Fig 1.
A flow chart providing an overview of our methodological approach.
Fig 2.
Habitus of Lagriinae n. gen. KK0290.
Image taken after DNA extraction. Scale bar is 1 mm.
Fig 3.
Habitus of HTS carabid specimens.
Images taken after DNA extraction. Scale bar is 1 mm.
Table 1.
Specimens sequenced using Illumina methods, with details about specimen histories.
Table 2.
Museum specimens that were assessed with a Qubit and Bioanalyzer but not Illumina sequenced, with details about specimen histories.
Table 3.
Museum specimens assessed with a Qubit but not with a Bioanalyzer or Illumina sequenced, with details about specimen histories.
Table 4.
Quality and quantity of DNA for specimens sequenced using Illumina methods.
Table 5.
Quality and quantity of DNA for specimens assessed with a Qubit and with a Bioanalyzer but not Illumina sequenced.
Table 6.
Quantity of DNA for specimens that assessed with a Qubit but not with a Bioanalyzer or Illumina sequenced.
Fig 4.
Electropherograms of DNA extracted from older museum specimens that were subsequently used in library preparation.
Pale spikes at 35 and 10380 bases represent standards included in each analysis. Dark shaded regions, when present, correspond to range of fragments that were selected and sequenced on the HiSeq 2000.
Fig 5.
Electropherograms of DNA extracted from younger museum specimens that were subsequently used in library preparation.
Pale spikes at 35 and 10380 bases represent standards included in each analysis. Dark shaded regions, when present, correspond to range of fragments that were selected and sequenced on the Illumina HiSeq 2000. Regions are not shown for Bembidion musae or Bembidion “Inuvik” 3984 as the DNA in those samples was sonicated prior to library preparation. For each specimen, age and total DNA in the extraction is also shown.
Table 7.
Library preparation details.
Table 8.
Predictions about phylogenetic placement of museum specimens.
Table 9.
Summary of success of PCR of four gene fragments.
Table 10.
De novo assembly statistics.
Table 11.
Results from CEGMA analyses between contigs from de novo assemblies and 248 core Eukaryotic genes (CEGs).
Fig 6.
Recovery success of 67 low-copy nuclear protein-coding gene fragments in HTS museum specimens.
Darkness of cell corresponds to percentage of the length of that fragment that was recovered, with black cells corresponding to 100% recovery. Gene fragments are ordered by average recovery as measured across both de novo and reference-based assemblies. Gene abbreviations are those used in Regier et al. [25]. Specimen abbreviations: Lag: Lagriinae n. gen. KK0290, subf: Bembidion subfusum 3977, snt1: B. sp. nr. transversale 3021, Lchi: Lionepha chintimini 4002, lach: B. lachnophoroides 3022, Bdrs: Bembidarenas 3983, ori1: B. orion 2831, inu1: B. "Inuvik" 3285, lapp: B. lapponicum 3974, aric: B. "Arica" 3242, dspt: B. cf. "Desert Spotted" 3978, mus: B. musae 3239, inu2: B. "Inuvik" 3984, ori2: B. orion 3079, snt2: B. sp. nr. transversale 3205. Four specimens with less than 34 million reads have specimen abbreviation and age shown in gray. Numbers under the specimen abbreviations are years between death and extraction.
Table 12.
Comparison of 67-gene set recovery between de novo assemblies and reference-based assemblies.
Fig 7.
Recovery success of seven focal genes, with comparison of de novo and reference-based assemblies.
For protein-coding genes, values in cells are the fractional recovery of the query sequence (for de novo assemblies) or reference sequence (for reference-based assemblies). Cells are shaded in a gray-scale ramp with black recovery of 100% of the fragment length and white 0%. For ribosomal genes, values in cells are the fractional recovery of the query sequence (for de novo assemblies), and for reference-based assemblies, values in cells represent the percentage recovery of the assembly relative to the de novo assembly (as opposed to the query or reference sequence). Values less than 1.0 indicate that some bases were missing from the reference-based assembly. A comparison of the de novo assembly sequence to the reference sequence shows that those missing regions are very different between the museum sample and the reference, and thus that region of the reference-based assembly failed. If there are no base differences between the reference-based and the de novo assemblies, the cell is colored using a blue ramp, with pure blue indicating 100% recovery. If there are base differences, the cell is colored red, with the number of base differences shown in parentheses. An asterisk (*) indicates that the sequence so marked is not in the predicted place in the maximum likelihood tree including the DeNovo, FarRef, and NearRef sequences; two asterisks (**) indicates that this placement failure is supported by a bootstrap value of over 50%. “-”indicates that no attempt was made to find this fragment in the assemblies. Four specimens with less than 34 million reads have specimen abbreviation and age shown in gray.
