Post on 28-Apr-2022
11/23/200911/23/2009
11
MERGANSER:MERGANSER:A Model for Predicting MercuryA Model for Predicting MercuryA Model for Predicting MercuryA Model for Predicting Mercury
Levels in Wildlife in New EnglandLevels in Wildlife in New England
Alison C. Simcox, PhDAlison C. Simcox, PhD2009 Mercury Science & Policy Conference Great 2009 Mercury Science & Policy Conference Great
Lakes & NE RegionsLakes & NE RegionsggNovember 18, 2009November 18, 2009
AcknowledgmentsAcknowledgments
EPA: Jeri Weiss, Diane Nacci, John Johnston, ORD (AMI)EPA: Jeri Weiss, Diane Nacci, John Johnston, ORD (AMI)
NESCAUM: John Graham, NESCAUM: John Graham, Kathleen FaheyKathleen Fahey
USGS: Jamie Shanley, Richard Moore, Craig Johnston, USGS: Jamie Shanley, Richard Moore, Craig Johnston, Keith Robinson, Richard SmithKeith Robinson, Richard Smith
VT DEC: Neil KammanVT DEC: Neil Kamman
Biodiversity Research Institute (BRI): David EversBiodiversity Research Institute (BRI): David EversBiodiversity Research Institute (BRI): David EversBiodiversity Research Institute (BRI): David Evers
Ecosystems Research Group Ltd: Eric MillerEcosystems Research Group Ltd: Eric Miller
NEIWPCC: Susy KingNEIWPCC: Susy King
Others on MERGANSER Project teamOthers on MERGANSER Project team
11/23/200911/23/2009
22
OutlineOutlineGenesis of idea for MERGANER: New England Genesis of idea for MERGANER: New England SPARROWSPARROWConceptual modelConceptual modelAlternative model forms/estimated equationAlternative model forms/estimated equationPredictor & Response VariablesPredictor & Response VariablesIssues & Current focusIssues & Current focusEndpoints Linking MERGANSER ith policEndpoints Linking MERGANSER ith policEndpoints: Linking MERGANSER with policyEndpoints: Linking MERGANSER with policy
Genesis of idea for MERGANER: Genesis of idea for MERGANER: New England SPARROWNew England SPARROW
“Spatially Referenced Regressions on Watershed “Spatially Referenced Regressions on Watershed Attributes” Attributes” -- uses regression equations to relate uses regression equations to relate total N & total N & P (nutrient) stream loads to nutrient sources and P (nutrient) stream loads to nutrient sources and watershed characteristics (USGS, watershed characteristics (USGS, Smith et al, 1993 & Smith et al, 1993 & 1997;1997; Moore et al, 2004)Moore et al, 2004)Produces estimates, Produces estimates, with uncertainty estimates,with uncertainty estimates, of nutrient of nutrient loads loads -- flux (kg/yr), yields(kg/kmflux (kg/yr), yields(kg/km22/yr), and conc (mg/l) /yr), and conc (mg/l) -- in in unmonitored stream reaches/watersheds in modeledunmonitored stream reaches/watersheds in modeledunmonitored stream reaches/watersheds in modeled unmonitored stream reaches/watersheds in modeled region (e.g., New Eng, US)region (e.g., New Eng, US)Used in TMDL and nutrientUsed in TMDL and nutrient--criteria programs, tracking criteria programs, tracking nutrient sources/delivery in Gulf of Mex., Chesapeake Bay, nutrient sources/delivery in Gulf of Mex., Chesapeake Bay, Long Island Sound Study, etc.Long Island Sound Study, etc.
11/23/200911/23/2009
33
NE SPARROW Input Nutrient Sources
Point Sources
Atm deposition of N(Ollinger,1992)
National Land CoverDataset 1992
- Agriculture- Urban- Forest
ProcessesLand to water deliveryy
Soil permeability –STATSGO
In-stream lossMean ann. streamflowReservoir detention
SPARROW Estimated Equation
SPARROW Estimated Equation
Load Land-to-water Error
Sourcesdelivery
Instreamdecay
11/23/200911/23/2009
44
RF1
2,462 Reaches
NE SPARROWEnhanced NHD
42,000 Reaches
11/23/200911/23/2009
55
Modifying New England Modifying New England SPARROWSPARROWfor Mercuryfor Mercury
Key insights:Key insights:
–– View fishView fish--tissue (and tissue (and piscivore) piscivore) data from lake areas data from lake areas as dependent variable (analogous to data from WQ as dependent variable (analogous to data from WQ monitoring stations)monitoring stations)
–– View mercury model as risk model rather than View mercury model as risk model rather than transport model transport model –– i.e., predict mercury levels in i.e., predict mercury levels in wildlife for any lake in New Englandwildlife for any lake in New Englandwildlife for any lake in New Englandwildlife for any lake in New England
Fish Consumption Advisories Fish Consumption Advisories –– 20082008(EPA (EPA -- National Fish and Wildlife Contamination ProgramNational Fish and Wildlife Contamination Program
11/23/200911/23/2009
66
Mercury in freshwater fishMercury in freshwater fish(Northeast North America)(Northeast North America)
Four species with the highest mean Hg concentrations:Four species with the highest mean Hg concentrations:Four species with the highest mean Hg concentrations: Four species with the highest mean Hg concentrations: muskellunge, walleye, white perch, and northern pikemuskellunge, walleye, white perch, and northern pike
Waterbodies exceeding EPA criterion for fish Hg (0.3 Waterbodies exceeding EPA criterion for fish Hg (0.3 ppm) ranged from 14% for standardppm) ranged from 14% for standard--length brook trout length brook trout fillets to 42% for standardfillets to 42% for standard--length yellow perch fillets.length yellow perch fillets.
