at SciVerse ScienceDirect

Estuarine, Coastal and Shelf Science 115 (2012) 63e74

Contents lists available

Estuarine, Coastal and Shelf Science

journal homepage: www.elsevier .com/locate/ecss

Dynamics of biogeochemical properties in temperate coastal areas of freshwaterinfluence: Lessons from the Northern Adriatic Sea (Gulf of Trieste)

Gianpiero Cossarini a,*, Cosimo Solidoro a,1, Serena Fonda Umani b,2

a Istituto Nazionale di Oceanografia e di Geofisica Sperimentale, OGS, Borgo Grotta Gigante 42/c, 34010 Sgonico (TS), ItalybUniversità di Trieste, via Valerio 28/1, 34127 Trieste, Italy

a r t i c l e i n f o

Article history:Received 14 August 2011Accepted 7 February 2012Available online 19 February 2012

Keywords:biogeochemical cycleclimatologycoastal zoneriver dischargeMarine ecologyAdriatic SeaGulf of Trieste

* Corresponding author.E-mail addresses: gcossarini@ogs.trieste.it (G. Coss

(C. Solidoro), s.fonda@units.it (S. Fonda Umani).1 Tel.: þ39 040 2140315.2 Tel.: þ39 040 5582007.

0272-7714/$ e see front matter � 2012 Elsevier Ltd.doi:10.1016/j.ecss.2012.02.006

a b s t r a c t

High spatial and temporal variabilities of biogeochemical properties are prominent features of regionsunder freshwater influence as a result of multiple factors. Understanding the ecological functioning ofthese ecosystems, which provide important services for humans, is challenging since it requires adequateobservational strategies and efforts.

Multi-years (1999e2006) continuous observations in the northernmost part of the Adriatic Sea (Gulfof Trieste) allowed us to compute a climatological description of seasonal dynamics of biogeochemicalproperties for three relevant sites: a coastal area directly influenced by a river, an off-shore area locatedin the centre of the Gulf and a coastal area located far from potential source of external nutrients.

The analysis of the climatologies provides a quantitative corroboration of the conceptual scheme forbiogeochemical and ecological seasonal dynamics of temperate coastal areas under freshwater influencealready proposed in literature, highlighting the role of river input, lateral transport, stratification regimeand interaction with bottom environment as driving factors.

While all areas follow a common pattern of succession of ecological processes, spatial variabilityaccounts for a significant decrease of the absolute trophic state, and for a phase delay in biogeochemicaldynamics. Results show that spatial heterogeneity is an inherent structural feature of coastal ecosystems,suggesting that the evaluation of the quality status of coastal ecosystems should be made by usingdifferent reference terms for different sub-areas.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Regions of freshwater influence (ROFI, Simpson, 1997) are sub-jected to variability from a range of sources that determine theirseasonal and interannual highly dynamic patterns. Seasonal, dailyand tidal cycles are among the more easily recognised sources ofnatural variability, but less regular influences can be also relevant.Riverine freshwater inputs and nutrient discharges alter thebuoyancy of the area, which favours water column stratification,enriches the environment with additional nutrients, and generatesdynamic fronts that divide the area into markedly different sub-areas. In addition to triggering surface circulation, winds can mixthe water column, bringing nutrient enriched bottomwater massesto the surface and homogenising the vertical distributions of dis-solved substances. Benthic pelagic coupling may influence bottom

arini), csolidoro@ogs.trieste.it

All rights reserved.

water properties. The presence of cities, which are often localisedalong the coast and close to river mouths, represents anothersource of anthropogenic pressures, such as urban-derived nutrientand pollutant loads, industrial wastes, fishery impacts and aqua-culture, tourism and other economic activities. Global changes,such as changes in run-off regimes (Howarth et al., 2000), aresuperimposed on the locally driven pressures along with environ-mental modifications caused by culturally driven policy strategies,such as oligotrophication (Stockner et al., 2000).

In such a heterogeneous context, paradigms forbiologicalephysical interactions that drive biogeochemicaldynamics in the open seas (e.g., the GraneSverdrup effect, Riley,1942; Sverdrup, 1953) do not always apply e mainly because ofthe prevalence of coastal inputs and lateral transport on verticalprocesses and because of the much shallower water depth e anddifferent physical and ecological structures and dynamics are inplace (Cushing, 1989; Legendre and Rassoulzadegan, 1995; MannandLazier,1998),whichoftendependon site specific characteristics.

Since the 1980s, possibly driven by problems related to coastaleutrophication and water quality, many studies have explored the

Fig. 1. Morphology of the Gulf of Trieste (a), position of sampling stations, and ofrelevant sites: FARCOAST (solid line box), OFFSHORE (dashed line box) and RIVER(dotted line box); monthly consistency of data at the nine sampling stations (b): blank,grey and black indicate no, one or more than one data per month at surface,respectively.

G. Cossarini et al. / Estuarine, Coastal and Shelf Science 115 (2012) 63e7464

relationships among macronutrient (nitrogen, phosphorus) loads,their ambient concentrations and the standing stock of phyto-plankton or macrophytobenthos (Nixon, 1995; Cloern, 2001). Morerecently, the attention of the scientific community has shifted tothe analysis of interannual changes and trends in the phenologyand composition of planktonic communities (Smetacek andCleorn, 2008). However, there have been relatively few studieson the seasonal dynamics of coastal systems from a multivariateecological perspective and that document, using long termobservations, the seasonal patterns in the structure and func-tioning of food web components as a function of the evolution ofenvironmental pressures and conditions (Legendre andRassoulzadegan, 1995).

Based on 3 years of biogeochemical data, Fonda Umani et al.(2007) and Solidoro et al. (2007) developed a conceptual schemeof the seasonal dynamics of the biogeochemical properties of theGulf of Trieste that may be applicable to other temperate ROFIs.According to this scheme, river inputs and wind conditions aremajor factors in defining the trophodynamics of these areas. Afterwinter, during which heterotrophic conditions prevail, the springbrings important outflows of river freshwater throughout thecoastal areas that favour the stratification of the water column and,depending on wind driven circulation, the nutrient enrichment inthe euphotic layer. Typically, a spring diatom-dominated phyto-planktonic bloom develops, which also results from the seasonalwarming and increase in photoperiod length, followed by bloomsof highly efficient, smaller autotrophic plankton. Beginning withthe reduction of river discharge in late spring, and possibly influ-enced by an increase in stratification, external nutrient supplies donot compensate the depletion in the pools of dissolved inorganicnutrients that are consumed by photosynthetic activity, particularlyin the surface layers. This depletion of nutrients causes both anincrease in the exudation of dissolved organic matter anda decoupling between primary production and bacterial carbondemand with a consequent accumulation of dissolved organiccarbon (DOC), which results also from degradation of particulateorganic carbon (POC). Concurrently, but more markedly at depth,bacteria mediated remineralisation of organic matter and themicrobial trophic web conveys most of the energy cycling in theecosystem. In late summer/early autumn, new river inputs andmixing of water column may trigger a second, less intense diatombloom, followed by increased microbial food web activity, andremineralisation prevails again in late autumn and winter.

The aim of this paper is to present a climatological annual cycleof biogeochemical properties in the Gulf of Trieste based on a largee 8 years e and partly still unpublished multivariate datasets, andto test against it the validity of the proposed conceptual scheme ofthe seasonal dynamics of biogeochemical properties in thistemperate ROFI area.

We hypothesise that the main biogeochemical patterns arecommon to the whole area but that there are differences amongsub-areas in terms of magnitude and timing, possibly as a result ofthe relative importance of the forcings that coexist (Solidoro et al.,2007; Lipizer et al., 2011).

The results provide reference values for future investigation andsite comparison, and point out that the ecological status of a givenarea cannot be evaluated by comparing biogeochemical parametersagainst a single value that is common to all seasons and sub-areas,because of the need to recognise the natural variability inherent inhighly dynamic environments.

