The lagoon of Venice (https://deims.org/f7d94927-17be-4d3d-9810-e3c9bc91829c) is a polyhedric shallow water body located in the northern Adriatic Sea which has a water surface of ca. 432 km² and a mean depth of ca. 1.2 m (Fig. 1).The lagoon is connected to the sea through three large (400–900 m) and deep (15–50 m) mouths which divide it into three hydrological basins separated by watersheds which shift according to tides and winds. The present study refers to the three morphological basins: northern, central and southern lagoon. Burano and Torcello tidal marshes mark the separation between the central and the northern basin and the deep Malamocco-Marghera artificial canal the separation between the central and the southern one. The central basin (ca. 132 km²) has been the most studied because it suffered the impact of industrial waste, urban sewage and other anthropogenic pressures such as commercial and touristic activities and the illegal fishing of Manila clam (Tapes philippinarum Adams & Reeve) by disruptive fishing gears (Sfriso et al. 2003).

Figure 1.

Map of the Venice Lagoon.

Information on data collected before 1980 is reported in the cited papers whereas the surveys carried out by our research team, sampling procedures and analytical methods are summarized in the following pages.

Data on nutrients, macroalgal biomass and environmental parameters of the water column have been collected in the whole central lagoon since 1987 (34 sites). Monitoring surveys were carried out in 1993 (34 sites), 1998 (52 sites), 2003 (65 sites), 2011 (45 sites) and 2014 (34 sites). Data of ammonium concentration were also reported for the whole lagoon in 2011 (118 sites).

Thirty-four sites in early summer 1987, 1993, 1998 and 2003, and 31 sites in 2011 were monitored by collecting the 5 cm surface sediment top layer for phosphorus, total nitrogen analyses and determination of the sediment density.

At each site, dissolved oxygen was measured with a portable instrument (Oxi 196 oximeter, Wissenschaftlich-Technische Werkstätten GmbH, Weilheim, Germany). Oxygen concentrations were reported as percentage of saturation (%DO) taking into account water temperature and salinity. Water transparency was measured with the Secchi disk. Transparency measurements were reported as a percentage of water column visibility because the waters are shallow (0.5–2.5 m) and the tidal excursion is relatively high (0.5–1 m). A value corresponding to 100% means that the bottom was visible, 50% means that the disk disappeared half way to the bottom. pH and Eh were measured with a portable pH-meter (pH 25) of the CRISON Instruments (Barcelona, Spain).

Six water column samples were collected with a home-made cylindrical sampler (length: 1.50 cm, diameter: 4 cm) which was repeatedly plunged into the water and poured to a tank. Sub-samples of 0.1–1.0 L were filtered through GF/F Whatman glass fiber filters (porosity: 0.7 mm). Filters and water samples were stored frozen at -18 °C for chlorophyll-a (Chl-a) analyses (Lorenzen 1967) and nutrient (reactive phosphorus (RP) and dissolved inorganic nitrogen (DIN) as the sum of ammonium, nitrite and nitrate) determination according to Strickland and Parsons (1984). Other water sub-samples were collected for salinity, which was determined in the laboratory by chlorine titration according to Oxner (1962).

At each site three sediment cores (first 5 cm top layer) were collected with a Plexiglas corer (i.d. 10 cm) and carefully mixed together. Sub-samples were retained to determine density and nutrient (total nitrogen and total, inorganic and organic phosphorus) concentrations. Density was obtained as g cm^-3 of wet and dry sediment according to Sfriso et al. (2003).

Total nitrogen (TN) concentrations were obtained by a Flash 2000 CHNS Analyser (Thermo Fisher Scientific spa), sediment freezing, lyophilization and pulverization. Inorganic (P_inorg) and total phosphorus (P_tot) were, respectively, determined before and after combustion at 550 °C for two hrs, with dissolution in 1N HCl and spectrophotometric measurements according to Aspila et al. (1976). Organic phosphorus (P_org) was obtained by difference. All the analyses were carried out in duplicate on two different days and values were retained when the difference was <5%. To compare the nutrient concentrations of the surface sediments of each site, values were normalized with reference to the amount of dry sediment per volume unit (dry density).

The macroalgal biomass was sampled between late May and July, which is the period when it is most abundant. At each site, biomass samples were collected with a 71×71 cm square frame or with a rake when the biomass was low (3–6 replicates) according to Sfriso et al. (1991, 2014a). Samples were weighted with a mechanical/electronic balance according to the biomass amount. The number of sampling sites ranged from 65 (1987), when macroalgae were very abundant and widespread, to 34 (2014).

