIME, AND A VALUE OF 0 IF THEY DO NOT. (A) MACROEVOLUTIONARY COHORT OF B. HIGGINSAE AND B. NEVINII. (B) MACROEVOLUTIONARY COHORT OF TEMPERATE SOUTH AMERICAN BERBERIS. (C) MACROEVOLUTIONARY COHORT OF MOST BERBERIS.

38

A D

E B

C F

FIGURE 8. HISTOGRAMS OF RELATIVE RATES OF SPECIATION FOR BERBERIS CLADES.THE RELATIVE RATES OF SPECIATION ARE ESTIMATED DIVIDING THE MEAN SPECIATION RATE OF THE STUDY CLADE BY THE BACKGROUND SPECIATION RATE ACROSS THE ENTIRE BERBERIS EXCLUDING THE STUDY CLADE. RELATIVE VALUE OF ONE REFERS TO CASES WHERE NO DIFFERENCE BETWEEN THE CLADE AND THE BACKGROUND SPECIATION RATE CAN BE FOUND. (A) TROPICAL CLADE, INCLUDING ONLY SPECIES FROM TROPICAL ANDES. (B) TEMPERATE CLADE, INCLUDING SPECIES FROM TEMPERATE SOUTH AMERICA AND SEPTENTRIONALES. (C) TEMPERATE SOUTHERN SOUTH AMERICAN CLADE. (D) SEPTENTRIONALES CLADE. (E) HIMALAYAN CLADE. (F) ANDEAN CLADE, INCLUDING BOTH TEMPERATE AND TROPICAL ANDEAN SPECIES.

39

TABLE 7. RELATIVE SPECIATION RATES FOR CLADES OF INTEREST IN BERBERIS. RELATIVE SPECIATION RATES ARE OBTAINED DIVIDING THE SPECIFIC CLADE RATES BY THE BACKGROUND RATES OF BERBERIS, WITHOUT CONSIDERING THAT PARTICULAR CLADE.

Clade Relative λ 90% HPD Southern South America 1.14 0.62-1.25 (temperate) Tropical South America 1.20 1.08-1.31 Septentrionales 1.39 1.19-1.72 Himalayan 1.29 1.14-1.51 Andeans 1.19 0.84-1.28 Temperate (temperate South 1.19 1.09-1.22 America + Septentrionales)

A D

B E

C F

Time before present (Ma)

FIGURE 9 SPECIATION RATES THROUGH TIME OF BERBERIS COMPARED WITH LINEAGES OF INTEREST. BLACK LINE INDICATES THE MEAN SPECIATION RATE AT ANY INDICATED TIME ACROSS BERBERIS. GREY SCALE INDICATE 90% BAYESIAN CREDIBLE REGIONS ON THE DISTRIBUTION OF RATES. (A) TROPICAL CLADE, INCLUDING ONLY SPECIES FROM TROPICAL ANDES. (B) TEMPERATE CLADE, INCLUDING SPECIES FROM TEMPERATE SOUTH AMERICA AND SEPTENTRIONALES. (C) TEMPERATE SOUTHERN SOUTH AMERICA CLADE. (D) SEPTENTRIONALES CLADE. (E) HIMALAYAN CLADE. (F) ANDEAN CLADE, INCLUDING BOTH TEMPERATE AND TROPICAL SOUTH AMERICAN SPECIES. 40

4. DISCUSSION

4.1 DIVERSIFICATION OF BERBERIS

Our results obtained by BAMM suggest that Berberis has incurred an important diversification burst soon after the divergence of B. nevinii. All other species can be assigned to a single Macroevolutionary cohort (Figure 7). Berberis speciation dynamics can be defined by a single important shift in speciation rates, and a tendency of the lineage to increase their speciation rates towards the present. The difference in diversification rates between temperate and tropical or Andean and Himalayan clades seem insignificant compared with this early rate shift identified close to the crown node of Berberis (Figure 5, Figure 6). Ranzania, the sister genus of Berberis (Kim, 2004b), contains only a single species (R. japonica) similar to most other genera in Berberidaceae. In contrast, Berberis has diversified more, possibly in response to mountain uplift (Ahrendt, 1961). The finding of rate shifts by BAMM suggest an early important change in the macroevolutionary dynamics of speciation in Berberis.

Possible morphological differences could explain the observed rate shift, but no clear morphologic or ecologic characters divide B. higginsae and B. nevinii from the rest of Berberis, reducing the chances to explain this burst in speciation by means of ecological interactions or key innovations. Similar evidence is found in Andean Lupinus radiation, where no evident key morphological traits have been identified that could triggered the diversification (Hughes & Eastwood, 2006). These findings suggest that ecological opportunities, rather than key innovations may be the cause of species diversification in these lineages. It is possible, however, that the basal lineages of Berberis have undergone severe extinctions, as suggested by diverse fossil flora found in Miocene Western North America where the extant taxa sampled here (B. higginsae and B. nevinii) occur (Ramı́rez & Cevallos-Ferriz, 2000). An early mass extinction leaving only a few extant lineages, followed by a posterior diversification of the sister clade may be a possible explanation for the detected basal rate shift in Berberis.

Nevertheless, comparing the clade-specific speciation rates, we can unveil differences that might give insights into the heterogeneity in the diversification rates among different lineages of Berberis. Considering the important diversification shift close to the crown node of Berberis, this only shift cannot fully explain diversity patterns found in this group, 41 and the considerably higher number of species in montane areas in the Himalayas and the tropical Andes. Although BAMM is able to model heterogeneity in rates across the phylogeny, studies have suggested that BAMM may not have enough power to detect rate shifts when speciation and extinction rates are not constant, giving to imbalanced topologies (Shi & Rabosky, 2015; Moore et al. 2016). All analyzed lineages exceed the background speciation rate of Berberis (Figure 8, Table 7), with Septentrionales clade showing the highest rate (rel. λ= 1.39). Within this lineage, the Himalayan clade has the second highest relative speciation rate (rel. λ=1.29). Within the South American lineages, the relative speciation rates are considerably lower, with the lowest value found in the Southern South America lineage (rel. λ=1.14). The tropical lineage has slightly higher values (rel. λ=1.195) but the overall pattern in the Andes is similar. (rel. λ=1.194). Relative speciation rates in all temperate lineages (that is, excluding the Andean tropical lineage) are not significantly different from the rates observed in the tropical clade (rel. λ= 1.18). The results for Berberis therefore are not those that one would expect based on the hypothesis that the LDG is due to differences in diversification rate between tropical and temperate lineages. They are therefore more consistent with alternative hypotheses, suggested in the introduction that, for example, the LDG might be due to tropical lineages having existed for longer periods of time and having occupied greater areas for longer periods than temperate ones.

Within the tropical Berberis, our findings supports previous studies that show that colonization of the Andes have caused no changes in speciation rates. Findings by Cardinal-McTeangue et al. (2016) on Brassicales found no shifts in speciation rates in Caricaceae, Capparaceae and Tropaeolaceae, widespread and species-rich elements found mostly in tropical Andes. Salariato et al (2016) found similar patters among the tribes Cremolobeae, Eudemeae and Schizopetalae (Brassicaceae). These groups colonized the Andes in the early Pliocene and speciated towards the South until the Mediterranean Matorral of Chile without important shifts in the speciation rates. These clades showing no response to Andean orogeny also diversified before the clades that show an adaptive radiation (Madriñán et al 2013; Hughes & Eastwood, 2006; Nürk et al., 2013a, b; Salariato et al., 2016). For Berberis, although our age estimates derived here fall within the Pleistocene, older fossil evidence from Late Miocene (10.66 Ma) from Bolivia (Gregory- Wodzicki et al. 1998) and Oligocene beds in Chile (23-24 Ma; Selles & Hinojosa, 1997), show that the genus was present in South America before Pleistocene radiations. These 42 fossils were not used here to time-calibrate the phylogeny because of the difficulty of placing them in the phylogeny. Future studies should explore the phylogenetic placement of these fossils in Berberis in order to derive more robust age estimates for South American species.

