vvEPA
United States
Environmental Protection
Agency ,
Office of Research and
Development
Washington DC 20460
EPA/620/R-99/QQ4
October 1999
-
r--- •''-" • • --' •
ffPlfe^^itTS54
$*X;vT'I'.""! HiTTS^^fe^s^^r^
jv> [rfj ..'.-*.: -i.*rt,4i'rj^c-.j5i,vrHfi .1^- -
, .„-.
'''5'
EMAP-Virginian
Province Four-Year
Assessment
(1990-93)
Environmental Monitoring and
Assessment Program
-------
-------
EPA/620/R-99/004
October 1999
EMAP-Virginian Province
Four-Year Assessment (1990-93)
John F. Paul
Atlantic Ecology Division
National Health and Environmental Effects Research Laboratory
U.S. Environmental Protection Agency
Narragansett, Rl 02882
John H. Gentile
Center for Marine and Environmental Analyses
Rosenstiel School of Marine and Atmospheric Science
University of Miami
Miami, FL 33149
K. John Scott
Science Applications International Corporation
Newport, Rl 02840
Steven C. Schimmel, Daniel E. Campbell,
and Richard W. Latimer
Atlantic Ecology Division
National Health and Environmental Effects Research Laboratory
U.S. Environmental Protection Agency
Narragansett, Rl 02882
United States Environmental Protection Agency
National Health and Environmental Effects Research Laboratory
Atlantic Ecology Division
27 Tarzwell Drive
Narragansett, Rl 02882
Printed on Recycled Paper
-------
NOTICE
The information in this document had been funded wholly or in part by the United
States Environmental Protection Agency (Environmental Monitoring and Assessment
Program, Office of Research and Development) through the Atlantic Ecology Division.
Support for SAIC was provided by EPA Contract #68-01-0005, M.P. Gant, Project
Officer. It has been subjected to the Agency's peer and administrative review, and has
received approval for publication as an EPA document. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
This is contribution no. 1837 of the Atlantic Ecology Division, National Health and
Environmental Effects Research Laboratory.
The suggested citation for this report is: Paul, J.F., J.H. Gentile, K.J. Scott, S.C.
Schimmel, D.E. Campbell, and R.W. Latimer. 1999. EMAP-Virginian Province Four-
Year Assessment Report (1990-93). EPA 600/R-99/004. U.S. Environmental Protection
Agency, Atlantic Ecology Division, Narragansett, Rhode Island.
ABSTRACT
Assessments with the four years (190-93) of ecological condition data collected
by the USEPA Environmental Monitoring and Assessment Program (EMAP) in
estuaries of the Virginian Biogeographic Province (Cape Henry to Cape Cod) were
presented. EMAP data were used to quantify, with confidence, the condition of
ecological resources within the broad-expanse of estuarine waters comprising the
Virginian Province, as well as its large and small estuarine systems and five major tidal
rivers. Over the four-year period, sufficient sampling sites were available to
characterize the condition of ecological resources for four major watershed systems
within the province (Chesapeake Bay, Delaware Bay, Hudson-Raritan system, and
Long Island Sound), and three tidal rivers in Chesapeake Bay (Potomac,
Rappahannock, and James Rivers). Results clearly showed that the EMAP objectives
were not only reasonable but were achievable with available indicators collected with a
probability-based sampling design. It was shown that the EMAP design can be used to
quantify with confidence the condition of ecological resources. Reducing the
uncertainties in the assessment should be approached through a systematic program of
directed research.
Key Words: EMAP; Environmental Monitoring; Virginian Province; estuaries; Indicators
(biology); Estuarine Assessment; Chesapeake. Bay; Delaware Bay; Long
Island Sound
-------
EXECUTIVE SUMMARY
Introduction and Purpose
The scientific community and the public have become increasingly concerned
with the apparent widespread extent of environmental impacts from anthropogenic
pollutants. Global climate change, acidic deposition, ozone depletion, non-point source
pollution, and habitat alteration threaten our ecological systems at regional, national,
and global scales. Despite these concerns, the status of the nation's ecological
resources has not been well documented, making it difficult to establish quantitatively
whether environmental policies and programs designed to limit anthropogenic impacts
on natural ecosystems are effective.
In 1988, the U.S. Environmental Protection Agency's Science Advisory Board
(SAB, 1988) recommended the implementation of a long-term (10-12 years) program to
monitor the status and trends of the nation's ecological resources to identify emerging
environmental problems before they reach crisis proportions. The Environmental
Monitoring and Assessment Program (EMAP) is the Agency's response to the Science
Advisory Board's recommendation.
The Environment Monitoring and Assessment Program is a long-term,
nationwide program initiated by EPA's Office of Research and Development (ORD).
EMAP's goals are: (1) estimate the current status, trends, and changes in selected
indicators of the condition of the nation's ecological resources on a regional basis with
known confidence; (2) estimate the geographic coverage and extent of the nation's
ecological resources with known confidence; (3) seek associations among selected
indicators of natural and anthropogenic stress and indicators of ecological condition;
and (4) provide annual statistical summaries and periodic assessments of the nation's
ecological resources.
The implementation of EMAP will provide answers to several environmental
questions and in so doing provide the information required to formulate environmental
protection policies for the 1990s and beyond. For example,
• What are the status, areal extent, and geographical distribution of the nation's
ecological resources? How much confidence do we have in such estimates?
• What proportion of these resources is declining or improving? Where? At what
rate?
• What factors are contributing to declining condition of ecological resources, that
is, can knowledge of exposure and habitat indicators be used to explain
observed ecological condition?
-------
• Are pollution control, reduction, mitigation, and prevention programs achieving
overall improvement in ecological condition?
To answer these questions EMAP developed a flexible probability-based sampling
design that can be scaled to the problem setting, has the power to determine status
with known confidence, and is capable of detecting temporal and spatial patterns and
trends. In addition, EMAP employed a suite of biotic and abiotic condition indicators to
characterize the ecological resources, principal stressors, and habitat conditions.
The EMAP ecological condition data presented in this report were collected over a
four-year period (1990-93) from the estuarine waters of the Virginian Biogeographic
Province (Cape Henry to Cape Cod). The results clearly demonstrate that the EMAP
data can be used to quantify, with confidence, the condition of ecological resources
within the broad-expanse of estuarine waters comprising the Virginian Province, as well
as its large and small estuarine systems and five major tidal rivers. Analyses of
individual year data were used to estimate inter-annual variability. Although individual
year data could be used to estimate ecological condition across the province, it results
in larger uncertainties because of the smaller sample size. Over the four-year period,
sufficient sampling sites were available to characterize the condition of ecological
resources for four major watershed systems within the province (Chesapeake Bay,
Delaware Bay, Hudson-Raritan system, and Long Island Sound), and three tidal rivers
in Chesapeake Bay (Potomac, Rappahannock, and James Rivers). The uncertainties of
single-year estimates for these systems were not always acceptable due to the reduced
number of sampling sites, particularly in systems like the Hudson-Raritan and the
individual tidal rivers.
Areal and Spatial Patterns of Indicators
The benthic communities in the Virginian Province were estimated to be impacted,
over the four-year period, in 25+3% of the estuarine area, with annual estimates
ranging from a low of 23+7% in 1991 to a high of 28+7% in 1990. Analysis of the three
resource classes indicates that the large estuarine systems had the smallest percent
area but largest absolute area of impacted benthos, 19+4% (3099 km2), with greater
percent areal impacts in the small estuarine systems and tidal rivers, 37+6% and
38+14%, respectively. Chesapeake Bay, which exhibited benthic impacts of 23+5%,
accounted for 45% of the impacted benthic area within the Virginian Province. The
Potomac and Rappahannock Rivers have identical percent impacted areas, 44+22%
and 44+33%, respectively. The James River, however, exhibited a smaller percent
area of benthic impact (19+23%). Delaware Bay and Long Island sound exhibited
benthic impacts over 24+12% and 28+11% of their areas, respectively. The percent
area of benthic impact in the Hudson-Raritan system, however, was the greatest of the
major estuarine systems examined (72+8%). Together, the four major estuarine
systems account for 79% of the impacted benthic area within the Virginian Province,
iv
-------
The percent area of bottom waters in the Virginian Province, in toto, with low
dissolved oxygen conditions (DO < 2 ppm) was 5+2%. Small estuarine systems had
only 1+1%, large systems had 5+2%, and tidal rivers 14+6%. These data suggest that
the tidal rivers are most at risk from low dissolved oxygen. Most of the low dissolved
oxygen is focused in the main stem of Chesapeake Bay and the mouths of the Potomac
and Rappahannock Rivers. Using moderate to severe hypoxia as the criterion (DO < 5
ppm), impacted conditions were observed in 24+3% of the province area, 17±5% of the
small systems area, 27+4% of the large systems, and 18+7% of the tidal rivers. These
analyses suggest the large estuarine systems are potentially at risk from moderate
reductions in dissolved oxygen. The major estuarine system analyses showed that
approximately 31% of Chesapeake Bay and 48% of Long Island Sound areas have
moderate to severe hypoxia.
