Trace DNA recovery rates from firearms and ammunition as revealed by casework data

ABSTRACT Understanding casework DNA recovery rates per evidence type is crucial for advancing forensic methodologies and practices. This study assesses DNA recovery and profile data for 17 firearm parts (including cartridges, bullets and cases: CBCs) using New South Wales firearms casework data from 2015 to 2019. CBCs were further classified as shotgun or metallic cartridges and as unfired or fired. Quantitative data analysis showed that the least DNA was recovered from the hammer, safety and CBCs, while a single swab of multiple firearm parts resulted in the highest DNA recovery. Shotgun and unfired cartridges yielded more DNA than metallic and fired cartridge cases, respectively. Additionally, DNA collected from exhibits after fingerprinting yielded more DNA than exhibits sampled for DNA without any fingerprint examination. Profile data analysis showed that at least 64% of samples were unusable in casework. Cumulatively swabbed areas were most likely to produce usable profiles. Up to 22% and 38% of usable and unusable profiles, respectively, were mixed. Internal parts of the firearms had lower percentages of mixed profiles. The results from the data suggest the areas to prioritize for DNA recovery from firearms and highlights the need for further research to improve sample collection from firearms and ammunition.


Introduction
The ability to recover trace DNA from firearms and ammunition (including cartridges, bullets, and cases: CBC's) is useful for making sub-source level interpretations of a crime event involving a firearm and/or linking individuals to the item in question.
However, DNA recovery from firearms has not been as widely examined, apart from a few studies exploring swabbing 1,11,12 and tape lifting 13,14 . The routinely used swabbing and tape lifting methods do not often yield usable profiles from firearm related evidence in casework, even though tape lifting in some situations may be more successful at recovering DNA [11][12][13] . Therefore, identifying casework DNA recovery rates is imperative to find areas where advancements are required.
In this study, historical casework data relating to firearms and ammunition in New South Wales (NSW), Australia, were assessed to better understand the DNA recovery rates given the current DNA subsampling technique employed in NSW. DNA case work data relating to firearms and ammunitions between 2015 and 2019 were analysed to: • determine the quantities of DNA recovered from various firearm and ammunition components, • identify the proportions of usable, 'intelligence only' and not usable DNA profiles obtained to ascertain value, and • identify the proportions of mixed profiles.
Understanding which components of firearms and ammunition give informative results can help triage the types of evidence processed (including the areas of items to prioritize for sampling), as well as allow comparisons with data across jurisdictions/research studies to help identify methodological and/or procedural improvement opportunities and determine directions of future research.
Previously published studies 11,15 have associated a degree of 'success' to their DNA recovery rates, as defined by the authors. Both Mapes et al 11 and Krosch 12 further discuss two metrics of 'success', the success of obtaining a full DNA profile and the success of obtaining suspect identification. These two metrics are not mutually exclusive.
Success from a law enforcement perspective could be defined as detecting a profile of high probative value. For example, detecting the victim's DNA profile from a bloodstain in the car of a suspect, when no previous interaction between victim and suspect is known. However, detecting the suspect profile in the car of the suspect, although meeting the criteria of success as discussed by Mapes et al 11 and Krosch 12 , may not hold as high a value for the investigating officer. Likewise, exculpatory evidence can be equally informative to an investigation. Therefore, the definition of 'success' is difficult to determine and is variable from a literature and a law enforcement perspective. In all cases however, the requirement to obtain a profile that can be used to identify an individual is imperative. As a result, we have simply used DNA recovery as a metric for comparison.

