Using a time-calibrated phylogenetic hypothesis including 768 Gesneriaceae species (out of ~3300 species) and more than 29,000 aligned bases from 26 gene regions, we test Gesneriaceae for diversification rate shifts and the possible proximal drivers of these shifts: geographic distributions, growth forms, and pollination syndromes. Bayesian Analysis of Macroevolutionary Mixtures analyses found five significant rate shifts in Beslerieae, core Nematanthus, core Columneinae, core Streptocarpus, and Pacific Cyrtandra. These rate shifts correspond with shifts in diversification rates, as inferred by Binary State Speciation and Extinction Model and Geographic State Speciation and Extinction model, associated with hummingbird pollination, epiphytism, unifoliate growth, and geographic area. Our results suggest that diversification processes are extremely variable across Gesneriaceae clades with different combinations of characters influencing diversification rates in different clades. Diversification patterns between New and Old World lineages show dramatic differences, suggesting that the processes of diversification in Gesneriaceae are very different in these two geographic regions.
Suppl. Append. 1
DNA sequence data from GenBank.
SupplApp1.csv
Suppl. Append. 2
Scored character states and literature sources.
SupplApp2.csv
Suppl. Append. 3
Biogeographic models and model fit statistics. Results for (A) Gesneriaceae, (B) Gesnerioideae, and (C) Didymocarpoideae. Abbreviations: Par. = free parameters; lnLik = log-likelihood; AIC = Akaike Information Criterion; AICc = Akaike Information Criterion, corrected; ΔAICc = change in AICc; AICw = AIC weights; BIC = Bayesian Information Criterion; ΔBIC = change in BIC; DEC = Dispersal Extinction Cladogenesis model; DIVALIKE = BioGeoBEARS implementation of DIVA model; BAYAREALIKE = BioGeoBEARS implementation of BayArea model; s = subset sympatry; J = founder-event speciation.
SupplApp3.pdf
Suppl. Append. 4
Summary of gene sequences used in the present study.
SupplApp4.pdf
Suppl. Append. 5
Taxonomic comments and conclusions of the revised phylogenetic hypotheses for the Gesneriaceae.
SupplApp5.pdf
Suppl. Append. 6
Stem and crown age estimates for Gesneriaceae clades and outgroups. For comparison, the ages of stems and crowns from Petrova et al. (2015), Perret et al. (2013), Woo et al. (2011), Bell et al. (2010), and Roalson et al. (2008) are provided. Estimation methods are indicated below reference names. Dates are indicated as Mean (Minimum, Maximum). Abbreviations: BEAST, Bayesian Evolutionary Analysis Sampling Trees; PL, penalized likelihood.
SupplApp6.pdf
Suppl. Append. 7
GeoSSE model testing. Results for (A) Africa and Madagascar, (B) Temperate and Tropical Andes, (C) Amazon and Atlantic Brazil, (D) Caribbean and West Indies, and (E) Pacific and Southeast Asia. Gray boxes denote the model with the best-fit. Significance of constrained models versus unconstrained (full) model is assessed as follows: N.S., P>0.1; *, P<0.1; **, P<0.05; ***, P<0.001. Rate categories: λA, speciation in focal area (endemic species); λB, speciation in all other areas combined; λAB, speciation in widespread species; μA, extinction in focal area (endemic species); μB, extinction in all other areas combined; qA, dispersal out of focal area; qB, dispersal out of all other areas into focal area. Abbreviations: Df = degrees of freedom; lnLik = log-likelihood; AIC = Akaike Information Criterion; AICc = Akaike Information Criterion, corrected; ΔAICc = change in AICc; AICw = Akaike weights; LRT = likelihood ratio test; BIC = Bayesian Information Criterion; ΔBIC = change in BIC.
SupplApp7.pdf
Suppl. Append. 8
SIMMAP ancestral character estimations of flower characters. Results for flower color in (A) Gesneriaceae, (B) Gesnerioideae, (C) Didymocarpoideae; corolla shape in (D) Gesneriaceae, (E) Gesnerioideae, (F) Didymocarpoideae; pollination syndrome (G) Gesneriaceae, (H) Gesnerioideae, (I) Didymocarpoideae.
SupplApp8.pdf
Suppl. Append. 9
SIMMAP ancestral character estimations of epiphytism and growth form characters. Results for Gesneriaceae for (A) epiphytism and (B) unifoliate growth form.
SupplApp9.pdf
Suppl. Append. 10
Geiger statistics for phylogenetic signal (λ), trait evolution at speciation (κ), and rate increase over time (δ). Significance of model fit with the addition of λ, κ, and δ parameters against the null model is assessed as follows: N.S., not significant; *, P<0.01; **, P<0.001. Corolla gibbosity is abbreviated "gibb." and epiphytism is abbreviated "epi."
