Oikos OIK-05142
Martins, V. F., dos Santos Seger, G. D., Wiegand, T. and dos Santos, F. A. M. 2018. Phylogeny contributes more than site characteristics and traits to the spatial distribution pattern of tropical tree populations. – Oikos doi: 10.1111/oik.05142
Appendix 1 Review of published studies that evaluated the influence of species traits or phylogenetic relationships among taxa on the spatial distribution pattern of populations of tree species at local scale in different forest types.
1
Reference Bleher et al. (2002) Clark et al. (2017) Condit et al. (2000) Type of study Simulation Data collection Data collection Forest type ---- Tropical Tropical Sexual system, population density (this is not a species trait), spatial distribution pattern of Dispersal syndrome, parental population (also Population density (this canopy layer (canopy, not a species trait), mean is not a species trait), midcanopy, understory, dispersal distance, and dispersal syndrome, and and shrub; in this sense, distribution of dispersal growth form (canopy Trait can be understood as distance (negative and understory; in this species height class), stem exponential and lognormal sense, can be DBH as a proxy for height, curve; mean dispersal understood as species seed mass, wood density, distance and distribution height class) and shade tolerance of dispersal distance are partially determined by the species dispersal syndrome) Phylogeny No No No Six plots ranging from Study design One 25-ha plot One 50-ha plot 25 ha to 52 ha Index C derived from the ratio of squared distances between pairs of points Relative neighbourhood Measure of spatial calculated by the T- Wavelet variance density (pair correlation distribution pattern Square-method and function) Morisita's index of dispersion 40 log-evenly spaced Spatial scale of interest Not informed scales between 2 m and 0–250 m and 0–500 m 115 m Dioecious species were more aggregated than Species with explosive monoecious species; rare dispersal were more species were more aggregated, followed by aggregated that common animal dispersal, and then species; the spatial Rare species were more by wind dispersal; distribution pattern of aggregated than aggregation decreased parental population had common species; from shrubs, then only weak effects on the species not dispersed by understory, to midcanopy pattern of next animals were more Main results and canopy species; generations; species with aggregated than animal- shorter species were more short dispersal distances dispersed species; short aggregated than taller were more aggregated species were more species; aggregation varied than species with large aggregated than tall in no obvious pattern with dispersal distances; the species seed mass; aggregation distribution of dispersal declined with wood distances had only weak density and shade- effects on the spatial tolerance distribution patterns of populations Sexual system, and Dispersal syndrome, Population density and especially population canopy layer, species dispersal syndrome are density and mean height, wood density, and Conclusions important drivers of dispersal distance are the shade tolerance are spatial distribution main drivers of spatial important drivers of spatial patterns distribution patterns distribution patterns
2 Reference Flügge et al. (2012) He et al. (1997) Hubbell (1979) Li et al. (2009) Simulation and data Type of study Data collection Data collection Data collection collection Forest type Tropical Tropical Tropical Subtropical Population density Population (this is not a Population density Population density density (this is species trait), (this is not a species Trait (this is not a species not a species dispersal trait) and dispersal trait) trait) syndrome, and syndrome sexual system Phylogeny Yes No No No Study design One 50-ha plot One 50-ha plot One 13.44-ha plot One 20 ha-plot Clumping index based on Relative Relative Donnelly nearest- Measure of spatial neighbourhood Morisita's index neighbourhood neighbour distribution pattern density (pair of dispersion density (pair distance statistic correlation function) correlation function) corrected for edge effects Spatial scale of 0–10 m 0–250 m 0–196 m 0–10 m interest Rare species were more aggregated Rare species were No effect of than common more aggregated than phylogeny; rare species; intermediate and species were more aggregation was common species; aggregated than Common species more pronounced wind- or explosively common species, so were slightly in mammal- dispersed species were Main results that recently growing more aggregated dispersed species, more aggregated than populations were than rare species then on wind, and species dispersed by more aggregated than last on species animals, followed by declining populations dispersed by species dispersed both of the same current birds/bats; no by wind/explosion and density effect of sexual animals system Both current density Population density Population density and recent changes in Population and dispersal and dispersal density leave their density has small syndrome are syndrome are Conclusions mark on the degree of effects on spatial important drivers important drivers of aggregation in locally distribution of spatial spatial distribution growing or declining patterns distribution patterns populations patterns
3
Reference Mou et al. (2005) Nanami et al. (1999) Pastor et al. (1999) Type of study Data collection Data collection Simulation Forest type Subtropical Temperate ---- Regeneration strategy (via seed Dispersal distance (local and bank, newly dispersed seeds, widespread; dispersal distance Trait sprouting, and mixed Dispersal syndrome is partially determined by the sprouting/newly dispersed species dispersal syndrome) seeds) Phylogeny No No No Study design Four 0.25-ha plots One 40 × 40 m plot Two landscapes of 5 × 5 plots Measure of spatial Global variability and visual L-function Visual inspection of maps distribution inspection of kriged maps pattern Spatial scale Sampling unit (plot) of 32 × 32 Sampling unit of 2.5 × 2.5 m 0–15 m of interest m After disturbance, seed bank species increased rapidly and formed large patches, and then quickly declined; species that rely on newly dispersed seeds Large distance seed occurred at first in a few dispersal by birds patches and became scatters new individuals Local seed dispersal resulted in widespread later; sprouters had in the environment, more aggregated spatial Main results rapid increases in cover, but whereas seed dispersal distribution patterns than their spatial distribution by gravity creates clumps widespread seed dispersal patterns were largely of new plants near seed determined by their pre- sources disturbance patterns; species that regenerate via sprouting/newly dispersed seeds had moderate cover and a random distribution The impact of regeneration strategy on the spatial distribution patterning may not have been fully expressed Result expected from previous because (1) spatial distribution studies; simulation assumed no patterns were highly variable Large dispersal distances underlying environmental within regeneration groups; (2) can weaken the effects of Conclusions heterogeneity, which can factors beyond regeneration dioecy on the spatial change the correspondence strategy can also affect the distribution patterns between seed dispersal patterns spatial distribution pattern; (3) and recruitment patterns low number of species/regeneration group; and (4) crude classification of regeneration groups
4
