High-throughput sequencing methods have facilitated obtaining large amounts of data from degraded DNA, thus resulting in a dramatic increase in destructive sampling requests to museums. Because the tissues taken from museum specimens as sources of DNA are destroyed during analysis, consideration of the costs and benefits of loss of valuable specimen material relative to knowledge gained is required for any project utilizing destructive sampling. Variation exists in the preservation of DNA in historical specimens due to specimen age and type of museum preparation, among other factors. Thus, it is important to assess DNA yield and quality from different sources of museum specimens when considering the needs of a particular molecular project. We compared DNA derived from several common sources of museum specimens including bone, claw, skin, and soft tissue adherent to skeletal preparations. To account for differences in preparation type and therefore specimen preservation, we tested the performance of samples representing 3 taxonomic groups: mephitids, rodents, and marsupials. We also compared yields from 2 commonly used DNA extraction techniques. DNA quality was assessed by comparing average fragment size, concentration, and copy number of template DNA (for mitochondrial and nuclear markers) in genomic DNA extracts, as well as mitochondrial genome sequence coverage resulting from shotgun sequencing. We show that DNA quality derived from historic museum samples differs depending on specimen and sample type; however, all samples yielded high mitochondrial copy number except the skin and nail from the tanned specimen. Overall, claw samples produced the greatest number of high-quality sequencing reads with the least amount of bacterial contamination. We also found that high DNA concentrations did not necessarily result in high percentages of on-target reads; in fact, the samples that yielded the highest DNA quantities also had the highest amount of exogenous bacterial DNA. Our results indicate that most historical tissue types can be suitable for next-generation sequencing approaches, therefore providing multiple options for natural history collection staff and researchers when considering destructive sampling requests.
