The electrochemical method used here is chronoamperometry of dissolved oxygen at a charged DME (Barradas and Kimmerle, 1966; Zvonarić et al., 1973). Chronoamperometry at a mercury electrode is an important and convenient tool for the in situ single particle analysis in an aqueous electrolyte solution. Mercury, as a liquid substrate, has several very unique advantages: it is atomically smooth, fluid and chemically inert with a large set of interfacial data in various aqueous electrolyte solutions, necessary for the analysis of the amperometric signal of an organic particle. Fluidity of mercury perfectly mimics interactions with other fluid interfaces in contact (e.g. biological membranes, vesicles, bubbles). Electrochemical measurement is performed by immersing the DME directly into the electrochemical vessel containing a fresh seawater aliquot of 15 mL. The mercury electrode has a surface area of only 4.7 mm2 with drop lifetime of 2 s. Every 2 s, a new and reproducible mercury surface is formed at the end of the glass capillary and the analysis can be repeated many times. This is an important aspect of the method, since the arrival of particles to the interface is a stochastic process and the representative behaviour can be determined only by analysing a larger set of data collected under identical experimental conditions (Kovač et al., 2000). The electrochemical method is based on the sensing of the interfacial properties, hydrophobicity and supramolecular organization of particles rather than on the chemical composition (Žutić et al., 2004). The main advantage of the electrochemical approach is simultaneous and direct detection of different organic constituents in seawater based on their different electrochemical responses (Fig. 2) while inorganic particles remain undetected (Ivošević and Žutić, 1997; Baldi et al., 1999; Kovač et al., 1999; Svetličić et al., 2006; Ivošević DeNardis et al., 2007a). Organic constituents of seawater can be classified according to their electrochemical responses at the DME based on the molecular adsorption, adhesion and spreading, and particle collision (Žutić et al., 1990, 1993; Tsekov et al., 1999; Svetličić et al., 2006). On the chronoamperometric curve, adsorption of dissolved organic matter and submicron particles is manifested cumulatively as a gradual decrease of reduction current which is proportional to surfactant concentration in the sample, known as surfactant activity (Zvonarić et al., 1973; Žutić et al., 1977, 1999). Surfactant activity of seawater is expressed as the equivalent amount of the nonionic synthetic surfactant Triton-X-100 (polyethylene glycol tert-octylphenyl ether, MW = 600) in milligrams per litre. The polarographic maximum of Hg(II) ions offers an alternative approach to measuring dissolved organic carbon of seawater (Hunter and Liss, 1981). However, adhesion of fluid SAP with hydrophobic properties is detected on the single particle level through spike-shaped signal (Žutić et al. 1993; Svetličić et al., 2001; Ivošević DeNardis et al., 2007b). Collision of gel microparticle with the flexible three-dimensional network, thus possessing a hydrophilic character, is detected also on the single particle level through specific dip-shaped signal (Svetličić et al., 2006; Ivošević DeNardis et al., 2007a). The dropping mercury electrode as an in situ sensor enables characterization of fluid SAP in the aqueous environment (Žutić et al., 1984, 2004; Žutić and Legović, 1987) in terms of concentration, size, reactivity and hydrophobicity (Žutić et al., 1993; Kovač et al., 2000; Ivošević DeNardis et al., 2007b). Figure 3 shows the attractive interaction between an oil droplet and the charged electrode–seawater interface. Oil droplets, as a subclass of surface-active particles, adhere and spread at the charged interface, causing double layer charge displacement and establishment of the flow of the compensating current known as adhesion signals. Signal amplitude reflects the particle size, signal shape reflects the physicochemical properties of the particle and signal frequency indicates particle concentration in the seawater sample. Signal frequency is translated into particle concentration using the corresponding calibration curve with Dunaliella tertiolecta cells as standard particles (Kovač et al., 2000). Chronoamperometric measurements meet the requirements for monitoring based on the following advantages: (i) simple and fast analysis (100 s per sample) of soft micrometre particles in the size and concentration ranges of 1–500 µm and 105–108 particles L-1, respectively, and (ii) neither sample pretreatment nor sample fixation. Measurements on land were done within 24 h after sampling in the laboratory. Previous mesocosm experiments in seawaters show a major difference in the transformation of the surface-active matter not on the timescale of hours but rather on the seasonal level (Žutić et al., 2004).

In order to access more quantitative information from the amperometric signal of an organic particle, the measurement has to be performed in a previously deaerated solution under nitrogen purging for a few minutes so as to remove redox reaction of dissolved oxygen. In this way, it is possible to retrieve information about: organic particle diameter, particle surface area at the interface, number of molecules in the monolayer, surface charge of the particle, critical interfacial tensions of adhesion and kinetic parameters of the adhesion process through analysis of the amperometric signal using a reaction kinetics model and the corresponding methodology (Ivošević DeNardis et al., 2012, 2015).