Reference Plotkin et al. (2000) Réjou-Méchain et al. (2011) Type of study Data collection Data collection Forest type Tropical Tropical Dispersal syndrome, wood density, shade Population density (this is not a species tolerance (pioneer, non-pioneer light- Trait trait) demanding, and shade-tolerant species), and population density (this is not a species trait) Phylogeny No Yes Study design Three 50-ha plots Five sites with transects containing 0.5-ha plots Measure of k-statistic compiling Ripley's K-function spatial information, and number of clumps and As (minus the slope of the pair correlation distribution clump size calculated from the fit of a function on spatial scale r on ln(r)) pattern Poisson cluster process 20 distances equally spaced between 0 m Spatial scale and 250 m (for one plot) or 385 m (for the Local (0.2–1 km), meso (1–10 km), and of interest other plots) for the calculation of k- landscape (10–80 km) statistics k-statistics showed that rare species were slightly more aggregated than common species; no effect of population density on Populations tended to be aggregated and the number of clumps; there was a slight aggregation patterns were partly explained by negative relationship between population species identity; family level had a significant Main results density and clump size, meaning rare effect on As values, but there was no influence species would be less aggregated than of traits; rare species were more aggregated common species; nevertheless, the than common species relationship between aggregation and population density arises from a statistical artifact Aggregation is partly context-dependent and There is no biological effect of population partly explained by traits displaying Conclusions density on spatial distribution patterns phylogenetic conservatism; further studies are necessary to clearly identify them
5
Reference Seidler and Plotkin (2006) Wang et al. (2010) Type of study Data collection Data collection Forest type Tropical Temperate Population density (this is not a species trait), growth form (overstory, midstory, and understory; in this sense, can be understood Trait Dispersal syndrome and species height as species height class), dispersal syndrome, and shade tolerance (light-demanding, mid- tolerant, and shade-tolerant species) Phylogeny Yes No Study design Two 50-ha plots One 25-ha plot Measure of Ripley's K-function, and number of clumps spatial Relative neighbourhood density (pair and clump size calculated from the fit of a distribution correlation function) Poisson cluster process pattern Spatial scale 250 m for the calculation of K-function 0–10 m of interest Aggregation decreased in the following order Rare species were more aggregated than in one plot: species dispersed ballistically, by intermediate and common species; short gravity, gyration, wind, and animals (fruits species were more aggregated than tall less than 2 cm in diameter > 2–5 cm > 5 cm); species; gravity- and wind-dispersed species Main results aggregation decreased in the following order were more aggregated than species in the other plot: species dispersed dispersed by animals; light-demanding and ballistically, by wind, mammals, and shade-tolerant species were more birds/bats; no effect of species height on aggregated than mid-tolerant species population spatial distribution patterns Population density, dispersal distance (as Dispersal syndromes with a greater potential indicated by dispersal syndrome and species Conclusions for long-distance seed transport result in height), and shade tolerance are important decreased population aggregation drivers of spatial distribution patterns
References
Bleher, B. et al. 2002. Seed dispersal, breeding system, tree density and the spatial pattern of trees – a simulation approach. – Basic Appl. Ecol. 3: 115–123. Clark, A. T. et al. 2017. Functional traits of tropical trees and lianas explain spatial structure across multiple scales. – J. Ecol. 00: 1–12 doi: 10.1111/1365-2745.12804. Condit, R. et al. 2000. Spatial patterns in the distribution of tropical tree species. – Science 288: 1414–1418. Flügge, A. J. et al. 2012. The memory of spatial patterns – changes in local abundance and aggregation in a tropical forest. – Ecology 93: 1540–1549. He, F. et al. 1997. Distribution patterns of tree species in a malaysian tropical rain forest. – J. Veg. Sci. 8: 105–114. Hubbell, S. P. 1979. Tree dispersion, abundance and diversity in a tropical dry forest. – Science 203: 1299–1309. Li, L. et al. 2009. Spatial distributions of tree species in a subtropical forest of China. – Oikos 118:
6 495–502. Mou, P. et al. 2005. Regeneration strategies, disturbance and plant interactions as organizers of vegetation spatial patterns in a pine forest. – Landscape Ecol. 20: 971–987. Nanami, S. et al. 1999. Dioecy-induced spatial patterns of two codominant tree species, Podocarpus nagi and Neolitsea aciculata. – J. Ecol. 87: 678–687. Pastor, J. et al. 1999. Generation of spatial patterns in boreal forest landscapes. – Ecosystems 2: 439–450. Plotkin, J. B. et al. 2000. Species-area curves, spatial aggregation and habitat specialization in tropical forests. – J. Theor. Biol. 207: 81–99. Réjou-Méchain, M. et al. 2011. Spatial aggregation of tropical trees at multiple spatial scales. – J. Ecol. 99: 1373–1381. Seidler, T. G. and Plotkin, J. B. 2006. Seed dispersal and spatial pattern in tropical trees. – PLoS Biol. 4: 2132–2137. Wang, X. et al. 2010. Spatial distributions of species in an old-growth temperate forest, northeastern China. – Can. J. For. Res. 40: 1011–1019.
7 Appendix 2
Characteristics of fourteen 1-ha plots established at the Atlantic Rainforest along the elevation gradient of the state park “Parque Estadual da Serra do Mar”, SE Brazil, traits included in the analysis of the spatial distribution pattern of tree populations, and distribution of populations in the study plots.
8 Table A1. Characteristics of fourteen 1-ha plots established at the Atlantic Rainforest along the elevation gradient of the state park “Parque Estadual da Serra do Mar”, SE
Brazil. Variables marked with * were not included in the site dataset for the analysis of the spatial distribution pattern of tree populations (plot A = Restinga Forest; plot F =
Disturbed Lowland Atlantic Rainforest; plots C, L, and M presented less than five populations each).
Vegetation Stand structure and richness Disturbance Topography No. multistemmed Total basal Average No. No. No. Total basal Total gap Total elevation Mean slope Plot Forest type1 individuals area of dead elevation (m, individuals2 families2 species2 area (m2)2 area (m2)3 range (m, asl)5 (o)5 (≥ 2 stems)4 stems (m2)4 asl)5 A* restinga 1634 31 83 28.12 2052.9 117 0.40 10 1 1.1 B lowland 1143 37 147 28.93 2598.9 69 0.44 46 23 14.7 C* lowland 1167 36 122 26.43 2588.1 83 1.71 64 12 15.7 D lowland 1324 40 158 30.94 1870.9 65 0.89 57 26 12.6 E lowland 1253 40 139 30.92 1908.1 75 1.03 73 25 11.1 disturbed F* 1264 37 105 29.50 2418.3 184 1.12 110 27 14.3 lowland G submontane 1513 40 149 36.92 1280.4 46 1.95 190 19 14.2 H submontane 1519 40 159 35.97 1196.2 56 1.73 209 16 11.5 I submontane 2003 48 203 45.09 939.0 48 0.83 327 48 27.4 J submontane 1823 47 208 43.62 1232.2 31 2.11 352 46 28.2 K montane 1767 42 177 39.94 1531.0 44 3.16 1037 47 27.3 L* montane 1664 37 168 35.90 1599.0 54 2.40 1020 50 26.1 M* montane 1822 38 171 43.45 1317.8 74 3.23 1024 49 26.3 N montane 1397 38 145 35.63 2303.2 64 3.14 1019 34 16.6 1Joly et al. (2012); 2data compiled from Joly et al. (2012) and updated from the Functional Gradient Project’s database; 3C.J.Caron, M.A.S. Scaranello, F.A.M. Santos and
L.F. Alves (unpublished data); 4Alves et al. (2010), and L.F. Alves, S.A. Vieira, M.A.S. Scaranello, C.A. Joly and L.A. Martinelli (unpublished data); 5Alves et al. 2010 -
plots F to N modified after Leitold (2014).
