Long-term patterns and drivers of microbial organic matter utilization in the northernmost basin of the Mediterranean Sea

https://doi.org/10.1016/j.marenvres.2020.105245Get rights and content

Highlights

  • A 21-year time series was analysed coupling canonical methods with machine learning.

  • Three coherent periods were identified in microbial and organic matter time series.

  • A prolonged period of reduced river runoff was the main driver of this structure.

  • Temperature and particulate matter were the main drivers of microbial metabolism.

  • Microbial metabolism was substrate limited in the years following the draught event.

Abstract

Marine heterotrophic prokaryotes degrade, transform, and utilize half of the organic matter (OM) produced by photosynthesis, either in dissolved or particulate form. Microbial metabolic rates are affected by a plethora of different factors, spanning from environmental variables to OM composition. To tease apart the environmental drivers underlying the observed organic matter utilization rates, we analysed a 21 year-long time series from the Gulf of Trieste (NE Adriatic Sea). Heterotrophic carbon production (HCP) time series analysis highlighted a long-term structure made up by three periods of coherent observations (1999-2007; 2008-2011; 2012-2019), shared also by OM concentration time series. Temporal patterns of HCP drivers, extracted with a random forest approach, demonstrated that a period of high salinity anomalies (2002–2008) was the main driver of this structure. The reduced river runoff and the consequent depletion of river-borne inorganic nutrients induced a long-term Chl a decline (2006–2009), followed by a steady increase until 2014. HCP driving features over the three periods substantially changed in their seasonal patterns, suggesting that the years following the draught period represented a transition between two long-term regimes. Overall, temperature and particulate organic carbon concentration were the main factors driving HCP rates. The emergence of these variables highlighted the strong control exerted by the temperature-substrate co-limitation on microbial growth. Further exploration revealed that HCP rates did not follow the Arrhenius’ linear response to temperature between 2008 and 2011, demonstrating that microbial growth was substrate-limited following the draught event. By teasing apart the environmental drivers of microbial growth on a long-term perspective, we demonstrated that a substantial change happened in the biogeochemistry of one of the most productive areas of the Mediterranean Sea. As planktonic microbes are the foundation of marine ecosystems, understanding their past dynamics may help to explain present and future changes.

Introduction

By producing, utilizing, and processing one of the largest organic matter (OM) pools on the Earth (Hansell et al., 2009), marine heterotrophic prokaryotes represent the gears spinning beneath the oceans’ biogeochemical engine. With half of the global primary production occurring in the ocean (Field et al., 1998), photosynthetic microorganisms represent one of the major OM sources in marine environments. While most of the phytoplankton-derived OM ends up in dissolved form (DOM, Wagner et al., 2020), particulate OM (POM) may derive either from DOM aggregation or from plankton itself (reviewed by Kharbush et al., 2020). Prokaryotes may access both DOM and POM either by direct uptake of small enough molecules (<600 Da, Weiss et al., 1991) or by enzymatic breakdown of high molecular weight OM (Chróst et al., 1992). The uptake of these low molecular weight organic molecules and their subsequent incorporation into microbial biomass converts DOM into POM, which is then suitable for the consumption by higher trophic levels (Cole and Pace, 1995). Therefore, measuring the heterotrophic carbon production (HCP) under different conditions provides an estimation of both prokaryotic growth and of the rates at which OM is moved from one pool to another (Ducklow and Kirchman, 2000).

Organic matter in the marine environment exists in a wide array of physical (e.g., size fraction, Verdugo et al., 2004) and chemical (e.g., number of organic compounds, Riedel and Dittmar, 2014) forms. This diversity is enhanced in coastal areas, where the proximity to the land and the reduced water column depth determine soil- and seabed-derived OM inputs. Moreover, a collateral, freshwater-mediated, flux of inorganic nutrients may fuel OM production by phytoplankton, making coastal zones hot spots of microbe-mediated OM processing (Celussi et al., 2019). Marine heterotrophic prokaryotes rely, therefore, on a plethora of different OM arrangements to fuel their growth. Like many other features of marine plankton, metabolic rates are controlled by environmental drivers such as temperature, pH, salinity, and organic matter features (Pomeroy and Wiebe, 2001; Arnosti, 2011; Morán et al., 2017). These factors are inherently spatially and temporally variable in coastal areas, posing therefore a challenge to the disentanglement of their role in microbial OM processing rates.

The intrinsic variability of coastal areas reflects the internal biological forcing as well as the intense terrestrial, offshore, and atmospheric forcing affecting these boundary zones. Therefore, inadequate temporal and spatial sampling strategies may over- or underestimate the magnitude and the effects of extreme events and even miss them (Ribera d’Alcalà et al., 2004). Multiyear fixed observations are the key tool for carrying out a reliable estimation of plankton dynamics over time, as their analysis allows to distinguish recurrent patterns from exceptional events (Fuhrman et al., 2015) as time represents the ecosystem's path towards its actual state. Therefore, analysing the ecological processes from a temporal perspective allows to quantify, characterize, and compare the fluxes of energy, matter and information underlying the observed changes (Ribera d’Alcalà, 2019).

