Decades-old formalin museum insects surrender their mitochondrial DNA

Natural history museums have long been treasure houses for evolutionary research, but one of their most common preservation methods has also been one of the most destructive to genetic material. Formalin, the formaldehyde-based fixative that has kept countless zoological specimens intact for decades, fragments DNA, chemically modifies its bases and creates cross-links that make extraction extraordinarily difficult. A new pilot study published in the journal Natural History Collections and Museomics now shows that even arthropods preserved in formalin for up to four decades can still yield mitochondrial DNA of sufficient quality for high-throughput sequencing and phylogenetic analysis, opening a door that many researchers had assumed was permanently closed.

The research team, working with material from the Zoological Museum Kiel, focused on two taxonomically distant groups of commonly collected arthropods: ground beetles of the family Carabidae and freshwater crabs of the family Potamidae. The beetle specimens had been caught in pitfall traps containing 4% formalin by the former Ecology Centre of Kiel University in the 1960s and 1980s, meaning they had been preserved in formalin for either 35 to 41 years or as long as 61 years. The crab samples, belonging to various species of the family Potamidae, had been preserved for approximately 12 to 25 years, with a documented fixation in 10% formalin for two individuals and a standard 10% fixation assumed for the rest. Formalin was still detectable in the storage media of all samples used, measured with QUANTOFIX test strips, allowing the researchers to assume continuous exposure since collection.

Formalin-preserved museum samples differ from the better-studied formalin-fixed paraffin-embedded material in important ways. FFPE tissue is typically fixed for minutes and then thoroughly washed, whereas museum specimens often sit in variable formalin concentrations for decades without subsequent washing. The damage patterns, however, resemble those found in ancient DNA, including extensive fragmentation and cytosine deamination. This similarity allowed the team to adapt methodological approaches originally developed for aDNA research, a strategy that proved central to their success.

Because most published protocols for formalin-fixed material focus on vertebrate tissue, the researchers faced a knowledge gap regarding invertebrates, despite their numerical dominance in museum collections. Invertebrates differ from vertebrates in tissue structure, body size and the high proportion of insoluble components such as chitin, making known protocols difficult to transfer. Many aquatic species and pitfall-trapped insects are predominantly preserved in formalin rather than as dried mounted specimens, creating a preservation bias that has limited genetic studies of these groups.

To identify a robust method, the team tested multiple DNA extraction protocols, all following the same basic procedure of sample preparation, digestion, purification and downstream library preparation. Sample preparation involved rinsing specimens with tap water for about two hours, followed by washing in GTE buffer, a solution of 100 mM glycine, 10 mM Tris-HCl, 1 mM EDTA, at room temperature for two hours, repeated once and then overnight. Samples were then rinsed in 99% ethanol, 70% ethanol and sterile water. For digestion, the researchers tested proteinase K digestion with various buffer compositions, hot alkaline lysis at pH 13 with 0.1 M sodium hydroxide and 1% SDS, and a commercial Roche kit designed for FFPE tissue. Purification methods included spin columns from QIAGEN and Roche, and phenol:chloroform extraction under a clean fume hood.

The winning combination proved to be proteinase K digestion followed by QIAquick spin column purification. For beetles, whole specimens of Dicheirotrichus gustavii or single longitudinal halves of Pterostichus melanarius were used, with optional homogenisation using a micro-pestle to break down chitinous structures. For freshwater crabs, approximately 1 to 2 cubic millimetres of leg muscle tissue was dissected, and only tissue that had completely lysed during digestion was used, avoiding the need for homogenisation. Digestion was carried out overnight at 56 degrees Celsius with proteinase K and DTT, a reducing agent that helps reverse formalin-induced cross-links. The researchers preferred the commercially available ATL buffer from QIAGEN over in-house prepared buffers due to its lower contamination risk, even though both yielded comparable DNA concentrations.

Library preparation used a partial uracil DNA glycosylase treatment, which removes partly de-aminated cytosines while retaining characteristic terminal damage patterns. This approach allows library preparation from highly degraded DNA while preserving authentic aDNA-like signatures that can be used to confirm the origin of the sequences. Due to extremely low DNA concentrations in the extracts, an additional non-specific PCR amplification step was necessary for most samples to raise concentrations sufficiently for Illumina sequencing. For example, one beetle sample was increased from 0.143 nanograms per microlitre to 13.1 nanograms per microlitre after unspecific amplification.

