Correspondence: Tomislav Radić, Center for Marine Research, ‘Ruđer Bošković’ Institute, G. Paliage 5, 52 210 Rovinj, Croatia. Tel.: +385 52 804 761; fax: +385 52 814 396; e-mail: email@example.com
The influence of inorganic nutrient concentrations on the ability of bacterioplankton to produce and degrade polysaccharidic transparent exopolymer particles (TEPs) and proteinaceous Coomassie-stained particles (CSPs) was investigated in an 11-day experiment. The dynamics of these particles were followed in prefiltered (1 μm) northern Adriatic seawater enclosures enriched either with 1 μM orthophosphate (main limiting nutrient in this area), 10 μM ammonium or both orthophosphate and ammonium. These enclosures were referenced to a nonenriched control. A high potential for bacterial TEP and CSP production was observed (104–105 L−1 for particles larger than 4 μm). In conditions of high orthophosphate concentration (either orthophosphate enriched or both orthophosphate and ammonium enriched), lower abundances and surface areas of CSPs were obtained, whereas TEP dynamics were more affected by unbalanced enrichments where only orthophosphate or ammonium was added. The impact of unbalanced nutrient ratios on TEPs was indicated by their higher abundance but low capacity for Alcian blue absorption, implying a change in their structure. Inorganic nutrient availability was thus proven to affect the bacterial potential for producing and degrading bacterially derived TEPs and CSPs.
The growth and activity of heterotrophic bacteria in seawater are not only carbon/energy limited, but may also be limited by inorganic nutrients, due to the requirement for balance in the ratios of elements. In conditions of deficiency of inorganic components (nitrogen or phosphorus), their efficiency in decomposing organic matter is reduced (Obernosterer & Herndl, 1995; Zoppini et al., 1998; Azam et al., 1999; Pettine et al., 2001), including degradation of polysaccharidic transparent exopolymer particles (TEPs) and proteinaceous Coomassie-stained particles (CSPs). The northern Adriatic, a shallow phosphorus-limited basin, largely influenced by the large Po river discharges, enables significant primary production (Degobbis, 1990; Degobbis et al., 2005), but under conditions of high N/P ratios large amounts of refractory polysaccharides are released by phytoplankton cells (Puddu et al., 2003; Vadstein et al., 2003). Since bacteria are capable of faster uptake of inorganic phosphorus than algal cells, their presence accentuates phosphorus depletion, resulting in higher polysaccharide production and accumulation (Obernosterer & Herndl, 1995; Guerrini et al., 1998; Thingstad et al., 1998; Vadstein et al., 2003). These processes may lead to massive accumulation of organic matter and macroaggregate formation in the northern Adriatic (Degobbis et al., 1995), wherein TEPs have a significant role due to their high concentrations and intrinsic stickiness (Alldredge, 1999; Radićet al., 2005). Although CSPs are also as abundant in the marine environment as TEPs, their ecological roles are still underexplored.
The hypothesis that heterotrophic bacteria would degrade polysaccharidic and proteinaceous particles more efficiently under conditions of greater inorganic phosphorus availability (unlike the conditions in the northern Adriatic Sea) was tested in this research. Consequently, lower quantities of TEPs and CSPs would be produced than under phosphorus-limiting conditions, where accumulation of TEPs and CSPs might occur. A further goal of this study was to examine the bacterial potential for producing CSPs and to test the hypothesis that the addition of ammonium (as a source of nitrogen) to bacteria would cause less degradation of these particles, with consequent accumulation of CSPs. In addition, an attempt was made to measure the impact of cyanobacteria on particle dynamics.
Materials and methods
Surface seawater was collected in June 2004 off the west Istrian coast, Croatia, in the northern Adriatic Sea in polycarbonate NALGENE containers, and filtered through polycarbonate filters with a pore size of 1 μm, allowing collection of only picoplankton in the filtrate. The filtrate was partitioned between eight 20 L transparent polycarbonate bottles, with 12 L placed in each. Previously, the bottles were filled with 0.5% HCl and rinsed with distilled water. After they had been filled with filtered seawater, three different enrichments were performed. Two enclosures were supplemented with 10 μM ammonium, two enclosures with 1 μM orthophosphate, and two enclosures with both 10 μM ammonium and 1 μM orthophosphate. The remaining two bottles were not supplemented with nutrients, and served as control bottles. One bottle from each pair was kept in darkness, and the other was exposed to normal daylight cycles. The first sampling was done immediately after the addition of nutrients; sampling was then done after 8, 12 and 24 h and then every second day, resulting in a total of eight sampling events.