The percent area of the Virginian Province sediments having moderate to severe
sediment toxicity (survival < 80%) was 9+2%, Examination of individual resource
classes showed that the percent area exhibiting sediment toxicity was 4+4% of the tidal
rivers, 9+3% of the large systems, and 12+6% of the small systems. These analyses
indicate that small estuarine systems are at greatest risk from toxic sediments.
Moderate to severe sediment toxicity was observed in 15+14% of the Hudson-Raritan
system, 13+7% of Long Island Sound, 6±3% of Chesapeake Bay, and 2+2% of
Delaware Bay. Severe toxicity (survival < 60%) occurred in only 1% of the estuarine
sediments of the province and was distributed across resource classes. Severe toxicity
was observed in Delaware Bay (1+2%), Hudson-Raritan system (7+10%), and Long
Island Sound (5+4%).
Sediment contamination condition in the Virginian Province was determined using
multiple thresholds for comparison with observed concentration of contaminants and
distribution of effects. Anthropogenic enrichment of metals was examined by
determining the metals enrichment above crustal levels. Metals enrichment was
observed in 49+4% of the province area. Small systems and tidal rivers had a greater
enrichment than large systems, 64+7% and 69±10% compared with 42+5%,
respectively. Metals enrichment ranged from 85±11% and 86±8% in the Hudson-
Raritan system and Long Island Sound, respectively, to 44+5% and 39+14% in
Chesapeake and Delaware Bays, respectively.
Sediment contaminant condition for potential biological effects was determined by
using exceedence of Effects Range-Median (ER-M) and Effects Range-Low (ER-L)
values (Long et a/., 1995). Exceedence of an established ER-M for selected trace
metals and organic chemicals has been postulated as one method of ranking the
contamination of marine sediments. The ER-M exceedence for metals was the greatest
in small estuarine systems, 9+4% of the area. The Hudson-Raritan exhibited the
greatest percent area with ER-M metals exceedence (27+9%), whereas the other major
systems exhibited 1-5% areal exceedence.
-------
Exceedence of at least one ER-M value for organics was found in 4+1 % of the area
of the Virginian Province. The percent area was relatively low in large and small
estuarine systems (2%) and higher in tidal rivers (14+2%). The Delaware Bay system
was the least impacted of the four major systems, less than 1 % of its area contained
sediments exceeding organic ER-M values. Values for Chesapeake Bay and Long
Island Sound were 2-4%, while the Hudson-Raritan system exhibited 44+14% of the
area containing sediments with organic concentrations that exceeded at least one ER-
M value. This is consistent with the documented PCB contamination in the Hudson
River.
Approximately half the entire province (50±4%) had sediment contaminant
concentrations below ER-L levels. Small systems and tidal rivers exhibited lower areal
extent of sediments below ER-Ls than large systems, 35+6% and 25+12% vs. 58±5%.
Only 1+3% of the sediments in the Hudson-Raritan system were observed to have
contaminant concentrations below ER-L levels. In other words, almost the entire
Hudson-Raritan system has sediments above levels observed to potentially elicit
biological responses. In contrast, half or more of Chesapeake Bay (50+5%) and
Delaware Bay (62+14%) were observed to have sediment concentration levels below
ER-L values. In Long Island Sound, 24+12% of the sediments were below ER-L levels.
Overall, 52+5% of the estuarine waters of the Virginian Province were in good
condition, i.e., these waters exhibited unimpacted benthic conditions and bottom
dissolved oxygen > 5 ppm and sediment toxicity acute survival > 80% and no ER-M
exceedence for sediment contaminants. Small estuarine systems had 55±14% of the
area in good condition, tidal rivers had 52+13%, and large systems 51+5%.
Indicator Associations
Analyses of associations were conducted between ecological condition indicator
and both stressor and habitat indicators to provide possible explanations for the
observed condition of ecological resources. Analysis of benthic communities and
habitat indicators show that impacted benthic communities tend to be associated with
muddy (> 80% silt-clay), moderate TOC content (1-3%) sediments, and polyhaline
bottom waters (> 18 o/oo). In contrast, unimpacted benthic communities tend to
associate with sandy (< 20% silt-clay), low TOC content (< 1%) sediments, and with
polyhaline bottom waters (> 18 o/oo).
Association analysis of benthic communities with stressors for the entire province
indicates that moderate to severe hypoxia (DO < 5 ppm), sediment toxicity (survival <
80%), and sediment contamination (ER-M exceedence) together account for 54% of the
impacted benthic area. The remaining 46% of impacted benthos is not associated with
any of the three stressor indicators. Low dissolved oxygen is the principal stressor of
concern in large systems and tidal rivers, being associated with 56% and 45% of the
vi
-------
impacted benthos, respectively. Moderate to severe hypoxia occurs primarily in benthic
impacted area of western Long Island Sound, while severe hypoxia (DO < 2 ppm)
occurs primarily in the main stem of Chesapeake Bay. The analysis suggests that
resources in the large estuarine systems of the Virginian Province, which have the
capacity to stratify and receive large inputs of nutrients, are, on the whole, at greater
risk from low dissolved oxygen (e.g., eutrophication) than from sediment contamination.
In small estuarine systems, sediment toxicity and ER-M exceedence are the principal
stressors of concern, accounting for approximately 26% of the observed impacted
benthos. This analysis suggests that the risk to benthic resources in small estuarine
systems is from sediment contamination and not from low dissolved oxygen conditions.
However, there is a large portion unexplained by the three measured stressors.
Conclusions
The four-year assessment of the EMAP-Virginian Province Demonstration Project
illustrates several important contributions to environmental monitoring in the areas of
sampling design and indicator research. In particular, the sampling design is both
systematic and probabilistic in nature; is extremely flexible; provides areal estimates,
with confidence, of the condition of all indicators; is spatially explicit allowing for a
variety of spatial analyses; accommodates post-stratification of data; can be scaled to
the problem setting without losing its defining characteristics; and, most important, has
comparability, which permits the direct comparison of results across differing spatial
scales and between different categories/populations of resources (e.g., large and small
systems and tidal rivers).
The four-year assessment of Virginian Province data affords the opportunity to
evaluate the applicability, sufficiency, and effectiveness of EMAP's indicator program.
The multi-indicator approach used by EMAP has proven both practical and necessary.
Traditional monitoring programs often measure only one type of indicator, either
exposure or effect. EMAP, however, by including the patterns of both natural and
anthropogenic stressors and habitat factors, provides information critical for forming
hypotheses that explain the observed ecological observations. There are limitations,
however, to the indicator program in the Virginian Province as illustrated by the
uncertainties in the analyses. First, only one ecological indicator, benthic community
condition, was successfully developed to assess the status of ecological condition in
the Province. Second, there was no development of stressor indicators for enrichment
and physical perturbation. This assessment emphasizes the need for additional
research on ecological and stressor indicators to reduce uncertainties in the
assessments.
The value of the EMAP design and indicator program is illustrated by its ability to
identify successfully and quantify the major environmental problems in the estuarine
waters of the Virginian Province. When the EMAP conclusions are compared with
vii
-------
analyses using other environmental data from the states and federal agencies, the
general conclusions are the same. The agreement between conclusions drawn from
EMAP and those from existing data could be viewed as an initial validation of the EMAP
concept. This is important because it provides evidence that the EMAP design and
indicators can capture the major ecological problems successfully when applied to data
poor environmental areas. The EMAP design supplements many other studies
because it allows the quantification of the degree of uncertainty (confidence limits) of
the results, and provides for quantitative comparisons across systems.
