ANNAUES • Ser. hist. nat. 11 2001 • 1 (23) original scientific paper UDC 574.5(262.3-17) received: 1.6,2001 QUANTIFICATION OF MUCILAGE-ASSOCIATED SUSPENDED MATTER IN THE GULF OF TRIESTE (ADRIATIC SEA) Alenka MALE!, Boris PETELIN & Branko ČERMELJ Marine Biology Station, National Institute of Biology, SI-6330 Piran, Fornače 41 ABSTRACT Mucilage-associated suspended matter in the Gulf of Trieste has been quantified using different approaches during the 2000 event. Remote sensing, video recording and image analysis were used to assess the quantity of surfaced mucilage. The quantity of water column mucilage was estimated taking quantitative samples of mucilage using a peristaltic pump and using in situ photography, video recording and image analysis. Samples of surface and water column mucilage were assessed for wet mass immediately after sampling, while for further analyses they were deep-frozen and analysed later. Our results indicate that during culmination of the 2000 event on average less than 2% of sea surface was covered by mucilage in offshore areas. The quantity of particulate matter in offshore surface mucilage was on average 220 mg carbon/L, however, mucilage was limited to less than 0.5 cm thick surface layer. Concentrations were significantly lower for the water column mucilage (47.6+1.7 mg C/L for mucilage clouds, 5.9+4.2 mg C/L for mucilage network). Integrating obtained values for the water column to a depth of 25 m we estimated the average quantity of mucilage-associated particulate matter in mid-June to be about SOg C/m2. Key words: mucilage, Adriatic Sea, quantification, video recording, remote sensing, image analysis, particulate matter accumulation QUANTIFiCAZIONE DELLA MATERIA SOSPESA ASSOCIATA ALLE MUCILLAGíNI NEL GOLFO Di TRIESTE (MARE ADRIATICO) SINTESI Durante il manifestarsí delle mucillagini nell'anno 2000, con diversi approcci gli autori hanno quantificato la materia sospesa associata al fenomeno. A tale scopo son o state utilizzate immagini telerilevale da elicottero successivamente elabórate con programmi grafici (Adobe Photoshop, CorelPhoto-Paint) e con il programma per l'analisi di immagini satellitari ENVI. í campioni per l'analisi quantitativa delle mucillagini nella colonna d'acqua sono stati raccolti con l'aiuto di una pompa peristáltica. La quantitá di mucillagini é stata calco/ata come massa umida, sostanza secca e carbonio. I hsultati dimostrano che durante la comparsa del fenomeno nel giugno 2000. in media meno del 2% della superficie marina é rimasta coperta dalle mucillagini. Nella colonna d'acqua la rete mucosa ha avuto un conten uto medio di 5.9+4.2 mg C/L, menlre le nubi mucose di 47.6±1.7 mg C/L. Gli autori hanno ínoltre calcolato che nella colonna d'acqua di 25 metri la quantitá di carbonio ha raggiunto, all'apice del fenomeno nel giugno 2000, valori medi di 80 g C/m2. Parole chrave: mucillagini, mare Adriático, quantificazione, remote sensing, eiaborazione di immagini, accumulo di materia sospesa 43 ANNALES • Ser. hist. nat. 11 2001 ■ 1 (23) Alenka MA L t'J et ûl : QUANTIFICATION OF MUCII.AGF-ASSOCIaTED SUSPENDED MATTER IN THE CULf Of TRIESTE (ADRIATIC SEA), 43-52 INTRODUCTION Recent events (1988, 1989, 1991, 1997, 2000) of massive mucilage accumulation in the northern Adriatic have inflicted acute changes on the ecological dynamics of this marine system and have created serious problems for different human activities in the region-Due to the frequency of occurrence, geographical scale, subsequent ecological and socio-economic impacts as well as their intricacy, mucilage events have become a matter of high environmental concern. All this has prompted extensive research in the last decade and mucilage has become one of the priorities of the national marine research programmes of the Adriatic countries as weil as that of major EU funded projects (Hopkins ef al., 1999). Numerous scientific papers have been published and up-to date knowledge has been synthesised in special volumes of scientific journals (Vollenweider ef al., 1995; Funari ef al., 2000). Yet, many aspects of this peculiar phenomenon, which in its extensiveness seems to be unique to the northern Adriatic, have remained unresolved. Many researchers