Ancestrality and Evolution of Trait Syndromes in Finches (Fringillidae)

Received: 8 March 2017 | Revised: 27 July 2017 | Accepted: 19 August 2017 DOI: 10.1002/ece3.3420

ORIGINAL RESEARCH

Ancestrality and evolution of trait syndromes in finches (Fringillidae)

1Muséum National d’Histoire Naturelle, Mécanismes adaptatifs et évolution Abstract (MECADEV UMR 7179), Sorbonne Species traits have been hypothesized by one of us (Ponge, 2013) to evolve in a cor- Universités, MNHN, CNRS, Brunoy, France related manner as species colonize stable, undisturbed habitats, shifting from “ances- 2Muséum National d’Histoire Naturelle, Service de Systématique Moléculaire, Paris, tral” to “derived” strategies. We predicted that generalism, r-selection, sexual France monomorphism, and migration/gregariousness are the ancestral states (collectively 3Muséum National d’Histoire naturelle, Institut called strategy A) and evolved correlatively toward specialism, K-selection, sexual de Systématique, Évolution, Biodiversité (ISYEB UMR 7205), CNRS, MNHN, UPMC, dimorphism, and residence/territoriality as habitat stabilized (collectively called B EPHE, Sorbonne Universités, Paris, France strategy). We analyzed the correlated evolution of four syndromes, summarizing the 4Muséum National d’Histoire Naturelle, Centre d’Écologie et des Sciences de la Conservation covariation between 53 traits, respectively, involved in ecological specialization, r-K (CESCO UMR 7204), MNHN, CNRS, UPMC, gradient, sexual selection, and dispersal/social behaviors in 81 species representative Sorbonne Universités, Paris, France of Fringillidae, a bird family with available natural history information and that shows Correspondence variability for all these traits. The ancestrality of strategy A was supported for three of Jean-François Ponge, Muséum National d’Histoire Naturelle, CNRS UMR 7179, the four syndromes, the ancestrality of generalism having a weaker support, except for Brunoy, France. the core group Carduelinae (69 species). It appeared that two different B-strategies Email: [email protected] evolved from the ancestral state A, both associated with highly predictable environments: one in poorly seasonal environments, called B1, with species living permanently in lowland tropics, with “slow pace of life” and weak sexual dimorphism, and one in highly seasonal environments, called B2, with species breeding out-of-the-tropics, migratory, with a “fast pace of life” and high sexual dimorphism.

KEYWORDS ancestral state, evolution, Fringillidae, phylogeny, phylogeography, strategies

1 | INTRODUCTION collectively called strategy A and previously called “barbarian” strategy in Ponge (2013), includes generalism, r-selected traits, dispersiveness/ Ponge (2013) suggested that a wide array of traits related to eco- gregariousness, and more generally traits under natural selection (as logical niche requirements (e.g., specialist vs. generalist), life history opposed to sexual/social selection). It is associated with disturbed (e.g., K- vs. r-selection), behavior (e.g., residents vs. dispersers, ter- and unpredictable environments (Sallan & Galimberti, 2015; Sheldon, ritorial vs. gregarious), and selection mode (e.g., sexual vs. natural 1996). Conversely, opposite trait modalities (values taken by a given selection) co-evolved along gradients of environmental predictability, trait under selection) include specialism, K-selected traits, residence/ forming two suites of generalized syndromes or evolutionary strate- territoriality, and traits under sexual/social selection, here collectively gies. According to this theory, the ancestral suite of syndromes, here called strategy B and previously called “civilized” strategy in Ponge

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2017 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.

www.ecolevol.org | 9935 Ecology and Evolution. 2017;7:9935–9953. 9936 | PONGE et al.

(2013). They are predicted to evolve in stable environments with a partial presence). When the same character exhibited different trait high level of exploitative competition through character displacement modalities (e.g., foraging height), data were scored according to an (Brown & Wilson, 1956) and convergence (Laiolo et al., 2015), allowing arbitrary scale (e.g., foraging height was scored 1 for ground, 2 for species to segregate and become finely tuned to their environment. shrubs, 3 for trees). The variance of these numerical values (weighted Given the number of evolutionary steps necessary for being finely by their occurrence in literature) was complemented to 1 and was tuned to the environment (Poisot, Bever, Nemri, Thrall, & Hochberg, used as synthetic index of specialization (e.g., foraging height spe- 2011), trait modalities of strategy B would thus be in derived posi- cialization in Appendix S3). Altitudinal specialization of a given spe- tions in phylogenetic trees (see Raia, Carotenuto, Passaro, Fulgione, cies was measured by dividing its altitudinal range (maximum minus & Fortelius, 2012; for a discussion about the derived position of eco- minimum altitude) by the maximum altitude found in our dataset (i.e., logical specialization). Conversely, trait modalities of strategy A would 4,950 m) and complementing to 1 this ratio. The same method based thus be in ancestral position along phylogenetic trees. This suggests on weighted occurrence was used to measure the intra-specific vari- a macroevolutionary trade-off between the need to move and/or re- ability of a behavioral trait when it exhibited different trait modali- produce and the need to specialize to stable resources and habitats ties (e.g., breeding dispersion and migration). Missing data (20 ± 23% (see Berg et al., 2010; Bonte et al., 2012; Rottenberg, 2007). Tropical of total records, Appendix S3) were interpolated according to a lowland areas, in particular tropical rainforests, known for their greater nearest-neighbor method (Holmes & Adams, 2002) prior to Principal stability and lower energetic cost for organisms compared to areas Component Analysis (PCA). with seasonal stress (Janzen, 1967), are thus expected to harbor more B-strategy traits, thought to be derived (Cardillo, 2002). 2.2 | Phylogenetic tree In this article, we want to test simultaneously, for the first time, in a monophyletic group, the following hypotheses: generalism, r-selection, The Fringillidae phylogeny by Zuccon et al. (2012) included 93 in- natural selection, and dispersiveness/gregariousness (strategy A) are group taxa of 205 extant species (sensu Dickinson & Christidis, 2014), ancestral and tend to shift toward derived (strategy B) attributes in representing all major lineages, genera, and species groups and was the course of evolution (Hypothesis 1), and these four trait syndromes based on a combination of three nuclear and two mitochondrial genes. evolve correlatively (Hypothesis 2). At last, we hypothesize that state Here, we used the substitution rates for a clade of Fringillidae, the B traits are favored by two kinds of predictable environments: stable drepanids, obtained by Lerner, Meyer, James, Hofreiter, and Fleischer and benign tropical environments favor state B traits only, while stable (2011), to time-calibrate the phylogeny of Fringillidae. Because 12 but seasonal or harsh environments favor a combination of state A and species had <50% of scored ecological traits, they were excluded state B traits (Hypothesis 3). To test these hypotheses, we chose to from the analysis. The time-calibrated phylogeny was generated with work with birds because their traits are particularly well-documented, BEAST 1.8.0 (Drummond & Rambaut, 2007), as implemented in the thanks to a long tradition of natural history documentation by orni- CIPRES Science Gateway (Miller, Pfeiffer, & Schwartz, 2010). We as- thologists (Schmeller, Henle, Loyau, Besnard, & Henry, 2012). We se- sumed an uncorrelated molecular clock model, a speciation yule tree lected the bird family Fringillidae (i.e., true finches) because (1) it is prior and a GTR+Γ or GTR+Γ+I substitution model (Posada & Crandall, distributed worldwide, (2) it is known for its wide range of variation 2001), for nuclear or mitochondrial genes, respectively. Markov chain in terms of ecological specialization, life history, social/dispersal be- Monte Carlo (MCMC) simulations were run for 100 million genera- havior, and sexual dimorphism (Clement, Harris, & Davis, 2011), and tions with sampling every 10,000 generations. The convergence was (3) its phylogeny is well-established on the basis of representatives of evaluated in Tracer 1.6, and the maximum clade credibility tree was all existing lineages (Zuccon, Prŷs-Jones, Rasmussen, & Ericson, 2012). summarized using TreeAnnotator v1.8.0 (both packages implemented This bird family is a good model to follow the evolution of multiple in BEAST), excluding the first 25% trees as burn-in. suites of traits, as attested by several studies performed on cardueline This phylogeny was used for the reconstruction of ancestral syn- finches (Badyaev, 1997a,b,c; Badyaev & Ghalambor, 1998; Badyaev, dromes, for the phylogenetic Principal Component Analysis (pPCA) Hill, & Weckworth, 2002). and for modeling correlations between transitions from ancestral to derived states, as explained in the following sections. The tree that we obtained covers about 40% of Fringillidae diversity. No global phylog- 2 | MATERIAL AND METHODS eny for the entire family is available, but a number of complete or al- most complete phylogenies have been published for some subclades/ 2.1 | Data collection and preparation genera, for example, Pyrrhula (Töpfer et al., 2011), Carpodacus sensu The literature on Fringillidae (153 references, Appendix S1) was used lato (Tietze, Päckert, Martens, Lehmann, & Sun, 2013), Haemorhous to collect the necessary natural history data (Appendix S2). Fifty-three (Smith, Bryson, Chua, Africa, & Klicka, 2013), and Spinus (Beckman variables were documented (Appendix S3). For each quantitative & Witt, 2015). We evaluated the representativeness of the family variable (measurements, count data) and each species, we used the diversity in our phylogeny by a qualitative comparison against a su- weighted arithmetic mean value across all available data. Qualitative permatrix tree. Using the dataset used by Zuccon et al. (2012) as start- data (coded as “Yes” or “Yes minus” in Appendix S2) were coded ing point, we assembled a supermatrix for three nuclear introns and as 1 for the presence of the character, 0 for its absence, 0.5 for its five mitochondrial genes by comparison with the datasets of other PONGE et al. | 9937 published phylogenies and searching in Genbank for any additional 1) a model where transitions in syndrome i were independent from finch sequence (Appendix S4). The supermatrix tree was estimated transitions in syndrome j (independent model) and 2) a model where by maximum likelihood using RAxML version 7.0.3 (Stamatakis, 2006), transitions in syndrome i depended on transitions in syndrome j (de- applying a gene partition, a GTR+Γ+I model, and random starting tree, pendent model, Pagel & Meade, 2006). We then selected the best with α-shape parameters, GTR-rates, and empirical base frequencies model by performing a likelihood ratio test (LRT). If traits evolve under estimated and optimized for each partition. Nodal support was esti- a dependent model, this means that their evolution is correlated, in a mated using 100 bootstrap replicates. way depicted by the estimated transition rates (see section 3). We calculated phylogenetic distances between two species in the supermatrix tree as the sum of branch lengths on the smallest path 2.5 | Phylogenetic principal components analysis that connects the two species. Next we calculated the mean phylo- (pPCA) genetic distance (MPD) between two species among the 81 species selected for our analysis. We analyzed whether the selected species The four syndromes (the first principal components of the four original were more phylogenetically clustered than expected randomly using PCAs) were used as continuous variables in a pPCA (Jombart, Pavoine, a null model. We simulated 1,000 null communities by selecting ran- Devillard, & Pontier, 2010), allowing to discern a phylogenetic signal domly 81 species in the supermatrix tree (outgroup excluded). The p- common to all four syndromes. Phylogenetic signal (autocorrelation) value (quantile) was calculated as the proportion of the MPD of the was tested on each synthetic variable by Abouheif’s test (Abouheif, null communities that were lower than the observed MPD. We ob- 1999; Pavoine, Ollier, Pontier, & Chessel, 2008). These methods were tained observed MPD = 0.319 and p = .837. Our selection of 81 spe- performed using the adephylo package of R (R Core Team, 2016). The cies therefore represents a random sample of Fringillidae phylogenetic reconstruction of ancestral states of the four syndromes suggested diversity. that strategy B should be subdivided in two groups. Thereby the first three components of pPCA were used to split the species in three strategies A, B1, and B2 by k-means clustering. 2.3 | Construction of syndromes

We selected four sets of variables that together describe a common 2.6 | Association of syndromes with tropical affinity life history, ecological, or behavioral pattern, which we call syndromes according to Sih, Bell, and Johnson (2004). Each one describes either The tropical affinity of species was determined according to geo- r-K-gradient, ecological specialization, sexual selection, or dispersal/ graphical records (Appendix S2). Species were considered as having social behavior (Appendix S3). To maximize independence between a tropical affinity when more than half of their distribution area was syndromes, we took care not to include the same variable in the cal- located in the tropics. To test the association of each syndrome with culation of different syndromes. For each syndrome, all attributed tropical affinity while correcting for phylogenetic autocorrelation, a variables were submitted to PCA, with Spearman’s coefficient as a phylogenetic ANOVA was performed on the variables describing the measure of correlation, to extract the correlated variation between four syndromes (first principal component of PCA, see section 3), dominant variables, that is, the functions of the principal components. using tropical affinity as a fixed, two-level factor, with the function Ideally, only the first principal component would have a distinctively aov.phylo implemented in the R package geiger (Harmon, Weir, Brock, high eigenvalue (Sih et al., 2004) and could be used as single proxy Glor, & Challenger, 2008). for ecological specialization, r-K gradient, sexual selection, and dispersal/social behavior, respectively. For each synthetic variable, species 3 | RESULTS were split into two groups by k-means clustering (Steinhaus, 1956), maximizing the ratio of between-group versus within-group variance. 3.1 | Phylogeny and representativeness of species These two groups were considered as two levels of each syndrome, diversity which could then be treated as a discrete variable. Calculations were performed using XLSTAT® for Excel® (Addinsoft®, Paris). As expected, the topology of the time-calibrated phylogeny is congruent with that obtained by Zuccon et al. (2012). Also the topology of the tree generated using the supermatrix, which includes 169 of 2.4 | Reconstruction of the ancestral states of 204 (82%) species, is largely congruent and recovers the same major syndromes clades. Minor differences involve a few nodes with low or no support Reconstruction of ancestral states (H1) and test for correlated evolu- and the branching order of a few clades. The comparison of the time- tion of the four syndromes (H2) were performed using BayesTraits 2.0 calibrated and supermatrix-based trees confirms that the 81 species (Pagel & Meade, 2006; Pagel, Meade, & Barker, 2004). Specifically, for in the time-calibrated tree have been sampled across the entire family, each syndrome ancestral states and transition rates between states that all major lineages are represented in the reduced dataset and that were estimated by maximum likelihood (H1). To test for correlated the species retained in the subsequent analysis are a fair representa- evolution of the four syndromes (H2), for each pair of syndrome i and j tion of the family diversity and disparity. This result is a logical conse- we computed the transition rates between states of syndrome i under quence of the taxa choice operated by Zuccon et al. (2012), which was 9938 | PONGE et al. driven by taxonomic incentive without taking into account ecological specific index of altitudinal specialization, from species restricted to or life history traits. lowlands (lowland specialists) to species with a large altitudinal range (altitudinal generalists). We selected PC1 as the synthetic variable summarizing best the information given by all traits describing eco- 3.2 | Construction of proxies for syndromes logical specialization (called PC1spec hereafter).

