The phylogeographic structure of tropical plant communities and populations – insights on environmental filtering, dispersal limitation and biogeographic barriers
Olivier Hardy
Université Libre de Bruxelles Olivier Hardy
Lab: Evolutionary Biology and Ecology (http://ebe.ulb.ac.be/ebe)
Scientific background: population genetics Research: biodiversity of African plants (trees in particular), including: - species delimitation => impact on biodiversity/endemism - gene flow + seed and pollen dispersal processes => impact of logging, pollinators/dispersers - genetic structure => insights on populations' history, past climate changes - phylogeny => evolution of traits or niches - plant communities => phylogenetic structure & species assembly rules - invasive species (Cecropia in Africa) => evolution + ecology + eco. services - some modelling (simulations) => link between processes and patterns
Phylogeographic structure
1. Inter-specific scale (communities)
Phylogenetic tree for 4 species
No species Species turnover Species turnover Species turnover turnover but + + & No phylogenetic Phylogenetic Phylogenetic No phylogenetic turnover clustering overdispersion turnover (attraction) (eveness, repulsion)
Phylogeographic structure
2. Intra-specific scale (populations)
Phylogenetic tree for 4 alleles
No genetic Genetic Genetic Genetic differentiation differentiation differentiation differentiation & but + + No phylogeographic No phylogeographic Phylogeographic "negative" structure structure structure phylogeographic structure => unexpected The phylogenetic structure of plant assemblages in tropical Africa: from local community to biogeographical scales
Olivier Hardy, Gilles Dauby, Thomas Couvreur, Vincent Droissart, Maxime Rejou-Mechain, Steven Janssens, William Hawthorne, Cicely Marshall, Olivier Maurin, Michelle van der Bank, Duncan W Thomas, David Kenfack, George B Chuyong, Denis Beina, Fidele Baya
Local (50 ha plot) Continental (bioregions) Inter-continental Phylogenetic turnover caused by ecological & evolutionary processes
among habitat environmental filtering Spatial adaptation to ≠ elevation phylogenetic + clustering habitat conservatism
among habitat environmental filtering adaptation to ≠ elevation Spatial + phylogenetic habitat convergence overdispersion e.g. radiation of several clades (= evenness = repulsion) in a set of new habitats
within habitat competitive exclusion Spatial + phylogenetic niche conservatism overdispersion sister species cannot coexist (= evenness = repulsion) because of niche overlap Phylogenetic turnover caused by biogeographic barrier
Scenario A: Scenario B: 1. barrier (allopatric speciation) 1. sympatric speciation 2. sympatric speciations 2. barrier (allopatric speciations) phylogeny
1. 1.
time
2. 2.
phylogenetic phylogenetic clustering overdispersion Integration communities – traits – ecology – biogeography – evolution x
Phylogeny Evolution
Traits (niche): x y x x y z z species: 1 2 3 4 5 6 7
Biogeography Historical dispersal 1 4 2 3 2 4 5 6 7 region 1 region 2
Ecology Niche structure in region 2 : 3 4 2 5 6 7 niche X niche Y niche Z No dispersal Ecological filtering or competitive exclusion Recent dispersal 4 2 5
1 community Competition (habitat OK for X+Y)
Integrating phylogeny into community studies evolutionary perspective testing community neutrality How to quantify phylogenetic turnover from species incidence data ? many methods developed; here just one presented 1. From a phylogeny phyletic distances between species
divergence time 0 1 2 2 (molecular phylogeny) dkl 1 0 2 2 2 2 0 1 (taxonomic level) 2 2 1 0
2. From community inventories mean dkl between species
within sites MPDw d(2_ species_ from_ same_ site) sampled among sites MPDa d(2_ species_ from_ different_ sites)
relative excess of relatedness between ST 1 MPDw / MPDa species co-occurring within a site (relative to species pairs sampled in distinct sites) Hardy & Senterre 2007 (J. Ecol. 95: 493–506) -> phylogenetic turnover Hardy & Jost 2008 (J. Ecol. 96: 849–852) How to test phylogenetic turnover?
