1 Sediment Quality Objectives for California Enclosed Bays and Estuaries Benthic Indicator Development Scientific Steering Committee 26 th July 2005.

1

Sediment Quality Objectivesfor California Enclosed Bays and Estuaries

Benthic Indicator Development

Scientific Steering Committee

26th July 2005

2

Overview

• Why Benthos and Benthic Indices?

• The Index Development Process– Define Habitat Strata

– Calibrate Candidate Benthic Indices

– Validate and Evaluate Candidate Indices

• Proposed Next Steps

3

Why Benthos?

• Benthic organisms are living resources– Direct measure of what legislation intends to protect

• They are good indicators– Sensitive, limited mobility, high exposure, integrate impacts,

integrate over time

• Already being used to make regulatory and sediment management decisions– Santa Monica Bay removed from 303(d) list

• Listed for metals in the early 1990’s

– 301(h) waivers granted to dischargers– Toxic hotspot designations for the Bay Protection and Toxic Cleanup

Program

4

Benthic Assessments Pose Several Challenges

• Interpreting species abundances is difficult– Samples may have tens of species and hundreds of organisms

• Benthic species and abundances vary naturally with habitat– Different assemblages occur in different habitats

– Comparisons to determine altered states should vary accordingly

• Sampling methods vary– Gear, sampling area and sieve size affect species and individuals

captured

5

Benthic Indices Meet These Challenges

• Benthic Indices– Remove much of the subjectivity associated with data

interpretation– Account for habitat differences– Are single values – Provide simple means of

• Communicating complex information to managers• Tracking trends over time• Correlating benthic responses with stressor data

– Are included in the U.S. EPA’s guidance for biocriteria development

6

Overview






7

Define Habitat Strata

• Rationale– Species and abundances vary naturally from

habitat to habitat• Benthic indicators and definitions of reference

condition should vary accordingly

• Objectives– Identify naturally occurring benthic

assemblages, and– The habitat factors that structure them

8

Approach

• Identify assemblages by cluster analysis– Standard choices

• Species in ≥ 2 samples

• ³√ transform, species mean standardization

• Bray Curtis dissimilarity with step-across adjustment

• Flexible sorting ß=-0.25

• Evaluate habitat differences between assemblages– Salinity, % fines, depth, latitude, longitude, TOC

– Using Mann-Whitney tests

9

Data• EMAP data enhanced by regional data sets

– Comparable methods• Sampling, measurements, taxonomy

– OR and WA data included• Potential to increase amount of data for index development

– 1164 samples in database

• Eliminated potentially contaminated sites– ≥ 1 chemical > ERM or ≥ 4 chemicals > ERL– Toxic to amphipods– Located close to point sources– DO < 2 ppm

• 714 samples analyzed

10

Identified Eight Assemblages

A Puget Sound Fine Sediments

B Puget Sound Coarse Sediments

C Southern California Euhaline Bays

D Polyhaline San Francisco Bay

E Estuaries and Wetlands

F Very Coarse Sediments

G Mesohaline San Francisco Bay

H Limnetic or Freshwater

11

SalinityS

alin

ity (

psu

)

0

10

20

30

40

Assemblage

A B C D E F G H

% Fine Sediments

Fin

e s

ed

ime

nts

(%

)

0

20

40

60

80

100

Assemblage

A B C D E F G H

Depth

Bo

tto

m d

ep

th (

m)

0

50

100

150

200

Assemblage

A B C D E F G H

Latitude

La

titu

de

(d

eci

ma

l de

gre

es)

30

35

40

45

50

Assemblage

A B C D E F G H

12

13

Overview






14

Six Candidate Indices

Acronym Name

IBI Index of Biotic Integrity

RBI Relative Benthic Index

BRI* Benthic Response Index

RIVPACSRiver Invertebrate Prediction and Classification System

BQI Benthic Quality Index

*: Two variations

15

Candidate IndicesComponents

Candidate Index Data

IBI Community measures

RBI Community measures

BRI-TC Species abundances

BRI-MNDF Species abundances

RIVPACS Presence/absence of multiple species

BQISpecies abundances & community measures

16

Index Development Teams

Candidate Index

Index Leader Reference

IBI Bruce ThompsonThompson and Lowe (2004)

RBI Jim Oakden Hunt et al. (2001)

BRI* Bob Smith Smith et al. (2001, 2003)

RIVPACS David Huff Wright et al. 1993

BQI Bob Smith Rosenberg et al. (2004)

*: Two variations

17

Common Definitions

• A common set of definitions were established– For “Good” and “Bad” sites

• Used in two ways– Identify data to be withheld from index development

• Subsequently used to validate index

• Goal: A set of clearly affected or reference sites to evaluate index performance

– “A Gold Standard”

– Identify reference and degraded condition for index calibration

18

Common Criteria

“Good” (Reference) Sites• Meet all the following criteria:

– Far from known point sources– Data available for sediment chemistry and at least one

amphipod toxicity test– No ERM* exceedences– No more than 3 ERL* exceedences– No toxicity

• Amphipod survival > 83%

– Species abundance list does not indicate bad biology (In progress)