3.2.1 | Ecological specialization 3.2.2 | r-K gradient The PCA bi-plot for ecological specialization shows that all indicators of specialization (“foraging height specialization,” “food specializa- The PC1-PC2 bi-plot for the “r-K-gradient” syndrome (Figure 1b) tion,” “habitat specialization,” “nest height specialization,” “altitudinal shows that “nesting stage” and “clutch size” (see Appendix S2 for specialization”) are positively correlated with the first principal com- definitions) display the highest correlation value with PC1, which ex- ponent PC1 (explaining 20% of the total variance) while indicators of plains 36% of the total variation (rs ≈ 0.8, Table 1). All indicators of generalism (tolerance), to the exception of “drought tolerance,” are K-selection (“nesting stage length,” “incubation length,” “number of negatively correlated with it (Figure 1a, Table 1). The second principal broods per season”) but one (“life duration”) are positively correlated component (PC2, explaining 19% of the total variance) opposes “cold with PC1, while indicators of r-selection (“clutch size” and “metabolic tolerance” to “altitudinal specialization” and can be interpreted as a rate per unit body mass”) are negatively correlated with it. PC2, which ) ) (a) 1.0 (b) 1.0 Metabolic rate per unit Cold tolerance body mass PC2 (23%

0.8 PC2 (19% 0.8

Broods per season

0.6 0.6

Nest height specialization

0.4 0.4 Clutch size

Breeding season 0.2 0.2 Habitat specialization Drought tolerance Nesting stage

0.0 0.0 PC1 (20%) –1.0 –0.8 –0.6 –0.4 –0.2 0.00.2 0.40.6 0.81.0 –1.0–0.8–0.6–0.4–0.20.0 0.20.4 0.60.8 1.0 Foraging height PC1 (36%) specialization –0.2 –0.2 Incubation Food specialization Shy –0.4 –0.4 Disturbance tolerance

–0.6 –0.6 Life duration

–0.8 –0.8 Altitudinal specialization

–1.0 –1.0

1.0 1.0 ) (c) ) (d)

Gregariousness plasticity 0.8 0.8 PC2 (15%

PC2 (13% Male plumage brightness

Sexual dimorphism 0.6 present 0.6 Frequency range Female plumage Number of notes Head dichromatism brightness Migration Nests dispersed % melanin 0.4 0.4 Breast dichromatism Mixed parties Breeding dispersion Nest building by males Bill relative length plasticity Song mimicry Rump dichromatism Monogamy 0.2 Duration of juvenile Migration plasticity 0.2 Tarsus relative length Territorial plumage % carotenoid Wing relative length Incubation by males 0.0 Song length 0.0 Female plumage –1.0 –0.8 –0.6 –0.4 –0.2 0.00.2 0.40.6 0.81.0 –1.0 –0.8 –0.6 –0.4 –0.2 0.00.2 0.40.6 0.81.0 drabness PC1 (28%) PC1 (36%) Breeding dispersion –0.2 Tail relative length –0.2 Feeding of youngs by males Juvenile plumage Nest outer diameter Gregariousness distinctness –0.4 –0.4

Size Weight

–0.6 –0.6

–0.8 –0.8

–1.0 –1.0

FIGURE 1 Projection in the plane of the first two principal components of four PCAs of the trait variables describing (a) ecological specialization, (b) r-K gradient, (c) sexual selection, (d) social/dispersal behavior PONGE et al. | 9939 explains 23% of the total variation, is more trait specific, opposing gradient from resident/territorial/solitary species to dispersive/poorly “metabolic rate per unit weight” to “life duration,” displaying the trivial territorial/gregarious species. PC2 (explaining only 15% of total varia- fact that species with high metabolic rate have low life duration. It tion) was trait specific, opposing “gregarious plasticity” to “gregarious- should be noted that the two variables which are highly correlated ness.” PC1 was selected as the synthetic variable best describing the with PC2 were poorly documented (for only 38% and 28% of spe- dispersal/social behavioral syndrome, hereafter called PC1beh. cies, respectively, see Appendix S3) compared to the other variables describing the r-K gradient (68% to 96%). We thus selected PC1 as a 3.3 | Ancestrality and evolution of syndromes (H1) synthetic descriptor for the r-K-gradient, which will be further called PC1rK. PC1spec was used to split the 81 species into 33 “specialists” and 48 “generalists” (Fig. S1). The transition rate from generalism toward specialism is 0.087, and 0.097 in the reverse direction. The ancestral state 3.2.3 | Sexual selection of ecological specialization is therefore uncertain (Figure 2). However, In the PC1-PC2 bi-plot (Figure 1c), the variables “rump dichromatism,” the most likely ancestral state for clade 1 (Fringillinae), the most basal “% carotenoid,” and “nest outer diameter” exhibit the highest correla- clade, and for the core group (Carduelinae) is generalism, with a prob- tion value (rs ≈ 0.8) with PC1 (explaining 28% of the total variance). ability of 0.82 and 0.66, respectively, as shown by pie charts (Figure 2). All variables featuring sex dimorphism (including “% carotenoid”) as Uncertainty at the root of the phylogenetic tree is mainly due to clade well as those related to size (“size,” “weight,” “nest diameter”) are 2 (Euphoninae) which is currently composed of specialists. Within the positively correlated with PC1 while variables featuring the absence core group (Carduelinae) specialism is nearly always in a derived posi- or weakness of sex dimorphism (“female plumage brightness,” “nest tion, and the only cases of reversal being within clade 5, while most building by males”), including “% melanin,” are negatively correlated extant species in clades 3, 8, 9, 10, 13, 14, and 15 are generalists. with it (Table 1), suggesting that PC1 can be used as a proxy for sexual PC1rK was used to split the 81 species into 22 “K-selected” and (Fisherian) selection. However, it must be noted that variables meas- 59 “r-selected” species (Fig. S1). Maximum-likelihood reconstruction uring the elaboration of male song (“song length,” “frequency range,” showed that r-selection is the most likely ancestral state for the whole “number of notes”) are negatively correlated with PC1, suggesting group and for most clades, with a transition rate of 0.031 from r- that this component opposes two bodies of sexual-selected traits, selection toward K-selection and nil in the reverse direction (Figure 3). rather than sexual versus natural selection, as previously thought K-selection appears as a derived state within Euphoninae tanagers (Ponge, 2013). It must also be noted that among variables featuring (clade 2), Hawaiian honeycreepers (clade 4), African serins (clade 9), the shape of the body, “wing relative length” and “tail relative length” and crossbills (clade 11). (on the negative side) are opposed to “tarsus relative length” and “bill PC1sex was used to split the 81 species into 25 “sexually dimorphic” relative length” (on the positive side) along this axis, suggesting that and 56 “sexually monomorphic” species (Fig. S1). Sexual monomor- the development of appendages used for flight (wing, tail) is opposed phism is the most likely ancestral state for the whole group and for most to those used for other mechanical functions (bill, tarsus) along this clades, with a transition rate of 0.041 from monomorphism toward di- gradient of sexual dimorphism. morphism and nil in the reverse direction (Figure 4). Sexual dimorphism The significance of this complex factor, which we will provisionally appears as a derived state limited to rosefinches (clades 5 and 7), cross- refer to “sexual dimorphism” rather than to “sexual selection,” will be bills (clade 11) and isolated species within some other clades. developed in section 4. PC1beh was used to split the 81 species into 31 “resident/territo- The second principal component, explaining only 13% of the total rial” and 50 “dispersive/gregarious” species (Fig. S1). Dispersiveness/ variation, opposes “size” and “weight” to variables describing sexual gregariousness is the most likely ancestral behavioral state, with dimorphism (all with positive scores along PC1), indicating that among a transition rate of 0.077 from dispersiveness/gregariousness to- sexually dimorphic species smaller species exhibit a lowest level of ward residence/territoriality and only 0.033 in the reverse direction sexual differentiation. This subsidiary PC2 component represented a (Figure 5). Similarly to K-selection, residence/territoriality appears as a weak proportion of trait variation linked to sexual selection and is not derived state in Euphoninae tanagers (clade 2), Hawaiian honeycreep- used in the following analyses. ers (clade 4), and African serins (clade 9), but not in crossbills (clade 11). Derivation toward residence/territoriality is also apparent within rosefinches belonging to clade 5. 3.2.4 | Dispersal/social behavior

PC1 explains 36% of the total variation (Figure 1d). This first princi- 3.4 | Correlated evolution of syndromes (H2) pal component is highly correlated with “breeding dispersion” on the positive side and “gregariousness” and “breeding dispersion plastic- 3.4.1 | Modeling and graphical study of the ity” on the negative side (r ≈ ±0.8, Table 1). Indicators of territoriality s correlated evolution of syndromes (“territorial,” “nests dispersed,” “breeding dispersion”) exhibit positive scores, in opposition to “migration,” “gregariousness” and two indica- We modeled correlated transitions between states (ancestral vs. de- tors of plasticity on the negative side. PC1 represents a behavioral rived) in couples of syndromes (Table 2). Several syndromes tend to 9940 | PONGE et al.

TABLE 1 List of variables selected for the description of four syndromes, with their loadings (= Spearman’s rank correlation coefficients) along the first PCA component (one separate analysis per syndrome)

Ecological specialization syndrome PC1spec r-K gradient syndrome PC1rK

Disturbance tolerance −0.39 Clutch size −0.76 Cold tolerance −0.17 Metabolic rate per unit body mass −0.31 Shy 0.03 Life duration −0.24 Drought tolerance 0.22 Broods per season 0.41 Altitudinal specialization 0.30 Incubation 0.68 Nest height specialization 0.48 Breeding season 0.70 Food specialization 0.55 Nesting stage 0.81 Habitat specialization 0.63 Foraging height specialization 0.74

Sexual dimorphism syndrome PC1sex Dispersal/social behavior syndrome PC1beh

Wing relative length −0.63 Gregariousness −0.80 Song length −0.62 Breeding dispersion plasticity −0.75 Female plumage brightness −0.60 Migration plasticity −0.69 Frequency range −0.56 Migration −0.57 % melanin −0.45 Mixed parties −0.23 Tail relative length −0.41 Gregariousness plasticity 0.26 Number of notes −0.36 Territorial 0.40 Nest building by males −0.32 Nests dispersed 0.64 Juvenile plumage distinctness −0.27 Breeding dispersion 0.76 Monogamy −0.27 Song mimicry −0.23 Feeding of youngs by males −0.06 Incubation by males 0.26 Bill relative length 0.27 Male plumage brightness 0.28 Tarsus relative length 0.32 Female plumage drabness 0.53 Weight 0.58 Sexual dimorphism present 0.59 Size 0.62 Duration of juvenile plumage 0.71 Head dichromatism 0.72 Breast dichromatism 0.75 Nest outer diameter 0.78 % carotenoid 0.79 Rump dichromatism 0.81

FIGURE 2 Reconstruction by maximum-likelihood inference of the ancestral state of the ecological specialization syndrome, using the distribution of species in two groups (see Fig. S1a). Generalism (state A) is in gray while specialism (state B) is in black. Transition rates from one group to the other at the base of the phylogenetic tree are indicated with arrows of proportional thickness. At each node of the phylogenetic tree, a pie chart indicates the confidence given to specialism or generalism as the ancestral state. Transition rates Clade numbers according to Zuccon et al. (2012) are indicated in a column on the right side of the figure. A molecular clock is represented at the bottom of the figure. Geographic distribution and tropical affinity are indicated for each species (see included legend for the significance of codes) PONGE et al. | 9941

Passer luteus Passer montanus Petronia petronia Montifringilla ruficollis Motacilla alba Anthus trivialis Plectrophenax nivalis Ammodramus humeralis Parula pitiayumi Leistes superciliaris Fringillinae Fringillacoelebs PAL 1 Fringillamontifringilla PAL Chlorophoniacyanea SAM Euphoniinae Euphoniamusica NAM SAM Euphonia chlorotica SAM Fringillidae Euphoniacayennensis SAM 2 Euphoniaxanthogaster SAM Euphonia minuta NAM SAM Euphonialaniirostris NAM SAM Euphoniaviolacea SAM Mycerobascarnipes PAL Eophona migratoria PAL nam 3 Hesperiphona vespertina NAM Coccothraustescoccothraustes PAL Paroreomyzamontana HAW Loxioidesbailleui HAW 4 Chlorodrepanis virens HAW Erythrina erythrina PAL ORI Haematospizasipahi PAL ori Carduelinae Carpodacus synoicus PAL ara Carpodacus sibiricus PAL Carpodacus puniceus PAL Carpodacus roseus PAL 5 Carpodacus thura PAL Carpodacus rubicilloides PAL Carpodacus rubicilla PAL Carpodacus pulcherrimus PAL Carpodacus rodochroa PAL Carpodacus rhodochlamys PAL Pinicolaenucleator PAL NAM Procarduelis nipalensis PAL ORI Pyrrhula pyrrhula PAL Pyrrhulaerythaca PAL ORI Rhodopechyssanguineus PAL Bucanetes githagineus PAL ARA AFR Eremopsaltria mongolica PAL 6 Callacanthisburtoni Agraphospiza rubescens PAL Leucostictenemoricola PAL Leucostictebrandti Leucostictetephrocotis NAM Leucostictearctoa Haemorhous mexicanus NAM 7 Haemorhous purpureus NAM Rhodospiza obsoleta PAL ARA 0.087 Chlorischloris PAL Chlorissinica PAL ori 8 Chlorismonguilloti ORI 0.097 Chlorisspinoides PAL ori State A State B Chlorisambigua PAL ORI Spinus olivaceus PAL Crithagra striolata AFR Crithagra burtoni AFR Crithagra mennelli AFR 9 Leistes superciliaris = Outgroup species Crithagra sulphurata AFR Crithagra rufobrunnea AFR Acanthis hornemanni = Generalist species Crithagra citrinelloides AFR Crithagra mozambica AFR Loxiapytyopsittacus = Specialist species Crithagra leucopygia AFR Linaria cannabina PAL 10 Linaria flavirostris PAL = In the tropics Acanthis flammea PAL NAM Acanthis hornemanni PAL NAM = Out of the tropics Loxialeucoptera PAL NAM 11 Loxia pytyopsittacus PAL PAL pal = Palearctic distribution (lower case = minor) Loxiacurvirostra PAL ORI NAM sam Cardueliscarduelis PAL Carduelis citrinella PAL 13 ORI ori = Oriental distribution Serinuscanicollis AFR Serinus syriacus PAL ARA ara = Arabian distribution Serinuspusillus PAL 14 Serinus serinus PAL AFR afr = African distribution Serinuscanaria PAL Spinus psaltria NAM SAM NAM nam = North-Central American distribution Spinus tristis NAM Spinus spinus PAL ori SAM sam = South American distribution Spinus pinus NAM 15 Spinus barbatus SAM HAW = Hawaiian distribution Spinus cucullatus SAM Spinus magellanicus SAM Spinus atratus SAM

15 12.5 10 7.5 5 2.5 0Ma 9942 | PONGE et al.

