The second half of 2018 produced a familiar mix of genuinely useful methods, important environmental observations, clever experiments, and papers whose conclusions deserved more caution than their titles suggested. The subjects ranged from microplastics and indoor pollution to non-target screening, climate impacts, occupational exposure, and the limits of machine learning in environmental toxicology.

July: New ways to read complicated environmental systems

Environmental processes are difficult to reduce to a few variables. Life-cycle assessments are a good example: households do not consume goods in the same way, so a single “average” consumer can hide important patterns. One study used self-organizing maps, an unsupervised-learning method, to identify household consumption profiles and extract their characteristic features. There is a little methodological showmanship in the choice of technique, but the application is genuinely suggestive. Environmental assessment may benefit from recognizing different behavioral patterns instead of forcing everyone into one model household.

Linking marine microplastics with hydrophobic organic pollutants is another sensible direction. Microplastics are often less important as isolated particles than as carriers, surfaces, or transport media for other chemicals. Some approaches developed for soil contamination may therefore be worth adapting to the study of plastic particles in the marine environment.

Only a limited number of rivers empty into the Arctic Ocean from within the Arctic Circle. A long-term investigation of these rivers found that carbonate and bicarbonate concentrations had risen steadily over approximately four decades. The result suggests that ocean acidification is not driven only by processes occurring in the ocean itself; riverine inputs may also make a meaningful contribution.

Perchlorate is a common subject in environmental chemistry. By analyzing ice cores from Greenland covering roughly 300 years, researchers found that perchlorate concentrations had remained relatively stable and did not appear to track human activity closely for most of that period. Several historical volcanic eruptions, however, coincided with high atmospheric concentrations. The increase observed during the most recent two decades may have a different explanation, possibly involving the production and transformation of organochlorine compounds.

A viewpoint by Chinese environmental chemists reviewed the improvement in air quality in China and the work that remains. Fine-particle concentrations have continued to decline, but they are still above World Health Organization guidelines. The next stage of air-pollution control may need to focus less on particle mass alone and more on the toxic substances carried by those particles.

Non-target analysis can overwhelm new users with tens of thousands of peaks. One study did not introduce a particularly unusual overall workflow, but its feature-filtering strategy reduced the number of peaks from about 60,000 to roughly 1,500. That reduction is important in practice: before a signal can be identified, the analyst has to decide which signals are worth pursuing.

An economic analysis discussed in the media examined pollution alongside household consumption data in China. The results suggested that the cost of air-pollution control is not limited to government spending. Individual behavior changes as well. Residents in heavily polluted areas spent more on health care and medicine, while spending on everyday necessities fell relatively. The latter interpretation is not entirely convincing; housing prices and the resulting debt burden might also be responsible for lower spending on basic goods.

One direct consequence of antibiotic misuse is the emergence and spread of antibiotic-resistance genes. A global survey measured their relative abundance in total suspended particles from 19 cities and examined changes in Xi’an over a ten-year period. The profiles varied substantially between cities and among different antibiotics. It is not impossible that air-quality forecasts will eventually need to include some kind of resistance-related indicator.

High mercury concentrations have been reported in municipal sewage sludge in Tibet. Using mass-flow analysis and pharmacokinetic modeling, researchers estimated that mercury in traditional Tibetan medicines could account for as much as 45% of the mercury contamination in the environment. Tibet may also act as a cold-trapping region for mercury because of its geography and climate. If the estimate is reliable, future mercury pollution in the region could become an unusually serious environmental phenomenon.

August: Exposure pathways, analytical workflows, and unusual samples

The detection of biocides and their transformation products in groundwater is not, by itself, especially surprising. The more valuable part of one study was its sampling design. By placing sampling points in a way that allowed the researchers to follow the movement of water, the study showed that the pollutants were being carried into groundwater by stormwater runoff.

A more unusual case involved racing pigeons. A winning pigeon in a European race was accused of having received morphine as a performance-enhancing drug. The complication was that poppy seeds are common in both pigeon feed and human food—there may even be some on a bagel. The owner asked scientists to investigate. Morphine and other opiates could indeed be detected in pigeon droppings, but the elevated concentrations lasted only a short time, less than two days. Since the birds had left their owner roughly a day before the race, the evidence was not sufficient to establish effective doping. Laboratory simulations, of course, cannot perfectly reproduce the conditions of an actual flight.