*: As, Cd, Cu, Pb, Hg, Ag, Zn, Hmw(8) & Lmw(11) PAH, Total PCB

19

Common Criteria

“Bad” (Degraded) Sites

• Meet both of the following criteria– 1 or more ERM exceedences, or

3 or more ERL exceedences, and– >50% mortality in an acute amphipod test

20

National vs. CA data

SouthNorth

21

Data For Benthic Index Development

Habitat# Samples

Good Bad

C Euhaline California Bays 85 17

D Polyhaline San Francisco Bay 18 12

E Estuaries and Wetlands 102 3

F Very Coarse Sediments 56 0

G Mesohaline San Francisco Bay 20 4

H Tidal Freshwater 65 0

22

Data For Benthic Index DevelopmentNumbers of samples

HabitatCalibration Validation

G B G B

C Euhaline California Bays 75 9 10 8

D Polyhaline San Francisco Bay 9 6 11 6

23

The Calibration Process

• Identify habitats with sufficient data– “Good” and “Bad” sites– For index calibration and validation

• Distribute calibration data– Teams calibrate candidate indices

• Distribute independent data for validation– Teams apply candidates to data

• Results compiled for evaluation

24

Overview






25

Index Validation Approaches• Classification accuracy

– Chemistry and toxicity– Biologist best professional judgment

• Repeatability– Same day– Same site on different days

• Independence from natural gradients• Correlations with other information

– Species richness– Other indices

26

Overall Classification AccuracyValidation Data (%)

Index Overall(n=35)

RIVPACS 83

BRI-TC 77

IBI 70

BRI-MNDF 63

BQI 63

RBI 51

27

Habitat Classification Accuracy Validation Data (%)

IndexSouthern California

(n=18)

San Francisco Bay(n=17)

RIVPACS 72 94

BRI-TC 72 82

IBI 67 73

BRI-MNDF 56 71

BQI 50 76

RBI 22 82

28

Status Classification Accuracy Validation Data (%)

Index“Good”

Sites(n=21)

“Bad”

Sites(n=14)

RIVPACS 86 79

BRI-TC 81 71

IBI 100 29

BRI-MNDF 67 57

BQI 81 36

RBI 52 50

29

Potential Reasons for Low Classification Accuracy

• Do threshold and scaling problems exist?– Does an index correlate well with condition,

but an incorrect threshold lead to the wrong interpretation?

• Are chemistry-toxicity “bad” definitions inadequate?– Chemistry criteria were less stringent than

many other benthic index efforts

30

RIVPACS vs Amphipod Mortality - San Francisco BayA

mp

hip

od

Mo

rta

lity

(%)

-20

0

20

40

60

80

100

RIVPACS Score

-1.5 -1.0 -0.5 0.0

31

RBI vs Amphipod Mortality - San Francisco BayA

mp

hip

od

Mo

rta

lity

(%)

-20

0

20

40

60

80

100

RBI Score

-1.0 -0.8 -0.6 -0.4 -0.2 0.0

32

BQI vs Amphipod Mortality - Southern CaliforniaA

mp

hip

od

Mo

rta

lity

(%)

-20

0

20

40

60

80

100

BQI Score

-20 -15 -10 -5 0

33

Are Validation Sites Misclassified?

• Is our “Gold Standard” correct?– Are multiple indices disagreeing?– How do index disagreements relate to biology?

• Samples with multiple disagreements evaluated– Using biologist best professional judgment

34

Disagreements with Status Designations

Number of Candidates Disagreeing

N(Σ=35)

0 8

1 9

2 5

3 6

4 4

5 2

6 1

35

Biology Comparison

• For six of seven samples– Biologists agreed that the chemistry-toxicity

status was incorrect• All four biologists agreed for four samples

• 75% agreement for other two

• “Gold Standard” is tarnished

36

Effect of Status Changeon Overall Classification Accuracy

Index Original After Change

RIVPACS 83 83

BRI-TC 77 89

IBI 70 76

BRI-MNDF 63 74

BQI 63 80

RBI 51 63

37

Overview






38

Complete the Index Validation Process

• Classification accuracy– Chemistry and toxicity– Biologist best professional judgment

• Repeatability– Same day– Same site on different days

• Independence from natural gradients• Correlations with other information

– Species richness– Other indices

39

Biology Classification

• Panel of six external experts– Evaluate 20-25 samples– Samples where 5 of 6 experts agree will

establish a new “Gold Standard”• To be used in the same way as the chemistry-

toxicity classification

40

Repeatability

• Identify sites where– Multiple samples were collected on the same

visit– Multiple visits to the same site

• Evaluate candidate index stability

41

Summary

• We will be able to develop benthic indices for two habitats– Some indices validating well

• Validation rates with sediment toxicity and chemistry data are low– Need to re-visit our scaling methods for some indices– Need to establishing biology-based good and bad criteria

• Best professional judgment of an independent panel of experts

• Have more validation steps to complete before making final selections

1 Sediment Quality Objectives for California Enclosed Bays and Estuaries Benthic Indicator Development Scientific Steering Committee 26 th July 2005.

Documents