Underwater hyperspectral classification of deep sea corals exposed to 2-methylnaphthalene.

Coral reefs around the world are under threat due to anthropogenic impacts on the environment. It is therefore important to develop methods to monitor the status of the reefs and detect changes in the health condition of the corals at an early stage before severe damage occur. In this work, we evaluate underwater hyperspectral imaging as a method to detect changes in health status of both orange and white color morphs of the coral species Lophelia pertusa. Differing health status was achieved by exposing 60 coral samples to the toxic compound 2-methylnaphthalene in concentrations of 0 mg L-1 to 3.5 mg L-1. A machine learning model was utilized to classify corals according to lethal concentration (LC) levels LC5 (5% mortality) and LC25 (25% mortality), solely based on their reflectance spectra. All coral samples were classified to correct concentration group. This is a first step towards developing a remote sensing technique able to assess environmental impact on deep-water coral habitats over larger areas.

Automated multivariate analysis of multi-sensor data submitted online: Real-time environmental monitoring.

A pilot study demonstrating real-time environmental monitoring with automated multivariate analysis of multi-sensor data submitted online has been performed at the cabled LoVe Ocean Observatory located at 258 m depth 20 km off the coast of Lofoten-Vesterålen, Norway. The major purpose was efficient monitoring of many variables simultaneously and early detection of changes and time-trends in the overall response pattern before changes were evident in individual variables. The pilot study was performed with 12 sensors from May 16 to August 31, 2015. The sensors provided data for chlorophyll, turbidity, conductivity, temperature (three sensors), salinity (calculated from temperature and conductivity), biomass at three different depth intervals (5-50, 50-120, 120-250 m), and current speed measured in two directions (east and north) using two sensors covering different depths with overlap. A total of 88 variables were monitored, 78 from the two current speed sensors. The time-resolution varied, thus the data had to be aligned to a common time resolution. After alignment, the data were interpreted using principal component analysis (PCA). Initially, a calibration model was established using data from May 16 to July 31. The data on current speed from two sensors were subject to two separate PCA models and the score vectors from these two models were combined with the other 10 variables in a multi-block PCA model. The observations from August were projected on the calibration model consecutively one at a time and the result was visualized in a score plot. Automated PCA of multi-sensor data submitted online is illustrated with an attached time-lapse video covering the relative short time period used in the pilot study. Methods for statistical validation, and warning and alarm limits are described. Redundant sensors enable sensor diagnostics and quality assurance. In a future perspective, the concept may be used in integrated environmental monitoring.

Integrated environmental monitoring and multivariate data analysis-A case study.

The present article describes integration of environmental monitoring and discharge data and interpretation using multivariate statistics, principal component analysis (PCA), and partial least squares (PLS) regression. The monitoring was carried out at the Peregrino oil field off the coast of Brazil. One sensor platform and 3 sediment traps were placed on the seabed. The sensors measured current speed and direction, turbidity, temperature, and conductivity. The sediment trap samples were used to determine suspended particulate matter that was characterized with respect to a number of chemical parameters (26 alkanes, 16 PAHs, N, C, calcium carbonate, and Ba). Data on discharges of drill cuttings and water-based drilling fluid were provided on a daily basis. The monitoring was carried out during 7 campaigns from June 2010 to October 2012, each lasting 2 to 3 months due to the capacity of the sediment traps. The data from the campaigns were preprocessed, combined, and interpreted using multivariate statistics. No systematic difference could be observed between campaigns or traps despite the fact that the first campaign was carried out before drilling, and 1 of 3 sediment traps was located in an area not expected to be influenced by the discharges. There was a strong covariation between suspended particulate matter and total N and organic C suggesting that the majority of the sediment samples had a natural and biogenic origin. Furthermore, the multivariate regression showed no correlation between discharges of drill cuttings and sediment trap or turbidity data taking current speed and direction into consideration. Because of this lack of correlation with discharges from the drilling location, a more detailed evaluation of chemical indicators providing information about origin was carried out in addition to numerical modeling of dispersion and deposition. The chemical indicators and the modeling of dispersion and deposition support the conclusions from the multivariate statistics. Integr Environ Assess Manag 2017;13:387-395. © 2016 SETAC.

Computational Visual Stress Level Analysis of Calcareous Algae Exposed to Sedimentation.

