Repeated asynchronous evolution of single‐species endemics of ivies (<i>Hedera</i> L.) in Macaronesian archipelagos

Evolutionary studies of oceanic island endemics are usually focused on lineages that have experienced in situ radiation, while those speciated once the or archipelago but not further diversified (single-species endemics) remain understudied. The Macaronesian archipelagos, Atlantic Ocean, home to significant numbers single-species endemics. genus Hedera L. (12 species) includes three from archipelagos with putatively independent origins. Here, we tested role phylogenetic niche conservatism their evolution. To end, (1) reconstructed spatio-temporal origin ivies using phylogenomics, and (2) assessed climatic during colonization speciation. Azores, Canary Islands, Madeira, western Mediterranean. spp. We used 166 samples representing all species generate genotyping-by-sequencing (GBS) libraries performed time-calibrated phylogenomic biogeographic analyses. Climatic preferences evolution were a geo-referenced database ivy (2,297 records). Independent asynchronous speciation estimated for ivies, resulting different degrees differentiation: H. canariensis displayed an early divergence (7.5–12 Ma) high isolation; azorica had intermediate isolation its closest relative helix (4.4–6.8 maderensis was embedded within climatically similar iberica (2.8–4.6 Ma). A strong signal suggested clade Hedera, where placed. represent evolutionary stages leading emergence appears be involved endemics, first by facilitating colonization, then limiting diversification. Los estudios evolutivos sobre endemismos de islas oceánicas se enfocan habitualmente en linajes que han experimentado radiaciones situ, mientras aquellos especiaron al llegar la isla o archipiélago pero no diversificaron después (endemismos monoespecíficos) sido poco estudiados. archipiélagos macaronésicos, el océano Atlántico, albergan un gran número monoespecíficos. El género (hiedras, 12 especies) incluye tres monoespecíficos macaronésicos con orígenes aparentemente independientes. En este estudio hemos evaluado papel conservación filogenética nicho evolución estos endemismos. Para ello, reconstruido origen espaciotemporal las hiedras macaronésicas mediante filogenómica y estudiado del climático durante su colonización especiación. Islas Canarias, Mediterráneo occidental. Empleamos muestras representaban todas especies para generar bibliotecas genotipado por secuenciación llevamos cabo análisis filogenómicos biogeográficos datados. Las preferencias climáticas estudiaron una base datos georreferenciada occidentales (2.297 registros). Se estimó especiación asíncronas macaronésicas, resultado diferentes grados diferenciación climático: mostró divergencia temprana (7,5–12 elevada nicho; divergió posteriormente pariente más próximo (4,4–6,8 Ma), muestra intermedias; recientemente (2,8–4,6 dentro clado iberica, apenas diferencia climáticamente. sugirieron señal fuerte occidental donde incluyen macaronésicas. representan distintas fases evolutivas conducen emergencia La parece implicada monoespecíficos, primero facilitando islas, limitando diversificación situ. Oceanic islands considered ideal environments study processes due habitat diversity they exhibit (Emerson, 2002). Macaronesia is region located North Ocean close northern Africa southern Europe, including five archipelagos: Savage Islands Cape Verde (Freitas et al., 2016; Helfgott 2000). influence cool humid winds wide altitudinal range give these atypical climates found mainland areas at latitudes, lead variation among (Cropper & Hanna, 2014). These also exceptional variety patterns (Patiño 2014) size, age, distance (Fernández-Palacios 2011). single multiple events, without back-colonization (Kim 2008), spectacular radiations (e.g. Echium, García-Maroto 2009; Cheirolophus, Vitales 2014; Argyranthemum, White 2020; Helianthemum, Albaladejo 2021) Cistus monspeliensis, Coello endemic monospecific genera Visnea Vieraea, Bramwell, 1976; Navaea, Escobar García Canarina canariensis, Mairal 2015; Captoloma canariense, Culshaw 2021; Campylanthus salsoloides, Pokorny Kleinia neriifolia, García-Verdugo, Caujapé-Castells, Illera, 2019). combination geographical variability associated make locations While exhibiting been extensively studied, research focusing (i.e. ‘single-species endemics’ sensu Patiño limited few works mentioned above, despite proportion endemism some Carine Schaefer, 2009). Single-species tend dominate certain ecosystems, such as subtropical laurel forests 2021), which habitats highest Madeira Several hypotheses proposed explain