Principal Component Analysis Loading Plots Research Articles

Metabolomics-based component profiling of hard and semi-hard natural cheeses with gas chromatography/time-of-flight-mass spectrometry, and its application to sensory predictive modeling

Gas chromatography/time-of-flight mass spectrometry (GC/TOF-MS) was used to analyze hydrophilic low molecular weight components, including amino acids, fatty acids, amines, organic acids, and saccharides, in cheese, and the sensometric application for practical metabolomic studies in the food industry is described. Derivatization of target analytes was conducted prior to the GC/TOF-MS analysis. Data on 13 cheeses, six Cheddar cheeses, six Gouda cheeses and one Parmigiano-Reggiano cheese, were analyzed by multivariate analysis. The uniqueness of the Parmigiano-Reggiano cheese metabolome was revealed. Principal component analysis (PCA) showed no grouping of the Cheddar cheeses and Gouda cheeses according to production method or country of origin. The PCA loading plot confirms that many amino acids contribute positively to PC1, suggesting that PC1 is closely related to degradation of proteins, and that lactic acid contributed positively to PC2, whereas glycerol contributed negatively to PC2, suggesting that factors regarding degradation of carbohydrates and fats were expressed in PC2. Partial least squares (PLS) regression models were constructed to predict the relationship between the metabolite profile and two sensory attributes, "Rich flavor" and "Sour flavor", which were related to maturation. The compounds that play an important role in constructing each sensory prediction model were identified as 12 amino acids and lactose for "Rich flavor", and 4-aminobutyric acid, ornithine, succinic acid, lactic acid, proline and lactose for "Sour flavor". The present study revealed that metabolomics-based component profiling, focusing on hydrophilic low molecular weight components, was able to predict the sensory characteristics related to ripening.

Multisensor system based on bisphthalocyanine nanowires for the detection of antioxidants

Electrophoretic deposition has been used to prepare thin films based on nanowires of three lanthanoid bisphthalocyaninates (including dysprosium, gadolinium and lutetium). Nanowires of similar structural characteristics have been obtained for the three compounds by tuning the electrophoretic conditions according to the redox properties of each phthalocyanine. The three electrodes have been used to form an array of sensors that has been employed to discriminate phenolic antioxidants of interest in the food industry including caffeic, gallic, vanillic and ferulic acids. The Principal Component Analysis (PCA) and the Partial Least Squares Discriminant Analysis (PLS-DA) of the electrochemical signals has allowed a clear discrimination of the four phenols analyzed according to the number of phenolic groups attached to the structure (monophenol, diphenol or triphenol). The PCA loading plots indicate that the three electrodes bring complementary information facilitating the discrimination of the studied solutions. In addition, good correlations between the intensity of the redox processes observed in the electrodes and the concentration of phenolic compounds have been found with detection limits in the range of 10−5–10−6molL−1 and good reproducibility.The fast preparation of these nanowires based films and their excellent performance offer a new sensing platform for the detection of antioxidants in a fast, reliable way.

Polydiacetylene paper-based colorimetric sensor array for vapor phase detection and identification of volatile organic compounds

Detection and identification of VOCs in their vapor phase is essential for safety and quality assessment. In this work, a novel platform of a paper-based polydiacetylene (PDA) colorimetric sensor array is prepared from eight diacetylene monomers, six of which are amphiphilic and the other two are bolaamphiphilic. To fabricate the sensors, monomers are coated onto a filter paper surface using the drop-casting technique and converted to PDAs by UV irradiation. The PDA sensors show solvent induced irreversible color transition upon exposure to VOC vapors. When combined into a sensing array, the color change pattern as measured by RGB values and statistically analyzed by principal component analysis (PCA) is capable of distinguishing 18 distinct VOCs in the vapor phase. The PCA score and loading plots also allow the reduction of the sensing elements in the array from eight to three PDAs that are capable of classifying 18 VOCs. Utilizing an array containing only two PDAs, various types of automotive fuels including gasoline, gasohol and diesel are successfully classified.

