Individualistic Evolutionary Responses of Central African Rain Forest Plants J

Downloaded by guest on October 3, 2021 n ln raia neiiypten 8 3,TF r sug- are TRFs 13), (8, patterns endemicity animal or plant (12) reconstructions and paleovegetation climate on the Based when unfa- 11). them (10, during from improves out refugia expand then into and contract periods scales, vorable Species climatic time during 11). persist evolutionary (10, to species fluctuations on a defined of operate are populations refugia viable that hypothe- with hypothesis, allowing areas former refugia forest the geographic (“no rain In 7). as height lowland (4, per- and open H2) but or as density contract, sis,” way canopy not diffuse decreased did more a or a con- (9)] in either H1 sisted TRFs or [hereafter 1). 8) hypothesis,” 7, (3, (Table “refugia refugia forest particular discrete into in significantly tracted Africa, Guinea Central Lower in fluctuations and climatic Pleistocene to responses (3–7). debated highly (LGM; remains maximum Ka), glacial 12 last to the 24 approximately during years Guinean particular million glacial–interglacial 0.01 in Lower to [Ma]), to (2.58 How ago Pleistocene responded the 2). of (TRFs) (1, cycles Milankovitch forests endemics are rain which tropical of quarter a L forests rain tropical TRFs conserve better more to and ecology challenging, species be worldwide. on will needed change is climate emphasis under TRFs dynam- future of the ics Predicting have diversity. genetic fluctuations in climatic patterns shaped Pleistocene individualistic to to Instead, responses simplistic. model too evolutionary refugia is including strict dynamics a TRF factors, past Using stability. explain environmental varia- habitat disparate and genetic soil, to in climate, linked patterns were Furthermore, tion expansion Demo- growth. and concordant. continuous contraction Pleis- repeated rarely to from the were ranged during trends events graphic primarily divergence emerged but and tocene, species com- structure among a genetic varied of within Levels histories framework. inferred phylogeographic demographic We parative Africa. and Central evolutionary generated in species’ Guinea we Lower from this, plants test tributed To for require- data characteristics. sequence could environmental genome-wide ecological species distinct or TRFs, change. of ments of climate because past diversity differently to unrivalled respond response the common given and refu- a However, into This implies fluctuations. TRFs climatic model of Here, Pleistocene gia contraction during Africa. dic- refugia and Central be of expansion to in out thought repeated found are by is structure tated genetic world of the patterns tropical in present-day continuous (TRF) of forest expanse rain second-largest dissim- The in comparable patterns. resulting to ilar individualistically, leading or climate change, dynamics, region climate of evolutionary a to face in similarly Species the respond biomes. may in hyper-diverse diversity in conservation particularly genetic 2020) species change, 27, January for of review for fundamental dynamics (received 2020 is evolutionary 21, October the approved and Norway, Understanding Oslo, Oslo, of University Stenseth, Chr. Nils by Edited Sup Normale Ecole Biologiques, a Helmstetter J. Andrew climatic Pleistocene fluctuations to plants forest rain African Central of responses evolutionary Individualistic www.pnas.org/cgi/doi/10.1073/pnas.2001018117 R,DAE nvMnplir 49 otele,Fac;and France; Montpellier, 34394 Montpellier, Univ DIADE, IRD, w anhptee aebe rpsdt xli TRF explain to proposed been have hypotheses main Two fCnrlArc,wt bu ,0 aclrpatspecies, plant vascular region 7,000 floristic about biodiverse with most Africa, the Central is of 1) (Fig. Guinea ower | lmt change climate a,1,2 rer,Universit erieure, ´ ei B Kevin , | 0 niiul fsvncodis- seven of individuals >700 phylogenetics ethune ´ eYaound de e ´ | demography a acseG Kamdem G. Narcisse , ,BP 4,Yaound 047, B.P. I, e ´ b aoaor eBtnqeSyst Botanique de Laboratoire R ln aiis rcca pls(0;treseis and Gabon, dominant species) three ecologically (Cameroon, (20); two [palms Guinea from Arecaceae Lower Congo) families: plant of in TRF Republic species, the liana and shrubs, a trees, including and plants, forest rain lowland codistributed, a (7). under concordance Guinea species Lower the within plant test framework to multiple comparative out common among set histories has evolutionary study (H2) of refugia no plant no 15), the African (7, over (H1) Central hypothesis refugia refugia” the single) favor species-dependent (usually generally species phy- “relaxed of or genetic studies the population logeographic Although as 1). Table this (H1b; 16–18). hypothesis to (7, refer tolerances adapta- environmental We different and/or of phenotypes, could because tions, species individualistically hypothesis, species- responded refugia details). “strict have for the the 1 under as Table even this see Alternatively, to (H1a; refer hypothesis ecological We refugia” of independent (7). independent traits life-history events, or climatic Pleistocene shared 1). (Fig. that 15) 6, Lower mountainous (3, In in coast located Atlantic (3). the be areas to to refugial close thought regions discrete are several refugia into or most Africa Central Guinea, 14) in (6, 90% as single much a as contracted have to gested doi:10.1073/pnas.2001018117/-/DCSupplemental at online information supporting contains article This 2 1 the under Published Submission.y Direct PNAS a is article This interest.y competing no declare paper.y authors the The wrote T.L.P.C. and A.J.H. and K.B., data; A.J.H., analyzed T.L.P.C. tools; and and reagents/analytic N.G.K., new K.B., contributed T.L.P.C. A.J.H., research; research; performed designed T.L.P.C. T.L.P.C. and B.S., A.J.H., contributions: Author n nlsso idvriy ntttBuso etad 40 otele,France.y Montpellier, 34000 Bertrand, Bouisson Institut Biodiversity, of Analysis Biodiversit and la sur Recherche la pour Fondation address: [email protected] y Email: addressed. be may correspondence whom To ,Cameroon e, idvriyi eea,adi eta fiai particular. forest in Africa rain Central tropical in and of general, conservation impor- in similar has biodiversity the study to for Our led implications region. this tant change within climate dynamics forest past rain that the view challenge results long-standing Our traits. life-history and codis- factors environmental different of by driven range responses indi- evolutionary asynchronous, a vidualistic variable, reveal of dataset We dynamics species. genomic plant evolutionary tributed unparalleled the an Cen- estimate of generated to forests We rain tropical Africa. biodiverse tral or highly similar the either hypoth- within exhibit this esis tested could We area responses. conserva- evolutionary same individualistic biodiversity the in improved Species for tion. fundamental is climate change past to responses evolutionary species’ Understanding Significance ´ ee eifre h vltoayhsoiso ait of variety a of histories evolutionary the inferred we Here, implies 9) (3, hypothesis refugia the of interpretation strict A b all oaetr Sonk Bonaventure , pce epne nasmlreouinr a to way evolutionary similar a in responded species mtqee ’clge D d’Ecologie, et ematique ´ NSlicense.y PNAS e ´ b n hmsL .Couvreur P. L. Thomas and , preetdsSciences des epartement ´ . y https://www.pnas.org/lookup/suppl/ NSLts Articles Latest PNAS –etefrteSynthesis the for e–Centre ´ | f10 of 1 a