this accumulation (see Fernández-Palacios Some suggest limits diversification tropical plant nearby non-tropical (Crisp Cook, 2012), others indicate long-term stability may limit Alternatively, ecological based competition, preemption (Silvertown, 2004), priority effects (Shaw Gillespie, 2016) saturation (Connell, 1971), proposed. (ivies, Araliaceae) species, typical 2017). Therefore, provide system understand dynamics Each archipelago: Carrière Wild. K. Koch ex A. Rutherford Madeira. Although several dealt history (Ackerfield Wen, 2003; Green 2011; Valcárcel 2003, 2017; Vargas, 2013; Vargas 1999), remains poorly understood. nuclear ITS (Vargas 1999) already colonized each independently. Based ploidy levels, authors polyploid ancestor hexaploid Madeiran diploid ancestors Canarian Azorean species. Subsequent analyses plastid haplotypes supported Mediterranean Iberian Peninsula, European origins (Valcárcel Moreover, presence ancestral haplotype derived might predated other two Despite low resolution Sanger sequencing (Green 2013), placement major clades, morphological traits, relationship between L., African algeriensis Hibberd maroccana McAll., hibernica (G. Kirchn.) Bean (McAll.) Ackerf. J. Wen Nevertheless, regarding adequately evaluated given lack well-supported sister-group relationships. recent next-generation technique clarified (Alonso 2022). This revealed phylogenetically nested south-western described pattern budding 2022), new originated extinction parental (Mayr Bock, An analysis preference warm places isothermality summers. Interestingly, largely overlap acquired ancestor, thus example pre-adaptation. similarity paralleled phenotypic functional traits (e.g., fruit seeds dry mass specific leaf area, Alonso related defining Two appear followed paths archipelagos. hypothetical makes sense thermophilous (Andergassen Bauer, 2002; Parker, 1962) has indicator periods Europe ancient (late Eocene, Axelrod, 1975) times Holocene, Mühller 2005). Indeed, part Tertiary flora (Axelrod, 1975; Herrera, 1992; Verdú 2003). However, latest can characterized temperate (Coca-de-la-Iglesia hypothesis still needs evaluated. Here aimed analyse exemplary case repeated islands. hypothesized general dominated ivies. Our objectives clarify obtaining well-resolved reconstruction GBS technique; compare niches relatives georeferenced (3) context through dating biogeographical total 100 analysed (Figure S1), 85 herbarium (Data S1) previous included 62 individuals 33 populations ivies: 19 populations), (11 populations) 24 (10 populations). 41 18 closely 2017): (2 individuals, 2 (13 8 (26 completed sampling 32 distribution widespread (27 20 hibernica) 13 Asian (5 nepalensis, 4 pastuchovii H.colchica rhombea). Four outgroup (Brassaiopsis, Dendropanax, Eleutherococcus Kalopanax; 2019) included. DNA extraction modified CTAB protocol (Cullings, Doyle Doyle, 1987) concentration Qubit 3.0 Fluorometer. 500 ng per sample prepare library. replicated four replicates al. (2022) confirm consistent libraries. For library preparation, Fernández-Mazuecos (2018), Elshire (2011) modifications (Escudero Grabowski After digestion, ligation pooling, pool purified AMPure XP magnetic beads (Beckman Coulter, USA) 1:1 beads: ratio. Afterwards, 70 amplified PCR cycles NEB Taq 2x Master Mix (New England Biolabs, BioRad MyCycler thermocycler (Bio-Rad Laboratories, USA), 0.8:1 Quantification quality control done 2100 Bioanalyzer (Agilient, CA, USA). Paired-end (Illumina HiSeq X 150 bp; Illumina Inc., Macrogen Inc. (Seoul, South Korea). reads processed ipyrad v.0.9.81 (Eaton, Eaton Overcast, 2020). Demultiplexing separately, remaining assembly steps both datasets combined. Data treated single-end novo done. employed parameters (2020) optimized (2022), clustering threshold 0.85 taxon coverage 15. dataset comprised levels (2x, 4x, 6x 8x; them (2x: canariensis; 6x: maderensis). Assembling mixed presents challenges, methodology developed avoid ploidy-related issues. allow maximum number alleles locus set six. value provided branch support optimization values (2, 4, 6) producing topologies. minimize impact polyploidy-derived paralogy, paralogue filtering documentation. Low recovery loci expected >20 years old, increase uncertainty (Fernández-Mazuecos 2018). preliminary run uncertain (<500 loci), well replicates, excluded final ran concatenation-based Maximum Likelihood (ML) Bayesian Inference (BI). ML RAxML-HPC 8.2.12 (Stamatakis, under GTR+G substitution model, default nucleotide model RAxML commonly 2022; 2018, bootstrap bootstopping criterion (Pattengale 2010). BI ExaBayes v.1.4.1 (Aberer