Our results contrast with the findings of recent radiations in the Andean tropics from multiple studies, especially in the northern region. Large adaptive radiations found in Lupinus L. (Hughes & Eastwood, 2006), Jamesonia H. & G. (Sánchez-Baracaldo 2014) and Draba L. (Koch & Al-Shehbaz, 2002) support the idea that the Andes have acted as a “cradle” of species, i.e., that most lineages have originated in recent times with high diversification rates (Pennington & Dick, 2010). The results by Nürk et al. (2013a, b) in Hypericum colonizing the northern Andes give more insights into these recent diversification events. Following Hypericum adaptation towards temperate climates in the Oligocene, the colonization of lowland tropics occurred first, and then a posterior colonization and radiation occurred in the high Andean lineages. Särkinen et al. (2012) recognized the same pattern, where lineages occurring at high elevation in the Andes show significantly younger ages compared to lower montane lineages.

Moreover, the evidence for adaptive radiation in Páramo region, confirms the propositions for a Pleistocene origin for these biomes caused by recent climatic events (Luebert & Weigend 2014). Madriñán (2013) reported incredibly high speciation rates for 14 genera occurring in the Páramo region in the northen Andes, including Berberis. The values obtained in the present study for tropical Berberis are considerably higher compared with Madriñán et al. (2013). While the net diversification rate in the Tropical lineage was 1.21 species Ma-1 (0.89-1.55) (Table 6), their values were 0.73 species Ma-1, although the average diversification rate for all Berberis lineage is more similar. Reasons for these observe differences in net diversification rates between the two studies may be many. The crown node is within the 95% HPD found by Madriñán et al. (2013) in the Pleistocene (3.61 Ma, HPD= 2.21-5.29, 3.80 Ma, 0.07-9.70, respectively). Given the similar divergence times, a likely explanation for these results would be the different taxon sampling and due to different models used for inferring diversification rates. While BAMM allows speciation and extinction rates to change through time, Madriñán et al. (2013) used a pure-birth model, which assumes constant rates and no extinction (Magallón & Sanderson, 2001). The same model has been used for Hughes & Eastwood (2006) for 43 evidence of radiation of Lupinus in the northern Andes and by Nürk et al. (2013a) for Hypericum.

Due to the presence of Berberis in South America well before the final uplift of the Andes, the lack of key innovations, and the lack of evidence for a Pleistocene radiation, we suggest that Berberis does not fit with the model of alpine radiation proposed by Hughes & Atchinson (2015) in that although the diversification of the genus might have been driven by ecological opportunities provided by new montane habitats, the genus diversified early on and less rapidly than genera such as Lupinus. Berberis is genus that probably originated in temperate regions of the Northern hemisphere, with the oldest known fossil belonging to the Paleocene found NE China (Li et al., 2010). We find most likely that this genus was pre-adapted to cold and temperate environments, and the Andes served as a “biotic corridor” rather than a “species pump” (Antonelli & Sanmartín, 2011) for Berberis.

The genus Berberis was proposed at the beginning of this work as an excellent model group to compare speciation rates between the tropics and temperate regions. We hypothesized according to taxonomic revision and recent phylogenetic studies (Adhikari et al., 2015) that tropical lineages would form a monophyletic group, probably as sister of the Temperate South American clade. Our study not only found tropical species (B. laurina) nested within the temperate South American clade, but we also demonstrated the non- monophyly of the tropical South American species. This means that the Septentrionales clade is not entirely composed of Northern temperate species, but includes possibly more than one tropical clade from South America. The question is now raised whether mountain uplift of the Himalayas has led to the increased diversification rates observed in this clade. Further studies should consider the heterogeneity in the phases of mountain uplift in different montane regions of the Himalaya (Favre et al., 2015).

Another possible explanation for lower speciation rates in the tropics may be the more intensive taxonomic studies and collections in temperate regions compared with the tropics. Ahrendt’s (1961) revision was published after his collections in Bhutan, Assam, southern Tibet, Upper Burma and (Adhikari et al. 2012). So far, there is no extensive work of Berberis comprising the whole diversity of the Andes, except for the work of Landrum (1999) on temperate species of the Andes. However, BAMM appears to be reliable for detecting rate shifts even under phylogenetic uncertainty (Rabosky, 2016), although the exact position of the shift may vary. 44

An important source of error could be inaccurate corrections for the incomplete taxon sampling of this study. Sample percentages were taken from Ahrendt (1961), but recent revisions (Landrum 1999) and regional checklists (Bernal et al., 2015) were also taken into account. Most of the times, the numbers do not match. This work tried to be as conservative as possible with the estimations, but considering the non-monophyly of the sections proposed by Ahrendt (1961) confirmed by all phylogenetic studies on Berberis (Kim et al. 2004a; Adhikari et al. 2015) including ours, the numbers may be inaccurate. The taxon sampling in this study is low (ca. 16%) and biased towards the northern hemisphere. In addition, we do not know the phylogenetic affinities of the basal species of Berberis (B. higginsae and B. nevinii) and the estimate number may be inaccurate as well. Past studies have shown the sensitivity of these models to incomplete taxon sampling (Spriggs et al., 2015). Although the analysis assuming complete taxon sampling gave similar results, in both cases the clade with more species is Septentrionales. Further studies should aim to increase the number of species sampled and an overall revision of South America should be considered.

In similar manner, further taxonomic and molecular research in Berberis should explore the possibility of cryptic species, as found in many other tropical groups in the tropics (Rabosky & Huang, 2015). Given the complex topography of the Andes, and the evidence for fragmentation of habitat driven by climatic changes in the Pleistocene, it may be possible that species found across wide ranges in the Andes can be distinct taxonomic units (Pennington et al., 2010; Gill et al., 2016). Recently systematic samplings across wide regions of the Andes have found several cryptic species (Pennington et al. 2010; Särkinen et al., 2012; Gagnon et al. 2015). We found evidence of this for at least one tropical species, B. rigida, as demonstrated by different accession showing sequence divergence equal to values observed between species, resulting in the species not appearing monophyletic in the molecular phylogeny (Figure 12). Further studies should include densely sampled phylogenies with multiple accessions per species in order to study the presence of cryptic species in the Andes.

Recently, a list of concerns has been raised about the results given by BAMM. Moore et al. (2016) claimed errors in the likelihood function of BAMM, suggesting that the model is only able to compute correct likelihoods when rates of speciation and extinction are constant. Moreover, critics about the extremely sensitivity of the model to the prior cast doubts about the number of rate shifts given by BAMM (Moore et al. 2016). The critics 45 have been revised by Rabosky in the past (Rabosky, 2014) and with more details in the BAMM project webpage (Rabosky, 2016). Indeed, it is suggested that studies using the method would test the data with different priors in order to test the robustness of the data. My analyses show that the Berberis data is robust to the prior. Results showed a consistent single shift at the base of the crown node of Berberis independent whether 1 or 100 shifts were used as the prior.

4.2 PHYLOGENETIC RELATIONSHIPS IN BERBERIS AND DIVERGENCE TIME

ESTIMATES

The major tree topologies for Berberis presented in this study are consistent with Adhikari et al. (2015). The simple-leaved Berberis remains as a monophyletic group sister to most compound-leaved Berberis. The compound-leaved B. higginsae and B. nevinii, appear as sister to all other species of Berberis. All newly sampled simple-leaved Chinese species fall within the Eurasian Septentrionales clade.

Neotropical Berberis are nested within the simple-leaved clade with ten species found in the tropical Andes forming a clade with strong support. Berberis ovalifolia, which is also found in tropical Andes has an unresolved position within this South American clade. The Southern Temperate clade has less overall support, but an important finding was the presence of B. laurina in this clade. Berberis laurina is one of the four species of Berberis found in the montane cloud Forest in Southern Brazil and Rio de la Plata region in Argentina, at elevations above 1,150 m (de Barcellos & Voltolini, 1995). These findings are not unexpected, given the well-known Pacific-Atlantic disjunctions (Moreira-Munoz, 2011). Similar disjunctions are found in many taxa, such as Drimys J.R Forst & G. Forst (Marquínez et al. 2009), Alstroemeria L. (Chacón et al. 2012), Araucaria Juss., Escallonia Mutis ex. L.f., Weinmannia L., Persea Mill. (Moreira-Munoz, 2011; Villagran & Hinojosa 1997; Landrum, 1981) and Myrceugenia O. Berg (Murillo et al. 2012) among others. The common pattern has been considered as evidence for a past continuous mixed subtropical forest, with components from the tropical and temperate biomes, extending across southern South America during the Oligocene-Miocene (Moreira-Munoz, 2011). These lineages then became fragmented during the increase in aridity in the Miocene, caused by the Andean uplift and the rain shadow produced in the oriental slope, forming what is known as the “dry diagonal” (Hinojosa & Villagran, 1997; Murillo et al. 2016). These groups 46 later migrated towards the Atlantic and Southern regions of Chile and Argentina, where similar climates occur. This would be the most likely explanation as well for the presence of Berberis in the Atlantic coast, without considering long-distance dispersal. A better understanding of the relationship between Andean and Southern Atlantic Berberis would be achieved including the remaining Southern Atlantic taxa to posterior studies.