consider massive accumulations of mucilage as an extreme case of particulate matter aggregation. Aggregates are ubiquitous in marine and limnetic environments and are formed mainly from smaller particles in the water column. Coagulation is an important mechanism controlling aggregate size distribution. However, other features are also considered to be important in large aggregate formation such as abundance of TEP (transparent exopolymer particle) (Alldredge ef. al., 1993), presence of particular phytoplankton group/species (Kiorboe & Hansen, 1993), spontaneous assembly of dissolved polymers forming stable gels (Chin ef al., 1998), and sustained residence of aggregates in surface waters (Riebesel, 1992). Classical coagulation theory underlines three main mechanisms for particle collision, an important step in formation of aggregates: laminar and turbulent shear, Brownian diffusion and differential sedimentation. To increase their size particles have to adhere upon collision. Adhesion rate depends mainly on the stickiness of particles; thus, formation of large aggregates is determined by both quantity and quality of available particles (Alldredge & Jackson, 1995). Resistance to disaggregation and reduced settling velocity are further conditions for accumulation of aggregates in the water column. Particle spectra in the oceans were found to be in general agreement with those expected from coagulation theory (Jackson & Burd, 1998). This theory, however, appears unsatisfactory in the case of the mucilage phenomenon. Due to the dramatic spatial extent of mucilage, some events affected areas of over 10,000 km2 and accumulated at the surface in layers that were several cm thick (Stachowitsch et a/., 1990; Rinaldi et a!., 1995) it was suggested that mucilage-associated particulate matter reached very high values, although there were no field data to support this presumption. Variable forms and changeable vertical and horizontal distribution in the water column and at the surface make quantification of total mucilage-associated matter a very difficult task. The most promising approaches seemed to be in situ visual monitoring using underwater cameras (Rinaldi ef al., 1995), although classical methodology for assessment of suspended matter was also used (Faganeli ef al., 1995). Satellite remote sensing of the event in 1989 followed temporal changes of surface mucilage (Zambian-chi eta I., 1992) while Berthon et al., (2000) used a bio-optical data set collected by an off-shore oceanographic tower off Venice to study optical properties of the water column in the presence of mucilage. Neither of these studies tried to estimate the quantity of mucilage-associated particulate matter. In order to provide quantitative estimates of mucilage accumulation in the water column and on the surface we investigated this phenomenon in the Gurlf of Trieste. Using different approaches i.e. remote sensing, video recording and quantitative sampling we sottght to quantify different forms of mucilage aggregates. MATERIALS AND METHODS We have made quantification of mucilage-associated suspended matter on the basis of data collected during event in early summer (June-July) of 2000. Quantification of surfaced mucilage The quantity of surfaced mucilage was assessed from pictures taken by video camera that was mounted on an aircraft. The eastern part of the Gulf of Trieste was over flown by helicopter and the sea surface surveyed for muciiage on 10, 11 and 14 June 2000. The survey on 14 June followed the path depicted in figure 1 (Plate I), it was carried out between 1 and 2 p.m. The height of this flight was between 400 and 500 m, and the velocity of the aircraft between 110 and 120 km h"1. The snapshots of the video were taken every 15 seconds arid we got 234 pictures; among them we selected 154 for further statistical analysis. The remaining pictures were not taken into account because they either partly showed land or displayed the same scenes due to aircraft manoeuvring or path repetition. To estimate the spatial coverage of mucilage in the studied area, analysis of pictures taken from the aircraft, was carried out using two procedures, in the first, we utilized graphical software (Adobe