[Source: Kamman et al., 2005 Ecotoxicology 14(1[Source: Kamman et al., 2005 Ecotoxicology 14(1--2)]2)]
Biological Mercury Hotspots Biological Mercury Hotspots (piscivores)(piscivores)
Evers et al. 2007. Biological mercury hotspots in the Northeastern United Statesand Southeastern Canada. Bioscience 57: 29-43
11/23/200911/23/2009
77
MERGANSER:MERGANSER:Structure and TimelineStructure and Timeline
MERcury GeoMERcury Geo--spatial AssessmeNtS for the New England spatial AssessmeNtS for the New England R iR iRegionRegion
EPA ORD funding source EPA ORD funding source ––Advanced Monitoring Initiative. Advanced Monitoring Initiative. Grant awarded Dec 06; NESCAUM contract April 08Grant awarded Dec 06; NESCAUM contract April 08
–– John Johnson, EPAJohn Johnson, EPA--Athens, and I are coAthens, and I are co--P.O.s; John P.O.s; John Graham is NESCAUM contract managerGraham is NESCAUM contract manager
Project team includes federal (EPA USGS) interstateProject team includes federal (EPA USGS) interstate–– Project team includes federal (EPA, USGS), interstate Project team includes federal (EPA, USGS), interstate (NESCAUM), state (VT), and academic Hg researchers (NESCAUM), state (VT), and academic Hg researchers (BRI, ERG)(BRI, ERG)
–– Projected model completion: June 2010Projected model completion: June 2010
–– Products: reports, journal articles, web accessProducts: reports, journal articles, web access
Alternative Model FormsAlternative Model Forms
(a) Standard multiple linear regression equation (least (a) Standard multiple linear regression equation (least squares formulation)squares formulation)
(b) “Multiplicative" model (b) “Multiplicative" model -- explanatory variables explanatory variables separated into two types: mercury "source" and separated into two types: mercury "source" and "process" variables"process" variables
–– Source terms each multiplied by one or more process Source terms each multiplied by one or more process terms (factors that quantify transfer of mercury fromterms (factors that quantify transfer of mercury fromterms (factors that quantify transfer of mercury from terms (factors that quantify transfer of mercury from each of sources to fish tissue)each of sources to fish tissue)
11/23/200911/23/2009
88
MERGANSERMERGANSER estimated equationestimated equation(multiplicative model)(multiplicative model)
nn
j=1j=1
mm
k=1k=1
nn
j=1j=1
mm
k=1k=1
MERGANSERMERGANSER Response VariablesResponse Variables
Match fish data points to MERGANSER lakesMatch fish data points to MERGANSER lakesYellow Perch Equivalent (YPE): MERGANSER model(s) Yellow Perch Equivalent (YPE): MERGANSER model(s) will predict mercury levels in standard fish units for New will predict mercury levels in standard fish units for New England Lakes (MEngland Lakes (Mii))–– Exploring inclusion of fish species and lengthExploring inclusion of fish species and length
Approaches for loons (possibly extend to other Approaches for loons (possibly extend to other piscivores):piscivores):
Direct con ersion from YPE sing empirical relationsDirect con ersion from YPE sing empirical relations–– Direct conversion from YPE using empirical relationsDirect conversion from YPE using empirical relations–– Model using full predictor variable setModel using full predictor variable set
11/23/200911/23/2009
99
Projections vs Loon Projections vs Loon observationsobservations
Mercury Risks to New England LoonsMercury Risks to New England Loons
Projected Loon Projected Loon ResponsesResponses
40
60
80
100
120
140
160
Per c
ent F
ledg
lings
Hg effects integrated through Loon population model (Grear et al 2009)
Spatial- and temporal projections of Hg risks to
NE Loons
1515
Dietary Hg effects on Loons and other
avian species
0
20
40
0 1 2 3 4 5 6
Maternal Dietary MeHg, mg/kg dry weight
MERGANSER model(s) predicted standard Fish Hg for NE Lakes
MERGANSERMERGANSER Predictor VariablesPredictor Variables
–– Dry deposition breakdown (mercury form affects Dry deposition breakdown (mercury form affects bioavailability)bioavailability)y)y)
–– Total mercury deposition to lakeTotal mercury deposition to lake–– pH (proxy) pH (proxy) –– Sulfur depositionSulfur deposition–– NWI wetlands categoriesNWI wetlands categories–– Percent wetland area contiguous to lake Percent wetland area contiguous to lake –– SPARROW phosphorusSPARROW phosphorus–– SPARROW phosphorusSPARROW phosphorus–– Percent watershed as upgradient lake (=total “water” Percent watershed as upgradient lake (=total “water”
