Research ArticleECOLOGY

Quaternary climate changes as speciation drivers in the Amazon floodplains

See allHide authors and affiliations

Science Advances  11 Mar 2020:
Vol. 6, no. 11, eaax4718
DOI: 10.1126/sciadv.aax4718
  • Fig. 1 Geographic distribution, population structure, and species tree of studied lineages.

    Numbers in the map are localities as in table S12. Barplots represent the coefficient of ancestrality per individual obtained with sNMF. Numbers in the species tree are posterior probabilities <0.95, and all other nodes presented maximum posterior probabilities. The red lines in the species tree represent different topologies, sampled in the MCMC. Top: M. lugubris. Middle: T. nigrocinereus/T. cryptoleucus. Bottom: M. assimilis. Gray, Solimões population; pink, Madeira population; green, Tapajós population; purple, Negro population; yellow, Amazonas population (Amazonas + Negro for M. assimilis).

  • Fig. 2 EEMS for M. lugubris, T. nigrocinereus/T. cryptoleucus, and M. assimilis.

    Color bar shows the effective migration rate on a log10 scale relative to the average migration rate over the entire range of the corresponding species. The darker the blue, the higher the average effective migration rate. The darker the orange, the lower the average effective migration rate. The size of the black circles represents the number of sampled individuals in a given locality.

  • Fig. 3 Population relationship graphs for M. lugubris, T. nigrocinereus/T. cryptoleucus, and M. assimilis obtained with TREEMIX.

    Colors represent populations as in Fig. 1. Pie charts represent the percentage of coefficient of ancestrality obtained with sNMF.

  • Fig. 4 Graphical representation of the explored demographic models.

    Isolation-migration model implemented on G-PhoCS (A). Demographic models simulated in FSC (B). Note that for M. lugubris and T. nigrocinereus/T. cryptoleucus, these models included four and five populations, respectively. Names of the models are described in Table 1. The model mig_siz_ALL_G-PhoCS is not depicted in the panel. Topologies tested with Fastsimcoal2 for M. lugubris (C), T. nigrocinereus/T. cryptoleucus (D), and M. assimilis (E). θ, effective population size; Tdiv, divergence time; Sol, Solimões population; Mad, Madeira population; Neg, Negro population; Ama, Amazonas population; Tap, Tapajós population.

  • Table 1 Composite likelihood [max ln(L)], Akaike information criterion (AIC), and relative contribution (weights) for each of the demographic models tested for M. lugubris, T. nigrocinereus/T. cryptoleucus, and M. assimilis.

    In bold: best model. 0mig_0siz, demographic model assuming no migration between populations and no population size changes; 0mig_siz, demographic model assuming no migration between populations but with population size changes; mig_0siz, demographic model assuming migration between populations but no population size changes; mig_siz, demographic model assuming migration between populations and population size changes; mig_siz_ALL_G-PhoCS, demographic model assuming migration between populations and population size changes with all divergence times fixed on G-PhoCS mean estimations (tables S3 to S5); 2*nparam, number of parameters in the models.

    TaxonModelMax ln(L)nparamAICRelative weights
    M. lugubris0mig_0siz−15,003.699969,112.592.72 × 10−201
    0mig_siz−14,986.311769,048.512.23 × 10−187
    mig_0siz−14,880.812168,570.661.29 × 10−83
    mig-siz−14,804.562968,235.527.72 × 10−11
    mig_siz_All_G-PhoCS−14,795.3162768,188.951.00
    T. nigrocinereus/
    T. cryptoleucus
    0mig_0siz−17,880.9131282,368.659.37 × 10−239
    0mig_siz−17,790.1542281,970.692.44 × 10−152
    mig_0siz−17,777.8022681,921.81.00 × 10−141
    mig-siz−17,672.8513681,458.494.08 × 10−41
    mig_siz_All_G-PhoCS−17,633.7653381,272.491.00
    M. assimilis0mig_0siz−3,415.642615,741.615.25 × 10−66
    0mig_siz−3,398.8391215,676.238.27 × 10−52
    mig_0siz−3,347.7671215,441.049.76 × 10−01
    mig-siz−3,346.7761815,448.472.37 × 10−02
    mig_siz_All_G-PhoCS−3,352.5391615,471.013.02 × 10−07
  • Table 2 Parameter values of simulated models in FSC with the best composite likelihood for each of the three studied species and 95% confidence interval (in parentheses) obtained with parametric bootstraps.