The paper is organised as follows: the study site (Gulf of Trieste)and data are presented in Section 2, biogeochemical climatogies forselected sub-areas are described in Section 3 along with biologicalclimatologies at one only site and results of principal componentanalysis on biogeochemical data. In Section 4 the conceptual

scheme of ecological functioning of this ROFI area is discussedtogether with its comparisonwith the Adriatic Sea, and the effect ofdistance from river sources. Some conclusions are drawn inSection 5.

2. Data and methods

2.1. Study site

The Gulf of Trieste, the northernmost part of the Adriatic Sea(Mediterranean Sea), is a semienclosed coastal area withamaximum depth of less than 25 m (Fig.1a). The Isonzo River (withits mouth on the northwestern shore of the Gulf) is the majorsource of freshwater and nutrients (Cozzi et al., 2012). Freshwaterinput can either spread throughout the basin or be outwardtransported to the western end of the Gulf depending on thesurface circulation (Querin et al., 2006). Circulation is generallycyclonic, forced by the incoming Istrian coastal current at thesouthern border, but intense and frequent wind conditions (fromthe northeastern quadrant) produce an east-to-west current at thesurface layer (Malacic and Petelin, 2009).

2.2. Data collection

Data were collected at 9 sampling locations (see Fig. 1a), almostat monthly frequency, from 1999 to 2006. Coverage and frequencyof sampling vary among stations (Fig. 1b). Station T21, sometimereferred to also as C1, which presents the longest and the mostcomplete timeseries, was sampled almost 200 times (includingweekly and biweekly sampling in several periods) while stationT26, which was sampled regularly only since 2002, has only 33records. At each station water samples were collected at three orfour depths depending on the station: one at surface, one or two atintermediate depths and one near the bottom.

G. Cossarini et al. / Estuarine, Coastal and Shelf Science 115 (2012) 63e74 65

Temperature (TEMP), salinity (SAL), and oxygen concentrationand saturation (OXY) data were recorded using an Idronaut OceanSeven (model 401) or an SBE 19plus SEACAT Profiler multi-parametric probes, calibrated every 6 months.

Samples for dissolved organic and inorganic components werecollected with Niskin bottles, filtered on board (pre-combustedWhatman GF/F filters), and kept frozen (�20 �C) until laboratoryanalysis. Nitrate (N-NO3), ammonia (N-NH4), phosphate (P-PO4)and silicate (Si-SiO4), were determined colorimetrically witha segmented flow autoanalyzer according to Grasshoff (1983).

Dissolved organic nitrogen (DON) was calculated as the differ-ence between total dissolved nitrogen (measured as N-NO3 afterUV decomposition of organic matter) and total dissolved inorganicnitrogen. Similarly, dissolved organic phosphorus (DOP) wascalculated as the difference between total dissolved phosphorus(measured as P-PO4 after UV decomposition of organic matter andoxidation by addition of hydrogen peroxide) and phosphate. Dis-solved organic carbon (DOC) was determined with an HTCOmethod using a Shimadzu TOC 5000A (Cauwet, 1994).

Particulate phosphorus (PP) was determined as phosphateafter high temperature combustion (450 �C) and acid hydrolysis(1 M HCl) of organically-bound phosphorus compounds, accord-ing to Solorzano and Sharp (1980). Particulate organic carbon(POC) and particulate nitrogen (PN) were determined with a CHNElemental Analyzer Carlo Erba Model EA1110, after acidificationwith 1 mol L-1 HCl to remove the inorganic carbonate(Sharp, 1974).

Chlorophyll a (CHLA) was measured with a Perkin Elmer LS50Bfluorometer after extraction (90% acetone) and centrifugation ofsamples kept at dark, according to Lorenzen and Jeffrey (1980).Further details on chemical analytical methods are given in Lipizeret al. (2011, 2012).

DOP data earlier than 2002 and PP data earlier than 2000 werenot used in the present work, since analysis showed sharp changesbetween periods, which coincide with changes in monitoringprogram and possibly procedure or operator which would needa specific investigation.

With the same frequency but only at the station T21/C1 bio-logical data of autotrophic components (microphyto, autotrophicnano, and autotrophic picoplankton) and heterotrophic ones(mesozoo, microzoo, heterotrophic nano and heterotrophic pico-plankton) are available. We distinguish only diatoms from the restof the microphytoplankton. Water samples were collected by Nis-kin bottles at 4 discrete depths (0 m, 5 m, 10 m, near the bottom).We consider here only the surface (mean of the 0 m and 5 mdepths) and bottom climatology of their total biomasses. Micro-plankton and picoplankton samples were preserved with formalin(2% final concentration), nanoplankton with glutaraldehyde (1%final concentration). Microplankton was analysed at the Leitzinverted microscope (Utermohl, 1958), nanoplankton and pico-plankton at the epifluorescent one after staining the filters with 40,6diamidino-2-phenylindole (DAPI). Abundance data were convertedin biovolumes by measuring relevant dimension with an eyepiece,and then in biomass (carbon content) by using appropriateformulae. Mesozooplankton was collected by vertical tows fromnear the bottom to surface by using aWP2e 200 mmmesh size net.Samples were preserved with formalin (4% final concentration) andanalysed at the stereo-microscope. Mesozooplankton biomass wasestimated as carbon content, which was determined with aCHNS 2400 Perkin Elmer Elemental Analyzer. For more detailson sampling, conservation and analyses see Fonda Umani et al.(2012).

It is worth to note that, since the significant long term avail-ability of data, the Gulf of Trieste has been included in the LongTerm Ecological Research Site net (http://www.lteritalia.it/siti.php)

2.3. Statistical methods

According to geographic position (Fig. 1a) and results ofa previous study (Solidoro et al., 2007), the sampling stations weregrouped into 3 sets that represent different morphological andtrophodynamic conditions. This aggregation helps in filtering outthe differences in seasonal evolution observed in a given samplingstation in different years that arise from small differences in thespatial distribution of biogeochemical properties and thus itincreases the robustness and reliability of the analysis.

The sampling stations in the most freshwater influenced site(RIVER) include T24, T25 and T03, located in front of the Isonzo/So�ca River mouth, which is the major source of freshwater andnutrients in the Gulf (Cozzi et al., 2012). None of these stationsexceeds 10 m in depth.

Stations T21/C1, T22 and T23 belong to the FARCOAST site, whichis representative of a coastal area far from the river input.

The third site, OFFSHORE, consists of stations T08, T26 and T11,which are all located in the centre and deepest (up to 25 m) area ofthe Gulf.

For each site, the overall monthly climatology (i.e. referencevalues of the 12 months) was computed according to the followingsteps:

(1) For each station, a regular monthly timeseries was computed,averaging data in case of multiple samplings during a month;

(2) The climatology of each site was computed as the median of allmonthly data and stations belonging to the given site.Nonparametric statistics were used to compute the referencevalue of a given month so that no assumptions about the datadistribution were required.

Climatologies and interquartile range of monthly values arereported in the table in the Appendix.

Pearson correlations were computed among couples of siteclimatologies in order to assess a common seasonal pattern ofbiogeochemical variables within the ROFI area.

A principal component analysis (PCA, Davis, 1973) was appliedto each of the three surface sites separately, considering the N-NO3,N-NH4, Si-SiO4, P-PO4, CHLA, DOC, POC, DON, DOP, PN, PP and OXYmonthly climatologies, after the standardisation of the variables.For bottom-layer analyses, a PCA was performed for only the twodeepest sites.

