Aim
To investigate the cryptic diversity and diversification timing in the putatively low‐dispersal Amazonian leaf‐litter lizard Loxopholis osvaldoi, and to ask how geography (rivers, isolation by distance, IBD), ecological drivers (isolation by environment, IBE) and historical factors (climatic refugia) explain intraspecific genetic variation.
Location
Central Amazonia, Brazil.
Taxon
Squamata; Gymnophthalmidae; Loxopholis osvaldoi.
Methods
We sequenced two mitochondrial and two nuclear markers in 157 individuals. Phylogeographic structure and the occurrence of independent evolving lineages where explored through phylogenetic and coalescent analyses. A species tree and divergence dates of lineages were inferred with *BEAST, employing multiple DNA substitution rates. The potential genetic impacts of geographical distance among localities, the environment and the position of localities in relation to main rivers were tested by redundancy analysis (RDA).
Results
We detected 11 independently evolving and largely divergent intraspecific lineages. Lineage distribution patterns are complex and do not match any conspicuous barrier to gene flow, except for the Amazon River. Most lineages appear to have originated in the lower Miocene and Pliocene, in disagreement with the Pleistocene refuge hypothesis. IBD, IBE and rivers appear to have acted in concert establishing and maintaining genetic structure. However, when controlling for other explanatory variables, IBD explains significantly more variation than rivers, IBE or historical factors.
Main Conclusions
Our results strongly suggest that L. osvaldoi is a species complex. Future taxonomic work should use an integrative approach to explore whether morphological variation is present and congruent with the genetic data. While the use of a sensitive dating analysis allowed us to better describe the diversification history of L. osvaldoi, the lack of a spatial model of Neogene river dynamics prevents the test of specific, more informative river barrier hypotheses. The data suggest that nonlinear correlation analyses (e.g. RDA) should be preferred to detect factors that affect phylogeographic patterns in the Amazon, instead of linear multiple regressions (e.g. Mantel tests). Given the high level of cryptic diversity detected within this and other Amazonian species, we caution against hypothesis tests based solely on the distribution of nominal taxa, which can provide a rather incomplete view of the processes behind Amazonian diversity.