9 The traits included in the analysis of the spatial distribution pattern of populations are:
• Sexual system: dioecy, homoecy, and dioecy.
• Dispersal syndrome within the hierarchical levels of classification proposed by Martins
et al. (2014):
• Level 1: biotic and abiotic seed dispersal.
• Level 4: bird and mammal dispersal.
• Level 5: seeds with fleshy appendages attached, drupoid diaspores, diaspores with
attractive colours (primate dispersal), dust diaspores, and winged diaspores.
• Species height.
• Wood density.
60
50
40
30
(%) Species 20
10
0 1 2 3 4 5 6 7 8 9 10 Number of plots
Figure A1. Number of plots in which tree species with 30 or more individuals/plot are found at
the Atlantic Rainforest in southeastern Brazil.
References Alves, L. F. et al. 2010. Forest structure and live aboveground biomass variation along an elevational 10 gradient of tropical Atlantic moist forest (Brazil). – For. Ecol. Manage. 260: 679–691. Joly, C. A. et al. 2012. Florística e fitossociologia em parcelas permanentes da Mata Atlântica do sudeste do Brasil ao longo de um gradiente altitudinal. – Biota Neotrop. 12: 123–145. Leitold, V. 2014. Airborne lidar-based estimates of tropical forest structure and ground topography in a mountainous area of the Brazilian Atlantic forest. – MSc thesis, Brazilian Institute of Space Research. Martins, V. F. et al. 2014. Dispersal spectrum of four forest types along an altitudinal range of the Brazilian Atlantic Rainforest. – Biota Neotrop. 14: 1–22.
11 Appendix 3 Potential drivers of the spatial distribution pattern of tree populations sampled in fourteen 1-ha plots established at the Atlantic Rainforest along the elevation gradient of the state park ‘Parque Estadual da Serra do Mar’, southeastern Brazil.
Dispersal syndrome and sexual system We used specialized literature, herbarium vouchers, and the expertise of plant taxonomists to determine the dispersal syndrome and sexual system of each of the 46 species studied. Even though dispersal syndromes cannot be entirely interpretable as adaptations to seed dispersal agents, at least to frugivores (Jordano 1995, Howe 2016), dispersal syndromes are still the best explanation for the wide variety of diaspore morphology (Howe 2016) and are thoroughly used as a potential driver of the spatial pattern of tree populations (Hubbell 1979, Nanami et al. 1999, Condit et al. 2000, Seidler and Plotkin 2006, Li et al. 2009, Wang et al. 2010, Réjou-Méchain et al. 2011). Diaspore morphology presents different levels of detail and consequently possible different degrees of specificity to seed dispersers. Because of that, Martins et al. (2014) organized the dispersal syndromes into five hierarchical levels. The two syndromes in the first, most general hierarchical level are biotic and abiotic seed dispersal. Biotic dispersal can be performed by animals and by the parent plant itself, while abiotic dispersal can be performed either by wind or water. Hence, these four syndromes compose the second hierarchical level of the seed dispersal syndrome classification. In the third hierarchical level, there are endozoochory, epizoochory, autochory, and barochory; in the forth level, seed dispersal syndromes related to different frugivores, and, in the most specific level, there are 12 syndromes, all related to the morphology of diaspores dispersed by birds, mammals, autochory, wind, and water. Restraining diaspore morphology and consequently its most likely primary dispersal agent should enable the evaluation of underlying patterns of seed dispersal as opposed to the great variation that emerges when syndromes from different hierarchical levels are considered together (Martins et al. 2014). Therefore, to accurately determine the influence of seed dispersal syndromes on the spatial distribution pattern of populations, the degree of overdispersion or aggregation should be compared among syndromes within the same hierarchical level. Many plant species are subject to secondary dispersal by animals or water (Seidler and Plotkin 2006). However, for the purpose of this study, we examined only the primary phase of dispersal, as did other authors (Hubbell 1979, Nanami et al. 1999, Condit et al. 2000, Seidler and Plotkin 2006, Li et al. 2009, Wang et al. 2010, Réjou-Méchain et al. 2011).
12 We classified the sexual system as dioecious, homoecious or monoecious (sensu Cruden and Lloyd 1995). We could only determine the dispersal syndrome and sexual system of morphotypes when its immediate upper taxonomic level (genus) had only one dispersal syndrome or sexual system.
Species height For our analysis, we need to attribute a height value for each population studied, as populations are the sampling unit where the spatial distribution pattern is characterized. To do so, we used the 98th percentile value of stem DBH of each tree species in order to calculate the maximum species height (Réjou-Méchain and Cheptou 2015). By using this value instead of the largest stem DBH, we avoid the influence of exceptionally large trees on our results. We used height-diameter Weibull models to calculate species height. Because the allometry of tree species changes with altitude at the study site, different models were built for each forest type (Scaranello et al. 2012). Therefore, we first pooled all individuals of a given tree species present at plots of the same forest type. Then we inserted the 98th percentile value of stem DBH of the species in the model built for the correspondent forest type. For the palm species included in our dataset, Euterpe edulis (Arecaceae), we obtained height measurements from Scaranello et al. (2012). The authors used a telescopic ruler or a laser distance meter to measure plant height. We determined the maximum species height as the 98th percentile of height value of individuals present in the same forest type. The maximum height of trees and palms, as here calculated, ranged from 9.2 m for Stylogyne lhotskyana (Primulaceae) to 26.8 m for Hieronyma alchorneoides (Phyllanthaceae).