Fig 8.
A maximum likelihood tree for the six-gene concatenated dataset of Lagriinae.
The museum specimen is marked with a star symbol. The branches and taxon names of Lagriinae n. gen. and its predicted closest relatives (based on morphological characters) are colored in blue.
Fig 9.
A portion of the maximum likelihood tree of carabids for Topo with all de novo assembly contigs included.
An example in which our BLAST searches for target genes within HTS museum specimen assemblies returned multiple contigs, all of which were nearly identical and within the prediction group. The prediction group is shown in blue. The contig chosen by our filtering criteria is marked by an asterisk.
Fig 10.
A maximum likelihood tree of carabids for 28S with all contigs included from the de novo assembly.
An example in which our BLAST searches for target genes within HTS museum specimen assemblies returned multiple contigs, many of which were placed on long branches at unexpected positions across the tree. The behavior of the multiple contigs is highlighted using two museum specimens, Bembidion lapponicum 3974 and Bembidion lachnophoroides 3022. In both cases, our filtering criteria appear to select the best of the multiple contigs. Red star: chosen contig for Bembidion lapponicum 3974. Red arrows: other contigs from B. lapponicum 3974. Blue star: chosen contig for Bembidion lachnophoroides 3022; blue arrows: other contigs from B. lachnophoroides 3022.
Fig 11.
A portion of the maximum likelihood tree of carabids for CAD with all contigs included from the de novo assembly.
An example in which our BLAST searches for target genes within HTS museum specimen assemblies returned multiple contigs, however our filtering criteria failed to accept a best contig, despite two contigs falling in the prediction group (shown in pink), and being nearly identical to the PCR-based sequence of a conspecific specimen (Bembidion “Inuvik” 3984).
Fig 12.
A portion of the maximum likelihood tree of carabids for ArgK with all contigs included from the de novo assembly.
An example in which our BLAST searches for target genes within HTS museum specimen assemblies returned multiple contigs, however our criteria for choosing the best among multiple contigs selected a contig which was not inferred to be where predicted in the phylogeny.
Fig 13.
A maximum likelihood tree of carabids from seven focal genes and “Three Separate” assembly sequences.
The placement of the DeNovo, NearRef, and FarRef sequences is shown relative to their prediction groups in a concatenated analysis of seven focal genes. Each prediction group is marked by a black arrow, and a unique color for branches and taxon names of all specimens in the prediction group. The placement of the three assembly sequences is indicated with a black star.
Table 13.
Support for or against de novo and reference-based sequences of each museum specimen forming a clade.
Fig 14.
A maximum likelihood tree of carabids from seven focal genes and IlluminaMerged sequences.
The placement of the IlluminaMerged sequences is shown relative to their prediction groups in a concatenated analysis of seven focal genes. Each prediction group is marked by a black arrow, and with a unique color for branches and taxon names of all specimens in the prediction group. The placement of each IlluminaMerged sequences is indicated with a black star.
Table 14.
Support for IlluminaMerged sequence being in predicted location in phylogeny.
Table 15.
Comparison of the IlluminaMerged sequences of museum specimens to sequences from conspecific specimens.
Fig 15.
Four maximum likelihood gene trees for the Bembidion transversale group.
The placement of DeNovo, NearRef, FarRef, and IlluminaMerged sequences of museum specimen Bembidion sp. nr. transversale 3021 is shown in a matrix of conspecific specimens and close relatives in the transversale species group. The museum specimen assembly sequences are shown in red with the IlluminaMerged sequence marked with a black star. All other specimens are colored by species.
Fig 16.
Squared correlation coefficients from univariate linear regression analyses between success measures and potential explanatory variables.
Measures of success of acquiring protein-coding gene fragments are NPDN50 (de novo assembly, percent of gene fragments for which at least 50% of the bases were recovered), NPDN80 (same, but at least 80% of the bases), NPRef50 (reference-based assembly, percent of gene fragments for which at least 50% of the bases were recovered), and NPRef80 (same, but at least 80% of the bases). On the left are analysis with all samples included; on the right are analyses with only samples with more than 60 million reads included. Symbols outlined in red indicate that the correlation is significant in a single-variable analysis; symbols outlined in pale pink indicate that the correlation is significant as a secondary variable in a bivariate analysis. Note x-axis orientations are mirrored in the two graphs.