Building Interpretable Machine Learning Models to Identify Chemometric Trends in Seabirds of the North Pacific Ocean.

Marine environmental monitoring efforts often rely on the bioaccumulation of persistent anthropogenic contaminants in organisms to create a spatiotemporal record of the ecosystem. Intercorrelation results from the origin, uptake, and transport of these contaminants throughout the ecosystem and may be affected by organism-specific processes such as biotransformation. Here, we explore trends that machine learning tools reveal about a large, recently released environmental chemistry data set of common anthropogenic pollutants measured in the eggs of five seabird species from the North Pacific Ocean. We modeled these data with a variety of machine learning approaches and found models that could accurately determine a range of taxonomic and spatiotemporal trends. We illustrate a general workflow and set of analysis tools that can be used to identify interpretable models which perform nearly as well as state-of-the-art "black boxes." For example, we found shallow decision trees that could resolve genus with greater than 96% accuracy using as few as two analytes and a k-nearest neighbor classifier that could resolve species differences with more than 94% accuracy using only five analytes. The benefits of interpretability outweighed the marginally improved accuracy of more complex models. This demonstrates how machine learning may be used to discover rational, quantitative trends in these systems.

Remote blubber sampling paired with liquid chromatography tandem mass spectrometry for steroidal endocrinology in free-ranging bottlenose dolphins (Tursiops truncatus).

Liquid chromatography tandem mass spectrometry allows for the measurement of steroid hormone suites in the blubber of marine mammals. By combining this technology with minimally invasive techniques such as remote biopsy, endocrine profiles can be assessed, allowing for studies of hormonal profile variation over time. In this study, we explored associations among different steroidogenic pathways and seasonal differences in blubber hormone profiles of free-ranging common bottlenose dolphins along the coast of South Carolina, USA. Male dolphins experience a peak in testosterone, androstenedione, progesterone, and 17-hydroxyprogesterone in the spring, likely related to an upregulation of the androgen steroidogenic pathway during mating season. We also observed increased cortisol concentrations during summer compared to winter. Among females, there was an increase in androstenedione with elevated progesterone concentrations indicative of pregnancy, highlighting another potential endocrine marker for pregnancy in free-ranging dolphins. This work emphasizes the importance of selecting the appropriate season for studies on endocrine status to effectively uncover physiological variation or disruption in free-ranging cetaceans.

LipidQC: Method Validation Tool for Visual Comparison to SRM 1950 Using NIST Interlaboratory Comparison Exercise Lipid Consensus Mean Estimate Values.

As advances in analytical separation techniques, mass spectrometry instrumentation, and data processing platforms continue to spur growth in the lipidomics field, more structurally unique lipid species are detected and annotated. The lipidomics community is in need of benchmark reference values to assess the validity of various lipidomics workflows in providing accurate quantitative measurements across the diverse lipidome. LipidQC addresses the harmonization challenge in lipid quantitation by providing a semiautomated process, independent of analytical platform, for visual comparison of experimental results of National Institute of Standards and Technology Standard Reference Material (SRM) 1950, "Metabolites in Frozen Human Plasma", against benchmark consensus mean concentrations derived from the NIST Lipidomics Interlaboratory Comparison Exercise.

Using procedural blanks to generate analyte-specific limits of detection for persistent organic pollutants based on GC-MS analysis.

Several methods are used to generate a limit of detection for organic pollutants measured by gas chromatography-mass spectroscopy (GC-MS); all have theoretical and practical drawbacks. The current project investigated two common existing methods (statistical and empirical) for applicability to chromatographic properties from real samples, comparing these with a new proposed method using procedural blanks to estimate a minimum detectable peak area. Weaknesses of all three methods are discussed. The proposed method was superior to other examined methods in that it provided analyte-specific limits of detection linked to the recovery of mass-labeled internal standards for every analyte within every sample. Other identified quality assurance benefits included the following: enhanced protection against false positives; providing a sensitivity performance metric across batch, analyst, and instrument; enabling chemists with discretionary decisions specific to every analyte regarding detectability and interferences; and some strengths of both statistical and empirical techniques without major drawbacks of either. In marine sediment samples, the proposed method of calculating the limit of detection increased reporting of trace level (low- to subppb) GC-MS data for polychlorinated biphenyls (PCBs), polybrominated diphenyl ethers (PBDEs), organochlorine pesticides (OCPs), and polycyclic aromatic hydrocarbons (PAHs) by up to 400% compared with the statistical method.