Air-pollution research has focused heavily on fine particles in recent years, but one study argued that China’s ground-level ozone pollution is also severe when compared with that of developed countries. Treating ozone as a secondary issue would therefore be a mistake.

Dissolved organic matter has fluorescent properties, and these properties change with temperature. That observation may be more useful than a simple description of fluorescence. The temperature dependence could potentially provide information about more complicated environmental processes, although the study leaves much of that possibility unexplored.

Different fields often give different names to essentially similar analytical strategies. The homologue-searching method used in one paper would be described in lipidomics as a form of semi-targeted analysis. What makes the work worth attention is that it was connected to an exposure-assessment workflow. Environmental science does not necessarily lack analytical technologies; it often lacks practical ways to combine existing technologies into a coherent workflow.

A distinctly physical study examined the consequences of global warming for drinking-water treatment. As temperatures rise, indicators such as disinfection by-product formation and electricity consumption at water-treatment plants are expected to increase. Many existing facilities were not designed with these conditions in mind. Future treatment equipment may need something resembling climate control of its own.

The accuracy of environmental analysis is often questioned, and results are sometimes treated as reliable only to the nearest order of magnitude. A multi-laboratory comparison of emerging flame retardants offered a more precise diagnosis. Instrumental analysis was not the principal reason for poor accuracy. Matrix effects and sample preparation appeared to be more important sources of error. Better instruments alone will not solve problems introduced before the sample reaches the instrument.

In many small and medium-sized North American cities, residential areas are arranged in blocks, with commercial districts containing restaurants and shops located between groups of blocks. A study of restaurant emissions found a clear increase in fine particles within 200 meters downwind of restaurants. Because around one-fifth of Pittsburgh’s population lived in areas with this kind of layout, restaurants could have a larger effect on exposure than their modest size would suggest—and in some places may matter more than factories.

DNA aptamers are a popular topic in environmental analysis. These modified DNA molecules are designed to bind particular contaminants and produce a fluorescent response on a suitable sensing platform. A review, however, noted that the practical limitations remain substantial. Sensitivity is not yet sufficient to measure many contaminants at environmentally relevant concentrations, and DNA fragments naturally present in environmental samples can interfere with the measurement.

Some studies are memorable mainly because of their samples. One examined endocrine-disrupting chemicals in hosiery collected from different countries. The concentrations of several known endocrine disruptors were not negligible. In particular, hosiery containing Lycra or Spandex showed bisphenol S and bisphenol A at concentrations in the milligram-per-gram range. Because these products are worn directly against the skin, the potential for close-contact exposure deserves attention.

Another non-target study analyzed mummies and the jars used to store their organs. Beeswax components appeared in several samples, while some substances not previously mentioned in the literature as preservatives—including salicylic acid—were also detected. The difficult part is interpretation. After thousands of years, it is not easy to determine whether a detected chemical was deliberately added during preservation or formed through long-term degradation and transformation.

A useful occupational-exposure study compared three groups: office workers, taxi drivers, and security guards. Researchers collected hand-wipe samples, surface-wipe samples, and dust, then examined relationships among the different sample types for flame retardants. Based on the target compounds, the apparent exposure pattern was taxi drivers greater than office workers, followed by security guards. Hand-wipe and surface-wipe results were well correlated, whereas hand-wipe results and dust concentrations were not. The next question should be which sample type best reflects concentrations inside the body, since internal dose is ultimately the point of exposure assessment.

September: Tracing products, materials, and environmental histories

A metabolomics-like approach was used to characterize the chemical profiles of raw milk, pasteurized milk, and UHT milk. Oxidized lipids emerged as potential markers for distinguishing the different products. The study is a reminder that food processing can leave a measurable chemical signature even when the products appear similar in routine testing.

If a city is treated as a single system, the movement of agricultural products between urban areas and farmland can be viewed as a kind of import-and-export network. A map of agricultural trade for U.S. metropolitan areas found that nearly 80% of them depended on imported agricultural products. Comparable urban–rural transfers clearly exist in China as well. Economic exchange does not necessarily mean that resources are distributed spatially in a sustainable or rational way, and these data could support more realistic urban-planning decisions.