Although uncertainties remain, the results of the four-year Virginian Province
assessment are encouraging. The Demonstration Project clearly showed that the
EMAP objectives were not only reasonable but were achievable with available
indicators collected with a probability-based sampling design. It was shown that the
EMAP design can be used to quantify with confidence the status and condition of
ecological resources. Reducing the uncertainties in the assessment should be
approached through a systematic program of directed research.
VIII
-------
TABLE OF CONTENTS
NOTICE ii
EXECUTIVE SUMMARY iii
TABLE OF CONTENTS . ix
LIST OF FIGURES xi
LIST OF TABLES xiii
ACKNOWLEDGMENTS xvii
SECTION 1: INTRODUCTION 1
1.1 Overview of EMAP . . 2
1.2 Overview of Virginian Province 4
1.3 Scope of Report 5
SECTION 2: EMAP APPROACH 7
2.1 Resource Classification . . 7
2.2 Sampling Design 8
2.3 Ecological Indicators 11
SECTION 3: INDICATOR STATUS AND ASSESSMENT 13
3.1 Benthic Community Condition Indicator 14
3.1.1 Benthic Condition 16
3.1.2 Benthic Condition: Major Estuarine Systems 19
3.1.2.1 Chesapeake Bay 19
3.1.2.2 Delaware Bay 22
3.1.2.3 Hudson-Raritan . 22
3.1.2.4 Long Island Sound 22
3.2 Dissolved Oxygen Condition Indicator 22
3.2.1 Dissolved Oxygen Condition 23
3.3 Sediment Toxicity Condition Indicator 26
3.3.1 Sediment Toxicity Condition 30
3.4 Sediment Contamination Condition Indicators 38
3.4.1 Sediment Contamination Condition: Areal Patterns 40
3.4.1.1 Metals enrichment . 40
3.4.1.3 Organic ER-M exceedence 44
3.4.1.4 Organic carbon normalized total PAHs 44
3.4.1.5 Organic carbon normalized total DDT ; 47
3.4.1.6 Organic sediment quality criteria 47
ix
-------
3.4.1.7 ER-L exceedence 47
3.4.2. Sediment Contamination Condition: Estuarine System Analysis 47
3.4.2.1 Metals enrichment 47
3.4.2.2 Metal ER-M exceedence 48
3.4.2.3 Organic ER-M exceedence 52
3.4.2.4 Organic carbon normalized total PAHs 52
3.4.2.5 ER-L exceedence 56
3.5 Indicator Co-occurrence 56
3.5.1 Statistical Patterns for Benthic Impact and Habitat Condition 56
3.5.1.2 Total organic carbon content of sediments 58
3.5.1.3 Bottom water salinity 59
3.5.1.4 Summary for habitat associations 61
3.5.2 Statistical Pattern for Benthic Impact and Stressors 61
3.5.2.1 Association with low dissolved oxygen 61
3.5.2.2 Association with sediment toxicity 62
3.5.2.3 Association with ER-M exceedence 63
3.5.2.4 Association with multiple stressors 64
3.5.2.5 Regression of the benthic index with individual variables 66
SECTION 4: DISCUSSION 68
4.1 Is there a problem? 68
4.2 What is the magnitude, extent, and distribution of the problem? 70
4.3 What factors are associated with the observed problems? 71
4.4 Are the observed problems consistent with existing knowledge? 72
4.5 What are the uncertainties? 77
4.6 How effective was the Four-Year EMAP Virginian Province Demonstration
Project in meeting the program objectives? 81
4.7 Summary and Conclusions ,82
SECTION 5: REFERENCES 85
APPENDIX A: OVERVIEW OF SAMPLING AND ANALYTICAL METHODS 97
APPENDIX B: REFINEMENTS TO BENTHIC INDEX 100
-------
LIST OF FIGURES
Figure 1-1. Biogeographic provinces used by EMAP to delineate coastal resources of
the United States 5
Figure 2.1. Distribution of probability-based sampling sites in the Virginian
Biogeographic Province during the period 1990-1993 10
Figure 3-1. Benthic impact for Virginian Province and resource classes across the
years 1990-93 17
Figure 3-2. Condition of benthic communities for Virginian Province for the period
1990-93 18
Figure 3.3. Benthic impact for major estuarine systems of Virginian Province for the
period 1990-93 : . . . 19
Figure 3-4. Condition of benthic communities for major estuarine systems in Virginian
Province for the period 1990-93 21
Figure 3-5. Dissolved oxygen condition of bottom waters in Virginian Province for the
period 1990-93 25
Figure 3-6. Dissolved oxygen condition of bottom waters for major estuarine systems
in Virginian Province for the period 1990-93 29
Figure 3-7. Toxicity condition of bottom sediments, as determined from acute
amphipod bioassays, in Virginian Province for the period 1990-93 ... 32
Figure 3-8. Toxicity condition of bottom sediments, as determined from acute
amphipod bioassays, for major estuarine systems in Virginian Province for
the period 1990-93, amphipod survival < 80% 33
Figure 3-9.
Figure 3-10.
Figure 3-11.
Figure 3-12. Sediment contaminant condition of bottom sediments for major estuarine
Relationships between benthic organism abundance and sediment toxicity
bioassay . 37
Sediment contaminant condition of bottom sediments in Virginian
Province for the period 1990-93, ER-M metals exceedence 43
Sediment contaminant condition of bottom sediments in Virginian
Province for the period 1990-93, ER-M organics exceedence 46
XI
-------
systems in Virginian Province for the period 1990-93, ER-M metals
exceedence 51
Figure 3-13. Sediment contaminant condition of bottom sediments for major estuarine
systems in Virginian Province for the period 1990-93, ER-M organics
exceedence 54
Figure 4-1. Benthic community condition in Chesapeake Bay as reported by
Chesapeake Bay Program (from USEPA, 1995a) 74
Figure 4-2. Sediment contamination and risk to aquatic life in Chesapeake Bay as
reported by Chesapeake Bay Program (from USEPA, 1995a) 75
Figure 4-3. Bottom dissolved oxygen in Chesapeake Bay as reported by Chesapeake
Bay Program (from USEPA, 1995a) 76
Figure B-2. The 1990-91 benthic index versus bottom water salinity for 1990-91
calibration data set (Schimmel et a/., 1994) 102
Figure B-1. The 1990-91 benthic index versus bottom water salinity for 1990-93
EMAP Virginian Province data set 102
Figure B-4. The 1990-93 benthic index versus sediment silt-clay content for 1990-93
calibration data set 116
Figure B-3. The 1990-93 benthic index versus bottom water salinity for 1990-93
calibration data set 116
Figure B-5. The 1990-93 benthic index versus bottom water salinity for 1990-93
validation data set 117
Figure B-6. The 1990-93 benthic index versus sediment silt-clay content 1990-93
validation data set 117
Figure B-7. The 1990-91 benthic index versus bottom salinity for combined four-year
calibration and validation data sets : 118
XII
-------
LIST OF TABLES
2-1. Summary of the Total Areas (km2) for Resource Classes and Major Estuarine
Systems in Virginian Province 8
3-1. Areal Estimates for Impacted Benthic Communities, as Determined by EMAP-
Virginian Province Benthic Index, for the Virginian Province 16
3-2. Areal Estimates for Impacted Benthic Communities, as Determined by EMAP-
Virginian Province Benthic Index, for Major Estuarine Systems in the Virginian
Province 20
3-3. Areal Estimates for Dissolved Oxygen Condition of Bottom Waters in the
Virginian Province 24
3-4. Areal Estimates for Dissolved Oxygen Condition (DO < 2 ppm) of Bottom Waters
for Major Estuarine Systems in the Virginian Province 27
3-5. Areal Estimates for Dissolved Oxygen Condition (DO < 5 ppm) of Bottom Waters
for Major Estuarine Systems in the Virginian Province 28
3-6. Areal Estimates for Bottom Sediment Toxicity Condition, as Determined from
Acute Amphipod Bioassay, in the Virginian Province 31
3-7. Areal Estimates for Bottom Sediment Toxicity Condition (Survival < 80%) for
Major Estuarine Systems in the Virginian Province 34
3-8. Areal Estimates for Bottom Sediment Toxicity Condition (Survival < 60%) for
Major Estuarine Systems in the Virginian Province 35
3-9. Chemical Measurements Conducted for Sediments of the Virginian Province
39
3-10. Areal Estimates for Enriched Metal Concentrations in Bottom Sediments in the
Virginian Province 41
3-11. Areal Estimates for Bottom Sediment Contaminant Condition for Metals in the
Virginian Province . -.....- 42
3-12. Areal Estimates for Bottom Sediment Contaminant Condition for Organics in the
Virginian Province 45
3-13. Areal Estimates for Bottom Sediment Contaminant Condition (No ER-L
XIII
-------
Exceedence) in the Virginian Province 48
3-14. Areal Estimates for Enriched Metal Concentrations in Bottom Sediments for the
Major Estuarine Systems in the Virginian Province 49
3-15. Areal Estimates for Bottom Sediment Contaminant Condition (Any ER-M Metal
Exceedence) for Major Estuarine Systems in the Virginian Province 50
3-16. Areal Estimates for Bottom Sediment Contaminant Condition (Any ER-M