3. Results

The physical and biogeochemical climatologies are reported inFigs. 2e5, which present the seasonal changes in organic carbonand chlorophyll a in the panel a, changes in nutrients in the panel b,changes in nitrogen and phosphorus particulate and dissolvedorganic compartments in panel c and changes in salinity, temper-ature and dissolved oxygen in the panel d. Climatologies arecomputed for the surface and bottom layers, except for site RIVER,which shows only the surface layer because of its shallowness.Values of monthly climatologies along with interquartile range aregiven in the appendix in order to be available as reference values forfuture studies. The seasonal changes in the biomass of the biolog-ical components at site T21/C1 are illustrated in Fig. 6.

3.1. Hydrodynamics variables

Temperature (TEMP) climatology, Figs. 2e5 panel d, depictsa well-marked seasonal cycle for both the surface and bottomlayers. In the surface layer, the maximum temperature is observedin July (up to 26 �C), and the minimum is in February

Fig. 3. Monthly evolution of biogeochemical properties at surface of the OFFSHORE site. Curves are the overall monthly climatologies.

Fig. 2. Monthly evolution of biogeochemical properties at surface of the RIVER site. Curves are the overall monthly climatologies.

G. Cossarini et al. / Estuarine, Coastal and Shelf Science 115 (2012) 63e7466

Fig. 4. Monthly evolution of biogeochemical properties at surface of the FARCOAST site. Curves are the overall monthly climatologies.

Fig. 5. Monthly evolution of biogeochemical properties at bottom of the OFFSHORE (blue lines) and FARCOAST (black lines) sites. Curves are the annual climatologies. (Forinterpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

G. Cossarini et al. / Estuarine, Coastal and Shelf Science 115 (2012) 63e74 67

Fig. 6. Monthly climatology for autotrophic and heterotrophic biological compartments for surface layer (a and b) and bottom layer (c and d) for station T21/C1. Right axes of plotsa and c refer to diatoms, right axes of plots b and d refer to heterotrophic picoplankton.

G. Cossarini et al. / Estuarine, Coastal and Shelf Science 115 (2012) 63e7468

(approximately 7.5 �C). No significant spatial difference is observ-able among the sites. The sites show some differences in the bottomlayer as a result of the differences in water column depths (Fig. 5d).Stratification, which lasts from April to September, morecompletely separates the surface and bottom layers at theOFFSHORE site than at the shallower FARCOAST site.

The seasonal cycle of salinity at the RIVER site (Fig. 2d) showswide variations from 32 in April to more than 37 in December. Asecond minimum occurs in October. In the other two sites (Figs. 3dand 4d), salinity values range between 35 (April) and 38 (Decemberand January), and no autumn minimum is detected. At the bottomlayer (Fig. 5d), salinity shows lower fluctuations, with valuesaround 38.

3.2. Nutrients

The surface annual cycle of nitrate and silicate (Figs. 2e4 panelb) is characterised by a maximum in April (up to 5e8 mM and up to3e6 mM for N-NO3 and Si-SiO4, respectively), a minimum insummer (down to less than 0.5 mM for both nutrients) anda secondarymaximum in autumn (approximately 2.5 mMand 4 mM,respectively). The RIVER site is characterised by values higher thanthe other two sites for both nutrients, with differences of 2e3 mM inwinter and 0.5e1 mM for the rest of the year for both nutrients. Inthe bottom layer (Fig. 5b), the annual nitrate cycle is relativelysmooth (concentrations higher than 1 mM are recorded only in theautumn months and in January), whereas the silicate annual cycleshows a significant increase from spring to autumn (up to 6 mM inOFFSHORE site in August and September).

Generally, ammonia reaches its maximum at the surface inwinter, in July and in autumn, whereas minima are noted in earlywinter and early spring (Figs. 2e4 panel b). Spatial variabilityamong the sites is not high (Table 1), and the sites have differentsequences of local minima and maxima in the winteresummer

period. At the bottom (Fig. 5b), seasonal evolution is characterisedby an increase in ammonia concentration from May untilAugusteSeptember. This feature is particularly relevant in thedeepest site (OFFSHORE, concentration up to 3 mM in August).

Phosphate, which is currently regarded as the limiting nutrientin the northern Adriatic Sea (Solidoro et al., 2007; Lipizer et al.,2011), ranges between 0.02 and 0.13 mM. The temporal variabilityof phosphate is higher than that of other nutrients (see Table 1 andTable in Appendix), and a seasonal pattern is not always recognis-able (Figs. 2e4 panel b). However, maxima are generally recordedin late autumn, and minima in the summer. At the bottom (Fig. 5b),the concentration of phosphate is similar to the surface one. Thetemporal pattern in this layer seems to be more definite than in theother layer: maxima occur in December, January and summer(particularly in the OFFSHORE site), and minima occur in March.Among the sites, the OFFSHORE one experiences the highestamplitude of the annual phosphate cycle.

3.3. Chlorophyll a and dissolved oxygen

In general, surface chlorophyll a (CHLA) seasonal changes(Figs. 2e4 panel a) show two peaks, in winter-early spring and inautumn (with a concentration up to 1.0 mg l�1), and a minimum insummer. However, spatial differences among sites are found withregard to the number and intensity of monthly peaks. At the RIVERsite, in particular, we can observe the highest peak in February andsecondary peaks in April, October and December. The OFFSHOREand FARCOAST groups present maxima in October, a secondarypeak in April and a third in February.

The annual change in CHLA in the bottom waters of theOFFSHORE and FARCOAST groups is mainly characterised bya single oscillation with a summer maximum and a winterminimum (Fig. 5a). Dissolved oxygen (not shown) presents clearseasonal patterns with a winter maximum (up to 6.5 cm3 l�1) and

Table 1Annual means and standard deviations of the monthly median values for the three sites and the two layers.

Surface layer Bottom layer

FARCOAST OFFSHORE RIVER FARCOAST OFFSHORE

Mean std Mean std Mean std Mean std Mean std

TEMP [�C] 16.0 6.3 16.5 6.4 16.2 6.4 14.4 5.3 13.9 4.7SAL 37.3 0.5 37.0 0.7 35.5 1.2 37.8 0.2 37.9 0.2N-NO3 [mM] 1.3 1.0 1.5 1.3 4.0 2.3 0.8 0.5 0.7 0.3N-NH4 [mM] 0.9 0.2 0.8 0.3 0.9 0.3 1.1 0.3 1.4 0.8Si-SiO4 [mM] 2.1 0.9 1.9 0.7 3.7 1.6 2.7 0.9 3.3 1.3P-PO4 [mM] 0.06 0.01 0.05 0.02 0.06 0.04 0.07 0.02 0.07 0.03DOC [mg C l�1] 1.2 0.2 1.1 0.2 1.2 0.2 1.1 0.2 1.1 0.1DON [mM] 11.8 1.9 11.5 2.3 13.0 3.2 12.0 2.2 11.2 2.3DOP [mM] 0.4 0.1 0.5 0.2 0.5 0.2 0.4 0.1 0.5 0.2POC [mg C l�1] 270.9 51.9 265.4 50.7 293.2 60.6 217.3 48.5 259.8 70.3PN [mM] 3.0 0.7 3.0 0.8 3.6 1.1 2.7 0.7 3.0 1.1PP [mM] 0.10 0.03 0.09 0.03 0.12 0.03 0.11 0.04 0.12 0.04CHLA [mg l�1] 0.7 0.3 0.6 0.4 0.8 0.3 0.8 0.2 0.8 0.3OXY [% SAT] 99.1 4.1 99.6 4.5 99.9 5.0 96.4 5.4 90.8 7.7

G. Cossarini et al. / Estuarine, Coastal and Shelf Science 115 (2012) 63e74 69

minima in late summer (down to 4.5 cm3 l�1), reflecting thetemperature-mediated solubility cycle. It is far more interesting toanalyse the deviation from saturation. Positive deviations fromsaturation (Figs. 2e5 panel d) occur during the maxima in May,when values up to 105% are registered in both layers and at all sitesexcept at the OFFSHORE one. Maximum under saturation occurs insummer (bottom) and autumn (surface). No occurrence of anoxia isevident. Only at the bottom of the deepest OFFSHORE stations inlate summer does the concentration of oxygen drop to lower than4 cm3 l�1, to less than 80% saturation (Fig. 5d).