The three remaining South American species of Berberis and their position within the Eurasian Septentrionales clade is an unexpected finding, considering past taxonomic work (Ahrendt, 1965; Landrum 1999; Adhikari, 2010) and recent phylogenetic studies (Kim et al, 2004a; Adhikari et al. 2015). Although we could only obtain ITS sequences for B. glauca and B. beauverdiana, B. quindiuensis show consistency of its position within a clade composed by species from Nepal and China for both ITS and ndhF regions and a close relationship to B. sargentiana, from China. The division between Septentrionales and Australes in Ahrendt’s work (1961) appears to be based mainly on geographic divisions (Landrum 1999). Some characters, such as red berries and deciduous are commonly found in Septentrionales, but are rare in Australes, while others, such as foliaceous spines, dentate and orange flowers are common in Australes but almost absent in Septentrionales (Ahrendt, 1961; Ladrum, 1999). No morphological characters appear to be synapomorphic for each clade. For example, most South American species are evergreen, but B. montana is deciduous like most species in Septentrionales. Similarly, the South American species falling within the Asian clade, such as B. quindiuensis, have racemose inflorescenes and large flowers similar to most Asian species in the subsect. Wallichianae where they fall in the molecular phylogeny (Ahrendt, 1961). The morphological characterization of these clades recovered in the phylogeny presented here should be a focus of a future study.

These findings open new insights about the biogeography of Berberis and its origin. The findings of past studies (Kim et al. 2004a) and the present study show the implausible colonization route of Berberis through North American species, due to the distant position of B. canadensis and B. fendleri to the Australes group (Figure 2). In the present study, we confirm the non-monophyly of the Neotropical group known as Australes by Ahrendt (1961), with some species forming a clade sister to the temperate and Southern Atlantic species, and other species belonging to the Eurasian Septentrionales group. In order to confirm the number of colonization events of Berberis to South America, it is necessary to 47 improve the resolution of Berberis phylogeny. A more extensive taxon sampling and more loci should be added to future studies.

The primary purpose of this study was to assess relative diversification rates and absolute dates were not necessary for this. Divergence times obtained in this work contrasts with the findings by Adhikari (2010) using a ndhF gene based phylogeny. Using the same calibration points, we obtained an estimate for the crown node of Berberis s.l. of 10.74 Ma (5.64-16.64), compared with 37.7 Ma (29.29-44.30). The age of split between Septentrionales and Australes is 4.39 Ma (2.66-6.24), compared with 29.1 Ma (19.89- 38.13). The origin of the South American clade for this study dates back to 3.61 Ma (2.21- 5.29), while for Adhikari (2010) is 19.1 Ma (8.79-29.50). These results give dates considerably more recent compared to Adhikari (2010).

Fossils of Berberis are well represented in Western North America Tertiary floras, with compound-leaved Berberis found in large quantities on the fossil record since the Middle Eocene (47.8 – 38 Ma) (Manchester, 1999) and Oligocene (Ramı́rez & Cevallos-Ferriz, 2000). The presence of Berberis in South America dates back to the Oligocene (Selles & Hinojosa, 1997), Late-Oligocene (Cesari et al., 2015) and early Neogene (Torres et al. 2013) in the Southern Andes and Patagonia. In the Central Andes, the oldest record is found in the Jakokkota flora from west-central Bolivia, with an estimate age of 10.66 Ma (10.72-10.60) (Gregory-Wodzicki et al. 1998). These dates are substantially older compared to the age estimates for the crown node of South American Berberis, or even for the crown node of Berberis itself.

The differences between the divergence times presented in Adhikari (2010) and this study are likely due to missing taxa, but following the recommendations by Zheng & Wiens (2015), a chronogram without considering taxa with missing data shows similar age estimates. The most likely explanation would be the different priors used for divergence time estimation. While the work by Adhikari (2010) used a Yule (pure-birth) process, which does not take into account extinction, the current study used the Birth-death incomplete taxon sampling process (Stadler, 2009) that considers both speciation and extinction. However, both Bayesian and PL methods give similar age estimates, suggesting a problem with the lack of calibration within Berberis. Many Berberis fossils cannot be assigned to any extant lineages occurring in the same area (Ramı́rez & Cevallos-Ferriz, 2000), and new findings of a simple-leaved Berberis fossil from the Paleocene (ca. 60 Ma) 48 in North East Asia (Li et al. 2010), conflicts with the younger ages of the Berberis stem node presented in this study. Although the present study might not present the final word on the absolute age estimates for Berberis, the chronogram presented here is still a useful framework for the purposes of studying relative differences and similarities in species diversification between clades within Berberis, and for comparing speciation rates between Berberis lineages. Such analyses do not rely on absolute but relative age estimates.

49

5. CONCLUSION

The LGD pattern has been suggested to be due to higher speciation in the tropics compared with temperate regions. Our results suggest that there has been a significant rate shift in speciation rates at the base of Berberis, followed by a constant increase in speciation rates. We found no evidence of higher speciation rates in the tropics compared with temperate regions. There is evidence for higher speciation rates in the Himalayan clade compared with the Andes, but no significant rate shifts related to adaptive radiation to mountain orogeny. In the case of Berberis, the LGD cannot be explained by greater diversification rates in the tropics compared to temperate regions. More studies of groups that span both regions need to be undertaken to confirm the generality of this pattern. The monophyly of the groups proposed by Ahrendt (1961) has been refuted, finding species from the Australes nested within the Northern Hemisphere Septentrionales. Further studies in Berberis should increase sampling of tropical species in order to improve the understanding of Berberis colonization in South America.

50

6. REFERENCES

Aberer, A.J., Krompass, D. and Stamatakis, A., 2013. Pruning rogue taxa improves phylogenetic accuracy: an efficient algorithm and webservice. Systematic biology, 62(1), pp.162-166.

Adhikari, B., 2010. Systematics and phylogeographic studies of Berberis L.(Berberidaceae) in the Nepal Himalaya (Doctoral dissertation, University of Edinburgh).

Adhikari, B., Milne, R., Pennington, R.T., Särkinen, T. and Pendry, C.A., 2015. Systematics and biogeography of Berberis sl inferred from nuclear ITS and chloroplast ndhF gene sequences. Taxon, 64(1), pp.39-48.

Adhikari, B., Pendry, C.A., Pennington, R.T. and Milne, R.I., 2012. A revision of Berberis ss (Berberidaceae) in Nepal. Edinburgh Journal of Botany, 69(3), p.447.

Ahmad, I., Day, J.P., MacDonald, M.V. and Ingram, D.S., 1991. Haploid culture and UV mutagenesis in rapid-cycling Brassica napus for the generation of resistance to chlorsulfuron and Alternaria brassicicola. Annals of Botany, 67(6), pp.521-519. Ahrendt, L.W.A., 1961. Berberis and Mahonia. Journal of the Linnean Society of London, Botany, 57(369), pp.1-410. Aitchison, J.C., Ali, J.R. and Davis, A.M., 2007. When and where did India and Asia collide?. Journal of Geophysical Research: Solid Earth, 112(B5). Alfaro, M.E., Santini, F., Brock, C., Alamillo, H., Dornburg, A., Rabosky, D.L., Carnevale, G. and Harmon, L.J., 2009. Nine exceptional radiations plus high turnover explain species diversity in jawed vertebrates. Proceedings of the National Academy of Sciences, 106(32), pp.13410-13414. Andreasen, K. and Baldwin, B.G., 2001. Unequal evolutionary rates between annual and perennial lineages of checker mallows (Sidalcea, Malvaceae): evidence from 18S–26S rDNA internal and external transcribed spacers. Molecular Biology and Evolution, 18(6), pp.936-944.

Antonelli, A. and Sanmartín, I., 2011. Why are there so many plant species in the Neotropics?. Taxon, 60(2), pp.403-414.

Arita, H.T. and Vázquez‐Domínguez, E., 2008. The tropics: cradle, museum or casino? A dynamic null model for latitudinal gradients of species diversity. Ecology Letters, 11(7), pp.653-663. Bates, J.M. and Zink, R.M., 1994. Evolution into the Andes: molecular evidence for species relationships in the genus Leptopogon. The Auk, pp.507-515.

