August has seen a range of exciting research published across MDPI’s Open Access journals.

In this roundup, we highlight research exploring how microbial inoculants can reduce reliance on phosphate fertilisers in sandy soils. We also feature insights into the effects of smoking electronic cigarettes (E-cigs) on a newborn’s stem cells via a mouse model, alongside a study uncovering how juvenile pygmy raccoons may learn behaviour in human-modified environments. Finally, we explore research revealing how large language models such as ChatGPT are still a long way from being able to accurately predict medical outcomes.

Can Beneficial Microorganisms Reduce Phosphorus Application Rates in Common Bean (Phaseolus vulgaris L.) Cultivation in Sandy Soils?

Published in Plants, August 2026 | São Paulo State University (UNESP), Department of Plant Health, Department of Plant Protection

Plants use phosphorus in nearly all biological processes, either directly as a constituent of organic compounds or indirectly through ATP-mediated reactions. Unlike carbon and nitrogen, phosphorus lacks any significant gaseous phase and mainly becomes available to plants from the weathering of rocks where only a small fraction is available to plants at any time.

Phaseolus vulgaris L. has a shallow root system, which can limit its access to water and nutrients. Improving root efficiency through microbial inoculants could therefore support more sustainable agricultural practices, contributing to SDG 2: Zero Hunger, which promotes sustainable agriculture.

This limited availability means modern agriculture relies heavily on phosphate fertilisers. However, much of the phosphate added to soil quickly becomes chemically bound into forms that plants cannot absorb.

Phosphate-solubilising microbes could offer a way to improve phosphorus use. These microorganisms release phosphorus from insoluble forms in the soil, making it more available to plants and potentially reducing the amount of fertiliser required.

Beneficial microorganisms can function as biostimulants, through promoting plant root development, and as biofertilisers, by solubilising phosphorus in the soil into forms plants can absorb. Therefore, the use of microbial inoculants can contribute to more sustainable agricultural systems.

Researchers in Brazil explored this approach by inoculating common bean plants (Phaseolus vulgaris L.) with the fungus Trichoderma harzianum and the bacterium Bacillus amyloliquefaciens. The study was conducted in sandy soil, where common beans are particularly sensitive to soil conditions because of their shallow and poorly developed root systems.

Prof. Fernando Galindo, corresponding author of the study explains:

Inoculation with Trichoderma harzianum and/or Bacillus amyloliquefaciens enhanced plant growth, root development, nutrient accumulation, and grain yield compared with non-inoculated plants. Importantly, microbial inoculation enabled a 40% reduction in the recommended P fertilization rate without compromising crop productivity under greenhouse conditions. These findings highlight the potential of microbial-based strategies to improve phosphorus use efficiency and reduce dependence on mineral P fertilizers, contributing to more sustainable and regenerative agricultural systems.

The findings suggest that microbial inoculants could help crops make better use of the phosphorus already present in soil, potentially reducing the amount of phosphate fertiliser needed.

You can read more about microbial inoculants in Exploring the Field: Soil Science.

Maternal E-Cigarette Vaping Drives Persistent Reprogramming of Bone Marrow Hematopoietic and Mesenchymal Stem Cells and Promotes Transcriptional and Metabolic Dysregulation-Associated Inflammaging and Disease Risks in Rat Offspring

Published in Cells, August 2026 | Department of Medicine, School of Medicine, Loma Linda University

Although conventional tobacco smoking has declined in recent decades, the use of electronic nicotine delivery systems, such as electronic cigarettes (E-cigs), has increased substantially. Despite being publicly perceived as a safer alternative to tobacco cigarettes, the long-term biological effects of E-cigs remain largely unknown.

As they are often perceived as being safer than smoking tobacco, there is evidence showing that E-cig use during pregnancy is not uncommon.

Further, the use of E-cigs among women of a reproductive age continues to rise. This raises additional concerns regarding the effect of maternal E-cig vaping on offspring health. A particular concern is the effects that in utero exposure to toxins from vaping have on offspring stem cells and blood development.

Bone marrow stem cells are essential for maintaining health throughout life. Hematopoietic stem cells continuously produce blood and immune cells, while mesenchymal stem/stromal cells help maintain the bone marrow environment and support bone formation. Because these cells are responsible for continual tissue renewal and repair, changes to their development or function early in life could have consequences that persist into adulthood.

Researchers in California, USA, used a rat model of maternal E-cig exposure during gestation alongside complementary ex vivo studies of human cells to assess the effect of nicotine containing E-cig exposure on offspring stem cell function. The results highlight that the potential health consequences of maternal E-cig vaping may extend far beyond pregnancy.