EVOLUTION Stability Is Linked to Genetic Structure. We recovered evidence for significant geographic structure in all species (SI Appendix, Fig. S1). Our genetic clustering analyses using Discriminant Analysis of Principal Components (DAPC) (28) and TESS3 (29) revealed that intraspecific diversity was highly structured, with four species exhibiting four or more distinct genetic clusters (Fig. 2 and SI Appendix, Figs. S2–S4). The topology of individual-level phylogenetic trees built using the entire dataset revealed groupings of individuals that closely matched those from clustering approaches (Fig. 2 and SI Appendix, Figs. S5–S11). To test the impact of stability on genetic structure across species, we used three main definitions of “stable areas.” The first consisted of the traditional refugia proposed by Maley (3) or Anhuf et al. (6) (Fig. 1) that are common to all species. For the two others, we estimated stable areas based on regional climate and species-specific habitat (30): 1) We identified climatically stable areas at the regional level by calculating differences in temperature and rainfall between the LGM and present (31); and 2) we inferred stable habitats between the LGM and present Fig. 1. Major vegetation types of Lower Guinea and bathymetry of the (e.g., ref. 15) by building ecological niche models (ENMs; see SI Gulf of Guinea. Refugial areas proposed by Maley (3) and Anhuf et al. (6) Appendix for methodological details) for each species. for our study area are shown. Vegetation layers are taken from ref. 19. Inset Initially, we tested whether traditional refugia (3, 6) were on the top right shows the area in the context of the African continent. Bathymetry contour values for every 100 m for the Gulf of Guinea (north: important for determining patterns of present-day genetic diver- 5.5; south: −7.0; west: 6.0; east: 14) were downloaded from ref. 95. The sity (H1). We expected individuals in or close to these tradi- light blue region indicates ocean floor greater than 100 m in depth. Map tional refugia to be more genetically diverse (10). We calculated was drawn with QGIS (v3.123). individual-level expected heterozygosity (He ), observed heterozygosity (Ho ), and allelic richness and compared these to distance from traditional refugia (Fig. 1 and SI Appendix, Figs. S12– Annonaceae [custard-apples (21); four species]. Our species S18). A significant relationship was found in five of the seven sampling covers a range of habit types, ecological tolerances, and species, depending on the distance and diversity metrics used. animal-mediated dispersal syndromes (SI Appendix, Tables S1 Decreasing distance to the coast [representing the single coastal and S2). We used a comparative phylogeographic approach (22– refugium hypothesis of Anhuf et al. (6)] was significantly asso- 24) to test how similar evolutionary dynamics are among Lower ciated with higher levels of genetic diversity in Annickia affinis Guinea TRF plant species. and Podococcus barteri. Both concepts of traditional refugia were If species follow the expectations of the refugia hypothesis linked to increased genetic diversity in Anonidium mannii and (H1; Table 1), then, under the strict species-independent refu- Sclerosperma mannii. No significant relationship was found for gia hypothesis (H1a), we expect to find high levels of genetic the remaining species. In a second analysis, we used analysis of structure, concordance in the timing of the emergence of this molecular variance (AMOVA) to test how much genetic vari- structure, and demographic trends showing similar patterns of ance is explained by the three different definitions of stability contraction (i.e., genetic bottlenecks) and expansion of popu- defined above. We found that proximity to stable areas gener- lations across all sampled species, independent of their life- ally explained a small amount of genetic variance (<10%; SI history traits or environmental preferences (7, 25). In contrast, Appendix, Fig. S19). Even so, stable areas explained a significant under the relaxed species-dependent refugia hypothesis (H1b; amount of variance in all species. Most notably, in A. mannii, all Table 1), we expect to infer responses to Pleistocene climatic three versions of stability explained large amounts of variance, events, but no evolutionary or temporal concordance, with each while for P. barteri, we found that climatic stability explained species responding in its own individualistic manner (7, 16, 18). substantially more variance than the others. Using family-tailored nuclear-marker baiting kits (26, 27), we sequenced hundreds of nuclear loci in more than 750 individuals. A Common Equatorial Genetic Discontinuity. We found a conspic- By doing so, we reduced biases linked to the use of different types uous latitudinal genetic discontinuity at the climatic inversion or few molecular markers, enabling us to undertake evolution- located 0 to 3◦ North, in Southern Cameroon/Northern Gabon ary and temporal analyses with increased accuracy in a common (Fig. 2H). It was recovered in all species distributed across framework. the equator (6/7) (Fig. 2 and SI Appendix, Figs. S2 and S4). We then examined how important this barrier is to explaining Results genetic diversity in each species. We performed a hierarchical Generated sequence data were analyzed in four main steps. AMOVA to show that, though ubiquitous, this North–South bar- First, we inferred intraspecific genetic structure for each species rier explains different amounts of variance among species (SI using 2,000 (2k) to 6k single-nucleotide polymorphisms (SNPs) Appendix, Table S3): It is important in A. mannii and P. barteri, (SI Appendix, Table S1). Second, phylogenetic relationships but less so in the other species. between resulting genetic clusters were reconstructed by using whole-sequence datasets of up to 365 nuclear loci (SI Appendix, Asynchronous Population Divergence. To estimate the timing Table S1). We used a subset of nuclear loci together with fossil- of population divergence, we took a secondary calibration calibration information to date the origin of intraspecific genetic approach. We first produced time-calibrated backbone trees at structure. Third, we inferred the demographic history of each a higher taxonomic level, which allowed us to make use of species and their genetic clusters separately. Finally, we inves- fossils that are not available for the focal groups (SI Appendix, tigated associations between intraspecific genetic variation and Figs. S20 and S21). We inferred divergence times between a range of environmental factors related to geography, climate, genetic clusters using multiple phylogenetic approaches. We stability, and soil characteristics. built population-level phylogenetic trees using divergence times

2 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.2001018117 Helmstetter et al. Downloaded by guest on October 3, 2021 Downloaded by guest on October 3, 2021 xaddoto n otatdit tbeaes Therefore, al. et areas. Helmstetter stable species into as contracted and Pleistocene, of the out during expanded repeatedly fluctuated Trends. have Demographic there Major that Two indicating pattern. common Ma, this driving 4.2 event diver- to single these no 0.3 the of likely from during was ages ranged commonly median events the most gence However, species, in 3). per occurred (Fig. event Pleistocene inversion For major climatic ways. one the different only across in divergence divergence climate example, past population that the affected hypothesis However, the change supporting Ma. species, consider- 1 among differed ably time around bound- through events and divergence Pliocene–Pleistocene of Ma), distribution the 2.5 points—during (approximately ary time two frequently and most 3 at occurred (Fig. divergence topolo- species Population all similar S24–S29). in Figs. reconstructing produced times when divergence approaches analyses and two migration gies StarBEAST These of our trees. levels species upon low (DENIM) expands incorporating migration by approach and This (ILS) sorting notwithstand- generally (33). estimation lineage topologies divergence incomplete used their ing also and reason- we 2), comparison, were ( individuals loci trees trees (Fig. five individual-level Species clock-like supported reflected subsampled cluster. and most well genetic species ably 32 each our represent the of to each selected for or We habitat (32) inferred (83). approach on coalescent based Bayesian StarBEAST the or areas and refugia trees stable backbone defined discrete in traditionally inferred to the refers either Stability are dynamics. and and times structure adverse genetic during influence even also species might a mechanisms of stability. other climate populations that harbor note to We inferred flora. forest rain Guinean species-independent Strict Subhypotheses of assumption the under expectations population genetic dynamics temporal species and impacting) spatial not resulting (or with impacting tested fluctuations hypotheses climatic Main 1. Table eei structure Genetic pta distribution Spatial eei diversity Genetic eorpi history Demographic eprloii fpplto litcn,temporally Pleistocene, population of origin Temporal niomna factors Environmental References fgntcclusters genetic of species within ihnspecies within ihnspecies within mn species among and within tutr mn species among structure nuniggntcvrainsaiiyepandistribution explain stability variation genetic influencing ihnspecies within hs xettosaecnee rudtehptei htsal ra rrfgapae nipratrl nteeouinr yaiso Lower of dynamics evolutionary the in role important an played refugia or areas stable that hypothesis the around centered are expectations These High loarcor Allopatric ihrwithin Higher vdneo population of Evidence atr eae to related Factors 3 ,7) 6, (3, eui oe H) uigavreciai eid eg,LGM), (e.g., periods climatic adverse During (H1): model Refugia .For S5–S11). Figs. Appendix, SI pce agsaetogtto thought are ranges Species parapatric tbeareas stable n expansion and (bottleneck) contraction congruent fgntcdiversity. genetic of l pce epne na in responded species All eui oe (H1a): model refugia risadevrnetlfcos n niomna factors environmental and factors. environmental and traits life-history of independent iia vltoayway evolutionary similar Rssgicnl otatdars oe unait n or one arid more into by ecosystems surrounded Guinea areas stable Lower ecologically discrete across several contracted significantly TRFs IAppendix, SI eae pce-eedn o applicable Not species-dependent Relaxed o ohigh to Low loarcor Allopatric ihrwti tbeaesN relationship No areas stable within Higher vdneo population of Evidence litcn,temporally Pleistocene, ifrn atr across factors Different 7 16) (7, atr eoee o n atclrarea. particular (H1a); any demographic for species common recovered no different pattern to found in we expect examination, trends par- might upon size however, we If population refugia, change. similar cross-species climate infer as areas past period acted different to areas time from differently ticular this populations responding over that also indicates constant are This relatively while 4B). Ka, was 70 (Fig. and cluster 20 final between expansion the in and trends decline of demographic evidence similar in have example, For not cases. did all species belonging same clusters the different to but patterns, species-level 4 (Fig. reflected species our within cluster ranges species’ as periods. bottlenecks glacial may during population which contracted LGM species, of one the but representative all to in be decline close rapid years of declines evidence 200,000 found last for the evidence over in including growth Ka), and decline (200 of fluctuat- experienced patterns species ing Annonaceae the time second, through 4A); increase size (Fig. population of experienced rate species constant palm relatively three a the First, (34): species. approach forest Plot rain way on impact an of less had cycles size these that population effective pattern in cyclical 1) similar (N (Table a decreases and expected increases we of model, refugia the under pce epne ndifferent in responded Species (H1b): model refugia nlf-itr traits life-history on dependent ways evolutionary orltosi nothers in relationship no n oeiec nohr expansion others in evidence no and nldn stability. including parapatric nsm pce and species some in xaso nsm pce btlncs or (bottlenecks) contraction population species some in expansion and (bottleneck) contraction incongruent pce xli itiuinseisexplain species distribution explain species fgntcdiversity genetic of eas siae ouainsz hnefrec genetic each for change size population estimated also We Stair- the using patterns demographic major two recovered We e .affinis, A. .Amr osattedin trend constant more A ). Hypotheses and mannii, A. he ffu lsesshowed clusters four of three affinis, A. We suaveolens. Greenwayodendron N orfgamdl(2:During (H2): model refugia No Low oprior No oeiec of evidence No o eesrl Pleistocene, necessarily Not ifrn atr across factors Different 4 7) (4, B–H e iest.N role No diversity. fstability. of expectations eprlyincongruent. temporally itiuino genetic of distribution vrti eidmyindicate may period this over Guinea Lower across contract not did TRFs periods, climatic adverse .I eea,teeresults these general, In ). NSLts Articles Latest PNAS | f10 of 3