Bernal, R., S.R. Gradstein & M. Celis (eds.). 2015. Catálogo de plantas y líquenes de Colombia. Instituto de Ciencias Naturales, Universidad Nacional de Colombia, Bogotá. [www document] < http://catalogoplantasdecolombia.unal.edu.co > (accessed 18th August 2016) 51

Bradford, J.C., 2002. Molecular phylogenetics and morphological evolution in Cunonieae (Cunoniaceae). Annals of the Missouri Botanical Garden, pp.491-503. Bremer, B., Bremer, K., Chase, M., Fay, M., Reveal, J., Soltis, D., Soltis, P. and Stevens, P., 2009. An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG III. Botanical Journal of the Linnean Society. Cardinal-McTeague, W.M., Sytsma, K.J. and Hall, J.C., 2016. Biogeography and diversification of Brassicales: A 103million year tale. Molecular phylogenetics and evolution, 99, pp.204-224. Césari, S.N., Panti, C., Pujana, R.R., Francis, J.E. and Marenssi, S.A., 2015. The late Oligocene flora from the Río Leona Formation, Argentinian Patagonia. Review of Palaeobotany and Palynology, 216, pp.143-158. Chacón, J., de Assis, M.C., Meerow, A.W. and Renner, S.S., 2012. From east Gondwana to Central America: historical biogeography of the Alstroemeriaceae. Journal of Biogeography, 39(10), pp.1806-1818. Chown, S.L. and Gaston, K.J., 2000. Areas, cradles and museums: the latitudinal gradient in species richness. Trends in Ecology & Evolution, 15(8), pp.311-315.

Colwell, R.K., Rahbek, C. and Gotelli, N.J., 2004. The Mid‐Domain Effect and Species Richness Patterns: What Have We Learned So Far?. The American Naturalist, 163(3), pp.E1-E23. Currie, D.J., Mittelbach, G.G., Cornell, H.V., Field, R., Guégan, J.F., Hawkins, B.A., Kaufman, D.M., Kerr, J.T., Oberdorff, T., O'Brien, E. and Turner, J.R.G., 2004. Predictions and tests of climate‐based hypotheses of broad‐scale variation in taxonomic richness. Ecology letters, 7(12), pp.1121-1134.

Davies, T.J., Savolainen, V., Chase, M.W., Moat, J. and Barraclough, T.G., 2004. Environmental energy and evolutionary rates in flowering plants. Proceedings of the Royal Society of London B: Biological Sciences, 271(1553), pp.2195-2200. de Barcellos Falkenberg, D. and Voltolini, J.C., 1995. The montane cloud forest in southern Brazil. In Tropical Montane cloud forests (pp. 138-149). Springer US. dos Reis, M., Donoghue, P.C. and Yang, Z., 2016. Bayesian molecular clock dating of species divergences in the genomics era. Nature Reviews Genetics, 17(2), pp.71-80. Dowle, E.J., Morgan-Richards, M. and Trewick, S.A., 2013. Molecular evolution and the latitudinal biodiversity gradient. Heredity, 110(6), pp.501-510.

Drummond, A.J., Ho, S.Y., Phillips, M.J. and Rambaut, A., 2006. Relaxed phylogenetics and dating with confidence. PLoS Biol, 4(5), p.e88.

Drummond, C.S., Eastwood, R.J., Miotto, S.T. and Hughes, C.E., 2012. Multiple continental radiations and correlates of diversification in Lupinus (Leguminosae): testing for key innovation with incomplete taxon sampling. Systematic Biology, 61(3), pp.443- 460.

52

Favre, A., Päckert, M., Pauls, S.U., Jähnig, S.C., Uhl, D., Michalak, I. and Muellner‐ Riehl, A.N., 2015. The role of the uplift of the Qinghai‐Tibetan Plateau for the evolution of Tibetan biotas. Biological Reviews, 90(1), pp.236-253.

Fischer, A.G., 1960. Latitudinal variations in organic diversity. Evolution, 14(1), pp.64- 81.

Flenley, J.R., 2011. Why is pollen yellow? And why are there so many species in the tropical rain forest?. Journal of Biogeography, 38(5), pp.809-816.

Forest, F., 2009. Calibrating the Tree of Life: fossils, molecules and evolutionary timescales. Annals of Botany, p. mcp192.

Gagnon, E., Hughes, C.E., Anne, G.P. and Bruneau, A., 2015. A new cryptic species in a new cryptic genus in the Caesalpinia group (Leguminosae) from the seasonally dry inter- Andean valleys of South America. Taxon, 64(3), pp.468-490. Gaston, K.J., 2000. Global patterns in biodiversity. Nature, 405(6783), pp.220-227. Gentry, A.H., 1982. Neotropical floristic diversity: phytogeographical connections between Central and South America, Pleistocene climatic fluctuations, or an accident of the Andean orogeny?. Annals of the Missouri Botanical Garden, 69(3), pp.557-593. Gill, B.A., Kondratieff, B.C., Casner, K.L., Encalada, A.C., Flecker, A.S., Gannon, D.G., Ghalambor, C.K., Guayasamin, J.M., Poff, N.L., Simmons, M.P. and Thomas, S.A., 2016, June. Cryptic species diversity reveals biogeographic support for the ‘mountain passes are higher in the tropics’ hypothesis. In Proc. R. Soc. B (Vol. 283, No. 1832, p. 20160553). The Royal Society. Graur, D. and Martin, W., 2004. Reading the entrails of chickens: molecular timescales of evolution and the illusion of precision. TRENDS in Genetics, 20(2), pp.80-86. Gregory-Wodzicki, K.M., McIntosh, W.C. and Velasquez, K., 1998. Climatic and tectonic implications of the late Miocene Jakokkota flora, Bolivian Altiplano. Journal of South American Earth Sciences, 11(6), pp.533-560. Guerrero, P.C., Durán, A.P. and Walter, H.E., 2011. Latitudinal and altitudinal patterns of the endemic cacti from the Atacama Desert to Mediterranean Chile. Journal of Arid Environments, 75(11), pp.991-997.

Hall, T.A., 1999, January. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. In Nucleic acids symposium series (Vol. 41, pp. 95-98).

Harber, J., 2016. Three new species of Berberis from Yunnan. Curtis's Botanical Magazine, 33(1), pp.24-24. Harrison, T.M. and Copeland, P., 1992. Raising tibet. Science, 255(5052), p.1663. Hawkins, B.A. and Porter, E.E., 2001. Area and the latitudinal diversity gradient for terrestrial birds. Ecology letters, 4(6), pp.595-601. 53

Hawkins, B.A., Rodríguez, M.Á. and Weller, S.G., 2011. Global angiosperm family richness revisited: linking ecology and evolution to climate. Journal of Biogeography, 38(7), pp.1253-1266.

Hillebrand, H., 2004. On the generality of the latitudinal diversity gradient. The American Naturalist, 163(2), pp.192-211.

Hinojosa, L.F. and Villagrán, C., 1997. Historia de los bosques del sur de Sudamérica, I: antecedentes paleobotánicos, geológicos y climáticos del Terciario del cono sur de América. Revista Chilena de Historia Natural, 70(2), pp.225-240.

Ho, S.Y. and Duchêne, S., 2014. Molecular‐clock methods for estimating evolutionary rates and timescales. Molecular ecology, 23(24), pp.5947-5965. Hoorn, C., Wesselingh, F.P., Ter Steege, H., Bermudez, M.A., Mora, A., Sevink, J., Sanmartín, I., Sanchez-Meseguer, A., Anderson, C.L., Figueiredo, J.P. and Jaramillo, C., 2010. Amazonia through time: Andean uplift, climate change, landscape evolution, and biodiversity. science, 330(6006), pp.927-931.