Samples for TEP concentration determination (20 mL) were preserved with formaldehyde (final concentration 2%) and stored in a refrigerator (4°C) until determination at the end of the experiment. Samples were filtered through polycarbonate filters (0.4 μm), stained with a prefiltered (0.2 μm) 0.02% solution of Alcian blue stain (8GX, Sigma) in 0.06% acetic acid (pH 2.5) for 3 s, and transferred into beakers with 80% H2SO4 (Passow & Alldredge, 1995). The absorption of the solution was measured in a spectrophotometer at 787 nm against distilled water as a reference. The concentration of TEPs was expressed as Xanthan Gum equivalent (μg L−1):
where Asample is the absorption of the sample, Ablank the absorption of the blank, Vf the filtered volume of the sample, and fc the calibration factor. Calibration was performed with Xanthan Gum as an equivalent and calculated according to the formula:
where W is the dry weight of Xanthan Gum per litre, Aequiv its average absorption, Ablank absorption of the blank, and Vequiv the volume of the filtered equivalent.
After filtration, one subsample's filtrate was collected and subjected to turbulent conditions (200 r.p.m. for 5 min). These conditions were chosen on the basis of the preliminary experiments with several variations of r.p.m. and duration. After mixing, the filtrate was again passed through a 0.4 μm filter, and the staining and standard calibrating procedure was repeated in order to determine whether any new particles had formed.
Visualization of TEPs and CSPs was achieved by staining with Alcian blue, prepared as previously described, and Coomassie brilliant blue (G-250, Sigma), respectively. A Coomassie brilliant blue stock solution (1% w/v) in sterile Mill-Q water was prepared. A working solution was prepared daily by diluting the stock solution in filtered (0.2 μm) sterile distilled water to 0.04% and passed through 0.2 μm filters. Seawater samples were passed through a polycarbonate filter (0.4 μm) at low vacuum (<50 mmHg). These filters (for determination of TEP abundance and area) were stained with 500 μL of Alcian blue solution for 3 s, upon which the dye was vacuum filtered (Alldredge et al., 1993). Filters for the determination of CSP abundance and area were stained with 350 μL of a working solution of Coomassie brilliant blue for 30 s (Long & Azam, 1996). Afterwards, the filters were transferred to Cyto-clear slides and stored at 4°C until microscopic examination (Logan et al., 1994). At least 50 particles were counted, and the area of the particles was measured (by putting them into one of two classes: circle or rectangle) on each slide, with each sample being measured in triplicate.
For bacterial abundance determination, 2 mL of samples were stained with 4′,6′-diamidino-2-phenylindol (1 μg mL−1 final concentration) for 10 min, and then passed through 0.2 μm black polycarbonate filters (Nucleopore). UV excitation with epifluorescent microscopy (Leitz Laborlux D) was used for counting (Porter & Feig, 1980). Seawater samples (15 mL) were passed through black filters (0.6 μm), and cyanobacteria were counted under green light excitation with an epifluorescence microscope and distinguished by the orange autofluorescence of phycoerythrin (Takahashi et al., 1985); staining with Primulin for 15 min was used for picoflagellate and nanoflagellate counting under blue excitation with an epifluorescence microscope (Caron, 1983). All samples were measured in triplicate.
3H-Labeled leucine (specific activity >140 Ci mmol−1, 20 nM final concentration) incorporation was also measured in triplicate. Subsamples (1.7 mL) were incubated for 1 h at the same temperature as used in sampling bottles. The incubation period was terminated by adding 100% trichloroacetic acid (TCA), upon which the labeled material was extracted with cold 5% TCA and collected by centrifugation. Radioactivity was measured with a liquid scintillation counter (Kirchman et al., 1985; Smith & Azam, 1992).