AFM is based on a relatively simple principle: it involves raster scanning of a sharp and hard tip (probe) located at the free end of a flexible cantilever. The tip is scanned over the surface of a sample, sensing the interaction forces between the tip and sample (Fig. 4). The sample is mounted on a piezoelectric scanner, which allows three-dimensional positioning with subnanometre accuracy. Interaction between the tip and the surface of the sample leads to cantilever bending, which is measured by laser light reflected from the cantilever to a position sensitive photodetector. As changes in cantilever deflection result in variation of the distance between the tip and sample, a constant distance is re-established with a feedback loop between the sample–tip positioning system and a computer-controlled piezoelectric scanner. Registered values of cantilever deflection are electronically converted into a pseudo 3-D image of the sample. As a result, AFM gives real 3-D images of the sample with a vertical resolution of 0.1 nm and lateral resolution of 1 nm. The main advantages of AFM over conventional light or electron microscopy in studies of marine organic matter include (i) 3-D high-resolution imaging of different seawater constituents (biotic and abiotic) connecting micrometric and nanometric dimensions; (ii) samples being non-destructively imaged in the air and under near-natural aqueous conditions; (iii) nanomechanical mapping (i.e. Young's modulus, deformation, hydrophobicity, adhesion); and (iv) determination of intra- and inter-molecular forces in heterogeneous molecular assemblies (i.e. Pletikapić et al., 2014).

After a dramatic mucilage event in the northern Adriatic in the summer of 1997, which had negative impact on the fishery and tourism, the monitoring programme “Systematic Study of the Adriatic Sea as a Base for Sustainable Development of the Republic of Croatia” was launched in 1998. Monitoring of organic matter dynamics was carried out to examine the environmental status in the northern Adriatic and to predict mucilage event. We focus on naturally occurring, non-living and micrometre-sized surface-active particles. This fraction mainly involves soft, hydrophobic, hydrophilic and reactive particles susceptible to degradation. Surface-active particles are described as vesicle or micelle-like structures formed by self-assembly of organic matter, primarily of lipid, polysaccharide and proteinaceous components deriving “mostly” from excreted cells and/or from their decomposed products (Žutić and Svetličić, 2000). They have an affinity to accumulate at the interfaces, and they can be measured directly electrochemically on the single particle level based on their interfacial properties, structural organization and hydrophobic–hydrophilic character. We introduced a direct electrochemical method to probe the state of marine organic matter without perturbing its original heterogeneous distribution. Figure 5 shows distribution of SAP (hydrophobic), gel microparticles (GeP) (hydrophilic) and surfactant activity measured over a period of 10 years at sampling station SJ 105. This particular sampling station was selected because it is distant from the coastline and thus the influence of direct input of inland freshwater is minimized (Fig. 1b). SAP concentrations ranged from 5 × 105 to 6 × 106 L,-1 showing spatiotemporal variability with special patterns, which depends on the sampling depth in the water column and/or season. For example, significantly (about 2 times) higher SAP concentrations occurred in the region under the direct influence of nutrient-rich freshwater inputs of the Po River (SJ 101, 103, 108). In the first 3-year period (2002, 2003, 2004), a mucilage event appeared. In the period just before the mucilage event, SAP concentration rose to 1 × 107 L-1 in the subsurface layer (0–10 m, SJ 101, 103, 108), indicating the forthcoming event (Svetličić et al., 2005). In contrast, low concentration of SAP in the subsurface layer at station SJ105 in 2007 is in line with the absence of mucilage events. Hence, SAP was identified and considered as a reactive mucilage precursor in seawater. Concentrations of GeP were usually around 105 L-1 in the years without mucilage events (0.2 × 105 L-1 is the lower electrochemical detection limit). An elevated concentration of GeP, by about 1 order of magnitude (15 × 106 L-1), was determined in all sampling stations at the time of mucilage events (2002–2004). Surfactant activity reflecting the concentration of submicron and dissolved organic fractions varied from 0.5 to 2 mg L-1 in the years without mucilage events and decreased with the depth profile. Elevated surfactant activity up to 3.4 mg L-1 was determined during the mucilage event. In addition, a net increase of surfactant activity in the whole water column was observed at the end of years without mucilage events as a consequence of prolonged accumulation of organic matter in the water column (unpublished results). It was therefore concluded that the likelihood of mucilage formation depends mainly on new primary production, specific structural organization of the newly produced organic matter and, above all, hydrographic and meteorological conditions in the seawater.