Casework data
Firearms casework data between 2017 and 2018 and ammunitions data between 2015 and 2019 were obtained from the NSW Police Force's enterprise data warehouse (EDW). A longer time frame was searched for ammunitions data to provide sufficient sample numbers for meaningful data analysis. The casework data included information such as types of offences, time of exhibit collection, exhibit descriptions (free text fields), DNA quantitation results and STR profile results but did not include whether or not a profile matched a suspect, a victim or other person of interest (POI). To ensure confidentiality, all private or identifiable information was deidentified/excluded by the NSW Police Force Science and Research Unit (SRU) prior to analysis.
Power Query (Microsoft® Excel®) was used to clean, combine and deidentify the data. STR profiles were re-classified into the following categories: 'usable', 'DNA intel', 'unusable', and 'not tested'. These categories are used exclusively within the SRU for the means of data interrogation and are defined below: • 'Usable' -Refers to a DNA sample result containing one or more of the following: ○ Full single source profile ○ Mixed DNA profile which could be interpreted to determine a contributor/s ○ Identifiable partial profile ○ Sample result suitable for upload to the NSW DNA database ○ Sample result suitable for identification of an individual(s) without a reference profile • 'DNA Intel' -Refers to a DNA sample result that did not meet the 'Useable' criteria, and contains one or more of the following: ○ Mixed or single source sample result not suitable for upload to DNA database, but may be suitable for comparison with a reference profile (nominated individual) ○ Sample result where a known individual cannot be excluded as a contributor ○ Sample result not suitable for upload to DNA database, but a known individual can be excluded as a contributor • 'Unusable' -Refers to a DNA sample result that did not meet the 'Usable' or 'DNA Intel' criteria and contains one or more of the following: ○ Profile is too weak (minimal DNA recovered and sample not suitable for comparison with a reference sample) ○ No DNA was recovered from the sample ○ Only one or two alleles recovered from the sample and are not suitable for comparison ○ Amount of DNA recovered from the sample is below the analytical threshold for further testing ○ Sample result is complex. Sample may originate from multiple contributors; the number of contributors may not be clear, and an individual profile cannot be determined with confidence. • 'Not Tested' -Refers to one or more of the following: ○ Sample result is complex, and two biologists have determined that further analysis will not provide any new information in this case ○ Circumstance has led to the sample being withdrawn from testing, e.g. an individual has pled guilty to a crime before the sample was tested, so no further DNA testing is required.
Exhibits with usable profiles obtained from a mixture and unusable profiles resulting from a complex mixture, were further classified as 'mixed'. All 'DNA Intel' profiles in this study, except for a single profile from a trigger, were mixed profiles.
To compare trace DNA recovery rates, the firearms data were further categorized into individual parts of the firearm (Supplementary Table 1 and Supplementary Figure 1) that were swabbed including CBCs (as one separate category). Categories with less than 15 samples were excluded to reduce sampling effects. Where free text information was available, CBCs were further classified as unfired or fired cartridges, and shotgun or 'other' metallic cartridges.
Data filtering/categorizations were based on free text fields due to limitations with the available IT and EDW systems. This may introduce some errors into the analysis (i.e. not all samples submitted in each category may have been captured due to typographical errors in data entry, and some samples may have been miscategorized for the same reason). To reduce these errors, the data were manually reviewed to ensure accuracy.
In NSW, most firearm and ammunition exhibits are triaged for fingerprint examination prior to sampling for DNA. A subset of data were analysed to compare swabs taken from exhibits where fingerprint examination had been performed, to swabs taken from exhibits where no fingerprint examination was performed. These data were interrogated to identify any effect that fingerprint examination may have on DNA recovery rates.
Finally, for this study, only trace DNA swabs (DNA material not related to a discrete biological stain, e.g. blood, semen, or saliva) were included for data analysis. Any swabs identified to be from discrete biological stains were excluded from the data set. The swabs were processed at the Forensic & Analytical Science Service (FASS), NSW Health Pathology.

Swabbing method
Ethylene oxide treated rayon swabs (Medical Wire & Equipment, UK) were used according to the NSW Police Force standard operating procedure for trace DNA. Wearing a mask and double gloves, the swab kits were opened. After opening the kit, the outer gloves were exchanged for a fresh pair, and the swab seals broken to release the rayon swab stick. For trace DNA sampling, one drop of sterile water (10 mL sterile water for injection, Pfizer, AUS) was placed onto one side of the swab head. The exhibit was firmly swabbed using the moistened side of the swab followed by the dryer side. Swab heads were broken off into AutoLys tubes (Hamilton®, USA) for automated DNA profiling.

DNA lysis and extraction
DNA lysis was achieved on the AutoLys Microlab STAR (Hamilton®, USA) workstation using the PrepFiler™ Automated DNA Extraction Lysis buffer (Applied Biosystems, USA). DNA was extracted on the Tecan Freedom EVO 150 workstation (Tecan Life Sciences, Switzerland) using the PrepFiler™ Automated DNA Extraction Kit (Applied Biosystems, USA), according to the manufacturer's standard protocol. The final elution volume of the DNA was 50 µL.