Data of different datasets of the whole central lagoon have been compared and means, standard deviations, maximum and minimum values have been determined. The Shapiro-Wilk test differentiated non-normal data, and then Friedman one-way ANOVA values (p <0.05) and Spearman’s non-parametric coefficients have also been calculated. The basic statistics and correlation analyses of the data collected in the 5 periods were carried out both separately (1987, 1993, 1998, 2003, 2011) and by considering the whole data set (1987–2011). The total information has been summarized in a table reporting the number of significant (p<0.05) correlations per single parameter and period. The principal component analysis (PCA) has been applied to log-transformed data to visualize the variance and the association between parameters of both single and total periods, and to explore the scores with a loading >0.7, which is the generalized standard used for this statistical analysis. Total data are visualized in a bi-plot, whereas the loadings >0.7 of the single variables of each period are reported in Table 5 showing the parameters with the highest variance in decreasing order.

The similarities and differences between the stations during the 5 sampling periods have been investigated by analyzing the same data in a transposed matrix and bi-plotting the results. The stations are grouped according to their ecological characteristics and comparisons between the separated or overlapped periods highlight their differences.

Data were processed by Statistica software, Release 10 (StatSoft Inc. Tulsa, USA) provided by an academic license.

The first detailed distribution of macroalgae over the whole lagoon took place in 1980 and was replicated in 2003 (Sfriso and Facca 2007). During this period the biomass decreased by ca. 90% in the whole lagoon and by over 96% in the central lagoon. However, additional information is available for the central lagoon (ca. 132 km²) because the biomass was mapped in 1987, 1993, 1998, 2003, 2011 and 2014, together with some physico-chemical parameters and the concentrations of nutrients in the water column and surface sediments. The highest mean biomass, mainly represented by Ulva rigida C. Agardh and Ulva laetevirens Areschoug, was recorded in late spring in 1987. It showed a mean value of 4.78 kg FW m^-2 (65 sites, Fig. 2) and a density that in many sites ranged between 10 and 20 kg FW m^-2, reaching even 25 kg FW m^-2 in one site. The mean biomass markedly decreased to 0.69 kg FW m^-2 in 1993 and to 0.11 kg FW m^-2 in 1998 (one-way ANOVA: p<0.001). In the following years, the biomass slowly increased, reaching a mean value of 1.05 kg FW m^-2 in 2014 (one-way ANOVA: p<0.001). In the same periods, phytoplankton, recorded as Chl-a, was expected to increase. On the contrary, the mean concentration of Chl-a showed similar values in 1987 (3.61±5.62 µg dm^-3), 1993 (4.01±5.43 µg dm^-3) and 1998 (3.36±4.45 µg dm^-3) and decreased significantly (1.32±1.58 µg dm^-3) in 2014 (one-way ANOVA: p <0.05) (Fig. 2).

Figure 2.

Trends of the mean, standard deviation, maximum, minimum and median values of the macroalgal biomass, Chl-a, dissolved oxygen, pH and water transparency in the central lagoon of Venice.

The environmental parameters mainly affected by the presence of macroalgae (i.e. the concentration of dissolved oxygen, pH and water transparency) showed a strongly related trend to the biomass variation (Fig. 2). In 1987, due to the high presence of macroalgal biomass, the mean %DO saturation was 274%, ranging between 159 and 394%. In 1993, when the biomass decreased significantly, the mean %DO halved to 135% (one-way ANOVA: p<0.001) and in the following periods fluctuated between 113 and 150% (one-way ANOVA: p<0.001). In 1987, the lowest %DO was higher than the mean values of the following periods. Moreover, since 1993 the lowest values have always been undersaturated. The pH values showed a similar decreasing trend (Fig. 2) with the highest mean value in 1987 (8.81) and the lowest one in 2003 (7.64) (one-way ANOVA: p<0.001). In contrast, the presence of high biomass, especially represented by laminar Ulvaceae that hamper sediment resuspension, enhanced water transparency which decreased significantly in 1998 and 2003 when the effort of clam harvesting was at its highest.