3.4. Dissolved and particulate organic matter

Production and consumption processes drive a clearly recog-nisable DOC seasonal cycle (Figs. 2e4 panel a) characterised bya minimum in February, 0.8 mg C l�1 and a maximum of approxi-mately 1.5 mg C l�1 in July (RIVER and OFFSHORE sites) or August(FARCOAST site). DOC annual cycles in the bottom layer (Fig. 5a) aresimilar to those observed at the surface, but accumulation lasts forone month more and concentrations are generally 5e10% lower inthe bottom than in the surface layer during summer (see also meanannual values in Table 1).

A small spatial variability of DOC is observed among sites; inparticular, the RIVER site is characterised by concentrations higher(0.05e0.15 mg C l�1) than the other two sites in spring and summer,indicating the higher productivity of the site under riverineinfluence.

The spatial and temporal fluctuations of DON and DOP(Figs. 2e4 panel c) are high, and the climatological annual cycledoes not show the clear seasonal process found for DOC. Betweenthe two variables, DON shows amore recognisable seasonal patternwith maxima (values higher than 15 mM at RIVER site and up to15 mM at the other two sites) in spring or summer and minima(down to 7 mM) in winter. DOP mean values are approximately0.4e0.5 mM (Table 1), with occasional monthly peaks that vary fromsite to site.

At surface, the annual cycle of POC (Figs. 2e4 panel a) showsa general accumulation from January until the end of summer/beginning of autumn, followed by a decrease during autumn.Bottom POC accumulation ends at the beginning of summer (earlierrespect to the surface layer) and concentrations at the bottom aregenerally lower than at surface except for the summer values at theOFFSHORE site.

PN cycle is quite similar to POC one (correlation coefficient ofapproximately 0.70 at all sites, p< 0.05), whereas PP has a more

irregular pattern, indicating a possible decoupling of the nutrientcycles and multiple sources and sinks.

A surface spatial gradient for the particulate compartment ispresent among sites, being the RIVER characterised by the highestvalues. At the bottom the OFFSHORE site presents concentrationhigher than at the other site for all particulate variables (Table 1).

3.5. Biological variables

Information on biological components (biomasses of differentsize classes of autotrophic and heterotrophic components) e not sooften available because of the cost and skill necessary e is availablefor station C1/T21 (Fig. 6), from which an uncommonly compre-hensive dataset has been compiled in the last decade (Fonda Umaniet al., 2012).

Diatom biomass exceeds all other autotrophic biomasses at boththe surface and the bottom layers by one order of magnitude(Fig. 6a and c). At the surface, two high peaks are observed: the firstin February and the second in July. At the bottom, only the first peakhas the same intensity and is in fact even higher than at the surface.This is probably attributable to the senescent part of the surfacecommunity that is not grazed by either microzooplankton ormesozooplankton at the surface (Fonda Umani et al., 2005, 2012)and therefore settles to the bottom. The export to the bottom of theJuly surface peak is probably reduced by both column water strat-ification and by more efficient grazing. Microflagellates thatcompose most of the rest of the microphytoplankton peak inAprileMay, followed by an increase in autotrophic nanoplankton(<10 mm) at the surface; at the bottom, the two peaks are simul-taneous, and nanoplankton biomass is higher than that of themicroflagellates. Autotrophic picoplankton reaches its maximum inSeptember after a continuous increase beginning in March at boththe surface and the bottom levels. Heterotrophic picoplanktonmake up the most abundant heterotrophic biomass, amounting toapproximately twice the biomass accounted for by the rest of theheterotrophic organisms over the whole year. The bacteria reachtheir maximum in September in conjunction with the autotrophiccomponent of the picoplankton. At this time, as shown in Fig. 4aand c, organic compartments decrease as consumption processesexceed production processes, and oxygen falls under saturation(Fig. 4d).

Mesozooplankton was the second most important standingstock. Because the mesozooplankton was collected by vertical towsfrom near the bottom to the surface, we cannot distinguish thesurface values from the bottom ones. In this case, we used the same

G. Cossarini et al. / Estuarine, Coastal and Shelf Science 115 (2012) 63e7470

average values for both layers. Mesozooplankton presented twomaxima: the highest in May and the second, less intense maximumin August. The microzooplankton biomass shows a small peak inFebruary in correspondence with the first diatom bloom, but theseorganisms did not substantially impact the production of diatoms,as experimentally assessed by Fonda Umani and Beran (2003),Fonda Umani et al. (2005, 2012). In summer at both depths, themicrozooplankton biomass was high following the increase in thepicoplankton numbers and decreased again in autumn. Theincrease of POC observed from spring to summer (see Fig. 4) can beexplained by the transfer of organic matter from the autotrophiccompartment to the heterotrophic compartment. Heterotrophicnanoplankton biomass was always low without any significantfluctuation. By comparing autotrophic and heterotrophic biomasssuccession, we can infer some possible trophic interactions inaddition to the lack of predation control on diatom bloom at thebeginning of the year. The mesozooplankton population increasefollows the increase of microflagellates and autotrophic nano-plankton, but in spring, these organisms rely on diatoms as well,possibly in this case controlling their development. The decrease ofmesozooplankton observed in July does not correspond toa decrease in the number of organisms but in a change ofcommunity structure, which passes from the dominance of cope-pods to that of cladocerans (e.g., Penilia avirostris (Lipej et al., 1997)).The latter is a fine filter feeder and is not able to consume largediatoms. Therefore, the surface secondary peak of the larger auto-trophic fraction may be caused not only by ephemeral nutrientinputs but also by the lack of an efficient top-down control.Whatever the cause, the diatom secondary peak is quickly levelledoff by both microzooplankton and mesozooplankton grazing. Theformer, in particular, as well as heterotrophic nanoplankton, alsopreys upon both autotrophic and heterotrophic picoplankton,which in the second part of the year become the most importantsource of biomass available to consumers.

3.6. Principal component analysis (PCA) on biogeochemicalvariables

The principal component analysis was performed separately foreach of the three sites and at both depths, and the results are jointlyshown in Fig. 7 (loadings of the original variables in the newprincipal components) and Fig. 8 (monthly changes in the scores ofthe new components).

At surface the first components (PC1s) explain approximately40% of the total variance in all sites (Table 2) and they are a linear

Fig. 7. Loadings of original variables (different symbols and colours) into the first two princip(b) layers.

combination of nutrients (N-NO3, Si-SiO4, P-PO4, all with negativeloadings) and organic matter (DOC, POC and PN) and oxygen (withpositive loadings). CHLA and DOP (with negative loadings) andDON, N-NH4 and PP (with positive loadings) constitute the secondPC at all sites, Fig. 7a, which accounts for 20e25% of the totalvariance (Table 2).

PC1 scores of the three surface sites (Fig. 8a) depict a commonseasonal cycle with maximum in summer and minimum in winter.A local maximum is present in winter: in February for RIVER andOFFSHORE sites and in March for FARCOAST one. PC2 temporalpatterns of the three sites, Fig. 8b, show the maximum in autumn,the minimum in winter, a local peak in spring and a stationarypattern in summer. However the time of the two peaks andminimum slightly differs among sites.

At the bottom layer (Fig. 7b) the dynamics of the chlorophylla and organic matter compartments (POC, DOC, PN, DON and PP)are coupled at both sites, as verified by the positive loadings of thePC1 (approximately 50% of the total variance, Table 2). The timing ofthe evolution of PC1, Fig. 8c, is slightly different between the twosites, with the OFFSHORE site delayed with respect to the FAR-COAST site. Oxygen and DOP have a negative value in PC2 for bothsites, however at the OFFSHORE site oxygenweights negatively alsoin PC1 showing a negative correlation with organic matterevolution.

