Periodic Reporting for period 1 - PLASTICSCAN (Combining CT Scanning and Spectroscopy to Characterise Micro-Plastics, a Growing Threat to Seabirds and other Marine Animals)
Okres sprawozdawczy: 2022-02-01 do 2024-01-31
Plastic waste is now a major source of pollution in the world’s oceans and is often eaten by wildlife that mistake plastic for food. The flesh-footed shearwater is one species of seabird that is particularly affected: foraging adults routinely feed bits of plastic to their chicks, which leads to poor health, stunted development, and even death. This project has focused on understanding exactly what about plastic is causing the most harm to shearwaters, is it simply the amount of plastic, or is it something about the size or shape, or what they are made of? To answer that question, we need a large quantity of plastics from the digestive tracts of several birds, we can then carefully record and analyse the various properties of those plastics and relate those properties to measurements of how healthy each bird was at the time. I set out to analyse 1000 pieces of ingested plastic from flesh-footed shearwaters, and I have collected data on over 1400 plastics. Although the analysis is not finished yet, I should be able to identify which properties of plastic (if any) are having a significant effect on seabird health.
Another challenge we face when monitoring plastic pollution and its impact on wildlife is making sure we can accurately and reliably measure the physical and chemical properties of plastics. As part of this project, I set out to develop several methods for collecting data on the properties of plastics, including photography, micro-CT scanning, and spectroscopy. Some of these techniques are already well known and often used for studying plastic pollution but can be improved in reliability by applying machine learning and other automated data processing methods, while other methods I have developed are entirely new and represent a significant improvement over conventional, human-driven observations.
Another challenge we face when monitoring plastic pollution and its impact on wildlife is making sure we can accurately and reliably measure the physical and chemical properties of plastics. As part of this project, I set out to develop several methods for collecting data on the properties of plastics, including photography, micro-CT scanning, and spectroscopy. Some of these techniques are already well known and often used for studying plastic pollution but can be improved in reliability by applying machine learning and other automated data processing methods, while other methods I have developed are entirely new and represent a significant improvement over conventional, human-driven observations.
I received 1453 samples of ingested material taken from the digestive tracts of 76 flesh-footed shearwaters as part of a previous fieldwork study by my colleagues. I have analysed the size, shape, colour, weight, volume, and surface area of more than 1400 samples, and the chemical composition of 246 samples, with another 750 samples planned for chemical analysis in the months following the end of the project. Once completed, this dataset of 1000+ samples will be analysed to determine what properties of plastic are significantly impacting seabird health.
As part of this project, I have developed and tested several analytical methods for measuring the properties of plastic. In collaboration with researchers at the University of Tasmania and Oxford Brookes University I devised a novel method using photography and automated image analysis to measure the size, shape, and colour of ingested plastics. We showed that the photography method is faster and more reliable than humans for counting objects and provides far more accurate information. I have also described how Raman and infrared spectroscopy can be used to measure chemical composition and identify different types of plastic. I have shown that ingested plastics can be difficult to identify because of significant contamination by biological material, but machine learning algorithms can be trained to reliably identify plastics, even those with substantial contamination. I have also analysed samples in bulk with micro-CT scanning, creating 3D models to calculate volumes and surface areas for plastic on a bird-by-bird basis, which I intend to compare to data from another technique, 3D laser scanning, shortly after the end of the project.
Finally, I have started a 6-month laboratory experiment to see how pristine plastics (made from polyethylene, polypropylene, and polystyrene) change over time under different environmental conditions, including exposure to seawater and/or ultraviolet light. Some samples are already substantially changed in terms of colour and density, indicating substantial degradation after just a few months. Once completed, I will determine whether changes in spectra can be used to estimate how long plastic has been in the ocean.