Ectoenzymatic activities were measured using six artificial fluorogenic substrates (Hoppe, 1983; Hoppe et al., 1988). For determination of phosphatase, α-d-glucosidase, α-d-galactosidase, β-d-glucosidase, β-d-glucopyranosidase and aminopeptidase activities, 4-methylumbelliferyl phosphate (final concentration 125 μM), 4-methylumbelliferyl-α-d-glucoside (final concentration 25 μM), 4-methylumbelliferyl-α-d-galactoside (final concentration 25 μM), 4-methylumbelliferyl-β-d-glucoside (final concentration 25 μM), 4-methylumbelliferyl-β-d-glucopyranoside (final concentration 25 μM) and l-leucine-4-methylcoumarinile-7-amide (final concentration 250 μM), were used, respectively. The fluorescence of triplicates was measured fluorometrically immediately after substrate addition and after incubation (of up to 1 h), under an excitation wavelength of 365 nm, using the emission at 455 nm for methylumbelliferyl and excitation at 380 nm and emission at 440 nm for methylcoumarinile. Specific enzymatic activity was obtained by dividing the average enzymatic activity per liter by bacterial abundance per liter, and standard deviation was calculated with the equation:
where SDx is the SD of enzymatic activity (μmol L−1 h−1), x is the average enzymatic activity, y is the average bacterial abundance, and SDy is the SD of bacterial abundance.
Standard oceanographic methods for determination of nutrients (orthophosphate, nitrite and nitrate) in seawater were used (Parsons et al., 1985), based on the formation of stained complexes and subsequent measurement of the intensity of staining by spectrophotometry at an appropriate wavelength, using distilled water as a reference. Ammonium was measured according to a modification of Solorzano's method (Ivančić & Degobbis, 1984).
For statistical analyses of the data, nonparametric Friedman and Wilcoxon tests were used, considering probability levels P<0.05 as significant.
Orthophosphate concentration was low in the control bottles during the entire experiment, approaching the detection limit of the method (0.03 μmol L−1), whereas the initial ammonium concentrations were 1–2 μmol L−1 and gradually reached 4–5 μmol L−1 by the end of experiment (Fig. 1). Large variations in orthophosphate and ammonium concentrations were observed in the enriched bottles. In the orthophosphate-enriched bottles, the orthophosphate concentration decreased during the first 24 h from the initial 1 to c. 0.5 μmol L−1 and remained at that level for the next 3 days. Thereafter, phosphorus regeneration occurred, and the concentration subsequently almost reached the initial level. In ammonium-enriched bottles (12 μmol L−1 initially), bacteria had already used 5–6 μmol L−1 within the 2 h needed for the initial measurement. After the 3–4 days, ammonium in ammonium-enriched bottles started to regenerate rapidly, whereas in all other bottles ammonium increased from the seventh day (Fig. 1). Nitrate and nitrite concentrations were found to remain constant in all bottles (2–4 μmol L−1 and c. 0.1 μmol L−1, respectively; data not shown).
Bacterial abundance and leucine incorporation rate
Heterotrophic bacterial abundances followed similar dynamics in all bottles, irrespective of enrichment (Friedman test: F=3.15, P=0.369; Fig. 2). During the first 24 h, abundances increased by three- to fivefold. A further, slower increment took place for 3 more days, and on the seventh day abundance sharply decreased in all bottles. The rate of leucine incorporation reached its maximum 8 h from the beginning of the experiment, with significantly higher values in bottles enriched with orthophosphate and both orthophosphate and ammonium (Friedman test: F=15.83, P=0.001; Fig. 2). After this, leucine incorporation progressively decreased, reaching values similar to the initial ones on the seventh day, with the simultaneous regeneration of nutrients.
Cyanobacterial abundance started to decrease after the first day in all conditions from 2–3 × 107 L−1, and eventually reached c. 5 × 106 L−1 from the seventh day on (data not shown). Moreover, their abundance was not significantly higher in the day–night cycle enclosures compared to the enclosures kept in darkness (Wilcoxon test, P=0.640), and also did not differ between the different nutrient regimes (Friedman test: F=3.53, P=0.318). As cyanobacterial abundance did not show any response to different nutrient or light regimes, and was two to three orders of magnitude lower than bacterial abundance, we concluded that this parameter did not have a significant role in the dynamics of the analyzed particles.
Throughout the experiment, the presence of picoflagellates and nanoflagellates in the samples was checked, and was not detected.