no phylogenetic Phylogenetic Phylogenetic structure clustering overdispersion
ST = 0 ST > 0 ST < 0
Testing ST <> 0 randomize phylogeny
assess distribution of ST for 1000 permutations
P-value = % of ST after permutation > or < ST observed More elaborated randomization tests: Hardy 2008 (J Ecol 96: 914–926) 1. Local scale
Korup CTFS: Cameroon Community inventory 500x1000m2 c. 330 000 stems >1cm dbh c. 450 tree spp
Habitat characterization -> Chuyong et al. 2011, Plant Ecology
ULB: Phylogenetic tree (rbcL + matK) for 270 species (95% of stems)
Barcode efficiency -> Parmentier et al. 2013, PLoS ONE
Phylogenetic structure: -> Parmentier et al. 2014, J Ecol 1250 local communities (20 x 20m)
=> compute ST for each pair Phylogenetic turnover between sub-communitites: impact of spatial distance vs topographic habitat types
ST Korup - Cameroon 0.0020 among habitats within habitats Habitat categories explain part but not all the 0.0015 phylogenetic turnover
0.0010 Phylogenetic clustering
(ST>0) occurs at all scales (no overdispersion) 0.0005
0.0000 16 32 64 128 256 512 1024 spatial distance (log scale, meters) Parmentier et al. 2014, J Ecol Does phylogenetic overdispersion occur at smaller scales? individual-based analyses
Computing a ‘co-occurrence excess’ of hetero-specific pairs
eA,i =1 if ind i belongs to species A, otherwise 0
A B wi,j,c =1 if distance i-j belongs to interval c, otherwise 0 B B B B D B A B A C D D C B B B D c B C B A C B A D C frequency of B at frequency of B in the B C A A B D D distance c from A plot (excluding A) C A D A (excluding A) Phylogenetic structure at the individual level: Are related species aggregating or segregating in space ?
0.025
Cola cauliflora
0.020
0.015
specific specific pairs -
0.010 Beilschmiedia jacquesfelixii
0.005
0.000
-0.005 Rinorea gabunensis
-0.010
-0.015
0.5 5 50 500 Standardized proportion of hetero of proportion Standardized Spatial distance (meters, log scale) Average trends over all pairs of species according to their divergence time
Divergence between
species pairs
0.0010
<2Ma Trend of aggregation between related 0.0008 <10Ma
species
specific specific pairs - <50Ma (no overdispersion) 0.0006 >100Ma
0.0004
et al. 2014, J. Ecol. 2014, al. et 0.0002
0.0000
Parmentier (averages according to divergence time) divergence to according (averages - 0.0002 Slight repulsion between distant species Standardized proportion of hetero of proportion Standardized 0.5 5 50 500 Spatial distance (meters, log scale) Korup - Cameroon Repulsion between related species at <1m
Divergence between
species pairs
0.0010
<2Ma Trend of aggregation between related 0.0008 <10Ma
species
specific specific pairs - <50Ma (no overdispersion) 0.0006 >100Ma
0.0004
et al. 2014, J. Ecol. 2014, al. et 0.0002
0.0000
Parmentier (averages according to divergence time) divergence to according (averages - 0.0002 Slight repulsion between distant species Standardized proportion of hetero of proportion Standardized 0.5 5 50 500 Spatial distance (meters, log scale) 2. Continental scale
Plant community data => bioregions Phylogenetic tree RAINBIO project (T Couvreur et al.) S Janssens, unpublished Tropical Africa
Sample size per grid cell rbcL+matK phylogeny of world angiosperms (c. 36 000 ‘species’) c. 4000 species from Africa (90% of RAINBIO genera represented)
Missing species from genera represented added ‘randomly’ to existing congeneric species phylogenetic tree of 19500 African species (S Janssens, unpublished) > 600,000 georeferenced records > 24,000 species Dauby et al. 2016, PhytoKeys 74: 1–18 Identifying bioregions for tropical Africa from RAINBIO database
http://bioregions.mapequation.org/
Bipartite network with species and localities as nodes => clustering => bioregions => 16 recognized bioregions
Each bioregion contains >5% endemic species
= Transition areas (<5% endemics)
Droissart et al., J Biog (2018) Floristic vs lineage turnover among bioregions?
Floristic
2 3 4
1
Bioregion Bioregion Bioregion Bioregion
Sp. 1 1 2 6 8 Sp. 2 0 5 0 3 Sp. 3 1 0 10 4 Pairwise measure of turnover: Sp. 4 0 9 0 7 Morisita-Horn index Sp. 5 6 2 7 0 Sp. 6 7 8 1 2 Floristic vs lineage turnover among bioregions?