Quantification
Reaction plates for quantification were set up on the Tecan Freedom EVO 150 workstation. DNA was quantified on the 7500 Real-Time PCR instrument (Applied Biosystems, USA) using Quantifiler™ Trio DNA Quantification Kit (or Quantifiler™ Human DNA Quantification Kit before April 2017) (Applied Biosystems, USA) according to the manufacturer's standard protocol. Samples containing DNA below the FASS amplification threshold of 0.004 ng/µL did not proceed further, as previous validation has indicated they are unlikely to produce an STR profile.

STR amplification
Reaction plates for amplification were set up on the Tecan Freedom EVO 150 workstation. STRs were amplified on the 9700 Thermal Cycler instrument (Applied Biosystems, USA) using amplification components of the PowerPlex® 21 System (Promega, USA) according to the manufacturer's standard protocol, with a DNA template amount of 0.7 ng and with 29 PCR cycles. For samples with a DNA concentration equal to or less than 0.047 ng/µL (0.7 ng in 15 μL), the full volume (15 µL) of DNA extract was added to the reaction.

Capillary electrophoresis and analysis of DNA profiles
Reaction plates for capillary electrophoresis were set up on the Tecan Freedom EVO 150. Amplified fragments were separated on the 3500xL Genetic Analyser (Applied Biosystems, USA) using capillary electrophoresis components of the PowerPlex® 21 System (Promega, USA) according to the manufacturer's standard protocol. Profiles were generated and analysed using the GeneMapper® ID-X Software (ThermoFisher Scientific, USA).

Statistical analysis of data
General data analysis was performed in Microsoft Excel. Prior to statistical analysis, highly skewed quantitation data were log transformed after adding the lowest recorded concentration of 0.0001 ng/µL to all concentrations. This was to avoid log transforming the many zero concentrations in the data. Statistical analysis was performed via SPSS Statistics (IBM, Australia) using a Kruskal-Wallis or Mann-Whitney U test for non-parametric data. Pairwise comparisons following the Kruskal-Wallis tests were conducted using a Dunn-Bonferroni post hoc test.