Passer luteus Passer montanus Petronia petronia Montifringillaruficollis Motacilla alba Anthus trivialis Plectrophenaxnivalis Ammodramus humeralis Parula pitiayumi Leistes superciliaris Fringillinae Fringillacoelebs PAL 1 Fringillamontifringilla PAL Chlorophoniacyanea SAM Euphoniinae Euphoniamusica NAM SAM Euphonia chlorotica SAM Fringillidae Euphoniacayennensis SAM 2 Euphoniaxanthogaster SAM Euphonia minuta NAM SAM Euphonialaniirostris NAM SAM Euphoniaviolacea SAM Mycerobascarnipes PAL Eophona migratoria PAL nam Hesperiphona vespertina NAM 3 Coccothraustes coccothraustes PAL Paroreomyzamontana HAW Loxioidesbailleui HAW 4 Chlorodrepanis virens HAW Erythrina erythrina PAL ORI Haematospizasipahi PAL ori Carduelinae Carpodacus synoicus PAL ara Carpodacus sibiricus PAL Carpodacus puniceus PAL Carpodacus roseus PAL 5 Carpodacus thura PAL Carpodacus rubicilloides PAL Carpodacus rubicilla PAL Carpodacus pulcherrimus PAL Carpodacus rodochroa PAL Carpodacus rhodochlamys PAL Pinicolaenucleator PAL NAM Procarduelis nipalensis PAL ORI Pyrrhula pyrrhula PAL Pyrrhulaerythaca PAL ORI Rhodopechyssanguineus PAL Bucanetes githagineus PAL ARA AFR Eremopsaltria mongolica PAL 6 Callacanthisburtoni PAL Agraphospiza rubescens PAL Leucostictenemoricola PAL Leucostictebrandti PAL Leucostictetephrocotis NAM Leucostictearctoa PAL Haemorhous mexicanus NAM Haemorhous purpureus NAM 7 Rhodospiza obsoleta PAL ARA Chlorischloris PAL 0.031 Chlorissinica PAL ori 8 Chlorismonguilloti ORI Chlorisspinoides PAL ori State A Chlorisambigua PAL ORI State B Spinus olivaceus PAL Crithagra striolata AFR Crithagra burtoni AFR Crithagra mennelli AFR 9 Leistes superciliaris = Outgroup species Crithagra sulphurata AFR Crithagra rufobrunnea AFR Acanthis hornemanni = r-selected species Crithagra citrinelloides AFR Crithagra mozambica AFR Loxiapytyopsittacus = K-selected species Crithagra leucopygia AFR Linaria cannabina PAL 10 Linaria flavirostris PAL = In the tropics Acanthis flammea PAL NAM Acanthis hornemanni PAL NAM = Out of the tropics Loxialeucoptera PAL NAM 11 Loxia pytyopsittacus PAL PAL pal = Palearctic distribution (lower case = minor) Loxiacurvirostra PAL ORI NAM sam Cardueliscarduelis PAL Carduelis citrinella PAL 13 ORI ori = Oriental distribution Serinuscanicollis AFR Serinussyriacus PAL ARA ara = Arabian distribution Serinuspusillus PAL 14 Serinusserinus PAL AFR afr = African distribution Serinus canaria PAL Spinus psaltria NAM SAM NAM nam = North-Central American distribution Spinus tristis NAM Spinus spinus PAL ori SAM sam = South American distribution Spinus pinus NAM 15 Spinus barbatus SAM HAW = Hawaiian distribution Spinus cucullatus SAM Spinus magellanicus SAM Spinus atratus SAM

15 12.5 10 7.5 5 2.5 0Ma

FIGURE 3 Reconstruction by maximum-likelihood inference of the ancestral state of the r-K-gradient syndrome, using the distribution of species in two groups (see Fig. S1b). r-selection (state A) is in gray while K-selection (state B) is in black. At each node of the phylogenetic tree, a pie chart indicates the confidence given to K-selection or r-selection as the ancestral state. Otherwise as for Figure 2 PONGE et al. | 9943

Passer luteus Passer montanus Petronia petronia Montifringilla ruficollis Motacillaalba Anthus trivialis Plectrophenaxnivalis Ammodramus humeralis Parula pitiayumi Leistes superciliaris Fringillinae Fringillacoelebs PAL 1 Fringillamontifringilla PAL Chlorophoniacyanea SAM Euphoniinae Euphoniamusica NAM SAM Euphonia chlorotica SAM Fringillidae Euphoniacayennensis SAM 2 Euphoniaxanthogaster SAM Euphonia minuta NAM SAM Euphonialaniirostris NAM SAM Euphoniaviolacea SAM Mycerobascarnipes PAL Eophona migratoria PAL nam Hesperiphona vespertina NAM 3 Coccothraustes coccothraustes PAL Paroreomyzamontana HAW Loxioidesbailleui HAW 4 Chlorodrepanis virens HAW Erythrina erythrina PAL ORI Haematospizasipahi PAL ori Carduelinae Carpodacus synoicus PAL ara Carpodacus sibiricus PAL Carpodacus puniceus PAL Carpodacus roseus PAL 5 Carpodacus thura PAL Carpodacus rubicilloides PAL Carpodacus rubicilla PAL Carpodacus pulcherrimus PAL Carpodacus rodochroa PAL Carpodacus rhodochlamys PAL Pinicolaenucleator PAL NAM Procarduelis nipalensis PAL ORI Pyrrhula pyrrhula PAL Pyrrhulaerythaca PAL ORI Rhodopechyssanguineus PAL Bucanetes githagineus PAL ARA AFR Eremopsaltria mongolica PAL 6 Callacanthisburtoni PAL Agraphospiza rubescens PAL Leucostictenemoricola PAL Leucostictebrandti PAL Leucostictetephrocotis NAM Leucostictearctoa PAL Haemorhous mexicanus NAM Haemorhous purpureus NAM 7 Rhodospiza obsoleta PAL ARA Chlorischloris PAL Chlorissinica PAL ori 8 0.041 Chlorismonguilloti ORI Chlorisspinoides PAL ori Chlorisambigua PAL ORI State A State B Spinus olivaceus PAL Crithagra striolata AFR Crithagra burtoni AFR Crithagra mennelli AFR 9 Leistes superciliaris = Outgroup species Crithagra sulphurata AFR Crithagra rufobrunnea AFR Acanthis hornemanni = Sexual-monomorphic species Crithagra citrinelloides AFR Crithagra mozambica AFR Loxiapytyopsittacus = Sexual-dimorphic species Crithagra leucopygia AFR Linaria cannabina PAL 10 Linaria flavirostris PAL = In the tropics Acanthis flammea PAL NAM Acanthis hornemanni PAL NAM = Out of the tropics Loxialeucoptera PAL NAM 11 Loxia pytyopsittacus PAL PAL pal = Palearctic distribution (lower case = minor) Loxiacurvirostra PAL ORI NAM sam Cardueliscarduelis PAL Carduelis citrinella PAL 13 ORI ori = Oriental distribution Serinuscanicollis AFR Serinus syriacus PAL ARA ara = Arabian distribution Serinuspusillus PAL 14 Serinusserinus PAL AFR afr = African distribution Serinuscanaria PAL Spinus psaltria NAM SAM NAM nam = North-Central American distribution Spinus tristis NAM Spinus spinus PAL ori SAM sam = South American distribution Spinus pinus NAM 15 Spinus barbatus SAM HAW = Hawaiian distribution Spinus cucullatus SAM Spinus magellanicus SAM Spinus atratus SAM

15 12.5 10 7.5 5 2.5 0Ma

FIGURE 4 Reconstruction by maximum-likelihood inference of the ancestral state of the sexual dimorphism syndrome, using the distribution of species in two groups (see Fig. S1c). Sexual monomorphism (state A) is in gray while sexual dimorphism (state B) is in black. At each node of the phylogenetic tree, a pie chart indicates the confidence given to sexual monomorphism or dimorphism as the ancestral state. Otherwise as for Figure 2 9944 | PONGE et al.

Passer luteus Passer montanus Petronia petronia Montifringilla ruficollis Motacillaalba Anthus trivialis Plectrophenaxnivalis Ammodramus humeralis Parula pitiayumi Leistes superciliaris Fringillinae Fringillacoelebs PAL 1 Fringillamontifringilla PAL Chlorophoniacyanea SAM Euphoniinae Euphoniamusica NAM SAM Euphonia chlorotica SAM Fringillidae Euphoniacayennensis SAM 2 Euphoniaxanthogaster SAM Euphonia minuta NAM SAM Euphonialaniirostris NAM SAM Euphoniaviolacea SAM Mycerobascarnipes PAL Eophona migratoria PAL nam Hesperiphona vespertina NAM 3 Coccothraustes coccothraustes PAL Paroreomyzamontana HAW Loxioidesbailleui HAW 4 Chlorodrepanis virens HAW Erythrina erythrina PAL ORI Haematospizasipahi PAL ori Carduelinae Carpodacus synoicus PAL ara Carpodacus sibiricus PAL Carpodacus puniceus PAL Carpodacus roseus PAL 5 Carpodacus thura PAL Carpodacus rubicilloides PAL Carpodacus rubicilla PAL Carpodacus pulcherrimus PAL Carpodacus rodochroa PAL Carpodacus rhodochlamys PAL Pinicolaenucleator PAL NAM Procarduelis nipalensis PAL ORI Pyrrhula pyrrhula PAL Pyrrhulaerythaca PAL ORI Rhodopechyssanguineus PAL Bucanetes githagineus PAL ARA AFR Eremopsaltria mongolica PAL 6 Callacanthisburtoni PAL Agraphospiza rubescens PAL Leucostictenemoricola PAL Leucostictebrandti PAL Leucostictetephrocotis NAM Leucostictearctoa PAL Haemorhous mexicanus NAM Haemorhous purpureus NAM 7 Rhodospiza obsoleta PAL ARA Chlorischloris PAL 0.077 Chlorissinica PAL ori 8 Chlorismonguilloti ORI Chlorisspinoides PAL ori Chlorisambigua PAL ORI 0.033 State A State B Spinus olivaceus PAL Crithagra striolata AFR Crithagra burtoni AFR Crithagra mennelli AFR 9 Leistes superciliaris = Outgroup species Crithagra sulphurata AFR Crithagra rufobrunnea AFR Acanthis hornemanni = Migrant/gregarious species Crithagra citrinelloides AFR Crithagra mozambica AFR Loxiapytyopsittacus = Resident/territorial species Crithagra leucopygia AFR Linaria cannabina PAL 10 Linaria flavirostris PAL = In the tropics Acanthis flammea PAL NAM Acanthis hornemanni PAL NAM = Out of the tropics Loxialeucoptera PAL NAM 11 Loxia pytyopsittacus PAL PAL pal = Palearctic distribution (lower case = minor) Loxiacurvirostra PAL ORI NAM sam Cardueliscarduelis PAL Carduelis citrinella PAL 13 ORI ori = Oriental distribution Serinus canicollis AFR Serinus syriacus PAL ARA ara = Arabian distribution Serinus pusillus PAL 14 Serinusserinus PAL AFR afr = African distribution Serinus canaria PAL Spinus psaltria NAM SAM NAM nam = North-Central American distribution Spinus tristis NAM Spinus spinus PAL ori SAM sam = South American distribution Spinus pinus NAM 15 Spinus barbatus SAM HAW = Hawaiian distribution Spinus cucullatus SAM Spinus magellanicus SAM Spinus atratus SAM

15 12.5 10 7.5 5 2.5 0Ma

FIGURE 5 Reconstruction by maximum-likelihood inference of the ancestral state of the dispersal/social behavioral syndrome, using the distribution of species in two groups (see Fig. S1d). Dispersiveness/gregariousness (state A) is in gray while residence/territoriality (state B) is in black. At each node of the phylogenetic tree, a pie chart indicates the confidence given to residence/territoriality or dispersiveness/ gregariousness as the ancestral state. Otherwise as for Figure 2 PONGE et al. | 9945 (Continues) 1.40558E-05 0.092664518 0.395159442 0.006022008 0.047947376 0.456700628 LRT p - value K gradient, sexual dimorphism, b 7.970508 9.589146 4.080964 3.641438 27.743912 14.437506 a c d e f Dependent transition rate 0.060825 0.052294 0.044915 0.068022 0.000164 0.169471 0 0.049974 0.47751 1.616921 0 1.472791 0.015748 0 0 0.072745 0.046407 0.083369 0.142005 0 LRT statistic specialization, r- −83.446792 −72.309863 −84.850098 −77.793806 −71.241161 −62.977532 Likelihood dependent value was fixed to .05. Bold types indicate significant relationships between p - −97.318748 −79.528616 −88.835352 −82.588379 −73.281643 −64.798251 K gradient A → B K gradient B → A K gradient A → B K gradient B → A Likelihood independent Dependent syndrome transition Sexual dimorphism A → B Dispersal/social behavior A → B r- Ecological specialization A → B Sexual dimorphism B → A Dispersal/social behavior B → A r- Ecological specialization B → A Dispersal/social behavior A → B Ecological specialization A → B Dispersal/social behavior B → A Ecological specialization B → A Sexual dimorphism A → B Sexual dimorphism B → A r- r- Dispersal/social behavior A → D Dispersal/social behavior D → A Dispersal/social behavior A → D Dispersal/social behavior D → A K gradient Ecological specialization Dispersal/social behavior Ecological specialization Ecological specialization r- Dispersal/social behavior Syndrome 2 = A = A = B = B = B = B = A = A = A = A = D = D = A = A = B = B = A = A = B = B K gradient K gradient K gradient K gradient Ecological specialization r- Dispersal/social behavior Sexual dimorphism Ecological specialization r- Dispersal/social behavior Sexual dimorphism r- Sexual dimorphism r- Sexual dimorphism Ecological specialization Ecological specialization Dispersal/social behavior Dispersal/social behavior Ecological specialization Ecological specialization Ecological specialization Ecological specialization K gradient and dispersal/social behavior K gradient Conditional syndrome state r- Sexual selection and ecological specialization Ecological specialization and dispersal/social behavior r- Dispersal/social behavior r-K gradient Sexual dimorphism Sexual dimorphism Sexual dimorphism Syndrome 1 TABLE 2 Modeling correlations between transitions from a state (ancestral A or derived B) to the opposite for four syndromes (ecological dispersal/social behavior). Likelihood ratio test (LRT) was used to for significance. Significance level LRT transition states of syndromes. Details are then given on rates for pairs syndromes where transitions between character significantly correlated 9946 | PONGE et al.

follow common paths, in particular (1) sexual dimorphism and ecological specialization, (2) r-K-gradient and dispersal/social behavior, and (3) ecological specialization and dispersal/social behavior. Reversal from derived to ancestral state never or hardly occurs when one member of these three couples of syndromes is in derived state (see transition rates in Table 2). The correlated evolution of the four syndromes was then visu- ally addressed by reporting on the same graph in the form of colored trajectories (Figure 6) the maximum-likelihood patterns displayed by g Figures 2–5. To the exception of generalism, for which uncertainty re- mains at the base of the tree, the ancestrality of r-selection, sexual h h Dependent transition rate 0 0 3.81304 9.535607 monomorphism, and dispersiveness/gregariousness is more likely than the respective alternative state of each syndrome. Although the correlated evolution of more than two syndromes could not be modeled by our method, Figure 6 shows that three syndromes can follow the same path, although all the four syndromes never follow a common path. Hence, our former, simplistic hypothesis of the existence of two opposite strategies, evolving from an ancestral strategy A (generalists, r-selected, sexually monomorphic, migratory/gregarious) to a derived strategy B (specialists, K-selected, sexually dimorphic, resident/territorial) should be refined by considering that these four suites of traits diverged in the course of evolution, generating more than one derived strategy.