AFM as a high-resolution imaging technique that requires only easy sample preparation has a huge potential in investigating marine organic matter, particularly in elucidating the structure, transformations and dynamics of organic matter in the marine environment. Of all forms of organic matter, marine fibrils and fibrillar networks were the most studied form. For example, Santschi et al. (1998) were the first to image individual fibrillar polysaccharides in marine macromolecular organic matter. From there on, AFM has been applied to study both (i) the networks secreted by marine bacteria and algae (Nishino et al., 2004; Svetličić et al., 2005, 2013; Malfatti and Azam, 2009; Malfatti et al., 2010; Urbani et al., 2012; Bosak et al., 2012; Villacorte et al., 2015b) and (ii) marine fibrils (Mišić Radić et al., 2011; Svetličić et al., 2011) forming giant gel macroaggregates (Vollenweider and Rinaldi, 1999; Giani et al., 2005a, b), including their self-assembly and molecular structure at different levels of association (Mišić Radić et al., 2011; Pletikapić et al., 2014). Application of AFM imaging for monitoring by itself is often not suitable. This is due to the large number of samples required in the monitoring studies, making AFM analysis, if used alone, a time-consuming process. It is therefore recommended to analyse seawater samples first with rapid and “bulk” techniques (i.e. chronoamperometry) in order to identify samples of interest. For AFM imaging, the drop deposition method modified for marine samples and imaging in air was found to be most adequate for visualization of macromolecular organic matter organization (Pletikapić et al., 2011; Mišić Radić et al., 2011, for more detail). Typically, AFM measurements of marine organic samples are performed at room temperature and 50–60 % of relative humidity, which leaves samples with a small hydration layer adhering to the substrate, helping to maintain the native structure (Balnois and Wilkinson, 2002). It was found that direct deposition of a drop of seawater (5 µL) on freshly cleaved mica, followed by the rinsing of sea salts and evaporation of water excess, was the procedure that caused the least impact on the original structures of biopolymers and other assemblies in seawater (Mišić Radić, 2011; Svetličić et al., 2013). Such an approach could offer an alternative way to visualize fragile organic structures. Although an aqueous environment would be preferable, direct AFM imaging of marine polymers, soft vesicular structures and organic droplets is often hampered by their weak interaction with the interface. While interpreting the data, one should consider possible surface and/or dehydration artefacts. Representative images of marine fibril forming networks with different levels of biopolymer entanglements and with different pore sizes are shown in Fig. 6. It is interesting to note that the appearance of marine gel networks at the nanometric scale, as shown in the corresponding figure, coincided with the occurrence of gel microparticles detected electrochemically at the same sample, as shown in Fig. 5. Image analysis revealed that biopolymer heights were in the 0.5–2.5 nm range, while pore sizes ranged from ∼ 40 nm to more than 1µm along the longest axis. Svetličić et al. (2011) found that these parameters (pore size, fibril height) can be related to the aging of gel aggregates in the water column; i.e. older aggregates contained smaller pores and thicker fibrils, while “young” aggregates had a looser structure with smaller fibrils and bigger pore sizes. One of the most comprehensive chemical characterization studies of northern Adriatic gel macroaggregates showed that carbohydrates are the main organic fraction and that the average ratio of polysaccharides to total organic carbon is 41.5 % (Giani et al., 2005b). Exopolysaccharides released by the diatom Cylindrotheca closterium under the conditions of increased production comprised a simpler system, which had the same chemical composition of polysaccharides as marine gels, indicating that the diatom exopolymeric substance was the most likely origin of marine gels (Urbani et al., 2005). In addition, a striking similarity was found in both marine gels and reconstituted polysaccharide networks (obtained from the same diatom) in terms of their mechanical properties as well as morphological features (Pletikapić et al., 2014). Marine gels were characterized in the non-imaging mode of AFM, the so-called “fishing”, where single polymers and/or their more complex associations are “picked up from the surface (fished) by the AFM probe and extended and/or unfolded/unzipped by increasing the distance of the AFM tip from the surface. This approach enabled relating mechanical properties to different biopolymer morphologies (from single fibril, helical structures to more complex junction zones in networks with a higher degree of cross-linking), revealing new information about network architecture and connectivity. As native marine gels are constantly exposed to mechanical perturbations caused by local hydrodynamics (turbulence and shear) as well as large-scale physical processes (strong winds and currents), the application of hundreds of piconewtons of force is highly biologically relevant. Networks were characterized as physical gels in line with DCS measurements (Mišić et al., 2011), where the level of association can be controlled by dilution, adjusting the ionic strength and stirring (Pletikapić et al., 2014). Marine gels are defined as three-dimensional networks of solvated biopolymers embedded in seawater. Their importance in marine ecosystems has been widely recognized among scientists from different fields due to their role in the microbial loop and sedimentation processes, biogeochemical carbon cycling, marine carbohydrate chemistry and particle dynamics in the ocean (Žutić and Svetličić, 2000; Azam and Long, 2000; Passow, 2002; Verdugo et al., 2004; Verdugo, 2012). These gels are also bioreactive since they compartmentalize nutrients and can thus enhance microbial activity (Kepkay, 1994; Azam, 1998, Del Negro et al., 2005). Due to their dense 3-D network structure and “sticky” nature, marine gels show a potential to scavenge nanoparticles that accumulate in the junction zones of the network (Svetličić et al., 2011; Pletikapić et al., 2012b). In this way, marine gels might stabilize nanoparticles and prevent aggregation and export from the water column. Recent review highlights the potential of AFM in revealing new information about the interaction of inorganic nanoparticles with living and non-living organic matter (Ciglenečki and Svetličić, 2015; Wang et al., 2016).