As part of this project, I have developed and tested several analytical methods for measuring the properties of plastic. In collaboration with researchers at the University of Tasmania and Oxford Brookes University I devised a novel method using photography and automated image analysis to measure the size, shape, and colour of ingested plastics. We showed that the photography method is faster and more reliable than humans for counting objects and provides far more accurate information. I have also described how Raman and infrared spectroscopy can be used to measure chemical composition and identify different types of plastic. I have shown that ingested plastics can be difficult to identify because of significant contamination by biological material, but machine learning algorithms can be trained to reliably identify plastics, even those with substantial contamination. I have also analysed samples in bulk with micro-CT scanning, creating 3D models to calculate volumes and surface areas for plastic on a bird-by-bird basis, which I intend to compare to data from another technique, 3D laser scanning, shortly after the end of the project.
Finally, I have started a 6-month laboratory experiment to see how pristine plastics (made from polyethylene, polypropylene, and polystyrene) change over time under different environmental conditions, including exposure to seawater and/or ultraviolet light. Some samples are already substantially changed in terms of colour and density, indicating substantial degradation after just a few months. Once completed, I will determine whether changes in spectra can be used to estimate how long plastic has been in the ocean.
Despite rising public awareness of the potential harm of plastic pollution to marine ecosystems and wildlife, there is still limited research and little consensus on what mechanisms are most important when determining how ingesting plastic harms wildlife. Individual studies have shown that microplastics can directly damage tissue and cause disease (named ‘plasticosis’), that plastics can release toxic chemicals and pollutants such as mercury, and that plastics may simply cause malnutrition by replacing nutritious food. This project presents a broad analysis that considers several possible aspects of plastic ingestion, including numerical counts (total number and weight of plastic), their physical properties (e.g. size and shape) and chemical properties (e.g. the type of plastic) to see which may be producing the most significant impact on health. By analysing 1400+ pieces of plastic from 76 birds, ranging from birds that have ingested just 1 piece of ingested plastic to birds that have ingested 200 or more, I should be able to conduct a statistically meaningful assessment of plastic impact on seabird health. This analysis is expected to be completed about 8-10 months after the end of the project, pending collection of some remaining spectroscopy data. The results from that study will be highly pertinent to both conservation scientists and policy makers who need to understand exactly how plastics are impacting wildlife to influence how we produce, use, and dispose of plastic materials.
As part of the project, I have developed advanced new analytical methods for collecting data on the properties of plastic pollution, which so far represent a substantial improvement over current methods. The size, shape, and colour of individual pieces of plastic have typically been measured by hand or eye, which is time-consuming and highly subjective, but by using photography and automated image analysis I was able to more reliably and more accurately measure hundreds of pieces of plastic at a time. I have also shown that ingested plastics are highly contaminated with biological material, which makes identifying the type of plastic using conventional spectroscopy more difficult, but that machine learning algorithms can be used to reliably identify certain types of plastic despite contamination. The use of CT scanning and 3D laser scanning to do volumetry of plastics is highly novel and represents a new approach to studying plastic. Overall, the methodologies developed during this project represent a marked improvement in accuracy and reliability over current methods and are specifically designed to be accessible and applicable by other researchers studying plastics. Should these methods be put into action, they will improve the data yielded by future studies monitoring the properties of plastic pollution and make data more comparable between studies.
As part of the project, I have developed advanced new analytical methods for collecting data on the properties of plastic pollution, which so far represent a substantial improvement over current methods. The size, shape, and colour of individual pieces of plastic have typically been measured by hand or eye, which is time-consuming and highly subjective, but by using photography and automated image analysis I was able to more reliably and more accurately measure hundreds of pieces of plastic at a time. I have also shown that ingested plastics are highly contaminated with biological material, which makes identifying the type of plastic using conventional spectroscopy more difficult, but that machine learning algorithms can be used to reliably identify certain types of plastic despite contamination. The use of CT scanning and 3D laser scanning to do volumetry of plastics is highly novel and represents a new approach to studying plastic. Overall, the methodologies developed during this project represent a marked improvement in accuracy and reliability over current methods and are specifically designed to be accessible and applicable by other researchers studying plastics. Should these methods be put into action, they will improve the data yielded by future studies monitoring the properties of plastic pollution and make data more comparable between studies.