Tree species distribution and phylogenetic diversity across southern South America

The present-day southern South American flora comprises a group of lineages with southern temperate affinities, which have been suggested to have evolved during and after the breakup of Gondwana, as well as a group of Neotropical elements which are largely found in northern South America. Here, I aim to assess the floristic coherence of the main vegetation types that inhabit this region and determine the principal drivers of variation in the evolutionary diversity of these forest communities. I compiled a database of 3662 species of shrubs and trees at 781 sampling sites spread over six countries in South America. To understand the vegetation types that occur in the region I conducted clustering and ordination analyses. I then compared the defined vegetation types using taxonomic distinction, indicator species and phylogenetic diversity analyses. My results indicated a high diversity of vegetation types in terms of woody floristic composition, with a large number of indicator species, many of which are endemic. Contradicting the idea that temperate floras are recently derived and evolutionarily poor subsets of tropical floras, I demonstrated that the forests of the far south of South America are characterized by exceptional evolutionary diversity.


Tree species distribution and phylogenetic diversity across southern South America
Vanessa Leite Rezende

Thesis Abstract
Abstract. The present-day southern South American flora comprises a group of lineages with southern temperate affinities, which have been suggested to have evolved during and after the breakup of Gondwana, as well as a group of Neotropical elements which are largely found in northern South America. Here, I aim to assess the floristic coherence of the main vegetation types that inhabit this region and determine the principal drivers of variation in the evolutionary diversity of these forest communities. I compiled a database of 3662 species of shrubs and trees at 781 sampling sites spread over six countries in South America. To understand the vegetation types that occur in the region I conducted clustering and ordination analyses. I then compared the defined vegetation types using taxonomic distinction, indicator species and phylogenetic diversity analyses. My results indicated a high diversity of vegetation types in terms of woody floristic composition, with a large number of indicator species, many of which are endemic. Contradicting the idea that temperate floras are recently derived and evolutionarily poor subsets of tropical floras, I demonstrated that the forests of the far south of South America are characterized by exceptional evolutionary diversity. Keywords. Temperate forest, tropical forest, tropical conservatism hypothesis, phylogenetic diversity, latitudinal gradients.  Wiens and Donoghue 2004), the latitudinal diversity gradient (the pattern whereby species richness declines from equatorial regions towards the poles) is largely governed by evolutionarily conserved ancestral preferences (Wiens and Graham 2005, Kozak and Wiens 2010, Romdal et al. 2013. This hypothesis assumes that most clades originated in tropical conditions that were widespread from the beginning of the Cretaceous to the end of the Eocene (Davies et al. 2004, Ruddiman 2006, and that most species in these clades have at least partly retained their ancestral physiological tolerances Donoghue 2004, Jansson et al. 2013). Thus, according to the TCH, species richness, evolutionary diversity and lineage age will be high in regions characterized by warm, nonfreezing temperatures, since these conditions match the ancestral niches of many extant clades, while temperate regions will be occupied by fewer, younger and less diverse clades, because the transitions to cold, temperate environments have been relatively infrequent and recent (Wiens and Donoghue 2004).

Methods
The floristic data were retrieved from the Ne-oTropTree database (Oliveira-Filho 2015), which is a compilation of tree species checklists gathered from the literature with additional occurrence records obtained from verified herbarium specimens. Each site comprises a 5 km radius circle (see description, history and protocol at http: // www.icb.ufmg.br/treeatlan). NeoTropTree defines trees as freestanding plants with stems that can reach over 3 m in height. Thus, the NeoTropTree database includes, palms, tree ferns and bamboos, when they fit the inclusion criteria. My database consisted of 781 plant inventories of tree communities (I refer to all plant species sampled in these plots as trees) spread over six countries (Brazil, Bolivia, Argentina, Chile, Uruguay and Paraguay) with 249,374 occurrence records.
To define the main vegetation types, I performed a cluster analysis (with support assessed via 1000 replicates), non-metric multidimensional scaling ordination (NMDS) and taxonomic distinction analyses. I generated the agglomerative dendrogram using the Jaccard coefficient of floristic similarity. I chose this coefficient because it produced the highest cophenetic correlation. To determine which species are responsible for the distinction between the main vegetation types, I performed an indicator species analysis using the method proposed by Tichý and Chytrý (2006).
For the phylogenetic analysis, I generated an ultrametric phylogeny using Phylocom (v.4.2), starting with the following reference tree: R20120829.new (from http://phylodiversity.net/ phylomatic/). For each tree community (each site), I calculated phylogenetic diversity sensu stricto (PD), which is the total sum of branch lengths in a phylogeny comprising species in the community and its equivalent, standardized for species richness (ses.PD). For this analysis I opted to study the pattern across multiple vegetation types, rather than just the gradient as a whole, as different vegetation types may have different histories and show different patterns of phylogenetic diversity. For this analysis, I excluded Pacific Insular Forest (only 1 sampling site) and combined Mediterranean Chile Forests with Temperate Pacific Forest, Dry Chaco with Wet Chaco and Pampa Deciduous Forest with Pampa Riverine Forest, since these vegetation types have a low number of sampling sites. I also divided Atlantic Semideciduous Forest into tropical and subtropical portions (see Rezende 2017 for details).