Huey, R.B., 1978. Latitudinal pattern of between-altitude faunal similarity: mountains might be" higher" in the tropics. The American Naturalist, 112(983), pp.225-229. Hughes, C. and Eastwood, R., 2006. Island radiation on a continental scale: exceptional rates of plant diversification after uplift of the Andes. Proceedings of the National Academy of Sciences, 103(27), pp.10334-10339. Hughes, C.E. and Atchison, G.W., 2015. The ubiquity of alpine plant radiations: from the Andes to the Hengduan Mountains. New Phytologist, 207(2), pp.275-282. Hughes, C.E., Pennington, R.T. and Antonelli, A., 2013. Neotropical plant evolution: assembling the big picture. Botanical Journal of the Linnean Society, 171(1), pp.1-18. Jablonski, D., Roy, K. and Valentine, J.W., 2006. Out of the tropics: evolutionary dynamics of the latitudinal diversity gradient. Science, 314(5796), pp.102-106. Janzen, D.H., 1967. Why mountain passes are higher in the tropics. The American Naturalist, 101(919), pp.233-249. Jetz, W. and Rahbek, C., 2001. Geometric constraints explain much of the species richness pattern in African birds. Proceedings of the National Academy of Sciences, 98(10), pp.5661-5666.

Jiang, W., Chen, S.Y., Wang, H., Li, D.Z. and Wiens, J.J., 2014. Should genes with missing data be excluded from phylogenetic analyses?. Molecular phylogenetics and evolution, 80, pp.308-318.

Joppa, L.N., Roberts, D.L., Myers, N. and Pimm, S.L., 2011. Biodiversity hotspots house most undiscovered plant species. Proceedings of the National Academy of Sciences, 108(32), pp.13171-13176.

Kim, Y.D. and Jansen, R.K., 1998. Chloroplast DNA restriction site variation and phylogeny of the Berberidaceae. American journal of Botany, 85(12), pp.1766-1778. 54

Kim, Y.D., Kim, S.H. and Landrum, L.R., 2004a. Taxonomic and phytogeographic implications from ITS phylogeny in Berberis (Berberidaceae). Journal of plant research, 117(3), pp.175-182. Kim, Y.D., Kim, S.H., Kim, C.H. and Jansen, R.K., 2004b. Phylogeny of Berberidaceae based on sequences of the chloroplast gene ndhF. Biochemical systematics and ecology, 32(3), pp.291-301. Koch, M. and Al-Shehbaz, I.A., 2002. Molecular data indicate complex intra-and intercontinental differentiation of American Draba (Brassicaceae). Annals of the Missouri Botanical Garden, pp.88-109. Kozak, K.H. and Wiens, J.J., 2007. Climatic zonation drives latitudinal variation in speciation mechanisms. Proceedings of the Royal Society of London B: Biological Sciences, 274(1628), pp.2995-3003. Lambers, J.H.R., Clark, J.S. and Beckage, B., 2002. Density-dependent mortality and the latitudinal gradient in species diversity. Nature, 417(6890), pp.732-735.

Landrum, L.R., 1981. The phylogeny and geography of Myrceugenia (Myrtaceae). Brittonia, 33(1), pp.105-129. Landrum, L.R., 1999. Revision of Berberis (Berberidaceae) in Chile and adjacent southern Argentina. Annals of the Missouri Botanical Garden, pp.793-834. Li, Y.L., Kvaček, Z., Ferguson, D.K., Wang, Y.F., Li, C.S., Yang, J., Ying, T.S., Ablaev, A.G. and Liu, H.M., 2010. The fossil record of Berberis (Berberidaceae) from the Palaeocene of NE China and interpretations of the evolution and phytogeography of the genus. Review of Palaeobotany and Palynology, 160(1), pp.10-31. Luebert, F. and Muller, L.A., 2015. Effects of mountain formation and uplift on biological diversity. Frontiers in genetics, 6. Luebert, F. and Weigend, M., 2014. Phylogenetic insights into Andean plant diversification. Frontiers in Ecology and Evolution, 2, p.27. Macpherson, E., 2002. Large–scale species–richness gradients in the Atlantic Ocean. Proceedings of the Royal Society of London B: Biological Sciences, 269(1501), pp.1715- 1720. Maddison, W. P. and D.R. Maddison. 2016. Mesquite: a modular system for evolutionary analysis. Version 3.10 [www document] (Accessed on 20 of June of 2016 ) Madriñán, S., Cortés, A.J. and Richardson, J.E., 2013. Páramo is the world's fastest evolving and coolest biodiversity hotspot. Frontiers in genetics, 4, p.192. Magallón, S.A., 2004. Dating lineages: molecular and paleontological approaches to the temporal framework of clades. International Journal of Plant Sciences, 165(S4), pp.S7- S21. Magallón, S. and Sanderson, M.J., 2001. Absolute diversification rates in angiosperm clades. Evolution, 55(9), pp.1762-1780. 55

Manchester, S.R., 1999. Biogeographical relationships of North American tertiary floras. Annals of the Missouri Botanical Garden, pp.472-522. Marquínez, X., Lohmann, L.G., Salatino, M.L.F., Salatino, A. and González, F., 2009. Generic relationships and dating of lineages in Winteraceae based on nuclear (ITS) and plastid (rpS16 and psbA-trnH) sequence data. Molecular Phylogenetics and Evolution, 53(2), pp.435-449. Martin, A.P. and Palumbi, S.R., 1993. Body size, metabolic rate, generation time, and the molecular clock. Proceedings of the National Academy of Sciences, 90(9), pp.4087- 4091. Mateo, R.G., Broennimann, O., Normand, S., Petitpierre, B., Araújo, M.B., Svenning, J.C., Baselga, A., Fernández-González, F., Gómez-Rubio, V., Muñoz, J. and Suarez, G.M., 2016. The mossy north: an inverse latitudinal diversity gradient in European bryophytes. Scientific reports, 6. McCain, C.M., 2009. Vertebrate range sizes indicate that mountains may be ‘higher’in the tropics. Ecology letters, 12(6), pp.550-560. Miehe, G., Gravendeel, B, Kluge, J., Kuss, P., Long D., Opgenoorth, L., Pendry, C., Rajbhandari, K., Rajbhandary, S., Ree, R., Schmidt, J., Søchting, U., Subedi, A., Watson, M., & Welk, E. 2015. Chapter 8. Flora. In Miehe, G., C.A. Pendry & Chaudhary, R. Nepal: An Introduction to the natural history, ecology and human environment of the Himalayas: 135-202. Edinburgh. Royal Botanic Garden Edinburgh. Miller, M.A., Pfeiffer, W. and Schwartz, T., 2010, November. Creating the CIPRES Science Gateway for inference of large phylogenetic trees. In Gateway Computing Environments Workshop (GCE), 2010 (pp. 1-8). IEEE. Mittelbach, G.G., Schemske, D.W., Cornell, H.V., Allen, A.P., Brown, J.M., Bush, M.B., Harrison, S.P., Hurlbert, A.H., Knowlton, N., Lessios, H.A. and McCain, C.M., 2007. Evolution and the latitudinal diversity gradient: speciation, extinction and biogeography. Ecology letters, 10(4), pp.315-331. Moller, M. and Cronk, Q., 1997. Origin and relationships of Saintpaulia (Gesneriaceae) based on ribosomal DNA internal transcribed spacer (ITS) sequences. American Journal of Botany, 84(7), pp.956-956. Moore, B.R., Höhna, S., May, M.R., Rannala, B. and Huelsenbeck, J.P., 2016. Critically evaluating the theory and performance of Bayesian analysis of macroevolutionary mixtures. Proceedings of the National Academy of Sciences, p.201518659. Moreira-Muñoz, A., 2011. Plant geography of Chile (Vol. 5). Springer Science & Business Media. Morlon, H., 2014. Phylogenetic approaches for studying diversification. Ecology letters, 17(4), pp.508-525. Morlon, H., Parsons, T.L. and Plotkin, J.B., 2011. Reconciling molecular phylogenies with the fossil record. Proceedings of the National Academy of Sciences, 108(39), pp.16327-16332. 56

Murillo-A, J., Ruiz-P, E., Landrum, L.R., Stuessy, T.F. and Barfuss, M.H., 2012. Phylogenetic relationships in Myrceugenia (Myrtaceae) based on plastid and nuclear DNA sequences. Molecular phylogenetics and evolution, 62(2), pp.764-776.