Age threshold Phylogenetic
Floristic
2 3 4
1
4 2 3
1
Bioregion Bioregion Bioregion
Bioregion
Sp. 1 1 2 6 8
0 5 0 3
Bioregion Bioregion
Sp. 2 Bioregion Bioregion Sp. 3 1 0 10 4 Pairwise measure of turnover: Lineage 1 1 0 10 4 Sp. 4 0 9 0 7 Morisita-Horn index Lineage 2 0 9 0 7 6 0 Sp. 5 2 7 Lineage 3 6 2 7 0 Sp. 6 7 8 1 2 Lineage 4 7 8 1 2 Question: Floristic and lineage turnover between bioegions
Using Upper Guinea as a focal bioregion, compare similarity with:
- spatially adjacent vs disjoint bioregions
- ecologically similar vs contrasted bioregions Floristic similarity between Upper Guinea bioregion and 4 other bioregions
Morisita-Horn similarity index based
on species relative abundances
1
0.9
Horn) - 0.8
0.7 Morisita
( 0.6 Lower Guinean forests 0.5 Adjacent bioregions share many species 0.4 Guineo-Soudanian savannas
0.3 similarity
East African Coastal forests 0.2 Highly disjoint bioregions have contrasted relative Angola 0.1 species abundances
Species 0
Lineage similarity between Upper Guinea bioregion and 4 other bioregions
1
0.9
0.8
Horn) - 0.7
0.6
Morisita
(
0.5 Ecologically similar bioregions share similar 0.4 abundances of old lineages
similarity 0.3
0.2 Ecologically contrasted bioregions differ in their relative
0.1 abundances of old lineages Lineage 0 0 10 20 30 40 50 lineages age threshold (Ma) Floristic and lineage similarity between all pairs of bioregions
0.5 y = -4E-05x + 0.25 0.5 y = -0.048 x + 0.28 R² = 0.26 R² = 0.44 0.4 y = -0.10 ln(x) + 0.93 0.4 y = -0.124 ln(x) + 0.26 R² = 0.34 R² = 0.43
level) 0.3 0.3 -
0.2 0.2
(species 0.1 0.1 Floristic similarity Floristic similarity 0 0 0 2000 4000 6000 0 1 2 3 4 5 6 7 8
1 y = -4E-05x + 0.59 1
0.9 R² = 0.06 0.9 0.8 y = -0.09 ln(x) + 1.20 0.8 R² = 0.5688
0.7 0.7 ) R² = 0.10 0.6 0.6 0.5 0.5
0.4 0.4 t=20 t=20 Ma ( 0.3 0.3
0.2 0.2 Lineage similarity Lineage similarity 0.1 0.1 0 0 0 2000 4000 6000 0 1 2 3 4 5 6 7 8 Spatial distance (km) Ecological distance (Euclid dist - BIO 2, 5, 7, 12, 18 ) Correlation between lineage similarity and ecological/spatial distance according to phylogenetic depth
0.8
0.7
0.6 ecological dist (5 BIO var) BIO2 Mean diurnal range 0.5 BIO5 Max temperature of the warmest month BIO7 Annual temperature range 0.4 BIO12 Annual rainfall BIO18 Rainfall of the warmest quarter 0.3 Dispersal limitation 0.2 spatial dist (log) effect disappears
across bioregions 16 across 0.1 after ≈8Ma ecological/spatial distances distances ecological/spatial
Correlation between lineage lineage betweenCorrelation 0 and and 0 10 20 30 40 50 Time threshold to define lineages (Ma)
16000 8000 4000 2000 1000 500 # lineages # 0 10 20 30 40 50 Correlation between lineage similarity and ecological/spatial distance according to phylogenetic depth
0.8 Divergence between magnoliids, eudicots,
0.7 monocots
0.6 BIO2 Mean diurnal range 0.5 BIO5 Max temperature of the warmest month BIO7 Annual temperature range BIO12 Annual rainfall 0.4 BIO18 Rainfall of the warmest quarterecological dist (5 BIO var) 0.3 Dispersal limitation effect disappears
across across bioregions 16 0.2 after ≈8Ma
Correlation Correlation between lineage 0.1 and and ecological/spatial distances spatial dist 0 (log) 0 50 100 150 200 Time threshold to define lineages (Ma)
16384 4096 1024 256 64
lineages 16
# # 4 1 0 50 100 150 200 Inter-continental scale: phylogenetic turnover across 1ha (semi-)evergreen forest plots (trees DBH>10cm)
O Hardy, P Couteron, F Munoz, BR Ramesh, R Pélissier