Changes of nutrient concentrations in the surface sediments of the Venice lagoon are reported in Table 1. The first data of total phosphorus (P_tot) and total nitrogen (N_tot) in the surface sediments date back to 1948–49 (Perin 1975, Perin et al. 1983), before the 2^nd post-war industrial development. The highest changes were recorded for P_tot concentrations. During the period 1948–49, P_tot presented very low values: 24±16 µg g^-1 (mean±SD). Twenty years later, in the period 1968–73, this value increased remarkably (164±79 µg g^-1) and even more in 1983 (454±126 µg g^-1), when it reached the highest mean concentration, i.e. 18.9 times the concentration measured in 1948–1949. In the following years, P_tot mean concentration decreased slightly, showing average values between 339 and 375 µg g^-1 (Cossu and De Fraja-Frangipane 1985; Sfriso et al. 2003, Facca et al. 2014). The maximum value was recorded by Cossu and De Fraja-Frangipane (1985): 1102 µg g^-1. No record for the organic and inorganic fractions is available.

Total Nitrogen (N_tot) exhibited a mean value of 1.00±0.86 mg g^-1 in 1948–49, increasing up to 1.86±2.20 mg g^-1 in the period 1968–73, followed by a progressive decrease to 0.69 µg g^-1 in 2011. The maximum value was recorded in the period 1968–73 by Perin (1975) with 3.56 mg g^-1.

Some more detailed information is available for the central basin of the Venice lagoon where the same operators monitored the surface sediment top layer (5 cm) of 31–34 stations in successive late spring-early summer periods, characterized by different scenarios (1987: presence of high algal biomass; 1993: sharp reduction of macroalgal biomass; 1998: intense Manila clam (Tapes philippinarum Adams and Reeve) harvesting; 2003: the highest Manila clam harvesting; 2011: sharp reduction of clam stocks (from ca. 40,000 tonnes in 2010 to ca. 2000 tonnes in 2012) and decrease of clam fishing activities).

During the period between 1987 and 2011 the mean concentrations of P_tot per volume unit (µg cm^-3 of sediment) did not change significantly, except for the maximum value of 720 µg cm^-3 in 1987, then progressively decreased to 473 µg cm^-3 (-34%) in 2011 (Table 2). In contrast, the mean value of the organic fraction (P_org) lowered by ca. 40% (from 104 to 62 µg cm^-3, one-way ANOVA: p<0.001). On the whole, the ratio P_org/P_tot decreased from 26.9% to 16.2% and in 2011 the maximum P_org concentration was more than halved (from 246 to 113 µg cm^-1).

In the same period (1987–2011), the mean N_tot concentration decreased from 1.21 to 0.35 mg cm^-3 (ca. -71.1%, one-way ANOVA: p<0.001) whereas the peak concentration lowered by 32%, from 3.00 to 2.05 mg cm^-3, although the lowest value was recorded in 1998 with 1.37 mg cm^-3.

Figs 3, 4 highlight the variation of P_org and N_tot loads in surface sediments of the central lagoon. In the first survey (1987) organic phosphorus showed high concentrations in 4 areas, one of which was affected by the industrial waste of Porto Marghera and the others by the urban discharges of Venice, Lido, Mestre and its hinterland (Fig. 3). In the following years, P_org concentrations decreased progressively and in 2011 the highest values were recorded along the mainland coast. Total nitrogen showed a temporal decrease even greater. Except for a station located at the north of the island of S. Erasmo, it generally decreased below 1 mg cm^-3 (Fig. 4).

Figure 3.

Maps of the organic phosphorus distributions in the 5 cm surface top layer of the central lagoon basin in 1987, 1993, 1998, 2003 and 2011.

Figure 4.

Maps of the total nitrogen distributions in the 5 cm surface top layer of the central lagoon basin in 1987, 1993, 1998, 2003 and 2011.

A significant decrease was also observed for the concentrations of nutrients in the water column Fig. 5). The first records date back to the period 1962–64 and deal with the concentration of ammonium in the canals of Porto Marghera industrial area and the lagoon surface in the vicinity. These values were very high (1000–2500 µM), reaching even 3800 µM in the industrial canals (Fig. 5). Ammonium concentration decreased to 500–1000 µM in the period 1970–72, lowering to 50–100 µM in 1985–8 (Fig. 5). In 2011 (Fig. 6) and 2014 the highest value recorded in that area was <10 µM.

Figure 5.

Distribution of ammonium in the water column in the central lagoon from 1962 to 1986 (from Cossu and De Fraja-Frangipane 1985).

Figure 6.