Nutrients have a different behaviour at the two sites: at theOFFSHORE site, Si-SiO4 and N-NH4 have positive loadings in PC1and N-NO3 and P-PO4 have positive loadings in PC2, whereas atFARCOAST the first pair of nutrients has positive loadings in PC2and the second pair has negative loadings in PC1.

3.7. Biogeochemical variable correlations among sites

The correlations between couples of sites (Table 3) show thatmost of the biogeochemical variables have similar evolution at thedifferent sites. Dissolved organic phosphorus and nitrogen andparticulate phosphorus are the only variables that displaya heterogeneity among sites. The climatologies have already shownthat these variables do not present any very clear seasonal patternbut high variabilities associated to the monthly values (see inter-quartile range of monthly values in the Table in the appendix).RIVER chlorophyll a shows a poor correlation with the other twosites because of the shift of one month in the spring and autumnblooms.

al components of the three sites (different line styles) and of the surface (a) and bottom

Fig. 8. Evolution of the scores of the first two principal components (upper and lower panel) of the three sites (different colours) and of the surface (a and c) and bottom (b and d)layers.

G. Cossarini et al. / Estuarine, Coastal and Shelf Science 115 (2012) 63e74 71

4. Discussion

Analysing a shorter dataset (three years long) of the Gulf ofTrieste, Solidoro et al. (2007) had shown that the high variability ofbiogeochemical properties can be summarised by identifyingbiogeochemical homogeneous water masses. Analysing theirspatial and temporal patterns of variability, the authors had beenable to identify a conceptual scheme of the biogeochemical func-tioning of the ecosystem and a subdivision of the Gulf of Trieste inhomogeneous areas. Using a much larger datasets we corroboratedthat scheme by computing the climatologies at three selected sites.Site climatologies (Figs. 2e5) show a common seasonal pattern formost of the variables, albeit sites can differ for the absolute valuesof variables or the timing of several minima and maxima. Thecommon seasonal pattern of biogeochemical properties isconfirmed by the high correlations of variables between eachcouple of sites (Table 3) and, in a multivariate picture, by thecommon pattern of loadings (Fig. 7) and scores (Fig. 8) of PCAs,which have been performed separately at each of the three sites.The common biogeochemical pattern supports the conceptualscheme of the functioning of this ROFI suggested by Fonda Umaniet al. (2007) and Solidoro et al. (2007).

Furthermore, the biological biomass climatology (identified forone only out of the nine stations, Fig. 6) corroborates the inter-pretation of the ecological dynamics hypothesised on biogeo-chemical data only. Since differences among sites are highlysignificant in term of intensity of the biological processes, butdefinitively less important in term of ecological succession, it ispossible to speculate that the biological biomasses climatology canbe indicative of the whole ROFI area.

Accordingly to surface climatologies (Figs. 2e4), all nutrients butammonia accumulated as a result of input from external sources(i.e. river discharges, see Cozzi et al., 2012), degradation of organic

Table 2Relative explained variance by the first 2 principal components at the different sitesand depths.

Surface Bottom

RIVER OFFSHORE FARCOAST OFFSHORE FARCOAST

PC1 43 39 42 50 39PC2 25 19 24 24 27

matter and reductions in consumption (Fonda Umani et al., 2007;Solidoro et al., 2007) from autumn to January.

In February, the high nutrient content and the increasing lightavailability stimulate a bloom (a peak of chlorophyll a) mostlyvisible in RIVER and FARCOAST sites. This bloom, due to diatoms(Fig. 6a), is rapidly interrupted in March, possibly because of theconsumption of nutrient stocks, which shows a concomitantminimum, and to a lesser extent, because of grazing by zooplankton(Fonda Umani et al., 2012). A new input of nutrients (nitrate andsilicate mainly), concomitant with spring Isonzo run-off (Cozziet al., 2012), stimulates a bloom in April. This bloom extends tothe whole Gulf (peak of CHLA at all sites), the chlorophyll a:carbonratio peaks to 0.03 mg CHLA (mg C)�1 which indicates fast growingrates of phytoplankton under favourable nutrient conditions(Cloern et al., 1995; Geider et al., 1997). The total carbon biomass ofprimary producers at C1/T21 at this time is lower than that ofFebruary and microphytoplankton contributes to the total biomassalmost as diatoms (Fig. 6a) which are possibly controlled by thegrowth of mesozooplankton (Fig. 6b).

During this time, when river influence is more important, largerorganisms (diatoms and mesozooplankton) dominate the structureof the food web, and the energy andmatter flows are mainly driventowards particulate compartments and the export production(Cossarini and Solidoro, 2008; Fonda Umani et al., 2012).

In fact, POC accumulates from spring until the end of summer,even if chlorophyll a content decreases to the summer minimum(see the uncoupling of PC loadings of POC and CHLA), possibly sug-gesting a transfer of biomass from phytoplankton to zooplanktonthrough grazing and the consequent production of faecal pellets (asdescribed for the classic food web in Cushing, 1989).

Further, PC1 results (loadings and scores at the three sites)support the hypothesis that at all sites the production of neworganic matter begins in late winter-early spring and indicatea reduction of nitrate, silicate and phosphate. POC accumulation(and nutrient uptake) stops (at the most river-influenced site) orreverses (at the other sites) from February to April and then startsagain until June when mineralisation processes prevail (see scoreevolutions at Fig. 8a).

The accumulation of organic matter in its dissolved form is evenhigher than that in the particulate compartment in summer. DOCconcentration increases of more than 50% respect to the wintervalues at all sites. A significant summer DOC accumulation is

Table 3Correlation of climatologies between couples of sites for each of the biogeochemical variables. Correlations marked in bold are significant at p< 0.05.

Surface Bottom

RIVER-OFFSHORE RIVER-FARCOAST OFFSHORE-FARCOAST OFFSHORE-FARCOAST

CHLA 0.241 0.539 0.900 0.728N-NO3 0.824 0.801 0.918 0.899N-NH4 0.771 0.489 0.344 0.661P-PO4 0.703 0.626 0.755 0.493Si-SiO4 0.921 0.557 0.681 0.394DOC 0.970 0.918 0.924 0.862POC 0.820 0.701 0.570 0.673DON 0.216 0.517 0.158 0.610DOP 0.401 0.127 -0.167 0.219PN 0.592 0.562 0.390 0.735PP 0.882 0.652 0.739 0.710OXY 0.975 0.862 0.853 0.608SAL 0.772 0.598 0.809 0.877TEMP 0.996 0.992 0.997 0.978

G. Cossarini et al. / Estuarine, Coastal and Shelf Science 115 (2012) 63e7472

a characteristic already observed in the Gulf (De Vittor et al., 2008),in the Northern Adriatic Sea (Pettine et al., 2001) and in severalcoastal systems (see references in Giani et al., 2005). DOC can orig-inate from hydrolysis of POC or from direct production by primaryproducers in intense light but nutrient depleted conditions. On thisissue it is worth to note that, in summer, the chlorophyll a:carbonratio decreases to the minimum value of 0.0025 mg CHLA (mg C)�1

highlighting the adaptation of diatoms, the major contributors ofphytoplankton community, to high irradiance and low nutrientconditions (Falkowski, 1994; Geider et al., 1997). It is also suggestedthat DOC is produced during grazing of protists and copepods (DeVittor et al., 2008), which actually show a significant increaseduring summer (Fig. 6b). During autumnDOC concentration returnsto winter minimum, showing a good capacity of this ROFI area inreacting to the significant external organic matter loads throughdegradation and exchange processes, as it has been pointed out forthe Northern Adriatic Sea (Pettine et al., 2001).