Quantitative data analysis
A total of 2,674 swabs relating to 1,587 individual exhibits and 731 individual criminal events were analysed. These exhibits came from various case types ( Figure 1) ranging from volume and major crimes (approximately 19% and 18%, respectively) to located property -either physically located or through search warrant (approximately 10%). A large percentage (approximately 47%) of the exhibits analysed were obtained through firearms legislation which includes seizure/surrender of firearms, illegal possession of firearms, illegal modification of firearms, or unlawful discharge of a firearm. The remaining 6% of exhibits did not have an incident type reported. The number of swabs taken per exhibit ranged from one to seven, with a median of one. There were no trends regarding the prioritization of target areas except when two or more swabs were taken from an exhibit, the two areas most frequently targeted for sampling on firearms were the grip and trigger.
For analysis, the dataset was split into 16 different categories ( Figure 2) to help distinguish the various areas of sampling on a firearm (Supplementary Figure 1 and  Supplementary Table 1). Non-firearm parts (CBCs) were combined into a separate category. Figure 2 shows the quantities of DNA recovered from firearm parts and CBCs (see Supplementary Table 1 for tabulated results). The quantities of DNA recovered were found to be highly variable, as observed from the presence of outliers from the barrel, CBCs, cylinder, grip and lever, and the range covered by the boxplots. High variability from trace DNA swabs is typically expected due to differences in DNA shedding among individuals 16,17 , how the areas were contacted 18-20 and the relative size or texture of the areas sampled. In casework situations, other factors that may also influence DNA recovery include whether the handler was wearing gloves or the environmental conditions under which the firearm(s) and ammunition were recovered.
The greatest quantity of DNA recovered was 120 ng from a fired cartridge case. A posthoc Dunn-Bonferroni test showed no significant differences between CBCs and the hammer, bolt handle, safety, and cylinder (p > 0.05). However, CBCs yielded significantly less DNA than all other firearm parts (p < 0.05) (Supplementary Table 2). Both unfired and fired ammunition are known to be items from which it is difficult to recover trace DNA. Metals, more specifically brass, have been shown to interact with the negatively charged phosphate backbone or high electron density nucleobases of DNA thereby changing its structural conformation 21 . Brass cartridges/cases account for approximately two thirds of the ammunition casework samples in NSW with the remaining being nickel or other compositions (personal communication, Dr Jennifer Raymond, NSW Police Force). As a result, it was expected that CBCs would yield less DNA compared to other firearm parts which are made of a combination of metals and plastic, require more handling to fire the weapon, and/or have larger surface areas, and/or have texture to trap DNA material. Firearms may also be handled recurrently, whereas ammunition is generally only handled in respect to one loading or shooting event.
The hammer yielded significantly less DNA than the trigger, grip, forestock, slide and multiple (or cumulative swab -a single swab of multiple firearm parts) (p < 0.05). Swabs of multiple parts also yielded more DNA compared to the bolt, bolt handle, hammer, magazine and trigger (p < 0.05) (Supplementary Table 2). From the data provided, there were not any specific target areas for cumulative swabs, as these are selected at the discretion of individual crime scene officers depending on the case context provided to them. The hammer of the firearm -the mechanism used to strike the firing pin -is not a part as frequently touched or handled as the grip, trigger, or other parts of the firearm. It is also smaller than other firearms parts such as the grip which may also influence DNA recovery 20 . It is therefore predictable that less DNA would also be recovered from the hammer compared to other more frequently touched and larger components of a firearm.
CBCs included both plastic shotgun cartridges and metallic* cartridges, as well as unfired and fired cartridges. Therefore, based on the free text fields in our data, these categories were separated for analysis to gain a better understanding of DNA recovery between metal compositions and firing status. Figure 3 shows the amounts of DNA recovered from predominantly plastic shotgun cartridges/cases and 'other' metallic cartridges/cases. As information on the specific metal compositions of the cartridges was not available from the data, we could not further classify the 'other category' based on metal types. However, previous research under controlled conditions has shown that metallic composition also influences DNA recovery with brass cartridges and cases yielding less DNA potentially due to the reactive nature of the metal 8,9 . The median quantities of DNA obtained from shotgun and other cartridges/cases were 0.08 ng and 0 ng, respectively (mean values 1.20 ng and 0.32 ng, maximum values 2.9 ng and 4.44 ng, respectively). A Mann-Whitney U test showed a statistically significant difference (p < 0.05), with shotgun cartridges yielding more DNA than metallic (other) cartridges. Shotgun cartridges contain a metal head (base of ammunition) and plastic hull (outer case of ammunition). Based on the free text field within our data; a combination of the head and/or hull were sampled for shotgun cartridges. Previous research under simulated case work conditions has indicated that the hull of shotgun cartridges yields more DNA than the head 1 , although the sample sizes were limited. Other research has also shown better trace DNA recovery from plastic substrates over metal substrates 22 . Our case work results are therefore consistent with other research data and suggests that predominantly plastic shotgun cartridges/cases yield more DNA than metal cartridges/cases. Furthermore, the relatively larger size of shotgun cartridges may also influence DNA recovery.
Where information was available, metallic 'other' cartridges were further classified as unfired or fired. Figure 4 shows the total amounts of DNA recovered from unfired and fired cartridges. The median quantities of DNA recovered from unfired and fired cartridges were 0.005 ng and 0 ng, respectively (mean values 0.199 ng and 0.425 ng, maximum values 20 ng and 120 ng, respectively). A Mann-Whitney U test showed a statistically significant difference between them (p < 0.05). Research utilizing various DNA collection techniques under controlled conditions has demonstrated that firing decreases the quantities of DNA recovered from cartridge cases 1,3,5,8 , Our case work data also reflect the same trend. Furthermore, in casework situations, several other factors may contribute to the persistence of DNA on unfired and fired ammunition. These include the temperatures and conditions to which exhibits are exposed at crime scenes, time between DNA deposition and collection, time between DNA collection and processing, and the general nature of trace DNA (i.e. differences in shedder status or how the ammunition was handled prior to firing).
As exhibits in NSW are routinely examined for fingerprints before swabbing for DNA, we also assessed if there was any difference in the quantity of DNA retrieved from exhibits (n = 90) that were swabbed for DNA after fingerprinting (n = 64) and those that were only swabbed for DNA with no fingerprinting techniques applied (n = 26). As the sample size for this part of the study is smaller, we combined the data, regardless of firearm part or CBC, to understand the impact of fingerprint analysis on DNA recovery as a whole. Figure 5 shows the quantities of DNA recovered from firearm parts and CBC's after fingerprinting or where no fingerprint technique was applied. Swabs taken from exhibits in the 'fingerprinting before DNA' category yielded higher quantities of DNA than samples which were only swabbed for DNA (p < 0.05). Although certain fingerprinting techniques such as physical developer and silver nitrate have been shown to negatively impact further DNA analysis 23 , the techniques used in NSW do not appear to significantly impact the recovery of genetic material.
Based on our subset of data (n = 64), the typical sequence of analyses for fingerprinting firearms and ammunition in NSW is alternative light source examination followed by cyanoacrylate fuming. Where needed, cyanoacrylate fuming is followed with Rhodamine 6 G staining (46% of samples) and/or Gun Blue or Ardox staining (8% of samples, respectively). The remaining samples ceased fingerprint development at cyanoacrylate fuming (43% of samples) or were fingerprint powdered (3% of samples). Conventional cyanoacrylate fuming has been shown to not affect subsequent DNA analysis 24 and in some instances shown to yield more DNA than samples with no treatment applied 25 , consistent with our results. One potential reason why cyanoacrylate fumed samples yield more DNA than those with no treatment may be that fuming allows the DNA trace to be easily located therefore allowing optimal DNA collection. In addition, the genetic material trapped within the cyanoacrylate polymer could potentially be more readily released from the swab fibres due to its increased size 24 . The possibility of very low levels of cross contamination have also been observed within superglue chambers 26 .
A larger scale study allowing comparison of samples taken 'post fingerprinting' and 'without fingerprinting' of the same type of firearm components may provide greater insight into the impacts of fingerprinting methods applied on quantities of DNA recovered and the quality of the profiles they produce.