3.4.2 | Multivariate analysis of phylogenetic signaling

A phylogenetic PCA (pPCA) performed on the four syndromes (PC1spec, PC1rK, PC1sex, PC1beh) was used for a finer delineation of groups of species that are associated to syndromes having evolved correlatively. The four syndromes exhibit each a significant phyloge-

to A. netic signal (Abouheif’s test, p < .01). The first three principal com- Dependent syndrome transition Ecological specialization A → D Ecological specialization D → A Ecological specialization A → D Ecological specialization D → A ponents of pPCA exhibit positive eigenvalues (Table 3, Figure 8c), indicating the existence of a phylogenetic signal on each of them. The first component of pPCA displays a phylogenetic signal common to all from D to A. four syndromes (Figure 8a,b), while second (Figure 8a) and third components (Figure 8b) partial out phylogenetic signals of some groups of syndromes. The first principal component pPC1 opposes A-strategists (generalist, r-selected, sexually monomorphic, gregarious/migratory species), to B-strategists (species with opposite trait modalities), as expected from our Hypothesis H2 (Figure 8a,b). However, the two following K gradient never shifts from A to D. K gradient never shifts from D to A. principal components pPC2 and pPC3 show that strategy B could be divided into two strategies, which we call B1 and B2 (Figure 8a,b). Strategy B1 is characterized by a combination of sexual dimorphism = A, then r- = A, then ecological specialization often shifts from D = D, then sexual dimorphism hardly shifts = A, then sexual dimorphism never shifts from D to A. = D, then dispersal/social behavior never shifts from D to A. and r-selection (Figure 8b) while strategy B2 is characterized by the = D = D = A = A common occurrence of residence/territoriality (Figure 8a) and K- selection (Figure 8b). The originality of strategy B1 is that it combines trait modalities of strategy A (r-selection) and strategy B (sexual dimorphism), while strategy B2 is characterized by trait modalities of strategy B, to the exception of sexual dimorphism.

K gradient = D, then dispersal/social behavior never shifts from D to A. We used the 3D space of the first three principal components of Dispersal/social behavior Dispersal/social behavior Dispersal/social behavior Dispersal/social behavior pPCA to split the whole set of species into those three strategies (A, Conditional syndrome state When ecological specialization When dispersal/social behavior When dispersal/social behavior = D, then ecological specialization neither shifts from A to D nor A. When dispersal/social behavior = D, then r- When dispersal/social behavior When ecological specialization When r- When ecological specialization TABLE 2 (Conttinued) b c d e f g h a B1, B2) with the k-means method of variance partition (with k = 3). PONGE et al. | 9947

Fringillinae Fringillacoelebs PAL 1 Fringillamontifringilla PAL Chlorophoniacyanea SAM Euphoniinae Euphonia musica NAM SAM Euphonia chlorotica SAM Fringillidae Euphoniacayennensis SAM 2 Euphoniaxanthogaster SAM Euphonia minuta NAM SAM Euphonialaniirostris NAM SAM Euphoniaviolacea SAM Mycerobascarnipes PAL Eophona migratoria PAL nam Hesperiphona vespertina NAM 3 Coccothraustes coccothraustes PAL Paroreomyzamontana HAW Loxioidesbailleui HAW 4 Chlorodrepanis virens HAW Erythrina erythrina PAL ORI Haematospiza sipahi PAL ori Carduelinae Carpodacus synoicus PAL ara Carpodacussibiricus PAL Carpodacus puniceus PAL Carpodacus roseus PAL 5 Carpodacus thura PAL Carpodacus rubicilloides PAL Carpodacus rubicilla PAL Carpodacus pulcherrimus PAL Carpodacus rodochroa PAL Carpodacusrhodochlamys PAL Pinicola enucleator PAL NAM Procarduelis nipalensis PAL ORI Pyrrhula pyrrhula PAL Pyrrhulaerythaca PAL ORI Rhodopechyssanguineus PAL Bucanetes githagineus PAL ARA AFR Eremopsaltria mongolica PAL 6 Callacanthisburtoni PAL Agraphospiza rubescens PAL Leucostictenemoricola PAL Leucostictebrandti PAL Leucostictetephrocotis NAM Leucostictearctoa PAL Haemorhous mexicanus NAM Heamorhous purpureus NAM 7 Rhodospiza obsoleta PAL ARA Chlorischloris PAL Chlorissinica PAL ori 8 Chlorismonguilloti ORI Chlorisspinoides PAL ori Chlorisambigua PAL ORI Spinus olivaceus PAL = Specialist Crithagra striolata AFR Crithagra burtoni AFR Crithagra mennelli AFR 9 = K-selected Crithagra sulphurata AFR Crithagra rufobrunnea AFR Crithagra citrinelloides AFR = Sexually dimorphic Crithagra mozambica AFR Crithagra leucopygia AFR Linaria cannabina PAL 10 = Resident/territorial Linaria flavirostris PAL Acanthis flammea PAL NAM Acanthis hornemanni PAL NAM Spinus tristis = State A species Loxialeucoptera PAL NAM 11 Crithagra leucopygia = State B1 species Loxia pytyopsittacus PAL Loxiacurvirostra PAL ORI NAM sam Loxia leucoptera = State B2 species Carduelis carduelis PAL = in the tropics Carduelis citrinella PAL 13 = out of the tropics Serinus canicollis AFR Serinus syriacus PAL PAL pal = Palearctic distribution (lower case = minor) Serinus pusillus PAL 14 ORI ori = Oriental distribution Serinus serinus PAL ARA ara = Arabian distribution Serinus canaria PAL Spinus psaltria NAM SAM AFR afr = African distribution Spinus tristis NAM NAM nam = North-Central American distribution Spinus spinus PAL ori Spinus pinus NAM 15 SAM sam = South American distribution Spinus barbatus SAM HAW = Hawaiian distribution Spinus cucullatus SAM Spinus magellanicus SAM Spinus atratus SAM

15 12.5 10 7.5 5 2.5 0Ma

FIGURE 6 Common representation of paths followed by ecological specialization, r-K-gradient, sexual dimorphism, and dispersal/social behavior along the phylogenetic tree of 81 fringillid species. Color lines are for derived states, gray lines for ancestral states, they were dashed in case of incertitude. Species names are in gray (A strategy), black italics (B1 strategy), or black roman (B2 strategy). Otherwise as for Figure 2 9948 | PONGE et al.

TABLE 3 Phylogenetic principal PC1 PC2 PC3 components analysis. Scores of the four Ecological specialization −0.4976956 −0.03918335 0.26095389 syndromes along the first three principal r-K gradient −0.22146655 −0.44248631 0.77233494 components. Eigen values are given in the last row Sexual dimorphism −0.80095985 −0.12831694 −0.47739305 Dispersal/social behavior −0.248425 0.88668218 0.3278684 Eigen value 0.1587 0.0789 0.0642

The three strategies are indicated on Figure 6. The reconstruction of et al., 2014). We may add that once bursts of speciation appear in ancestral traits gives a full support to the ancestrality of strategy A and novel environments, ancestral traits associated with migration (in par- to derived positions for strategies B1 and B2 (Figure 7). The highest ticular orientation and memorization of geographic features, together transition rate is from A toward B1 (0.067), followed by A toward B2 with collective behavior) may disappear in favor of sophisticated sig- (0.027), while reversal rates are nil. Transition from B2 toward B1 is naling traits associated with territoriality if the colonized environment possible but less frequent than transition from A to either strategy B turns out to be more stable than the original environment, a case of (transition rate 0.014), while transition from B1 toward B2 never oc- relaxed selection pressure (Wiersma, Nowak, & Williams, 2012). curs. Strategy A is expressed in clades 1, 8, 10, 13–15, partly in clades The ancestral reconstruction of the “sexual dimorphism” syndrome 3 and 11. Strategy B1is expressed in clades 5 and 7, partly in clades 6 shows that a group of sexually selected traits related to male song elab- and 11, while the B2 strategy is expressed in clades 2, 4, and 9. oration (Table 1) is in ancestral position while a group of sexually selected traits related to plumage color and body size is in derived position (Figure 4). This is not in agreement with our original hypothesis of nat- 3.5 | Association of syndromes with tropical affinity ural selection opposed to sexual selection, the former process being a (H3) response to fluctuations of the environment while the latter would drive We tested whether there was a latitudinal signal (tropical affinity) directional evolution in stable environments (Møller & Garamszegi, in the distribution of syndrome states by performing phylogenetic 2012). However, it has been shown that sexual selection is also an adap- ANOVAs. K-selection and residence/territoriality exhibit the same tive response to environmental constraints and fluctuations (Badyaev significant affinity for tropical areas, while sexual dimorphism exhibits & Ghalambor, 1998), and that synergistic combinations of natural and a higher affinity for nontropical areas, and specialism does not display sexual selection are widespread (Botero & Rubenstein, 2012). The op- any significant latitudinal signal (Table 4). position between visual (size, color) and acoustic signals (song display) has been already observed by Badyaev et al. (2002) in Carduelinae. Such a reversal from one type of sexually selected signal to another is 4 | DISCUSSION known as the “transfer hypothesis” (Shutler & Weatherhead, 1990). The observed contrast between song elaboration and color display is paral- 4.1 | Ancestral syndromes in Fringillidae leled by a contrast between melanin and carotene pigments in plumage Results support our H1 hypothesis: r-selection and dispersiveness/gre- coloration (Table 1). Both melanin- and carotene-based plumage colors gariousness are ancestral and tend to shift toward derived attributes have been shown to predict success in dominance interactions between in the course of evolution. Even though most species/traits are spread males and influence favorably female choices (Hill, 1990; Tarof, Dunn, along a continuous r-K gradient (Jones, 1976), it has been theoretically & Whittingham, 2005) and are honest signals of male good condition (Blanck, Tedesco, & Lamouroux, 2007) and empirically demonstrated and resistance to parasites (Roulin, 2015; Safran, McGraw, Wilkins, (Flegr, 1997) that species whose reproductive strategy lies in the mid- Hubbard, & Marling, 2010). However, carotene must be found in the dle of the r-K continuum are at disadvantage, justifying the distinction environment while melanin is produced by birds themselves (Griffith, of two opposite strategies. Our qualitative approach shows two groups Parker, & Olson, 2006; Roulin, 2015). We hypothesize that in environ- of r-selected and K-selected species. Comparisons between tropical ments where food resources are scarce, the ability of males to find high- and nontropical bird species showed an association of K-strategy (ap- quality resources and allocate this quality to offspring is advantageous proximated by basal metabolic rate) with climatically stable environ- to survival and is favored by mates (carotene-based sexual selection). ments (Wiersma, Muñoz-Garcia, Walker, & Williams, 2007), supporting Conversely, in environments where resources are abundant, at least in theoretical expectations (Southwood, May, Hassell, & Conway, 1974). breeding areas, melanin, which is not context dependent but pleiotrop- The reconstruction of ancestral traits for the behavioral syndrome ically linked to body condition and strongly heritable (Roulin, 2015), is shows that behavioral plasticity and gregariousness have a basal po- favored by mates (melanin-based sexual selection). sition while rather standardized and solitary behaviors are in derived The ancestrality of dispersiveness (see above) suggests that an- position (Figure 5), as shown by Simpson, Johnson, and Murphy (2015) cestral true finches (Fringillidae) were monomorphic (melanin-based in Parulidae. Here too, this syndrome is clearly related to the stability conspicuous plumage in both sexes) had elaborated song and were of the environment, gregariousness and dispersal ability being advan- well-equipped for flying to remote breeding areas where resources tageous for motile organisms facing environmental hazards (Stevens are seasonally abundant, while dimorphic (carotene-based colored) PONGE et al. | 9949

Passer luteus Passer montanus Petronia petronia Montifringillaruficollis Motacillaalba Anthus trivialis Plectrophenaxnivalis Ammodramus humeralis Parula pitiayumi Leistes superciliaris Fringillinae Fringillacoelebs PAL 1 Fringillamontifringilla PAL Chlorophoniacyanea SAM Euphoniinae Euphoniamusica NAM SAM Euphonia chlorotica SAM Fringillidae Euphoniacayennensis SAM 2 Euphoniaxanthogaster SAM Euphonia minuta NAM SAM Euphonialaniirostris NAM SAM Euphoniaviolacea SAM Mycerobascarnipes PAL Eophona migratoria PAL nam 3 Hesperiphona vespertina NAM Coccothraustes coccothraustes PAL Paroreomyzamontana HAW Loxioidesbailleui HAW 4 Chlorodrepanis virens HAW Erythrina erythrina PAL ORI Haematospizasipahi PAL ori Carduelinae Carpodacus synoicus PAL ara Carpodacus sibiricus PAL Carpodacus puniceus PAL Carpodacus roseus PAL 5 Carpodacus thura PAL Carpodacus rubicilloides PAL Carpodacus rubicilla PAL Carpodacus pulcherrimus PAL Carpodacus rodochroa PAL Carpodacus rhodochlamys PAL Pinicolaenucleator PAL NAM Procarduelis nipalensis PAL ORI Pyrrhula pyrrhula PAL Pyrrhulaerythaca PAL ORI Rhodopechyssanguineus PAL Bucanetes githagineus PAL ARA AFR Eremopsaltria mongolica PAL 6 Callacanthisburtoni PAL Agraphospiza rubescens PAL Strategy A Leucostictenemoricola PAL Leucostictebrandti PAL Leucostictetephrocotis NAM Leucostictearctoa PAL Haemorhous mexicanus NAM 7 Haemorhous purpureus NAM Rhodospiza obsoleta PAL ARA 0.014 Chlorischloris PAL Chlorissinica PAL ori 8 Strategy B1 Strategy B2 Chlorismonguilloti ORI Chlorisspinoides PAL ori Chlorisambigua PAL ORI Leistes superciliaris = Outgroup species Spinus olivaceus PAL Crithagra striolata AFR Spinus tristis = State A species Crithagra burtoni AFR Crithagra mennelli AFR 9 Crithagra sulphurata AFR Loxialeucoptera = State B1 species Crithagra rufobrunnea AFR Crithagra citrinelloides AFR Crithagra mozambica AFR Crithagra leucopygia = State B2 species Crithagra leucopygia AFR Linaria cannabina PAL 10 = In the tropics Linaria flavirostris PAL Acanthis flammea PAL NAM Acanthis hornemanni PAL NAM = Out of the tropics Loxialeucoptera PAL NAM 11 Loxia pytyopsittacus PAL PAL pal = Palearctic distribution (lower case = minor) Loxiacurvirostra PAL ORI NAM sam Carduelis carduelis PAL ORI ori = Oriental distribution Carduelis citrinella PAL 13 Serinuscanicollis AFR ARA ara = Arabian distribution Serinus syriacus PAL Serinuspusillus PAL 14 AFR afr = African distribution Serinusserinus PAL Serinuscanaria PAL NAM nam = North-Central American distribution Spinus psaltria NAM SAM Spinus tristis NAM SAM sam = South American distribution Spinus spinus PAL ori Spinus pinus NAM 15 Spinus barbatus SAM HAW = Hawaiian distribution Spinus cucullatus SAM Spinus magellanicus SAM Spinus atratus SAM

15 12.5 10 7.5 5 2.5 0Ma

FIGURE 7 Reconstruction by maximum-likelihood inference of the ancestral state of the three strategies A (green), B1 (blue), and B2 (red), using the distribution of species in three groups according to k-means clustering based on the first three principal components of phylogenetic PCA. At each node of the phylogenetic tree, a pie chart indicates the confidence given to residence/territoriality or dispersiveness/ gregariousness as the ancestral state. Otherwise as for Figure 2 9950 | PONGE et al.

TABLE 4 Relationship between syndrome groups and tropical (r-K-gradient, ecological specialization, and dispersal/social behavior, affinity, tested by phylogenetic ANOVA (ANOVA corrected for Table 3, Figure 6). Sexual dimorphism appears to have evolved in as- phylogenetic autocorrelation) sociation with ancestral trait modalities of the three other syndromes Tropical affinity F-value Probability (Figures 6 and 8) and is not associated with K-selection and residence/ territoriality (with the exception of the genus Loxia within clade 11, Generalist vs. specialist 0.41 vs. 0.53 0.17 0.81 NS associated with K-selection, see Figure 3). This feature explains why a r-Selected vs. 0.35 vs. 0.77 13.25 0.025* K-selected common evolution of the four syndromes was not detected. However, r-selected traits, sexual monomorphism, and dispersiveness/gregari- Sexual monomorphism 0.56 vs. 0.23 12.52 0.026* vs. sexual dimorphism ousness are all present in ancient lineages in the phylogenetic tree, Dispersive/gregarious 0.33 vs. 0.68 12.02 0.032* supporting at least partially our hypothesis H2 that syndromes evolved vs. resident/territorial correlatively along the phylogenetic tree of Fringillidae.