Organics Exceedence) for Major Estuarine Systems in the Virginian Province
53
3-17. Areal Estimates for Sediment Contamination Condition (Organic Carbon-
Normalized PAHs > 200 Mg/g-OC) for Major Estuarine Systems in Virginian
Province 55
3-18. Areal Estimates for Bottom Sediment Contaminant Condition (No ER-L
Exceedence) for Major Estuarine Systems in the Virginian Province 57
3-19. Areal Estimates for Association of Silt-clay Content of Sediments with Benthic
Condition for the Virginian Province 58
3-20. Areal Estimates for Association of Total Organic Carbon (TOG) Content of
Sediments with Benthic Condition for the Virginian Province 59
3-21. Areal Estimates for Association of Bottom Water Salinity with Benthic Condition
for the Virginian Province '. 60
3-22. Association of Benthic Condition with Enriched Metal Concentrations in
Sediments for the Virginian Province 61
3-23. Association of Benthic Condition with Bottom Dissolved Oxygen for the Virginian
Province 62
3-24. Areal Estimates for Association of Benthic Condition with Sediment Toxicity for
the Virginian Province 63
3-25. Areal Estimates for Association of Benthic Condition with ER-M Exceedence in
Sediments for the Virginian Province 64
3-26. Co-occurrence of Stressors with Province-wide Impacted Benthic Communities
(percent of impacted area) 65
XIV
-------
3-27. Co-occurrence of Stressors with Resource Class Impacted Benthic Areas
(percent of impacted area). Dissolved Oxygen Criterion of 5 ppm 66
3-28. Co-occurrance of Stressors with Resource Class Impacted Benthic Areas
(percent of impacted area). Dissolved Oxygen Criterion of 2 ppm 66
3-29. Results of Stepwise Regression for Benthic Index Against Habitat and Stressor
Indicators 67
4-1. Expansion of Table 3-2 to Include Estuarine Classes for Impacted Benthic
Communities tp Illustrate Effect of Small Sample Size on the Mean and
Uncertainty Estimates 78
4-2. Means and 95% Confidence Intervals for Four-year Virginian Province Estuarine
Condition Estimates for Entire Province, Estuarine Classes, and Major Estuarine
Systems . . 83
B-1. Candidate Benthic Measures Used to Formulate the Benthic Index 104
B-2. Summary of Correlations Between Habitat Indicators and the Candidate Benthic
Measures for the Entire 1990-93 Data Set, Using Pearson Correlation
Coefficients (significance for p < 0.05) 105
B-3. Regression Coefficients and Correlation Coefficients (r2) for Bottom Water
Salinity Normalization Functions for Benthic Measures 106
B-4. Impacted Sites from 1990-93 Used to Calibrate the Benthic Index. Information for
Each Site Includes Estuarine System, Date of Sampling, EMAP Station Number,
and Geographic Coordinates 108
B-5. Unimpacted Sites from 1990-93 Used to Calibrate the Benthic Index. Information
for Each Site Includes Estuarine System, Date of Sampling, EMAP Station
Number, and Geographic Coordinates 109
B-6. Impacted Sites from 1990-93 Used to Validate the Benthic Index. Information for
Each Site Includes Estuarine System, Date of Sampling, EMAP Station Number,
and Geographic Coordinates . . . 111
B-7. Unimpacted Sites from 1990-93 Used to Validate the Benthic Index. Information
for Each Site Includes Estuarine System, Date of Sampling, EMAP Station
Number, and Geographic Coordinates 112
B-8. Results of Discriminant Analyses Conducted to Combine Candidate Benthic
xv
-------
Measures into an Index of Benthic Condition for the 1990-93 EMAP Virginian
Province Data Set 113
XVI
-------
ACKNOWLEDGMENTS
The results of the program presented in this report could not have been conducted
without the efforts of literally hundreds of individuals, who assisted in design,
development, field sampling, laboratory analyses, total quality management,
administrative support, information management, and data analysis. These are the
individuals, whose dedication and perseverance in conducting the EMAP-Virginian
Province activities, that made our job of producing this assessment report enjoyable.
We would like to acknowledge in particular two individuals, Rick Linthurst and Jay
Messer, whose leadership made EMAP a reality and permitted us the opportunity to
"just do it". Special thanks to Sandi Benyi and Melissa Hughes for assistance with the
data analysis. Thanks to Jerry Pesch, Rich Pruell, Charles Strobel, and Steve Weisberg
for their critical reviews of this report. This is contribution no. 1837 of the Atlantic
Ecology Division, National Health and Environmental Effects Research Laboratory.
EMAP has evolved as a program since the conduct of the Four-Year Virginian
Province Demonstration Project. The information presented in this report concerning
EMAP was current during the conduct of the Demonstration Project. It should not be
implied that this information represents the present status of EMAP.
XVII
-------
-------
SECTION 1: INTRODUCTION
The scientific community and the public have become increasingly concerned that
the impact of pollutants extends far beyond the local scale. Global climate change,
acidic deposition, ozone depletion, non-point source pollution, and habitat alteration
threaten our ecological systems at regional, national, and global scales. Years of
scientific study have convinced us that ecosystem responses to natural and
anthropogenic disturbances are often complex and difficult to characterize, even as
these studies have heightened awareness of these environmental problems. The
status of the nation's ecological resources has not been well documented in the past,
and establishing quantitatively whether environmental policies and programs designed
to limit anthropogenic impacts on natural ecosystems are effective has been difficult.
Despite the implementation of stricter environmental control programs in coastal
regions, the perception of the scientific community and the informed public is that water
and sediment quality and the abundance and quality of living marine resources have
declined in the past 10 to 15 years. The perceived decline in estuarine and coastal
environmental quality has been noted in the popular press and the scientific literature
(Smart et a/., 1987; Morganthau, 1988; Toufexis, 1988). These problems are
exemplified by the following:
1. Increases in the frequency, duration, and extent of water containing insufficient
oxygen to sustain living resources (USEPA, 1984; Officer et a/., 1984; Parker et
a/., 1986; Rabalais et a/., 1985; Whitledge, 1985);
2. Accumulation of contaminants in bottom sediments and in the tissues of fish and
shellfish to levels that threaten human health and the sustainability offish and
shellfish populations (OTA, 1987; NRG, 1989);
3. Increased evidence that many restoration and mitigation efforts have not
replaced losses of critical habitats (Sanders, 1989; The Conservation
Foundation, 1988);
4. Increased incidence of pathological problems in fish and shellfish (Sinderman,
1979; O'Connor et a/., 1987; Buhler and Williams, 1988; Capuzzo etal., 1988);
5. Increased frequency and persistence of algal blooms and associated decreases
in water clarity (USEPA, 1984; Pearl, 1988; Smayda and Villareal, 1989);
6. Increased incidence of closures of beaches, shellfishing grounds, and fisheries
because of pathogenic and chemical contamination (Smart et a/., 1987; FDA,
1971, 1985; Hargis and Haven, 1988; Broutman and Leonard, 1988; Leonard et
a/., 1989); and
1
-------
7. Increased incidence of human health problems from consuming contaminated
fish and shellfish or swimming in contaminated waters (Fein et a/., 1984; Malins,
1989).
In 1988, the U.S. Environmental Protection Agency's Science Advisory Board (SAB,
1988) recommended the implementation of a program to monitor ecological status and
trends that would identify emerging environmental problems before they reach crisis
proportions. The Environmental Monitoring and Assessment Program (EMAP) is the
Agency's response to the Science Advisory Board's recommendation. Coincidentally,
the National Research Council's Marine Board, in a review of marine and estuarine
monitoring systems (NRC, 1990), recommended the creation of a national network of
regional monitoring programs for estuarine and coastal environments. This review
recognized the need for new monitoring programs that build on existing information and
expand the information base landward in order to identify the factors contributing to
coastal pollutant problems.