Of the 37 samples evaluated with the final protocol, comprising 32 ground beetles and five freshwater crabs, 26 produced DNA concentrations of at least 0.1 nanograms per microlitre at a fragment length of approximately 180 base pairs after library preparation, sufficient for successful Illumina sequencing. All freshwater crab samples and 20 ground beetle samples met this threshold. The highest library concentrations among the crab samples were found in two individuals of the genus Himalayapotamon, with values of 0.45 and 0.321 nanograms per microlitre. Among the beetles, the highest concentration came from a specimen of D. gustavii at 3.33 nanograms per microlitre. Crab samples generally yielded higher DNA concentrations after library preparation than beetle samples, likely due to differences in tissue structure and preservation history, including the shorter formalin exposure and higher fixation concentration of the crabs.

Sequencing generated approximately 11.3 to 56.2 million raw reads per specimen on an Illumina HiSeq4000. Bioinformatic analysis used the EAGER pipeline, designed for degraded aDNA, with competitive mapping that assigns each read exclusively to the best-matching reference genome. Because species-specific mitochondrial genomes were unavailable for the sampled taxa, reads were mapped to reference genomes of closely related species. Mitochondrial genome coverage ranged from approximately 3X to 17X, reflecting low endogenous mitochondrial DNA content. The proportion of reads mapping to target mitochondrial reference genomes ranged from 0.0067 to 0.0106% in crab specimens and 0.0042 to 0.0375% in beetle samples. Most sequencing reads mapped to the respective target reference genomes, with remaining reads predominantly aligning to closely related species or the human reference sequence, a pattern expected given that modern human DNA is often better preserved than highly degraded formalin-stored material.

Despite the relatively low coverage, the recovered mitochondrial data proved sufficient for downstream phylogenetic analyses. Maximum Likelihood phylogenetic analysis with 100 bootstrap iterations placed the beetle samples within their expected clades, with P. melanarius grouping within the Pterostichus clade and D. gustavii clustering with closely related Bradycellus ruficollis. The two Himalayapotamon specimens also grouped as expected, forming clades with H. robertsianum and H. emphysetum. This accurate phylogenetic placement demonstrates that the recovered sequences are authentic and originated from the target organisms rather than from contamination, a conclusion further supported by negative controls that yielded only very low DNA concentrations and were not sequenced.

The study demonstrates that high-throughput sequencing is feasible from invertebrate museum specimens preserved in formalin for decades and provides a practical protocol adaptable to a broad taxonomic range. The researchers emphasise that their samples covered a wide range of conditions, including different preservation periods, different formalin concentrations and phylogenetically distant taxa, reflecting the heterogeneity of historical museum collections rather than systematically controlled variables. They acknowledge limitations, including the small and uneven number of samples available for different methodological approaches and the incomplete documentation of preservation history, which limits the ability to distinguish the effects of specific conservation conditions on DNA extraction. Nevertheless, the results provide evidence for the potential applicability of the protocol across taxonomically diverse formalin-preserved arthropod specimens, and the researchers have made their alternative tested protocols available for other laboratories whose conditions may favour different approaches. Further testing with larger numbers of biological replicates and systematically selected samples will be needed to confirm the robustness of the method, but the door to decades of formalin-preserved genetic diversity in museum collections now stands demonstrably ajar.

Subject of Research: Recovering mtDNA from long-term formalin-preserved museum arthropods: A pilot study

Article Title: Recovering mtDNA from long-term formalin-preserved museum arthropods: A pilot study

Article References: Mack, A., Grützke, S., Krause-Kyora, B., Brandis, D., Kuhlmann, M., & Ewers, C. (2026). Recovering mtDNA from long-term formalin-preserved museum arthropods: A pilot study. Natural History Collections and Museomics, 3, Article e200681. https://doi.org/10.3897/nhcm.3.200681

Image Credits: AI Generated

DOI: 10.3897/nhcm.3.200681

Keywords: museomics, mitochondrial DNA, formalin, natural history collections, ground beetles, freshwater crabs, ancient DNA methods, phylogenetics, Recovering, mtDNA, long-term, formalin-preserved