Geography and ecology shape the phylogenetic composition of Amazonian tree communities

Aim: Amazonia hosts more tree species, from numerous evolutionary lineages both young and ancient, than any other biogeographic region. Previous studies have shown that tree lineages colonised multiple edaphic environments and dispersed widely across Amazonia, leading to a hypothesis, which we test, that lineages should not be strongly associated with either geographic regions or edaphic forest types.

Location: Amazonia.

Taxon: Angiosperms (Magnoliids; Monocots; Eudicots).

Methods: Data for the abundance of 5,082 tree species in 1,989 plots were combined with a mega-phylogeny. We applied evolutionary ordination to assess how phylogenetic composition varies across Amazonia. We used variation partitioning and Moran’s eigenvector maps (MEM) to test and quantify the separate and joint contributions of spatial and environmental variables to explain the phylogenetic composition of plots. We tested the indicator value of lineages for geographic regions and edaphic forest types and mapped associations onto the phylogeny.

Results: In the terra firme and várzea forest types, phylogenetic composition varies by geographic region, but the igapó and white-sand forest types retain a unique evolutionary signature regardless of region. Overall, we find that soil chemistry, climate, and topography explain 24% of the variation in phylogenetic composition, with 79% of that variation being spatially structured (R 2 = 19% overall for combined spatial/environmental effects). Phylogenetic composition also shows substantial spatial patterns not related to the environmental variables we quantified (R 2 = 28%). A greater number of lineages were significant indicators of geographic regions than forest types.

Main conclusions: Numerous tree lineages, including some ancient ones (>66 Ma), show strong associations with geographic regions and edaphic forest types of Amazonia. This shows that specialization on specific edaphic environments has played a long-standing role in the evolutionary assembly of Amazonian forests. Furthermore, many lineages, even those that have dispersed across Amazonia, dominate within a specific region, likely because of phylogenetically conserved niches for environmental conditions that are prevalent within regions. 

Methods
Geography and ecology shape the phylogenetic composition of Amazonian tree communities

Data files derived from the abundance of 5,082 tree species in 1,989 plots across Amazonia and used to support the findings described in Luize et al., 2024 Geography and ecology shape the phylogenetic composition of Amazonian tree communities, Journal of Biogeography.

This dataset includes three files:

1) Data contains the location of 1,989 tree inventory plots in Amazonia, associated evoPCA scores, and environmental data for each plot. To produce the evoPCA scores, we combined the abundance of 5,082 tree species with the phylogenetic relationships between species present in each plot. The environmental data for each plot was extracted from referenced data sources based on the locations of the plots. For soil attributes, we obtained point-data from soil profiles available in Soterlac (Dijkshoorn et al., 2005), ISRIC wise (Batjes, 2016), RAINFOR sites (Quesada et al., 2010), and Poels (1987), Zuquim et al (2019), and van Kekem et al. (1996) and created a loess model based (with a span of 0.2, a degree of 2, using Gaussian fitting) on all data available (data availability differed per parameter), to estimate the gradient of the soil chemical parameters across the entire Amazon region.

File = “Table_1_Evo_Assemb_Amaz_simpleCC0.csv” can be opened with R, text editor, or exported to Excel.

Rows are the values obtained for each plot

Column headers are as follows:

Axis 1 to Axis1983           The evoPCA scores for all axes

PlotCode           Code of plot as in ATDN database

Latitude           Plot latitude in decimal degrees

Longitude        Plot longitude in decimal degrees

2) a phylogenetic chronogram for the tree species observed in the tree inventories. To produce the phylogenetic hypothesis, we used the GBOTB phylogeny as the base topology (Smith & Brown, 2018) and the Scenario 3 in the V.phylomaker v.2 R package (Jin & Qian, 2022) to prune and bind species.

File = “Phylo_Amazonian_trees_ATDN_1989plots.phy” can be opened with R using library “ape” or with the Figtree app

3) Indicator analysis table with the results of species and lineages (above species level) that were determined as indicators of geographic regions or forest types within Amazonia. (Table 3. Nodes and tips determined as significant indicators for Amazonian forest types and geographic regions. All those taxa determined as indicator has IndVal p.value<0.01 after 999 permutations. Indicator of: Forest types considered: TF = Terra-firme; PZ=White-sand forest; VA=Várzea; IG=Igapó; SW=Swamp; and Geographic regions considered: CA=Central Amazonia; SA=Southern Amazonia; EA=Eastern Amazonia; NWA=Northwestern Amazonia; SWA=Southwestern Amazonia; GS=Guiana Shield. ) Results are described in “Geography and ecology shape the phylogenetic composition of Amazonian tree communities”, Luize et al. 2024 Journal of Biogeography.

File = “Table_2_Evo_Assemb_Amaz.xlsx” Excel file with the results of indicator analysis

rows are the values for each lineage or species

columns are as follows:

Nodes or tips   The lineages (species or higher level lineages) determined as indicators

rank         Ranking for the lineages indicator values

ran_test_for   If the indicator value was tested for geographic regions or forest types

indicator_of          The geographic region or the forest type that the lineage was an indicator

Specificity_A        The specificity statistic as described in Legendre & Dufrene original indicator analysis publication

Fidelity_B              The fidelity statistic as described in Legendre & Dufrene original indicator analysis publication

indval      The Indicator value statistic as described in Legendre & Dufrene original indicator analysis publication

p.value      Associated p values for the indval statistic

descendentesGenera           The genera that were descendants from lineages determined as indicators

Sharing/Access information – Open access

Links to other publicly accessible locations of the data: https://sites.google.com/naturalis.nl/amazon-tree-diversity-network/homepage?authuser=0

Data was derived from the Amazon Tree Diversity Network (ATDN) database. The abundance matrix of species per location is available upon request following ATDN agreements.