Narragansett, Rhode Island

2007 Peer Review2007 Peer Review

September 2007

RESEARCH GOALS

The goal of AED’s nutrient effects research program is to construct quantitative relationships between nitrogen loads to estuaries (nitrogen is the limiting nutrient in most estuarine systems) and important estuarine biological responses for application by the Office of Water, the states and authorized tribes as “translators”for effects-based criteria. This poster describes our research concerning the relationship between the spatial extent of eelgrass (Zostera marina) and nitrogen loads in small to medium-sized estuaries in Southern New England.

Objective:

To construct nitrogen load-ecological response models for estuaries of the East Coast, beginning with southern New England.

Specific steps:

• Estimate nitrogen load to estuaries

• Determine eelgrass extent metrics along N gradient

• Estimate residence time

• Construct and revise nitrogen load-ecological response models using residence time or other factors to minimize uncertainty

James Latimer, Giancarlo Cicchetti, Steve Rego, Carol Pesch

Stressor-Response Relationships for Nutrients: Comparative Systems Approach for Development of Nitrogen Load-Eelgrass Response Models for Estuaries

KEY REFERENCES

Nitrogen load-eelgrass response models can be used by the states/tribes in southern NE as part of weight of evidence to determine critical nitrogen loading limits protective of designated uses.

Process:1. Select response value for criteria2. Apply load-response model3. Set nitrogen limit based on model output

Advisors to/collaborators:• EPA LIS Study• EPA OW National Nutrient Coordinator• EPA OW National Estuarine Experts Workgroup• EPA Regions 1, 2• NOAA NCCOS• States of RI, NH, CT

IMPACTS

• Develop estuary eelgrass-based indicators that are not subject to cross-correlation

• Increase number of estuaries in model

• Validate model with other estuaries in same class

• Expand to other estuarine classes (by working with NOAA-NCCOS and other partners)

FUTURE DIRECTIONS

1

10

100

1000

10000

FW

RC

SR

TC

MM

PM

PC

RC

PM

SH

MM

CR

KB

MH

HM

PO

MA

OM

SH

RG

PR

FU

RS

TM

LM

MW

FM

LP

MO

HR

PH

MM

GM

VH

MN

PR

OB

ML

BM

LTM

SA

MM

HM

NR

CC

CM

BS

RG

BR

BT

MK

RR

AH

RA

CM

HR

CW

CR

EB

RB

HC

GC

RF

HM

GC

CW

WM

BH

RA

CR

PR

CB

RC

kg N/ha/yr

EPA needs to provide approaches and methods to develop and apply nutrient criteria that will support designated uses for aquatic systems.

Specifically, EPA needs to define quantitative and causal relationships between varying levels of nutrients and the biological responses of aquatic ecosystems and the resulting services such systems provide.

AGENCY PROBLEM

Nutrients are a major cause of water quality impairments in estuaries, leading to low dissolved oxygen, fish kills, overabundance of nuisance macrophytes, loss of beneficial macrophytes, likely increased sedimentation, and detrimental species shifts of both flora and fauna. The Agency needs to provide tools to assist states and tribes in developing nutrient criteria for estuaries.

Decline of seagrass beds is one of the damaging effects of nitrogen overenrichment. Excess nitrogen leads to a variety of system changes that tend to favor growth of phytoplankton and macroalgae over SAV, leading to a decline in the spatial extent of seagrass.

One of several approaches outlined by U.S. EPA for a state or authorized tribe to develop nutrient criteria is the “effects-based” approach, which relies on nutrient load-response models using the following procedure:

1. Establish numeric criteria for response variables such as eelgrass extent

2. Adopt a procedure to quantitatively address causal parameters (i.e., nitrogen and phosphorus) and determine nutrient loads in specific water body segments that will achieve the response variable criteria (step 1).

This procedure could be a mathematical loading-response model that is referenced in the state or tribal water quality standards as a “translator” for water quality parameters. This translator procedure, together with numeric criteria for response variables, provide a state or authorized tribe with the means to set targets for permit limits, assessment, and total maximum daily loads.

Study System Selection• Selected 40 systems in 2001• Sampled 13 systems in 2002 for eelgrass indicator• Added 9 systems in 2006• Adding 20 more systems in 2007

2001 systems

The load-response translator, together with numeric criteria for response variables, provide a state or authorized tribe with the means to set targets for permit limits, assessment, and total maximum daily loads which will improve water quality in estuaries.