NS = not significant at 0.05 level; *p < .05. Our results point to the existence of a common ancestor to Fringillidae with all features of the A strategy (generalism, r-selection, monochromatism with elaborated song, dispersiveness/gregarious- resident birds are better equipped for foraging in areas where re- ness), with a further evolution of a group of clades where sexual di- sources are scarcely distributed. The ancestrality of monochroma- morphism is prominent while other features of the A strategy are kept tism (males and females both harboring a bright coloration) has been (strategy B1), and another group where most features of the B strategy demonstrated in other bird groups (Friedman, Hofmann, Kondo, & evolved to the exception of sexual dimorphism (strategy B2; Figure 7). Omland, 2009; Simpson et al., 2015). Our hypothesis of generalism as ancestral in the Fringillidae cannot 4.3 | Geographical distribution and past be rejected, because of two arguments. First, generalism is an ancestral history of the Fringillidae state in Carduelinae, the core group and, second, the two species of the most basal subfamily, the Fringillinae, which are included in our study We may wonder whether the observed discrepancy between a lineage (F. coelebs and F. montifringilla), are clearly generalists. It should also be with sexually selected dimorphism (B1) and another with K-selected, noted that reversal from specialism to generalism is observed in clade 5 specialist, and resident/territorial traits (B2) could be explained by only (Figure 2). Therefore, despite uncertainty of ancestral reconstruc- geographical affinities of the species, and the way different envi- tion, specialism is a derived state in all clades but clade 5. An important ronments were colonized in the course of the evolution of this now point is that generalism is still the commonest state in the crown group worldwide bird family. If we examine the relationship between tropi- (clades 7 to 15, Figure 2), that is, in clades most remote from the common cal affinity and the balance between ancestral and derived states of ancestor and thus resulting from a high number of speciation events. our four syndromes (Table 4), it appears that most tropical species are The ancestrality of generalism has been demonstrated in a variety K-selected and resident territorial while most nontropical species are of monophyletic groups, among plants (Schneeweiss, 2007; Tripp & dimorphic sexually, and ecological specialization is undifferentiated. Manos, 2008) and animals (Hwang & Weirauch, 2012; Kelley & Farrell, This result supports only partly our third hypothesis (H3) that derived 1998; Loiseau et al., 2012; Prinzing, D’Haese, Pavoine, & Ponge, 2014; attributes are associated with tropical affinity, suggesting that the B2 Yotoko & Elisei, 2006), birds included (Brumfield et al., 2007; Jønsson, type lives (and probably evolved) mainly in tropical areas, while the Fabre, Ricklefs, & Fjeldså, 2011), while fewer studies conclude to the B1 branch is located (and probably evolved) mainly out of the tropics. ancestrality of specialism (Sedivy, Praz, Müller, Widmer, & Dorn, 2008; The existence of a sexually dimorphic branch living out of the tropics Stireman, 2005). Despite repeated assessment of phylogenetic con- is supported by studies on other bird families (Bailey, 1978; Friedman servatism of niche requirements (Brumfield et al., 2007; Peterson, et al., 2009), while Price, Lanyon, and Omland (2009) pointed on sing- Soberón, & Sánchez-Cordero, 1999; Prinzing, Durka, Klotz, & Brandl, ing by both sexes as the tropical ancestral state in New World black- 2001; Prinzing et al., 2014; Wiens & Graham, 2005), confirmed in the birds. Species we classified in the sexually dimorphic group (Fig. S1c) are present study, it has even been shown that habitat specialization is mostly adapted to life in harsh environments. An extraordinary variety of a labile ecological trait that may change in the short-term within the true rosefinches (clade 5) are living in the Himalayas, which are thought same species (Barnagaud, Devictor, Jiguet, & Archaux, 2011). This to be a source of species radiation (Tietze et al., 2013). Within clade might explain why ancestral attributes of ecological specialization are 6, the trumpeter finch Bucanetes githagineus and the Mongolian finch less clear-cut than for r-K, sexual selection, and behavioral syndromes. Eremopsaltria mongolica live in deserts or semi-deserts, and the Blanford’s rosefinch Agraphospiza rubescens lives in the Himalayas. Other species of this group have a circumpolar distribution and live in coniferous forests 4.2 | The correlated evolution of syndromes in (the white-winged crossbill Loxia leucoptera, the red crossbill Loxia cur- Fringillidae virostra). However, mountain finches (Leucosticte) live in Asian tundras Even though we do not record a correlated evolution of all four syn- and high mountains and display no or only weak sexual dimorphism, dromes, a common phylogenetic signal is detected (Table 3) and three thus tolerance of harsh habitats does not necessarily entail sexual dimor- syndromes exhibit a correlated evolution along the phylogenetic tree phism. Contrary to erroneous ideas resulting from the ordinary confusion PONGE et al. | 9951

(a) Residence (b) territoriality Strategy A Strategy A pPC2 pPC3 Strategy B1 Strategy B1 Strategy B2 Strategy B2

K-selecon

Specialism pPC1 Residence Sexual dimorphism territoriality Specialism

K-selecon

pPC1

0.2 (c) Sexual dimorphism 0.15

0.1

0.05

0 pPC1 pPC2 pPC3 pPC4

FIGURE 8 Projection in the plane of the first and second (a), and first and third, (b) principal components of phylogenetic PCA of the synthetic variables (see text for details) describing the four studied syndromes. Eigenvalues are diagrammatically represented (c) between stress and disturbance (Borics, Várbiró, & Padisák, 2013), such with most benign habitats, mostly in lowland tropics. Both habitats of B- harsh habitats are highly predictable (Greenslade, 1983), allowing sea- organisms are predictable, although in B1 the high seasonality forces spe- sonal short-distance (altitudinal) or long-distance (latitudinal) migration cies to migrate seasonally (either altitudinal or latitudinal migration) while to cope with foraging and breeding bird requirements. According to in B2 the ancestral migratory behavior has been lost, being unnecessary. Ponge (2013) cyclic, seasonal processes to which organisms are adapted allows anticipation, a key adaptive trait of strategy B. ACKNOWLEDGMENTS The existence of K-selected, resident/territorial (also specialist?) B2-strategists living in the tropics is attested by many studies on birds We are grateful to Peter Clement, Claire Doutrelant, and Doris Gomez (Jetz, Freckleton, & MvKechnie, 2008; Sol et al., 2010; Wiersma et al., for constructive comments on a first draft of the manuscript. 2007, 2012). We may thus consider that strategy B2 is associated with stable benign environments, better exemplified near the Equator in CONFLICT OF INTEREST tropical rainforests. None declared.

5 | CONCLUSIONS AUTHORS CONTRIBUTION

Our results provide strong support to strategy A (r-selection, sexual Jean-François Ponge performed most part of the redaction, part of the monomorphism, migratory/gregarious) as ancestral in the fringillid fam- calculations (construction of the synthetic variables as proxies for trait ily. However, our former idea of two opposite strategies previously called syndromes), made graphical representations of the results, and collected “barbarians” (ancestral) and “civilized” (derived) by Ponge (2013), and the original data in published literature. Dario Zuccon performed the now called A and B, respectively, must be refined. Strategy B, associated phylogeny and contributed to the redaction. Marianne Elias performed with predictability of the environment, should be subdivided in B1 (sexu- the reconstruction of ancestral trait syndromes and some other calcula- ally dimorphic, r-selected, migratory/gregarious), associated with harsh tions, and contributed to the redaction. Sandrine Pavoine performed the habitats (mountain and boreal habitats), mostly out of the tropics, and phylogenetic PCA and some other calculations and contributed to the B2 (sexually monomorphic, K-selected, resident/territorial), associated redaction. Pierre-Yves Henry contributed to the redaction and to fruitful 9952 | PONGE et al. discussions about methods and concepts. Marc Théry and Éric Guilbert Flegr, J. (1997). Two distinct types on natural selection in turbidostat-like and contributed to the redaction and to fruitful discussions. chemostat-like ecosystems. Journal of Theoretical Biology, 188(1), 121–126. Friedman, N. R., Hofmann, C. M., Kondo, B., & Omland, K. E. (2009). Correlated evolution of migration and sexual dichromatism in the New ORCID Word orioles (Icterus). Evolution, 63(12), 3269–3274. Greenslade, P. J. M. (1983). Adversity selection and the habitat templet. Jean-François Ponge http://orcid.org/0000-0001-6504-5267 American Naturalist, 122(3), 352–365. Griffith, S. C., Parker, T. H., & Olson, V. A. (2006). Melanin- versus carotenoid-based sexual signals: Is the difference really so black and REFERENCES red? Animal Behaviour, 71, 749–763. Harmon, L. J., Weir, J. T., Brock, C. D., Glor, R. E., & Challenger, W. (2008). Abouheif, E. (1999). A method for testing the assumption of phylogenetic GEIGER: Investigating evolutionary radiations. Bioinformatics, 24(1), independence in comparative data. Evolutionary Ecology Research, 1(8), 129–131. 895–909. Hill, G. E. (1990). Female house finches prefer colourful males: Sexual selec- Badyaev, A. V. (1997a). Covariation between life history and sexually tion for a condition-dependent trait. Animal Behaviour, 40(3), 563–572. selected traits: An example with cardueline finches. Oikos, 80(1), Holmes, C. C., & Adams, N. M. (2002). A probabilistic nearest neighbour 128–138. method for statistical pattern recognition. Journal of the Royal Statistical Badyaev, A. V. (1997b). Avian life history variation along altitudinal gradi- Society, Series B, Statistical Methodology, 64(2), 295–306. ents: An example with cardueline finches. Oecologia, 111(3), 365–374. Hwang, W. S., & Weirauch, C. (2012). Evolutionary history of assassin bugs Badyaev, A. V. (1997c). Altitudinal variation in sexual dimorphism: A new (Insecta: Hemiptera: Reduviidae): Insights from divergence dating and pattern and alternative hypotheses. Behavioral Ecology, 8(6), 675–690. ancestral state reconstruction. PLoS ONE, 7(9), e45523. Badyaev, A. V., & Ghalambor, C. K. (1998). Does a trade-off exist between Janzen, D. H. (1967). Why mountain passes are higher in the tropics. sexual ornamentation and ecological plasticity? Sexual dichromatism American Naturalist, 101, 233–249. and occupied elevational range in finches. Oikos, 82(2), 319–324. Jetz, W., Freckleton, R. P., & MvKechnie, A. E. (2008). Environment, migra- Badyaev, A. V., Hill, G. E., & Weckworth, B. V. (2002). Species divergence tory tendency, phylogeny and basal metabolic rate in birds. PLoS ONE, in sexually selected traits: Increase in song elaboration is related to de- 3(9), e3261. crease in plumage ornamentation in finches. Evolution, 56(2), 412–419. Jombart, T., Pavoine, S., Devillard, S., & Pontier, D. (2010). Putting phylog- Bailey, S. F. (1978). Latitudinal gradients in colors and patterns of passerine eny into the analysis of biological traits: A methodological approach. birds. Condor, 80(4), 372–381. Journal of Theoretical Biology, 264(3), 693–701. Barnagaud, J. Y., Devictor, V., Jiguet, F., & Archaux, F. (2011). When species Jones, J. M. (1976). The r-K-selection continuum. American Naturalist, 110, become generalists: On-going large-scale changes in bird habitat spe- 320–323. cialization. Global Ecology and Biogeography, 20(4), 630–640. Jønsson, K. A., Fabre, P. H., Ricklefs, R. E., & Fjeldså, J. (2011). Major global Beckman, E. J., & Witt, C. C. (2015). Phylogeny and biogeography of the radiation of corvoid birds originated in the proto-Papuan archipelago. New World siskins and goldfinches: Rapid, recent diversification in the Proceedings of the National Academy of Sciences of the United States of Central Andes. Molecular Phylogenetics and Evolution, 87, 28–45. America, 108(6), 2328–2333. Berg, M. P., Kiers, T., Driessen, G., Van der Heijden, M., Kooi, B. W., Kuenen, Kelley, S. T., & Farrell, B. D. (1998). Is specialization a dead end? The phy- F., … Ellers, J. (2010). Adapt or disperse: Understanding species per- logeny of host use in Dendroctonus bark beetles (Scolytidae). Evolution, sistence in a changing world. Global Change Biology, 16(2), 587–598. 52(6), 1731–1743. Blanck, A., Tedesco, P. A., & Lamouroux, N. (2007). Relationships between Laiolo, P., Seoane, J., Illera, J. C., Bastianelli, G., Carrascal, L. M., & Obeso, life-history strategies of European freshwater fish species and their J. R. (2015). The evolutionary convergence of avian lifestyles and their habitat preferences. Freshwater Biology, 52(5), 843–859. constrained coevolution with species’ ecological niche. Proceedings of Bonte, D., Van Dyck, H., Bullock, J. M., Coulon, A., Delgado, M., Gibbs, the Royal Society of London, Series B, Biological Sciences, 282, 20151808. M., … Travis, J. M. (2012). Costs of dispersal. Biological Reviews, 87(2), Lerner, H. R. L., Meyer, M., James, H. F., Hofreiter, M., & Fleischer, R. C. 290–312. (2011). Multilocus resolution of phylogeny and timescale in the extant Borics, G., Várbiró, G., & Padisák, J. (2013). Disturbance and stress: Different adaptive radiation of Hawaiian honeycreepers. Current Biology, 21(21), meanings in ecological dynamics? Hydrobiologia, 711(1), 1–7. 1838–1844. Botero, C. A., & Rubenstein, D. R. (2012). Fluctuating environments, sexual Loiseau, C., Harrigan, R. J., Robert, A., Bowie, R. C. K., Thomassen, H. A., selection and the evolution of flexible mate choice in birds. PLoS ONE, Smith, T. B., & Sehgal, R. N. M. (2012). Host and habitat specialization 7(2), e32311. of avian malaria in Africa. Molecular Ecology, 21(2), 431–441. Brown, W. L. Jr, & Wilson, E. O. (1956). Character displacement. Systematic Miller, M. A., Pfeiffer, W., & Schwartz, T. (2010).Creating the CIPRES Zoology, 5(2), 49–64. Science Gateway for inference of large phylogenetic trees. In Gateway Brumfield, R. T., Tello, J. G., Cheviron, Z. A., Carling, M. D., Crochet, N., & Computing Environments Workshop (GCE 2010), Proceedings of a Rosenberg, K. V. (2007). Phylogenetic conservatism and antiquity of meeting held on 14 November 2010, New Orleans, LA. Red Hook, a tropical specialization: Army ant-following in the typical antbirds Curran Associates. Retrieved from http://ai2-s2-pdfs.s3.amazonaws. (Thamnophilidae). Molecular Phylogenetics and Evolution, 45(1), 1–13. com/e27a/caf97f5b2eae4257bb5d8278fbe0e6405c39.pdf, 7 pp. Cardillo, M. (2002). The life-history basis of latitudinal diversity gradients: Møller, A. P., & Garamszegi, L. Z. (2012). Sexual selection, range size and How do species traits vary from the poles to the equator. Journal of population size. Ornis Hungarica, 20(1), 1–25. Animal Ecology, 71(1), 79–87. Pagel, M., & Meade, A. (2006). Bayesian analysis of correlated evolution Clement, P., Harris, A., & Davis, J. (2011). Finches and sparrows (p. 512). of discrete characters by reversible-jump Markov chain Monte Carlo. London: Bloomsbury. American Naturalist, 167(6), 808–825. Dickinson, E. C., & Christidis, L. (2014). The Howard and Moore com- Pagel, M., Meade, A., & Barker, D. (2004). Bayesian estimation of ancestral plete checklist of the birds of the World. II. Passerines (752 pp, 4th ed.). character states on phylogenies. Systematic Biology, 53(5), 673–684. Eastbourne: Aves Press. Pavoine, S., Ollier, S., Pontier, D., & Chessel, D. (2008). Testing for phyloge- Drummond, A. J., & Rambaut, A. (2007). BEAST: Bayesian evolutionary netic signal in phenotypic traits: New matrices of phylogenetic proxim- analysis by sampling trees. BMC Evolutionary Biology, 7(1), 214. ities. Theoretical Population Biology, 73(1), 79–91. PONGE et al. | 9953

