The dual role of Andean topography in primary divergence: functional and neutral variation among populations of the hummingbird, Metallura tyrianthina

Abstract

Background: The ridges and valleys of the Andes create physical barriers that limit animal dispersal and cause deterministic local variation in rainfall. This has resulted in physical isolation of animal populations and variation in habitats, each of which has likely contributed to the evolution of high species diversity in the region. However, the relative influences of geographic isolation, ecoclimatic conditions, and their potential interactions remain poorly understood. To address this, we compared patterns of genetic and morphological diversity in Peruvian populations of the hummingbird Metallura tyrianthina.

Results: Phylogenetic and variation partitioning analyses showed that geographic isolation rather than climatic dissimilarity explained the greatest proportion of genetic variance. In contrast, bill length variation was explained by climatic seasonality, but not by genetic divergence. We found that mutation-scaled migration rate (m) between persistently humid and semi-humid environments was nearly 20 times higher when the habitats were contiguous (m = 39.9) than when separated by a barrier, the Cordillera de Vilcanota (m = 2.1). Moreover, the population experiencing more gene flow exhibited a lesser degree of bill length divergence despite similar differences in climate.

Conclusions: Geographic isolation is necessary for genetic divergence. Ecological differences, represented here by climate characteristics, are necessary for functional divergence. Gene flow appears to hinder the evolution of functional traits toward local adaptive optima. This suggests that functional diversification requires geographic isolation followed or accompanied by a shift in ecological conditions. Andean topography causes both isolation and climatic variation, underscoring its dual role in biotic diversification.

Fig. 1

a Map illustrating the distribution of all three Metallura tyrianthina subspecies that occur in Peru. Circles denote sampling sites for both genetic and bill length data; squares represent sampling sites for bill length data only. Blue points are sites where the habitat was characterized as humid montane forest and brown points semi-humid montane scrub. b Map of sampling sites from the department of Cusco, Peru. One pair of sites was from the humid (Carrizales) and drier (Urubamba) sides of the Cordillera de Vilcanota, separated by ~20 km. A second pair of sites was along the Manu Road, with a humid (Pillahuata) and drier (Paucartambo) site ~5 km apart. c Topographic profile of the Cordillera de Vilcanota, which exceeds 5000 m, and the Manu Road where a cordillera less than 4000 m in elevation separates the two sites along the Manu Road

Fig. 2

a ND2 Bayesian phylogeny for Peruvian populations of Metallura tyrianthina. Posterior probabilities for each node printed above branches (* signifies 1.0 pp). The six geographically structured clades are lettered on all figures. Each clade is colored by subspecies as in panel c. b Median-joining haplotype network of ND2. Brown color indicates semi-humid montane scrub habitat; blue indicates humid montane forest. c Geographic distribution of each clade (A-F). Red dotted lines signify putative physical barriers that isolate the six clades

Fig. 3

Linear regression of bill length versus a pairwise genetic differences (D_A) and b linearized Fst among all populations. No significant relationships were found across all comparisons. c Boxplot of mean divergence in bill length between all humid sites, semi-humid sites, and among sites in different habitats. Mean divergence between habitats was significantly greater than that between habitats, whereas within habitat divergence did not differ

Fig. 4

Bill length as a function of PC1 of seasonality (82.9 % of the variation). The linear regression model was highly significant (p < 0.0001) with an R² of 0.44

Fig. 5

a Topographic profiles of the two cordilleras separating sampling sites in the Vilcanota region and along the Manu Road (see Fig. 1c). Arrows with red lettering indicate mean migration rates (m) estimated from IMa2 analyses (Table 4). b Bill length differences among the four Cusco localities. Blue represents the two humid sites and brown the two drier sites. Asterisks indicate p-values for the comparisons of bill length between two pairs of adjacent populations in humid and semi-humid environments (* < 0.05; ** < 0.001; *** < 0.0001)