During summer, as river run-off decreases (Cozzi et al., 2012)and the system becomesmore closed as a result of the stratification,the microbial food web becomes the principal energy pathway(Legendre and Rassoulzadegan, 1995; Cossarini and Solidoro,2008), and remineralisation becomes the principal source ofnutrients. The marked increase of heterotrophic picoplanktonduring summer (Fig. 6b) support this interpretation. Indeed,oxygen content, which exceeds the saturation from March to July,falls under the saturation in summer when consumption processesprevail.

In autumn the river discharge increases again (Cozzi et al., 2012)and wind and negative heat fluxes cause the vertical mixing of thewater column. The consequent increase of the availability ofnutrient at surface stimulates an autumn phytoplankton bloom(mainly diatoms, Fig. 6a) until December when the picoplanktonbecomes the dominant component of the planktonic community(Fig. 6a and c).

At this time, a new, but short in time, accumulation of particu-late organic matter occurs. However since the higher increaseregistered at the RIVER site, it can be argued that importantexternal input contribute to the autumn accumulation.

At the bottom, dynamics are slightly different, confirming thatthe main drivers in ROFI areas are the external input and theinteraction with the bottom (Solidoro et al., 2007). During themixing period (NovembereApril), mixing and sinking processestransfer organic matter produced at the surface down to the bottomin the absence of local blooms (low values of chlorophyll a but highvalues of carbon and biological compartments Figs. 5 and 6). Duringstratification period fertilisation resulting from nutrient reminer-alisation (see increase of N-NH4, P-PO4 and Si-SiO4 in Fig. 5b)

stimulates bottom phytoplankton blooms throughout the warmermonths (high values of chlorophyll a). Among the nutrients, N-NO3does not showany significant signal because nitrogen is prevalentlyremineralised into N-NH4. The greatest rate of local POC productionis recorded from April to June. The accumulation of nutrientsduring late summer, even in the presence of high levels of chloro-phyll a, demonstrates that the rate of secondary production andnutrient remineralisation is higher than that of primary productionand nutrient consumption. This is confirmed by the reducedoxygen saturation during summer and by the concomitantconsumption of DOC, which shows concentration of 10e15% lowerat the bottom.

The dissolved organic pools are the major reservoir of nutrients(P and N; Table 1), however their seasonal dynamics can be maskedby great variabilities (see table in the appendix) due to the manyprocesses involved (e.g., excretion and exudation from the biolog-ical compartments of different trophic levels, recycling and trans-port; Hansell and Carlson, 2002). In addition, the low correlationamong the couples of sites (Table 3) highlights also the presence ofa certain degree of spatial heterogeneity, which is possibly linked tothe impact of external forcings (Lipizer et al., 2011). DOP and DONpresent opposite seasonal dynamics (see loadings in the PC1ePC2plane), highlights a substantial decoupling of the nitrogen andphosphorus cycles. DON presents a certain degree of affinity withthe cycle of production and consumption of organic carbon,whereas DOP has a less clear affinity with other variable patterns(Fig. 7). A negative correlation between DOP and PP in surface(loadings in Fig. 7a) and negative correlation between DOP and P-PO4 at the bottom (loadings in Fig. 7b) might indicate a promptrecycling highlighting a possible limiting role of phosphorus in theGulf. The comparison with published results about surroundingareas does not help in clarifying the seasonal pattern and dynamicsof dissolved organic pool. DON and DOP data of the Gulf are slightlyhigher than that reported by Danovaro et al. (2005) and Ivancicet al. (2011). A DOP seasonal cycle with summer minimum re-ported for the Po-Rovinj transect (Ivancic et al., 2011) is a charac-teristic not observed in the Gulf.

The Gulf of Trieste is amarginal area of the Northern Adriatic Sea(NAS), which was for long regarded as an eutrophic area because ofthe consistent inputs from Po River (Degobbis et al., 2000). Indeednow it is recognised that only the westernmost part can beconsidered eutrophic and that a large part should be consideredmesotrophic or oligotrophic (Solidoro et al., 2009). A trend in oli-gotrophication has been recently recognised (Solidoro et al., 2009;Mozetic et al., 2010) possibly due to changes in river nutrient loads(Cozzi and Giani, 2011) and oceanographic conditions (Degobbiset al., 2000). The Gulf of Trieste and the NAS, despite the

G. Cossarini et al. / Estuarine, Coastal and Shelf Science 115 (2012) 63e74 73

difference in size, share some common features: such as the pres-ence of a relative important river and a similar seasonal cycle ofoceanographic forcings (Cushman-Roisin et al., 2001). Indeed,a succession of typical ecological season identified for the NAS(Degobbis et al., 2000) fairly matches with the conceptual schemeof the biogeochemical evolution of the Gulf of Trieste here pre-sented. The two seas may differ for the absolute level of the trophicstatus. The comparison of the RIVER site with the area in front ofthe Po River (zone [1] and [5] in Solidoro et al., 2009) shows that themean concentration of all nutrients is lower, whereas at FARCOASTand OFFSHORE nutrient concentrations, but phosphate, are slightlyhigher than that of an off-shore area in the NAS (zone [8] in Solidoroet al., 2009). Phosphate concentrations are 10% lower in the Gulf ofTrieste. Therefore, we suggest that the Isonzo River has a lowerimpact on the Gulf than that of the Po on NAS; the Gulf is evenmoreP-limited than the off-shore area of NAS and the surplus of theother nutrients are exported out of the Gulf to the NAS. Thishypothesis (P-limitation and lower trophic status in the Gulf) issupported by the fact that chlorophyll a and DOC concentrations inthe Gulf are substantially lower than that of the NAS. Chlorophylla at RIVER site is almost halved with respect to that registered infront of the Po River (Solidoro et al., 2009), while at OFFSHORE andFARCOAST the values are 15% lower than that of the off-shore areain the NAS (Solidoro et al., 2009). The seasonal cycle of DOC in theGulf is similar to that observed in the NAS, however backgroundand maximum values are lower. Taking as background values forthe Gulf of Trieste the range interquartile of the minimum monthand site (i.e. February of the OFFSHORE site, see the table inappendix), it comes out a DOC concentration of 62e79 mM, thatupdates a previous estimation of 59� 7 mM (De Vittor et al., 2008).This value is slightly lower than the estimation of 76�10 mMproposed for the NAS (Pettine et al., 2001) and the values of around80e90 mM (minimum value of the monthly averages) recorded inthree transects in the northern and central part the Adriatic Sea(Giani et al., 2005). Further, the 75th percentile of summer valuesnever exceed 150 mM in the Gulf whereas the maximum of themonthly averages reported by Giani et al. (2005) is generally higherthan that limit.

Although the common ecological pattern within this ROFI area,the distance from the river and the depth of the water columnintroduce important differences on the trophic level and possiblydelays in the timing of the ecological processes among the sites. Asthe distance from the river mouth increases, the mean contents ofnutrients and organic matter decrease, highlighting the importanceof the river as a fundamental driver of the ecosystem functioning(Mozetic et al., 1998; Solidoro et al., 2007; Lipizer et al., 2011).However, the distance has to be considered here not as the simplegeographical distance, but as a function of circulation and transportprocesses. Since the typical cyclonic circulation in the Gulf (Malacicand Petelin, 2009), the FARCOAST site, located in the eastern part ofthe Gulf, results to be the one less influenced by the river input. Themost oligotrophic site, FARCOAST, presents a more reducedproductive season with respect to the other sites: accumulation ofPOC ends in June, whereas in the other sites remains till August.

The shift in the organic matter production/consumption cycle iswell described by the difference in the timing of theFebruaryeMarch relative peak of PC1s scores (Fig. 8a), which appearfirst at the more river-influenced site and later at the FARCOAST site.