Peterson, A. T., Soberón, J., & Sánchez-Cordero, V. (1999). Conservatism Sol, D., Garcia, N., Iwaniuk, A., Davis, K., Meade, A., Boyle, W. A., & Székely, of ecological niches in evolutionary times. Science, 285, 1265–1267. T. (2010). Evolutionary divergence in brain size between migratory and Poisot, T., Bever, J. D., Nemri, A., Thrall, P. H., & Hochberg, M. E. (2011). resident birds. PLoS ONE, 5(3), e9617. A conceptual framework for the evolution of ecological specialization. Southwood, T. R. E., May, R. M., Hassell, M. P., & Conway, G. R. (1974). Ecology Letters, 14(9), 841–851. Ecological strategies and population parameters. American Naturalist, Ponge, J. F. (2013). Disturbances, organisms and ecosystems: A global 108, 791–804. change perspective. Ecology and Evolution, 3(4), 1113–1124. Stamatakis, A. (2006). RAxML-VI-HPC: Maximum likelihood-based Posada, D., & Crandall, K. A. (2001). Selecting the best-fit model of nucleo- phylogenetic analyses with thousands of taxa and mixed models. tide substitution. Systematic Biology, 50(4), 580–601. Bioinformatics, 22(21), 2688–2690. Price, J. J., Lanyon, S. M., & Omland, K. E. (2009). Losses of female song Steinhaus, H. (1956). Sur la division des corps matériels en parties. with changes from tropical to temperate breeding in the New World Bulletin de l’Académie Polonaise des Sciences, Classe Troisième, 4(12), blackbirds. Proceedings of the Royal Society of London, Series B, Biological 801–804. Sciences, 276, 1971–1980. Stevens, V. M., Whitmee, S., Le Galliard, J. F., Clobert, J., Böhning-Gaese, K., Prinzing, A., D’Haese, C., Pavoine, S., & Ponge, J. F. (2014). Species living Bonte, D., … Baguette, M. (2014). A comparative analysis of dispersal in harsh environments have low clade rank and are localized on former syndromes in terrestrial and semi-terrestrial animals. Ecology Letters, Laurasian continents: A case study of Willemia (Collembola). Journal of 17(8), 1039–1052. Biogeography, 41(2), 353–365. Stireman, J. O. III (2005). The evolution of generalization? Parasitoid flies Prinzing, A., Durka, W., Klotz, S., & Brandl, R. (2001). The niche of higher and the perils of inferring host range evolution from phylogenies. plants: Evidence for phylogenetic conservatism. Proceedings of the Journal of Evolutionary Biology, 18(2), 325–336. Royal Society of London, Series B, Biological Sciences, 268, 2383–2389. Tarof, S., Dunn, P. O., & Whittingham, L. A. (2005). Dual functions of a R Core Team (2016). R: A language and environment for statistical computing. melanin-based ornament in the common yellowthroat. Proceedings Vienna: R Foundation for Statistical Computing. Retrieved from http:// of the Royal Society of London, Series B, Biological Sciences, 272, www.R-project.org/ 1121–1127. Raia, P., Carotenuto, F., Passaro, F., Fulgione, D., & Fortelius, M. (2012). Tietze, D. T., Päckert, M., Martens, J., Lehmann, H., & Sun, Y. H. (2013). Ecological specialization in fossil mammals explains Cope’s rule. Complete phylogeny and historical biogeography of true rosefinches (Aves: American Naturalist, 179(3), 328–337. Carpodacus). Zoological Journal of the Linnean Society, 169(1), 215–234. Rottenberg, H. (2007). Exceptional longevity in songbirds is associated Töpfer, T., Haring, E., Birkhead, T. R., Lopes, R. J., Severinghaus, L. L., with high rates of evolution of cytochrome b, suggesting selection for Martens, J., & Päckert, M. (2011). A molecular phylogeny of bullfinches reduced generation of free radicals. Journal of Experimental Biology, Pyrrhula Brisson, 1760 (Aves: Fringillidae). Molecular Phylogenetics and 210(12), 2170–2180. Evolution, 58(2), 271–282. Roulin, A. (2015). Condition-dependence, pleiotropy and the handicap Tripp, E. A., & Manos, P. S. (2008). Is floral specialization an evolutionary principle of sexual selection in melanin-based colouration. Biological dead-end? Pollination system transitions in Ruellia (Acanthaceae). Reviews, 91(2), 328–348. Evolution, 62(7), 1712–1737. Safran, R. J., McGraw, K. J., Wilkins, M. R., Hubbard, J. K., & Marling, J. Wiens, J. J., & Graham, C. H. (2005). Niche conservatism: Integrating evo- (2010). Positive carotenoid balance correlates with greater reproduc- lution, ecology, and conservation biology. Annual Review of Ecology and tive performance in a wild bird. PLoS ONE, 5(2), e9420. Systematics, 36, 519–539. Sallan, L., & Galimberti, A. K. (2015). Body-size reduction in vertebrates Wiersma, P., Muñoz-Garcia, A., Walker, A., & Williams, J. B. (2007). Tropical following the end-Devonian mass extinction. Science, 350, 812–815. birds have a slow pace of life. Proceedings of the National Academy of Schmeller, D. S., Henle, K., Loyau, A., Besnard, A., & Henry, P. Y. (2012). Sciences of the United States of America, 104(22), 9340–9345. Bird-monitoring in Europe: A first overview of practices, motivations Wiersma, P., Nowak, B., & Williams, J. B. (2012). Small organ size contrib- and aims. Nature Conservation, 2, 41–57. utes to the slow pace of life in tropical birds. Journal of Experimental Schneeweiss, G. M. (2007). Correlated evolution of life history and host Biology, 215(10), 1662–1669. range in the nonphotosynthetic parasitic flowering plants Orobanche Yotoko, K. S. C., & Elisei, C. (2006). Malaria parasites (Apicomplexa, and Phelipanche (Orobanchaceae). Journal of Evolutionary Biology, 20(2), Haematozoea) and their relationships with their hosts: Is there an evo- 471–478. lutionary cost for the specialization? Journal of Zoological Systematics Sedivy, C., Praz, C. J., Müller, A., Widmer, A., & Dorn, S. (2008). Patterns of host- and Evolutionary Research, 44(4), 265–273. plant choice in bees of the genus Chelostoma: The constraint hypothesis Zuccon, D., Prŷs-Jones, R., Rasmussen, P. C., & Ericson, P. G. P. (2012). The of host-range evolution in bees. Evolution, 62(10), 2487–2507. phylogenetic relationships and generic limits of finches (Fringillidae). Sheldon, P. R. (1996). Plus ça change: A model for stasis and evolu- Molecular Phylogenetics and Evolution, 62(2), 581–596. tion in different environments. Palaeogeography, Palaeoclimatology, Palaeoecology, 127(1), 209–227. Shutler, D., & Weatherhead, P. J. (1990). Targets of sexual selection: Song SUPPORTING INFORMATION and plumage of wood warblers. Evolution, 44(8), 1967–1977. Sih, A., Bell, A., & Johnson, J. C. (2004). Behavioral syndromes: An ecolog- Additional Supporting Information may be found online in the sup- ical and evolutionary overview. Trends in Ecology and Evolution, 19(7), porting information tab for this article. 372–378. Simpson, R. K., Johnson, M. A., & Murphy, T. G. (2015). Migration and the evolution of sexual dichromatism: Evolutionary loss of female coloration with migration among wood-warblers. Proceedings of the Royal How to cite this article: Ponge J-F, Zuccon D, Elias M, et al. Society of London, Series B, Biological Sciences, 282, 20150375. Ancestrality and evolution of trait syndromes in finches Smith, B. T., Bryson, R. W. Jr, Chua, V., Africa, L., & Klicka, J. (2013). (Fringillidae). Ecol Evol. 2017;7:9935–9953. https://doi. Speciational history of North American Haemorhous finches (Aves: Fringillidae) inferred from multilocus data. Molecular Phylogenetics and org/10.1002/ece3.3420 Evolution, 66(3), 1055–1059. Figure S1. Scores (loadings) of 81 fringillid species along the first axis of the four PCAs featuring ecological specialization (a), r-K gradient (b), sexual selection (c) and dispersal/social behaviour (d).

Spinus magellanicus Fringilla montifringilla Erythrina erythrina Acanthis flammea Carpodacus roseus Carduelis carduelis Acanthis flammea Acanthis hornemanni (a) Erythrina erythrina (b) Crithagra striolata Fringilla montifringilla Leucosticte arctoa Haemorhous purpureus Carpodacus rubicilloides Pyrrhula pyrrhula Carpodacus rodochroa Fringilla coelebs Eophona migratoria Chloris sinica Euphonia cayennensis Haematospiza sipahi Fringilla coelebs Spinus psaltria Agraphospiza rubescens Pinicola enucleator Spinus tristis Spinus tristis Haematospiza sipahi Carduelis carduelis Carpodacus thura Spinus pinus Linaria flavirostris Spinus barbatus Pyrrhula pyrrhula Chloris spinoides Bucanetes githagineus Chloris chloris Coccothraustes coccothraustes Serinus pusillus Linaria cannabina Carpodacus rodochroa Carpodacus puniceus Crithagra sulphurata Chloris chloris Rhodospiza obsoleta Carpodacus sibiricus Carduelis citrinella Rhodopechys sanguineus Crithagra rufobrunnea Carpodacus pulcherrimus Crithagra burtoni Chloris sinica Carpodacus rubicilla Carpodacus rhodochlamys Crithagra mozambica Leucosticte nemoricola Crithagra mennelli Spinus spinus Carpodacus rhodochlamys Euphonia xanthogaster Eophona migratoria Spinus atratus Chlors ambigua Haemorhous purpureus Serinus serinus Spinus psaltria Serinus canaria Rhodospiza obsoleta Coccothraustes coccothraustes Eremopsaltria mongolica Spinus olivaceus Carpodacus synoicus Linaria flavirostris Serinus pusillus Acanthis hornemanni Spinus cucullatus Spinus spinus Pinicola enucleator Linaria cannabina Chloris spinoides Hesperiphona vespertina Serinus syriacus Serinus canicollis Leucosticte brandti Procarduelis nipalensis Chloris monguilloti Loxia curvirostra Hesperiphona vespertina Haemorhous mexicanus Euphonia musica Mycerobas carnipes Carpodacus rubicilla Chloris monguilloti Serinus serinus Leucosticte nemoricola Carpodacus roseus Spinus atratus Spinus pinus Crithagra citrinelloides Chloris ambigua Carpodacus rubicilloides Leucosticte tephrocotis Rhodopechys sanguineus Procarduelis nipalensis Carpodacus pulcherrimus Carduelis citrinella Carpodacus puniceus Euphonia chlorotica Spinus cucullatus Callacanthis burtoni Pyrrhula erythaca Haemorhous mexicanus Chlorodrepanis virens Spinus barbatus Euphonia laniirostris Serinus canicollis Carpodacus thura Spinus olivaceus Bucanetes githagineus Serinus canaria Carpodacus sibiricus Crithagra mennelli Callacanthis burtoni Euphonia violacea Serinus syriacus Crithagra striolata Loxia leucoptera Crithagra leucopygia Carpodacus synoicus Crithagra sulphurata Euphonia xanthogaster Crithagra citrinelloides Euphonia chlorotica Pyrrhula erythaca Leucosticte brandti Spinus magellanicus Crithagra leucopygia Chlorodrepanis virens Euphonia musica Mycerobas carnipes Euphonia violacea Euphonia laniirostris Eremopsaltria mongolica Euphonia minuta Loxia pytyopsittacus Loxia leucoptera Leucosticte tephrocotis Loxia pytyopsittacus Agraphospiza rubescens Loxia curvirostra Euphonia cayennensis Paroreomyza montana Leucosticte arctoa Crithagra mozambica Chlorophonia cyanea Crithagra rufobrunnea Paroreomyza montana Chlorophonia cyanea Euphonia minuta Crithagra burtoni Loxioides bailleui Loxioides bailleui -6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6 8 First component of PCA (generalists → specialists) First component of PCA (r-selected → K-selected)

Carduelis carduelis Crithagra leucopygia Spinus pinus Serinus pusillus Linaria cannabina Spinus pinus Eremopsaltria mongolica Paroreomyza montana (c) Linaria flavirostris (d) Crithagra striolata Spinus tristis Serinus citrinella Leucosticte arctoa Leucosticte nemoricola Chloris chloris Euphonia laniirostris Acanthis hornemanni Crithagra rufobrunnea Spinus spinus Crithagra mennelli Loxia pytyopsittacus Euphonia minuta Loxia leucoptera Crithagra mozambica Leucosticte brandti Euphonia violacea Haemorhous mexicanus Serinus serinus Loxia curvirostra Chlorodrepanis virens Serinus canaria Spinus atratus Bucanetes githagineus Leucosticte arctoa Acanthis flammea Loxioides bailleui Spinus barbatus Crithagra citrinelloides Carduelis carduelis Serinus syriacus Rhodopechys sanguineus Serinus canicollis Coccothraustes coccothraustes Serinus canaria Rhodospiza obsoleta Chloris ambigua Carduelis citrinella Euphonia musica Chloris spinoides Spinus psaltria Leucosticte nemoricola Rhodospiza obsoleta Erythrina erythrina Chloris sinica Fringilla montifringilla Crithagra burtoni Spinus psaltria Acanthis hornemanni Serinus syriacus Chloris monguilloti Carpodacus roseus Chloris spinoides Serinus pusillus Leucosticte brandti Serinus serinus Spinus magellanicus Serinus canicollis Euphonia cayennensis Leucosticte tephrocotis Linaria flavirostris Chloris sinica Spinus cucullatus Carpodacus rubicilla Chlorophonia cyanea Chloris ambigua Spinus spinus Spinus magellanicus Haemorhous purpureus Euphonia chlorotica Fringilla coelebs Spinus barbatus Carpodacus synoicus Procardualis nipalensis Chloris monguilloti Spinus olivaceus Carpodacus thura Fringilla coelebs Crithagra mozambica Crithagra sulphurata Pyrrhula pyrrhula Chloris chloris Spinus cucullatus Fringilla montifringilla Carpodacus pulcherrimus Spinus tristis Procarduelis nipalensis Pyrrhula erythaca Crithagra citrinelloides Acanthis flammea Hesperiphona vespertina Eophona migratoria Carpodacus puniceus Linaria cannabina Spinus atratus Leucosticte tephrocotis Mycerobas carnipes Coccothraustes coccothraustes Callacanthis burtoni Callacanthis burtoni Pyrrhula erythaca Rhodopechys sanguineus Crithagra mennelli Pyrrhula pyrrhula Pinicola enucleator Eremopsaltria mongolica Paroreomyza montana Loxia leucoptera Chlorophonia cyanea Haemorhous purpureus Crithagra sulphurata Mycerobas carnipes Euphonia violacea Bucanetes githagineus Euphonia chlorotica Hesperiphona vespertina Carpodacus rhodochlamys Carpodacus roseus Haematospiza sipahi Heamorhous mexicanus Carpodacus rodochroa Carpodacus synoicus Euphonia musica Euphonia xanthogaster Loxioides bailleui Loxia pytyopsittacus Euphonia minuta Erythrina erythrina Carpodacus rubicilloides Loxia curvirostra Eophona migratoria Carpodacus rodochroa Carpodacus sibiricus Carpodacus pulcherrimus Euphonia cayennensis Carpodacus thura Euphonia laniirostris Agraphospiza rubescens Crithagra leucopygia Pinicola enucleator Agraphospiza rubescens Carpodacus puniceus Euphonia xanthogaster Haematospiza sipahi Crithagra burtoni Carpodacus rubicilloides Crithagra striolata Carposacus sibiricus Crithagra rufobrunnea Carpodacus rubicilla Chlorodrepanis virens Carpodacus rhodochlamys Spinus olivaceus -6 -4 -2 0 2 4 6 8 -6 -4 -2 0 2 4 6 8 First component of PCA (sexual monomorphism → sexual dimorphism) First component of PCA (migrants → residents) Appendix S4. Supermatrix tree of Fringillidae for three nuclear introns and five mitochondrial genes.