In addition, gel network formation was visualized at station SJ 107 in August 2010, when concentrations of micrometre-sized GeP and SAP were low (0.2 and 2.3 × 106 L-1, respectively). Significantly higher concentration of submicron organic content (surfactant activity, 2.88 mg L-1) was detected 2 months prior to the mucilage event at the depth of 10 m, in 2010, which could act as a pool for biopolymer self-organization into marine vesicles that under specific conditions transform to giant gel (Svetličić et al., 2005). For identification of vesicle-like structures, AFM can be a very suitable tool due to the high-resolution imaging and easy sample preparation. The existence of marine vesicles has already been inferred by Žutić et al. (1984) and later by Nagata and Kirchman (1997). Recently, it was reported that membrane vesicles of 100 nm in diameter can be released by marine cyanobacteria and are frequently found in marine ecosystems in the concentration range 105 to 106 vesicles mL-1 (Soler et al., 2015). Such vesicles may entrap nutrients, toxins and a variety of other organic molecular species present in seawater. These components may be concentrated and preserved from degradation in the inner cavity of the vesicle and/or attached to the membrane of the vesicle. This ability of vesicles to transport and deliver diverse compounds in discrete packages adds another layer of complexity to the flow of information, energy and biomolecules in marine microbial communities (Biller et al., 2014). Such vesicles (Fig. 7) were also periodically encountered during the diatom bloom experiment and during the northern Adriatic field studies by AFM (Žutić et al., 2004; Svetličić et al., 2008). It is likely that the abundance of such structures is highly underestimated. Due to the small size and relatively soft and fragile character of these structures, during AFM imaging with relatively higher forces, particularly when contact mode is used, vesicles are swept away by the AFM tip and remain undetermined. Proper care while setting imaging parameters, scanning at low force (i.e. peak force tapping mode) and coupling AFM with other bulk methods such as electrochemistry to identify the sample of interest with higher concentrations of vesicles may resolve the issue. Such an approach was used to study vesicle-like particles in the Bay of Kotor (Ivošević DeNardis et al., 2013, 2014). The most important findings show that both surface analytical methods, electrochemical and AFM, provide comparable results on different scales.

However, despite this, AFM is not yet exploited up to its full potential. Significant advances in the field are expected by integrating AFM into versatile hybrid devices that would combine two or three complementary techniques in one instrument, allowing a more detailed and comprehensive analysis of marine samples. While simultaneous AFM imaging and mechanical mapping (stiffness, friction, dissipation and/or adhesion) is already showing its potential by increasing the number of studies conducted on biological samples (Dufrêne et al., 2013, and references therein), including marine samples (Francius et al., 2008; Pletikapić et al., 2012a), of particular interest for investigation of marine organic matter is coupling AFM with different optical, spectroscopic and/or interfacial techniques (Moreno and Toca-Herrera, 2009). In line with the topic covered in this paper, integration of mercury as a substrate in the AFM set-up is drawing particular attention (Schon et al., 2013). An AFM cantilever has been developed using a mercury fountain pen probe allowing simultaneous probing of mechanical and electrical properties, for instance in biological membrane research. In addition, a hybrid AFM-optical (fluorescence) microscope (Kassies et al., 2005; Geisse et al., 2009) would be extremely useful for deepening the research on different classes of organic particles traditionally detected by staining and microscopic analysis. An additional step forward in the field is expected by combining AFM and IR to provide simultaneous correlation of topographical (organizational) and chemical data (Dazzi et al., 2012; Amenabar et al., 2013; Kulik et al., 2014). Alternatively, one could use chemically functionalized tips and, by simultaneously performing imaging and force curve acquisition, generate chemical maps (Blanchette et al., 2008). Such studies would allow relating molecular organization of marine organic matter with chemical information on the micro- and nanometre scales, which would undoubtedly widen the spectrum of information about samples of interest.