Chemometric analysis of gas chromatographic data-investigation of enological parameters of a bag-in-box white wine as affected by storage time and temperature

In this study, a bag-in-box white wine was stored at 22, 35, and 45 °C for up to 48 days to produce a series of samples that exhibited different enological parameters (absorbance at 420 nm, free SO2, total SO2, total phenol, and total aldehyde). Wine samples were extracted with dichloromethane and analyzed using gas chromatography (GC) to generate volatile fingerprints. Principal component analysis (PCA) score plots of the first three principal components showed grouping trends that were influenced by storage time and temperature. PCA loading plots revealed that changes in chemical profiles were different for wines held at different storage temperatures. Storage time could be predicted accurately by partial least squares (PLS) regression of the GC data. Coefficients of determination (R2) were >0.99, and the standard error of prediction values were 0.4, 0.5, and 1.9 days over the test period of 15, 30, and 48 days, respectively. Using the same GC data with PLS analyses, the enological parameters could be accurately predicted from GC fingerprints, except for the predictions of SO2 in a wine stored at 22 °C and total phenol in a wine stored at 45 °C. Copyright © 2011 John Wiley & Sons, Ltd.

Metabolite profiling and fingerprinting of commercial cultivars of Humulus lupulus L. (hop): a comparison of MS and NMR methods in metabolomics

Hop (Humulus lupulus L. Cannabaceae) is an economically important crop. In addition to its role in beer brewing, its pharmaceutical applications have been of increasing importance in recent years. Bitter acids (prenylated polyketides), prenylflavonoids and essential oils, are the primary phytochemical components that account for hop medicinal value. An integrated approach utilizing nuclear magnetic resonance (NMR) and mass spectrometry (MS) techniques was used for the first large-scale metabolite profiling in Humulus lupulus. Resins and extracts prepared from 13 hop cultivars were analysed using NMR, liquid chromatography (LC)-MS and fourier transform ion cyclotron resonance (FTICR)-MS in parallel and subjected to principal component analysis (PCA). A one pot extraction method, compatible with both MS and NMR measurement was developed to help rule out effects due to differences in extraction protocols. Under optimised conditions, we were able to simultaneously quantify and identify 46 metabolites including 18 bitter acids, 12 flavonoids, 3 terpenes, 3 fatty acids and 2 sugars. Cultivars segregation in PCA plots generated from both LC-MS and NMR data were found comparable and mostly influenced by differences in bitter acids composition among cultivars. FTICR-MS showed inconsistent PCA loading plot results which are likely due to preferential ionisation and also point to the presence of novel isoprenylated metabolites in hop. This comparative metabolomic approach provided new insights for the complementariness and coincidence for these different technology platform applications in hop and similar plant metabolomics projects.

Potential 1H NMR urinary metabolomic biomarkers for Lung cancer in hamster model

The objective of this study was to use the metabolomic approach to investigate early potential biomarkers for lung cancer. Golden Syrian hamsters were randomized into 2 groups. Lung cancer was induced in treatment group by a chemical carcinogen, N‐nitrosobis‐2‐(oxopropyl) amine (BOP). Proton nuclear magnetic resonance (1H NMR) spectra, acquired using 500 MHz magnet were processed using ACD labs software. The processed, digitized NMR spectral data was analyzed using multivariate data analysis software, SIMCA P+. Principal Component Analysis (PCA) score plots showed significant separation between the cancer and control groups. The metabolites responsible for the differences between the two groups as observed from the PCA loading plots, were quantified using CHENOMX NMR metabolite database. The metabolites (n=26) found to be significantly different in the two groups were traced back to their respective biochemical pathways in an effort to gain mechanistic insight into lung cancer progression from the metabolite view point. Some of these metabolites such as glycine, sarcosine, and alanine have been previously reported to be increased in other cancer types. These could potentially serve as potential biomarkers of cancer in general. In addition, some metabolites may be of specific relevance to progression of lung cancer and may therefore be further explored as possible biomarkers of this aggressive cancer type.Grant Funding Source: This work was funded by start up funds from Wayne State University.

Metabolomics Study on Quality Control and Discrimination of Three Curcuma Species based on Gas Chromatograph–Mass Spectrometry

Metabonomic analysis is an important molecular phenotyping method for characterising plant ecotypic variations; hence, it may become a powerful tool for quality control and discrimination of traditional Chinese medicine (TCM). To discriminate and assess the quality of Curcuma phaeocaulis, C. kwangsiensis and C. wenyujin from different ecotypes. The identification of the compositions of essential oils from the three Curcuma species was included in this study. Metabolomics analysis was carried out on all samples by gas chromatography-mass spectrometry (GC-MS) coupled with multivariate statistical analysis. Characterisation of phytochemicals in essential oils was performed by automated matching to the MS library and comparison of their mass spectra (NIST05 database). Principal component analysis (PCA) effectively distinguished the samples from different species and ecotypes. Partial least squares discrimination analysis (PLS-DA) was successfully employed in classifying the GC-MS data of authentic, commercial and introduction cultivation samples. Furthermore, the components contributing significantly to the discrimination, namely curzerenone, germacrone, curdione and epicurzerenone, were screened by PCA and PLS-DA loading plots and further can be used as chemical markers for discrimination and quality control among different groups of samples.