Photoshop, Corel Photo-Paint) to manually paint mucilage areas white and the mucilageless sea surface black. The second method of picture elaboration was thresholding using the software too! ENVI for image elaboration. We used contour plot to determine the threshold between muciiage and ANNALES • Ser. hist. nat. 11 2001 ■ 1 (23) Alenka MA L t'J et ûl : QUANTIFICATION OF MUCII.AGF-ASSOCIaTED SUSPENDED MATTER IN THE CULf Of TRIESTE (ADRIATIC SEA), 43-52 the sea surface without mucilage, applying one contour on which we varied the levels. Before further classification of images we carried out the "decorrelation stretch" to remove the high correlation commonly found in mul-tispectra! datasets. The 'isodata' method, i.e. unsupervised classification was then applied to destretched images. This method calculates class means and iteratively clusters the remaining pixels using minimum distance technique (Tou, 1974). From ail classified images, descriptive statistics and histograms were calculated. Example of application of these procedures to the same situation at sea is presented in figures 2-4 showing the original video snapshot of mucilage masses at sea (Fig. 2); manual elaboration of the picture, using graphical software (Fig. 3); and corrected 'isodata' (unsupervised classification) picture after thresholding and application of decorrelation stretch (Fig. 4). Results obtained by different methods were rather similar (Tab. 1). Tab. 1: Comparison of different methods for assessment of mucilage coverage applied to the situation in figure 2. Tab. t: Pokrovnost morske površine s sluzjo na primeru s slike 2 (ocenjena z različnimi metodami). On 11 June we also followed changes in the coverage of the sea surface by mucilage over a day (from 8 a.m. to 8 p.m.) using still photography of the same near-shore location. Pictures were elaborated manually using graphical software (Fig, 6). Surfaced mucilage was sampled manually using 0.5 to 1.5 f polyethylene bottles for quantification of suspended matter and its quantity determined using the same procedures as for water column samp!es (see below). Quantification of the water column mucilage Field surveys, in situ video-recording and sampling of the water column mucilage were carried out approx. weekly from 6 )une, when mucilage was first observed by divers, to 26 July, 2000, when no mucilage aggregates of any form were seen in the studied area. Field observations were most frequent from 11 to 21 June, when the mucilage event peaked in the studied area. Therefore, quantification of different mucilage forms and layers (surfaced mucilage, water column mucilage net- work and clouds) was based on the data collected during this period. Sampling. Concentration of the mucilage-associated suspended matter in the water column was estimated on the basis of quantitative sampling using a peristaltic pump that was operated by SCUBA diver. While sampling, we distinguished two types of water column mucilage (Malej et a I., 2000): 1) loose network of mucous material stretching in layers up to 10 m thick, that were considered "quasi homogeneous suspended matter field", and 2) mucilage "clouds", that were spatially well defined and were surrounded by particle-poor water, in both cases, 35 litres of mucilage matter were sampled into large glass containers that were brought to the laboratory within one hour. Quantification of the mucilage network was derived from the premise that sampling by pump provided an average sample of the homogeneous layer. In the case of mucilage clouds our methodology provided data on particulate matter associated with these clouds. Additionally, traditional sampling using 5 L Niskin bottles at standard depths was carried out. Analyses. Collected samples were analysed in the laboratory using the same procedure for all three types of mucilage (surfaced, mucilage network, mucilage clouds). Mucilage-associated suspended matter, that was collected in large (3.5 litre) containers, in all cases compacted in the surface layer. Further analysis consisted of determination of the volume of this compacted mucilage layer (cms-concentrated mucilage sample) using measuring cylinders, followed by an estimate of the remaining, presumably