1.1 Overview of EMAP
The Environment Monitoring and Assessment Program is a nationwide program
initiated by EPA's Office of Research and Development (ORD). EMAP was developed
in response to the demand for information about the degree to which existing pollution
control programs and policies protect the nation's ecological resources. EMAP is an
integrated federal program; ORD is coordinating, planning, and implementing EMAP in
conjunction with other federal agencies, including the Agricultural Research Service, the
Bureau of Land Management, the U.S. Fish and Wildlife Service, the U.S. Forest
Service, the U.S. Geological Survey, and the National Oceanic and Atmospheric
Administration (NOAA), and with other offices within EPA (e.g., Office of Water). These
other agencies and offices participate in the collection and analysis of EMAP data and
will use it to guide their policy decisions, as appropriate.
The goal of EMAP is to contribute to decisions on environmental protection and
management by monitoring and assessing the condition of the nation's ecological
resources. To accomplish this goal, EMAP has worked to accomplish four objectives:
(1) estimate the current status, trends, and changes in selected indicators of the
condition of the nation's ecological resources on a regional basis with known
confidence; (2) estimate the geographic coverage and extent of the nation's ecological
resources with known confidence; (3) seek associations among selected indicators of
natural and anthropogenic stress and indicators of ecological condition; and (4) provide
annual statistical summaries and periodic assessments of the nation's ecological
resources.
EMAP has been designed to provide the information required to formulate
environmental protection policies for the 1990s and beyond by providing answers to the
-------
following questions:
1. What are the status, extent, and geographical distribution of the nation's
ecological resources?
2. What proportion of these resources is declining or improving? Where? At what
rate?
3. What factors are likely to be contributing to declining condition?
4. Are pollution control, reduction, mitigation, and prevention programs achieving
overall improvement in ecological condition?
Assessment of status and trends in the condition of the nation's ecological
resources requires collecting data in a standardized manner, over large geographic
scales, and for long periods of time. Such assessments cannot be accomplished by
aggregating data from the many individual, short-term monitoring programs conducted
in the past or are being conducted currently (NRC, 1990). Differences in the
parameters measured, collection methods, timing of sample collection, and program
objectives limit the value of historical monitoring data and existing monitoring programs
for making integrated regional and national assessments.
EMAP was proposed and developed by EPA/ORD because an integrated
monitoring and assessment program that samples ecological resources in proportion to
their abundance (probability-based) offers considerable advantages over historical
monitoring approaches. These include:
1. EMAP was designed to improve the definition of the extent and magnitude of
pollution problems at regional and national scales.
2. Simultaneous, probability-based sampling of pollution exposure, environmental
condition, and biological resources is important for associating environmental
stressors with impacted ecological condition.
3. EMAP data are geographically referenced; therefore, the distribution and spatial
patterns for impacted ecological conditions can be analyzed.
4. The EMAP sampling design is flexible in that it does not restrict the aggregation
or post-stratification that can be applied to the data. The only restriction is on the
number of sample points that may be available to conduct the aggregation.
5. The temporal revisits to sampling sites permits information that can be used to
capture the direction of ecological change resulting from remediation programs.
-------
In summary, EMAP can provide objective assessments of the severity and extent of
environmental problems and the degree to which impacted resources are responding to
efforts to protect or restore them.
1.2 Overview of Virginian Province
EMAP identified boundaries for seven estuarine regions (Figure 1-1) based upon
biogeographic provinces defined by NOAA and the U.S. Fish and Wildlife Service using
major climatic zones and prevailing ocean currents (Bailey, 1983; Terrell, 1979). The
four years of data collected over 1990-93 in the estuarine waters of the Virginian
Province, which covers approximately 23,573 km2 and includes the wide expanse of
irregular coastline from Cape Cod, Massachusetts, to the mouth of Chesapeake Bay
(Cape Henry, Virginia), were to be used to evaluate the feasibility of regional sampling
and to evaluate and improve the sampling design and indicators prior to the actual
nation-wide implementation of EMAP in estuaries.
The Virginian Province was selected as the testing ground for the EMAP estuarine
monitoring effort because there is a public perception that estuaries in this region of the
country are deteriorating more rapidly than in other regions. Many estuaries in this
province have been investigated intensively by scientists, and a considerable amount of
information was available for use in designing the Virginian Province monitoring
activities. Six EPA National Estuary Programs were in place in the Virginian Province in
1990. In addition, many management decisions were forthcoming, including
development of a restoration plan for the New York Harbor complex, and development
of management plans and evaluation of previous management actions for many large
estuaries, including Delaware Bay, Chesapeake Bay, and Long Island Sound.
Development of such plans presented an opportunity to demonstrate how EMAP
monitoring data can assist in the formulation of environmental programs and policies.
Estuaries were selected as one of the first resources to be sampled by EMAP.
Estuaries are among the most productive of ecological systems. Historically, more than
70% of commercial and recreational landings offish and shellfish were taken from
estuaries (NOAA, 1987). Estuaries also provide feeding, spawning, and nursery
habitats, and are part of migratory routes for many commercially and recreationally
important fish and wildlife, including threatened and endangered species (Lippson ef
a/., 1979; Olsen et a/., 1980). The public values estuarine ecosystems for recreation
(e.g., boating, swimming, hunting, and fishing) and aesthetic appeal. Approximately $7
billion in public funds are spent annually on outdoor marine and estuarine recreation in
the 33 coastal states (NOAA, 1988). Millions of tourists visit coastal beaches annually,
and coastal property is among the nation's most valuable real estate. About half the
nation's population now lives in coastal areas, and by the year 2010, the population in
these areas will have grown by almost 60% to 127 million people (Culliton et a/., 1990).
The estuarine and coastal environment also provides cooling waters for energy
-------
Columbian
Californian
Acadian
Virginian
Carolinian
West Indian
Figure 1-1. Biogeographic provinces used by EMAP to delineate coastal resources of the United
States.
production, transportation routes for ships, and space for economic development. Most
of the nation's major ports are located in estuaries.
Estuaries are complex transition zones between streams, rivers, and coastal
oceans. They have physical features that concentrate and retain pollutants, and they
tend to serve as repositories for the many pollutants released into the atmosphere and
the nation's surface waters. The ecological condition of estuaries is influenced strongly
by human activities in the watershed, particularly land use patterns and the release of
pollutants to the environment. In many coastal regions, water and sediment quality and
the abundance of living resources are perceived to have declined despite the
implementation of pollution control programs.
1.3 Scope of Report
EMAP monitoring activities were conducted in the Virginian Province for the four-
year period, 1990-93, using a probability-based sampling design. Data were collected
and analyzed in a consistent manner. An assessment of results from the first year's
activities was reported in Weisberg etal. (1993). The 1991, 1992, and 1993 monitoring
activities were reported in Schimmel etal. (1994), Strobel etal. (1994), and Strobel et
al. (1995), respectively. Strobel et a/. (1 ,"5) a'so included statistical summaries for all
four years of data. This report presents an assessment of all four years (1990-93) of
-------
EMAP monitoring data collected in the estuarine waters of the Virginian Province. In
particular, this report provides estimates of ecological resources for both the Province
and for the major estuarine systems in the Province, evaluates associations between
ecological condition and other indicators, and attempts an evaluation of the
effectiveness of the program in meeting its objectives. This report is EMAP's first
attempt at providing a multi-year interpretive assessment of regional-scale ecological
conditions. It builds upon the efforts of numerous investigators who have been involved
with EMAP.
The next section (Section 2) provides background on the EMAP approach for
resource class classification, sampling design, and ecological indicators. Section 3
summarizes the assessment for condition indicators and the association between
stressors and ecological condition for both the entire province and the major estuarine
systems in the province. Section 4 provides a discussion of results of the four-year
assessment organized around a series of environmental management questions.
Appendix A provides an overview of the sampling and analytical methods for the
parameters used in this report. Appendix B presents the refinements to the benthic
index that were developed for the four-year assessment.
6
-------
SECTION 2: EMAP APPROACH
There are three distinct elements of EMAP monitoring that were the guiding forces
in designing the sampling plan for the Virginian Province efforts. First, probability-
based sampling sites were selected in an unbiased manner so that resources (e.g.,
estuarine waters in the Virginian Province) were sampled in proportion to their
abundance in a resource size class (discussion in next section). Probability-based
sampling permits estimation of the condition of the portion of the resource that was not
sampled based on knowledge of the sampled portion. Estimates of the proportion of
the total area sampled that is impacted can be made with quantifiable confidence.