EVOLUTION AB C D

EF G H

Fig. 2. Genetic-clustering results of seven Central African rain forest species. For each species, collection points for individuals (where known) are shown, colored by genetic cluster, as inferred with DAPC. To the bottom left of A–G is a phylogenetic tree showing the relationships among genetic clusters (including outgroups in white). Pie charts at nodes on the trees indicate support (posterior probability). H shows the location of major genetic discontinuities associated with the climatic inversion that were inferred for each species by using TESS3, by drawing a line along contact of ancestry coefﬁcients among clusters in interpolation maps (SI Appendix, Fig. S2).

Distinct Sets of Environmental Factors Explain Patterns in Genetic rain forest plant species exhibiting a broad range of life-history Diversity Across Species. We tested how different environmen- traits. tal factors shaped intraspecific genetic variation in each of Overall, our results support the refugia hypothesis (H1; our species. Again, under the strict species-independent refu- Table 1), where species have undergone periods of contrac- gia model (H1a), we expected genetic variation across species to tion into stable areas and expansion out of them, mainly during be mainly explained by shared stable areas (Table 1). To inves- the Pleistocene. Indeed, as for most plant species studied to tigate this, we collated data relating to geographic distribution date in Lower Guinea (7, 15, 24, 36, 37), we recovered gener- of individuals, climate, soil type, and stability for each of the ally high levels of population structure: Four species had four seven species (SI Appendix, Table S5). We used a distance-based or more genetic clusters (Fig. 2 and SI Appendix, Fig. S2). redundancy analysis (db-RDA) framework (35) to build mod- Even when fixing the number of genetic clusters and com- els and determine the total amount of genetic variance that can paring across our species, we still found geographically dis- be explained by these variables and their relative importance. parate patterns of genetic clustering (SI Appendix, Fig. S4). As expected under H1b, species differed in the total amount In contrast to the previously mentioned studies, our results of variance explained and by the factors that were most impor- are based on the same set of genome-wide markers in each tant (Fig. 5). Indeed, for the liana species M. enghiana, stability family, rather than small numbers of species-tailored markers, explained negligible variance, while for the tree species A. affinis which can be biased toward increased levels of diversity and stability was shown to be more important (Fig. 5). structuring (38). Other major differences among species emerged: For exam- Geographically, most genetic clusters within species are para- ple, soil characteristics explained diversity patterns in G. suave- patric in distribution (Fig. 2). These results are in line with the olens, while the geographic distribution of individuals was key in prediction of past population contraction, leading to genetic dif- S. mannii. However, some variables were common across mod- ferentiation (H1; Table 1), followed by range expansion (7). els, shared by up to seven species (SI Appendix, Table S7). A Secondary contact after expansion out of putative refugia has number of these were soil variables, including pH, depth to not led to genetic homogenization. Barriers to gene flow appear bedrock, and the presence of particular soil types (inceptisols to have been maintained since forest contraction, leading to the and ultisols). In terms of present-day environment, precipitation patterns of genetic structure observed today. However, these was a common explanatory variable, particularly the amount of clusters do not generally overlap with traditional refugia (Fig. 1), precipitation during the driest and warmest periods of the year. nor are they concordant among species. For all of our sampled Stability in precipitation over time and ENM-based stability were species, genetic diversity was linked (SI Appendix, Fig. S19) to at also common to seven and five species respectively, but were gen- least one or more of our three definitions of stability [traditional, erally not important explanatory factors, explaining low amounts climate, or habitat, (30)], favoring the refugia hypothesis (H1). of variance. A previous phylogeographic study of the palm P. barteri based on plastome data also inferred a strong correlation between Discussion habitat stability (ENMs) and unique and high genetic diversity Our study provides a perspective on the impact of the Pleistocene (15) in areas along the Atlantic coast. While the same pattern climatic fluctuations on the highly diverse flora of the Lower was observed in this study for the same species, climate-based Guinean TRF. Using a common comparative phylogeographic stability actually explained more genetic variance (SI Appendix, framework, we tested hypotheses relating to species’ evolution- Fig. S19) and, therefore, played a more important role than ary responses (Table 1) simultaneously across seven lowland habitat stability in P. barteri.

4 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.2001018117 Helmstetter et al. Downloaded by guest on October 3, 2021 Downloaded by guest on October 3, 2021 emtte tal. et Helmstetter our with the 1), across Table relaxed diversity (H1b; the hypothesis genetic support data refugia in our species-dependent Rather, patterns flora. simplis- TRF current Guinean too such Lower appears explain regions, hypothesis TRF to this other tic in (45–48), suggested and Neotropics As (17) the Guinea 1). as Lower Table in frogs (H1a; for hypothesis refugia independent study. present the diver- in recovered population those to of similar in estimates the microsatellites nuclear age of probably based Indeed, chloroplast gences are of DNA. dynamics discrepancies struc- nuclear evolutionary age the population and in These of differences were of region. origin result results the pre-Pleistocene These in 3). a ture (Fig. suggest here struc- to population inferred of used older those emergence in than the (44) resulted of Ma) ture data, 4.5 plastome to complete (3.5 43). estimates on 42, in based (17, here, studies animals pled and single-species within 41) times 40, other divergence 36, Estimated from (15, plants inferred African Central intraspecific also generating was struc- in population turing Pleistocene-dated role 10). key (7, structure Pleistocene a population which play in (H1), fluctuations hypothesis refugia climate results the These with line (39). in magnitude are in increased when fluctuations 3), (Fig. climatic Pleistocene the during originated structure genetic North/South the events of divergence side indicate either on Stars inversion. situated climatic Pleistocene. Ma clusters 2.58 genetic the at DAPC of color between StarBEAST background start for of in the species change plots all The represents density (dashed). across represent DENIM divergence lines and population (dotted) black of represent Overlain age medians and these event. median of (triangles) the each side for StarBEAST either HPDs on using Lines 95% shown. by are estimated (circles) DENIM times divergence Median 3. Fig. 0246 u eut,hwvr ontfi ihasrc species- strict a with fit not do however, results, Our species, most in that, showed we dating, molecular Using * * ouaindvrec ie nLwrGienri oetspecies. forest rain Guinean Lower in times divergence Population * * * * * Time (Ma) Time * pce sam- species a suaveolens, G. * .suaveolens G. * 4)were (41) STARBEAST DENIM S. P. P. M. G. A. A. mannii barteri acaulis mannii affinis suaveolens enghiana eprloii fteNrhSuhgntcdsotniy(7) discontinuity genetic North–South the 2H (Fig. of origin temporal 49). species (47, on regions TRF impact other limited to had land contrast thus, continental in of and, dynamics, Gulf of 1) evolutionary that Fig. the increases km; in note important 70 m) than to We 100 (less lead 37). to not up (7, did (of species LGM Guinea the liana during and drops herb, sea-level tree, population-structure in single-species on analyses based this suggested flora, Guinean also Lower was the In Pleis- (16). to fluctuations climate responses tocene individualistic showing species plant sampled ,weesfrohr,i a climate was suave- it others, G. for enghiana, whereas barteri ), (M. P. and important acaulis, most P. olens, the were factor variables important soil R most of a the sum not as (largest was recovered our it supporting However, was species, results. each Stability previous for 5). variable con- explanatory (Fig. differed significant species that variance across genetic support explaining siderably not factors did as our (35) prediction, the However, a analyses 1). be (Table Under redundancy to species distance-based stability species. across expected factor across we explanatory main (H1a), diversity hypothesis genetic refugia strict of influenced distribution have factors the environmental what tested we structure, (H1b). hypothesis further species-specific differed—lending relaxed the populations to stable support 4 (Fig. those others of not but locations species, some of in patterns found fluctuating were rapidly detect to ability N our versus limits kb data in (270 Annonaceae disparity than dataset kb; a 740 data palm the sequence to However, less palms. point contained effect than greater species a results had Annonaceae periods Our on glacial where species. of families plant effect these between muted demographic of long-term a 4), the from trajectory suggests (Fig. on fauna fluctuations This Ka and climatic flora 53–56). 300 Pleistocene other (36, last in worldwide the inferred TRFs over commonly gen- also decline a is exhibited population which Palms of plants. lack African eral found Central been other have that demographic several (52), in bottlenecks Pleistocene population found and gradual (36), 1). (41), response declines expansions We a demographic (Table recent in including expected fluctuations changes, 4). be climatic pattern might as (Fig. cyclical to decreases, somewhat and patterns increases a of experienced major species Annonaceae two revealing inland. further northern regions be between southern could differences and it environmental but reduced unclear, to remains related pattern this this behind along mechanism structuring 2 genetic (Fig. a Eastern divide show in not sampled Nevertheless, did species inland. (50). three Cameroon/Gabon further for flora down recovered Guinean break clusters to Lower Genetic appears the discontinuity sup- in genetic further event this This barrier rare climatic 3). a this (Fig. across is colonization event species that divergence hypothesis each the major in ports single, inversion a the just across recovered eco- we in 51)]. (50, addition, differences inviability to In immigrant [e.g., linked populations dispersal of niches of limited logical side or either (7) on in barrier seasons differences the inverted pro- as the asynchronous, with such this associated traits, Instead, but phenology species-dependent (7). similar, to event linked shared from cesses single result a may of result discontinuity the not plants thus, Guinean Lower is, in pattern time genetic (3/4 dominant Pleistocene This the periods). during mainly posterior 3), Fig. highest [HPDs]; (nonoverlapping densities signif- We periods four 43). time to 42, dated different (17, are animals icantly splits in northern/southern common these less that be show to seems but 50), 24, e aei on fidvdaitcrsosspran othe to pertains responses individualistic of point in case A nadto oifrigteoii n yaiso genetic of dynamics and origin the inferring to addition In species, seven all of histories demographic the inferred We eei lseswt osiuul tbepplto sizes population stable conspicuously with clusters Genetic . ,adi a eta hsreduction this that be may it and S1 ), Table Appendix, SI .Ti are a fe enrcvrdi lns(,15, (7, plants in recovered been often has barrier This ). B–D 2 o n pce Fg ) o oespecies, some For 5). (Fig. species any for ) and i.S2 Fig. Appendix, SI NSLts Articles Latest PNAS .Tecausal The B–D). B–H ,adthe and ), | f10 of 5

