Comparing range evolution in two western Plethodon salamanders: glacial refugia, competition, ecological niches, and spatial sorting

Abstract

Aim Plethodon dunni and P. vehiculum are sister species with widely overlapping ranges, yet the distribution of P. vehiculum extends almost 400 km farther north than its sister species. We explore Pleistocene refugial structure, competition, physiological tolerances and dispersal ability as contributing factors to the range difference between these two species. Location The Pacific Northwest of North America, including Oregon, Washington and British Columbia – the full range of both species. Methods We used genetic, environmental and morphological data to test hypotheses that explain this range difference. Genetic data were used to explore range expansion dynamics and population structure. Species distributional models were used to compare current niches to mid‐Holocene and Pleistocene distributions. Morphological data were used to assess phenotypic differences between the species and test for evidence of spatial sorting. Results Both species underwent rapid range expansions since the Pleistocene and share similar population structure. Species distributional models are different in distribution at all times periods between species. The species are significantly different in all morphological measurements taken. Only P. vehiculum shows evidence of spatial sorting. Main conclusions Physiological tolerance and dispersal ability best explain the distributional difference between these species. Spatial sorting most likely plays a key role in the range expansion of P. vehiculum.

The impact of climate change on westernPlethodonsalamanders’ distribution

“…They are largely sympatric and diverged from one another approximately 10 mya (Kozak et al, 2009). These two species had similar southern ancestral distributions during the LGM, and there is evidence of extensive range expansion in both species, yet P. vehiculum was able to expand much farther north than P. dunni in the last 20,000 years, even though the effective population size of P. dunni grew faster (Pelletier & Carstens, 2016). Overall, P. vehiculum has the largest range with very little genetic structure, even onto Vancouver Island (Pelletier et al, 2011), and has a higher tolerance to extremes in temperature and humidity (Dumas, 1956), indicating this species is a good disperser.…”

A global analysis of bats using automated comparative phylogeography uncovers a surprising impact of Pleistocene glaciation

“…However, they are difficult to apply to this question due to differences among species and geographical regions. For example, despite the tenuous correlation between species dispersal abilities and geographical range size (Lester, Ruttenberg, Gaines, & Kinlan, 2007), long-distance dispersal appears to be a key component of population expansion following glaciation (Hewitt, 2004), which suggests that organismal traits related to dispersal at least partially influence the capacity of a particular species to respond to newly-opened habitat (e.g., Morales, Villalobos, Velazco, Simmons, & Piñero, 2016;Pelletier & Carstens, 2016). Similarly, the response of a given species to broad-scale climate change is influenced by both the organismal life history and the particular habitat that it occupies.…”

Unraveling climate influences on the distribution of the parapatric newts Lissotriton vulgaris meridionalis and L. italicus

“…Particular attention was given to a sympatric zone, because of the great importance accorded to these areas in many fields of theoretical and applied ecology. In fact, understanding the patterns of species’ overlapping occurrences represents a challenging topic of ecological research, ranging from the reconstruction of the species’ evolutionary history [ 28 ] to ecological modelling [ 29 , 30 ], especially when applied to the study of the relationships between past biogeographical patterns and climate in closely related taxa [ 9 , 12 , 31 ]. We built SDMs for both the target species under the current conditions using two machine-learning techniques, Boosted Regression Trees (BRT) and Maxent.…”