Shared evolutionary history To determine the phylogenetic relationships among the species studied, we used the phylogenetic hypothesis and clade age estimates of Magallón et al. (2015). These authors built a phylogenetic tree for 87% of angiosperm families based on five molecular markers (atpB, rbcL, matK, 18S and 26S) and 137 fossil-based calibrations of phylogenetic nodes. Tree resolution affects different phylogenetic metrics (Swenson 2009, Seger et al. 2013) and the megatree of Magallón et al. (2015) only solves relationships up to the family level. Because of that, we built nine phylogenetic trees to the genus level for families with more than one species sampled at the study site (accounting for 75% of the species) and incorporated them to the megatree. These nine trees were built based on recent studies that used molecular markers to solve relationships among clades (references at the end of this appendix). With
13 this procedure, our phylogenetic tree solved all relationships among genera and reduced polytomies from 17.1% (for a tree solved to the family level) to 10%. We defined the branch lengths of the megatree using the clade ages proposed by Magallón et al. (2015) and the age estimates available in recent studies for some of the families incorporated in the megatree (references at the end of this appendix). We selected these ages as long as they were not older than the age of the immediate upper clade according to Magallón et al. (2015). We estimated the nodes without clade age information through rate smoothing, using the non-parametric rate smoothing technique (Sanderson 1997) and setting the smoothing parameter (lambda) to one. This technique assumes that substitution rates on two contiguous branches are likely to be similar (i.e. they are autocorrelated). We calculated the node age estimates with the function chronopl using the correlated model at the package ape ver. 3.2 (Paradis et al. 2004) in R (
Variable selection for trait, phylogenetic relationship, and site components included in variation partitioning analysis of the degree of overdispersion or aggregation of populations Before performing the variation partitioning analysis of the spatial distribution pattern as measured by the standardized pair correlation function index err(r) for the 101 tree populations studied, we selected the variables to be included in our three explanatory components. In order to obtain the phylogenetic relationship component P, we first square root-transformed the phylogenetic distance values and then extracted the eigenvectors of the phylogenetic distance matrix using a principal coordinate analysis (PCoA) performed with the package APE ver. 3.0-1 (Paradis et al. 2004) of R. The eigenvectors are a transformed measurement of phylogenetic relationships among species and the first ones express divergences near the phylogenetic tree root (Diniz-Filho et al. 2012).
In order to obtain the trait and site components T and S, we first log10-transformed all quantitative variables, and then performed forward selection of trait and site variables separately. The
14 forward selection procedure starts with no explanatory variable in the model. The variables that are included in each component are those that produce the largest increase in the coefficient of multiple determination R2, provided this increase is significantly different from zero, using a pre-determined significance level α (Legendre and Legendre 1998). Forward selection is a simple procedure and most reduces sum-squared-errors. Additionally, it is the preferred variable selection method for variation partitioning analysis (Borcard et al. 1992, Desdevises et al. 2003, Gilbert and Bennet 2010, Liu et al. 2013, Duarte et al. 2014, Arellano et al. 2015). In order to reduce forward selection of redundant variables (type I error), we used the double-stopping procedure proposed by Blanchet et al. (2008). First, we carried out a global test using all explanatory variables within one component. If the global test was significant, we then selected the variables that significantly explained err(r) using two criteria: 2 2 2 the significance level (here, α = 0.05) and the adjusted R (R adj) of the global test. We used R adj instead of R2 because it considers the number of samples and explanatory variables in each analysis 2 2 (Peres-Neto et al. 2006). Only variables with p < 0.05 and R adj < R adj of the global model were included in T or S as explanatory variables. We performed the forward selection with the package packfor ver. 0.0-8 (Dray et al. 2007) in R using 9999 permutations.
References Arellano, G. et al. 2015. Disentangling environmental and spatial processes of community assembly in tropical forests from local to regional scales. – Oikos 125: 326–335. Blanchet, F. G. et al. 2008. Forward selection of explanatory variables. – Ecology 89: 2623–2632. Borcard, D. et al. 1992. Partialling out the spatial component of ecological variation. – Ecology 73: 1045–1055. Condit, R. et al. 2000. Spatial patterns in the distribution of tropical tree species. – Science 288: 1414– 1418. Cruden, R. W. and Lloyd, R. M. 1995. Embryophytes have equivalent sexual phenotypes and breeding systems: why not a common terminology to describe them? – Am. J. Bot. 82: 816–825. Desdevises, Y. et al. 2003. Quantifying phylogenetically structured environmental variation. – Evolution 57: 2647–2652. Diniz-Filho, J. A. F. et al. 2012. Exploring patterns of interspecific variation in quantitative traits using sequential phylogenetic eigenvector regressions. – Evolution 66: 1079–1090. Dray, S. et al. 2007. Packfor: forward selection with permutation (Canoco p.46). – R package ver. 0.0- 8.
15 Duarte, L. D. S. et al. 2014. Climate effects on amphibian distributions depend of phylogenetic resolution and biogeographical history of taxa. – Global Ecol. Biogeogr. 23: 213–222. Gilbert, B. and Bennett, J. R. 2010. Partitioning variation in ecological communities: do the numbers add up? – J. Appl. Ecol. 47: 1071–1082. Howe, H. F. 2016. Making dispersal syndromes and networks useful in tropical conservation and restoration. – Global Ecol. Conserv. 6: 152–178. Hubbell, S. P. 1979. Tree dispersion, abundance, and diversity in a tropical dry forest. – Science 203: 1299–1309. Jordano, P. 1995. angiosperm fleshy fruits and seed dispersers: a comparative analysis of adaptation and constraints in plant-animal interactions. – Am. Nat. 145: 163–191. Legendre, P. and Legendre, L. 1998. Numerical ecology, 2nd edn. – Elsevier Science B.V. Li, L. et al. 2009. Spatial distributions of tree species in a subtropical forest of China. – Oikos 118: 495–502. Liu, X. et al. 2013. The environment and space, not phylogeny, determine trait dispersion in a subtropical forest. – Funct. Ecol. 27: 264–272. Magallón, S. et al. 2015. A metacalibrated time-tree documents the early rise of flowering plant phylogenetic diversity. – New Phytol. 207: 437–453. Martins, V. F. et al. 2014. Dispersal spectrum of four forest types along an altitudinal range of the Brazilian Atlantic Rainforest. – Biota Neotrop. 14: 1–22. Nanami, S. et al. 1999. Dioecy-induced spatial patterns of two codominant tree species, Podocarpus nagi and Neolitsea aciculata. – J. Ecol. 87: 678–687. Paradis, E. et al. 2004. APE: analyses of phylogenetics and evolution in R language. – Bioinformatics 20: 289–290. Peres-Neto, P. R., et al. 2006. Variation partitioning of species data matrices: estimation and comparison of fractions. – Ecology 87: 2614–2625. Réjou-Méchain, M. and Cheptou, P.O. 2015. High incidence of dioecy in young sucessional tropical forests. – J. Ecol. 103: 725–732. Réjou-Méchain, M. et al. 2011. Spatial aggregation of tropical trees at multiple spatial scales. – J. Ecol. 99: 1373–1381. Sanderson, M. J. 1997. A nonparametric approach to estimating divergence times in the absence of rate constancy. – Mol. Biol. Evol. 14: 1218–1231. Scaranello, M. A. S. et al. 2012. Height-diameter relationships of tropical Atlantic moist forest trees in southeastern Brazil. – Sci. Agr. 69: 26–37. 16 Seger, G. D. S. et al. 2013. Discriminating the effects of phylogenetic hypothesis, tree resolution and clade age estimates on phylogenetic signal measurements. – Plant Biol. 15: 858–867. Seidler, T. G. and Plotkin, J. B. 2006. Seed dispersal and spatial pattern in tropical trees. – PLoS Biol. 4: 2132–2137. Swenson, N. G. 2009. Phylogenetic resolution and quantifying the phylogenetic diversity and dispersion of communities. – PLoS One 4: e4390. Wang, X. et al. 2010. Spatial distributions of species in an old-growth temperate forest, northeastern China. – Can. J. For. Res. 40: 1011–1019.