Map of ammonium distribution obtained for the whole lagoon by sampling 118 sites in spring and in autumn 2011.

By considering the whole central lagoon, information on the nutrient concentrations in the water column has been available since 1987. The mean concentration of reactive phosphorus (RP) decreased from 0.76 µM in 1987 (34 sites) to 0.19 µM in 2011 (45 sites), slowly increasing to 0.24 µM in 2014 (34 sites, Fig. 7) (one-way ANOVA: p<0.001). Conversely, the concentration of the dissolved inorganic nitrogen (DIN) was relatively low in 1987 (10.1 µM), in the presence of a very high macroalgal biomass, but it remarkably increased up to 27.6 µM in 1993 (32 sites) when macroalgae were negligible. Afterwards, DIN concentration progressively decreased to 5.9 µM in 2014 (one-way ANOVA: p<0.001). The mean concentrations of ammonium and nitrates showed the same pattern, increasing from 1987 (5.15 and 4.54 µM, respectively) to 1993 (12.2 and 14.2 µM, respectively) and decreasing in the following years to 2.01 µM (ammonium) and 3.51 µM (nitrates) (one-way ANOVA: p<0.001). Similarly, nitrites showed a mean concentration of 0.38 µM in 1987, increasing to 1.23 µM in 1993 and decreasing again to 0.37 µM in 2014 (one-way ANOVA: p<0.001).

Figure 7.

Trends of the mean, standard deviation, maximum, minimum and median values of the reactive phosphorus, dissolved inorganic nitrogen, ammonium, nitrate and nitrite concentrations in the central lagoon of Venice.

The Spearman non-parametric correlation matrices of data collected in the central lagoon were determined by considering both the whole period: 1987–2011 and the years 1987, 1993, 1998, 2003, 2011, separately. On the whole, the highest number of significant (p<0.05) direct or inverse correlations was shown by RP (13 out of 18), salinity, Chl-a and nitrite (12 out of 18) and pH (11 out of 18) Tables 3, 4). Results depended on the different scenarios. In 1987, the highest number of correlations was found with temperature; in succession, other parameters were nitrite, macroalgal biomass and some parameters associated with the presence of biomass such as %DO, P_tot and P_org accumulated in the surface sediments and DIN in the water column. In 1993, macroalgal biomass started to decrease significantly and in 1998, when the biomass was negligible and clam fishing activities were affecting a wide part of the central lagoon, the parameters with a high number of significant correlations were salinity (12), water transparency (8) and P_inorg, P_tot (10 and 8, respectively). In 2003, when clam fishing showed the highest efforts, water transparency (13) dropped down and high amounts of ammonium (15) and P_org (12) were released by the surface sediments. RP continued to decrease (12) in the water column whereas pH (13) showed the lowest values (Fig. 7). In 2011, the scenario changed and the number of correlations decreased (Table 4).

The principal component analysis highlighted the parameters with the highest variance (Table 5, Fig. 8). Their spatial distribution and association is displayed by bi-plotting the values of the first two components. The analysis of the whole dataset (1987–2011) shows the parameters with a loading >0.7 (Table 5). Among them, RP and nitrite (NO₂-) played a key role in the changes of the environment as well as %DO (O₂) and P_org (Fig. 8). Moreover, the macroalgal biomass was associated with %DO, pH, water transparency, salinity and sediment density, and opposed to nutrient concentrations, Chl-a and water temperature. The analysis of the parameters of each monitoring period is reported in Table 5. On the whole, salinity displayed a loading >0.7 each single year but the value changes when the whole period is considered. Similarly, nitrate and water transparency showed significant loadings (4 out of 5) in the single periods, but not when they referred to the whole period. Nitrite, pH, P_org loadings were significant in three periods and only P_org did not display any significant value. These trends are also highlighted when the significant values of each parameter are summed.

Figure 8.

PCA bi-plot of the first two components during the whole 1987–2011 period.

The bi-plot of the first two PCA components of the transposed matrix highlights the similarities/dissimilarities between the stations of the different sampling periods (Fig. 9). All the stations sampled in 1987 and 1993 are opposite to the ones sampled in 2011. In contrast, results of sampling carried out in 1998 and 2003 partly overlap the ones occurred in other periods highlighting that the environmental conditions in 1998 and 2003 were intermediate.

Figure 9.

Transposed PCA bi-plot of the first two components during the whole 1987–2011 period. The separation or overlapping of the different years are highlighted by different colours.