EVOLUTION A E

BCDH

Fig. 4. Demographic histories of seven Lower Guinean rain forest species. Population size changes were reconstructed by using the Stairway Plot method. Median effective population sizes (Ne) over time are shown by colored lines. A shows patterns inferred using all individuals in each species. B–H show demographic trends for each genetic cluster detected within each species, based on the same color code as A. Additional plots, including CIs, can be found in ref. 92.

and/or geography (A. affinis, A. mannii, and S. mannii). Over- study gives a temporal context to the evolutionary dynamics for all, our analyses showed that each species possesses a unique multiple Lower Guinean Central African TRF plant species. The set of environmental variables that explains patterns in genetic idiosyncratic nature of genetic structure, population divergence, variation (Fig. 5), favoring our hypothesis H1b. demographic histories, and the environmental traits underlying Finally, given our variety of sampled species (SI Appendix, patterns of diversity across our seven species demonstrates that Table S2), we show that differences in life-history traits and phe- each has responded to past climate change differently, favoring notypes resulted in alternative evolutionary histories, supporting a relaxed species-dependent refugia hypothesis (7, 16–18). We previous studies (18, 48, 57). However, it was also suggested that show that similar patterns of genetic structure across species, species sharing life-history traits or phenotypes, such as habit such as the North–South discontinuity, are likely the result of dif- or dispersal syndromes, could respond in a similar evolutionary ferent underlying causal factors (66) (i.e., asynchronous origins manner to past climate change (7). We show that this isn’t nec- [Fig. 3] and varied demographic histories [Fig. 4]). Therefore, essarily true. For example, the three tree species sampled here rather than aiming for general patterns, it may be more perti- show differences in terms of spatial genetic structure and fac- nent to understand individual species’ ecology and evolutionary tors explaining the distribution of diversity (Figs. 2 A–C and 5), history, before combining these to understand how TRFs have though the temporal origin of this structure did exhibit some sim- responded to past climate change. Our framework would allow ilarities (Fig. 3). Differences were also evident for species with similar studies in Southeast Asian or Amazonian rain forests potentially similar dispersal syndromes. Given fruit morphology to investigate whether similar, species-dependent patterns are (58, 59) and field observation (ref. 60 and personal observation found. Our understanding of individualistic responses will be by T.L.P.C.), A. mannii and S. mannii are often dispersed by ele- critical if we are to predict general responses to ongoing climate phants (SI Appendix, Table S2), which can travel long distances change and has important implications if we want to conserve and have a strong effect on population structure and recruitment those species most at risk. (61, 62). However, population structure and evolutionary dynamics of these species fall at the two extremes of our study (Figs. 2 B Materials and Methods and G, 3, and 4 A, C, and H), although geography is an important Sample Collection, Library Preparation, and DNA Sequencing. We sampled factor in both species (Fig. 5). and sequenced a total of 725 individuals from four Annonaceae and three TRFs contain an estimated 50% of the world’s plant biodiver- palm (Arecaceae) species across most of their ranges in Cameroon, Gabon, sity (63). As much as one-third of the tropical African flora is and Republic of the Congo (one specimen of P. barterii from Nigeria). At potentially threatened with extinction (64), so it is critical that each collection site, we surveyed for the presence of the seven species, sam- we understand the origins of diversity in this region (65). Our pling them if found. This typically resulted in three to seven species per