Here we used a 21-year-long (1999–2019) biogeochemical time series from the Gulf of Trieste (northern Adriatic Sea). We analysed monthly measurements of HCP, chlorophyll a, particulate and dissolved organic carbon, as well as the abundance of heterotrophic prokaryotes and Synechococcus. Long-term trends were extracted from the time series and clustered with a time-constrained algorithm, to identify coherent periods among the selected variables. The non-parametric, machine learning random forest approach, was used to investigate the relative importance and the partial effect of environmental drivers and OM sources on the observed HCP rates over time. The primary goal of the time series analysis was to identify and describe HCP long-term dynamics, underlining microbe-mediated OM processing. Furthermore, we aimed to assess the importance of OM sources, as well as key environmental drivers, in determining HCP rates across the temporal spanning of the series.

Section snippets

Study area and data

The Gulf of Trieste is a land-locked, river-influenced, shallow (<25 m) embayment occupying the northernmost part of the Adriatic Sea. The main freshwater runoff comes from the Isonzo river outflow, in the north-western part of the Gulf, while freshwater sources along its eastern boundary are of torrential nature (Comici and Bussani, 2007). The river runoff in the area is widely variable, driving broad seasonal and interannual salinity fluctuations with values spanning between 28.5 (Kralj et

Microbial and organic matter long-term dynamics

Descriptive statistics of the analysed time series is reported in Supplementary Table 1. Three coherent time periods were identified in the HCP time series by the time-constrained clustering (Fig. 1 a). Their underlying long-term structure consisted of two major cycles, spanning between 1999 and 2007 and between 2012 and 2019 (red and green clusters in Fig. 1 a, respectively). The cycles peaks (2005 and 2007 for the first cycle, 2014 for the second one, Fig. 1 a) were separated by a U-shaped

Discussion

Our overarching aim was to disentangle the dynamics underlying the microbe-mediated processing of organic matter. To pursue this aim, we used a Random Forest approach, allowing us to model and thus explain the observed HCP rates as a function of selected environmental drivers. The primary reason behind the use of Random Forest relies on its high performance and reliability in the estimation of variable importance. This was especially important as our goal was not to predict HCP rates in future

Conclusion

The analysis of a 21-year-long time series allowed us to investigate time-related patterns underlying the evolution of the “inseparable liaison” between microbes and organic matter (Dittmar and Arnosti, 2018). By means of canonical time series analysis coupled with a machine learning approach, we were able to tease apart, in a temporal fashion, long-term environmental drivers of microbial growth. Microbe-mediated organic matter processing in the northernmost part of the Mediterranean Sea is

CRediT authorship contribution statement

Vincenzo Manna: Conceptualization, Formal analysis, Data curation, Writing - original draft, Writing - review & editing. Cinzia De Vittor: Conceptualization, Data curation, Writing - review & editing. Michele Giani: Conceptualization, Data curation, Writing - review & editing. Paola Del Negro: Conceptualization, Data curation, Writing - review & editing. Mauro Celussi: Conceptualization, Data curation, Writing - review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors would like to thank the personnel of ARPA and of the Miramare MPA for providing the vessels and the crews for sampling activities. Data for this study have been collected thanks to the support of the Regione Friuli Venezia Giulia and of the programs INTERREG Italy-Slovenia 2 and 3. We highly acknowledge the work of all scientists that have been working in microbial ecology and biogeochemistry at our institute over the years and contributed to the dataset construction: M. Borin

References (100)

  • M. Giani et al.

    Recent changes in the marine ecosystems of the northern Adriatic Sea

    Estuar. Coast Shelf Sci.

    (2012)
  • E.C. Grimm

    CONISS: a FORTRAN 77 program for stratigraphically constrained cluster analysis by the method of incremental sum of squares

    Comput. Geosci.

    (1987)
  • M. Kralj et al.

    Hypoxia and dissolved oxygen trends in the northeastern Adriatic Sea (gulf of Trieste)

    Deep Sea Res. Part II Top. Stud. Oceanogr.

    (2019)
  • M. Lipizer et al.

    Effects of intense physical and biological forcing factors on CNP pools in coastal waters (Gulf of Trieste, Northern Adriatic Sea)

    Estuar. Coast Shelf Sci.

    (2012)
  • V. Manna et al.

    Effect of an extreme cold event on the metabolism of planktonic microbes in the northernmost basin of the Mediterranean Sea

    Estuar. Coast Shelf Sci.

    (2019)
  • M. Šolić et al.

    Impact of the 3 °C temperature rise on bacterial growth and carbon transfer towards higher trophic levels: empirical models for the Adriatic Sea

    J. Mar. Syst.

    (2017)
  • Y. Sugimura et al.

    A high temperature catalytic oxidation method for the determination of non-volatile dissolved organic carbon in seawater by direct injection of liquid sample

    Mar. Chem.

    (1988)
  • P. Verdugo et al.

    The oceanic gel phase: a bridge in the DOM-POM continuum

    Mar. Chem.

