Assessing the metabolic rates of eight hemiboreal lakes with high-frequency measurements and Bayesian modelling Fabien CREMONA, Alo LAAS, Peeter NÕGES, and Tiina NÕGES Centre for Limnology Institute of Agricultural and Environmental Sciences Estonian University of Life Sciences Tartu (Estonia)
Net Ecosystem Production (NEP) = Gross Primary Production (GPP) – Ecosystem Respiration (ER) NEP > 0 lake is autotrophic NEP < 0 lake is heterotrophic To calculate a complete diel NEP cycle (1) necessary to obtain estimates of GPP and ER + (2) computational procedure for constructing estimates (1) Measurements of changes in dissolved oxygen (DO) concentration in the surface layer corrected for O2 exchange at the air–water interface have been the most popular approach to date (Staehr et al., 2012) D: First we have to define what is lake metabolism. It can be summarized by NEP; F: on this photo you can see a HF buoy... HF measurement buoy in Lake Võrtsjärv, Estonia
the tracing of diel lake metabolic rates is possible (2) Two approaches are used for constructing estimates of lake metabolism: 2a) Back-calculating GPP and ER from an observed timeseries of DO measurements (Odum, 1956): „bookkeeping approach“. Unable to assess the daytime component of ER and thus cannot accurately trace dark–light cycles 2b) Process-based models of DO dynamics that predict time-series of DO concentrations from a variety of environmental variables (McNair et al., 2013). the tracing of diel lake metabolic rates is possible D: for the 2nd step of the analysis, two approaches can be used. F:This is this second approach that interests us here.
BaMM (Bayesian Metabolic Model, Holtgrieve et al. 2010) main features -can estimate simultaneously GPP, ER and their uncertainties based on changes in DO concentration, water temperature, and PAR irradiance -generates DO time-series, compares them with observed data and it is then able to select the most efficient parameter values for reconstructing the observed data -needs relatively few parameters; -has an irradiance sub-model; -BaMM has been tested with 10-15-min time steps making it suitable to use with HF data; D: Here is the model we have employed, which is called BaMM. F: lire
BaMM conceptual model Methods Cremona et al. 2014 Ecological Modelling Time-integrated output data GPP G ER d[O2]/dt BaMM estimated parameters P (I) G R (T) Measured diel data PAR Irradiance Water T D: the BaMM conceptual model. You must read this figure from bottom to top. F: ...the model calculates the time-integrated output data DO Pmax αP-I k20 Salinity R20 User-defined parameters Environmental dataa Areab Depthb Start. O2 conc.b I @ ½ Sat.b Cremona et al. 2014 Ecological Modelling
BaMM equations Methods D: Here are the model equations. Equations that are strikethrough are those related to O isotopes values. Holtgrieve et al. 2010 Limnology and Oceanography
-Lake mixing depth= thermocline -k set to uniform distribution along a wide range (2 orders of magnitude) -no use of δ18O -closest weather station data (except for Võrtsjärv and Valguta Mustjärv) -frequency of DO measurements: 10 min -> 144 time steps for 24 h -frequency of Irradiance measurements: 30 min -> irradiance curve smoothed by BaMM D: some important notes regarding the modelling processes. F: ...as you can see in this fig, so that it can be used as input data
D:our 8 study lakes located all over Estonia (in blue) Harku Ülemiste Väike-Maarja Äntu Sinijärv Tiirikoja Peipsi Saadjärv D:our 8 study lakes located all over Estonia (in blue) Mullutu Valguta Mustjärv Tõravere Võrtsjärv Roomassaare Erastvere Võru
Methods Cremona et al. 2014 Ecological Modelling HF measurement buoy in lakes: O2, T Weather stations: PAR, wind speed (-> k) D: the main steps of the modelling. F: the bars correspond to the min and max credible intervals Metabolic values displayed with 97.5% and 2.5% credible intervals Cremona et al. 2014 Ecological Modelling
Results Cremona et al. 2016 Inland Waters D: the model has captured quite well the diel changes of PP. Cremona et al. 2016 Inland Waters
Results Erastvere Cremona et al. 2016 Inland Waters
Results Mullutu Cremona et al. 2016 Inland Waters
Results Cremona et al. 2016 Inland Waters
Results Cremona et al. 2016 Inland Waters
Results Cremona et al. 2016 Inland Waters
Results Cremona et al. 2016 Inland Waters
Results Cremona et al. 2016 Inland Waters
Metabolic state and uncertainties Results D: in this table you can see the metabolic state of the lakes according to the model results Cremona et al. 2016 Inland Waters
Coupling between primary production and respiration Results Slope of the R / P relationship 0.24 (oligotrophic) 1.42 (eutrophic) 0.33 (mesotrophic) 0.96 (eutrophic) 0.91 (eutrophic) 0.84 (eutrophic) In this figure we have plotted modelled Respiration against modelled PP. Higher proportion of autochthonous OM consumed in eutrophic lakes Cremona et al. 2016 Inland Waters
-> priming of PP by benthic respiration ? If you look closer and compare coupling with limnological parameters... -> priming of PP by benthic respiration ? Cremona et al. 2016 Inland Waters
Comparison of BaMM outputs with „bookkeeping“ approach Cremona et al. 2016 Inland Waters
Conclusion -According to the model results, 2 lakes are autotrophic, 1 heterotrophic, 5 are at a balanced metabolic state -Strong coupling between primary production and respiration in most lakes -Higher respiration of autochthonous organic matter in eutrophic lakes than in oligotrophic / mesotrophic lakes
Thank you for your attention ! Literature cited Holtgrieve et al. (2010) Limnology & Oceanography 55:1047-1053. Cremona et al. (2014) PLoS One: e101845. Cremona et al. (2014) Ecological Modelling 294:27-35. Cremona et al. (2016) Inland Waters 6:352-363 Acknowledgements This research was supported by Start-Up Personal Research Grant PUT 777 to FC and IUT 21-2 of the Estonian Ministry of Education and Research, by Estonian Science Foundation grant ETF9102, by the Swiss Grant "Enhancing public environmental monitoring capacities" and by MARS project (Managing Aquatic ecosystems and water Resources under multiple Stress) funded under the 7th European Union Framework Programme, Theme 6 (Environment including Climate Change), Contract No.: 603378 (http://www.mars-project.eu).
Uncertainty assessment NEP A B C D NEPmax NEPmin A: Certainly autotrophic C: Likely heterotrophic B: Likely autotrophic D: Certainly heterotrophic Cremona et al. 2014 Ecological Modelling