DNA is generally assumed to be more stable than RNA. A photolysis experiment comparing single- and double-stranded DNA and RNA produced a counterintuitive pattern: double-stranded molecules were more stable than single-stranded ones, but RNA was more stable than DNA under the tested conditions. The result may be relevant to evolutionary questions and could help explain the environmental persistence and transformation of RNA viruses.

A study attracted attention by examining a possible relationship between PFAS exposure and hyperthyroidism in cats. Once the data were inspected, however, the difference between cats with and without hyperthyroidism was not especially large. The animals were all at least ten years old, which also raises the mundane but important question of how age should be treated when studying disease in older cats.

National urbanization statistics are usually reported at fairly broad scales. One study used nighttime lights as a proxy for urbanization at the kilometer scale and regressed the light data against flows of materials such as steel, cement, and aluminum. The results suggested that nighttime illumination can help investigate material flows across the land surface. There is room to push the idea further, perhaps by combining it with spectral analysis.

Tree rings do more than record a tree’s age. They can also preserve a history of changes in metal pollution. One study used tree rings to reconstruct mercury contamination during Europe’s historical gold-mining period, when amalgamation was widely used. Tree rings, ice cores, and sediment cores are all familiar archives in environmental chemistry because they preserve changes over time and provide relatively independent records. The catch is that collecting them is rarely easy.

As people spend more time indoors, indoor pollution may deserve at least as much attention as outdoor pollution. Common advice includes opening windows, airing out a room, or using photocatalytic paint to remove odors. One study found that some photocatalytic paints marketed for indoor deodorization could themselves release high concentrations of volatile organic pollutants. A treatment intended to improve indoor air may therefore introduce another exposure pathway.

Not every interdisciplinary paper deserves equal confidence. One study applying machine learning to an environmental problem contained conceptual errors in several places. To someone who actually works in machine learning, it might look more like an exercise assigned to students than a mature interdisciplinary contribution. The presence of a fashionable method does not make the underlying science sound.

For researchers entering non-target analysis, a review from the U.S. Environmental Protection Agency provided a useful overview of the available tools. As the field grows, clear explanations of software, data processing, and practical workflows are becoming at least as important as the introduction of yet another analytical platform.

October: Pollution controls and unintended consequences

Trace amounts of melamine were still detectable in powdered milk and liquid milk products sold in the United States in 2017. The concentrations were far below those associated with the 2008 contaminated-milk scandal, but liquid formula contained more melamine than powdered formula. Persistent low-level contamination may have originated from cattle feed.

Media coverage sometimes gives the impression that eutrophication is improving, yet the underlying situation continues to deteriorate in many places. A study of the Gulf of Mexico examined the spatial and temporal distribution of the marine low-oxygen zone. Similar problems occur in China’s coastal waters, where they can have serious consequences for fisheries.

The shale-gas revolution helped turn the United States into an energy exporter. Replacing coal and oil with natural gas can reduce carbon dioxide emissions, but methane leakage complicates the picture. Measurements near natural-gas extraction sites showed regional peaks in methane concentrations. Methane is itself a potent greenhouse gas, so a fuel transition is not automatically climate-friendly if leakage is poorly controlled.

Childhood obesity in the United States is deeply rooted. One study considered roughly 300 possible contributing factors and used random forests to identify the variables most strongly associated with risk. Most of the high-ranking factors were social—for example, the crime rate in a school district—rather than the conventional environmental measures, such as the number of free sports facilities nearby. The machine-learning component is eye-catching, but the broader lesson is that public-health risks are embedded in social environments as much as in physical surroundings.

A commentary by a senior researcher on non-target screening focused on reproducibility. The discussion included many sensible points, even if some remain open to debate. As non-target methods become more common, the field will need agreement on how results are confirmed, shared, and reproduced across laboratories.

Epidemiological research has increasingly appeared in journals once associated mainly with the natural sciences. One study found that childhood gut microbiota could help indicate the later risk of type 1 diabetes. Breastfeeding remained the most important factor associated with the development of a diverse gut community; having siblings and furry pets at home also appeared to matter. A companion paper from the same research program examined the establishment of children’s gut microbiota. These studies are especially useful because the gut microbiome, related metabolites, and enzyme systems are still developing in early childhood and are not yet stable. They offer a useful reference point for future omics research involving young children.