S 4.1. Assembling the supermatrix

We downloaded all Fringillidae sequences in Genbank and sorted them by taxon and locus. After counting how many taxa were available per locus, we decided to include in the supermatrix three introns and five mitochondrial genes (see locus list below), all of which were represented by at least 50 species. For each sequence, length, associated specimen/sample number and locality were retrieved from Genbank and/or the corresponding publication and combined in a preliminary table. Possible misidentification or anomalous sequences were checked using simple neighbor-joining tree. For each species we selected one sequence per locus. Whenever possible, for a given species we selected the sequences originated from the same specimen or minimized the number of specimens, and we tried also to choose sequences obtained from the same subspecies or with similar geographic origin.

Table S 4.1. Genes, alignment length and number of sequence per gene included in the supermatrix.

Locus Alignment length (bp) Number of sequences Glyceraldehyde-3-phosphate dehydrogenase (GAPDH), intron 11 327 116 (68.6 %) Myoglobin gene, intron 2 741 116 (68.69 %) Ornithine decarboxylase (ODC) gene, introns 6-7 697 112 (66. %) Cytochrome b (cytb) 1143 129 (76.3 %) NADH dehydrogenase subunit II (ND2) 1041 133 (78.7 %) NADH dehydrogenase subunit III (ND3) 351 129 (76.3 %) Cytochrome oxidase subunit I (COI) 648 59 (34.9 %) ATPase subunit 6 (ATP6) 684 81 (47.9 %)

Table S 4.2. Species and Genbank accession numbers included in the supermatrix.

Taxon Specimen GAPDH Myoglobin ODC cytb ND2 ND3 COI ATP6 Fringilla coelebs NRM 956301 JN715183 JN715273 JN715365 JN715455 JN715548

Fringilla teydea NHMO KU705760 KU705760 KU705760 KU705760 KU705760 Fringilla montifringilla NRM 20046395 JN715184 GU816941 GU816920 GU816851 GU816816

Chlorophonia cyanea NRM 20066989 JN715171 JN715263 JN715353 JN715537 Chlorophonia cyanea USNM 625305 JQ174424 Chlorophonia occipitalis no information EF529844 Chlorophonia occipitalis no information EF529956 Euphonia plumbea USNM 632543 JQ174819

Euphonia affinis 529-05 EU442311 Euphonia affinis no information EU442360 Euphonia luteicapilla USNM 612448 JQ174817

Euphonia chlorotica NRM 956750 JN715175 AY228298 JN715357 JN715448 JN715541

Euphonia finschi FMNH 389276 AF290143 Euphonia finschi NRM 20066306 JN715180 JN715270 JN715362 JN715545

Euphonia violacea NRM 966943 JN715181 JN715271 JN715363 JN715453 JN715546

Euphonia laniirostris LSU B18375 AF006232 Euphonia laniirostris LSU B18411 AF447330

Euphonia laniirostris NRM 20066309 JN715176 JN715266 JN715358 JN715449

Euphonia hirundinacea 528-15 EU442301

Euphonia hirundinacea no information EU442333

Euphonia elegantissima KU 87070 JQ174807 Euphonia cyanocephala MACN-Or-ct 913 FJ027571

Euphonia musica no information AF310067 Euphonia musica NRM 976696 JN715178 JN715268 JN715360 JN715451 JN715543

Euphonia fulvicrissa STRI PAEFU102 AF383014 AF383130 AF382975 Euphonia gouldi CU 44255 FJ231698

Euphonia chrysopasta KU 89079 JQ174806

Euphonia minuta NRM 20066307 JN715177 JN715267 JN715359 JN715450 JN715542 Euphonia anneae USNM 607624 JQ174801

Euphonia xanthogaster NRM 20066305 JN715182 JN715272 JN715364 JN715454 JN715547

Euphonia rufiventris NRM 20066310 JN715179 JN715269 JN715361 JN715452 JN715544 Euphonia pectoralis MACN-Or-ct 2867 FJ027574

Euphonia cayennensis NRM 20056062 JN715174 JN715265 JN715356 JN715540

Mycerobas icterioides FMNH 256357 KJ456356 Mycerobas affinis AMNH 5592 KJ456354

Mycerobas melanozanthos FMNH 222122 KJ456357

Mycerobas carnipes NRM 23363 JN715196 JN715284 JN715376 JN715466 JN715558 Hesperiphona vespertina NRM 23336 JN715186 JN715275 JN715367 JN715457 JN715550

Hesperiphona vespertina UMMZ 235311 KM078770 KM078770 KM078770

Coccothraustes coccothraustes NRM 976374 JN715172 AY228292 JN715354 AY228055 JN715446 JN715538 GU571828 Eophona migratoria no information AF342871

Eophona migratoria NRM 896473 JN715173 JN715264 JN715355 JN715447 JN715539

Eophona personata no information AF342872

Melamprosops phaeosoma BPBM 147112 KM112521 KM112827 KM078793 KM078793 KM078793 KM078793 KM078793

Paroreomyza montana NZP USFWS band 2141-67177 KM078771 KM078771 KM078771

Paroreomyza montana RCF 1984 JN715197 JN715285 JN715377 JN715467 JN715559

Dysmorodrepanis munroi NZP USFWS band 1581-75367 KM112533 KM112815 KM078774 KM078774 KM078774 KM078774 KM078774 Telespiza cantans NZP USFWS band 8061-85532 KM112508 KM112859 KM078777 KM078777 KM078777 KM078777 KM078777

Telespiza ultima Conant 92.7.26-1 KM112509 KM112811 KM078787 KM078787 KM078787 KM078787

Loxioides bailleui MRSNT 5783 JN715195 JN715283 JN715375 JN715465 JN715557

Loxioides bailleui NZP USFWS band 8031-75805 KM078805 KM078805

Psittirostra psittacea no information KU158196 KU158196 KU158196 KU158196 KU158196 Oreomystis bairdi MVZ 178406 KM112527 KC007674 KM112853 KM078807 KM078807 KM078807 KM078807 KM078807

Magumma parva MVZ 178407 KM112536 KM112805 KM078799 KM078799 KM078799 KM078799 KM078799

Loxops caeruleirostris MVZ 178405 KM112505 KM112834 KM078776 KM078776 KM078776 KM078776 KM078776 Loxops coccineus NZP USFWS band 2350-43262 KM112523 KM112807 KM078785 KM078785 KM078785 KM078785 KM078785

Manucerthia mana TKP 559 KM112532 KM112820 KM078768 KM078768 KM078768 KM078768 KM078768

Chlorodrepanis virens NZP RCF-3425 KM078788 KM078788 KM078788

Chlorodrepanis virens RCF 2913 JN715225 JN715313 JN715405 JN715496 JN715588

Chlorodrepanis flava NZP USFWS band 1680-10360 KM112537 KM112821 KM078780 KM078780 KM078780 KM078780 KM078780

Chlorodrepanis stejnegeri NZP RCF-2666 KM112512 KM112809 KM078801 KM078801 KM078801 KM078801 KM078801 Hemignathus wilsoni NZP RCF-3422 KM112535 KM112823 KM078802 KM078802 KM078802 KM078802 KM078802

Akialoa obscura no information KU158190 KU158190 KU158190 KU158190 KU158190

Pseudonestor xanthophrys NZP USFWS band 8051-40968 KM112541 KM112840 KM078809 KM078809 KM078809 KM078809 KM078809 Vestiaria coccinea NZP USFWS band 1471-19056 KM112546 KM112847 KM078797 KM078797 KM078797 KM078797 KM078797

Himatione sanguinea NZP RCF-3427 KM112501 KM112808 KM078773 KM078773 KM078773 KM078773 KM078773

Palmeria dolei BPBM 184556 KM112499 KM112801 KM078803 KM078803 Erythrina erythrina NRM 976373 JN715154 JN715246 JN715336 JN715428 JN715519 GU571800

Erythrina erythrina USNM B18937 KM078766 KM078766 KM078766

Haematospiza sipahi no information AF342875 Haematospiza sipahi NRM 23812 JN715185 JN715274 JN715366 JN715456 JN715549 Chaunoproctus ferreorostris BMNH 1855.12.19.71 JN715445 JN715536

Carpodacus synoicus MAR 5084 KF194098 Carpodacus synoicus NHMO 26633 JN715168 JN715260 JN715350 JN715442 JN715533

Carpodacus stoliczkae BMNH 1967.17803 KF194009 KF194103 KF194147

Carpodacus roborowskii NRM 23882 JN715161 JN715253 JN715343 JN715435 JN715526 Carpodacus rubicilloides MAR 5931 KF194089

Carpodacus rubicilloides NRM 23864 JN715167 JN715259 JN715349 JN715441 JN715532

Carpodacus rubicilla NRM 20016594 JN715166 JN715258 JN715348 JN715440 JN715531 Carpodacus rubicilla ZFMK J.II.22.f.l KF194088

Carpodacus sibiricus NRM 20076294 JN715224 JN715312 JN715404 JN715495 JN715587 Carpodacus sibiricus UWBM 47588 KM078763 KM078763 KM078763

Carpodacus puniceus FMNH 277071 KF194173 Carpodacus puniceus NRM 23927 JN715158 JN715250 JN715340 JN715432 JN715523

Carpodacus subhimachalus FMNH 277061 KF194180 Carpodacus subhimachalus NRM 23477 JN715199 JN715287 JN715379 JN715469 JN715561 Carpodacus roseus NRM 20026495 JN715164 JN715256 JN715346 JN715438 JN715529 Carpodacus roseus USNM B10230 KM078779 KM078779 KM078779 Carpodacus trifasciatus MTD C63834 KF194004 KF194019 KF194115 KF194166

Carpodacus thura AMNH DOT5609 KF194034 KJ455736 KF194181 KF194135

Carpodacus dubius MAR 369 KF194106

Carpodacus dubius NRM 20016581 JN715169 JN715261 JN715351 JN715443 JN715534

Carpodacus rhodochlamys NRM 20026491 JN715160 JN715252 JN715342 KF194174 JN715434 JN715525 Carpodacus grandis ZFMK J.II.22.g1.g KF194010 KF194051 KF194152

Carpodacus davidianus MTD C64235 KF193991 KF194018 KF194060 KF194156 Carpodacus pulcherrimus NRM 20026494 JN715157 JN715249 JN715339 KF194172 JN715431 JN715522

Carpodacus waltoni MAR 5936 KF193993 KF194017 KF194057 KF194141

Carpodacus edwardsii MTD C24119 KF194039 Carpodacus verreauxii CAS 95886 KF194033 KF194169 KF194128

Carpodacus verreauxii MNHN 19-37 EU880942

Carpodacus rodochroa AMNH DOT5671 KF194176 Carpodacus rodochroa NRM 553889 JN715162 JN715254 JN715344 JN715436 JN715527

Carpodacus rodopeplus RMNH 44517 JN715163 JN715255 JN715345 JN715437 JN715528

Carpodacus formosanus NMNST 7838 KF193996 KF194006 KF194119 KF194167

Carpodacus vinaceus MNHN 19.23 EU880943

Carpodacus vinaceus NRM 20026493 JN715170 JN715262 JN715352 JN715444 JN715535

Carpodacus vinaceus U Mainz MAR 961 HQ284695

Pinicola enucleator NRM 996174 JN715198 JN715286 JN715378 HQ284684 JN715468 JN715560 GU572046 Pyrrhula nipalensis MAR ES11 KF194184

Pyrrhula nipalensis NRM 23476 JN715202 JN715290 JN715382 JN715472 JN715564

Pyrrhula leucogenis CMC 36632 HQ284782 HQ284678

Pyrrhula aurantiaca BMNH 1949.25.3775 HQ284579

Pyrrhula erythrocephala MNHN 18-05 EU878710 EU881015 EU880949 Pyrrhula erythrocephala U Mainz MAR 90021 HQ284642

Pyrrhula erythaca NRM 20016568 JN715201 JN715289 JN715381 HQ284652 JN715471 JN715563

Pyrrhula murina U Sheffield TRB2 HQ284768 KX109679 KX109755 HQ284631 KX109700 Pyrrhula pyrrhula MNHN 95-25 EU880950

Pyrrhula pyrrhula NRM 20046541 JN715203 JN715291 JN715383 JN715473 JN715565 GU572069

Pyrrhula pyrrhula U Mainz MAR 4437 HQ284595

Rhodopechys sanguineus NRM 20026504 JN715207 JN715295 JN715387 JN715477 JN715569

Bucanetes githagineus MTD C64524 MAR1172 HQ284701

Bucanetes githagineus NRM 20046702 JN715204 JN715292 JN715384 JN715474 JN715566

Eremopsaltria mongolica NRM 23495 JN715205 JN715293 JN715385 JN715475 JN715567 Eremopsaltria mongolica USNM B10185 KM078792 KM078792 KM078792

Agraphospiza rubescens MART 90161 KF194120

Agraphospiza rubescens NRM 23826 JN715165 JN715257 JN715347 JN715439 JN715530

Callacanthis burtoni NRM 24869 JN715135 JN715227 JN715317 JN715409 JN715500

Pyrrhoplectes epauletta MNHN 16-2j EU880948

Pyrrhoplectes epauletta MTD C63109 MAR5549 HQ284689

Pyrrhoplectes epauletta NRM 23810 JN715200 JN715288 JN715380 JN715470 JN715562

Procarduelis nipalensis no information AF342866 Procarduelis nipalensis NRM 23824 JN715156 JN715248 JN715338 JN715430 JN715521

Leucosticte nemoricola IZAS uncat JN715189 JN715278 JN715370 JN715460 JN715553 Leucosticte nemoricola MTD C62013 MAR4246 HQ284700

Leucosticte brandti NRM 24575 JN715188 JN715277 JN715369 JN715459 JN715552 Leucosticte brandti UWBM 66356 KM078775 KM078775 KM078775