6 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.2001018117 Helmstetter et al. Downloaded by guest on October 3, 2021 Downloaded by guest on October 3, 2021 emtte tal. et Helmstetter all for analyses DAPC conducted also We off. increasing began flattened from values BIC values BIC best- in the when identify change examining to by or (BIC) clusters criterion of information number Bayesian of fitting the value used per function We iterations 20. the 100,000 = with used K-means We successive clusters. using ters genetic to performed individuals then find.clusters assign We to existed. (28) isolation-by-distance DAPC whether identify of to data patterns SNP broad-scale using individuals con-specific of Diversity. pairs between and kept Structure We Population (70)]. [v1.8 sites. bcftools monomorphic minor using excluded and and SNPs SNPs (>10), filter biallelic individuals only to across (>0.01) depth frequency average (>25), in allele depth (>2), (>40%) HaplotypeCaller depth quality program mapping by by the quality SNPs to using filtered reads We SNPs (69)]. Duplicates paired called [v4.0 GATK (68)]. cleaned, we [v0.7.12 and our BWA removed, mapped using were We com- references and file. species-specific built reference new, were these a contigs form consensus to out loci, filtering bined paralogous After against. and reads coverage map to low species each for psuedoreference a Calling. SNP and palms both for 27 ref. of See that species. 60. Annonaceae followed ref. cleaning e.g., read and data; sequencing, concrete morphology with S2 fruit Table available from Appendix, are inferred monkey, studies were ruminant, few syndromes rodent, because These (elephant, dispersed). syndromes bird dispersal or under and [ranges premontane m]) and 1,300 lowland, firme, terra forest, rain (inundated affinis, (A. ( barterii, trees species (P. liana as shrubs such understory forms, olens), life cover of but restricted, variety forest rain a lowland are species These location. sampling cate- each to belonging model variable. the in variables See gory. of proportion relative R the The total category. full the own the shows their in axis to included assigned x were were that which variables nonsignificant, of but sta- number models, or small soil, a geography, were climate, were There categories: bility. different model four the of one up to making assigned Variables analyses. redundancy distance-based 5. Fig. ilyslciganme fae htmxmzdcosvldto cr and score cross-validation maximized that ini- axes DAPC, of for number axes a of selecting number optimum tially the choosing identified, clusters function the the used clustering on then based We analyses results. phylogenetic downstream for species per event selected we and values, G. suaveolens Annonaceae Annonaceae Annonaceae Annonaceae M. enghiana Arecaceae Arecaceae Arecaceae P. acaulis S. mannii A. mannii P. barteri A. affinis tce a ltsoigtecmoet ftemdl from models the of components the showing plot bar stacked A al S7 Table Appendix, SI ocl Ns efis sdSCP vrin()114(7]t build to (67)] 1.1.4 (v) [version SECAPR used first we SNPs, call To nteRpcae“dgnt 7)t ne h ubro clus- of number the infer to (71) “adegenet” package R the in k .Te locvrarneo ifrn ecologies different of range a cover also They enghiana). M. to 2 = . . . . . 0.5 0.4 0.3 0.2 0.1 0.0 o eal.Mtoooyfrlbaypeaain DNA preparation, library for Methodology details. for k 2 .For 8. = k ftefl oe,adec a sdvddt show to divided is bar each and model, full the of oke tlatoeitapcfi divergence intraspecific one least at keep to 2 = bRAmdlR db−RDA model o h muto aineepandb each by explained variance of amount the for .acaulis, P. dapc eiiilycnutdMne tests Mantel conducted initially We and acaulis, P. 2)t en h iest among diversity the define to (28) 2 k and 1 = and mannii, A. k .mannii S. a iia BIC similar had 2 = ,adone and ), Variable type .suave- G. k Stability Soil Not significant Geography Climate pto up SI k sdauiompirdsrbto rm8 o12M ntero,as root, the on Ma 112 We to partition. 89 each from to distribution Separate assigned prior (78). were uniform (v0.1.3) models a modeltest-ng used clock using each lognormal by for models relaxed identified substitution were appropriate the partition and partitions, across analysis. lowest linked dating the molecular with our tree loci for gene subset 32 this each the used selected for and clock- We variance variance how (32). root-to-tip root-to-tip SortaDate estimated calculating program then the by We using was (27). locus Anaxagorea each on like trees our rooted and each above. detailed from inference as individual HybPiper, phylogenetic single using for a alignments producing included 20 species, and focal sampled in subfamilies Annonaceae we used all be tree, across to backbone from Annonaceae, Annonaceae this taxa for our to one built For and solution we calibration. palms species, a secondary focal for find seven one our To trees, of backbone tree. each two in the events divergence calibrate date to and issue available fossils rarely are Trees. Backbone taxa. outgroup individuals, related sequenced closely all three using to one species, to each replicates) addition (75). for in (100 (v8.2.9) separately using analysis RAxML done in locus bootstrap was search each This rapid maximum-likelihood thorough to a a assigned performed by were followed We models file. then GAMMA partition We + a sequence. GTR empty Separate the an (76). using loci with aligned alignment all each concatenated in target individuals the missing on based Trees. Maximum-Likelihood analyses, phylogenetic our of individuals. each of (27). 75% for species in reconstructed this length did exon their set supercon- We of those suitable 75% only a had selected identify we that To inference, tigs dataset. phylogenetic the reliable from for classified discarded loci was of and locus paralog this true so, and a three If “main” as clades. the than separate whether more examined formed we sequences concerning locus, alternative a sequences for When flagged loci were (75). corresponding individuals the (v8.2.9) of trees RAxML gene using inferred sequences we paralogous HybPiper, putative by If option flagged positions. default were –auto gap the allowed using the all (74) with (v0.91b) and (73) GBLOCKS parameters with (v7.305) alignments aligned MAFFT these We cleaned using locus. and supercontigs target of each exons set for target for data each 27, sequence the ref. off-target of in associated “supercontigs” described plus produced as data, This cleaned inference. our Alignment. phylogenetic process to Multisequence (72) (v1.2) and HybPiper Assembly Contig See climate. calculated past as and areas, present details. stable methodological for further 3) below) annual and (see in rainfall; ENMs change from annual from calculated and (31) temperature from stability mean distance climatic 1) per- 2) criteria: We different refugia; areas. three Maley’s stable on to based proximity with analyses As their separate possible inversion. formed by not individuals climatic grouped was the we at comparison Second, discontinuity this genetic result, species’ or North a the mostly North was of to cluster South the clusters whether or these on based assigned was indi- then classification sets This grouped (DAPC), South. two we cluster where performed genetic approach We hierarchical by data. a viduals the used of we First, stratification AMOVAs. given of a by explained linear be a performed AMOVA. We coastline. measure. distance the and statistic from approximately summary each an distance for from regression and distance as point) (calculated central refugia Maley’s from tance (H heterozygosity observed function the using by maps score identified geographic cross-validation We the of filetype. examining value Models by each for clusters Mixed dur- of Factor errors number Latent caused appropriate they the to as conversion dataset, the file from ing removed were SNPs projected maximum of the number a (29). with “tess3r” and clusters package algorithm R population the least-squares using inferring by when implemented was data TESS3 location geographic rating use. hnuigtefunction the using then epromdmlclrdtn sn ES2(244 7) re were Trees (77). (v2.4.4) BEAST2 using dating molecular performed We above as (75) (v8.2.9) RAxML using locus each for trees gene built We (H heterozygosity expected of values per-individual calculated also We incorpo- by DAPC from differed TESS3, approach, clustering second A hsapoc a eue ots o uhgntcvraincan variation genetic much how test to used be can approach This aigitapcfi iegneeet sdfcl,a there as difficult, is events divergence intraspecific Dating nepltoso netycefiinswr ipae on displayed were coefficients ancestry of Interpolations k. optim.a.score o fe utbelc eeietfid efildany filled we identified, were loci suitable After ,adallcrcns n oprdteet dis- to these compared and richness allelic and ), oslc nlnme fae to axes of number final a select to pxcat plot.tess3Q. k f1.I oeseis small a species, some In 10. of ucini h rga phyx program the in function NSLts Articles Latest PNAS .aalsadS mannii. S. and acaulis P. eue h program the used We IAppendix SI | f10 of 7 for e ),