17 Guatteria sp4 A 4.7 Annona sericea F Cryptocarya mandioccana G 73.4 Ocotea catharinensis N 127.7 0 Ocotea catharinensis K Mollinedia schottiana D Mollinedia schottiana B 105.1 Mollinedia schottiana E 0 Mollinedia schottiana F Mollinedia schottiana G Mollinedia schottiana H 2.2 Mollinedia salicifolia K Mollinedia argyrogyna N 0 Mollinedia argyrogyna K Euterpe edulis B Euterpe edulis A Euterpe edulis D Euterpe edulis E Euterpe edulis G 135.9 0 Euterpe edulis H Euterpe edulis I Euterpe edulis J Euterpe edulis K Euterpe edulis N 112.7 Sorocea hilarii J Lonchocarpus cultratus F 115.8 Maytenus litoralis A Garcinia gardneriana I Garcinia gardneriana A 112 0 Garcinia gardneriana J
102.6 Hieronyma alchorneoides F Pera glabrata A Alchornea triplinervia F 99.8 74.2 0 118.6 Alchornea triplinervia A 135.8 0.4 96.8 Alchornea glandulosa F Licania hoehnei N 0 Licania hoehnei K Guarea macrophylla A Eugenia verticillata A Eugenia sp15 F 4.1 116.4 Eugenia prasina H 0 Eugenia prasina G Myrceugenia myrcioides F 23.8 Myrcia spectabilis K Myrcia spectabilis E 0 Myrcia spectabilis N 20.4 Myrcia racemosa A 5.7 Myrcia multiflora A
123.7 Myrcia brasiliensis A Marlierea tomentosa H 11.3 0 Marlierea tomentosa A Calyptranthes lucida N 0 Calyptranthes lucida K 2.8 Calyptranthes grandifolia J Guapira opposita D Guapira opposita B Guapira opposita E 0 Guapira opposita F Guapira opposita H Guapira opposita K Stylogyne lhotskyana H 0 Stylogyne lhotskyana G 22.1 Myrsine venosa A
119.9 Pouteria psammophila J 94.4 0.3 Pouteria caimito K Ecclinusa ramiflora J 7.9 0 Ecclinusa ramiflora I
6 Chrysophyllum viride K Chrysophyllum flexuosum D 1 Chrysophyllum flexuosum B 112.3 0 Chrysophyllum flexuosum E Chrysophyllum flexuosum H Schefflera angustissima A 88.2 Cordia taguahyensis H Jacaranda puberula A Bathysa mendoncaei D 106.7 Bathysa mendoncaei B Bathysa mendoncaei E 0 Bathysa mendoncaei H 89.8 Bathysa mendoncaei I 0.2 Bathysa mendoncaei J Bathysa australis H Bathysa australis F 0 Bathysa australis K
64.8 Rudgea jasminoides H Rudgea jasminoides G 0 Rudgea jasminoides I Faramea pachyantha J 0 Faramea pachyantha I 62 Coussarea meridionalis D Coussarea meridionalis B Coussarea meridionalis E 20.7 Coussarea meridionalis G 0 Coussarea meridionalis H Coussarea meridionalis I Coussarea meridionalis J 10.3 Coussarea accedens D Coussarea accedens B Coussarea accedens E 0 Coussarea accedens G Coussarea accedens H
140 120 100 80 60 40 20 0
Figure A1. Phylogenetic tree of 101 populations of 46 tree species sampled in fourteen 1-ha plots (letters A–N after species names) established at the Atlantic Rainforest in SE Brazil. The horizontal axis shows the time of diversification (in million years).