    Models are described in Fig. 4. mig_0siz, demographic model assuming migration between populations but no population size changes; mig_siz_ALL_G-PhoCS, demographic model assuming migration between populations and population size changes with all divergence times fixed on G-PhoCS mean estimations (tables S3 to S5); Ne, effective population size (number of diploid individuals). Ancestral population sizes follow the phylogenetic relationship between populations; n-m, ancestral population of Negro and Madeira; n-m-a, ancestral population of Amazonas, Negro, and Madeira; s-m, ancestral population of Solimões and Madeira; a-t, ancestral population of Amazonas and Tapajos; a-t-n, ancestral population of Amazonas, Tapajos, and Negro; m-a/n, ancestral population of Madeira and Amazonas/Negro; anc, ancestral population of the species complex; Mx → y, estimated migration band from x to y forward in time (number of individuals per generation); Tdiv, divergence time in years; s, Solimões population; m, Madeira population; n, Negro population; a, Amazonas population; t, Tapajos population; Epsilon, the ratio between the ancestral population size in relation to the current one when size change was allowed; tau, time of population size change.

    TaxonModelParameters
    M. lugubrismig_siz_
    All_G-PhoCS
    Ne SolimoesNe NegroNe MadeiraNe AmazonasNe n-mNe n-m-aNe ancMs→nMn→sMs→m
    577,831
    (139,204–
    297,980)
    187,753
    (88,058–
    211,804)
    41,753
    (20,230–
    41,237)
    64,241
    (20,490–
    41,960)
    592,914
    (303,123–
    541,514)
    402,991
    (247,756–
    635,255)
    88,469
    (86,784–
    552,158)
    0.04
    (0.00–
    0.24)
    0.50
    (0.64–
    2.56)
    0.02
    (0.00–
    1.49)
    Mm→sMs→aMa→sMn→mMm→nMn→aMa→nMm→aMa→mEpsilon s
    0.02
    (0.01–0.23)
    1.55
    (0.00–3.70)
    0.07
    (0.01–0.10)
    1.67
    (0.03–4.68)
    0.003
    (0.00–0.04)
    1.88
    (0.00–4.35)
    0.001
    (0.002–
    0.08)
    0.10
    (0.09–
    6.39)
    5.77
    (0.00–
    10.49)
    0.22
    (0.22–
    1.38)
    Epsilon nEpsilon mEpsilon aTau sTau nTau mTau a
    3.96
    (0.65–8.74)
    5.50
    (3.65–8.49)
    0.3
    (0.28–5.18)
    20,203
    (18,539–
    105,746)
    21,632
    (2,545–
    59,913)
    20,255
    (11,054–
    31,161)
    20,025
    (2,489–
    23,263)
    T. nigrocinereusmig_siz_
    All_G-PhoCS
    Ne
    Solimoes
    Ne NegroNe
    Madeira
    Ne
    Amazonas
    Ne
    Tapajos
    Ne s-mNe a-tNe a-t-nNe ancMs→n
    231,432
    (88,960–
    238,231)
    196,767
    (50,976–
    178,292)
    127,012
    (40,956–
    134,760)
    100,777
    (30,694–
    77,297)
    116,840
    (20,538–
    141,563)
    311,298
    (63,291–
    407,013)
    447,604
    (312,473–
    707,007)
    211,583
    (85,335–
    550,942)
    356,013
    (93,022–
    534,331)
    1.36
    (1.02–
    2.48)
    Mn→sMs→aMa→sMs→mMm→sMn→aMa→nMn→mMm→nMa→m
    0.52
    (0.25–2.33)
    0.064
    (0.00–1.88)
    1.15
    (0.24–4.18)
    0.30
    (0.00–0.14)
    1.04
    (0.68–3.08)
    0.03
    (0.00–1.55)
    0.57
    (0.17–
    1.13)
    0.22
    (0.00–
    0.69)
    0.08
    (0.00–0.62)
    0.01
    (0.00–
    0.43)
    Mm→aMa→tMt→aEpsilon sEpsilon nEpsilon mEpsilon aEpsilon tTau sTau n
    0.00
    (0.00–5.46)
    0.26
    (0.06–0.51)
    0.01
    (0.00–1.52)
    0.27
    (0.38–5.88)
    1.09
    (2.01–7.15)
    6.27
    (3.62–
    15.70)
    4.43
    (1.43–
    18.93)
    0.34
    (0.22–
    5.06)
    79,733
    (34,392–
    138,714)
    60,520
    (7,006–
    113,621)
    Tau mTau aTau t
    19,378
    (3,871–
    18,503)
    11,112
    (1,429–
    12,285)
    13,927
    (9,354–
    46,256)
    M. assimilismig_0sizNe
    Solimoes
    Ne
    Amazonas/
    Negro
    Ne
    Madeira
    Ne m-a/nNe ancMm→sMs→mMm→a/nMa/n → mMs→a/n
    358,432
    (193,676–
    396,955)
    525,318
    (505,290–
    570,850)
    765,488
    (778,199–
    1,322,214)
    195,303
    (841,052–
    960,669)
    249,877
    (635,438–
    960,239)
    0.34
    (0.00–0.14)
    0.20
    (0.00–
    0.81)
    3.05
    (5.53–
    8.63)
    3.94
    (0.00–0.58)
    2.37
    (3.02–
    6.79)
    Ma/n→sTdiva/n-m
    3.07
    (3.93–5.74)
    166,612
    (24,182–
    173,088)