parsimony starting tree, parallel MCMC runs coupled chains 1.2 million generations model. Convergence Tracer 1.7.2 (Rambaut extended majority-rule consensus tree obtained burn-in 25%. coalescent-based SVDquartets (Chifman Kubatko, implemented PAUP 4.a168 (Swofford, 2002) summarized 50% tree. multispecies coalescent option exhaustive quartet used, one partition taxon. environmental conditions occur distributions (realized niche; Pearson Dawson, occurrences built S2) 1; see below; algeriensis, azorica, helix, hibernica, maroccana). macroclimatic variables because availability informativeness about broad-scale downloaded bioclimatic layers 30-second WorldClim 2.1 (Fick Hijmans, 2017) clipped extent raster package (Hijmans, 2021). calculated Pearson's pairwise variable correlations hierarchical cluster (‘hclust’ function, R Core Team, r ≥ 0.80, retained most synthetic biologically relevant S2): bio4 (temperature seasonality), bio8 (mean temperature wettest quarter), bio11 coldest bio16 (precipitation quarter) bio18 warmest quarter). extracted data our package. selected background input weighted (weight = 0 1 background) principal component (PCA) ecospat (Broennimann PCA results plotted alongside kernel density estimates 95% 100% available space ks (Duong, calculating Schoener's D (Schoener, 1970) performing equivalency tests 2012; Warren 2008). Niche test conservatism, is, if significantly higher than random shifts occurred. divergence, lower null simulated randomly reallocating ranges. consider area. non-significant geographically opposite distant When pair tests, it interpreted identical. ‘phyloecospace’ approach (Baldo 2017), applies ‘phylomorphospace’ method mapping diversification; Sidlauskas, 2008) variables. pruned include taxon, positions projected onto tool phytools (Revell, 2012). Pagel's lambda (Pagel, components position additional conservatism. TreePL (Smith O'Meara, 2012) estimate penalized likelihood complete input. Fossils literature unreliable too calibration (Anderson Góis-Marques Additionally, identification trichomes preserved fossils. root corresponding common (31.9–51.3 Ma, establish optimal parameter (gradient-based optimizer opt 2, autodifferentiation optad cross-validation optcvad 1), second sampled cross validation 200,000 iterations 5000 validation. best smoothing parameter, (meanage, 31.9–51.3 Ma; 1e−13). Since does measure uncertainty, additionally only minimum age (minage, 31.9 1, optvad 1e−13) (maxage, 51.3 1e−9) point, estimation nodes. ‘thorough analysis’ runs. Biogeographic BioGeoBEARS (Matzke, 2013) version meanage dated tip (the individual recently diverged sister species). Five defined considering natural ranges aim study: continental area (Europe Africa) eastern (Asia). Dispersal-extinction-cladogenesis (DEC) dispersal-vicariance (DIVALIKE) models +J option, accounting founder-event 2014, time-stratification: crown node (20–14.1 before (14.1–7.1 after (7.1–0 Azores. decided palaeoislands likely emerged (68 65–67 Ma oldest palaeo-island respectively; Carracedo Troll, period constructed ‘manual dispersal multipliers’ ‘allowed areas’ matrices S3) permitted (>14.1 reconstructions Araliaceae family pointed genus. AICc comparison. restricted currently occupied widest range. newly generated 740,539,092 raw reads. Bases score Q20 represented 95.4%. In dataset, 141 included, 7,393 retained, 943,611 bp, 49,885 SNPs, 21,626 informative 70.97% missing sites. percentage reported partly ingroup. mostly resolved relationships level structured main clades 2): (H. colchica, rhombea; BS 100%, posterior probability PP 1.0) maroccana; 73%, 1.0). Within clade, exhibited (BS 87%, formed 93%, highly 95%, clades: helix; maderensis; 0.98). (with support) reciprocally monophyletic except Italian earliest-diverging according analysis. group iberica. Phylogenetic very limited. Coalescent-based topology fully congruent level, nodes polytomy rhombea nepalensis subspecies S3). explained 82.2% 3a). axis (PC1) determined bio11, bio18, (PC2) bio8. PC1 cold rainy winters summer precipitations mild precipitation. PC2 seasonality. (Figures 3a) show extreme upper seasonality season comparison lowest precipitation (positive values) contrast, relatively (negative values). ranged algeriensis-H. (D 0) 0.475; Table 1). pairs azorica-H. canariensis-H. equivalent, neither nor equivalent (Table According maderensis), 0.167 ps 0.1181 pe 0.2772 0.007 0.2376 0.0099 0.012 0.1386 0.015 0.2574 0.0198 0.019 0.2277 0.003 0.2079 0.004 0.023 0.1188 0.2475 0.3762 0.207 0.0990 0.21782 0.229 0.0594 0.28713 0.067 0.0693 0.014 0.008 0.1683 0.052 0.0495 0.107 0.029 0.006