Dependence of forage quality and methanogenic potential of tropical plants on their phenolic fractions as determined by principal component analysis

This study was conducted to elucidate relationships among various phenolic fractions in, and methane (CH4) emissions from, tropical plants when incubated in ruminal fluid in vitro. As a second objective, principal component analysis (PCA) was tested for its utility in screening plants for their ability to reduce CH4 formation at simultaneously acceptable nutritional quality. Leaves from 27 tropical plants were analyzed for their nutritional composition and various phenolic fractions. They were incubated in vitro using the Hohenheim gas test method. Variables measured after 24h of incubation were total gas and CH4 production, and pH, ammonia, bacterial and protozoal counts, as well as short-chain fatty acids in the incubation fluid. In vitro organic matter (OM) digestibility was computed by a standard equation. The data obtained was subjected to analysis of variance, correlation, regression and PCA. Among phenolic fractions, total phenols had the closest relationship with CH4/digestible OM (r=−0.84, P<0.001). The total tannin fraction contributed strongly to this effect (r=−0.74, P<0.001) whereas the non-tannin phenol fraction was less important (r=−0.45, P<0.05). Methane reduction by the influence of non-tannin phenols was not associated with a negative effect on protein degradation, while this was the case with tannins. Condensed (r=−0.60, P<0.01) and hydrolysable tannins (r=−0.60, P<0.01) contributed to the decrease in CH4/digestible OM. The loading plot of PCA showed that dietary crude protein (CP) content and incubation fluid ammonia, total short-chain fatty acids, propionate, valerate, iso-butyrate, iso-valerate as well as in vitro OM digestibility were clustered. They had inverse directions to contents of fiber fractions and incubation fluid acetate proportion and acetate-to-propionate ratio. The methane-to-total-gas ratio had the opposite effect of the contents of any phenolic fraction. Plants possessing a favorable forage quality, based on the corresponding PCA score plot, were Carica papaya, Manihot esculenta, Morinda citrifolia, Sesbania grandiflora and Melia azedarach, whereas CH4 mitigating plants included Swietenia mahagoni, Acacia villosa, Eugenia aquea, Myristica fragrans and Clidemia hirta. All phenolic fractions studied reduced CH4 emissions from in vitro incubations with ruminal fluid and PCA seems useful to screen plants for high nutritional quality and low ruminal CH4 formation. However, high forage quality seemed to be partially associated with high CH4 emission. The search for plants rich in non-tannin phenols might be promising as these compounds appear to decrease CH4 while they obviously have less negative effect on protein degradation as compared to the tannin fractions.

Discrimination analysis of human lung cancer cells associated with histological type and malignancy using Raman spectroscopy

The Raman spectroscopic technique enables the observation of intracellular molecules without fixation or labeling procedures in situ. Raman spectroscopy is a promising technology for diagnosing cancers-especially lung cancer, one of the most common cancers in humans-and other diseases. The purpose of this study was to find an effective marker for the identification of cancer cells and their malignancy using Raman spectroscopy. We demonstrate a classification of cultured human lung cancer cells using Raman spectroscopy, principal component analysis (PCA), and linear discrimination analysis (LDA). Raman spectra of single, normal lung cells, along with four cancer cells with different pathological types, were successfully obtained with an excitation laser at 532 nm. The strong appearance of bands due to cytochrome c (cyt-c) indicates that spectra are resonant and enhanced via the Q-band near 550 nm with excitation light. The PCA loading plot suggests a large contribution of cyt-c in discriminating normal cells from cancer cells. The PCA results reflect the nature of the original cancer, such as its histological type and malignancy. The five cells were successfully discriminated by the LDA.