Furthermore, the level of confidence in the estimate can be increased by increasing the
number of sites sampled or through the incorporation of existing data.
Second, EMAP focuses on indicators of biological response and uses measures of
exposure from stress or pollution for interpreting biological response data. Traditionally,
estuarine monitoring has focused on measures of stress (e.g., concentration of
contaminants in sediment) and attempted to infer biological impacts based upon
laboratory bioassays. The advantage of the ecologically-based approach emphasized
in EMAP is that it can be applied to situations where multiple stressors exist, acting
separately or in combination, and where natural processes are complex and cannot be
easily described. This is certainly the case in estuarine systems, which are subject to
an array of anthropogenic inputs and exhibit great biotic diversity and complex physical,
chemical, and biological interactions.
Third, EMAP is conducted on regional and national scales using standardized
methods. Entire regions are sampled within a defined time window (index period) to
characterize the resource and to ensure comparability of data within and among
sampling years.
2.1 Resource Classification
For design considerations, EMAP classifies estuaries into three classes (or strata):
large estuarine systems, small estuarine systems, and large tidal rivers (Holland, 1990).
Large estuarine systems are defined as systems having surface areas greater than 260
km2 and aspect ratios (length/average width) less than 18. Applying these criteria to
the Virginian Province results in the identification of twelve large estuarine systems with
a total surface area of 16,097 km2 or 68% of the province's estuarine area (Table 2-1).
Large tidal rivers, defined as systems having surface areas greater than 260 km2 and
aspect ratios greater than 18, include the Hudson, Potomac, James, Delaware, and
Rappahannock Rivers. These five tidal rivers have a total surface area of 2,601 km2 or
11% of the total province area. Small estuarine systems are defined as systems having
surface areas less than 260 km2 but greater than or equal to 2.6 km2. Applying these
criteria to the Virginian Province results in the identification of one hundred forty-four
-------
Table 2-1. Summary of the Total Areas (km2) for Resource Classes arid Major Estuarine
Systems in Virginian Province.
Virginian Province
Chesapeake Bay
(including rivers listed below)
Potomac River
Rappahannock River
James River
Delaware Bay
Hudson-Raritan
Long Island Sound
Large
Systems
16,097
7,044
-
-
-
1,784
-
3,069
Small
Systems
4,875
2,315
127
26
100
30
451
276
Tidal Rivers
2,601
2,048
1,124
345
578
245
309
-
Total
23,574
11,407
1251
371
678
2,059
760
3,345
small estuarine systems with a total surface area of 4,875 km2, or 21% of the province.
2.2 Sampling Design
EMAP uses a probability-based sampling design over time and space to develop a
cost-effective monitoring program (Overton etal., 1991). The statistical approach is
similar in concept to other federal statistical programs or surveys, such as those
conducted by the Census Bureau, Bureau of Labor Statistics, and National Agriculture
Statistics Service. The principal distinction is that these programs focus on producing
estimates of the characteristics of human populations, business establishments, or
agricultural enterprises. In contrast, EMAP focuses on producing estimates of attributes
of ecological resource populations, such as ecological condition of estuarine waters of
the Virginian Province.
EMAP was designed to assess regional populations of ecological resources in the
United States. The design permits estimates of the condition, geographic coverage
(i.e., spatial distribution), and extent for regional populations of ecological resources.
The design permits population estimates to be provided with known statistical
confidence. EMAP intended to make these estimates not only for a specific point in time
(current status) but also repeated over time (trends). The design enables associations
(empirical relationships) to be investigated between condition indicators and stressor
indicators for the ecological resources.
8
-------
The EMAP sampling design provides unbiased estimates of the status and trends in
indicators of ecological condition with a known level of confidence. The value of the
EMAP sampling design is that it is both systematic in areal coverage yet probabilistic
relative to the sampling strategy (Overton et al., 1991). This design, therefore, is
capable of both determining areal extent (with confidence intervals) and the spatial
patterns of ecological indicators irrespective of the characteristics of their statistical
distributions. EMAP proposed to base its status assessments on data collected over a
four-year baseline (Holland, 1990). This multi-year cycle was chosen to dampen the
inter-annual variability resulting from natural phenomena such as extremely dry or wet
years and major climatic disturbances such as hurricanes.
EMAP does not attempt to fully characterize naturally-occurring seasonal variability
or to assess the status of ecological resources for all seasons. An index period (July-
September) was chosen for estuarine sampling to represent that portion of the year
when the measured ecological parameters are expected to show the maximum
response to pollutant stress (Connell and Miller, 1984; Sprague, 1985; Mayer et a/.,
1989); dissolved oxygen concentrations are lowest (Holland et a/., 1988; USEPA, 1984;
Officer et a/., 1984); fauna and flora are most abundant; and within-season variability is
expected to be minimized. A consequence of the index approach is that short-term,
episodic events may not be detected. This approach is consistent with EMAP's goals
of determining the long-term status and trends of ecological resources that can then be
used as the basis for intensive site-specific research to understand the reasons for the
observed deviations form baseline conditions.
Sampling sites in the large estuarine class were selected using a randomly-placed
systematic grid (Holland, 1990; Paul et al., 1992). The distance between the
systematically-spaced sampling points on the grid was approximately 18 km. The grid
is an extension of the systematic grid proposed for use by EMAP (Overton et al., 1991).
The center points of the grids are the sample sites. A linear analog of the systematic
grid was used for site selection in the large tidal rivers (Holland, 1990; Paul ef al.,
1992). A systematic linear grid was used to define the spine of the five large tidal rivers,
where the starting point of the spine was at the mouth of the river. The first transect was
randomly located between river kilometer 0 and 25. Additional transects were then
placed every 25 km up the river to the head of tide. The 144 small estuarine systems
were randomly sampled from the entire list of small systems in the province (Holland,
1990; Paul et al., 1992). They were ordered from north to south by combining adjacent
estuaries into groups of four. One estuary was selected randomly from each group
without replacement for each of the four years of sampling. The location of the sample
within each selected small system was randomly placed.
Application of the sampling design to the three estuarine resource classes resulted
in 425 sampling sites over the four-year period. The distribution of these sites is shown
in Figure 2-1. Additional sampling sites were visited over the four-year period to meet
-------
' Long Island Sound
Washington, D.C
Potomac River
Chesapeake Bay
Figure 2.1. Distribution of probability-based sampling sites in the Virginian Biogeographic Province
during the period 1990-1993.
10
-------
specific research needs, including indicator testing and evaluation, and spatial and
temporal variability estimates. Refer to Holland (1990), Weisberg ef a/. (1993),
Schimmel etal. (1994), and Strobel etal. (1994, 1995) for details on these additional
sampling sites.
2.3 Ecological Indicators
EMAP monitors ecological indicators to assess status, changes, and trends in the
condition of the nation's ecological resources (Hunsaker and Carpenter, 1990).
Indicators are defined as any characteristic of the environment that can provide
quantitative information on the condition of ecological resources, the magnitude of
stress, the exposure of a biological component to stress, or the amount of change in the
condition of the resource.
Ecological condition and response to perturbation are determined by the interaction
of all the physical, chemical, and biological components of the system. Because it is
impossible to measure all these components, EMAP's strategy has been to emphasize
indicators of ecological structure and function that represent the condition of ecological
resources, relative to societal values. EMAP has selected, developed, and evaluated
indicators for the following reasons: (1) to describe the overall condition of ecological
resources, (2) to permit the detection of changes and trends in condition, and (3) to
provide preliminary diagnosis of possible factors that might contribute to unacceptable
conditions caused by human or natural stressors.
EMAP defines two general types of ecological indicators, condition and stressor
indicators. A condition indicator is any characteristic of the environment that provides
quantitative estimates on the state of an ecological resource and is based on something
valued by society. There are two types of condition indicators: biotic and abiotic.
Condition indicators are used to estimate the status, trends, and changes in ecological
condition as well as the extent of ecological resources. EMAP estimates the regional
distribution of quantitative values for each of these indicators within and among
resource categories. All estimates are accompanied by specified levels of confidence
(95% confidence limits for this report) so the user knows the certainty of the estimates.