Another viewpoint considered the use of machine learning in environmental toxicology. The authors argued that machine learning should have a larger role in the field, but that both acceptance and the quality of current applications remain insufficient. Their phrase “ML-literate scientists” is particularly useful: environmental researchers do not all need to become machine-learning specialists, but they should understand what the methods can and cannot do.

Ocean-going vessels often switch to low-sulfur fuel to control major conventional pollutants. One study found that this change could increase emissions of intermediate-volatility organic compounds. Pollution-control measures frequently suppress one group of pollutants while creating another problem. Comprehensive assessment is necessary, although this leads directly into the difficulty of dealing with unknown unknowns. In practice, environmental control still often means moving forward while learning from incomplete information.

Environmental chemical behavior is closely tied to politics, economics, and technology. A change in China’s steel standards altered the amount of vanadium used in steel production, and vanadium-ore prices doubled in 2017. Greater demand for the ore could lead to greater environmental release. Historical records also suggested that vanadium pollution, after declining from the 1970s onward, had begun to rise again. These tangled relationships are precisely what make environmental research interesting—and make controlled experiments so difficult.

November: Scale, exposure, and the meaning of “real-world” risk

Environmental research has become increasingly specialized. We often ask about one contaminant in one location, but large-scale patterns still matter. A global study estimated daily concentrations of fine particulate matter across the world, providing a broad baseline for comparison. Such global estimates cannot replace local measurements, but they are valuable for understanding the overall scale of exposure.

Indoor-pollution studies conducted in China often borrow methods and assumptions developed elsewhere. Lifestyle differences need to be included more carefully. In China, carpets are less common in many homes, and households that use them may designate indoor slippers. In many Western homes, people may walk on carpets barefoot or in outdoor footwear. These habits alone can change exposure estimates. Building ventilation, heating systems, and construction standards also differ, so models should be adjusted rather than transferred unchanged.

Secondhand smoke is inhaled while someone else is smoking. Thirdhand smoke is what remains after the smoker has left: the room absorbs the chemicals and releases them slowly over time. One study found that different pollutants were distributed among different aerosol fractions. Understanding thirdhand smoke would benefit from linking these distributions to bioavailability rather than treating every measured concentration as equally available to the body.

Growing food demand is pushing agriculture toward more intensive production. A global assessment discussed the environmental consequences of agricultural intensification, with particular attention to nitrogen emissions. It provides useful baseline information, although the analysis would be stronger if it were more closely connected to global climate change.

Tire wear produces rubber particles, but their ecological significance cannot be inferred from particle presence alone. A study used benthic organisms to examine toxicity and found limited effects. The authors further considered how different leaching procedures might alter the apparent toxicity, suggesting that some very high toxicity estimates may result from overly aggressive sampling or extraction. Under realistic environmental conditions, contaminants in particles may not be highly bioavailable. Environmental analytical chemistry should therefore connect concentration measurements with field observations and biological availability.

Short-chain chlorinated paraffins already contain an enormous number of isomers and homologues; the analytical problem becomes even more difficult for medium- and long-chain chlorinated paraffins. For such chemically unruly contaminant groups, the research community needs common analytical methods and standards. The discussion is also relevant to other families of compounds, including bisphenols, PFAS, and environmental DNA, where inconsistent definitions and workflows can make results difficult to compare.

A commentary on climate politics made a straightforward point. Political leaders may reject climate science while supporting expensive infrastructure to protect coastlines from rising seas. Sea walls and other coastal defenses can cost as much as highly visible political construction projects, yet unlike a border wall, coastal barriers may be one of the few practical options for some communities. Even people who do not accept the causes of climate change still cannot prevent seawater from flooding low-lying areas. Reducing fossil-fuel use and developing alternative energy remain the more fundamental responses.

Isotopes are among the most powerful tools in environmental chemistry. Radioactive isotopes can be used as tracers; stable isotopes can identify sources when their signatures are sufficiently distinct, or indicate environmental processes when they are transformed. One study found that stable strontium isotopes could signal combustion sources and remained relatively stable in both water systems and fish. Such properties make them potentially useful for tracing pollution through aquatic food webs.