Evidence of vintage effects on grape wines using 1H NMR-based metabolomic study

The chemical composition of grape wines varies with grape variety, environmental factors of climate and soil, and bacterial strains, which can each affect the wine quality. Using 1H NMR analysis coupled with multivariate statistical data sets, we investigated the effects of grape vintage on metabolic profiles of wine and the relationship between wine metabolites and meteorological data. Principal component analysis (PCA) showed a clear differentiation between Meoru wines that were vinified with the same yeast strain and Meoru grapes harvested from the same vineyard but with a different vintage. The metabolites contributing to the differentiation were identified as 2,3-butandiol, lactic acid, alanine, proline, γ-aminobutyric acid (GABA), choline, and polyphenols, by complementary PCA loading plot. Markedly higher levels of proline, lactic acid and polyphenols were observed in the 2006 vintage wines compared to those of 2007 vintage, showing excellent agreement with the meteorological data that the sun-exposed time and rainfall in 2006 were approximately two times more and four times less, respectively, than those in 2007. These results revealed the important role of climate during ripening period in the chemical compositions of the grape. This study highlights the reliability of NMR-based metabolomic data by integration with meteorological data in characterizing wine or grape.

Time-Dependent Changes of Oxytocin Using 1H-NMR Coupled with Multivariate Analysis: a New Approach for Quality Evaluation of Protein/Peptide Biologic Drugs

A new method that combines (1)H-NMR and principal component analysis (PCA) was employed to obtain the quality evaluation of biopharmaceuticals, with regard to their quality, consistency, and differences in protein modification patterns. To assess the feasibility of the method, three (1)H-NMR spectra of oxytocin (OXT) were collected every 7 d (at Day 0, 7 and 14), and time-dependent changes in the spectra were found by PCA of the (1)H-NMR signals from 0.5-9.0 ppm, excluding the region around the water signal (4.6-5.0 ppm). Although the three OXT spectra seemed similar by simple visual inspection, time-dependent differences among the three spectra were clearly distinguished by a PCA scores plot. Peak changes indicating both OXT decomposition and the emergence of new OXT decomposition products within the timeframe of the experiment were also observed by a PCA loading plot. The results demonstrate that this method can evaluate the consistency of biopharmaceutical quality.

Shedding new light on the molecular architecture of oocytes using a combination of synchrotron Fourier transform-infrared and Raman spectroscopic mapping.

Synchrotron Fourier transform-infrared (FT-IR) and Raman microspectroscopy were applied to investigate changes in the molecular architecture of mouse oocytes and demonstrate the overall morphology of the maturing oocyte. Here we show that differences were identified between immature mouse oocytes at the germinal vesicle (GV) and mature metaphase II (MII) stage when using this technology, without the introduction of any extrinsic markers, labels, or dyes. GV mouse oocytes were found to have a small, centrally located lipid deposit and another larger polar deposit of similar composition. MII oocytes have very large, centrally located lipid deposits. Each lipid deposit for both cell types contains an inner and outer lipid environment that differs in composition. To assess interoocyte variability, line scans were recorded across the diameter of the oocytes and compared from three independent trials (GV, n = 91; MII, n = 172), and the data were analyzed with principal component analysis (PCA). The average spectra and PCA loading plots show distinct and reproducible changes in the CH stretching region that can be used as molecular maturation markers. The method paves the way for developing an independent assay to assess oocyte status during maturation providing new insights into lipid distribution at the single cell level.

1H Nuclear Magnetic Resonance-Based Metabolomic Characterization of Wines by Grape Varieties and Production Areas

(1)H NMR spectroscopy was used to investigate the metabolic differences in wines produced from different grape varieties and different regions. A significant separation among wines from Campbell Early, Cabernet Sauvignon, and Shiraz grapes was observed using principal component analysis (PCA) and partial least squares-discriminant analysis (PLS-DA). The metabolites contributing to the separation were assigned to be 2,3-butanediol, lactate, acetate, proline, succinate, malate, glycerol, tartarate, glucose, and phenolic compounds by PCA and PLS-DA loading plots. Wines produced from Cabernet Sauvignon grapes harvested in the continental areas of Australia, France, and California were also separated. PLS-DA loading plots revealed that the level of proline in Californian Cabernet Sauvignon wines was higher than that in Australian and French Cabernet Sauvignon, Australian Shiraz, and Korean Campbell Early wines, showing that the chemical composition of the grape berries varies with the variety and growing area. This study highlights the applicability of NMR-based metabolomics with multivariate statistical data sets in determining wine quality and product origin.