Profile data analysis
DNA profiles were classified as usable, DNA intel, unusable, and not tested, as described previously. Samples containing DNA below the FASS amplification threshold were not STR amplified, unless specifically requested. As a result, all samples that do not progress to STR amplification are auto-classified as unusable, as they never progress to the profiling stage of testing. However, our data show that 8 samples with DNA concentrations below the threshold were specifically requested to be profiled, and all produced informative profiles (7 usable, 1 DNA intel). There were 1020 samples with DNA concentrations above threshold of which 481 produced informative profiles (337 usable, 144 DNA intel). Our data also show that firearms and ammunition are more likely than not (56%) to yield DNA below the amplification threshold, however, 47% of samples with DNA concentrations above the threshold produced informative profiles (use for identification or intelligence only) to aid in investigations.
Information on whether the usable and unusable profiles were of mixed origin was also available in our data set, helping to distinguish which usable profiles were a major or minor component of a mixture, and which unusable profiles were too complex for investigation. Figure 5. Amounts of DNA (Log 10 (DNA conc (ng/µL)) recovered from firearm parts and CBC's when fingerprinting was conducted before DNA (n = 64) or no fingerprinting method was applied (n = 26). Boxes represent interquartile ranges, whiskers represent 1.5 times the interquartile range, '×' denotes the mean value and '-' represents the median value. Table 1 shows the percentages of usable, unusable (inclusive of those that did not progress to profiling), DNA intel and not tested trace swabs obtained from all 17 firearms parts and CBCs. The relative frequencies of usable profiles were less than 26% across all targeted areas. Cumulative swabs of 'multiple' areas gave the highest percentage of usable profiles, consistent with Figure 2, indicating that cumulative swabbing collects more of the available DNA. However, 22% of the cumulative swabbed samples were of mixed origin, as would be expected. Our casework data findings concur which previous research by Richert 27 who found sampling several smaller areas of a firearm provided better DNA results than swabs of individual areas. However, in contrast a recent study by Gosch et al 20 suggests sampling several smaller areas of a firearm to reduce profile complexity. Including individual sampling locations in the evaluation process is also suggested to further help in assessing competing hypothesis at the activity level 20 . Based on the competing findings, further research is required to determine whether cumulative swabbing or individual swabs of several smaller surfaces would be a better method of DNA collection from firearms.
From all of the usable profiles, all components except for the magazine, cylinder, and CBCs resulted in equal or slightly greater percentages of mixed profiles. Mixed profiles from the exterior of firearms were expected, as firearms used in crimes may have multiple handlers and/or may be concealed in various forms where non-handler DNA transfer is possible. Furthermore, due to the nature of trace DNA, it is possible that a single handler can deposit both self and non-self alleles on handled objects through secondary transfer 16 thereby resulting in mixed profiles.
The magazine and cylinder are where ammunition (CBCs) is loaded into a firearm. Given that the magazine, cylinder, and ammunition are concealed within the firearms, and that only one person is required to load a firearm, minimal adventitious alleles are expected from those parts compared to external areas of firearms. Whilst previous research under controlled settings have shown that adventitious DNA may transfer to fired cartridge cases within the firearm magazine or cylinder, our case work data has shown lower percentages of mixed profiles (6%, 5% and 2%, for usable, and 22%, 0%, 3% for unusable, respectively) from concealed parts of the firearm (magazine, cylinder and CBC) compared to non-mixed profiles (8%, 16% and 4% for usable, and 57%, 63% and 90% Table 1. Summary of profile data obtained from all 17 components of firearms including CBCs. Grey bars represent the relative (normalized) percentages of profiles in each column. 'Unusable' is also inclusive of samples that never progressed to profiling. for unusable, respectively) (data not shown). Nonetheless, profiles obtained from all parts of firearms and CBCs should be interpreted with caution due to the high percentages of mixed origin alleles.
Most samples (79%) from all 17 categories overall were unusable for casework. From these unusable samples, 72% were classified as 'unusable' because they resulted in no profile or a profile too weak or did not meet the threshold for profiling. The remaining 28% of unusable samples were mixtures that were too complex. Again, the occurrence of mixed profiles from casework samples suggests that either there are multiple handlers of firearms or adventitious alleles through indirect transfer readily persist on firearms. Our results are comparable to those of earlier studies, whereby a large percentage of casework exhibits (including firearms) produced incomplete, partial or no profiles 11,12,15 . Furthermore, research has also shown firearms overall have produced lower DNA recovery rates compared to robbery items, packaging and tools 15 .