EVOLUTION recommended (79, 80), based on fossil evidence from Futabanthus and performed ensemble modeling using all models. We then projected models Endressinia. (of all algorithms) into the past, performing a further ensemble modeling The analysis was run for 100 million generations, sampling every 10,000 step. Finally we converted ensemble projections to binary presencevabsence generations to generate a posterior distribution of 10,000 trees. Effective maps using KAPPA and TSS, allowing us to identify areas suitable for each sample-size values were examined by using Tracer (v1.7) (81), and the appro- species under a given statistic and present or past climate model. Further priate amount of burn-in was removed. Multiple runs were performed and information on our ENM approach can be found in SI Appendix, and details combined by using logcombiner. Treeannotator was used to generate a of model adequacy can be found in SI Appendix, Table S6. maximum-clade credibility tree, keeping target node heights. We repeated this process for 23 palm taxa, spread across subfamilies Demographic Inference. We used Stairway Plot (34) (v2), a model-flexible Arecoideae, Ceroxyloideae, and Coryphoideae and including two or three approach that uses site-frequency spectra (SFS), to infer changes in effec- samples for each focal species. We rooted trees on Coryphoideae taxa (21). tive population size (Ne) over time. We generated VCF files for each cluster, We used three fossils for calibration, following priors in Couvreur et al. (21), but did not apply a minor-allele frequency filter to avoid skewing the SFS. but modifying the prior for crown Hyphaeninae due to a recently published Stairway Plot uses SNP counts to infer changes in Ne, so the removal of fossil, Hyphaeneocarpon indicum (82) (SI Appendix, Table S4). SNPs with missing data may skew counts. We generated SFS for each cluster by calculating the minor-allele frequency at each SNP and multiplying Dated StarBEAST and DENIM Trees. To date divergence among genetic clus- this by the mean number of haploid samples, following Burgarella et al. ters in our focal species, we used a Bayesian approach involving the (87). This resulted in a new SFS that used all observed site frequencies and multispecies coalescent, implemented in StarBEAST (83). To improve mixing, minimized the number of SNPs removed. The total number of samples was we used the StarBEASTWithSTACEYOps template in BEAST2. When build- slightly reduced based on the amount of missing data. Breakpoints were ing a tree for each species, we included the closest outgroup taxa from the calculated as recommended in the manual. We used two-thirds of sites for backbone trees. We randomly chose five individuals per genetic cluster in training and performed 200 bootstraps. The number of observed sites was the focal species and defined each set of five as a different population. calculated as the total length of the SECAPR pseudoreference (see above) Subsetting in this way was previously shown to represent general patterns for each species. We used an angiosperm wide mutation rate of 5.35e−9 well (50). Outgroup individuals were assigned to a separate species (multiple sites per year (88) and a generation time of 15 y, based on the generation species if necessary). time of the Annonaceae species Annona crassiflora (89). For palms, we used Like for the backbone trees, 32 loci were selected for each species based a mutation rate of 2.5e−8 and a generation time of 10 y (90). A longer gen- on root-to-tip variance. We used a constant population function and a coa- eration time of up to 50 y has been proposed for some of our study species lescent exponential population tree model. Given that our analyses were (91). Previous work in A. affinis (50) has shown that increasing generation among very close relatives and that we limited root-to-tip variance, we time does not change demographic trends and has only minor effects on chose to use strict clock models for each locus. These models are appropri- the timing of events, though Ne was markedly scaled down as a result, so ate for recently diverged lineages, given the expected low substitution-rate values of Ne should be interpreted with caution. variation and improved computational feasibility (84, 85) when using relatively large sequence datasets. We also tested using lognormal uncorrelated db-RDA. We used db-RDA (35) to uncover how genetic variation could be relaxed clock models for each locus for A. affinis and found similar ages to attributed to different geographic, climatic, soil, and stability variables. those obtained with the strict clock (SI Appendix, Fig. S30). Genetic distances were calculated by using principal coordinates analy- We added uniform priors on the root where the upper and lower bounds sis and used as a response variable. We used Moran Eigenvector’s Maps were set to the 95% HPD intervals from the relevant node in the appropriate (MEMs) to summarize spatial structure in the data (referred to as geogra- backbone tree. The analysis was then run as specified previously for the phy throughout). Explanatory variables included those variables used for backbone trees. Given that the Podococcus are sister species, we analyzed ENMs, MEMs, several variables related to stable areas, and US Department them together, while all other species were analyzed separately. of Agriculture soil taxonomy (SI Appendix, Table S5). We pruned correlated We also used DENIM (33) to account for migration among different variables from each dataset and then built db-RDA models for each species genetic clusters. We used the same input dataset as for StarBEAST, modi- by adding and removing variables in order to maximize the explained fying priors in accordance with the DENIM manual. This yielded topologies variance. Further details about this approach can be found in SI Appendix. and divergence times contingent on migration events between taxa for individual loci, which can be compared to StarBEAST results. Initial runs revealed Data Availability. Data associated with this manuscript are stored on Dryad that uniform divergence-time priors were incompatible with palm species, (92). Scripts used for analyses can be found on GitHub (93). Sequence data so normal priors were used instead. In this case, the 2.5% and 97.5% val- have been deposited in the Sequence Read Archive (94). ues of the normal distribution encompassed the 95% HPD intervals of the corresponding node in the palm backbone tree. See SI Appendix for more ACKNOWLEDGMENTS. We thank Prof. Moutsambote, Raoul Niangadouma, information on our dating approaches. and Theophile´ Ayole for help in the field. We also thank Olivier Hardy and his team for sharing samples of G. suaveolens. We thank the Institut ENMs. We constructed ENMs for each of the seven species using the R pack- de Recherche pour le Developpement´ (IRD) itrop High-Performance Com- age “biomod2” (86) and climatic, altitude, and soil variables (SI Appendix, puting (HPC) (South Green Platform) at IRD Montpellier for providing HPC Table S5). We collated known occurrence points for each species from resources that have contributed to the research results reported within our field data, including additional individuals beyond those that were this paper. We thank the Centre National de la Recherche Scientifique et Technologique (CENAREST), the Agence National des Parques Nationaux sequenced for genetic data (1,691 individuals total; SI Appendix). The back- (ANPN), and Prof. Bourobou Bourobou for research permits (AR0020/16 and ground area for each species consisted of a 5◦ buffer around occurrence AR0036/15 [CENAREST] and AE16014 [ANPN]). Fieldwork in Cameroon was points. We inferred ENMs using five different model algorithms: gener- undertaken under the “accord cadre de cooperation” between the IRD and alized linear model, boosted regression trees, artificial neural networks, Ministere` de la Recherche Scientifique et de l’Innovation. We thank Prof. random forest, and MAXENT.Phillips (maxent). We evaluated model fit by Bouka Biona of the Institut National de Recherche en Sciences Excates et using Cohen’s kappa (KAPPA) and true skill statistic (TSS). Following this, we Naturelles of the Republic of Congo for research permits.

1. H. P. Linder et al., The partitioning of Africa: Statistically deﬁned biogeographical 7. O. J. Hardy et al., Comparative phylogeography of African rain forest trees: A review regions in sub-Saharan Africa. J. Biogeogr. 39, 1189–1205 (2012). of genetic signatures of vegetation history in the Guineo-Congolian region. Compt. 2. V. Droissart et al., Beyond trees: Biogeographical regionalization of tropical Africa. J. Rendus Geosci. 345, 284–296 (2013). Biogeogr. 45, 1153–1167 (2018). 8. A. W. Diamond, A. C. Hamilton, The distribution of forest passerine birds and 3. J. Maley, The African rain forest—main characteristics of changes in vegetation and quaternary climatic change in tropical Africa. J. Zool. 191, 379–402 (1980). climate from the Upper Cretaceous to the quaternary. Proc. Roy. Soc. Edinb. B Biol. 9. J. Haffer, Speciation in Amazonian forest birds. Science 165, 131–137 Sci. 104, 31–73 (1996). (1969). 4. S. A. Cowling et al., Simulated glacial and interglacial vegetation across Africa: Impli- 10. G. Hewitt, The genetic legacy of the Quaternary ice ages. Nature 405, 907–913 cations for species phylogenies and trans-African migration of plants and animals. (2000). Global Change Biol. 14, 827–840 (2008). 11. K. D. Bennett, J. Provan, What do we mean by ‘refugia’?Quat. Sci. Rev. 27, 2449–2455 5. A. M. M. Lezine,´ K. Izumi, M. Kageyama, G. Achoundong, A 90,000-year record of (2008). Afromontane forest responses to climate change. Science 363, 177–181 (2019). 12. R. Bonneﬁlle, “Rainforest responses to past climatic tropical Africa” in Tropical Rain- 6. D. Anhuf et al., Paleo-environmental change in Amazonian and African rainforest forest Responses to Climatic Change, J. Flenley, M. Bush, eds. (Springer, Berlin, during the LGM. Palaeogeogr. Palaeoclimatol. Palaeoecol. 239, 510–527 (2006). Germany, 2007), pp. 117–170.

