Library preparation protocols for high-throughput DNA sequencing (HTS) include amplification steps in which errors can build up. In order to have confidence in the sequencing data, it is important to understand the effects of different Taq polymerases and PCR amplification protocols on the DNA molecules sequenced. We compared thirteen enzymes in three different marker systems: simple, single copy nuclear gene and complex multi-gene family. We also tested a modified PCR protocol, which has been suggested to reduce errors associated with amplification steps. We find that enzyme choice has a large impact on the proportion of correct sequences recovered. The most complex marker systems yielded fewer correct reads and the proportion of correct reads was greatly affected by the enzyme used. Modified cycling conditions did reduce the number of incorrect sequences obtained in some cases, but enzyme had a much greater impact on the number of correct reads. Thus, the coverage required for the safe identification of genotypes using one of the low quality enzymes could be seven times larger than with more efficient enzymes in a biallelic system with equal amplification of the two alleles. Consequently, enzyme selection for downstream HTS has important consequences, especially in complex genetic systems.

Reduced representation genome sequencing has popularized the application of single nucleotide polymorphisms (SNPs) to address evolutionary and conservation questions in nonmodel organisms. Patterns of genetic structure and diversity based on SNPs often diverge from those obtained with microsatellites to different degrees, but few studies have explicitly compared their performance under similar sampling regimes in a shared analytical framework. We compared range‐wide patterns of genetic structure and diversity in two amphibians endemic to the Iberian Peninsula: Hyla molleri and Pelobates cultripes, based on microsatellite (18 and 14 loci) and SNP (15,412 and 33,140 loci) datasets of comparable sample size and spatial extent. Model‐based clustering analyses with STRUCTURE revealed minor differences in genetic structure between marker types, but inconsistent values of the optimal number of populations (K) inferred. SNPs yielded more repeatable and less admixed ancestries with increasing K compared to microsatellites. Genetic diversity was weakly correlated between marker types, with SNPs providing a better representation of southern refugia and of gradients of genetic diversity congruent with the demographic history of both species. Our results suggest that the larger number of loci in a SNP dataset can provide more reliable inferences of patterns of genetic structure and diversity than a typical microsatellite dataset, at least at the spatial and temporal scales investigated.

The hyperdiverse shrew genus Crocidura is one of few small mammal genera distributed across Sundaland and all of its boundaries. This represents a rare opportunity to study the geological history of this region through the evolutionary history of these shrews. We generate a phylogeny of all recognized species of Sundaland Crocidura and show that most speciation events took place during the Pleistocene, prior to the inundation of the Sunda Shelf around 400 000 years ago. We find east–west differentiation within two separate lineages on Borneo, and that the current taxonomy of its two endemic species does not reflect evolutionary history, but ecophenotypic variation of plastic traits related to elevation. Sulawesi shrews are monophyletic, with a single notable exception: the black-footed shrew (C. nigripes). We show that the black-footed shrew diverged from its relatives on Borneo recently, suggesting a human-assisted breach of Wallace’s line. Overall, the number of Crocidura species, especially on Borneo, probably remains an underestimate.