Results
The agglomerative dendrogram analysis showed a floristic differentiation among 16 Main Vegetation Types (Figs. 1, 2). There was a strong division into four major groups, which primarily differed in the phytogeographical domain to which they belong, thermal domain, leaf flush and the geographical location (east and west side of the Andes; Table  1). The Atlantic Rain Forest was the vegetation type with the highest number of indicator species. However, when I analyzed the proportion of the number of indicator species out of the number of species occurring in a given vegetation type, the Pacific Flora (Mediterranean and Temperate For-est) had the highest proportion of indicator species (38 and 34 %), whereas the Atlantic Rain Forest had only 4 %. Also, the highest indicator values (IV) for species (IV > 0.95) were found for Mediterranean Chile and the Temperate Pacific. When analysing species richness (SR), Atlantic Rain Forest has the highest SR, with an average of 323 species per site, whereas the lowest SR was found in the Pacific Forest (29 species per community).
Across all sites, I found phylogenetic diversity sensu stricto (PD) to be strongly correlated with SR. These metrics showed a strong latitudinal pattern, where the highest value of PD was found for the communities in Atlantic Rain Forest (178), while the lowest values of PD where found for communities in the Pacific Forest (306). However, the standardized metric (ses.PD) showed an opposite result to that found for PD and SR. The highest value was found for the Pacific Forest communities (0.97), while the Atlantic Rain Forest had the lowest value (-6.35; Fig. 3).

Discussion
My results indicate that there are many vegetation types in southern South America that are distinct from tropical biomes and should therefore not be included in the Neotropical floristic prov-   ince (Rezende et al. unpubl. data). The large number of vegetation types is due to the floristic complexity found in the region (Rezende et al. 2016, Rezende 2017. The highest number of indicator species was recorded for the Atlantic Rain Forests, which was expected because this vegetation type also has the largest number of species. However, the largest proportional values (indicator species in relation to total richness) and the largest IV values (IV > 0.95) were recorded for the Mediterranean and Temperate Pacific forests (Rezende et al. 2016, Rezende 2017. These vegetation types located west of the Andes have a wide spectrum of biogeographic elements, including genera with many endemic species that appear to have radiated within them (Arroyo et al. 2008).
As with species richness (SR), phylogenetic diversity (PD) declines with distance from the equator, which is expected given how tightly correlated SR and PD are (Rezende et al. unpubl. data, Rezende 2017). However, once I controlled for this correlation and evaluated phylogenetic diversity with a standardized metric (ses.PD), I found that lineage diversity increases further south. Thus tree communities in the far south of South America show greater lineage diversity than expected given their species richness (Rezende et al. unpubl. data, Rezende 2017). This unexpected result may be because the temperate flora has high richness at the family level with low generic and species diversity, as well as a high proportion of monotypic genera, with several of these genera even representing monogeneric families (e.g., Aextoxicaceae, Gomortegaceae, Desfontainiaceae and Eucryphiaceae; Vilagran and Hinojosa 1997).
Overall, my findings show that the flora of South America's extratropical regions is not a narrow, cold-adapted subset of that of tropical regions. Rather, as suggested by Segovia and Armesto (2015) the extratropical floras also include diverse lineages that likely have a temperate southern hemisphere origin. My findings demonstrate that these southern forests are evolutionarily distinct with a distinct evolutionary history from the tropics and allow us to refute the Tropical Conservatism Hypothesis (TCH). Moreover, my results do not support the idea that temperate floras are recently derived from tropical floras. Indeed, the distinctive ecology and biogeographical history of each biome could be a key to understanding the distribution of tree species in southern South America.