8 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.2001018117 Helmstetter et al. Downloaded by guest on October 3, 2021 Downloaded by guest on October 3, 2021 3 .L Charles L. K. 43. Portik M. D. 42. Pi R. inferred 41. dispersal gene and history Population Hardy, J. O. Doucet, L. J. Debout, D. G. Africa. in 40. change vegetation scale Orbital Dupont, L. 39. V U. 38. Ley C. A. 37. from environments Duminil how J. of 36. testing Integrative Knowles, L. L. Edwards, L. D. He, Q. 35. practical and sorting A lineage incomplete time: of presence little the so in estimation genes, Divergence Jones, many R. Svenning, G. So 33. Walker, C. J. F. J. Brown, Balslev, W. J. H. Smith, Dransfield, A. S. J. 32. Kissling, D. W. Blach-Overgaard, A. change. 31. climate from refugia Identifying Ashcroft, B. M. 30. spatial of inference Fast François, TESS3: O. Michel, O. Martins, H. Deist, M. T. Caye, K. A 29. components: principal of analysis Discriminant Balloux, F. Devillard, S. Jombart, T. 28. Couvreur P. L. of T. analyses 27. Phylogenomic Leebens-Mack, J. Barrett, F. C. Trapnell, W. D. Heyduk, K. pre- Stability 26. Moritz, C. Rodrigues, T. M. phy- Haddad, F. Comparative C. Hickerson, Hardy, J. M. J. Carnaval, O. C. A. Savolainen, 25. V. Dauby, G. Duminil, J. Heuertz, M. of 24. identification marine of models the phylogeographic comparative and Testing Meyer, P. phylogeography C. Hickerson, Comparative J. M. Faith, 23. P. D. Moritz, of patterns C. diversification 22. global and Origin Baker, J. W. group. Forest, model F. a Couvreur, as P. Palms L. evolution: T. forest 21. rain Tropical Baker, J. W. Couvreur, P. L. T. 20. 4 .Lu .X u xlrn ouainsz hne sn N rqec spectra. frequency SNP using changes size population Exploring Fu, X. Y. Liu, X. 34. for Les Valton, C. Chevillotte, H. Guillaumet, L. J. traits, 19. Species’ phylogeography: in Phenotypes Mason, A. N. Bell, C. R. Zamudio, R. K. 18. Bell C. R. 17. 6 .R twr,A .Lse,I ans .Dal L. Barnes, I. Lister, M. A. Stewart, R. J. 16. environmental Mid-Pleistocene Wefer, G. Faye Schneider, A. R. 15. Donner, B. Dupont, M. L. 14. phylogeny Taxonomy, begonias: Refuge V. Begoniaceae in “Studies Sosef, M. S. M. 13. emtte tal. et Helmstetter ogla anforest. rain Congolian frog leaf-folding Ice the during rainforest African the Ages. of fragmentation indicate trees shade-tolerant rpclfrs rguigdmgahcmdlselection. model demographic using (2017). frog forest tropical tree, timber African Central a of structure benthamianus genetic spatial from (2011). Ecol. populations?Mol. natural in (2008). 3817 diversity genetic genome-wide reflect rainforests. African central in forest the supports species tree forest rain hypothesis. African refuge in changes demographic and tion landscapes. across structure (2013). genetic 3402 shape present the to past the Genet. migration. era. genomic the (2018). in estimation divergence-time to Africa. approach in diversity species palm (2013). on 2426–2435 effects climatic Multimillion-year (2010). selection. for (2016). scans genome and structure population populations. structured genetically of (2010). analysis 94 the for method new genes. nuclear of enrichment targeted using Soc. Linn. J. Biol. genus the in relationships species hotspot. forest Atlantic Brazilian the (2009). in diversity genetic dicts Africa. Guinea, lower from trees (2014). rainforest in logeography approach. Bayesian (2008). hierarchical a using dispersal and vicariance conservation. for areas divergent genetically palms. of phylogeny genus-level Biol. complete a from Inferences forests: rain tropical Biol. BMC 2009). France, 115 Format 000; 000 6 Islands. Guinea of Gulf the and Africa Central from Ecol. frogs Mol. reed in incursions marine 0184 (2016). 8041–8048 diversification. 113, vertebrate and variation, environmental time. and space diversity. in species genetic of responses unique predicts stability Climate Evol. forests: Phylogenet. rain African tral hnei rpclArc ea serya .5Ma. 1.05 as early as began Africa tropical in change 94. University vol. Agricultural 1994), (Wageningen Netherlands, Africa” Wageningen, University, in Wageningen refuges in Papers, forest Scutobegonia rain sect. glacial and to Loasibegonia sect. relation Begonia of biogeography historical and er,G ab,E amk .J ad,Pesoeepplto xasosof expansions population Pleistocene Hardy, J. O. Kaymak, E. Dauby, G. neiro, l,A iaso,L at,H lern owa xetd irstliemarkers microsatellite do extent what To Ellegren, H. Waits, L. Einarsson, A. ali, ˜ 4(2011). 44 9, ¨ rc il Sci. Biol. Proc. 5–5 (2015). 555–559 47, hlgorpyo h genus the of Phylogeography al., et oprtv hlgorpyo ih eb n ins(Marantaceae) lianas and herbs eight of phylogeography Comparative al., et doycai epne ociaedie oetfamnainand fragmentation forest climate-driven to responses Idiosyncratic al., et 8(2013). 48 11, 2354 (2017). 5223–5244 26, aePesoeemlclrdtn fps ouainfragmenta- population past of dating molecular Pleistocene Late al., et yt Biol. Syst. vlaigmcaim fdvricto naGuineo-Congolian a in diversification of mechanisms Evaluating al., et k,sa n oetilns iesfiaini h African the in Diversification islands: forest and sea, Sky, al., et (Caesalpinioideae). hlgnmc ftemjrtoia ln aiyannonaceae family plant tropical major the of Phylogenomics al., et 0–2 (2016). 106–120 117, fiau paradorsalis Afrixalus 2–3 (2016). 126–138 105, .Biogeogr. J. 0780(2017). 20171800 284, 93 (2019). 19–31 68, × .Biogeogr. J. 5c Isiu eRcecepu eD le pour Recherche de (Institut cm 75 rn.Genet. Front. 4315 (2015). 1443–1454 42, Sabal Heredity 7119 (2018). 1781–1794 45, rc o.Sc .Bo.Sci. Biol. B. Soc. Roy. Proc. Aeaee sn agtdsqec capture. sequence targeted using (Arecaceae) Aua yeoide ftelwrGuineo- lower the of Hyperoliidae) (Anura: 0 (2014). 403 5, 89 (2011). 88–99 106, Podococcus rn.PatSci. Plant Front. o.Ecol. Mol. n eui eiie:Individualistic revisited: Refugia en, ´ t rpclshmdsArcie,1: Africaines, humides tropicales ets ˆ Geology o.Eo.Resour. Ecol. Mol. ut c.Rev. Sci. Quat. 1–2 (1998). 419–429 7, rc al cd c.U.S.A. Sci. Acad. Natl. Proc. Pla/rcca)i Cen- in (Palmae/Arecaceae) .Biogeogr. J. o.Ecol. Mol. 9–9 (2001). 195–198 29, 91(2019). 1941 9, LSOne PLoS vlpeet Bondy, eveloppement, ´ Science M vl Biol. Evol. BMC LSOne PLoS 6–7 (2010). 661–671 277, Evolution Distemonanthus M Genet. BMC 5245–5263 26, 3589–3602 30, 1407–1413 37, 785–789 323, e0197433 13, Ecology 540–548 16, e84307 9, 3808– 17, 3386– 67, 322 8, BMC Mol. Nat. 94, 11, 7 .Bouckaert Unix. for R. tools Phylogenetic 77. Phyx: Smith, A. S. Walker, F. of J. Brown, post-analysis W. and J. analysis phylogenetic 76. for tool A 8: version in RAxML use Stamatakis, their A. for alignments 75. multiple 7: from blocks version conserved software of Selection alignment Castresana, sequence J. multiple 74. MAFFT Standley, M. D. Katoh, K. 73. markers. genetic Johnson of G. analysis M. multivariate the 72. for package R A Adegenet: Jombart, T. associa- discovery, 71. mutation calling, SNP for framework statistical A Barrett, J. Li, H. 70. McKenna A. 69. 8 .L,R ubn atadacrt hr edainetwt Burrows-Wheeler with alignment read short accurate and Fast Durbin, R. Li, H. 68. Comparative Soltis, S. P. Manos, S. P. Andermann, McLachlan, T. S. 67. J. Morris, B. A. Soltis, E. D. 66. B M. Nowak, C. Pauls, U. S. 65. 3 .L iehrt .L .Curu,W .Bkr ln hlgn sawno nthe on window St a T. as 64. phylogeny Plant Baker, J. W. Couvreur, P. L. T. Eiserhardt, L. W. 63. planters Tree Terborgh elephants: Forest J. Walsh, P. 62. Maisels, F. Mossimbo, E. Deem, L. S. Blake, S. 61. Gautier-Hion A. 60. “Annonac Thomas, Le A. 59. 8 .J Lissambou J. B. 58. Duminil J. 57. for- Atlantic Paz an A. of Phylogeography 56. Miyaki, Y. C. Cabanne, S. G. Batalha-Filho, H. 55. Allen M. J. 54. 3 .Morgan K. for- tropical 53. African putative Testing selection Dawson, K. natural I. by Dormontt, E. caused Harris, D. isolation Lowe, Reproductive J. A. Funk, J. 52. D. Vines, H. T. Nosil, P. 51. Helmstetter J. A. 50. Raes N. 49. Ib R. Paz, A. 48. explain to hypothesis refugia the Leite of R. become L. has Y. What 47. Kaefer, L. I. Rocha, da G. com- D. plant neotropical 46. and Amazonian Bush, B. M. Oliveira, De E. P. Colinvaux, A. P. 45. Migliore J. 44. LSCmu.Biol. Comput. PLoS (2017). 1886–1888 33, phylogenies. large analysis. phylogenetic usability. and performance in Improvements enrichment. target using Sci. reads sequencing high-throughput from netics Bioinformatics data. sequencing from estimation Bioinformatics parameter genetical population and mapping tion data. sequencing DNA next-generation analyzing eune,fo a ed oainet.PeJPernshttps://doi.org/10.7287/ transform. Preprints PeerJ alignments. Illumina 2018). enriched to February targeted (13 reads of peerj.preprints.26477v3 raw processing user-friendly from and rapid sequences, the for pipeline ics America. North eastern (2006). unglaciated of phylogeography species. and populations within diversity genetic on extinction. vlto fhpriest ntetoia anoetbiome. rainforest tropical (2017). 1422 the in hyperdiversity of evolution forests. Congo. the of community. vertebrate forest tropical a 1–371. pp. 16, vol. 1969), France, Paris, Naturelle, d’Histoire (Annonaceae). Nat. traits?Am. matter. rivers and dynamics climate geography, maximum. glacial last the through Evol. stability Phylogenet. demographic reveals passerine est Evol. Ecol. hotspot. biodiversity Indo-Burma the (2011). nine across within diversity species genetic shaping in change environmental s eui sn hools n ula N phylogeography. DNA nuclear (2010). and 50–58 chloroplast using refugia est bioRxiv. habitats. species. divergent tree from immigrants forest against rain African an in 2019). October diversity (24 of https://doi.org/10.1101/807727 patterns shape refugia maxima. glacial (2014). Quaternary at phylogeographic trait-based A landscape: a across approach. variation genetic structuring in hypothesis. (2016). refuge challenges Evol. Amazonia?Ecol. in in diversity hypotheses biological refuge and aridity the of Rev. failure Sci. The Quat. time-scales: glacial on munities Biogeogr. from insights New trees: forest 601 (2016). 1600016 4, evart ´ Ecography :When ( Vitreorana): glassfrogs Forest Atlantic of Phylogeography al., et itrcldsrbto fSnaadsDpeoaprainforests Dipterocarp Sundaland’s of distribution Historical al., et hr ftetoia fia oai oetal hetndwith threatened potentially is flora African tropical the of third A al., et 1–2 (2019). 212–223 46, o.Ecol. Mol. 8924 (2012). 2829–2842 2, oprtv hlgorpyrvasasae mato Pleistocene of impact shared a reveals phylogeography Comparative al., et Bioinformatics c.Adv. Sci. rmt N ugssln-emsaiiyo nArcnrainforest. African an of stability long-term suggests DNA Primate al., et r-litcn rgno hlgorpia rasi fia rain African in breaks phylogeographical of origin Pre-Pleistocene al., et a ouaingntcsrcueb rdce rmlife-history from predicted be structure genetic population Can al., et a ´ eaanliflecso rercuteti fia equatorial African in recruitment tree on influences Megafaunal al., et e,K .Lp,A .Cafr,Tsigterl feooyadlf history life and ecology of role the Testing Crawford, J. A. Lips, R. K. nez, ˜ ES :Asfwr ltomfrBysa vltoayanalysis. evolutionary Bayesian for platform software A 2: BEAST al., et etoia oetepnindrn h atgailperiod glacial last the during expansion forest Neotropical al., et h eoeaayi oli:AMpeuefaeokfor framework MapReduce A toolkit: analysis genome The al., et Biotropica 4–6 (2000). 141–169 19, yPpr xrcigcdn euneaditosfrphyloge- for introns and sequence coding Extracting HybPiper: al., et 4310 (2008). 1403–1405 24, (2011). 2987–2993 27, PhytoKeys 6–7 (2007). 662–672 169, .Cn,A ik,C ao,A noel,SCP— bioinformat- SECAPR—A Antonelli, A. Bacon, C. Zizka, A. Cano, A. aooi eiino h fia genus African the of revision Taxonomic al., et ´ tal. et hlgnmcdt eelhwaciai neso n glacial and inversion climatic a how reveal data Phylogenomic al., et 9–0 (2012). 892–902 65, Bioinformatics 8–8 (2016). 180–186 39, 1057(2014). e1003537 10, ax44(2019). eaax9444 5, 7333 (2015). 3723–3737 24, ri hrcesa ai ffutcoc n eddseslin dispersal seed and choice fruit of basis a as characters Fruit , o.Bo.Evol. Biol. Mol. e”in ees” ´ 5–6 (2009). 459–468 41, 59 (2018). 55–93 114, 7416 (2009). 1754–1760 25, ln,M fnigr h mato lblciaechange climate global of impact The Pfenninger, M. alint, ´ Greenwayodendron .Aubr A. Gabon, du Flore 3211 (2014). 1312–1313 30, rc al cd c.U.S.A. Sci. Acad. Natl. Proc. rc al cd c.U.S.A. Sci. Acad. Natl. Proc. 4–5 (2000). 540–552 17, Oecologia 3240 (2019). 4302–4309 9, o.Bo.Evol. Biol. Mol. Evolution Heredity eoeRes. Genome NSLts Articles Latest PNAS 2–3 (1985). 324–337 65, o.Ecol. Mol. Annca)phylogenomics. (Annonaceae) 0–1 (2005). 705–719 59, vle d (Mus Ed. eville, ´ 4–5 (2019). 545–557 122, o.Ecol. Mol. o.Ecol. Mol. 7–8 (2013). 772–780 30, e Phytol. New 2–4 (2013). 925–946 22, 2710 (2010). 1297–1303 20, Anopheles Greenwayodendron rp ln Biol. Plant Trop. 16790–16795 111, 1008–1013 113, 4533–4549 20, 4261–4293 15, Bioinformatics u National eum ´ pl Plant Appl. 1408– 214, | mosquito f10 of 9 Mol. 3, J.