SupplApp10.pdf
Suppl. Append. 11
BiSSE model testing. Results for epiphytism in (A) Gesneriaceae, (B) Gesnerioideae, (C) Didymocarpoideae; ornithophily in (D) Gesneriaceae, (E) Gesnerioideae, (F) Didymocarpoideae; unifoliate growth in (G) Didymocarpoideae. Gray boxes denote the best fitting model. Significance of constrained models versus unconstrained (full) model is assessed as follows: N.S., P>0.1; *, P<0.1; **, P<0.05; ***, P<0.001. Rate categories: λ, speciation; μ, extinction; q, transition rate. In all cases, estimated rates for the characters of interest are indicated by λ1 and μ1, respectively. Abbreviations: Df = degrees of freedom; lnLik = log-likelihood; AIC = Akaike Information Criterion; AICc = Akaike Information Criterion, corrected; ΔAICc = change in AICc; AICw = Akaike weights; LRT = likelihood ratio test; BIC = Bayesian Information Criterion; ΔBIC = change in BIC.
SupplApp11.pdf
Suppl. Figure 1
Gesneriaceae phylogenetic hypothesis. Numbers above branches refer to (A) aLRT and (B) ML bootstrap percentages, respectively. (C) ML phylogram with branch lengths.
SupplFig1.pdf
Suppl. Figure 2
Calibrated Gesneriaceae phylogenetic hypothesis. Bars on branches reflect the 95% confidence interval on the time estimate. Circled numbers at nodes indicate fossil, geologic, and secondary calibration points, respectively.
SupplFig2.pdf
Suppl. Figure 3
Historical biogeographical hypothesis for Gesneriaceae using the best-fit model BAYAREALIKE+s+J. Geographic areas: A, Temperate and Tropical Andes; B = Amazon and Atlantic Brazil; C = Central America and West Indies; E = Europe; F = Africa and Madagascar; I = South Asia; N = East Asia; P = Pacific and Southeast Asia.
SupplFig3.pdf
Suppl. Figure 4
Historical biogeographic hypothesis for Gesnerioideae using the best-fit model BAYAREALIKE+s+J. Geographic regions: A = Tropical Andes; B = Atlantic Brazil; M = Central America; OW = Old World; T = Temperate Andes; W = West Indies, Z = Amazon and Guyana Highlands.
SupplFig4.pdf
Suppl. Figure 5
Historical biogeographic hypothesis for Didymocarpoideae using the best-fit model BAYAREALIKE+s+J. Geographic areas: E = Europe; F = Africa and Madagascar; I = South Asia; N = East Asia; NW = New World; P = Pacific and Southeast Asia.
SupplFig5.pdf
Suppl. Figure 6
Posterior distributions of trait-dependent rates estimated by GeoSSE and area map. A. Geographic areas map showing areas used by GeoSSE comparisons. B. Diversification rate parameters for Africa focal area versus all other areas under the best model (Suppl. Appendix 7A). C. Diversification rate parameters for Andean South America focal area versus all other areas under the best model (Suppl. Appendix 7B). D. Diversification rate parameters for Amazon and Atlantic Brazil focal area versus all other areas under the best model (Suppl. Appendix 7C). E. Diversification rate parameters for Central America and West Indies focal area versus all other areas under the best model (Suppl. Appendix 7D). F. Diversification rate parameters for Southeast Asia and the Pacific focal area versus all other areas under the best model (Suppl. Appendix 7E).
SupplFig6.pdf
Suppl. Figure 7
SIMMAP ancestral character estimations of primary flower color changes on the time-calibrated ML phylogeny. Branch colors approximate primary flower color, with gray branches used to denote white flowers.
SupplFig7.pdf
Suppl. Figure 8
Posterior distributions of trait-dependent rates estimated by BiSSE for the family. Separate New World and Old World plots are presented in Supplemental Fig. 10. A. Diversification rate parameters for epiphytes versus non-epiphytes under the best model (Suppl. Appendix 11A). B. Diversification rate parameters for ornithophilous versus non-ornithophilous under the best model (Suppl. Appendix 11D). C. Diversification rate parameters for unifoliate versus non-unifoliate under the best model (Suppl. Appendix 11G).
SupplFig8.pdf
Suppl. Figure 9
SIMMAP ancestral character estimations of epiphytism changes on the time-calibrated ML phylogeny. Red branches denote epiphytes, and gray branches denote terrestrial plants.
SupplFig9.pdf
Suppl. Figure 10
Posterior distributions of trait-dependent rates estimated by BiSSE separately for Gesnerioideae and Didymocarpoideae.
SupplFig10.pdf