A Portable and Reusable Database Infrastructure for Mass Spectrometry, and Its Associated Toolkit (The DIMSpec Project).

Nontargeted analysis (NTA) is a rapidly growing field of techniques that includes the identification of unknown chemical analytes in complex mixtures such as environmental, biological, and food matrices. The use of reference mass spectral databases is a key component of most NTA workflows, providing a high level of confidence for chemical identification when analytical standards are not available, yet effective interlaboratory sharing of research grade spectra remains challenging. The Database Infrastructure for Mass Spectrometry (DIMSpec) project focused on the creation of an open-source toolkit supporting storage and sharing of high-resolution mass spectra with attached sample and methodological metadata. As a demonstration of its utility, the DIMSpec toolkit was used to create a database of curated mass spectra for per- and polyfluoroalkyl substances (PFAS) generated from various sources. While the underlying toolkit is agnostic to analytical targets, this initial release (along with the database schema, mass spectral data, and database tools) should enable PFAS researchers to use these data for their own studies, including the identification of novel PFAS in the environment.

DIGITAL NIST: An examination of the obstacles and opportunities in the digital transformation of NIST's reference materials.

Early in 2022, NIST embarked on a pilot project to produce digital calibration reports and digital certificates of analysis for reference materials. The goal is to produce examples of digital reports and certificates to assess the scope and challenges of digital transformation in those particular measurement services. This paper focuses on the Reference Material Certificate effort of the pilot project. Our aims for this part of the pilot project are: to generate a digital Reference Material Certificate from certification data; descriptive information about the material, and other data and metadata as needed; to generate a human-readable report from the digital Reference Material Certificate; and to hold a workshop to gather stakeholder feedback. The challenges for NIST include the diverse and complex information presently contained in NIST certificates, converting values to non-SI units to match the needs of stakeholders, and format updates to NIST Reference Material Certificates necessary to allow for machine generation. Other practical challenges include the wide variety of Reference Materials offered by NIST, as well as the needs of internal and external stakeholders. This presentation will report on the progress of the NIST effort and discuss some of the challenges and solutions to producing Digital Reference Material Certificates.

Perfluoroalkyl substances in diamondback terrapins (Malaclemys terrapin) in coastal South Carolina.

This study focused on an estuarine wildlife species exhibiting high site fidelity and ubiquitous distribution in coastal environments along the Atlantic and Gulf coasts of the United States to monitor per- and polyfluoroalkyl substances (PFAS). A total of 75 diamondback terrapin (Malaclemys terrapin) plasma samples were collected from five creeks associated with Kiawah (Oyster Creek, Fiddler Creek, Sandy Creek, Gnat Creek) and Edisto (Townsend Creek) islands in Charleston County, South Carolina and investigated for 15 legacy PFAS. Of those, PFHxS was the only PFAS found in all terrapin plasma samples. Four additional PFAS were routinely detected (greater than 90% of the samples) and were included in statistical analyses: PFOS, PFNA, PFDA, and PFUnA. Sex-differences were observed for two creeks with male plasma containing higher PFAS than female plasma (PFHxS at Townsend Creek, PFOS at Oyster Creek). Sex-specific site differences in PFAS concentrations were observed primarily for males, suggesting male terrapins may be more sensitive indicators of localized contaminant profiles than females. Three PFAS were observed to have negative correlations with body mass: PFOS in males (p = 0.045, tau = -0.220), PFNA in males (p = 0.016, tau = -0.269), and PFHxS in both males (p = 0.007, tau = -0.302) and females (p = 0.001, tau = -0.379). No relationships for body mass and PFDA and PFUnA were observed.

Feasibility of using the National Marine Mammal Tissue Bank for retrospective exploratory studies of perfluorinated alkyl acids.