EVOLUTION 78. D. Darriba, D. Posada, A. M. Kozlov, A. Stamatakis, T. Flouri, ModelTest-NG : A new 88. A. R. De La Torre, Z. Li, Y. Van De Peer, P. K. Ingvarsson, Contrasting rates of molecular and scalable tool for the selection of DNA and protein evolutionary models. Mol. Biol. evolution and patterns of selection among gymnosperms and flowering plants. Mol. Evol. 37, 291–294 (2019). Biol. Evol. 34, 1363–1377 (2017). 79. M. D. Pirie, J. A. Doyle, Dating clades with fossils and molecules: The case of 89. R. G. Collevatti, M. P. Telles, J. S. Lima, F. O. Gouveia, T. N. Soares, Contrasting spatial Annonaceae. Bot. J. Linn. Soc. 169, 84–116 (2012). genetic structure in Annona crassiflora populations from fragmented and pristine 80. D. C. Thomas et al., The historical origins of palaeotropical intercontinental disjunc- savannas. Plant Systemat. Evol. 300, 1719–1727 (2014). tions in the pantropical flowering plant family Annonaceae. Perspect. Plant Ecol. Evol. 90. M. Gros-Balthazard et al., The discovery of wild date palms in Oman reveals a complex Systemat. 17, 1–16 (2015). domestication history involving centers in the Middle East and Africa. Curr. Biol. 27, 81. A. Rambaut, A. J. Drummond, D. Xie, G. Baele, M. A. Suchard, Posterior sum- 2211–2218.e8 (2017). marization in Bayesian phylogenetics using Tracer 1.7. Syst. Biol. 67, 901–904 91. T. R. Baker et al., Fast demographic traits promote high diversification rates of (2018). Amazonian trees. Ecol. Lett. 17, 527–536 (2014). 82. K. K. S. Matsunaga, S. R. Manchester, R. Srivastava, D. K. Kapgate, S. Y. Smith, Fossil 92. A. J. Helmstetter, K. Bethune,´ N. G. Kamdem, B. Sonke,´ T. L. P. Couvreur, Data palm fruits from India indicate a Cretaceous origin of Arecaceae tribe Borasseae. Bot. from: Individualistic evolutionary responses of central African rain forest plants J. Linn. Soc. 190, 260–280 (2019). to Pleistocene climatic fluctuations. Dryad. https://doi.org/10.5061/dryad.1rn8pk0rk. 83. J. Heled, A. J. Drummond, Bayesian inference of species trees from multilocus data. Deposited 16 November 2020. Mol. Biol. Evol. 27, 570–580 (2010). 93. A. J. Helmstetter, K. Bethune,´ N. G. Kamdem, B. Sonke,´ T. L. P. Couvreur, AFRO- 84. J. T. Weir, D. Schluter, Calibrating the avian molecular clock. Mol. Ecol. 17, 2321–2328 DYN. Github. https://github.com/ajhelmstetter/afrodyn. Deposited 13 November (2008). 2020. 85. R. P. Brown, Z. Yang, Rate variation and estimation of divergence times using strict 94. A. J. Helmstetter, K. Bethune,´ N. G. Kamdem, B. Sonke,´ T. L. P. Couvreur, and relaxed clocks. BMC Evol. Biol. 11, 271 (2011). Annickia affinis: AFRODYN data. Sequence Read Archive. http://www.ncbi.nlm. 86. W. Thuiller, B. Lafourcade, R. Engler, M. B. Araujo, BIOMOD-A platform for ensemble nih.gov/bioproject/PRJNA541395. Deposited 15 November 2020. forecasting of species distributions. Ecography 32, 369–373 (2009). 95. General Bathymetric Chart of the Oceans (GEBCO) Compilation Group, GEBCO 87. C. Burgarella et al., A Western Sahara centre of domestication inferred from pearl 2020 grid. http://doi.org/10.5285/a29c5465-b138-234d-e053-6c86abc040b9. Accessed millet genomes. Nat. Ecol. Evol. 2, 1377–1380 (2018). 24 November 2020.

10 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.2001018117 Helmstetter et al. Downloaded by guest on October 3, 2021