    (2004)
  • J.K. Apple et al.

    Temperature regulation of bacterial production, respiration, and growth efficiency in a temperate salt-marsh estuary

    Aquat. Microb. Ecol.

    (2006)
  • N. Arandia-Gorostidi et al.

    Testing the metabolic theory of ecology with marine bacteria: different temperature sensitivity of major phylogenetic groups during the spring phytoplankton bloom

    Environ. Microbiol.

    (2017)
  • C. Arnosti

    Microbial extracellular enzymes and the marine carbon cycle

    Ann. Rev. Mar. Sci.

    (2011)
  • F. Azam et al.

    Microbial structuring of marine ecosystems

    Nat. Rev. Microbiol.

    (2007)
  • S. Arrhenius

    Über die Reaktionsgeschwindigkeit bei der Inversion von Rohrzucker durch Säuren

    Z. Phys. Chem.

    (1889)
  • F. Azam et al.

    Significance of bacteria in the mucilage phenomenon in the northern Adriatic Sea

    Ann. Ist. Super Sanita

    (1999)
  • F. Bernardi Aubry et al.

    Picophytoplankton contribution to phytoplankton community structure in the gulf of venice (NW Adriatic Sea)

    Int. Rev. Hydrobiol.

    (2006)
  • F. Bernardi Aubry et al.

    Plankton communities in the northern Adriatic Sea: patterns and changes over the last 30 years

    Estuar. Coast Shelf Sci.

    (2012)
  • P. Biecek

    DALEX: explainers for complex predictive models in R

    J. Mach. Learn. Res.

    (2018)
  • L. Breiman

    Random forests

    Mach. Learn.

    (2001)
  • C.P.D. Brussaard

    Optimization of procedures for counting viruses by flow cytometry

    Appl. Environ. Microbiol.

    (2004)
  • M. Celussi et al.

    Effect of sunlight on prokaryotic organic carbon uptake and dynamics of pigments relevant to photoheterotrophy in the Adriatic Sea Aquat

    Microb. Ecol.

    (2015)
  • F. Cerino et al.

    Phytoplankton temporal dynamics in the coastal waters of the north-eastern Adriatic Sea (Mediterranean Sea) from 2010 to 2017

    Nat. Conserv.

    (2019)
  • R.J. Chróst

    Significance of bacterial ectoenzymes in aquatic environments

  • R.B. Cleveland et al.

    STL: a seasonal-trend decomposition procedure based on loess

    J. Off. Stat.

    (1990)
  • J.J. Cole et al.

    Why measure bacterial production? A reply to the comment by Jahnke and Craven

    Limnol. Oceanogr.

    (1995)
  • C. Comici et al.

    Analysis of the river Isonzo discharge (1998-2005)

    Boll. di Geofis. Teor. ed Appl.

    (2007)
  • S. Cozzi et al.

    Climatic and anthropogenic impacts on environmental conditions and phytoplankton community in the Gulf of Trieste (Northern Adriatic Sea)

    Water

    (2020)
  • B.C. Crump et al.

    Dominance of particle-attached bacteria in the Columbia River estuary, USA

    Aquat. Microb. Ecol.

    (1998)
  • D.R. Cutler et al.

    Random forests for classification in ecology

    Ecology

    (2007)
  • T. Dittmar et al.

    An inseparable liaison: marine microbes and nonliving organic matter

  • H.W. Ducklow et al.

    Oceanic bacterial production

  • H. Ducklow

    Bacterial production and biomass in the oceans

  • J. Faganeli et al.

    C: N: P ratios and stable C isotopic ratios as indicators of sources of organic matter in the gulf of Trieste (northern Adriatic)

    Oceanol. Acta

    (1988)
  • C.B. Field et al.

    Primary production of the biosphere: integrating terrestrial and oceanic components

    Science

    (1998)
  • S. Fonda Umani et al.

    Major interannual variations in microbial dynamics in the Gulf of Trieste (Northern Adriatic Sea) and their ecosystem implications

    Aquat. Microb. Ecol.

    (2007)
  • S. Fonda Umani et al.

    Carbon fluxes in the pelagic ecosystem of the gulf of Trieste (northern Adriatic Sea)

    Estuar. Coast Shelf Sci.

    (2012)
  • J.A. Fuhrman et al.

    Marine microbial community dynamics and their ecological interpretation

    Nat. Rev. Microbiol.

    (2015)
  • T. Galili

    Dendextend: an R package for visualizing, adjusting, and comparing trees of hierarchical clustering

    Bioinformatics

    (2015)
  • V. Galy et al.

    Efficient organic carbon burial in the Bengal fan sustained by the Himalayan erosional system

    Nature

    (2007)
  • M. Giani et al.

    Temporal variability of particulate organic carbon, nitrogen and phosphorus in the Northern Adriatic Sea

    Hydrobiologia

    (2003)
  • S.J. Giovannoni et al.

    Seasonality in ocean microbial communities

    Science

    (2012)
  • Cited by (8)

    View all citing articles on Scopus
    View full text