Murillo‐A, J.C., Stuessy, T.F. and Ruiz, E., 2016. Explaining disjunct distributions in the flora of southern South America: evolutionary history and biogeography of Myrceugenia (Myrtaceae). Journal of Biogeography. Myers, N., Mittermeier, R.A., Mittermeier, C.G., Da Fonseca, G.A. and Kent, J., 2000. Biodiversity hotspots for conservation priorities. Nature, 403(6772), pp.853-858. Nürk, N.M., Madriñán, S., Carine, M.A., Chase, M.W. and Blattner, F.R., 2013. Molecular phylogenetics and morphological evolution of St. John’s wort (Hypericum; Hypericaceae). Molecular phylogenetics and evolution, 66(1), pp.1-16. Nürk, N.M., Scheriau, C. and Madriñán, S., 2013. Explosive radiation in high Andean Hypericum—rates of diversification among New World lineages. Frontiers in genetics, 4, p.175. Nylander, J.A.A., 2004. MrModeltest v2. Program distributed by the author. Evolutionary Biology Centre, Uppsala University, 2. Ohta, T., 2002. Near-neutrality in evolution of genes and gene regulation. Proceedings of the National Academy of Sciences, 99(25), pp.16134-16137.

Ollerton, J. and Cranmer, L., 2002. Latitudinal trends in plant‐pollinator interactions: are tropical plants more specialised?. Oikos, 98(2), pp.340-350.

Pagel, M., Venditti, C. and Meade, A., 2006. Large punctuational contribution of speciation to evolutionary divergence at the molecular level. Science, 314(5796), pp.119- 121. Paradis, E., Claude, J. and Strimmer, K., 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics, 20(2), pp.289-290. Pennington, R.T. and Dick, C.W., 2010. Diversification of the Amazonian flora and its relation to key geological and environmental events: a molecular perspective. Amazonia, landscape and species evolution, 1st ed. Oxford: Wiley-Blackwell, pp.373-385. Pianka, E.R., 1966. Latitudinal gradients in species diversity: a review of concepts. American Naturalist, pp.33-46. Pliscoff, P. and Luebert, F., 2006. Sinopsis bioclimática y vegetacional de Chile. Santiago de Chile: Editorial Universitaria. Plummer, M., Best, N., Cowles, K. and Vines, K., 2006. CODA: Convergence diagnosis and output analysis for MCMC. R news, 6(1), pp.7-11. Qian, H. and Ricklefs, R.E., 2004. Taxon richness and climate in angiosperms: is there a globally consistent relationship that precludes region effects?. The American Naturalist, 163(5), pp.773-779. Quental, T.B. and Marshall, C.R., 2010. Diversity dynamics: molecular phylogenies need the fossil record. Trends in Ecology & Evolution, 25(8), pp.434-441. 57

R Development Core Team 2008. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, [www document] < http://www.R-project.org.> (accessed 18th of August 2016)

Rabosky, D.L., 2010. Extinction rates should not be estimated from molecular phylogenies. Evolution, 64(6), pp.1816-1824. Rabosky, D.L., 2014. Automatic detection of key innovations, rate shifts, and diversity- dependence on phylogenetic trees. PloS one, 9(2), p.e89543. Rabosky, D.L. 2016. Bayesian Analysis of Macroevolutionary Mixtures [www document] < http://www.bamm-project.org > (accessed 18th of August 2016) Rabosky, D.L., Donnellan, S.C., Grundler, M. and Lovette, I.J., 2014a. Analysis and visualization of complex macroevolutionary dynamics: an example from Australian scincid lizards. Systematic biology, 63(4), pp.610-627. Rabosky, D.L., Grundler, M., Anderson, C., Shi, J.J., Brown, J.W., Huang, H. and Larson, J.G., 2014b. BAMMtools: an R package for the analysis of evolutionary dynamics on phylogenetic trees. Methods in Ecology and Evolution, 5(7), pp.701-707. Rabosky, D.L. and Huang, H., 2015, June. Minimal effects of latitude on present-day speciation rates in New World birds. In Proc. R. Soc. B (Vol. 282, No. 1809, p. 20142889). The Royal Society. Rahbek, C., Gotelli, N.J., Colwell, R.K., Entsminger, G.L., Rangel, T.F.L. and Graves, G.R., 2007. Predicting continental-scale patterns of bird species richness with spatially explicit models. Proceedings of the Royal Society of London B: Biological Sciences, 274(1607), pp.165-174.

Rajbhandari, K.R., Rai, S.K. and Bhatt, G.D., 2016. Endemic Flowering Plants of Nepal: An update. Plant Resources, p.106. Rambaut, A. and Drummond, A.J., 2013. TreeAnnotator v1. 7.0. University of Edinburgh, Institute of Evolutionary Biology. Rambaut, A., Suchard, M., Xie, W. and Drummond, A., 2014. Tracer v. 1.6. Institute of Evolutionary Biology, University of Edinburgh.

Ramı́rez, J.L. and Cevallos-Ferriz, S.R., 2000. Leaves of Berberidaceae (Berberis and Mahonia) from Oligocene sediments, near Tepexi de Rodrı́guez, Puebla. Review of Palaeobotany and Palynology, 110(3), pp.247-257. Ramos, V.A. and Naipauer, M., 2014. Patagonia: where does it come from?. Journal of Iberian Geology, 40(2), p.367. Richardson, J.E., Pennington, R.T., Pennington, T.D. and Hollingsworth, P.M., 2001. Rapid diversification of a species-rich genus of neotropical rain forest trees. Science, 293(5538), pp.2242-2245.

Ricklefs, R.E. and He, F., 2016. Region effects influence local tree species diversity. Proceedings of the National Academy of Sciences, 113(3), pp.674-679. 58

Rohde, K., 1992. Latitudinal gradients in species diversity: the search for the primary cause. Oikos, pp.514-527. Rolland, J., Condamine, F.L., Jiguet, F. and Morlon, H., 2014. Faster speciation and reduced extinction in the tropics contribute to the mammalian latitudinal diversity gradient. PLoS Biol, 12(1), p.e1001775. Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D.L., Darling, A., Höhna, S., Larget, B., Liu, L., Suchard, M.A. and Huelsenbeck, J.P., 2012. MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space. Systematic biology, 61(3), pp.539-542. Rosenzweig, M.L., 1995. Species diversity in space and time. Cambridge University Press. Ruggiero, A. and Hawkins, B.A., 2008. Why do mountains support so many species of birds?. Ecography, 31(3), pp.306-315. Sanderson, M.J., 2002. Estimating absolute rates of molecular evolution and divergence times: a penalized likelihood approach. Molecular biology and evolution, 19(1), pp.101- 109. Särkinen, T., Pennington, R.T., Lavin, M., Simon, M.F. and Hughes, C.E., 2012. Evolutionary islands in the Andes: persistence and isolation explain high endemism in Andean dry tropical forests. Journal of Biogeography, 39(5), pp.884-900. Sánchez-Baracaldo, P. and Thomas, G.H., 2014. Adaptation and convergent evolution within the Jamesonia-Eriosorus complex in high-elevation biodiverse Andean hotspots. PloS one, 9(10), p.e110618. Sauquet, H., 2013. A practical guide to molecular dating. Comptes Rendus Palevol, 12(6), pp.355-367. Sauquet, H., Ho, S.Y., Gandolfo, M.A., Jordan, G.J., Wilf, P., Cantrill, D.J., Bayly, M.J., Bromham, L., Brown, G.K., Carpenter, R.J. and Lee, D.M., 2012. Testing the impact of calibration on molecular divergence times using a fossil-rich group: the case of Nothofagus (Fagales). Systematic Biology, 61(2), pp.289-313. Schenk, J.J., 2016. Consequences of secondary calibrations on divergence time estimates. PloS one, 11(1), p.e0148228.

Segovia, R.A. and Armesto, J.J., 2015. The Gondwanan legacy in South American biogeography. Journal of Biogeography, 42(2), pp.209-217.

Selles, D. and Hinojosa, L.F., 1997. Niveles sedimentarios y paleofloras del Oligoceno superior-Mioceno inferior en la Formación Abanico, noreste de Santiago. In Congreso Geológico Chileno (Vol. 8, pp. 580-584). Shi, J.J. and Rabosky, D.L., 2015. Speciation dynamics during the global radiation of extant bats. Evolution, 69(6), pp.1528-1545. Smith, S.A. and Donoghue, M.J., 2008. Rates of molecular evolution are linked to life history in flowering plants. science, 322(5898), pp.86-89. 59

Smith, S.A. and O’Meara, B.C., 2012. treePL: divergence time estimation using penalized likelihood for large phylogenies. Bioinformatics, 28(20), pp.2689-2690. Spriggs, E.L., Clement, W.L., Sweeney, P.W., Madriñán, S., Edwards, E.J. and Donoghue, M.J., 2015. Temperate radiations and dying embers of a tropical past: the diversification of Viburnum. New Phytologist, 207(2), pp.340-354. Stadler, T., 2009. On incomplete sampling under birth–death models and connections to the sampling-based coalescent. Journal of Theoretical Biology, 261(1), pp.58-66. Stamatakis, A., 2014. RAxML version 8: a tool for phylogenetic analysis and post- analysis of large phylogenies. Bioinformatics, 30(9), pp.1312-1313.