At the beginning of the experiment and during the first 3 days, specific phosphatase activity was high in all bottles, although, as expected, it was lower in bottles enriched with orthophosphate than in nonenriched bottles (Friedman test: F=13.11, P=0.004). Specific phosphatase activity decreased from the initial measurement (Fig. 3), with markedly lower values from the seventh day in all enclosures, and was related to the decline in bacterial and cyanobacterial abundances and bacterial activity as well as to orthophosphate regeneration. Cell-specific aminopeptidase activities, after initial increases, remained relatively constant in all bottles until the ninth day, when the highest values were measured (Fig. 3). These activities were higher in the bottles with elevated orthophosphate concentrations during the first 7 days (Friedman test: F=12.12, P=0.007). Among the analyzed glucosidases, no significant difference was found between different enclosures for α-glucosidase (Friedman test: F=2.26, P=0.519) or α-galactosidase (Friedman test: F=2.03, P=0.566). The latter showed increased activities from the initial zero to 0.4 amol cell−1 h−1 in some cases (Fig. 4). Data for β-glucosidase and β-glucopyranosidase activities are not shown, since they were similar in all conditions (Friedman test: F=0.40, P=0.934, and F=3.40, P=0.180, respectively), showing a decrement after the highest initial measurement from 0.3–0.6 to c. 0.1 amol cell−1 h−1 by the second day, and staying at that level in all bottles until the end of the experiment.
Particle abundances, total surface areas and concentrations
TEP abundances showed large variability (5 × 103–15 × 105 L−1), especially in bottles with only ammonium or only orthophosphate added, in which their increments were periodically noted during the experiment, including at the end. TEP abundance in these bottles was significantly higher than in control bottles or in bottles enriched with both nutrients (Friedman test: F=7.63, P=0.022; Fig. 5). CSP abundance was about twofold higher than that of TEPs during the entire experiment, and this pattern was more pronounced in the control bottles than in the other ones. The abundance in control bottles was somewhat higher than in others, although the difference was not highly statistically significant (Wilcoxon test: P=0.06–0.07; Fig. 5). Total TEP surface area was similar in all bottles and slowly decreased towards the end of the experiment (Friedman test: F=2.03, P=0.567), whereas the total CSP area periodically attained values up to 10-fold greater than that of TEPs in the control and ammonium-enriched bottles (Fig. 6). CSPs displayed lower surface area values in bottles with added orthophosphate or both orthophosphate and ammonium (Friedman test: F=10.88, P=0.012), although they still had higher surface areas than the corresponding TEP values.
The highest TEP concentration (c. 450 μg L−1) was measured in the control bottles (Fig. 7), but values during the experiment were not significantly different than in other bottles (Friedman test: F=3.00, P=0.223). After 2–3 days, absorption values decreased for all bottles and became constant. The concentrations of potential TEPs, which were produced after placing the <0.4 μm filtrate in turbulent conditions, had the sharpest peaks in the enrichments with one nutrient, but without statistical significance of differences between their general dynamics (Friedman test: F=1.00, P=0.801).
TEP abundance (>4 μm) in this experiment was in good agreement with the results of previous experiments on northern Adriatic bacterioplankton (105 L−1; Radićet al., 2003), and TEP concentrations were very similar to those found in other experiments with bacterial cultures (up to c. 400 μg L−1) (Stoderegger & Herndl, 1999; Passow, 2002a). Although the estimated contribution of bacteria in producing TEPs in the North Sea was 1–2% (Stoderegger & Herndl, 1999), TEP concentrations and abundances in our experiment are close to, or even surpass, average values from the actual northern Adriatic ecosystem: c. 300 μg L−1 (Radićet al., 2005) and 105 L−1 (Radićet al., 2003) respectively. Considering the presence of phytoplankton and other microorganisms as sources of TEPs in the natural ecosystem, these results confirm the importance of bacteria in TEP formation. To our knowledge, there has been only one previous report on bacterial ability to cause CSP formation (Bhaskar et al., 2005), which in this study was several times greater than that for TEPs. Although care must be exercised in relating this result to a natural system, due to different ecological conditions and the confinement of the experimental bottles compared to the in situ situation, this rather high potential of bacteria for TEP and CSP production might have important implications for interpretation of their dynamics in the sea.