mucilage-free, water (mlw). Quantitative subsamples were then taken from both (cms and mfw) for suspended matter determination. The ratio of total sample volume to cms volume was calculated and used in calculation of the integrated water column mucilage-associated particulate matter. For the purposes of this paper conversion factors from volume to dry mass and carbon content of mucilage were taken from our previous study, done during the 1997 event (Flander, 1999; Malej et at., 2000): mean dry mass content of mucilage was 1.6 (±0.9)% of wet mass; and mean carbon content of 10.7 (±3.2)% of dry mass. Two examples of pictures, taken in the water column by SCUBA divers using an underwater camera, were also analysed applying the same methodology of picture analysis as for the surfaced mucilage. The method applied to water column video snapshots was unsupervised "isodata" classification (see above). RESULTS AND DISCUSSION Time-course of the mucilage event in 2000 in contrast to previous mucilage events in the last decade (1991 and 1997), the 2000 phenomenon appeared earlier In the season. While in 1991 and 1997 Method Mucilage {% of surface) Mudlageless j (% of surface) Manual separation 6.9 93.i Thresholding 6.0 94.0 Unsupervised classification 7.0 93.0 ANNALES ■ Ser. hist. nat. -11- 2001 • 1 (23) Alenka MALEI el a/.: QUANTIFICATION OF MUCILAGE-ASSOCIATED SUSPENDED MATTER IN THE GULF OF TRIESTE (ADRIATIC SEA), 43-52 the water column mucilage accompanied by surface masses were observed in mid-July (Flander Putrle et at., 2000), during the 2000 event visible mucilage has already appeared in the water column in the first days of June with surface accumulations starting from 10 June. The phenomenon reached its peak in mid-June and dissipated by mid-July. Already by 28 June some portion of mucilage has settled to the bottom and after mid-July mucilage was not observed in the water column. Mucilage completely disappeared from the studied area by the end of July. Most results used for calculation of mucilage-associated matter are based on samples collected from June 11 to 21. In this period we carried out measurements using all reported techniques (remote sensing, quantitative sampling of different mucilage forms, vide-recording, standard oceanographic techniques). Quantification of surfaced mucilage The spatial extent of the surfaced mucilage was assessed when the phenomenon reached its culmination in the studied area, i.e. on 14 June. According to our previous observations, which indicated that ascension of mucilage to the surface occurred predominantly during late morning, we made our estimate of spatial coverage of the sea surface on the basis of measurements done between 1 and 2 p.m. We assumed that later during the day the quantity of surfaced mucilage did not increase significantly. Results of descriptive statistics of 154 snapshots taken from aircraft are given in table 2 and figure 5. Only about 10% (16) images showed mucilage coverage greater than 3% and the situation presented in Table 1 and figure 2 i.e. about 7% coverage was rather rare. In a majority of the images (120) coverage by mucilage was less than 2% of sea surface and it was nil in 12 snapshots. Applying an overall mean coverage of 1.6% of sea surface by mucilage (Tab. 2) to the whole Gulf of Trieste (about 600 km2) gives an estimated mucilage-covered area of about 10 km2. Tab. 2: Descriptive statistics of 154 snapshots, taken from aircraft on 14 June 2000. Tab. 2: Opisna statistika obdelanih videoposnetkov površinske sluzi junija 2000. Item % coverage by mucilage Minimum u Maximum 21.0 Mean 1.6 1st quartile (Q1) 0.3 Median 1.0 3rd quartile (Q3) 1.9 Variance 6.2 Standard deviation 2.5 Higher surface coverage was observed along shores during late afternoon hours where the surfaced mucilage accumulated because it was driven by sea breeze towards the shoreline. Figure 6 gives an indication of the variability of surface coverage by mucilage in such a near shore area. Coverage was nil before 9 a.m., increased to about 30% at 11 a.m. and reached over 50% between 01 and 02 p.m. These data indicate that past estimates, if based on near shore data, probably overestimated the quantity of surfaced