Replicate Phylogenies and Post-Glacial Range Expansion of the Pitcher-Plant Mosquito, Wyeomyia smithii, in North America

Clayton Merz,Victor Hanson-Smith,Christina M Holzapfel,William E Bradshaw,Kevin J Emerson,Julian M Catchen,William J Etges

doi:10.1371/journal.pone.0072262

Abstract

Herein we tested the repeatability of phylogenetic inference based on high throughput sequencing by increased taxon sampling using our previously published techniques in the pitcher-plant mosquito, Wyeomyia smithii in North America. We sampled 25 natural populations drawn from different localities nearby 21 previous collection localities and used these new data to construct a second, independent phylogeny, expressly to test the reproducibility of phylogenetic patterns. Comparison of trees between the two data sets based on both maximum parsimony and maximum likelihood with Bayesian posterior probabilities showed close correspondence in the grouping of the most southern populations into clear clades. However, discrepancies emerged, particularly in the middle of W. smithii's current range near the previous maximum extent of the Laurentide Ice Sheet, especially concerning the most recent common ancestor to mountain and northern populations. Combining all 46 populations from both studies into a single maximum parsimony tree and taking into account the post-glacial historical biogeography of associated flora provided an improved picture of W. smithii's range expansion in North America. In a more general sense, we propose that extensive taxon sampling, especially in areas of known geological disruption is key to a comprehensive approach to phylogenetics that leads to biologically meaningful phylogenetic inference.

Highlights

The term ‘‘resampling’’ in phylogenetics almost exclusively refers to bootstrapping or other methods of subsampling a single data set
We evaluated the validity of phylogenetic inference based on the 46-population maximum parsimony tree by comparing its topology with geological and historical biogeography of eastern North America
Populations The range of sampled North American populations extended from the Gulf Coast (30–31uN) to Newfoundland (50uN) and northwestward to Saskatchewan (54uN), from 30–4,000 ft elevation at 35uN in North Carolina, and from 30–2,000 ft elevation at 40–41uN in New Jersey and the Pocono Mountains of Pennsylvania

Summary

Introduction

The term ‘‘resampling’’ in phylogenetics almost exclusively refers to bootstrapping or other methods of subsampling a single data set. We present a different kind of resampling: a twostage phylogenetic analysis based on separate collections from near-by but geographically separate natural populations of the pitcher-plant mosquito, Wyeomyia smithii. We expected that using the same published methods, protocols and phylogenetic analyses would generate separate sets of phylogenetic trees whose major branches would align with each other. After comparing the two data sets, we merged the data into a combined set of 46 populations, which leads to new conclusions on the origin of northern and mountain populations of W. smithii. We evaluated the validity of phylogenetic inference based on the 46-population maximum parsimony tree by comparing its topology with geological and historical biogeography of eastern North America

Methods

Results

Conclusion