1H NMR and GC/MS metabolomics of earthworm responses to sub-lethal DDT and endosulfan exposure

The metabolic response of the earthworm Eisenia fetida to two pesticides, dichlorodiphenyltrichloroethane (DDT) and endosulfan, was characterized in contact tests using proton nuclear magnetic resonance (1H NMR) and principal component analysis (PCA). PCA loading plots suggested that maltose, leucine and alanine were important metabolites contributing to the differences in dosed and control earthworms for both compounds at doses of 0.5, 1.0 and 2.0 μg/cm2. Gas chromatography/mass spectrometry (GC/MS) was used to quantify the metabolites identified in E. fetida and determine if the changes in maltose, leucine and alanine following exposure to DDT and endosulfan (at 0.5 and 1.0 μg/cm2) were reproducible and greater than the natural variability. Quantification by GC/MS suggested that maltose was not a reliable biomarker since it both increased and decreased in earthworms exposed to DDT and increased by just 3% with exposure to endosulfan. Leucine was not stable with the GC/MS derivitization method used in this study and could not be confirmed as a reliable biomarker. However, alanine consistently increased for both DDT and endosulfan exposed E. fetida. Alanine showed considerable variability in control earthworms (±41.6%), yet the variability in alanine to glycine ratios was just ±10.5%. Increases in the alanine to glycine ratio were statistically significant at the P = 0.05 level for the 1.0 μg/cm2 DDT dose and both the 0.5 and 1.0 μg/cm2 endosulfan doses, suggesting that deviations from the normal homeostatic ratio of 1.5 for alanine to glycine is a potential biomarker of DDT and endosulfan exposure warranting further study.

Assessing organic potato cropping systems at the end of the first cycle of four-year rotations using principal component analysis

A single or a few variables may not be sufficient to evaluate management practice effects in a complicated cropping system, so six plant and 13 soil variables were integrated using principal component analysis (PCA) to examine nine 4-yr organic potato rotations. The rotations were combinations of three forage levels (0, 1, and 2 yr of forages) with three soil amendments (monogastric compost, ruminant compost, and alfalfa meal). Quantities of amendments were estimated by soil test recommendations and amendment nutrient availabilities. In the 4th potato year, one half of each original plot was not amended ("the 4th year unamended plots"), while the other half received soil amendments ("4th year amended plots"). The first three principal components explained 67 and 63% of the overall variation for the 4th-yr amended and unamended plots, respectively. PCA ordination plots indicated that, overall, the type of soil amendments had larger effects on soil and plant variables, but forage frequencies were influential for the amendments showing weaker effects. PCA loading plots indicated that plant nutrient uptake and potato total tuber weight would be the best single variables for characterizing the current cropping systems. Plant variables, except for potato petiole nitrate, were closely displayed, but they were not strongly correlated with soil variables, which may reflect the high background fertility of this site. Applications of soil amendments in the 4th yr affected the relationships among variables, most notably the strength of relationships between soil pH and soil N variables. The results suggest that PCA provides an effective way to compare complex cropping systems, especially in situations with high site heterogeneity. Key words: Principal component analysis, soil amendment, livestock system, forage, potato, organic crop rotation

Electronic nose based on the polymer coated SAW sensors array for the warfare agent simulants classification

A reference and three polymer coated surface acoustic wave sensors making an electronic nose were introduced. This system contained the pre-concentration unit in order for detection of low level warfare agent simulants. Principal component analysis (PCA), artificial neural network (ANN) and the combination of these two methods (PCA-ANN) were used for differentiation between the three classes including nerve and mustard simulants and mixtures of them. Different polymers including hexafluoro-2-propanol-substituted polysiloxane SXFA, poly(epichlorohydrin) PECH and phenyl methyl polysiloxane (75% phenyl and 25% methyl) OV 25 were used as chemical interfaces and the obtained results were satisfactory. The loading plot of PCA showed that there were no sensors with the similar loadings and thus these three polymers had the enough and effective separated roles in differentiating between above-mentioned various gas classes. It was also proved that the ANN and PCA-ANN had good capability in pattern recognition of aforementioned gas classes. The fabricated system was successfully used for identification of single and binary samples of the three gas classes above- mentioned.

Automated image analysis as a tool to quantify the colour and composition of rainbow trout ( Oncorhynchus mykiss W.) cutlets

The goal of this paper is to propose and evaluate automated image analysis methods for describing muscle cutlets in rainbow trout. The proposed automated image analysis methods were tested on a total of 983 scanned images of trout cutlets, and included quality traits such as fat percentage, flesh colour and the size of morphologically distinguishable subparts of the cutlet. A sub-sample of 50 images was randomly selected for manual segmentation of the cutlet, the dorsal fat depot and the red muscle and regions. The identification of these regions by manual and automatic image analysis correlated strongly ( r = 0.97, r = 0.95 and r = 0.91, respectively). The estimated fat percentage obtained from image analysis, based on the area of visible fat and the colour of the cutlet flesh, correlated well with chemical fat percentage measured by mid-infrared transmission spectroscopy (MIT) ( r = 0.78). The automated image analysis methods are therefore a reliable means of predicting the fat percentage of trout cutlets. Principal component analysis (PCA) loading plots were used to identify subsets of variables from the image analysis of special significance for further studies; cutlet area, dorsal fat depot area, red muscle area, back height, cutlet width, and width of left and right abdomen wall were among the variables selected. PCA loading plots of different colour variables indicated that simple statistical coefficients such as percentiles and mean values can be used to quantify different aspects of flesh colour. In conclusion, the methods presented here provide a powerful toolbox for describing important morphological structures and quality traits of trout cutlets.

