Synthesis of geological data and comparative phylogeography of lowland tetrapods suggests recent dispersal through lowland portals crossing the Eastern Andean Cordillera

Abstract

Vicariance is the simplest explanation for divergence between sister lineages separated by a potential barrier, and the northern Andes would seem to provide an ideal example of a vicariant driver of divergence. We evaluated the potential role of the uplift of the Eastern Cordillera (EC) of the Colombian Andes and the Mérida Andes (MA) of Venezuela as drivers of vicariance between lowland populations co-distributed on both flanks. We synthesized published geological data and provided a new reconstruction showing that the EC-MA grew from north to south, reaching significant heights and separating drainages and changing sediment composition by 38-33 million years ago (Ma). A few lowland passes across the EC-MA may have reached their current heights (~1,900 m a.s.l.) at 3-5 Ma. We created a comparative phylogeographic data set for 37 lineages of lowland tetrapods. Based on molecular phylogenetic analyses, most divergences between sister populations or species across the EC-MA occurred during Pliocene and the Quaternary and a few during the latest Miocene, and coalescent simulations rejected synchronous divergence for most groups. Divergence times were on average slightly but significantly more recent in homeotherms relative to poikilotherms. Because divergence ages are mostly too recent relative to the geological history and too asynchronous relative to each other, divergence across the northern Andes may be better explained by organism-environment interactions concomitant with climate oscillations during the Pleistocene, and/or dispersal across portals through the Andes.

Keywords: Andean portals; Andean uplift; Climate; Comparative phylogeography; Divergence times; Eastern Cordillera; Hierarchical approximate Bayesian computation; Mérida Andes; Orogeny.

Figure 1. Topography and main drainages in the northern Andes. Red triangles represent Pliocene and younger volcanic edifices.

Diagonal hatching represents elevations greater than 2,000 m. Inset shows sample localities from across South America and Central America per each of the four taxonomic classes of tetrapod studied here. The Eastern Cordillera (EC) and its northern extensions, the Santander Massif and Perijá Range (see inset), extend ~1,200 km from southern Colombia to northwestern Venezuela with summits reaching 5,300 m a.s.l. The Mérida Andes (MA) extend ~350 km to the NE reaching maximum elevations of ~5,000 m a.s.l. The EC joins the Central Cordillera across a low-elevation bridge (~2,000 m a.s.l., between the Putumayo corridor and Andalucia pass), while it splits from the MA at the low-elevation Táchira corridor (~1,000 m a.s.l.), constituting the lowest points of the eastern Andes.

Figure 2. Paleogeographic reconstruction at middle Miocene times (~13 Ma, modified from Montes et al., 2021).

Stippled areas in the figure represent sedimentary environments interpreted from preserved rock sequences; numbers indicate the stratigraphic sequence where it is preserved (1: Grosse, 1926; 2: Parnaud et al., 1995; Erikson et al., 2012; 3: Guerrero, 1997; Montes et al., 2021; 4: Gómez et al., 2005; 5: Montes et al., 2010; 6: Quiroz et al., 2010; 7: Borrero et al., 2012; 8: Barat et al., 2014; 9: van der Hammen, Werner & van Dommelen, 1973; Molnar & Pérez-Angel, 2021; 10: Echeverri et al., 2015; Gallego-Ríos et al., 2020 (10); Weber et al., 2020; 11: Moreno et al., 2015; 12: Farris et al., 2017; 13: Jaramillo et al., 2017; Hoorn et al., 2022; 14: León et al., 2018). Most clastic deposits at this time are fluvial sands and near-shore environments, and mark the segmentation of basins by rising mountain belts and northward-propagating magmatic belts. The absolute elevation of the areas shaded as “significant relief” is unknown, but somewhere in the northern Andes there was proto-paramo vegetation suggestive of ~3 km elevation at this time (Hoorn et al., 2022).

Figure 3. Assumptions of the comparative phylogeographic analysis in msBayes using hABC (see Methods) required sub-sampling the Bayesian phylogenetic consensus trees.

Arrows mark the subclades selected for analysis by hABC according to the possible tree topology. For reciprocally monophyletic groups with respect to the EC-MA (A) all data were used. For paraphyletic groups (B) we sampled only the clade that fit the two-population model assumed by msBayes. For polyphyletic groups (C) we selected the clade that adjusted to the two-population model with sampling localities geographically closest to the Eastern Cordillera (EC). COL = Colombia, ECU = Ecuador, GUY = Guiana.

Figure 4. Distribution of divergence time estimates between eastern and western lineages of (A) frogs, (B) birds, (C) mammals and (D) non-avian reptiles, estimated by Bayesian MCMC relaxed-clock phylogenetic analysis of mitochondrial DNA.

Dots indicate the mean divergence time and dotted lines indicate the median divergence time within each class. Median divergence time in frogs was 3.24 million years ago (Ma), 2.99 Ma in non-avian reptiles, 2.44 Ma in birds, and 1.44 Ma in mammals.

Figure 5. Mean divergence times (in million of years ago, Ma) of eastern and western populations as estimated by Bayesian molecular phylogenetic inference of mitochondrial DNA sequence data for homeotherm and poikilotherm tetrapods.

The points represent the mean divergence time of each of 37 data sets (GLM, df = 35, t = −2.43, p = 0.02, adjusted R² = 0.12; mean node age in poikilotherms = 2.35 Ma; mean node age in homeotherms = 4.46 Ma).