Panama canal Western Ghats
watershed 50 plots of 1ha 50 plots of 1ha
CTFS Area 50km x 60km Area 50km x 200km
data data Rainfall: Rainfall: French Institute French Institute 1500-3300mm Pondicherry 1400-6000mm
Elevation : data Elevation: 50-400m 50-1050m
Hardy et al. 2012 (Glob. Ecol. Biogeogr. 21:1007–1016) Phylogenetic turnover at regional scale
ST Panama canal ST Western Ghats 0.010 0.020
0.015
0.005 0.010
0.005 0.000
0.000
-0.005 -0.005 0 500 1000 1500 0 1000 2000 3000 4000 5000 D rainfall btw. plots D rainfall btw. plots
Phylogenetic clustering within plots mostly explained by rainfall differences Phylogenetic turnover impact of ecological gradients vs biogeography
Biogeographic level mean pairwise ST within within between Panama W. Ghats Panama & W. Ghats
Weak 0.001 0.003 0.010 (<500mm)
Strong
rainfall 0.008 0.011 0.017 (>1000 D D or >3500mm)
For forest tree communities, at the scale of angiosperms:
biogeography (inter-continental) Impacts on of are comparable ST meso-climatic gradient + seem cumulative Panama India Panama canal Western Ghats watershed 50 plots of 1ha 50 plots of 1ha Area 50x60 km2 Area 50x200 km2
Rainfall: Rainfall: 1500-3300mm 1400-6000mm data CTFS data French Institute No shared species but Pondicherry many shared clades
Independent reconstructions of ancestral climatic niche along the phylogenies Inter-continental correlation of clade ‘adaptations’ with respect to rainfall gradient
0.3 Wet forest adapted clades
Celastraceae Myristicaceae
0.2 Clusiaceae Mg Sapotaceae Boraginaceae Garcinia Calophyllum Mp Olacaceae Annonaceae Beilschmiedia Lauraceae Theaceae AID 0.1 MID Eugenia ER1 Euphorbiaceae Moraceae RID Ro Palmae Myrtaceae Ulmaceae Casearia My Apocynaceae ER2 Bignoniaceae Rhamnaceae Malvaceae Rubiaceae Achariaceae Fabaceae 0 Ficus Meliaceae -0.3 -0.2 -0.1 0 0.1 0.2 Sapindaceae Verbenaceae R = 0.62 Rutaceae
Combretaceae Zanthoxylum
Node trait values in Western Ghats Western invalues trait Node -0.1 Terminalia Anacardiaceae y =Dry 0.6546x forest + 0.0581 adapted clades R² = 0.3758
-0.2 Node trait values in Panama Canal -> (pre)adaptations can persist over long evolutionary time (ca. 100 Myrs) Hardy et al. 2012 (Glob. Ecol. Biogeogr.) Conclusions at community scale
Local scale - phylogenetic clustering due to topographic habitats - no phylogenetic overdispersion due to competitive exclusion between related species, except may be at very short distance (<1m)
Continental scale species turnover = f(isolation) + f(ecological distance) lineage turnover = f(ecological distance)
Inter-continental scale - deep phylogenetic conservatism of climate adaptations
Phylogeographic structures within African tree species
Olivier Hardy, Jeremy Migliore, Boris Demenou, Dario Ojeda, Sandra Cervantes, Arthur Boom
Phytogeographic subdivision - Centres of endemism (after White 1979)
Upper Guinea Lower Guinea
Congolia
gap Dahomey Dahomey
Guineo-Congolian rainforest
(from Leal, 2004) # species (# endemic species) of Caesalpinioideae Greenwayodendron (Annonaceae) => long-living tree species of mature rain forest
Phylogeny of Greenwayodendron chloroplast gemomes
suaveolens Migliore et al. 2018, J Biog G.
G. gabonicum G. suaveolens (CVL) G. usambaricum G. oliveri
Phylogeny of G. suaveolens chloroplast gemomes
Within species => clear phylogeographic structure:
different clades distributed in parapatry (+ some overlap)
When did they diverge?