A significant spatial decoupling within the Gulf occurs inautumn. Indeed, despite the maximum freshwater discharge is inNovember (Cozzi et al., 2012), the minimum of salinity is registeredin spring. This is because strong autumn easterly winds (Stravisi,1977a) drive the surface circulation, forcing the river plume outof the Gulf along a narrow strip near the northern-western coast(Stravisi, 1977b; Querin et al., 2006). Therefore, in autumn, only the

most coastal site shows a decrease in salinity, increase in nutrientsand a phytoplankton bloom. The other two sites show a delayedbloom in November at both layers (peak of chlorophyll a, Figs. 3 and4, and of diatoms, Fig. 6a and c), when nutrients are supplied to thephotic zone by mixing processes more than the external input (asdescribed in Degobbis et al., 2000). In fact, the increase of nutrientconcentration at surface occurs without a decrease of salinity(Figs. 3 and 4).

Other important differences within the ROFI area can be due tothe bathymetry which influences the level of mixing and conse-quently bottom processes. In fact, at the deepest site, OFFSHORE, itis registered, with respect to the FARCOAST, a higher rate of organicmatter accumulation (year-round higher POC concentration ofapproximately 50e100 mg C l�1), a higher consumption rate (highernegative deviation of saturation of oxygen during summer) anda delay in nutrients production/utilization processes (as shown bythe different nutrient loadings and scores evolution of the PCs).

The existence of differences among sites highlights that even ina geographically small ROFI region, such as the Gulf of Trieste,significant heterogeneity in biogeochemical properties is aninherent feature of coastal systems. Therefore, it is not sensible toassociate the evaluation of ecological quality of different sub-areasbased on a comparison against a single reference value common toall areas. Instead, different reference values should be used toclassify different sub-areas. This concept has already been com-mented in relation to transitional (Solidoro et al., 2004; Bandeljet al., 2008) and estuarine environments (Elliott and Quintino,2007). However, the concept described here is seldom incorpo-rated in the application of environmental legislation and is toooften overlooked, even in the scientific literature, where it is stilleasy to find analyses inwhichmarine influenced areas are classifiedas being of higher ecological quality than river-influenced areasrather than as being simply different.

5. Conclusions

An analysis of biogeochemical data regularly collected foralmost a decade at different sites in the Gulf of Trieste allowed us todraw a robust picture of the biogeochemical properties of a coastalarea under the influence of freshwater in the northernmost part ofthe Adriatic Sea, whichmay be useful for other areas. We computedan overall monthly climatology from a multi year, multi site coastalecosystem dataset, deriving and comparing the typical cycles atthree sites representative of areas that are differently influenced byriver discharge.

Our analysis confirmed, despite the spatial variability, the exis-tence of a common scheme of ecosystem functioning andphysicalebiological interactions, highlighting the role of river input,lateral transport, stratification regimes and interactions with thebottom environment as driving factors. The biogeochemical clima-tology showed a typical succession of trophodynamic processes: inlatewinter and spring, outflow from the river triggers a classic springphytoplankton bloom, which takes advantage of seasonal warmingand the increase in photoperiod length. During late springesummer,the reduction of river discharges leads to an increase in stratification,and autotrophic activities cause a depletion in the pools of dissolvedinorganic nutrients, particularly in surface layers. An increase in theproduction of dissolved organic matter and a decoupling betweenprimary production and bacterial carbon demand occur, causing anaccumulation of DOC. In late summer/early autumn, new inputs ofriver water and new mixing regime may trigger a second phyto-plankton bloom, which is followed by microbial food web activitiesthat prevail during winter, when consumption and inorganicnutrient regeneration processes dominate.

G. Cossarini et al. / Estuarine, Coastal and Shelf Science 115 (2012) 63e7474

Finally, since intensity and temporal succession of biogeo-chemical processes are modulated by the distance from the river,significant spatial heterogeneity in biogeochemical properties is aninherent feature of coastal systems which should be taken intoaccount explicitly in the evaluation of the good environmentalstatus.

Acknowledgements

The present work was partly funded by the European Commu-nity and the Friuli-Venezia Giulia Region ECOMADR and INTERREG3 projects. The authors thank all the technical and scientific staff atdepartment of biological oceanography of OGS (former Laboratoryof Marine Biology e LBM) involved in data sampling and analysis.The authors thank two anonymous reviewers and the guest editorfor their valuable comments, and Stefano Salon for revising themanuscript.

Appendix. Supplementary material

Supplementary data related to this article can be found online atdoi:10.1016/j.ecss.2012.02.006.

References

Bandelj, V., Socal, G., Park, Y.-S., Lek, S., Coppola, C., Camatti, E., Capuzzo, E.,Milani, L., Solidoro, C., 2008. Analysis of multitrophic plankton assemblages inthe Lagoon of Venice. Marine Ecology Progress Series 368, 23e40.

Cauwet, G., 1994. HTCO method for dissolved organic carbon analysis in seawater:influence of catalyst on blank estimation. Marine Chemistry 47, 55e64.

Cloern, J.E., 2001. Our evolving conceptual model of the coastal eutrophicationproblem. Marine Ecology Progress Series 210, 223e253.

Cloern, J.E., Grenz, C., Vidergar-Lucas, L., 1995. An empirical model of the phyto-plankton chlorophyll/carbon ratio d the conversion factor between produc-tivity and growth rate. Limnology and Oceanography 40, 1313e1321.

Cossarini, G., Solidoro, C., 2008. Global sensitivity analysis of a trophodynamicmodel of the Gulf of Trieste. Ecological Modelling 212, 16e27.

Cozzi, S., Giani, M., 2011. River water and nutrient discharges in the NorthernAdriatic Sea: current importance and long term changes. Continental ShelfResearch 31 (8), 1881e1893.

Cozzi, S., Falconi, C., Comici, C., �Cermelj, B., Kovac, N., Turk, V., Giani, M., 2012.Recent evolution of river discharges in the Northeastern Adriatic Sea and theirpotential response to climate changes and anthropogenic pressure. Estuarine,Coastal and Shelf Science 115, 14e24.

Cushing, D.H., 1989. A difference in structure between ecosystems in stronglystratified waters and in those that are only weekly stratified. Journal ofPlankton Research 11 (1), 1e13.

Cushman-Roisin, B., Gacic, M., Poulain, P.M., Artegiani, A., 2001. Physical Oceanog-raphy of the Adriatic Sea, Past, Present and Future. Kluwer Academic Publishers,Dordrecht, Netherlands, 340 pp.

Danovaro, R., Armeni, M., Luna, G.M., Corinaldesi, C., Dell’Anno, A., Ferrari, C.R.,Fiordelmondo, C., Gambi, C., Gismondi, M., Manini, E., Mecozzi, M.,Perrone, F.M., Pusceddu, A., Giani, M., 2005. Exo-enzymatic activities and dis-solved organic pools in relation with mucilage development in the NorthernAdriatic Sea. Science of the Total Environment 353, 189e203.

Davis, J.C., 1973. Statistics and Data Analysis in Geology. Wiley, New York, 550 pp.Degobbis, D., Precali, R., Ivancic, I., Smodlaka, N., Fuks, D., Kveder, S., 2000. Long-

term changes in the northern Adriatic ecosystem related to anthropogeniceutrophication. International Journal of Environment and Pollution 13 (1e6),495e533.

De Vittor, C., Paoli, A., Fonda Umani, S., 2008. Dissolved organic carbon variability ina shallow coastal marine system (Gulf of Trieste, northern Adriatic Sea). Estu-arine, Coastal and Shelf Science 78 (2), 280e290.

Elliott, M., Quintino, V., 2007. The Estuarine Quality Paradox, EnvironmentalHomeostasis and the difficulty of detecting anthropogenic stress in naturallystressed areas. Marine Pollution Bulletin 54, 640e645.

Falkowski, P.G., 1994. The role of phytoplankton photosynthesis in global biogeo-chemical cycles. Photosynthesis Research 39, 235e258.