Although in this experiment both initial ammonium and orthophosphate uptake were rapid in the enriched bottles, these enclosures attained practically the same bacterial abundance as the control bottles to which nutrients had not been added (Fig. 2). This apparent insensitivity to the enrichment regimes may indicate that control by predators in the original seawater sample was more important than nutrient levels. Removing the predators thus enabled similar growth of bacterial populations in all investigated conditions. Degradable organic matter was apparently sufficiently abundant in the initial water to support bacterial carbon demand, and high phosphatase activity ensured a supply of inorganic phosphorus, even in the control bottles (Fig. 3). However, some significant differences between treatments were noted in bacterial activity. The sharpest increments in leucine incorporation after the initial measurements were observed in treatments where only orthophosphate or orthophosphate and ammonium were added (Table 1). In these conditions, specific aminopeptidase activities were also higher than in other bottles (Table 1; Fig. 3). These enclosures with elevated orthophosphate concentrations were obviously the places where organic matter rich in nitrogen was more intensively degraded by bacteria. This was also evident in the reduced surface areas of CSPs and somewhat less significantly reduced CSP abundance (Table 1). However, aminopeptidase activity also did not stop in bottles enriched with ammonium, confirming that amino acids were utilized as a nitrogen source, as well as ammonium and nitrate (Kirchman et al., 1989; Kirchman, 2000).
Table 1. Summary table: major influences of different nutrient regimes on the bacterioplankton production of the polysaccharidic and proteinaceous particles; values significantly higher (+) or lower (−) compared to the control bottles
CSP surface area
On the other hand, the activities of four analyzed glycolytic enzymes were essentially the same in all bottles irrespective of the enrichment regime, in contrast to the initial hypothesis that addition of phosphorus will make possible more intensive degradation of polysaccharides. All glycolytic enzymes, however, did show marked changes in activities during the experiment, similar in all bottles. The reason for such changes may be related to bacterially derived dissolved organic matter attaining predominance after the initial conditions, where more dissolved matter of different origins was present. Such changes in the organic matter composition influenced decreases in β-glucosidase and β-glucopyranosidase activities from initial values, and increases in α-galactosidase activities from mostly zero to high values in all bottles (Fig. 4). That is not surprising, considering that galactose comprises a greater part of bacterial cells than in phytoplankton (Hoagland et al., 1993; De Philippis & Vincenzini, 1998; Bhaskar & Bhosle, 2005). Considering the prevalence of high molecular weight organic matter in bacterial exudates, primarily highly hydrated polysaccharides (Heissenberger & Herndl, 1994), high hydrolytic enzyme activities were expected. Likewise, α-glucosidase, β-glucosidase and phosphatase activities were one to two orders of magnitude higher than usual in the northern Adriatic, whereas aminopeptidase activity was similar to those previously reported in this area (Karner et al., 1992; Corinaldesi et al., 2003; Danovaro et al., 2005).
Intensive TEP formation occurred in parallel with the exponential growth phase in bacterial populations in all settlements, with similar values noted between enclosures. Organic matter in surrounding water, derived from dying cells from the fourth day, did not enhance the TEP concentration. This could be explained by different qualities of produced organic matter in the exponential and death phases (Nester et al., 1998). The changed chemical and physical structures of excreted polymers during the death phase of the population may be the reason for the change in TEP dynamics towards the end of the experiment. During this phase, different patterns of TEP concentration and TEP abundance occurred. TEP abundance showed large variations with periodically significant increases (Fig. 5), unlike TEP concentration (Fig. 7), particularly in bottles enriched only with ammonium or only with orthophosphate. This uncoupling between Alcian blue absorption (TEP concentration) and TEP abundance may be explained by the different physical structure of the polysaccharide network in these particles, which absorbed considerably less Alcian blue dye compared to that noted during the first days of the experiment.
Enclosures where only one nutrient was added were also the places where the highest values for potential TEPs were measured (Fig. 7). Potential TEPs may be defined as polysaccharides present in seawater that have not yet, but potentially may, form TEPs. Potential TEPs were quantified by placing the filtrate remaining after TEP filtration in turbulent conditions and subsequently measuring any new TEPs that had formed. Although potential TEPs gained much higher values in settlements with only one nutrient added, it did not seem to affect, or contribute to, greater TEP concentrations in these bottles. As glycolytic activities in all enclosures were similar, TEPs in bottles enriched with only one nutrient were obviously not degraded more efficiently. However, this glycolytic activity might have been directed more towards the transformation of the precursor polysaccharides, and/or they might have aggregated into TEPs in a different way than in control or nutrient-balanced bottles, manifesting in increased TEP abundance (Table 1). Interestingly, conditions such as those in enclosures with only ammonium added, with high N/P ratios, which caused large increases in TEP abundance, are often found in the northern Adriatic basin (Degobbis et al., 2000, 2005).