Perfluorinated alkyl acids (PFAAs) have been used for 50+ years in materials such as stain-resistant treatments for paper and clothing, lubricants, and foam fire extinguishers. PFAAs are characterized by a fully fluorinated alkyl chain with a terminal acid group. Their long half-lives and ubiquitous environmental distribution create considerable concern for wildlife and human exposure. There is interest in examining temporal trends of PFAAs using the National Marine Mammal Tissue Bank (NMMTB), but NMMTB tissues are frozen and cryohomogenized in polytetrafluoroethylene (PTFE)-based materials. Because PTFE supplies may leach PFAAs into samples, this study mimicked collection, processing and storage steps of NMMTB samples and measured PFAA leaching to determine the feasibility of using this sample archive for PFAA temporal trends. We also explored concentrations in Atlantic white-sided dolphin (Lagenorhynchus acutus, WSDs) and rough-toothed dolphin (Steno bredanensis, RTDs) blubber (n=3 and 0) and liver (n=48 and 12, respectively). The materials used in NMMTB protocols may add up to 0.968ng/g perfluorooctanoic acid (PFOA), 0.090ng/g perfluorononanoic acid (PNFA), and 0.221ng/g perfluorooctane sulfonate (PFOS) to each archived sample. Leaching of PFNA and PFOS from supplies compared to dolphin levels was negligible, but PFOA contributions were substantially higher than levels found in most dolphin liver samples. Therefore, monitoring PFOA temporal trends from the NMMTB would require careful consideration. RTDs had significantly higher levels of PFOS and PFNA than WSDs. Both species have similar life history, trophic status, and foraging behaviors in deep pelagic waters, so differences could be from latitudinal variation in contamination. RTDs stranded in Florida; WSDs stranded farther north mostly in Massachusetts. Juveniles had significantly higher levels of PFOS and PFNA than adults in both species, suggesting growth dilution as they approach maturity. PFOS significantly decreased after 2001 in both species as expected based on changes in production.

LipidPioneer : A Comprehensive User-Generated Exact Mass Template for Lipidomics.

Lipidomics, the comprehensive measurement of lipid species in a biological system, has promising potential in biomarker discovery and disease etiology elucidation. Advances in chromatographic separation, mass spectrometric techniques, and novel substrate applications continue to expand the number of lipid species observed. The total number and type of lipid species detected in a given sample are generally indicative of the sample matrix examined (e.g., serum, plasma, cells, bacteria, tissue, etc.). Current exact mass lipid libraries are static and represent the most commonly analyzed matrices. It is common practice for users to manually curate their own lists of lipid species and adduct masses; however, this process is time-consuming. LipidPioneer, an interactive template, can be used to generate exact masses and molecular formulas of lipid species that may be encountered in the mass spectrometric analysis of lipid profiles. Over 60 lipid classes are present in the LipidPioneer template and include several unique lipid species, such as ether-linked lipids and lipid oxidation products. In the template, users can add any fatty acyl constituents without limitation in the number of carbons or degrees of unsaturation. LipidPioneer accepts naming using the lipid class level (sum composition) and the LIPID MAPS notation for fatty acyl structure level. In addition to lipid identification, user-generated lipid m/z values can be used to develop inclusion lists for targeted fragmentation experiments. Resulting lipid names and m/z values can be imported into software such as MZmine or Compound Discoverer to automate exact mass searching and isotopic pattern matching across experimental data. Graphical Abstract ᅟ.

Persistent organic pollutants in blood plasma of satellite-tracked adult male loggerhead sea turtles (Caretta caretta).

Risks from persistent organic pollutants (POPs) remain largely a mystery for threatened loggerhead sea turtles (Caretta caretta). The present study examines regional-scale POP differences in blood plasma from adult male C. caretta based on movement patterns. Turtles were captured near Port Canaveral, Florida, USA, in April of 2006 and 2007 and fitted with satellite transmitters as part of a National Marine Fisheries Service-funded project. Residents (n = 9) remained near the capture site, whereas transients (n = 10) migrated northward, becoming established in areas largely from south of Pamlico Sound, North Carolina, to north of Cape May, New Jersey, USA. Blood was sampled from the dorsocervical sinus of each turtle and analyzed using gas chromatography-mass spectrometry for organochlorine pesticides (OCPs), polychlorinated biphenyls (PCBs), polybrominated diphenyl ethers (PBDEs), and toxaphenes. Blood plasma concentrations of OCPs and total PBDEs were elevated in transients (p < 0.05) and in some cases were correlated with turtle size. Migratory adults showed an atypical PBDE congener profile relative to other published studies on wildlife, with PBDE 154 being the dominant congener. Additionally, PCB congener patterns differed between groups, with total PCBs slightly elevated in transients. This supports the idea that foraging location can influence exposure to, and patterns of, POPs in highly mobile species such as C. caretta. Understanding patterns of contamination informs wildlife managers about possible health risks to certain subpopulations. The present study is the first to examine POPs in the rarely studied adult male sea turtle and to couple contaminant measurements with satellite tracking.