Stebbins, G.L., 1974. Flowering plants: evolution above the species level. London: Arnold xviii, 399p. Illustrations. General (KR, 197500089). Stehli, F.G., Douglas, R.G. and Newell, N.D., 1969. Generation and maintenance of gradients in taxonomic diversity. Science, 164(3882), pp.947-949.

Sullivan, J.H., Teramura, A.H. and Ziska, L.H., 1992. Variation in UV-B sensitivity in plants from a 3,000-m elevational gradient in Hawaii. American Journal of Botany, pp.737-743. Terborgh, J., 1973. On the notion of favorableness in plant ecology. American Naturalist, pp.481-501. Thomas, D.C., Hughes, M., Phutthai, T., Ardi, W.H., Rajbhandary, S., Rubite, R., Twyford, A.D. and Richardson, J.E., 2012. West to east dispersal and subsequent rapid diversification of the mega‐diverse genus Begonia (Begoniaceae) in the Malesian archipelago. Journal of Biogeography, 39(1), pp.98-113. Thompson, J.D., Higgins, D.G. and Gibson, T.J., 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic acids research, 22(22), pp.4673-4680. Torres, T., Gutiérrez, N.M., Bostelmann, E., Le Roux, J.P., Oyarzún, J.L., Ugalde, R., Otero, R. and Hervé, F., 2013. Exceptionally preserved fossil flora of the Río Leona Formation in Sierra Baguales, Magallanes, Chile: insights into the early Neogene Patagonian ecosystems. Bollettino di Geofisica teorica ed applicata (Supplement B), 54, pp.352-355. Van Hinsbergen, D.J., Lippert, P.C., Dupont-Nivet, G., McQuarrie, N., Doubrovine, P.V., Spakman, W. and Torsvik, T.H., 2012. Greater India Basin hypothesis and a two- stage Cenozoic collision between India and Asia. Proceedings of the National Academy of Sciences, 109(20), pp.7659-7664. Villagrán, C. and Hinojosa, L.F., 1997. Historia de los bosques del sur de Sudamérica, II: Análisis fitogeográfico. Revista Chilena de Historia Natural, 70(2), pp.1-267. Wang, W., Chen, Z.D., Liu, Y., Li, R.Q. and Li, J.H., 2007. Phylogenetic and biogeographic diversification of Berberidaceae in the northern hemisphere. Systematic Botany, 32(4), pp.731-742. 60

Wang, W., Lu, A.M., Ren, Y., Endress, M.E. and Chen, Z.D., 2009. Phylogeny and classification of Ranunculales: evidence from four molecular loci and morphological data. Perspectives in Plant Ecology, Evolution and Systematics, 11(2), pp.81-110. Wen, J., Zhang, J., Nie, Z.L., Zhong, Y. and Sun, H., 2014. Evolutionary diversifications of plants on the Qinghai-Tibetan Plateau. Frontiers in genetics, 5, p.4. Wiens, J.J. and Donoghue, M.J., 2004. Historical biogeography, ecology and species richness. Trends in ecology & evolution, 19(12), pp.639-644.

Willig, M.R., Kaufman, D.M. and Stevens, R.D., 2003. Latitudinal gradients of biodiversity: pattern, process, scale, and synthesis. Annual Review of Ecology, Evolution, and Systematics, pp.273-309.

Willis, K.J., Bennett, K.D. and Birks, H.J.B., 2009. Variability in thermal and UV‐B energy fluxes through time and their influence on plant diversity and speciation. Journal of Biogeography, 36(9), pp.1630-1644. Xie, H., Ash, J.E., Linde, C.C., Cunningham, S. and Nicotra, A., 2014. Himalayan- Tibetan plateau uplift drives divergence of polyploid poppies: Meconopsis viguier (Papaveraceae). PloS one, 9(6), p.e99177. Yan, L.J., Liu, J., Möller, M., Zhang, L., Zhang, X.M., Li, D.Z. and Gao, L.M., 2015. DNA barcoding of Rhododendron (Ericaceae), the largest Chinese plant genus in biodiversity hotspots of the Himalaya–Hengduan Mountains. Molecular ecology resources, 15(4), pp.932-944. Zhang, M.L., Kang, Y., Zhong, Y. and Sanderson, S.C., 2012. Intense uplift of the Qinghai-Tibetan Plateau triggered rapid diversification of Phyllolobium (Leguminosae) in the Late Cenozoic. Plant Ecology & Diversity, 5(4), pp.491-499. Zheng, Y. and Wiens, J.J., 2015. Do missing data influence the accuracy of divergence- time estimation with BEAST? Molecular phylogenetics and evolution, 85, pp.41-49. Zuckerkandl, E. and Pauling, L., 1965. Evolutionary divergence and convergence in proteins. Evolving genes and proteins, 97, pp.97-166.

61

7. APPENDICES

7.1 APPENDIX 1 – ACCESION DETAILS

TABLE 8: LIST OF ACCESIONS USED IN THIS STUDY. ABBREVIATIONS: NL = NICOLAS LAVANDERO; BA = BHASKAR ADHIKARI (ADHIKARI ET AL. 2015)

Taxon name EDNA no Origin Extracted by Berberis mekongensis W. W. Smith EDNA16-0045022 China NL Berberis pseudotibetica C. Y. Wu ex S. Y. Bao EDNA16-0045023 China NL Berberis brevipedicillata T. S. Ying EDNA16-0045024 China NL Berberis calcipratorum Ahrendt EDNA16-0045025 China NL Berberis grodtmanniana C. K. Schneid. EDNA16-0045026 China NL Berberis hersii Ahrendt EDNA16-0045027 China NL Hook. EDNA16-0045028 Chile/Argentina NL Berberis gagnepainii C.K.Schneid. EDNA16-0045029 China NL Berberis lecomtei C.K.Schneid. EDNA16-0045230 China NL Berberis dictophylla EDNA16-0045231 China NL Berberis julianae Schneid. EDNA16-0045232 China NL Berberis nummularia Bunge EDNA16-0045233 China NL Berberis phanera C.K.Schneid EDNA16-0045234 China NL Berberis cf. petriruizii EDNA16-0045422 Colombia NL Berberis beauverdiana C.K.Schneid. EDNA16-0045455 Peru NL Berberis grandiflora Turcz. EDNA16-0045456 Ecuador NL Berberis hispanica Boiss. & Reut. EDNA16-0045458 Morocco NL Berberis sp. EDNA16-0045460 Ecuador NL Berberis sp. EDNA16-0045461 Ecuador NL Berberis sp. EDNA16-0045462 Ecuador NL Berberis glauca Kunth EDNA16-0045495 Colombia NL Berberis lutea Ruiz & Pav. EDNA16-0045469 Ecuador NL Berberis rigida Hieron. EDNA16-0045474 Ecuador NL Berberis rigida Hieron. EDNA16-0045475 Ecuador NL Berberis quindiuensis Kunth ex DC. EDNA16-0045483 Colombia NL Berberis ovalifolia Rusby EDNA16-0045477 Bolivia NL Berberis morana L.A. Camargo EDNA16-0045478 Colombia NL Berberis goudotii Triana & Planch. ex Wedd. EDNA16-0045485 Colombia NL Berberis laurina Billb. ex Thunb. EDNA16-0045559 Brazil NL Berberis angulosa var. angulosa Wall. ex EDNA09-00526 Nepal: Rasuwa BA Hook.f. & Thoms. Berberis angulosa var. angulosa Wall. ex EDNA09-00563 Nepal: Rasuwa BA Hook.f. & Thoms. Berberis angulosa Wall. ex Hook.f. & Thoms. EDNA09-00564 Nepal: Rasuwa BA var fasciculata Ahrendt 62