Roads in snowy parts of the United States are kept passable partly by spreading enormous quantities of salt—an average of about 62.1 kilograms per person per year. At that scale, road salt is no longer an incidental material. It was found to affect the quality of water from private wells and to corrode building structures. The situation resembles a classic tragedy of the commons: each individual application seems reasonable, but the accumulated environmental cost is shared by everyone.

December: Better accounting for complex chemicals

PFAS are difficult to include in a complete mass balance because the group contains so many compounds. One existing approach used gamma radiation to measure total fluorine on the surface of paper products. A newer study applied X-ray photoelectron spectroscopy to measure surface fluorine on consumer goods. The unresolved challenge is how to determine total PFAS in a large, complex environmental sample rather than only measuring what is present at its surface.

Beijing’s PM2.5 contains a wide variety of chemical material, so the detection of antibiotic-resistance genes in those particles is not particularly surprising. The finding is nevertheless another indication that particulate pollution can carry more than conventional inorganic and organic pollutants.

Microplastics may not be highly toxic as particles by themselves, but their interactions with other chemicals can matter. Plastic fragments can adsorb pollutants and can also release additives into water. One study showed that phthalates leached from plastic fragments into seawater, providing a plausible exposure route for aquatic organisms.

Silicone wristbands have finally begun to be combined with non-target analysis for personal exposure assessment. One experiment used the bands over a short period of five days and was not aimed at a narrowly defined list of chemicals. Instead, it focused mainly on patterns in the compounds accumulated by the wristbands. This is still an early use of the method, but passive personal samplers may become useful for comparing exposure profiles across people and locations.

Electronic cigarettes have become popular in part because of their many flavors, relatively smooth sensation, and limited regulation in some settings. The resulting increase in vaping among adolescents has created new concerns. A chemical analysis of e-liquids examined the forms in which nicotine was present. In one prominent brand, sometimes described as the “iPhone of e-cigarettes,” the ratio of freebase nicotine to monoprotonated nicotine was unusually low. Freebase nicotine is the form most associated with throat irritation. Adjusting the pH so that less nicotine remains in the freebase form can therefore produce a smoother sensation, which may help explain the popularity of certain products.

Environmental-survey papers often struggle to connect measured concentrations with actual consequences. A study measured flame retardants in child car seats, but careful reading showed that the authors could not establish what level of toxicity or exposure would produce a health effect. Applying toxicity-prediction or exposure models can make a paper appear logically complete without solving that central problem.

In fact, flame retardants should generally be detectable in products such as child car seats. If they were absent, the bigger concern would be whether the seat could ignite easily during a fire. The broader issue is that so-called emerging contaminants are often replacements for substances that have been banned. Industry introduces a new product, researchers detect its chemicals, and the chemicals are labeled hazardous. But the deeper conflict is that the physical and chemical properties required to provide a useful function—such as flame resistance—may themselves create risks to people. Changing one product for another does not necessarily remove that underlying trade-off.

Sampling is the unglamorous part of environmental analysis, yet it may introduce the largest share of analytical uncertainty. Devices that combine sampling and extraction therefore remain useful. One study designed a visually elaborate, almost steampunk-like passive sampler by linking a spring, a syringe, and a solid-phase-extraction cartridge. The device operated for an extended period without external power. The principle is not new, but the combination offers considerable room for practical invention.

Mushrooms are common in cuisines around the world, but they can also accumulate arsenic. An analysis of commercially available mushroom varieties examined arsenic concentrations and their distribution. The authors estimated that inorganic arsenic in mushrooms could raise bladder- and lung-cancer risks at ordinary exposure levels. The statistical analysis, however, was not convincing, so the conclusion should be treated cautiously rather than accepted at face value.

As analytical tools improve, interpreting the resulting data becomes more difficult. The growth of different omics fields has created a need for interface software and integrated workflows. One study attempted to connect dose–response omics data with ecological risk assessment through a model-selection procedure. Environmental science currently contains two opposing tendencies: one builds walls between disciplines through specialized terminology, while the other tries to dismantle those walls and connect methods. The second approach is more productive. The first often turns scientific resources into an exercise in maintaining its own boundaries.