Fonda Umani, S., Beran, A., 2003. Seasonal variations in the dynamics of microbialplankton communities: first estimates from experiments in the Gulf of Trieste,Northern Adriatic Sea. Marine Ecology Progress Series 247, 1e16.

Fonda Umani, S., Tirelli, V., Beran, A., Guardiani, B., 2005. Relationships betweenmicrozooplankton and mesozooplankton: competition vs predation on naturalassemblages in the Gulf of Trieste (northern Adriatic Sea). Journal of PlanktonResearch 27 (10), 973e986.

Fonda Umani, S., Del Negro, P., Larato, C., DeVittor, C., Cabrini, M., Celio, M.,Falconi, C., Tamberlich, F., Azam, F., 2007. Major inter-annual variations in

microbial dynamics in the Gulf of Trieste (northern Adriatic Sea) and theirecosystem implications. Aquatic Microbial Ecology 46, 163e175.

Fonda Umani, S., Malfatti, F., Del Negro, P., 2012. Carbon fluxes in the pelagicecosystem of the Gulf of Trieste (Northern Adriatic Sea). Estuarine, Coastal andShelf Science 115, 170e185.

Geider, R.J., MacIntyre, H.L., Kana, T.M., 1997. Dynamic model of phytoplanktongrowth and acclimation: responses of the balanced growth rate and the chlo-rophyll a:carbon ratio to light, nutrient-limitation and temperature. MarineEcology Progress Series 148, 187e200.

Giani, M., Savelli, F., Berto, D., Zangrando, V., Cosovic, B., Vojvodic, V., 2005. Temporaldynamics of dissolved and particulate organic carbon in the northern Adriatic Seain relation to themucilage events. Science of the Total Environment 353,126e138.

Grasshoff, K.,1983.Determinationof nutrients. In:Grasshoff, K., Ehrhardt,M., Kremling, K.(Eds.), Methods of Seawater Analysis. Verlag Chemie, Basel, pp. 125e187.

Hansell, D.A., Carlson, C.A., 2002. Biogeochemistry of Marine Dissolved OrganicMatter. Academic Press, Amsterdam, Netherlands, 774 pp.

Howarth, R.W., Swaney, D.P., Butler, T.J., Marino, R., 2000. Climatic control oneutrophication of the Hudson River estuary. Ecosystems (IntergovernmentalPanel on Climate Change) 3, 210e215.

Ivancic, I., Fuks, D., Radic, T., Lyons, D.M., Silovic, T., Kraus, R., Precali, P., 2011.Phytoplankton and bacterial alkaline phosphatase activity in the northernAdriatic Sea. Marine Environmental Research 69, 85e94.

Legendre, L., Rassoulzadegan, F., 1995. Plankton and nutrient dynamics in marinewater. Ophelia 41, 153e172.

Lipej, L., Mozetic, P., Turk, V., Malej, A., 1997. The trophic role of the marine cla-docera Penilia avirostris in the Gulf of Trieste. Hydrobiologia 360, 197e203.

Lipizer, M., Cossarini, G., Falconi, C., Solidoro, C., Fonda Umani, S., 2011. Impact ofdifferent forcing factors on N:P balance in a semi-enclosed bay: the Gulf ofTrieste (North Adriatic Sea). Continental Shelf Research 31 (16), 1651e1662.

Lipizer, M., De Vittor, C., Falconi, C., Comici, C., Tamberlich, F., Giani, M., 2012. Effects ofintense physical and biological forcing factors on CNP pools in coastal waters (Gulfof Trieste, Northern Adriatic Sea). Estuarine, Coastal and Shelf Science 115, 40e50.

Lorenzen, C., Jeffrey, J., 1980. Determination of chlorophyll in seawater. UNESCOTechnical Papers in Marine Science 35, 1e20.

Malacic, V., Petelin, B., 2009. Climatic circulation in the Gulf of Trieste (northernAdriatic). Journal of Geophysical Research 114. doi:10.1029/2008JC004904.

Mann, K.H., Lazier, J.R.N., 1998. Dynamics of Marine Ecosystems. Blackwell Scien-tific, London, 466 pp.

Mozetic, P., Fonda Umani, S., Cataletto, B., Malej, A., 1998. Seasonal and inter-annualplankton variability in the Gulf of Trieste (Northern Adriatic). Journal of MarineScience 55, 711e722.

Mozetic, P., Fonda Umani, S., Solidoro, C., Cossarini, G., Socal, G., Precali, R., France, J.,Bianchi, F., De Vittor, C., Smodlaka, N., 2010. Recent trends towards oligo-trophication of the Northern Adriatic: evidence from chlorophyll a time series.Estuaries and Coasts 33 (2), 362e375.

Nixon, S.W., 1995. Coastal marine eutrophication: a definition, social causes, andfuture concerns. Ophelia 41, 199e219.

Pettine, M., Capri, S., Manganelli, M., Patrolecco, L., Puddu, A., Zoppini, A., 2001. Thedynamics of DOM in the Northern Adriatic Sea. Estuarine, Coastal and ShelfScience 52, 471e489.

Querin, S., Crise, A., Deponte, D., Solidoro, C., 2006. Numerical study of the role ofwind forcing and freshwater buoyancy input on the circulation in a shallowembayment (Gulf of Trieste, Northern Adriatic Sea). Journal of GeophysicalResearch 111. doi:10.1029/2006JC003611.

Riley, G.A., 1942. The relationship of vertical turbulence and spring diatom flow-ering. Journal of Marine Research 5, 67e87.

Sharp, J.H., 1974. Improved analysis for particulate organic carbon and nitrogenfrom seawater. Limnology and Oceanography 19, 984e989.

Simpson, J.H., 1997. Physical processes in the ROFI regime. Journal of MarineSystems 12 (1), 3e15.

Smetacek, V., Cleorn, J.E., 2008. On phytoplankton trends. Science 319, 1346e1348.Solidoro, C., Pastres, R., Cossarini, G., Ciavatta, S., 2004. Seasonal and spatial vari-

ability of water quality parameters in the lagoon of Venice. Journal of MarineSystems 51 (1-4), 179e189.

Solidoro, C., Bandelj, V., Barbieri, P., Cossarini, G., Fonda Umani, S., 2007. Under-standing dynamic of biogeochemical properties in the northern Adriatic Sea byusing self-organizing maps and k-means clustering. Journal of GeophysicalResearch 112. doi:10.1029/2006JC003553.

Solidoro, C., Bastianini, M., Bandelj, V., Codermatz, R., Cossarini, G., Melaku Canu, D.,Ravagnan, E., Salon, S., Trevisani, S., 2009. Current state, scales of variability, andtrends of biogeochemical properties in the northern Adriatic Sea. Journal ofGeophysical Research 114. doi:10.1029/2008JC004838.

Solorzano, L., Sharp, J.H., 1980. Determination of total dissolved phosphorus and partic-ulate phosphorus in natural waters. Limnology and Oceanography 25 (4), 754e758.

Stockner, J.G., Rydin, E., Hyenstrand, P., 2000. Cultural oligotrophication: causes andconsequences for fisheries resources. Fisheries 25 (5), 7e14.

Stravisi, F., 1977a. Il regime dei venti a Trieste (1951e1975). Bollettino della SocietàAdriatica di Scienze 61, 87e104.

Stravisi, F., 1977b. Bora driven circulation in Northern Adriatic. Bollettino di Geo-fisica teorica ed applicata 19, 95e102.

Sverdrup, H.U., 1953. On the conditions for the vernal blooming of phytoplankton.Journal du Conseil International pour l’ Exploration de la Mer 18, 287e295.

Utermohl, H., 1958. Zur Vervolkommung der quantitativen Phytoplankton-Meth-odik. Mitteilungen Internationale Vereiningung fuer Theoretische und Ange-wandte Limnologie 9, 1e38.