OUTCOMES

Expected outcomes:• Improved water column light intensity from nitrogen

causes• Improved light intensity at leaf from reduced epiphytes• Stable or increased eelgrass extent from nitrogen causes

N SourceAtmospheric Deposition (w+d)

Watershed Surface

Natural Vegetation

65% retained in plants & soil

Turf

62% ret ained in plants & soil (at mos)

Agriculture/Horti culture

62% retained in plant s & soil (atmos )

Impervi ous -roofs/dri veways

62% retained in plants & soil

N Source

Wastewater

Marine Embayment

I mpervi ous -roads/lots/runways

0% ret ained i n plants & soil

35% transported 38% transported (atmos) 38% transported (atmos) 38% transported 100% transported

39% lost as gases ( fert ilizer) 39% lost as gases (fert ilizer)

61% transported (ferti lizer) 61% transported (fert ili zer)

Vadose Zone61% los t

39% transported

Aquifer Zone35% l ost

Septic Tank/Leach Field40% lost

Septic Plumes34% lost

60% transported

66% transported

65% transported

Watershed Subsurface

WWTF

N SourceFertilizer Application

N SourceAtmospheric Deposition (w+d)

Watershed Surface

Natural Vegetation

65% retained in plants & soil

Turf

62% ret ained in plants & soil (at mos)

Agriculture/Horti culture

62% retained in plant s & soil (atmos )

Impervi ous -roofs/dri veways

62% retained in plants & soil

N Source

Wastewater

Marine Embayment

I mpervi ous -roads/lots/runways

0% ret ained i n plants & soil

35% transported 38% transported (atmos) 38% transported (atmos) 38% transported 100% transported

39% lost as gases ( fert ilizer) 39% lost as gases (fert ilizer)

61% transported (ferti lizer) 61% transported (fert ili zer)

Vadose Zone61% los t

39% transported

Aquifer Zone35% l ost

Septic Tank/Leach Field40% lost

Septic Plumes34% lost

60% transported

66% transported

65% transported

Watershed Subsurface

WWTF

N SourceFertilizer Application

Assumptions

• Whole estuary Zostera extent is a good system-level indicator sensitive to nitrogen loading

• Initially, Zostera linear extent is a good proxy for areal extent (subsequently abandoned when new data

became available)

• All other factors would cause additional variance in the ideal model but are not so severe as to require

explicit inclusion for this class of estuary (e.g., light intensity, sulfide levels, substrate type, currents)

• Average nitrogen loading to estuaries is reasonably estimated from watershed loading model

•Flushing time is reasonably estimated using an empirical model and can be estimated using estuarine

area

Rock Weed

Deep SA

Legend

B lack Rock Harbor Landuse

CATEGO RY

A CT-A GR

A GR

NATV EG

OTHER

RD- RN-CM

REC

RES D

�0 675 1,350337.5 Meters

Legend

Pawcatuck River LanduseCATEGO RY

A CT-A GR

A GR

NATV EG

OTHER

RD- RN-CM

REC

RES D

�0 7,200 14,4003,600 Meters

Estimation of Land-side Nitrogen Loading to Study Estuariesusing the Nitrogen Loading Model (NLM)

• Watershed sources: fertilizer, human waste, atmospheric deposition• Watershed surface sinks: natural vegetation, agriculture, pervious surfaces• Watershed subsurface sinks: vadose and aquifer • Atmospheric deposition onto estuarine surface

Source: Valiela et al. (1997)

Airplane derived Eelgrass images collected in 2002• Desired 40% endlap between images; 30% sidelap• Morning flights with low sun angle• Wind less than 10 knots• Tide: ~ 2 hrs. low• Cloud cover < 5%• Acquisition at or near ‘peak biomass’• Altitude ~ 500 ft. (AED specific)

Source: Finkbeiner et al. (2001) C-CAP protocol

Residence Time vs. Area

y = 0.0958x0.332

R2 = 0.89

0.01

0.1

1

0.1 1 10 100 1000

Area (sq. km)

Re

side

nce

Tim

e (m

o) NB

BHGB

OBGC

MC

LBSC

BCAP

(months)

y = 2.92x0.332 (days)

Empirically Derived Estuarine Flushing Time• Literature values for flushing time of 11 estuaries• Estuarine surface area explained 89% variance• Power-law functionality applied to EPA systems

Source: Dettmann (2002)

Dettmann E (2002) Empirical Estimation of Water Residence Times of Non-Riverine Embayments, US EPA/ORD/NHEERL/Atlantic Ecology Division

Finkbeiner M, Stevenson B, Seaman R (2001) Guidance for Benthic Habitat Mapping: an Aerial Photographic Approach. Report No. NOAA/CSC/20117-PUB, U.S. NOAA Coastal Services Center, Charleston, SC

Hauxwell J, Cebrián J, Valiela I (2003) Eelgrass Zostera marina loss in temperate estuaries: relationship to land-derived nitrogen loads and effect of light limitation imposed by algae. marine Ecology Progress Series 247:59-73