mucilage. Alternately, previous events were more intense. Wet, dry and carbon mass of surfaced mucilage. Measurements of mucilage wet mass gave rather different values for surfaced mucilage collected offshore as compared to values of more compacted mucilage that accumulated near shore. Offshore mucilage had an average wet mass of 130±2 g per litre of seawater in contrast to a significantly higher mean wet mass of that collected near shore (870±18 g per litre sea water). Applying conversion factors (dry mass to wet mass) of 1.6% and 10.7% carbon content, estimated average dry mass was 2.1 g DW/L and 0.22 g C/L for offshore surface mucilage. Respective values for mucilage that accumulated near shore were 13.9 g DW/L and 1.5 g C/L. However, these extremely high particulate mass values were limited only to top layers thinner than 0.5 cm offshore and 5 cm near shore. Surface mucilage assessed by this technique represented only a fraction of the water column mucilage that ascended to surface. Therefore we tried to quantify particulate mass of the water column aggregations. Quantification of the water column mucilage Particulate matter. Average concentration of particulate matter in the water column during the mucilage event estimated on the basis of Niskin samples was 3.8±1.8 mg DW/L. Values were higher below 20 m depth (5.6±2.1 mg DW/L) than in the upper 15 m (3.0±0.9 mg DW/L). Using the conversion factor of 3% carbon to dry mass, average estimated particulate carbon was 0.1 mg C/L. Mucilage network. Quantification of particulate matter associated with the mucilage network on the basis of sampling by pump indicated a mean mucilage wet mass of 3.5±2.6 g WW per litre seawater. Using the same conversion factors as for surface mucilage (1.6% wet to dry mass, 10.7% carbon content) we estimated that this mucilage layer on average contained 55.6±40 mg DW/L and 5.9±4.2 mg C/L. Over the time from June 10 to June 20, the wet mass of mucilage network increased from 1.2 g WW/L (2.0 mg C/L) to 7.1 g WW/L (12.1 mg C/L) indicating that "older" mucilage had a greater particulate mass per unit volume. In contrast, sampling this layer during the event using Niskin bottles gave significantly lower DW of 5.2±3.9 mg DW/L (0.2+0.1 mg C/L). 46 ANNALES • Ser. hist. nat. • 11 • 2001 • 1 (23) Alenka MALE) ef .it.: QUANTIFICATION OF MUCILAGE-ASSOCIATED SUSPENDED MATTER IN THE CULF OF TRIESTE (ADRIATIC SEA), 43-52 PLATE I/TABLA I: Fig. 1: The helicopter flight path during video recording of the southeastern part of the Gulf of Trieste. SI. 1: Pot helikopterja med snemanjem površinske sluzi v jugovzhodnem delu Tržaškega zaliva. Fig. 2: Original video snapshot of surface mucilage taken from aircraft on 14 June 2000. SI. 2: Originalen videoposnetek površinske sluzi 14. ¡unija 2000. Fig. 3: Manually elaborated video snapshot from figure 2 using graphical software. SI. 3: Grafično obdelan videoposnetek s slike 2. Fig. 4: Corrected unsupervised classification ('isodata' method) - decorrelation stretched situation on Fig. 2. SI. 4: Prikaz površine morja z videoposnetka na sliki 2 po opravljenem klasifikacijskem postopku. 47 ANNALES • Ser. hist. nat. ■ 11 - 2001 - 1 (23) Atenfea MALfJ etal.; QUANTIFICATION Of MUCH AGE-ASSOCIATED SUSPENDED MATTER IN THE ClJLF OF TRIESTE (ADRIATIC SEA), 43-52 Tab. 3: Comparison of wet mass, dry mass and carbon concentrations the during mucilage event in 2000: particulate matter (Niskin samples, taken at standard depths of 0.5, 5, 10, 15, 20, m above bottom); water column mucilage: mucilage network, mucilage clouds; and surfaced mucilage: a) offshore samples, b) concentrated mucilage along shore. Tab. 3: Primerjava mokre mase, suhe mase in ogljika v sluzi I. 2000: lebdeča snov (vzorčenje z Niskinovo posodo na standardnih globinah 0,5, 5, 10, 15, 20 m in nad dnom); sluz v vodnem stolpcu: sluzasta mreža in sluzasti oblaki; površinska sluz: a) na odprtem morju, b) ob obali. Wet mass (g/L) Dry mass (mg/L) Carbon (mg/L) particulate matter (Niskin) 3.8+1.8 0.1 ±0.09 mucilage network 3.5±2.6 55.6 ±40 5.9±4.2 mucilage clouds 27.8±1.8 445+16 47.6±1.7 a) 'offshore' surfaced mucilage 120±20 2,100 220 b) 'along shore' surf, mucilage 870+18 13,900 1,500 v \' >n" v v s' v s •