Dr Xu, corresponding author of the study explains:

Using a longitudinal animal model, we found that prenatal e-cigarette exposure was associated with persistent reprogramming of the offspring bone marrow, affecting hematopoietic and mesenchymal stem/stromal cell populations, inflammatory and immune homeostasis, and bone-forming capacity from early life into adulthood.

Together, these findings suggest that maternal E-cig exposure could have long-term effects on offspring bone marrow, immune, and skeletal function. Further research is needed to determine the underlying mechanisms and whether similar effects occur in humans. However, Dr Xu believes that the study provides much-needed evidence of the potential effects of vaping during pregnancy:

These findings raise important questions about how maternal vaping may shape the long-term hematopoietic, skeletal, and immune health of offspring and underscore the need for greater awareness of e-cigarette exposure during pregnancy.

Preliminary Evidence of Material Culture in the Critically Endangered Pygmy Raccoon (Procyon pygmaeus)

Published in Wild, August 2026 | Department of Biology, Miami University

Researchers in Mexico and the USA observed a previously undocumented behaviour in the critically endangered pygmy raccoons (Procyon pygmaeus). Two juvenile siblings collected paper waste from a beach club bin, manipulated it with water and their paws, and used the resulting paper balls during play.

A juvenile pygmy raccoon making a paper ball.

The observations suggest that pygmy raccoons may incorporate human-made materials into their behaviour outside of foraging. This provides an insight into how individual raccoons interact with their surroundings, particularly in environments shaped by human activity.

The authors explain:

By documenting previously unseen behaviours in the critically endangered pygmy raccoon, this study adds to the understanding of how individual raccoons experience and interact with their surroundings. It also highlights how individuals in anthropogenic environments may incorporate human-created objects into their daily lives.

For a critically endangered species found only on the island of Cozumel, Mexico, understanding how pygmy raccoons respond to human-modified environments could also help inform conservation. Tourism sites provide access to food, waste, and other human-made materials, particularly during important stages of juvenile development. While these resources may lead to novel behaviours, they can also increase the risk of habituation, injury, illness, and other human-related threats.

By documenting a previously unreported behaviour, the study shows how observations of individual animals can add to our understanding of pygmy raccoons and their interactions with human environments.

However, the authors are keen to highlight the study’s relevance outside of biology and conservation:

This study has implications in multiple fields, not only Biology and Conservation as understanding behaviour has applications for raccoon rehabilitation and preservation within Cozumel’s tourism industry; but also, in Anthrozoology, Human-Animal Studies, and Anthropology. This behaviour expands the list of species that demonstrate social learning and culture; as such, it challenges long-standing beliefs in human exceptionalism.

Large Language Model Chatbots Cannot Reliably Calculate Clinical Risk Scores, A Comparative Accuracy Study

Published in Cardiovascular Medicine, August 2026 | Department of Cardiology, University and Hospital Fribourg

The European System for Cardiac Operative Risk Evaluation II (EuroSCORE II) is used to estimate the risk of death following cardiac surgery. The score combines 18 patient-specific factors, including age, heart function, medical conditions, and the type of procedure, to provide an individualised estimate of operative mortality. These calculations can help heart teams assess treatment options and support decisions about patient care.

Accurate risk calculations are therefore important when clinicians are weighing different treatment options. However, large language models (LLMs) are increasingly being used for medical questions and decision support, despite being designed primarily to generate text rather than perform precise mathematical calculations. This raises questions about whether they can reliably reproduce established clinical risk scores.

Researchers in Germany compared four publicly available LLM chatbots, ChatGPT, Claude, Gemini, and DeepSeek, to gold standard EuroSCORE II calculations for 105 patients. Each model was asked to calculate the score both from the clinical information alone and with the official calculation formula provided.

The large language models struggled with the calculations when working without the formula, with the best performing approach agreeing with the reference calculations in only around half of cases. Providing the formula improved the performance for most models, with ChatGPT achieving the highest agreement at 51.4%. However, this remained well below the study’s predefined threshold of 90% agreement, and systematic biases were observed across the different approaches.

The findings highlight a gap between the ability of LLMs to process medical information and their ability to perform reliable quantitative calculations. While providing the formula improved accuracy, the models were still not reliable enough to replace validated clinical risk calculators. For calculations that directly influence treatment decisions, the researchers recommend using dedicated, validated tools rather than relying on general-purpose LLM chatbots.

 

If you would like to keep reading, see the full list of MDPI Journals where you can discover the latest science, from breakthroughs in sustainable agriculture to advances in animal behaviour, clinical medicine, and more.