Latimer JS, Rego S, Cicchetti G, Pesch C, Dettmann EH, McKinney R, Charpentier M (2007) The Relationship Between Land-Based Nitrogen Loading and Eelgrass Extent for Embayments in Southern New England: Initial Model Construction. Report No. EPA 600 R-07-021, USEPA Office of Research and Development, National Health and Environmental Effects Research Laboratory, Atlantic Ecology Division, Narragansett, RI

Short FT, Burdick DM (1996) Quantifying eelgrass habitat loss in relation to housing development and nitrogen loading in Waquoit Bay, Massachusetts. Estuaries 19:730-739

Valiela I, Collins G, Kremer J, Lajtha K, Geist M, Seely B, Brawley J, Sham CH (1997) Nitrogen loading from coastal watersheds to receiving estuaries: new method and application. Ecol Appl 7:358-380

Nitrogen loading rate (kg N/ha /yr)

100 200 300 400 500

Nitrogen loading rate (kg N/ha /yr)

100 200 300 400 500Nit rogen loading rate (kg N/ha/yr)

50 100 150 200 2500

Nit rogen loading rate (kg N/ha/yr)

50 100 150 200 2500

Published Models for Same Estuarine Type

Short and Burdick (1996) *Hauxwell et al (2003)

Derived Nitrogen Load-Eelgrass Response Model• Estuaries varied in size and loading rate • Scaling factors for eelgrass extent and nitrogen loading were

used to increase variance explained by model• Similar shape and ranges as other studies• Preliminary normalized model was significant at 95% CI and

R2 = 0.82

versus

Flushing Time = f (estuarine area) = f (total shoreline, depth)Eelgrass length, m = f (estuarine volume)

(Loading, mg/d) (Flushing Time, d)

Volume, m3Total Shoreline, m

Eelgrass Length, m

Summary

• The approach is based on comparative systems ecology

• Measured response indicators in many estuaries along a nutrient gradient

Hypothesis: ecological responses will be observable and that they vary according to the level of nutrient inputs

Eelgrass linear vs. area measures

Eelgrass linear indicator• Required because of inadequate photo/sampling• System-level indicator not bed-level• Proxy for areal measures• Exploring traditional areal measures

Nitrogen Loading to Estuaries

A

B

C

D

Ee

lgra

ss (%

of s

hore

line)

0

10

20

30

40

50

60

70

80

90

100

0 5 10 15 20 25 30 35 40 45

Loading Rate (un-normalized, kg N/ yr x103)

y = 1692.6 x -0.4726

r2 = 0.415

0 5 10 15 20 25 30

Loading Rat e (normalized to est uarine volume, g N/m3/yr)

y = 52.5139x- 0.8171

r2 = 0.688

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300 350

Loading Rat e (normalized to estuarine area, kg N/ha/ yr)

y = 551. 9x -0.7418

r2 = 0.331

0 25 50 75 100 125 150 175 200 225 250 275

Loading Rat e (normalized to est uari ne volume and f lushing time, mg N/ m3)

y = 280.1x -0.793

r2 = 0. 819

A

B

C

D

A

B

C

D

Ee

lgra

ss (%

of s

hore

line)

0

10

20

30

40

50

60

70

80

90

100

0 5 10 15 20 25 30 35 40 45

Loading Rate (un-normalized, kg N/ yr x103)

y = 1692.6 x -0.4726

r2 = 0.415

0 5 10 15 20 25 30

Loading Rat e (normalized to est uarine volume, g N/m3/yr)

y = 52.5139x- 0.8171

r2 = 0.688

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300 350

Loading Rat e (normalized to estuarine area, kg N/ha/ yr)

y = 551. 9x -0.7418

r2 = 0.331

0 25 50 75 100 125 150 175 200 225 250 275

Loading Rat e (normalized to est uari ne volume and f lushing time, mg N/ m3)

y = 280.1x -0.793

r2 = 0. 819

~60 kg N/ha/y r eelgrass gone*

~30 kg N/ha/y r eelgrass loss*

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

5,000

0 50 100 150 200 250 300 350

Loading Rate (normalized to estuarine area, kg N ha/yr)

y = 10935x -0.4397

r2 = 0.230

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

5,000

0 5 10 15 20 25 30 35 40 45

Loading Rate (un-normalized, kg N/yr x103)

y = 1070.1x -0.0559

r2 = 0.008

A

B

C

DEel

gras

s (m

)