Condition indicators discussed in this report include the benthic index (a combination of
structural properties of the benthic animal assemblages), dissolved oxygen
concentrations in bottom water, acute toxicity of sediments, and sediment contaminant
concentrations.
Stressor indicators are characteristics of the environment that cause changes in the
condition of an ecological resource. Both natural and human-induced stressors are
examined. Monitoring selected stressor and condition indicators allows EMAP to seek
associations between indicators of stress and observed ecological conditions.
Associations provide insight or possible causality, and lead to the formulation of
11
-------
hypotheses regarding factors (e.g., land use activities, sources, etc.) that might be
contributing to the unacceptable ecological conditions. Associations can provide
direction for regulatory, management, or research programs by suggesting the
possibility of causal relationships. In this report the benthic index is associated with the
following indicators of stress: dissolved oxygen, sediment toxicity, sediment
contaminant concentrations, salinity, sediment grain size, and sediment organic carbon
content. Note that a parameter can function as both a condition indicator and an
indicator of stress, depending on its context, e.g., low dissolved oxygen can be both an
indicator of stress and an indicator of condition.
12
-------
SECTIONS: INDICATOR STATUS AND ASSESSMENT
Areal and spatial analyses of EMAP data were conducted for the province and each
of the three estuarine resource classes at the regional and watershed scales. EMAP's
probability-based sampling design assigns an areal weight to each sampling station.
This design permits the calculation of areal-based cumulative distribution functions
(CDFs) for each of the indicators. The estimation procedures in Heimbuch et al.
(1995a) were used to calculate CDFs and 95% confidence intervals (C.I.) that are the
foundation for the analyses in this report. Each condition indicator is assigned a "critical
value" or threshold that separates impacted from unimpacted conditions. For example,
dissolved oxygen is assigned a critical value of 2.0 ppm based upon numerous
examples in the scientific literature that clearly suggest that dissolved oxygen values
less than 2.0 ppm cause both acute and chronic ecological effects. Estimates of the
relative (percent of total resource class or estuarine system area) and absolute (km2)
areal extent of "impacted condition" were calculated for each condition indicator and
resource category at both province-wide and major estuarine system scales (refer to
next paragraph for specific major estuarine systems). It is important to note that
analyses of both percent area and absolute area may be useful for developing different
management options. An impact is not necessarily due to anthropogenic stress alone.
Natural stress can also contribute to ecological condition. EMAP indicators, as currently
applied, cannot unambiguously discriminate between natural and anthropogenic stress.
While estimates of resource condition on an areal basis are one of the strengths of
the EMAP probabilistic design, EMAP data are also spatially explicit and, therefore,
amenable to a variety of landscape analyses. For example, stations below a critical
value can be spatially displayed for the province or specific estuarine systems. Spatial
representation of impacted indicator values is particularly useful because they can be
linked to land-use patterns that are potentially amenable to control or remediation
strategies. Although EMAP's primary objective is to describe status and trends at the
province level, estimates can also be generated for subpopulations or different
groupings of the estuarine resource. In addition to the three resource classes defined
for the sampling design, we have also aggregated and analyzed the sampling data from
stations in four major estuarine systems in the Virginian Province (Chesapeake Bay,
Delaware Bay, the Hudson-Raritan system [New York/New Jersey harbor area, Hudson
River, and Raritan Bay], and Long Island Sound), along with three major tidal river
systems within Chesapeake Bay (the James, Rappahannock, and Potomac Rivers).
This type of classification integrates multiple land-based sources of anthropogenic
pollutants (point and non-point) into riverine systems that discharge to the estuaries.
These analyses illustrate the flexibility and power of the EMAP design and can be used
to provide an assessment of relative ranking of the potential risks for each estuarine
system. However, there are practical limitations imposed by application of the design
due to small sample sizes for certain systems, as will be discussed in Section 4.5.
13
-------
These approaches, while illustrating the flexibility of the EMAP design, provide a
suite of analysis options that are scale-independent which can be applied to any
problem setting. We have made no presumption which of these approaches is most
useful from an assessment perspective. However, spatial analysis combined with areal
estimates of condition produce information that is comprehensive and consistent with
risk-based principles of assessment and management. In practice, the analysis
approach selected will be dependent upon the specific management question being
addressed.
Results from the analysis of each major indicator collected over the four-year period
of the study are presented and discussed in the remainder of this section. The
presentation for each indicator includes: (1) background on the importance of the
indicator; (2) percent and absolute estuarine area impacted for the Virginian Province
and the three resource classes; (3) spatial representation of the impacted estuarine
areas in the province; (4) percent and absolute area and spatial presentation illustrating
impacted estuarine sampling areas for the four major estuarine systems and the three
tidal river systems; and (5) a summary of the status of the indicator.
3.1 Benthic Community Condition Indicator
The status of biological resources in the Virginian Province was characterized by
evaluating the condition of bottom dwelling (benthic) invertebrate assemblages. The
importance of the role of benthic communities in estuarine ecosystems is well
established (Holland et a/., 1987, 1988; Rhoads etal., 1978; Pearson and Rosenberg,
1978; Sanders et a/., 1980; Boesch and Rosenberg, 1981), and for the purposes of this
report, the condition of benthic assemblages will be the only biological condition
indicator presented. Benthic assemblages were used as an indicator because previous
studies suggested that they are sensitive to pollutant exposure (Pearson and
Rosenberg, 1978; Boesch and Rosenberg, 1981). They also integrate responses to
exposure and disturbance over relatively long periods of time (months to years). Their
sensitivity to pollutant stress is, in part, because benthic organisms live in the sediment,
a medium that accumulates environmental contaminants overtime (Nixon etal., 1986;
Schubel and Carter, 1984). Their relative immobility also restricts benthic organisms
from avoiding pollutant exposure and environmental disturbances.
Three samples were collected at each site using a stainless steel, Young-modified
van Veen grab that samples a surface area of 440 cm2. They were sieved through a 0.5
mm screen and preserved for laboratory analysis. Organisms were identified and
counted. Biomass of key taxa was measured. Other taxa were grouped according to
taxonomic type (e.g., polychaetes, amphipods, decapods) for biomass determination.
Characteristics of benthic assemblages, often expressed as indices, have been
used to measure and describe ecological status and trends of marine and estuarine
14
-------
environments for several decades (Sanders, 1956, 1960; Boesch 1973; Pearson and
Rosenberg, 1978; Holland et a/., 1988). This literature has identified a diverse array of
benthic assemblage attributes that can be used to characterize ecological status and
trends, including: 1) measures of biodiversity/species richness, 2) changes in species
composition, 3) changes in the relative abundance or productivity of functional groups,
4) changes in relative abundance and/or productivity of "key" species, 5) changes in
biomass, and 6) relative size of biota (Weisberg et a/., 1993).
An index, based upon several structural properties of benthic assemblages, was
used to summarize the benthic data and characterize estuarine biological condition in
this report. Discriminant analysis was used to identify a combination of characteristics
of benthic assemblages that distinguishes between reference sites (i.e., clean sites)
and sites with known pollutant exposure. The sites used to develop this index were
distributed across the entire Province and encompass geographic diversity as well as a
variety of habitats. The algorithm for the EMAP Virginian Province 1990-93 benthic
index used for this report is:
1.389 * (salinity normalized Gleason's D based upon infauna and epifauna - 51.5) 728.4
- 0.651 * (salinity normalized tubificid abundance - 28.2) /119.5
- 0.375 * (spionid abundance - 20.0) / 45.4,
where
salinity normalized Gleason's D based upon infauna and epifauna =
Gleason's D / (4.283 - 0.498 * bottom salinity
+ 0.0542 * bottom salinity2
-0.00103* bottom salinity3)* 100,
Gleason's D = S/ln(N),
N = total number of individuals,
S = number of species,
salinity normalized tubificid abundance =
tubificid abundance - 500 * exp(-15 * bottom salinity),
ln(...) denotes natural logarithm,
and
exp(...) denotes the exponential function.
15
-------
This version of the benthic index is a refinement to that presented in Schimmel et al.
(1994) and was developed from data collected in all four years. The index represents
an attempt at reducing a complex set of biological measurements to a simple,
interpretable value. Details of the rationale and development of the benthic index and
the procedure used to distinguish impacted from unimpacted (reference) benthic
condition are presented in Appendix B.