Whereas TEPs exhibited more intensive formation in conditions where only ammonium or orthophosphate was added, CSP formation was more intensive in enclosures without additional orthophosphate (Figs 5 and 6 and Table 1). This supports the hypothesis of greater degradation of organic matter after enrichment with phosphorus in the case of CSPs. Another hypothesis of the study was that addition of ammonium to bacterial isolates would direct bacterial preference towards this inorganic source of nitrogen rather than hydrolysis of organic nitrogen compounds. In that case, accumulation of proteinaceous particles (CSPs) may have occurred. Our results indicate that orthophosphate had a much more important role in controlling CSPs, as their abundances and total surface areas were generally lower in bottles enriched with phosphate, or both phosphate and ammonium, than in other bottles. Accordingly, aminopeptidase activities were higher in bottles with added orthophosphate, as well as leucine incorporation. On the contrary, high ammonium concentrations did not cause greater CSP accumulation than in the controls.
Formation of CSPs is investigated rarely in marine ecosystems, and without any clarification of the bacterial role in this process. Our results show the ability of bacteria to produce CSPs, whether by directly producing particles or excreting dissolved matter that can later form particles. Bacterial capability in producing CSPs results in CSP abundances and total areas several times higher than for TEPs. Both CSPs and TEPs increased considerably during, or 1–2 days after, the occurrence of maximum bacterial abundance.
Apart from heterotrophic bacteria, the dynamics of the cyanobacterium Synechococcus were also followed. Synechococcus is phosphorus-limited in Mediterranean waters (Vaulot et al., 1996), although populations may grow even with only nanomolar concentrations of orthophosphate (Ikeya et al., 1997). However, in this experiment, the populations of these microorganisms stopped growing after the first day, even in conditions of 1 μmol L−1 orthophosphate. Thereafter, their number remained constant or decreased, even in bottles that were exposed to the day–night light regime. Jacquet et al. (2002) stated that in conditions where organic matter and inorganic nutrients are available, Synechococcus is outcompeted by bacteria for inorganic nutrient resources. The absence of expected growth may also be caused by virus infection, an important control factor for Synechococcus, since many species of this genus are known as cyanophage domadars, particularly in conditions of abundance greater than 106 L−1 (Wommack & Colwell, 2000; Mühling et al., 2005). The activity of aminopeptidase, as the only ectoenzyme that these species use (Martinez & Azam, 1993), was also the same in the dark as in the light. As cyanobacterial abundance was two to three orders of magnitude lower than bacterial abundance and decreased with time, irrespective of light and nutrient regimes, it may be concluded that it was not an important factor in particle dynamics in this experiment.
This experiment confirms significant bacterial potential in producing TEPs. It also highlights the bacterial potential for causing CSP formation with abundances and surface areas several times greater than those of TEPs. According to the available literature, little has been done up to now in studying CSP production by bacteria. Considering TEPs and CSPs larger than 4 μm, bacteria-derived organic matter may result in 104–105 L−1 TEPs and CSPs. Addition of orthophosphate resulted in higher degradation of CSPs by bacteria, whereas enrichment with ammonium did not prevent organic nitrogen compound degradation and hence did not cause enhanced accumulation of CSPs as an organic source of nitrogen. Although phosphorus enrichment affected CSP dynamics, it did not enable greater TEP degradation. Unbalanced nutrient enrichments were more important for TEP dynamics, and resulted in a periodic increase in the abundance of TEPs with low Alcian blue absorption capacity during the bacterial decline phase. Inorganic nutrient availability was thus shown to affect the bacterial potential for producing and degrading bacterially derived TEPs and CSPs. This is of importance for the interpretation of the dynamics of these abundant particles in the sea, especially in marine ecosystems such as the northern Adriatic Sea, characterized by periodically intensive organic matter accumulation.
We thank Dr D. M. Lyons for improving the English of the manuscript and Dr N. Supić for her support. This study is part of T.R.'s PhD thesis and was funded by the Ministry of Science, Education and Sports of the Republic of Croatia (Grant No. 0098111).