Berberis kumaonensis C.K.Schneid. EDNA08-01827 Nepal: Humla BA Berberis concinna Hook.f. var. concinna EDNA08-01858 Nepal: Rasuwa BA Berberis everestiana EDNA09-00549 Nepal: Mustang BA Berberis mucrifolia Ahrendt EDNA07-02122 Nepal: Mustang BA Berberis jaeschkeana C.K.Schneid. var. EDNA09-00555 Nepal: Mustang BA usteriana C.K.Schneid. Berberis pendryi Bh.Adhikari EDNA09-00545 Nepal: Mustang BA Berberis karnaliensis Bh.Adhikari EDNA08-01835 Nepal: Jumla BA Berberis sibirica Pallas EDNA09-00674 China: N of Beijing BA Berberis orthobotrys Bienert ex Aitch. var. EDNA09-00638 Iran BA orthobotrys Berberis orthobotrys Bienert ex Aitch. var. EDNA09-00527 Nepal: Rasuwa BA rubicunda Ahrendt Berberis asiatica Roxb. ex DC. EDNA09-00570 Nepal: Rasuwa BA Berberis asiatica Roxb. ex DC. EDNA08-01826 Nepal: Mugu BA Berberis koehneana C.K.Schneid. EDNA09-00531 Nepal: Rasuwa BA Berberis koehneana C.K.Schneid. EDNA07-02236 Nepal: Rasuwa BA Berberis aristata DC. EDNA09-00569 Nepal: Rasuwa BA Berberis petiolaris Wall.ex G.Don EDNA09-00595 Nepal: Mugu BA Berberis thomsoniana C.K.Schneid. EDNA08-01856 Nepal: Rasuwa BA Berberis canadensis Mill. BA Beberis fendleri A. Gray EDNA11-0023781 BA Berberis subsessiliflora Pamp. EDNA09-00678 China: Hubei, W BA Berberis aetnensis C. Presl EDNA09-00621 China: Hubei, W BA Berberis garciae Pau. EDNA09-00626 Spain: Guadalajara BA province Berberis maderensis Lowe EDNA09-00641 Unknown (Cultivated) BA Berberis tsarongensis Stapf EDNA09-00207 China: Yunnan BA Berberis tsarongensis Stapf EDNA09-00385 China: Tibet BA Berberis minutiflora C.K.Schneid. EDNA09-00679 China: Yunnan, NW BA Berberis cretica L. EDNA09-00620 Unknown (cultivated) BA Berberis heteropoda Schrenk EDNA09-00675 Kyrgyzstan: Tien- BA Shan Berberis integerrima Bunge EDNA09-00637 Turkey: Kastamonu BA Berberis integerrima Bunge EDNA09-00616 Kazakhstan: Ili BA Intermountain Valley Berberis prattii C.K.Schneid. EDNA09-00379 China: Sichuan, W BA Berberis gyalaica Ahrendt EDNA09-00677 China: Tibet BA Berberis tschonoskyana Regel EDNA09-00623 Japan: Shikoku BA Berberis koreana Palib EDNA09-00624 Korea: Kyonggi-do BA Berberis amurensis var. japonica ( Regel ) EDNA09-00622 Japan: Hokkaido BA Rehder Berberis vulgaris L. EDNA09-00625 Spain: N Pyrenees BA Berberis kawakamii Hayata EDNA09-00373 Taiwan: Miaoli BA Berberis bergmanniae C.K.Schneid. EDNA09-00374 Unknown (cultivated) BA Berberis wallichiana DC. EDNA07-02235 Nepal: Rasuwa BA 63

Berberis sargentiana C.K.Schneid. EDNA09-00117 China: Hubei, W BA Berberis hookeri Lem. EDNA08-00188 BA Berbeis coxii C.K.Schneid. EDNA09-00113 China: Tibet BA Berberis congestiflora C. Gay EDNA09-00366 Chile: Araucania BA Berberis actinacantha Mart. EDNA09-00617 Chile BA Lam. EDNA09-00399 Argentina, NW: BA Sierras Grandes de Córdoba Berberis hieronymi C.K.Schneid. EDNA09-00671 Argentina, NW: BA Sierras Grandes de Córdoba Berberis rotundifolia Poepp. & Endl. EDNA09-00618 Chile: Biobío BA Berberis chilensis Gillies ex Hook. EDNA09-00367 Chile: Maule BA Forst. EDNA09-00362 Chile: Los Lagos BA Berberis negeriana Tischler EDNA09-00619 Chile: Biobio BA Phil. EDNA09-00368 Chile: Biobio BA Forst. EDNA09-00395 Chile BA Berberis montana Gay EDNA09-00364 Chile: Los Lagos BA Berberis trigona Kunze ex Poepp. & Endl. EDNA09-00672 Chile: Araucanía BA Berberis polyodonta (Fedde) Laferr. EDNA09-00413 China: Yunnan, W. BA Berberis duclouxiana (Gagnep.) Laferr. EDNA09-00370 China: Yunnan, NW BA Berberis gracilipes Oliv. EDNA09-00394 China: Sichuan, W BA Berberis napaulensis (DC.) Laferr. EDNA08-00185 LKSRB 1 BA Berberis nervosa Pursh EDNA09-00371 Canada: British BA Columbia Berberis repens Lindl. EDNA09-00398 Canada: British BA Columbia Berberis aquifolium Pursh EDNA09-00372 Canada: British BA Columbia Berberis fremontii Torr. EDNA09-00400 Unknown (cultivated) BA Berberis nevinii A. Gray. EDNA09-00615 Unknown (cultivated) BA Berberis higginsae Munz Y-D Kim et al. (2004) Berberis pallida Hartw. ex Benth. EDNA09-00401 Mexico: Hidalgo BA Nandina domestica Thunb. Gen Bank Acc: AY145148 Caulophyllum robustum Maxim. Gen Bank Acc: AY145149 Ranzania japonica T.Ito Gen Bank Acc: AY145150 Hydrastis canadensis L. Gen Bank Acc: AY145146 Clematis drummondii Torr. & A.Gray Gen Bank Acc: AY145147 Glaucidium palmatum Siebold & Zucc. Gen Bank Acc: AY145145 Cocculus carolinus DC. Gen Bank Acc: AY145144 Akebia quinata Decne. Gen Bank Acc: AY145143

64

leaved leaved

-

7.2 APPENDIX 2 – ITS TREE

leaved leaved

leaved leaved

-

-

imple

Berberis Compound

Australes Southern + temperate + Andes Tropical Brazil S

Septentrionales Septentrionales Simple

FIGURE 10. BAYESIAN 50% MAJORITY RULE CONSENSUS TREE OBTAINED FROM ITS SEQUENCES OF BERBERIS. BAYESIAN POSTERIOR PROBABILITIES (PP) ARE SHOWN ON EACH BRANCH. SPECIES ENCLOSED IN BLACK BOXES CORRESPOND TO TROPICAL SOUTH AMERICAN SPECIES WITHIN SEPTENTRIONALES CLADE. BERBERIS HIGGINSAE WAS SELECTED AS OUTGROUP.

65

7.3 APPENDIX 3 – NDHF TREE leaved

-

leaved leaved

leaved leaved leaved

- -

-

Outgroup

imple imple

Septentrionales Septentrionales Simple

Australes Southern temperate + Brazil Australes Andes Tropical S S

Berberis Compound

FIGURE 11. BAYESIAN 50% MAJORITY RULE CONSENSUS TREE OBTAINED FROM NDHF SEQUENCES OF BERBERIS. BAYESIAN POSTERIOR PROBABILITIES (PP) ARE SHOWN ON EACH BRANCH. SPECIES ENCLOSED IN BLACK BOXES CORRESPOND TO TROPICAL SOUTH AMERICAN SPECIES WITHIN SEPTENTRIONALES CLADE.

66

7.4 APPENDIX 6 – BAYESIAN TREE WITH ALL SOUTH AMERICAN SPECIES

leaved leaved

-

leaved leaved

leaved leaved leaved leaved

-

- -

imple

imple

Outgroup

Septentrionales Septentrionales Simple

Australes Southern temperate + Brazil S

Australes Andes Tropical S

Berberis Compound

FIGURE 12. COMBINED BAYESIAN 50% MAJORITY RULE CONSENSUS TREE OBTAINED FROM NDHF AND ITS SEQUENCES OF 116 ACCESSIONS REPRESENTING 108 SPECIES OF BERBERIS. BAYESIAN POSTERIOR PROBABILITIES ARE SHOWN ON EACH BRANCH. ALL ACCESSION OF TROPICAL SOUTH AMERICAN SPECIES WERE CONSIDERERD IN THE ANALYSIS, IN ORDER TO FIND CRYPTIC SPECIES.