In addition to usable and unusable samples, up to 21% of samples produced DNA intel profiles whereby the profile could not be uploaded to the database but provided some useful information, e.g. could be used to determine whether a nominated individual can or cannot be excluded as a contributor.
As with the quantitative data, CBCs were further classified into shotgun cartridges and other (metallic cartridges), as well as unfired and fired cartridge cases, to assess any potential differences. Table 2 shows the profile data relating to shotgun cartridges/ cases and 'other' metallic cartridges/cases. Table 3 shows the profile data relating to unfired and fired metallic cartridge and cases. For both shotgun and other cartridges/ cases, and unfired and fired cartridges, a higher percentage of samples (up to 94 and 98%, respectively) were not usable for casework. Comparable to the firearm components, most unusable samples (52% shotgun, 92% other) were either too weak or produced no profiles or did not meet the threshold for profiling. From the usable profiles obtained from shotgun and other cartridges/cases, and unfired cartridges, 6% of shotgun, 2% of other, and 4% of unfired metallic cartridges produced mixed profiles. Over the entire four-year period examined, only six usable DNA profiles were obtained from fired cartridge cases. This is as expected from the general lack of DNA recovery success observed across multiple jurisdictions. A small percentage of samples from shotgun, other and unfired cartridges also produced DNA intel profiles where individuals could or could not be excluded.  DNA collection after fingerprinting or when no fingerprinting technique was applied did not appear to significantly influence the percentages of usable and unusable samples (Table 4). Although the quantities of DNA recovered from firearm parts and CBCs were greater following fingerprint analysis (p < 0.05), profile analysis showed that a large percentage (88%) of those samples were unusable for investigative purposes regardless of whether DNA was swabbed without having been fingerprinted or after fingerprinting (Table 4). From the DNA collected following fingerprinting, 70% of the samples above our jurisdiction's quantitation threshold were unusable (data not shown). From the DNA collected where no fingerprint technique was applied, 60% of the samples above the FASS quantitation threshold were unusable (data not shown).
The time between exhibit collection and DNA collection from the exhibits ranged between the same day to three years (mean within 1.8 months, median within 1 month). Two swabs (of the grip and 'multiple' areas of the firearm) collected and analysed after three years produced a usable mixed and unusable DNA profile, respectively. Seven swabs of various firearm components were collected and analysed two years after collection, with only one producing a usable profile. Approximately six swabs (all unfired cartridges) were collected and processed after a year with only one producing a usable profile. The remaining swabs were taken within a year of the exhibit being collected. Previous research reviewing trace DNA recovery rates from volume crime exhibits has shown that time between collection and laboratory submission does not appear to influence DNA quantity or quality 15 . Our data also shows that usable DNA profiles were still obtainable from exhibits over a year after collection.
Identifying casework DNA recovery rates relating to common evidence items is imperative in guiding sample triaging, informing prioritization of areas to target for sampling, identifying areas of improvements/research (including improved DNA recovery methods and/or the transfer and persistence of trace DNA) and managing stakeholder expectations. Additionally, comparing DNA recovery success rates across multiple jurisdictions can further drive considerations and efforts towards improvements.
Trace DNA profiling recovery rates from Queensland (QLD), Australia where swabbing is employed indicate suspect identification in up to 8% of profiles obtained from firearms (more specifically, the handle, barrel and trigger) and up to 4% of profiles from unfired and fired cartridge cases 12 . Meanwhile a large percentage of samples from firearms (86-100%) and ammunition (up to 97%) contained no DNA. Similar results have also been reported from the Netherlands where 74% of samples from firearm grips and 86% of samples from cartridge cases produced no DNA results 11 . Both the QLD and the Netherlands data represent similar proportions of usable and unusable DNA results to those in NSW, suggesting a need to investigate alternative/novel/new approaches to sample collection, and/or more sensitive robust amplification methods.
It is important to note that direct comparisons between jurisdictions are difficult to make due to a multitude of factors that can influence DNA recovery. These factors include DNA collection methods/techniques (both between jurisdictions and between person to person), DNA profiling methods and chemistry, profile interpretation, database upload criteria, and differences in data analysis approaches between publications. Therefore, what is considered a successful recovery result in one jurisdiction/publication may differ compared to another jurisdiction/publication.