18 Phylogenetic tree (newick format) (((annona_sericea_F:4.693361,guatteria_sp4_A:4.693361):123.007782,((cryptocarya_mandioccana_G: 73.357468,(ocotea_catharinensis_K:0,ocotea_catharinensis_N:0)NA:73.357468):31.736769,((mollined ia_argyrogyna_K:0,mollinedia_argyrogyna_N:0)NA:2.200043,mollinedia_salicifolia_K:2.200043,(mol linedia_schottiana_H:0,mollinedia_schottiana_G:0,mollinedia_schottiana_F:0,mollinedia_schottiana_E :0,mollinedia_schottiana_B:0,mollinedia_schottiana_D:0)NA:2.200043)mollinedia:102.894241)node3 8:22.606918)node22:8.211001,(((((schefflera_angustissima_A:106.741402,((jacaranda_puberula_A:88 .247772,cordia_taguahyensis_H:88.247765)node501:1.503291,(((bathysa_australis_K:0,bathysa_austra lis_F:0,bathysa_australis_H:0)NA:0.16406,(bathysa_mendoncaei_J:0,bathysa_mendoncaei_I:0,bathysa _mendoncaei_H:0,bathysa_mendoncaei_E:0,bathysa_mendoncaei_B:0,bathysa_mendoncaei_D:0)NA: 0.164065)bathysa:64.653371,((((coussarea_accedens_H:0,coussarea_accedens_G:0,coussarea_acceden s_E:0,coussarea_accedens_B:0,coussarea_accedens_D:0)NA:10.346412,(coussarea_meridionalis_J:0,c oussarea_meridionalis_I:0,coussarea_meridionalis_H:0,coussarea_meridionalis_G:0,coussarea_meridi onalis_E:0,coussarea_meridionalis_B:0,coussarea_meridionalis_D:0)NA:10.346429)coussarea:10.3464 12,(faramea_pachyantha_I:0,faramea_pachyantha_J:0)NA:20.692823):41.307175,(rudgea_jasminoides _I:0,rudgea_jasminoides_G:0,rudgea_jasminoides_H:0)NA:62)rub8:2.817429)rubiaceae:24.933611)no de500:16.990341)node376:5.599318,((myrsine_venosa_A:22.099327,(stylogyne_lhotskyana_G:0,stylo gyne_lhotskyana_H:0)NA:22.099329):72.25885,((((chrysophyllum_flexuosum_H:0,chrysophyllum_fle xuosum_E:0,chrysophyllum_flexuosum_B:0,chrysophyllum_flexuosum_D:0)NA:0.992521,chrysophyl lum_viride_K:0.992525)chrysophyllum:4.962756,(ecclinusa_ramiflora_I:0,ecclinusa_ramiflora_J:0)N A:5.955277):1.985309,(pouteria_caimito_K:0.330876,pouteria_psammophila_J:0.330868)pouteria:7.6 0971):86.417587)node576:17.982525)node375:7.533504,(guapira_opposita_K:0,guapira_opposita_H: 0,guapira_opposita_F:0,guapira_opposita_E:0,guapira_opposita_B:0,guapira_opposita_D:0)NA:119.87 4222)node373:3.860003,(((maytenus_litoralis_A:112.020554,((((licania_hoehnei_K:0,licania_hoehnei _N:0)NA:96.750183,((alchornea_glandulosa_F:0.429871,(alchornea_triplinervia_A:0,alchornea_triplin ervia_F:0)NA:0.42987)alchornea:73.800888,pera_glabrata_A:74.230766)node111:22.519398)node90: 3.058376,hieronyma_alchorneoides_F:99.808533)node68:2.799194,(garcinia_gardneriana_J:0,garcinia _gardneriana_A:0,garcinia_gardneriana_I:0)NA:102.607719)node67:9.41283)node64:3.764979,(lonch ocarpus_cultratus_F:112.70118,sorocea_hilarii_J:112.70121)node221:3.084369)node63:2.793039,(gua rea_macrophylla_A:116.37809,((((calyptranthes_grandifolia_J:2.833625,(calyptranthes_lucida_K:0,cal yptranthes_lucida_N:0)NA:2.833642)calyptranthes:8.499727,(marlierea_tomentosa_A:0,marlierea_to mentosa_H:0)NA:11.333353,(myrcia_brasiliensis_A:5.666676,myrcia_multiflora_A:5.666698,myrcia_ racemosa_A:5.666698,(myrcia_spectabilis_N:0,myrcia_spectabilis_E:0,myrcia_spectabilis_K:0)NA:5. 19 666698)myrcia:5.666676)myrciagroup:9.065247,myrceugenia_myrcioides_F:20.398598):3.40003,((eu genia_prasina_G:0,eugenia_prasina_H:0)NA:4.132181,eugenia_sp15_F:4.132193,eugenia_verticillata _A:4.132193)eugenia:19.666447):92.579437)node282:2.200511)node61:5.155634)node58:12.023827, (euterpe_edulis_N:0,euterpe_edulis_K:0,euterpe_edulis_J:0,euterpe_edulis_I:0,euterpe_edulis_H:0,eut erpe_edulis_G:0,euterpe_edulis_E:0,euterpe_edulis_D:0,euterpe_edulis_A:0,euterpe_edulis_B:0)NA:1 35.758037)node51:0.154122)node5;
References used to build the phylogenetic tree, separated by family
Annonaceae Chaowasku et al. 2014. – Am. J. Bot. 101: 691–709. Chatrou et al. 2012. – Bot. J. Linn. Soc. 169: 5–40. Arecaceae Baker et al. 2009. – Syst. Biol. 58: 240–256. Baker et al. 2011. – Ann. Bot. 108: 1417–1432. Cuenca et al. 2008. – Mol. Phylogenet. Evol. 46: 760–775. Eiserhardt et al. 2011. – Taxon 60: 485–498. Meerow et al. 2009. – PLoS One 4: e7353. Montúfar and Pintaud. 2008. – Rev. Peru. Biol. 15 (supl. 1): 73–78. Roncal et al. 2011. – Biotropica 43: 324–334. Roncal et al. 2013. – Bot. J. Linn. Soc. 171: 120–139. Euphorbiaceae Riina et al. 2014. – Syst. Bot. 39: 227–234. Sierra et al. 2010. – Taxon 59: 101–116. Wurdack et al. 2005. – Am. J. Bot. 92: 1397–1420. Lauraceae Alves and Souza. 2013. – Taxon 62: 281–290. Assis. 2009. Sistemática e filosofia: filogenia do complexo Ocotea e revisão do grupo Ocotea indecora (Lauraceae). – PhD thesis, Univ. of São Paulo. Chanderbali et al. 2001. – Ann. Miss. Bot. Gard. 88: 104–134. Li et al. 2008. – Ann. Miss. Bot. Gard. 95: 580–599 Li et al. 2011. – Am. J. Bot. 98: 1–17. Nie et al. 2007. – Plant Syst. Evol. 267: 191–203.
20 Rohwer and Rudolph. 2005. – Ann. Miss. Bot. Gard. 92: 153–178. Monimiaceae Renner et al. 2010. – J. Biogeogr. 37: 1227–1238. Myrtaceae Biffin et al. 2010. – Ann. Bot. 106: 79–93. Lucas et al. 2007. – Taxon 56: 1105–1128. Primulaceae Yesson et al. 2009. – J. Biogeogr. 36: 1234–1252. Rubiaceae Barrabé et al. 2014. – Mol. Phylogenet. Evol. 71: 15–35. Kainulainen et al. 2010. – Am. J. Bot. 97: 1961–1981. Rydin et al. 2009. – Taxon 58: 793–810. Wikström et al. 2015. – PloS One 10: e0126690. Sapotaceae Gautier et al. 2013. – Taxon 62: 972–983. Kümpers et al. 2016. – Bot. J. Linn. Soc. 180: 161–192. Richardson et al. 2014. – Bot. J. Linn. Soc. 174: 130–140.
References used to obtain family age estimates, separated by family Arecaceae Cuenca et al. 2008. – Mol. Phylogenet. Evol. 46: 760–775. Couvreur et al. 2011. – BMC Biol. 9: 44. Eiserhardt et al. 2011. – Taxon 60: 485–498. Meerow et al. 2009. – PLoS One 4: e7353. Roncal et al. 2011. – Biotropica 43: 324–334. Lauraceae Nie et al. 2007. – Plant Syst. Evol. 267: 191–203. Monimiaceae Renner et al. 2010. – J. Biogeogr. 37: 1227–1238. Myrtaceae Biffin et al. 2010. – Ann. Bot. 106: 79–93. Primulaceae
21 Yesson et al. 2009. – J. Biogeogr. 36: 1234–1252. Rubiaceae Wikström et al. 2015. – PloS One 10: e0126690. Sapotaceae Richardson et al. 2014. – Bot. J. Linn. Soc. 174: 130–140.
