Understanding Plastic Pollution: The potential health effects, abundance and classification of microplastics

PLOS ONE recently published a new Collection of research entitled Recent Advances in Understanding Plastic Pollution. Given the broad scope of this collection, and the potential implications this research has on both humans the rest of the biosphere globally, we are digging deeper into the findings with some of the authors from papers included in this collection. In this third installment of interviews, we learn more about how microplastics may affect metabolism, and how it is getting easier to use machine learning to analyse samples containing microplastics.

---

CJ O’Brien, Plastics Campaign Associate, Oceana

CJ O’Brien’s paper in this collection: O’Brien CJ, Hong HC, Bryant EE, Connor KM (2021) The observation of starch digestion in blue mussel Mytilus galloprovincialis exposed to microplastic particles under varied food conditions. PLoS ONE 16(7): e0253802. https://doi.org/10.1371/journal.pone.0253802

PLOS: In this paper, you studied the effects of microplastics on blue mussel Mytilus galloprovincialis during different food regimes. Why is this species particularly interesting to study in order to understand plastic pollution?

CJO: Mytilus galloprovincialis are small but mighty in their importance to the marine ecosystem and to plastic pollution research. Many researchers study this species because they are bioindicators which means they help us monitor the overall health of the environment. Mytilus galloprovincialis filter feed and are sessile creatures, making them extremely sensitive to pollution and other anthropogenic changes. Studying this species and its physiological reaction to the exposure of microplastic allowed us as researchers to get a better look at how microplastics are not only impacting them as a species, but how microplastic might be impacting the ecosystem as a whole. 

Additionally, Mytilus galloprovincialis are crucial to the marine environment and to humans as well. This species is constantly filtering the water column in which they live, creating more clean environments for their marine neighbors. They are also found all over the world and are cultivated for food in many different regions. Not to mention they make great lab subjects as they are easy to care for. I would say that intertidal filter feeders in general are extremely fascinating organisms and crucial in our understanding of plastic pollution, the health of the ocean, and the health of humans. 

PLOS: You found that enzyme activity was affected by the presence of microplastics in the high-food regime only. Was this a result you had foreseen? How is the high-food regime reflected in the real lives of this species?

CJO: This outcome was shocking to me. I expected amylase activity to be negatively affected by the presence of microplastic in both feeding regimes. I thought that since microplastic holds no nutrition for these organisms, that filtering microplastic particles would take up a large proportion of their energy to filter, increase toxicity, or reduce available organic content available for digestion. Theoretically, these perturbations could hinder their ability to make or secrete amylase and survive. However, mussels evolved a range of digestive related characteristics to cope with fluctuations in nutrients and understanding how they modulate them when exposed to microplastic pollution is an emerging field of science.

In our experiment, we subjected mussels to fluctuating feeding environments that differ, similar to that to mussels at different shore levels. Mussels fed high food concentrations represented mussels that live lower in the water column and are exposed to more feeding options than mussels high on the shore due to daily tidal variation. With that context, I thought that the amylase activity in mussels in the low food group would be impacted more than mussels in the high food group. This inference was not observed and in fact high microplastics led to unpredictably high amylase activity.

This was interesting to me because food digestion is positively related to food abundance–the digestive modulation hypothesis–and microplastics is not food. Mussels are adapted to conserve energy as much as they can due to unpredictable environments, such as tidal, thermal, and pH variation. Any change to their energy reserves in nature could impact their growth, survival, and fitness. However, our study showed that it is possible that even under very high microplastic exposures and presumably less organic content ingested, amylase activity was actually increased to compensate for diluted food. 

PLOS: Working to combat plastic pollution must be endlessly inspiring but occasionally daunting. What motivated you to work in this field, and what are the rewards that keep you going?

CJO: Growing up in Florida, I’ve always had a deep curiosity and connection to the ocean. My motivation for getting into this field was fueled by wanting to protect the place that I loved most. I increasingly saw plastic pollution on beaches that I spent time at and as I started to learn more, I realized just how big this problem is. I was utterly fascinated that a man-made material, made to last forever but oftentimes only used for a few moments has caused so much harm–especially microplastic which can be microscopic. It is so insidious!

Currently, I work on policies that reduce the production and use of single-use plastic. While I don’t work in research anymore, I’ve seen firsthand how research influences policies that reduce single-use plastic. It is so crucial that researchers continue to investigate how this pollutant impacts the health of our oceans and the health of us as humans. Plastic production is expected to increase and if we are to have any chance in fighting the plastic pollution crisis, we will need all hands on deck from scientists, policymakers, as well as artists, musicians, community members, and young people. I feel hopeful when I see collaborative, creative, and equitable approaches to this problem.