area minus lake area). Start at getting at nested lake area minus lake area). Start at getting at nested lake issueissue
11/23/200911/23/2009
1010
pH ProxypH Proxy
11/23/200911/23/2009
1111
Sulfur DepositionSulfur Deposition
Public Database: Lake Quality and Public Database: Lake Quality and Fish/wildlife Mercury DataFish/wildlife Mercury Data
http://www.epa.gov/aed/html/wildlife/index.htmlhttp://www.epa.gov/aed/html/wildlife/index.html
11/23/200911/23/2009
1212
MERGANSER IssuesMERGANSER Issues
Large lakesLarge lakesDivide into contributing watersheds? Problem: fewDivide into contributing watersheds? Problem: few–– Divide into contributing watersheds? Problem: few Divide into contributing watersheds? Problem: few lakes have fish Hg from multiple sites. (Large lakes lakes have fish Hg from multiple sites. (Large lakes generally have data so don’t need model predictions)generally have data so don’t need model predictions)
–– For loons, much of data are from multiple territories For loons, much of data are from multiple territories on lake, so largeon lake, so large--lake issue more pertinentlake issue more pertinent
–– Lake Champlain won’t include in model (would need Lake Champlain won’t include in model (would need too much data from outside New England)too much data from outside New England)too much data from outside New England) too much data from outside New England)
Nested lakesNested lakes–– At very least, flag them to evaluate their model At very least, flag them to evaluate their model
residuals as a groupresiduals as a group
Current FocusCurrent Focus
USGS is completing independent variable setUSGS is completing independent variable setUSGS is evaluating linear (least squares) andUSGS is evaluating linear (least squares) andUSGS is evaluating linear (least squares) and USGS is evaluating linear (least squares) and multiplicative modeling approachesmultiplicative modeling approaches–– Identify key independent variablesIdentify key independent variables
Evaluating "source additive" nonlinear model form (with Evaluating "source additive" nonlinear model form (with sourcesource--specific multiplicative processing terms)specific multiplicative processing terms)Preliminary results of linear and multiplicative modeling Preliminary results of linear and multiplicative modeling will be presented at AGU conference in Dec 2009will be presented at AGU conference in Dec 2009
11/23/200911/23/2009
1313
Endpoints:Endpoints:Linking Linking MERGANSERMERGANSER with policywith policy
MERGANSER will provide information about:MERGANSER will provide information about:–– Natural featuresNatural features that contribute to mercury risk (e.g., that contribute to mercury risk (e.g.,
t h d i f tl d l H)t h d i f tl d l H)watershed size, presence of wetlands, low pH) watershed size, presence of wetlands, low pH) –– HumanHuman--influenced conditionsinfluenced conditions that can be modified for that can be modified for
desired outcomes (e.g., location of mercury sources)desired outcomes (e.g., location of mercury sources)Model outputs/products:Model outputs/products:–– Mercury levels in standard fish units for New England Mercury levels in standard fish units for New England
lakes (possibly streams)lakes (possibly streams)–– Projected mercury levels in piscivorous birds (possibly Projected mercury levels in piscivorous birds (possibly
other piscivores)other piscivores)other piscivores)other piscivores)–– “Ecological risk” metrics (e.g., for loons)“Ecological risk” metrics (e.g., for loons)–– Optimal locations for longOptimal locations for long--term mercury monitoring (for term mercury monitoring (for
National Mercury Monitoring Network)National Mercury Monitoring Network)–– Predicted "hotspots" of deposition and exposure (including Predicted "hotspots" of deposition and exposure (including
those linked with specific sources)those linked with specific sources)
For More Information…For More Information…
EPA Region 1:EPA Region 1:Alison Simcox (617) 918Alison Simcox (617) 918--16841684simcox alison@epa govsimcox alison@epa govsimcox.alison@epa.govsimcox.alison@epa.gov
NESCAUM:NESCAUM:John Graham (617) 259John Graham (617) 259--20232023jgraham@nescaum.orgjgraham@nescaum.org
USGS:USGS:USGS:USGS:Jamie Shanley (802) 828Jamie Shanley (802) 828--44664466jshanley@usgs.govjshanley@usgs.gov