22 Appendix 4 Detailed results of the analysis of the degree of overdispersion or aggregation of tree populations sampled in fourteen 1-ha plots established at the Atlantic Rainforest along the elevation gradient of the state park “Parque Estadual da Serra do Mar”, SE Brazil.
We show here the detailed results of the variation partitioning analysis of the standardized pair correlation function index err(r), which describes the strength of population overdispersion or aggregation, by three components (traits, phylogenetic relationships, and site variables). We performed the variation partitioning without population density in the site dataset (results in Table A1 and A2) as well as including this variable as a site characteristic (results in Table A3 and A4). To test the generality of the results obtained for err(r), we repeated the variation partitioning analysis for other two summary functions commonly used to describe population overdispersion or aggregation, the untransformed values of the pair correlation function g(r) (results in Table A6 and A7) and the untransformed values of the L-function L(r) (results in Table A8 and A9). The L-function derives from Ripley’s K-function K(r), which is the expected number of points within a circle of radius r centred at an arbitrary focal point, divided by the overall density λ of the point pattern at the study site (= number of individuals divided by area). Because the expected number of points increases with increasing r, K(r) is usually transformed into L(r) = [K(r)/π]0.5 – r to stabilize its variance (Besag 1977). The L-function differs from g(r) by its cumulative nature, and therefore K(r) and L(r) confound overdispersion or aggregation at different distances from the focal point (Wiegand and Moloney 2014). Our models explained a much higher proportion of the variation in err(r) compared to g(r) and L(r). This outlines the high power of our newly developed summary function err(r) in capturing the spatial distribution pattern of populations. We attribute this to two factors: first, by using the z-scores with respect to the null model of complete spatial randomness (CSR), we reduced spurious high or low values in g(r) caused by stochastic noise in small populations; and second, we assigned a value of err(r) = 0 to all populations with patterns indistinguishable from CSR, thereby removing additionally spurious non-significant variation in the z-scores.
23 Table A1. Variation partitioning of the degree of overdispersion or aggregation of tree populations of the Atlantic Rainforest in SE Brazil, as measured by the standardized pair correlation function index err(r), including variable sets representing traits (T), phylogenetic relationships (P), and site characteristics (S), without population density in the site dataset. The standardized index err(r) was calculated at distances of 1, 10, 20, 30 40 and 50 m from focal trees. All the significance tests were based on 9999 permutations and a significance level α = 0.05. Significant portions are in boldface type.
Whole T Whole P Whole S Pure T Pure P Pure S Residuals component component component fraction fraction fraction 2 R adj NA 0.768 0.072 NA 0.722 0.025 err(r = 1 m) 0.207 p NA < 0.001 0.007 NA < 0.001 0.003 2 R adj 0.076 0.702 0.116 0.004 0.529 0.026 err(r = 10 m) 0.273 p 0.018 < 0.001 < 0.001 0.133 < 0.001 0.011 2 R adj 0.062 0.899 NA -0.001 0.836 NA err(r = 20 m) 0.102 p 0.027 < 0.001 NA 0.472 < 0.001 NA 2 R adj 0.046 0.807 0.038 -0.002 0.727 0.001 err(r = 30 m) 0.194 p 0.037 < 0.001 0.044 0.931 < 0.001 0.267 2 R adj 0.078 0.716 0.086 -0.002 0.550 -0.003 err(r = 40 m) 0.289 p 0.009 < 0.001 0.013 0.506 < 0.001 0.595 2 R adj 0.102 0.954 0.072 < 0.001 0.784 < -0.001 err(r = 50 m) 0.046 p 0.008 < 0.001 0.048 0.227 < 0.001 0.663
Table A2. Number of phylogenetic eigenvectors, traits, and site variables included by forward selection as explanatory variables in variation partitioning analysis of the degree of overdispersion or aggregation of tree populations of the Atlantic Rainforest in SE Brazil, as measured by the standardized pair correlation function index err(r). Population density was excluded from the site dataset. The standardized index was calculated at distances of 1, 10, 20, 30, 40 and 50 m from focal trees. N.S. means no trait or site variable was significant.
Phylogenetic Traits Site variables eigenvectors err(r = 1 m) 20 N.S. Restinga No. individuals, no. err(r = 10 m) 11 Biotic seed dispersal multistemmed individuals err(r = 20 m) 22 Biotic seed dispersal N.S. err(r = 30 m) 20 Dust diaspores Plot I err(r = 40 m) 17 Biotic seed dispersal Plot I err(r = 50 m) 31 Biotic seed dispersal Plot I
24 Table A3. Variation partitioning of the degree of overdispersion or aggregation of tree populations of the Atlantic Rainforest in SE Brazil, as measured by the standardized pair correlation function index err(r), including variable sets representing traits (T), phylogenetic relationships (P), and site characteristics (S), with population density in the site dataset. The standardized index err(r) was calculated at distances of 1, 10, 20, 30, 40 and 50 m from focal trees. All the significance tests were based on 9999 permutations and a significance level α = 0.05. Significant portions are in boldface type.
Whole T Whole P Whole S Pure T Pure P Pure S Residuals component component component fraction fraction fraction 2 R adj NA 0.768 0.126 NA 0.665 0.023 err(r = 1 m) 0.209 p NA < 0.001 0.003 NA < 0.001 0.009 2 R adj 0.071 0.702 0.168 0.013 0.488 0.032 err(r = 10 m) 0.265 p 0.026 < 0.001 < 0.001 0.027 < 0.001 0.006 2 R adj 0.062 0.899 0.067 < -0.001 0.787 -0.001 err(r = 20 m) 0.103 p 0.033 < 0.001 0.014 0.484 < 0.001 0.621 2 R adj 0.027 0.807 0.103 0.001 0.693 < -0.001 err(r = 30 m) 0.192 p 0.048 < 0.001 < 0.001 0.196 < 0.001 0.339 2 R adj 0.078 0.716 0.086 -0.002 0.550 -0.003 err(r = 40 m) 0.289 p 0.009 < 0.001 0.013 0.506 < 0.001 0.595 2 R adj 0.102 0.954 0.085 0.001 0.794 < 0.001 err(r = 50 m) 0.046 p 0.010 < 0.001 0.008 0.183 < 0.001 0.287
Table A4. Number of phylogenetic eigenvectors, traits, and site variables included by forward selection as explanatory variables in variation partitioning analysis of the degree of overdispersion or aggregation of tree populations of the Atlantic Rainforest in SE Brazil, as measured by the standardized pair correlation function index err(r). Population density was included in the site dataset. The standardized index was calculated at distances of 1, 10, 20, 30, 40 and 50 m from focal trees. N.S. means no trait or site variable was significant. Phylogenetic Traits Site variables eigenvectors err(r = 1 m) 20 N.S. Restinga, population density No. individuals, no. err(r = 10 m) 11 Biotic seed dispersal multistemmed individuals, population density err(r = 20 m) 22 Biotic seed dispersal Population density err(r = 30 m) 20 Dust diaspores Population density err(r = 40 m) 17 Biotic seed dispersal Plot I err(r = 50 m) 31 Biotic seed dispersal Population density
25 Table A5. Simple linear regressions between local population density and the degree of overdispersion or aggregation of tree populations of the Atlantic Rainforest in SE Brazil, as measured by the standardized pair correlation function index err(r) at different distances from focal trees.