Conclusion
DNA recovery was found to be highly variable among casework samples taken from different areas of firearms as well as samples from similar areas from different firearms. Nonetheless, historical data from casework exhibits can help prioritize the types of samples taken from crime scenes and firearms exhibits. Our results show that CBCs yield less DNA than most other firearm parts. Swabs of the hammer and safety of the firearm generate the least DNA while cumulative swabbing generates the most. From the CBCs, shotgun cartridges and unfired metallic cartridges returned greater quantities of DNA than from metallic and fired cartridge cases, respectively. Sample data show that 64-98% of samples from the various firearm parts and ammunition (including shotgun and metallic, and unfired and fired cartridges) were unusable for casework due to the samples either containing DNA lower than the concentration threshold for amplification or being too weak or complex for analysis. Areas of the firearm that are cumulatively swabbed are most likely to produce usable profiles, whilst the areas least likely to produce usable profiles are the bolt, butt, scope, and CBC's. In line with expectations, internalized components of firearms including the magazine, cylinder and CBCs produced slightly higher percentages of non-mixed profiles than mixed profiles.
Cumulative swabbing, which has demonstrated greater DNA recovery and higher percentages of usable profiles, may be a useful trace DNA sampling technique for firearms but will be expected to produce mixed source profiles. The use of probabilistic genotyping can be employed to determine likelihood ratios for propositions relating to individual contributors. Nonetheless, further research is needed for optimal sampling techniques for both firearms and cartridge cases, as well the prioritization of DNA sampling when triaging exhibits.

Note
*Metallic cartridges encompass all cartridges other than shotgun.