This paper presents a machine learning based approach for analyses of photos collected from laboratory experiments conducted to assess the potential impact of water-based drill cuttings on deep-water rhodolith-forming calcareous algae. This pilot study uses imaging technology to quantify and monitor the stress levels of the calcareous algae Mesophyllum engelhartii (Foslie) Adey caused by various degrees of light exposure, flow intensity and amount of sediment. A machine learning based algorithm was applied to assess the temporal variation of the calcareous algae size (∼ mass) and color automatically. Measured size and color were correlated to the photosynthetic efficiency (maximum quantum yield of charge separation in photosystem II, [Formula: see text]) and degree of sediment coverage using multivariate regression. The multivariate regression showed correlations between time and calcareous algae sizes, as well as correlations between fluorescence and calcareous algae colors.

Assessing the potential impact of water-based drill cuttings on deep-water calcareous red algae using species specific impact categories and measured oceanographic and discharge data.

The potential impact of drill cuttings on the two deep water calcareous red algae Mesophyllum engelhartii and Lithothamnion sp. from the Peregrino oil field was assessed. Dispersion modelling of drill cuttings was performed for a two year period using measured oceanographic and discharge data with 24 h resolution. The model was also used to assess the impact on the two algae species using four species specific impact categories: No, minor, medium and severe impact. The corresponding intervals for photosynthetic efficiency (ΦPSIImax) and sediment coverage were obtained from exposure-response relationship for photosynthetic efficiency as function of sediment coverage for the two algae species. The temporal resolution enabled more accurate model predictions as short-term changes in discharges and environmental conditions could be detected. The assessment shows that there is a patchy risk for severe impact on the calcareous algae stretching across the transitional zone and into the calcareous algae bed at Peregrino.

The effect of sediment mimicking drill cuttings on deep water rhodoliths in a flow-through system: Experimental work and modeling.

The impact of sediment coverage on two rhodolith-forming calcareous algae species collected at 100m water depth off the coast of Brazil was studied in an experimental flow-through system. Natural sediment mimicking drill cuttings with respect to size distribution was used. Sediment coverage and photosynthetic efficiency (maximum quantum yield of charge separation in photosystem II, ϕPSIImax) were measured as functions of light intensity, flow rate and added amount of sediment once a week for nine weeks. Statistical experimental design and multivariate data analysis provided statistically significant regression models which subsequently were used to establish exposure-response relationship for photosynthetic efficiency as function of sediment coverage. For example, at 70% sediment coverage the photosynthetic efficiency was reduced 50% after 1-2weeks of exposure, most likely due to reduced gas exchange. The exposure-response relationship can be used to establish threshold levels and impact categories for environmental monitoring.

Factorial design applied for multiple endpoint toxicity evaluation in Atlantic salmon (Salmo salar L.) hepatocytes.

The toxic equivalent (TEQ) approach is traditionally used in risk evaluation of dioxins. Non-dioxin-like PCBs are not included in this approach and TEQ can therefore underestimate toxicity. In this study, a factorial design and multiple endpoint strategy have been used to evaluate the combined toxicity and possible interactions between the non-dioxin-like PCB 138 and the potent AhR agonists 2,3,7,8-TCDF (TCDF) and 1,2,3,7,8-PeCDD (PCDD). Primary hepatocyte cultures from Atlantic salmon were exposed for 24h and qPCR was employed to create CYP1A dose-response curves and to quantify the transcriptional levels of eight genes (CYP1A, UDPGT, HSP70, GR, GPX, MnSOD, GST and p53). Principal component analysis (PCA) was used to evaluate response similarities between genes. PLS regression was used to model CYP1A and UDPGT responses to the three chemicals. The contour plot examinations of the CYP1A model indicated an antagonism between PCDD and TCDF and in the UDPGT model a possibly synergistic interaction between PCB 138 and PCDD. The results indicate that PCB 138, in combination with TCDF and PCDD, can contribute to the measured CYP1A and UDPGT responses. Using primary cell cultures, multivariate data analysis of qPCR data is shown to be a useful tool in toxicological studies. A multiple endpoints strategy can enhance the quality of risk evaluation of chemical compounds.

Relationship between composition and toxicity of motor vehicle emission samples.

In this study we investigated the statistical relationship between particle and semivolatile organic chemical constituents in gasoline and diesel vehicle exhaust samples, and toxicity as measured by inflammation and tissue damage in rat lungs and mutagenicity in bacteria. Exhaust samples were collected from "normal" and "high-emitting" gasoline and diesel light-duty vehicles. We employed a combination of principal component analysis (PCA) and partial least-squares regression (PLS; also known as projection to latent structures) to evaluate the relationships between chemical composition of vehicle exhaust and toxicity. The PLS analysis revealed the chemical constituents covarying most strongly with toxicity and produced models predicting the relative toxicity of the samples with good accuracy. The specific nitro-polycyclic aromatic hydrocarbons important for mutagenicity were the same chemicals that have been implicated by decades of bioassay-directed fractionation. These chemicals were not related to lung toxicity, which was associated with organic carbon and select organic compounds that are present in lubricating oil. The results demonstrate the utility of the PCA/PLS approach for evaluating composition-response relationships in complex mixture exposures and also provide a starting point for confirming causality and determining the mechanisms of the lung effects.