Migliore et al. 2018, J Biog
Phylogeny of G. suaveolens chloroplast gemomes
Within species => clear phylogeographic structure:
different clades distributed in parapatry (+ some overlap)
When did they diverge?
mostly 3.5 – 4.5 Myrs ago, but up to 9 Myrs ago for the Cameroon Volcanic Line ! Migliore et al. 2018, J Biog Phylogeography of Musanga cecropioides (Urticaceae)
=> short-living pioneer species
Mus UG Mus W-CVL Mus C-LG Phylogeny chloroplast gemomes Mus CONG
Mus SE-LG
Divergence between parapatric clades : 1-2 M yrs
Migliore, unpublished Anthonotha macrophylla Distemonanthus benthamianus short-distance seed dispersal (?) long-distance seed dispersal by wind
H14 2 H15 H35 4 H22 2 H29 4 H27 H1 H13 H21 H29 H5 4 4 7 H6 H7 10 11 5 7 H18 7 2 H28 H41 39 H19 3 H5 H8 2 H33 3 H14 3 H43 3 5 H15 5 27 H23 H10 6 H31 6 6 2 H26 2 H24 2 H25 5 5 7 6 H6 6 H28 28 H2 6 H23 H4 2 26 H20 3 4 H16 2 106 5 3 5 H26 H18 11 10 3 H42 H38 H36 H16 3 2 25 3 2 H 10 2 3 H7 54 H4 3 6 8 3 36 20 H2 11 23 H30 12 20 H24 3 12 26 8 9 2 2 H20 2 2 H9 6 2 H37 H8 17 H17 4 H21 13 H3 H13 H17 15 H1 H32 4 22 H39 H22 H25 20 2 5 H34 2 17
24 H30 2 H40 7 H12 10 2 3 5 H27 H12 H19 H10 H11 H9 H11
again, phylogeographic structure dominated by parapatric clades very similar structures in 2 species (except within Lower Guinea) shared colonisation history for the two species ?
B. Demenou, PhD thesis Anthonotha macrophylla Distemonanthus benthamianus
700 ka 40 ka
6 Ma 0.4 Ma Similar patterns but asynchronous Divergence between West Africa and Central Africa => distinct histories But is molecular dating reliable ? Anthonotha macrophylla Anthonotha +sister clades plastid phylogeny nuclear phylogeny
UG and LG diverged 6.1 Ma Anthonotha spp diverged 1.9 Ma UG and LG pop of A macrophylla diverged 1.2 Ma
Deep incongruence between plastid and nuclear genomes => errors or biology? Anthonotha "macrophylla" 6 samples were plastid phylogeny misidentified:
A. acuminata
Berlinia sp A. sp but phylogeographic pattern remained A. acuminata perfectly coherent
A. wijcampaensis
A. crassifolia
corrected species identification based on 300 nuclear genes
Chloroplasts ban be easily transfered between species of the "Berlinia clade"
=> a species could expand its range through pollen dispersal by recurrent back-crossing with related species Changing forest cover during Quaternary (after Dupont et al. 2000)
reconstruction based on pollen from marine sediments
Africanhumid
Holocene
period
LastGlacial
Maximum
(LGM)
If gene flow was extensive => genetic homogeneization during interglacials => divergence between pop dated to LGM (20 kyrs)
But phylogeographic patterns indicate much more ancient divergence (up to Miocene) => very slow population dynamics of (some) forest tree species Conclusions at intra-specific (intra-generic) scale African tree from the Guineo-Congolian region - phylogeographic patterns characterized by parapatric clades (spatial phylogeographic clustering)
- shared phylogeographic patterns occur between species: - differentiation between West Africa, Lower Guinea, Congolia - colonization of Dahomey Gap from the Cameroon Volcanic Line - ancient lineages in the CVL => conservation
- shared patterns are not necessarily contemporaneous => different histories but history repeats itself?
- divergence between populations (lineages) usually much older than Last Glacial Maximum (Plio-Pleistocene, Miocene) => slow population dynamics Phylogeographic structure
Inter-specific scale (communities) => dominated by niche conservatism
Intra-specific scale (populations) => dominated by historical contingency / biogeography
phylogenetic clustering >> phylogenetic overdispersion
Thank you for your attention