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/6/11/eaax4718/DC1

    Supplementary Material and Methods

    Study groups and floodplain environments

    Sequence capture of UCEs and bioinformatics

    Synchronicity of size changes among populations

    Summary of results

    Full versus composite likelihood

    Table S1. Summary of bioinformatics results.

    Table S2. Topologies tested for M. lugubris, T. nigrocinereus/T. cryptoleucus, and M. assimilis in FSC.

    Table S3. Demographic parameters estimated for M. lugubris obtained with G-PhoCS.

    Table S4. Demographic parameters estimated for T. nigrocinereus/T. cryptoleucus obtained with G-PhoCS.

    Table S5. Demographic parameters estimated for M. assimilis obtained with G-PhoCS.

    Table S6. Parameter values of simulated models in FSC with the best composite likelihood for M. lugubris.

    Table S7. Parameter values of simulated models in FSC with the best composite likelihood for T. nigrocinereus/T. cryptoleucus.

    Table S8. Parameter values of simulated models in FSC with the best composite likelihood for M. assimilis.

    Table S9. Model selection and parameter estimation of Multi-DICE single-population models for M. assimilis, M. lugubris, and T. nigrocinereus/T. cryptoleucus.

    Table S10. Posterior estimates obtained with the hABC procedure implemented in Multi-DICE using aSFS.

    Table S11. Results of the “leave one out” cross-validations obtained with the hABC procedure implemented in Multi-DICE using the aSFS.

    Table S12. Individuals sampled of M. lugubris, M. assimilis, and T. nigrocinereus/T. cryptoleucus.

    Fig. S1. Estimated effective diversity surface obtained in EEMS for M. lugubris, T. nigrocinereus/T. cryptoleucus, and M. assimilis.

    Fig. S2. Principal components analyses of the simulated single-population models with Multi-DICE.

    Fig. S3. Principal components analyses on the retained simulated SFS of each population.

    Fig. S4. Principal components analysis of simulated and observed aSFSs performed in Multi-DICE.

    References (4450)

  • Supplementary Materials

    This PDF file includes:

    • Supplementary Material and Methods
    • Study groups and floodplain environments
    • Sequence capture of UCEs and bioinformatics
    • Synchronicity of size changes among populations
    • Summary of results
    • Full versus composite likelihood
    • Table S1. Summary of bioinformatics results.
    • Table S2. Topologies tested for M. lugubris, T. nigrocinereus/T. cryptoleucus, and M. assimilis in FSC.
    • Table S3. Demographic parameters estimated for M. lugubris obtained with G-PhoCS.
    • Table S4. Demographic parameters estimated for T. nigrocinereus/T. cryptoleucus obtained with G-PhoCS.
    • Table S5. Demographic parameters estimated for M. assimilis obtained with G-PhoCS.
    • Table S6. Parameter values of simulated models in FSC with the best composite likelihood for M. lugubris.
    • Table S7. Parameter values of simulated models in FSC with the best composite likelihood for T. nigrocinereus/T. cryptoleucus.
    • Table S8. Parameter values of simulated models in FSC with the best composite likelihood for M. assimilis.
    • Table S9. Model selection and parameter estimation of Multi-DICE single-population models for M. assimilis, M. lugubris, and T. nigrocinereus/T. cryptoleucus.
    • Table S10. Posterior estimates obtained with the hABC procedure implemented in Multi-DICE using aSFS.
    • Table S11. Results of the “leave one out” cross-validations obtained with the hABC procedure implemented in Multi-DICE using the aSFS.
    • Table S12. Individuals sampled of M. lugubris, M. assimilis, and T. nigrocinereus/T. cryptoleucus.
    • Fig. S1. Estimated effective diversity surface obtained in EEMS for M. lugubris, T. nigrocinereus/T. cryptoleucus, and M. assimilis.
    • Fig. S2. Principal components analyses of the simulated single-population models with Multi-DICE.
    • Fig. S3. Principal components analyses on the retained simulated SFS of each population.
    • Fig. S4. Principal components analysis of simulated and observed aSFSs performed in Multi-DICE.
    • References (4450)

    Download PDF

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article