Toxicological evaluation of complex mixtures: fingerprinting and multivariate analysis.

The present paper describes a strategy for toxicological evaluation of complex mixtures based on chemical "fingerprinting" followed by pattern recognition (multivariate data analysis). The purpose is to correlate chemical fingerprints to measured toxicological endpoints, identify all major contributors to toxicity, and predict toxicity of additional mixtures. The strategy is illustrated with organic extracts of exhaust particles which are characterized by full scan gas chromatography-mass spectrometry (GC-MS). The complex GC-MS data are resolved into peaks and spectra for individual compounds using an automated curve resolution procedure. Projections to latent structures (PLS) is used for the regression modeling to correlate the GC-MS data to the measured responses; mutagenicity in the Ames Salmonella assay. The regression model identifies those peaks that co-vary with the observed mutagenicity. These peaks may be identified chemically from their spectra. Furthermore, the regression model can be used to predict mutagenicity from GC-MS chromatograms of additional samples.

Supplemental role of the Ames mutation assay and gap junction intercellular communication in studies of possible carcinogenic compounds from diesel exhaust particles.

This study presents a new strategy for the carcinogenic evaluation of complex chemical mixtures based on genotoxic and nongenotoxic assays. We studied the ability of organic extracts of diesel exhaust particles (DEP) to induce point mutations in five different Salmonella typhimurium strains (Ames test) and to inhibit gap junction intercellular communication (GJIC) in rat liver epithelial cell lines. A crude extract of DEP was prepared by extraction with dichloromethane (DCM), and fractionated according to polarity into five fractions: aliphatic hydrocarbons, polycyclic aromatic hydrocarbons (PAH), nitro-PAH, dinitro-PAH, and polar compounds. Statistical experimental design, multivariate data analysis, and modeling were used to quantify the mutagenicity of individual and combined DEP fractions in the Ames assay. Quantitative determination of GJIC was carried out using a recently described combination of scrape loading and digital image analysis. Both assays responded to the DEP extract, but the responses were due to different fractions. The nitro-PAH fraction showed the strongest mutagenic potential, followed by the dinitro-PAH fraction. The effect on GJIC was due to the fraction containing the polar components, followed by the dinitro-PAH fraction. The extract was found to induce both basepair substitutions and frameshift mutations, through activation by bacterial nitroreductases. Hyperphosphorylation of connexin43, the major connexin in the epithelial cell lines, was less evident for DEP extract than for other communication inhibitors such as phorbol esters and growth factors, and consequently inhibitors of the protein kinase C (PKC) and mitogen-activated protein (MAP) kinase pathway were unable to counteract the inhibition by DEP extract. Since the Ames test is a well accepted method to screen for substances with genotoxic activity while inhibition of GJIC is associated with effect of tumor promoters and nongenotoxic carcinogens, it is not surprising but encouraging and interesting that the present data indicate that the two endpoints supplement each other as screening tests and in the evaluation of hazardous compounds in complex mixtures.

Toxicological evaluation of complex mixtures by pattern recognition: correlating chemical fingerprints to mutagenicity.

We describe the use of pattern recognition and multivariate regression in the assessment of complex mixtures by correlating chemical fingerprints to the mutagenicity of the mixtures. Mixtures were 20 organic extracts of exhaust particles, each containing 102-170 individual compounds such as polycyclic aromatic hydrocarbons (PAHs), nitro-PAHs, oxy-PAHs, and saturated hydrocarbons. Mixtures were characterized by full-scan GC-MS (gas chromatography-mass spectrometry). Data were resolved into peaks and spectra for individual compounds by an automated curve resolution procedure. Resolved chromatograms were integrated, resulting in a predictor matrix that was used as input to a principal component analysis to evaluate similarities between mixtures (i.e., classification). Furthermore, partial least-squares projections to latent structures were used to correlate the GC-MS data to mutagenicity, as measured in the Ames Salmonella assay (i.e., calibration). The best model (high r2 and Q2) identifies the variables that co-vary with the observed mutagenicity. These variables may subsequently be identified in more detail. Furthermore, the regression model can be used to predict mutagenicity from GC-MS chromatograms of other organic extracts. We emphasize that both chemical fingerprints as well as detailed data on composition can be used in pattern recognition.