Use of principal component analysis to investigate the origin of heptadecenoic and conjugated linoleic acids in milk.

The aim of this paper was the application of principal component analysis (PCA) 1) to elucidate mutual metabolic relationships between milk fatty acids (FA) and 2) to illustrate the origin of milk FA, in particular C17:1 and cis-9,trans-11 conjugated linoleic acid. Data were combined from 3 experiments with lactating Holstein-Friesian cows offered diets based on grass or legume silage and concentrates. Loading plots of PCA based on milk FA concentrations showed 4 groups of milk FA, having similar precursors or metabolic pathways in the rumen and/or mammary gland: medium-chain saturated FA, de novo synthesized from acetate and beta-hydroxybutyrate; monoenoic milk FA, products of delta9-desaturase activity in the mammary gland; odd chain FA of rumen microbial origin and C18:0, n-6 C18:2, and n-3 C18:3 of dietary origin or the result of rumen biohydrogenation. Loading plots of PCA based on both milk and duodenal FA concentrations as well as on milk FA yields and duodenal FA flows further illustrated the importance of postabsorptive synthesis of the milk medium chain saturated and monoenoic FA and the direct absorption from the blood stream of odd chain FA, C18:0, n-6 C18:2, and n-3 C18:3. In all loading plots, milk oleic acid (C18:1) appeared intermediate between clusters of 18-carbon FA and monoenoic FA, illustrating its dual (dietary and endogenous production) origin. Milk C17:1 was suggested to be a desaturation product of C17:0, in common with other milk monoenoic FA. Finally, the PCA technique, based on milk FA patterns of one experiment, was applied to investigate factors determining cis-9,trans-11 conjugated linoleic acid concentrations in milk. Within the range of diets and cows studied here, we showed changes in cis-9,trans-11 conjugated linoleic acid to be mainly dependent on vaccenic acid supply and to a lesser extent on variation in desaturase activity.

Particulate Fluxes of Aliphatic and Aromatic Hydrocarbons in Near-shore Waters to the Northwestern Mediterranean Sea, and the Effect of Continental Runoff

Particulate fluxes of aliphatic and aromatic hydrocarbons were measured with a sediment trap moored at 80m depth offshore Monaco (200m water column) during an 18-month period. The highest fluxes of n -alkanes and polycyclic aromatic hydrocarbons (PAH) (c . 300 and 10μg m−2day−1, respectively) were noted following a strong rainfall event (March–April 1989) and were mostly accounted for by continental runoff and river outflows. Fluxes during periods of low precipitation (August 1989–August 1990) were one order of magnitude lower for PAHs (1·51±1·40μg m−2day−1) or two orders of magnitude lower for n -alkanes (4·79±3·3μg m−2day−1) than during the earlier period (March–April 1989). The total PAH and total particle fluxes exhibited a positive linear correlation during the entire sampling period (r =0·87, N =31, P< 0·05) underscoring the strong affinity of PAHs for particles. Examination of the seasonal variability of fecal pellet content, associated parameters (total organic carbon, total carbon, nitrogen), and individual hydrocarbon content of particles by Principal Component Analysis (PCA) showed that aliphatic and aromatic hydrocarbons were clustered in several subgroups in the PCA loading plots according to their origin. n -Alkanes were grouped in two clusters: (i) lower molecular weight (n -C16–19) and (ii) the higher molecular weight alkanes (n -C20–38) suggesting different pathways into the coastal zone (i.e. runoff vs atmospheric deposition). The distribution of lycopane, pristane and phytane indicated multiple origins. However, the closer location of the two isoprenoids, lycopane and pristane to fecal pellets, suggests a zooplanktonic origin but phytane to fossil fuel source. PAHs exhibited a variety of pyrolytic sources and only fluoranthene and pyrene were not grouped with the remaining PAHs suggesting multiple sources of pollution in these waters.