Distance R2 b p 1 m 0.087 0.020 0.003 10 m 0.068 0.025 0.008 20 m 0.058 0.015 0.015 30 m 0.092 0.009 0.002 50 m 0.079 -0.008 0.004
Table A6. Variation partitioning of the degree of overdispersion or aggregation of tree populations of the Atlantic Rainforest in SE Brazil, as measured by the pair correlation function g(r), including variable sets representing traits (T), phylogenetic relationships (P), and site characteristics (S), with population density in the site dataset. The pair correlation function g(r) was calculated at distances of 1, 10, 20, 30, 40 and 50 m from focal trees. All the significance tests were based on 9999 permutations and a significance level α = 0.05. Significant portions are in boldface type.
Whole T Whole P Whole S Pure T Pure P Pure S Residuals component component component fraction fraction fraction 2 R adj 0.026 0.656 0.205 0.002 0.469 0.015 g(r = 1 m) 0.330 p 0.030 0.003 0.001 0.210 0.001 0.047 2 R adj 0.026 0.330 0.241 0.024 0.171 0.053 g(r = 10 m) 0.575 p 0.048 0.013 0.001 0.037 0.024 0.008 2 R adj NA 0.508 0.175 NA 0.365 0.032 g(r = 20 m) 0.459 p NA 0.001 0.001 NA 0.001 0.024 2 R adj NA 0.275 NA NA NA NA g(r = 30 m) 0.725 p NA 0.001 NA NA NA NA 2 R adj NA 0.269 0.113 NA 0.203 0.046 g(r = 40 m) 0.684 p NA 0.001 0.003 NA 0.001 0.037 2 R adj NA 0.235 NA NA NA NA g(r = 50 m) 0.765 p NA 0.001 NA NA NA NA
26 Table A7. Number of phylogenetic eigenvectors, traits, and site variables included by forward selection as explanatory variables in variation partitioning analysis of the degree of overdispersion or aggregation of tree populations of the Atlantic Rainforest in SE Brazil, as measured by the pair correlation function g(r). Population density was included in the site dataset. The pair correlation function was calculated at distances of 1, 10, 20, 30, 40 and 50 m from focal trees. N.S. means no trait or site variable was significant. Phylogenetic Traits Site variables eigenvectors g(r = 1 m) 10 Wood density Restinga, population density g(r = 10 m) 7 Dioecy Restinga, average elevation, population density g(r = 20 m) 16 N.S. Restinga, average elevation g(r = 30 m) 8 N.S. N.S. g(r = 40 m) 7 N.S. Plot B, plot J, no individuals g(r = 50 m) 6 N.S. N.S.
Table A8. Variation partitioning of the degree of overdispersion or aggregation of tree populations of the Atlantic Rainforest in SE Brazil, as measured by the L-function L(r), including variable sets representing traits (T), phylogenetic relationships (P), and site characteristics (S), with population density in the site dataset. The L-function L(r) was calculated at distances of 1 m, 10 m, 20 m, 30 m, 40 m, and 50 m from focal trees. All the significance tests were based on 9999 permutations and a significance level α = 0.05. Significant portions are in boldface type.
Whole T Whole P Whole S Pure T Pure P Pure S Residuals component component component fraction fraction fraction 2 R adj NA 0.489 0.127 NA 0.385 0.023 L(r = 1 m) 0.488 p NA < 0.001 0.002 NA < 0.001 0.036 2 R adj 0.029 0.465 0.202 -0.004 0.294 0.029 L(r = 10 m) 0.511 p 0.037 < 0.001 < 0.001 0.560 < 0.001 0.046 2 R adj 0.016 0.387 0.142 -0.003 0.240 < -0.001 L(r = 20 m) 0.615 p 0.110 < 0.001 < 0.001 0.480 0.003 0.370 2 R adj 0.027 0.541 0.200 -0.006 0.333 -0.010 L(r = 30 m) 0.474 p 0.046 < 0.001 < 0.001 0.910 < 0.001 0.710 2 R adj 0.031 0.466 0.165 -0.005 0.308 0.006 L(r = 40 m) 0.534 p 0.040 < 0.001 < 0.001 0.630 < 0.001 0.270 2 R adj NA 0.430 0.148 NA 0.302 0.019 L(r = 50 m) 0.551 p NA < 0.001 0.002 NA < 0.001 0.127
27 Table A9. Number of phylogenetic eigenvectors, traits, and site variables included by forward selection as explanatory variables in variation partitioning analysis of the degree of overdispersion or aggregation of tree populations of the Atlantic Rainforest in SE Brazil, as measured by the L-function L(r). Population density was included in the site dataset. The L-function L(r) was calculated at distances of 1, 10, 20, 30, 40 and 50 m from focal trees. N.S. means no trait or site variable was significant.
Phylogenetic Traits Site variables eigenvectors L(r = 1 m) 12 N.S. Montane, no multistemmed individuals L(r = 10 m) 10 Wood density Lowland, Submontane, population density L(r = 20 m) 10 Monoecy Lowland, Submontane L(r = 30 m) 16 Monoecy Lowland, Submontane, population density L(r = 40 m) 12 Monoecy Lowland, Submontane, population density L(r = 50 m) 12 N.S. Lowland, Submontane, population density
References Besag, J. 1977. Contribution to the discussion of Dr. Ripley’s paper. – J. R. Stat. Soc. B Met. 39: 193– 195. Wiegand, T. and Moloney, K. A. 2014. A handbook of spatial point pattern analysis in ecology. – Chapman and Hall/CRC Press.
28