0 5 10 15 20 25 30

Loading Rate (normalized t o es tuarine volume, g N/m3/yr)

y = 2651.8x -0.3886

r2 = 0.346

0 25 50 75 100 125 150 175 200 225 250 275

Loading Rate (normalized to estuarine volume and flushing t ime, mg N/m3)

y = 4118.4x -0.2598

r2 = 0.176

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

5,000

0 50 100 150 200 250 300 350

Loading Rate (normalized to estuarine area, kg N ha/yr)

y = 10935x -0.4397

r2 = 0.230

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

5,000

0 5 10 15 20 25 30 35 40 45

Loading Rate (un-normalized, kg N/yr x103)

y = 1070.1x -0.0559

r2 = 0.008

A

B

C

D

A

B

C

DEel

gras

s (m

)

0 5 10 15 20 25 30

Loading Rate (normalized t o es tuarine volume, g N/m3/yr)

y = 2651.8x -0.3886

r2 = 0.346

0 25 50 75 100 125 150 175 200 225 250 275

Loading Rate (normalized to estuarine volume and flushing t ime, mg N/m3)

y = 4118.4x -0.2598

r2 = 0.176

~30 kg N/ha/y r eelgrass loss*

~60 kg N/ha/y r eelgrass gone*

AED’s Preliminary Load-Response Models Using Linear Eelgrass Indicator

y = 0.890x + 7720 r2 = 0.967***

050

100

150

200

250

300

350

400

450500

0 50 100 150 200 250 300 350 400 450 500

y = 0.938x + 5820 r2 = 0.729***

0

20

40

60

80

100

120

140

0 20 40 60 80 100 120 140

SPARROW (kg N/yr x 103)

NL

M (

kg

N/y

r x

103 )

A

B

y = 0.890x + 7720 r2 = 0.967***

050

100

150

200

250

300

350

400

450500

0 50 100 150 200 250 300 350 400 450 500

y = 0.938x + 5820 r2 = 0.729***

0

20

40

60

80

100

120

140

0 20 40 60 80 100 120 140

SPARROW (kg N/yr x 103)

NL

M (

kg

N/y

r x

103 )

y = 0.890x + 7720 r2 = 0.967***

050

100

150

200

250

300

350

400

450500

0 50 100 150 200 250 300 350 400 450 500

y = 0.890x + 7720 r2 = 0.967***

050

100

150

200

250

300

350

400

450500

0 50 100 150 200 250 300 350 400 450 500

y = 0.938x + 5820 r2 = 0.729***

0

20

40

60

80

100

120

140

0 20 40 60 80 100 120 140

SPARROW (kg N/yr x 103)

NL

M (

kg

N/y

r x

103 )

y = 0.938x + 5820 r2 = 0.729***

0

20

40

60

80

100

120

140

0 20 40 60 80 100 120 140

SPARROW (kg N/yr x 103)

NL

M (

kg

N/y

r x

103 )

A

B

NLM vs. SPARROW Loading to Estuaries

Preliminary Load-Response Model using Area (State of MA data)

Random Numbers Subjected to

Scaling Factors

The use of ratios in the X and Y axes can lead to X-Y cross-correlation:• Scaling factors (as ratios), if used in both the X and Y variables, may

introduce a pattern on the data• Ideally, data should show minimal co-dependence between X and Y

variables or their scaling factors

Nitrogen FromWatershed Nitrogen From

Sea

Phytoplankton

Epiphytes

Nitrogen FromAtmosphere

SuspendedSolids

Light fromSun

Light Attenuation

SedimentType

Sediment ProcessessNutrient Cycling Regenearation

WaterFlushingTime

Nitrogen FromWatershed

Nitrogen FromAtmosphere

WaterFlushingTimeEelgrass

Eelgrass

Nitrogen FromWatershed Nitrogen From

Sea

Phytoplankton

Epiphytes

Nitrogen FromAtmosphere

SuspendedSolids

Light fromSun

Light Attenuation

SedimentType

Sediment ProcessessNutrient Cycling Regenearation

WaterFlushingTime

Nitrogen FromWatershed

Nitrogen FromAtmosphere

WaterFlushingTimeEelgrass

EelgrassComplex:• multiple processes• multiple pathways

Simplified:processes combined and “captured” by: • stressor (N loading)• mitigating factors (flushing time, morphometry)

Correlation Matrix for Variables Nitrogen Loading from NLM to Embayments• Ranged from 20 – 5,000 kg N/ha/yr (20-325 for eelgrass)• Compared favorably with SPARROW estimates

Note x-axis scales are not equivalent

Note x-axis scales are not equivalent in all areal normalized models

Red bars indicate those included in seagrass model development

y = -91.753Ln(x) + 428.15

R2 = 0.6259

0

20

40

60

80

100

120

140

160

180

0 20 40 60 80 100 120Loading Rate (kg N/ha/yr)

Ee

lgra

ss E

xte

nt

(ha

)

20011995