Leucosticte sillemi ZMA.AVES 43449 KT444635

Leucosticte arctoa NRM 24559 JN715187 JN715276 JN715368 JN715458 JN715551 Leucosticte arctoa USNM 609762 KM078791 KM078791 KM078791

Leucosticte tephrocotis NRM 20016579 JN715190 JN715279 JN715371 JN715461 Leucosticte tephrocotis UWBM 66944 SVD 2371 KX109627 Haemorhous mexicanus NRM 20056140 JN715155 JN715247 JN715337 JN715429 JN715520

Haemorhous mexicanus NZP RCF-2645 KM078782 KM078782 KM078782

Haemorhous cassinii USNM B20075 KM112518 KM112813 KM078786 KM078786 KM078786 KM078786 KM078786

Haemorhous purpureus NRM 557743 JN715159 JN715251 JN715341 JN715433 JN715524

Haemorhous purpureus UMMZ 227048 T-126 KM078772 KM078772

Rhodospiza obsoleta NRM 20046707 JN715206 JN715294 JN715386 JN715476 JN715568

Rhynchostruthus socotranus NRM 23483 JN715208 JN715296 JN715388 JN715478 JN715570

Chloris chloris MNHN C34 EU880931 Chloris chloris NRM 986328 JN715141 JN715233 JN715323 JN715415 JN715506 GU571791

Chloris chloris SMTD C 62175 HQ284692

Chloris sinica NRM 20026538 JN715150 JN715242 JN715332 JN715424 JN715515

Chloris sinica USNM B19440 KM078783 KM078783 KM078783

Chloris spinoides NRM 20026503 JN715151 JN715243 JN715333 KJ456211 JN715425 JN715516 Chloris spinoides ZMUC 131646 EU880974 Chloris monguilloti NRM 546196 JN715147 JN715239 JN715329 JN715421 JN715512

Chloris ambigua no information U78322 Chloris ambigua NRM 20026539 JN715136 JN715228 JN715318 JN715410 JN715501

Chloris ambigua ZMUC 118651 EU880992

Linurgus olivaceus MNHN 40-23 EU880938

Linurgus olivaceus NRM 20086232 JN715191 JN715280 JN715372 JN715462 JN715554

Crithagra citrinelloides no information AY790888 Crithagra citrinelloides NRM 20026501 JN715212 JN715300 JN715392 JN715482 JN715574

Crithagra citrinelloides ZMUC 134740 EU880968 Crithagra hyposticta ZMUC 131127 EU878731 EU881039 EU880971

Crithagra capistrata ZMUC 135500 EU878742 EU881052 EU880984

Crithagra scotops no information AY790894

Crithagra leucopygia no information L76264 Crithagra leucopygia NRM 20106050 JN715214 JN715302 JN715394 JN715485 JN715577

Crithagra leucopygia ZMUC 118662 EU880993

Crithagra atrogularis no information L76267

Crithagra citrinipectus ZMUC 130559 EU878717 EU881025 EU880957 Crithagra mozambica MNHN 40-54 EU880953

Crithagra mozambica no information L76265

Crithagra mozambica NRM 20066026 JN715216 JN715304 JN715396 JN715487 JN715579

Crithagra flaviventris no information AY790887 Crithagra flaviventris ZMUC 132129 EU878732 EU881040 EU880972

Crithagra dorsostriata UMMZ 233806 T-810c KM112547 KM112837 KM078798 KM078798 KM078798 KM078798 KM078798

Crithagra sulphurata no information AY790889 Crithagra sulphurata NRM 20026498 JN715221 JN715309 JN715401 JN715492 JN715584

Crithagra albogularis UWBM 70375 KM112520 KM112836 KM078764 KM078764 KM078764 KM078764 KM078764

Crithagra reichardi ZMUC 122436 EU878734 EU881043 EU880975 Crithagra gularis no information L77556

Crithagra gularis ZMUC 131650 EU881041 EU880973

Crithagra mennelli NRM 20026500 JN715215 JN715303 JN715395 JN715486 JN715578 Crithagra mennelli ZMUC 118671 EU880990

Crithagra striolata no information AY790895

Crithagra striolata NRM 23718 JN715220 JN715308 JN715400 JN715491 JN715583

Crithagra striolata ZMUC 134756 EU880969

Crithagra burtoni no information AY790896 Crithagra burtoni NRM 20086267 JN715209 JN715297 JN715389 JN715479 JN715571

Crithagra burtoni ZMUC 123647 EU880997

Crithagra melanochroa ZMUC 118856 EU878737 EU881046 EU880978

Crithagra rufobrunnea NRM 857618 JN715218 JN715306 JN715398 JN715489 JN715581

Crithagra totta no information AY790892

Linaria flavirostris JM136 KJ456172 Linaria flavirostris NRM 20066634 JN715144 JN715236 JN715326 JN715418 JN715509 GU571794

Linaria cannabina MNHN 25-82 EU880939

Linaria cannabina NRM 966403 JN715139 JN715231 JN715321 JN715413 JN715504 GU571787

Linaria cannabina SMTD C 61033 HQ284691

Acanthis flammea no information L76386 Acanthis flammea NRM 20016449 JN715143 JN715235 JN715325 JN715417 JN715508 GU571793

Acanthis hornemanni AJN 000043 JN715145 JN715237 JN715327 JN715419 JN715510 GU571795

Acanthis hornemanni no information U83201

Acanthis hornemanni ZMUC 119818 EU880989

Loxia pytyopsittacus NRM 20046001 JN715194 JN715282 JN715374 JN715464 JN715556 GU571962 Loxia pytyopsittacus ZMUC 119824 EU880986

Loxia scotica no information AF171656

Loxia curvirostra NRM 976546 JN715192 AY228303 GU816921 AY228065 GU816852 GU816817 GU571958

Loxia leucoptera no information AF342878

Loxia leucoptera NRM 20026565 JN715193 JN715281 JN715373 JN715463 JN715555 GU571960

Loxia leucoptera ZMUC 116705 EU880962

Chrysocorythus estherae RMNH 44712 JN715483 JN715575 Carduelis carduelis NRM 996076 JN715140 JN715232 JN715322 JN715414 JN715505 GU571789

Carduelis carduelis NZP USFWS band 1760-76299 KM078790 KM078790

Carduelis citrinella MNHN C38 EU880933 Carduelis citrinella no information L77872

Carduelis citrinella NRM 553307 JN715211 JN715299 JN715391 JN715481 JN715573

Serinus serinus no information L76263

Serinus serinus NRM 20046491 JN715219 JN715307 JN715399 JN715490 JN715582 GU572089

Serinus serinus ZMUC 116715 EU880980

Serinus canaria NRM 20026502 JN715213 JN715301 JN715393 JN715484 JN715576 Serinus canaria NZP 105 KM078794 KM078794 KM078794

Serinus syriacus NRM 23600 JN715222 JN715310 JN715402 JN715493 JN715585

Serinus pusillus JM128 KJ456465 Serinus pusillus NRM 20046715 JN715217 JN715305 JN715397 JN715488 JN715580

Serinus pusillus ZMUC 135626 EU880983

Serinus alario no information AY790899

Serinus canicollis no information AY790890 Serinus canicollis NRM 20076189 JN715210 JN715298 JN715390 JN715480 JN715572

Serinus canicollis ZMUC 128624 EU880958

Serinus flavivertex no information AY790897

Spinus thibetanus BMNH 1948.34.64 JN715223 JN715311 JN715403 JN715494 JN715586 Spinus thibetanus no information L76279

Spinus spinus MNHN 2000.1651 EU880941

Spinus spinus no information L76391

Spinus spinus NRM 986184 JN715152 JN715244 JN715334 JN715426 JN715517 GU571799

Spinus pinus NRM 20016375 JN715148 JN715240 JN715330 JN715422 JN715513 Spinus pinus UMMZ 227858 KM078796 KM078796 KM078796 Spinus tristis MSB 29161 KT221304

Spinus tristis NRM 20016378 JN715153 JN715245 JN715335 JN715427 JN715518

Spinus psaltria NRM 20016376 JN715149 JN715241 JN715331 JN715423 JN715514 Spinus psaltria NZP USFWS band 2120-99480 KM078806 KM078806 KM078806

Spinus lawrencei MVZ 176985 KT221368 KT221302 KT221172

Spinus atriceps 01N6050 KT358782 Spinus atriceps AMNH DOT7151 KT221375 KT221309 KT221179 KT221243 Spinus spinescens ANSP 18807 KT221360 KT221230 KT221294

Spinus spinescens ZMUC 120509 EU880985

Spinus yarrellii no information U83200

Spinus cucullatus NRM 20026508 JN715142 JN715234 JN715324 JN715416 JN715507

Spinus cucullatus USNM B12867 KT221318

Spinus cucullatus ZMUC 135619 EU880982

Spinus crassirostris MSB 34207 KT221383 KT221317 KT221187 KT221251 Spinus crassirostris ZMUC 129671 EU880959

Spinus magellanicus FMNH 334723 KT221349

Spinus magellanicus NRM 986696 JN715146 JN715238 JN715328 JN715420 JN715511

Spinus magellanicus ZMUC 120822 EU880988

Spinus dominicensis AMNH DOT6930 KT221387 KT221322 KT221192 KT221256 Spinus siemiradzkii ANSP 19701 KT221358

Spinus siemiradzkii ZMUC 116681 EU878725 EU881033 EU880965

Spinus olivaceus ANSP 19746 KT221354 KT221224 KT221288 Spinus notatus 03N1083 KT358784

Spinus notatus FMNH 393899 KT221417 KT221352 KT221222 KT221286

Spinus xanthogastrus ANSP 18543 KT221365 Spinus xanthogastrus ZMUC 116687 EU878721 EU881029 EU880961 Spinus atratus MSB 34123 KT221305

Spinus atratus NRM 546071 JN715137 JN715229 JN715319 JN715411 JN715502

Spinus atratus ZMUC 118878 EU880994

Spinus uropygialis MSB 33471 KT221422 KT221364 KT221234 KT221298 Spinus uropygialis ZMUC 116685 EU880964

Spinus barbatus AMNH DOT10430 KT221310 Spinus barbatus NRM 546142 JN715138 JN715230 JN715320 JN715412 JN715503

Motacilla alba 100 Anthus trivialis Petronia petronia 100 96 Montifringilla ruficollis 100 Passer luteus 100 Passer montanus Plectrophenax nivalis 100 Ammodramus humeralis 91 Parula pitiayumi 68 Leistes superciliaris Fringilla montifringilla 100 Fringilla coelebs 90 Fringilla teydea Euphonia elegantissima 100 Euphonia cyanocephala

E 45 n 80 Euphonia musica t 99 Euphonia chrysopasta 49 Euphonia anneae 75 Euphonia pectoralis 36 31 Euphonia chlorotica 95 Euphonia finschi 92 Euphonia affinis 36 Euphonia luteicapilla Euphonia plumbea 25 95 100 Euphonia minuta 81 Euphonia hirundinacea 86 Euphonia laniirostris 99 Euphonia violacea 85 Euphonia gouldi 100 Euphonia fulvicrissa 97 Euphonia xanthogaster 56 Euphonia cayennensis 99 Euphonia rufiventris Coccothraustes coccothraustes 55 Hesperiphona vespertina Mycerobas melanozanthos 100 97 Mycerobas icterioides 99 Mycerobas affinis 31 Mycerobas carnipes 100 45 Eophona personata 100 Eophona migratoria Pinicola enucleator Pyrrhula nipalensis 98 Pyrrhula leucogenis 98 Pyrrhula pyrrhula 98 100 Pyrrhula murina 58 Pyrrhula aurantiaca Pyrrhula erythaca 81 48 88 Pyrrhula erythrocephala Rhodopechys sanguineus 100 Eremopsaltria mongolica 100 Bucanetes githagineus Agraphospiza rubescens 60 89 Pyrrhoplectes epauletta 100 Callacanthis burtoni 98 Procarduelis nipalensis Leucosticte nemoricola 100 90 Leucosticte brandti 100 Leucosticte arctoa 100 Leucosticte tephrocotis Melamprosops phaeosoma 32 Paroreomyza montana 99 99 64 Oreomystis bairdi Loxioides bailleui 100 75 Telespiza ultima 100 Telespiza cantans Akialoa obscura 100 Psittirostra psittacea 28 Vestiaria coccinea 46 100 Palmeria dolei 55 Himatione sanguinea Pseudonestor xanthophrys 35 37 Dysmorodrepanis munroi 21 Magumma parva 75 Manucerthia mana 69 100 Loxops caeruleirostris 94 Loxops coccineus 94 Chlorodrepanis stejnegeri 100 Chlorodrepanis flava 50 Chlorodrepanis virens 100 Hemignathus wilsoni Haematospiza sipahi 100 Erythrina erythrina Chaunoproctus ferreorostris Carpodacus synoicus 81 100 Carpodacus stoliczkae 32 Leucosticte sillemi 100 Carpodacus roborowskii Carpodacus puniceus 99 77 34 97 Carpodacus subhimachalus 63 Carpodacus sibiricus Carpodacus roseus 32 100 Carpodacus trifasciatus 69 100 Carpodacus thura 100 Carpodacus dubius Carpodacus rubicilloides 100 Carpodacus rubicilla Carpodacus grandis 18 100 Carpodacus rhodochlamys Carpodacus waltoni 70 98 Carpodacus pulcherrimus 96 Carpodacus davidianus 46 Carpodacus edwardsii Carpodacus rodochroa 44 99 Carpodacus verreauxii 81 Carpodacus rodopeplus 70 Carpodacus formosanus 94 Carpodacus vinaceus Haemorhous mexicanus 96 Haemorhous cassinii 100 Haemorhous purpureus Rhodospiza obsoleta 100 Rhynchostruthus socotranus 100 Chloris chloris Chloris sinica 100 91 Chloris ambigua 94 100 Chloris spinoides 53 Chloris monguilloti Linurgus olivaceus Crithagra rufobrunnea Crithagra scotops Crithagra capistrata 29 24 100 Crithagra hyposticta 95 23 95 Crithagra citrinelloides 98 Crithagra dorsostriata Crithagra mozambica 82 83 Crithagra citrinipectus 83 99 Crithagra atrogularis 96 Crithagra leucopygia Crithagra burtoni 17 Crithagra totta 26 Crithagra striolata 69 Crithagra melanochroa 16 Crithagra albogularis 89 97 Crithagra flaviventris 100 Crithagra sulphurata 98 Crithagra reichardi 99 Crithagra gularis 89 Crithagra mennelli Acanthis hornemanni 100 Acanthis flammea 100 Loxia leucoptera 100 Loxia pytyopsittacus 100 Loxia scotica 94 Loxia curvirostra 52 Linaria flavirostris 99 Linaria cannabina Chrysocorythus estherae 67 Carduelis carduelis 90 91 Carduelis citrinella Serinus serinus 100 Serinus canaria 31 99 Serinus alario Serinus syriacus 99 83 Serinus pusillus 24 Serinus canicollis 59 97 Serinus flavivertex Spinus thibetanus Spinus tristis 99 Spinus psaltria 97 69 Spinus lawrencei Spinus dominicensis 81 82 Spinus spinus 91 Spinus atriceps 100 Spinus pinus 100 Spinus notatus Spinus cucullatus 100 Spinus barbatus 100 Spinus olivaceus 4960 Spinus spinescens Spinus uropygialis 3157 Spinus siemiradzkii 98 Spinus crassirostris 12 Spinus xanthogastrus 19 Spinus atratus 16 Spinus yarrellii 48 Spinus magellanicus