PLOS: Several other studies in this Collection also look the effects of plastic pollution on living species. Has seeing these other research studies in the collection helped inspire any thoughts about future work you might do, or other advances your research community will make?

CJO: Our study subjected mussels to high concentrations of spherical microplastics that may have an effect on mussels in future microplastics conditions. Our results showed that these types of microplastics are not lethal over short exposures. I continue to monitor studies of microplastics on bivalves and other marine organisms in general in my role as the Plastics Campaign Associate. The Connor Lab at University of California-Irvine continues to deeply study how bivalves work from genome to phenome.

---

Ho-min Park, PhD Student, Ghent University

Ho-min Park’s paper in this collection: Park H-m, Park S, de Guzman MK, Baek JY, Cirkovic Velickovic T, Van Messem A, et al. (2022) MP-Net: Deep learning-based segmentation for fluorescence microscopy images of microplastics isolated from clams. PLoS ONE 17(6): e0269449. https://doi.org/10.1371/journal.pone.0269449

PLOS: You studied various machine learning techniques for annotating microplastics from fluorescence microscopy images, which is very promising for reducing the time and effort it takes researchers to analyze microscopy images. How close are we to where machine learning can truly analyze microscopy images as well as a human can?

HP: I think we are getting very close. For quite a few image analysis and annotation efforts that take up a lot of time, I even believe that machine learning techniques are already better than humans, given that humans tend to suffer from visual fatigue rather quickly. Furthermore, when targeting high-speed and high-quality image analyses, the ideal approach will most likely consist of first having machine learning analyze an image of interest, and then ask a domain expert to validate the analysis performed.

However, we still need to obtain a better understanding of the inherent limitations of data-driven approaches. Human-made data often contain biases and errors, and where these biases and errors can propagate to machine learning models that were trained on these human-made data. For example, while annotating our microscopy images, we were able to spot several image blobs that made it hard for humans to determine whether these blobs were denoting microplastics or light bleed artifacts, and where such ambiguities typically also affect the training and decision-making capabilities of machine learning models.

PLOS: You made all data and code publicly available for the software you developed for this project. What motivated you to do this? Do you know whether other researchers have used your code or software, maybe not yet for this project, but perhaps for any other code you’ve made available in the past?

HP: In imaging of microplastics, the acquisition of data requires several steps, and where most of these steps can be considered time-consuming and labor intensive, especially when they involve chemical processes. In particular, to obtain a set of microscopy images, we had to collect numerous clam samples, subsequently digesting the proteins and lipids, staining the remaining microplastics pieces, and performing image capturing with a microscope. As a result, most studies only make available the amount and the type of microplastics, and not the original images. However, this makes it challenging for other researchers to cross-validate experimental methods and results. We therefore took the decision to open up our data and our software, thus making it easier for other researchers to build on top of our work. In this respect, we also plan to post an introductory article on our work to the Papers with Code platform in the near future. Finally, although our paper was published only recently, we already received several inquiries regarding the usage of our data and our software.

PLOS: For this paper, you had two collaborating institutions and three “first authors” who contributed equally. Can you tell us more about how this collaboration worked?

HP: The idea of building a machine learning tool first came about when Maria Krishna, who is a PhD student in Food Chemistry at Ghent University Global Campus, encountered difficulties in manually counting microplastics in the fluorescence images she collected. After discussing these difficulties with me (Maria Krishna knew about my computer vision research), and after encouragement from our doctoral advisors, we decided to experiment with a few images and a number of deep learning models. This required a lot of work, both on the chemistry side (for the acquisition of microplastics from shellfish until image collection) and on the machine learning side (for model training and development of the GUI). In this context, we received a lot of help from two student interns, Sanghyeon Park and Jiyeon Baek, with Sanghyeon even staying on for the entire duration of the project.

PLOS: As a researcher, how do you hope to inspire other researchers, and the general public, to focus on plastic pollution as a social issue? What are some ways in which researchers who do not work directly in this field can help?

HP: With increasingly better methodologies to quantify microplastics pollution, including computational methodologies that leverage machine learning, we believe it will be easier to raise awareness about the seriousness of the spread of microplastics, and where this increased awareness will hopefully trigger more research and development efforts. These research and development efforts could for instance target the creation of biodegradable plastics, the discovery and possible engineering of organisms that can break down microplastics, and a better understanding of the risks posed by microplastics and their impact on human health, and where the latter effort would be of high interest to law and policy makers.

---

Cover image: Port of Dover, 2014 Beach Clean (CC-BY 2.0)

Disclaimer: Views expressed by contributors are solely those of individual contributors, and not necessarily those of PLOS.