Integrating Environmental Genomics Into Biogeochemical Models

Size: px
Start display at page:

Download "Integrating Environmental Genomics Into Biogeochemical Models"

Transcription

1 Integrating Environmental Genomics Into Biogeochemical Models SEED Fund Update Sarah Preheim, Department of Environmental Health and Engineering Anand Gnanadesikan, Department Earth and Planetary Sciences

2 Microorganisms are a dominant form of biomass on Earth: what they do matters 1 l of seawater Natural water: ~ bacteria/ μl

3 Photo credits at Chesapeake Bay Program Poor water quality is often related to microbial activity Algal blooms Dead zone Beach closures Loss of habitat

4 Organic enrichment/oxygen depletion, i.e., high levels of oxygen-demanding substances and/or low levels of dissolved oxygen (e.g., organic waste); and Oxygen depletion is a major issue Mercury, a toxic metal found in fish tissue, and, to a lesser extent, in the water column, often entering the aquatic environment via atmospheric deposition. impacting water quality in the US Toxic organics, nutrients, pesticides, and metals are also reported as top causes of impairment for estuarine waters. Note: Percents do not add up to 100% because more than one cause may affect a waterbody. Figure 8. Top 10 causes of impairment in assessed bays and estuaries. Report to Congress, 2004 Reporting Cycle

5 Varadharajan C MIT Thesis Microbial processes in anoxic environments have local and global impacts Change the mobility of contaminants Impact nutrient cycling (e.g. denitrification) Greenhouse gas production and emission Contaminant/nutrient remobilization from the gas-saturated sediments, rather than from the water column where dissolved gas concentrations are lower. Preexisting bubbles or gas cavities in the sediments can serve as sites David B. Senn, and Harold F. Hemond Science 2002;296: for bubble formation and nucleation, whereas very high overpressures are required for Greenhouse gas emission spontaneous bubble formation within the water column (Jones et al. 1999).

6 Major challenges in microbial ecology Develop a mechanistic understanding of community structure and function Linking metagenomic analysis to ecosystem-level models Predict changes in microbial community structure and function

7 Accurate models aid remediation efforts to restore water quality Restoration efforts: limiting sediment, fertilizer, waste and pollution Accurate predictions inform regulation and support remediation efforts Current models do not typically include many microbial processes Do missing processes impact predictions?

8 Measure Research Approach Chemical concentrations and bacterial diversity Model Bacterial population capabilities/interactions Biogeochemical model Predict Community response to and affecting environmental change

9 Upper Mystic Lake is a model system for studying microbial ecology

10 depth Mystic Lake is a model system for studying dead-zone microbiology chemical concentration summer 10*NO3 (mg N/L) HFO 2008 Average DO (mg/l) SO4 (mg S/L) thermocline C thermocline 4 C oxygen sulfate nitrate Iron(II) Illustration of typical lake mixing

11 Mar 26 May 10 June 17 July 17 Aug 15 Measure chemical and biological changes with depth at high resolution surface 0 m 1.5 m 3 m 4 m 5 m 6 m 7 m 8 m 9 m 10 m 11 m 13 m 14 m 15 m 16 m 17 m 19 m 20m 21m 22m Sampling schedule sediment Collect water at 1 meter intervals 2013

12 Observe changes in both population and gene distribution Populations 16S rrna libraries for 5 months Method 6 runs on Illumina MiSeq Genes Shotgun metagenomic libraries for 1 month Method 1 run on Illumina HiSeq

13 Research Approach Measure Chemical concentrations and bacterial diversity -At high spatial and temporal resolution Model Bacterial populations/ capabilities/interactions Biogeochemical model Predict Community response to and affecting environmental change

14 Assumption of population modeling: gene organization is informative microbial population Organism microbial interactions Gene narg nirk norb nosz Genome NO - 3 NO - 2 NO N 2 O N 2 microbially mediated process microbial community

15 Identify key populations and capabilities from sequence data Identify populations from genetics and ecology/distribution Infer capabilities from genome reconstruction and similarity organisms with known function

16 Genomic bins from metagenomic data using ecological information Assess population capabilities through genome reconstruction (metagenomic binning) Binning technique uses ecological distribution to cluster genes Albertsen et al Nature Biotechnology (6): 533

17 Arora-Williams et al In Prep Genome reconstruction data match population data

18 Modeling the feedback between biology and chemical environment Bacterial Populations Integrated metabolism Environmental Change Feedback Preheim, Olesen et al 2015 Submitted

19 Modeled microbial processes driving changes in lake chemistry primary redox reactions redox zones reaction fronts secondary redox reactions surface sediment Hunter et al Journal of Hydrology 209:53-80 Scott Olesen

20 Biogeochemical model provides mechanistic understanding of key population distributions Preheim, Olesen et al 2016 Nature Microbiology 1:16130

21 Metagenomic bins identified missing coupled processes Bin No. Carbon fixation Methane oxidation Denitrification Sulfur oxidation rcbl pmoab narg nirk nirs norb nosz dsra apra sat Arora-Williams et al In Prep

22 Adding processes to model significantly alters predicted biogeochemistry

23 Towards a mechanistic understanding of microbial processes in Chesapeake Bay Chesapeake Bay has more complicated hydrodynamics Lighter, fresh water on top of heavy, salt water Salt water isolated from atmosphere at the mouth of the Bay Mixing could influence microbial processes Mixing Crump et al 1999 AEM 65(7):3192

24 ChesROMS: a tool to model biogeochemical dynamics in the Chesapeake Bay Predictions for sea nettles, HAB and water quality Missing sulfur and iron cycles Banakar et al 2011 EcoHealth 8, , 2011

25 * Station CB CB CB CB3.3C CB4.1C 3 9 CB4.2C 3 9 CB4.3C 3 9 CB Mainstem sampling by: Maryland Department of Natural Resources (DNR) * * * ** CB CB CB CB CB CB CB CB CB CB CB CB Old Dominion Water Quality Lab (ODU) * * * * *

26 Chesapeake Bay sample sequencing effort Populations 16S rrna libraries for all samples Method 3 runs on Illumina MiSeq and counting Genes Shotgun metagenomic libraries for at least one profile and transect Method 1 run on Illumina HiSeq and counting

27 Dynamics of key populations shows community response to chemical changes Preheim lab preliminary data

28 Distribution of key genes in biogeochemical pathways with depth in Aug, 2015 Preheim lab preliminary data

29 Integrating Environmental Genomics Into Biogeochemical Models Microbial processes impact water quality Developing better models will aid remediation efforts Upper Mystic Lake: A model dead-zone ecosystem: Population and gene distributions support mechanistic model Neglecting processes could significantly impact predicted biogeochemistry Populations and processes in the Chesapeake Bay dead-zone: Sampling efforts and preliminary Community responsive to chemical changes

30 Acknowledgements Massachusetts Institute of Technology Eric Alm, Scott Olesen, Sarah Spencer, Harry Hemond, Ben Scandella, Kyle Delwich Johns Hopkins University Eric Sakowski, Keith Arora-Williams, Sonali Abraham, Chris Holder, Anand Gnanadesikan Maryland Department of Natural Resources Kristen Heyer, Laura Fabian Old Dominion University, Water Quality Lab Suzanne Doughten Resources provided Institute for Data Intensive Engineering and Science (IDIES) seed grant Maryland Advanced Research Computing Center (MARCC)