3.1.1 Benthic Condition
The condition of the benthic communities in the Virginian Province, as determined
by the benthic index, is summarized in Table 3-1 and illustrated in Figure 3-1. Impacted
benthic communities comprised 25+3% of the province area for the four-year period,
with annual values ranging from a low of 23+7% in 1991 to a high of 28+7% in 1992.
This means that, for the four-year period, 75+3% of the estuarine area of the Virginian
Province had unimpacted benthic communities. Of the three resource classes, the large
estuaries had the smallest percent area with impacted benthic communities, 19+4% for
the four-year period, with annual values ranging from 13+7% for 1991 to 27+18% in
1992. The percent of area with impacted benthic resources was markedly larger in the
small estuarine systems and tidal rivers, 37+6% and 38+14%, respectively, for the four-
year period. This is likely a reflection of the proximity of the small estuarine systems
and tidal rivers to urban areas which are the major sources of anthropogenic stress.
Table 3-1. Areal Estimates for Impacted Benthic Communities, as Determined by EMAP-Virginian
Province Benthic Index, for the Virginian Province. Values Are Mean Estimate and 95%
Confidence Interval (C.I.). (a) Relative Area (percent), (b) Absolute Area (km2).
(a)
1990
1991
1992
1993
1990-1993
%Area C.I. %Area C.I. %Area C.I. %Area C.I. %Area C.I.
Province
Large Systems
Small Systems
Tidal Rivers
26
15
44
58
7
23
22
27
23
13
50
35
7
7
18
34
28
27
35
25
7
18
14
12
25
22
28
35
10
9
33
4
25
19
37
38
3
4
6
14
(b)
1990
1991
1992
1993
1990-1993
Area C.I. Area C.I. Area C.I. Area C.I. Area C.I.
Province
Large Systems
Small Systems
Tidal Rivers
6,115
2,450
2,153
1,514
1,650
3,638
1,061
705
5,382
2,012
2,455
914
1,650
1,095
880
896
6,650
4,319
1,683
648
1,650
2,854
698
308
5,908
3,614
1,375
918
2,357
1,380
1,614
101
5,894
3,099
1,796
998
738
645
307
361
16
-------
Analysis of impacted benthic condition based on absolute area provides a
different perspective (Table 3-1). Absolute areal extent (km2) was calculated by
multiplying the area of each resource class (see Table 2-1) by the percent area
impacted for each indicator. On an area basis, the large systems contain the greatest
area (3,099 km2) of impacted benthic communities.
Two additional points are worth noting: 1) the variability of annual values was
similar among the three resource classes although the temporal patterns were different,
and 2) the percent area of impacted benthos was similar in the tidal river and small
estuarine classes despite an almost two-fold difference in values for the absolute area
(km2) impacted. These differences result from the difference in actual areas that each
resource class covers.
The spatial patterns of benthic impact, illustrated in Figure 3-2, provide additional
insight into the status and condition of benthic resources in the Virginian Province. It is
clear that impacted benthic communities are associated primarily with the major tidal
rivers and small bays and estuarine systems. These data confirm observations from
extant data that our riverine systems are often the focus for anthropogenic stress. It
should be noted that this spatial display (Figure 3-2) identifies the EMAP sampling sites
having impacted or unimpacted benthic communities, as measured by the benthic
index, when the sample was collected. Lack of data for a particular location indicates
either that no sample was taken or that an acceptable sample was not obtained.
Province
Large Systems Small Systems Tidal Rivers
Figure 3-1. Benthic impact for Virginian Province and resource classes across the years 1990-93.
17
-------
Benthic Community Condition
o missing values
• impacted site
* unimpacted site
Small Tidal Rivers
Systems
Figure 3-2. Condition of benthic communities for Virginian Province for the period 1990-93.
18
-------
Although the benthic index used in this report appears to work well for
distinguishing sites of differing benthic condition, it is not the only effective index for
assessing the condition of estuarine benthic resources (for an example, see
Ranasinghe et a/., 1993 or Weisberg et a/., 1997). An important point about the
benthic index development is that it is based upon empirical data and is designed to
distinguish between impacted and unimpacted conditions; conditions which were
determined by independent criteria.
3.1.2 Benthic Condition: Major Estuarine Systems
An alternative approach for analyzing EMAP monitoring data is to determine the
potential status and risks to benthic resources for specific geographic areas rather than
for the three estuarine resource classes at the province level. The geographic areas
used in the following analyses were classified based on their hydrology into major
watersheds or estuarine drainage areas. Analysis of EMAP data from Chesapeake
Bay, Delaware Bay, the Hudson-Raritan system, and Long Island Sound illustrate the
flexibility of the EMAP sampling design (Table 3-2 and Figures 3-3 and 3-4).
3.1.2.1 Chesapeake Bay
Chesapeake Bay comprises 11,406 km2 or approximately 48% of the Virginian
Province. Analyses were conducted to determine the condition of the benthic resource
for Chesapeake Bay in toto, large and small estuarine systems, and tidal rivers within
the boundaries of the watershed. These analyses indicate that 23±5% of the benthic
120
•o
o
^ 100
a.
E
Figure 3.3.
gRrwince
m Chesapeake Bay
g Delaware Bay
] Hudson-F&ritan
] Long Island Scind
Combined
Large Systems
Small Systems
Tidal Rxers
Benthic impact for major estuarine systems of Virginian Province for the
period 1990-93.
19
-------
Table 3-2. Areal Estimates for Impacted Benthic Communities, as Determined by EEMAP-Virginian Province
Benthic Index, for Major Estuarine Systems in the Virginian Province. Values Are Mean Estimate
and 95% Confidence Interval (C.I.). (a) Relative Area (percent), (b) Absolute Area (km2).
(a)
Chesapeake Bay
Potomac River
Rappahannock River
James River
1990 1991 1992 1993 1990-1993
%Area C.I. % Area C.I. %Area C.I. % Area C.I. %Area C.I.
33 10 20 11
22
27
Long Island Sound 30
22 39 24 39 24 12
11 23
44 22
44 33
19 23
Delaware Bay 12 4 13 23 26 24 44 32 24
Hudson-Raritan 78 13 59 Ot 83 19 70 20 72
17 28
12
8
11
(b)
1990 1991 1992 1993 1990-1993
Area C.I. Area C.I. Area C.I. Area C.I. Area C.I.
Chesapeake Bay 3,764 1,141 2,276 1,255 2,481 979 3,063 1,292 2,672 542
Potomac River * * * * * * * *551 275
Rappahannock River
James River
163 121
130 159
Delaware Bay 246 91 275 476 543 493 915 664 502 240
Hudson-Raritan 589 101 451 Of 633 145 529 154 551 61
Long Island Sound 988 723 1,299 786 1,299 786 397 569 941 359
* There can be a large uncertainty due to the small number of sampling sites for an individual year in a
specific tidal river. Therefore, estimates for the individual years are not presented.
t Due to assumptions in the estimation procedures, if one resource class entirely exceeds the criterion and
other resource classes have no exceedences, the C.I. becomes zero.
20
-------
Impacted jgOnimpactecl
Cheapeake Bay Potomac River Rappahannock James River
River
£
<
+j
c
S
100
80
60
o 40 -
20 -
Province Cheapeake Delaware
Bay Bay
Hudson- Long Island
Raritan Sound
Figure 3-4. Condition of benthic communities for major estuarine systems in Virginian Province for the
period 1990-93.
21
-------
resources of Chesapeake Bay are impacted (Table 3-2). Distribution of impacted
benthos among the resource class is as follows: 26±8% and 36±17% of the area of the
small estuarine systems and tidal rivers, respectively, and 19±6% for the large
estuarine class (Figure 3-4).
For the Potomac, Rappahannock, and James Rivers, 44±22%, 44±33%, and
19±23%, respectively, of the areas had impacted benthos. It should be noted that the
95% confidence intervals for these individual tidal river systems are large due to the
smaller number of sample points that were available to produce the estimates.
3.1.2.2 Delaware Bay
Delaware Bay comprises an area of 2,059 km2 or approximately 9% of the
Virginian Province. Benthic resources were impacted in 24±12% of this area (Table 3-
2). Examination of individual resource classes showed that 18±17% of the large
estuarine class had impacted benthic resources, and the small estuarine systems and
the tidal river with 52±22% and 70±2%, respectively. However, because the area of the
small systems and the tidal river, 275 km2, was only 12% of the estuarine area, the
overall influence on the analysis for the Delaware system is muted.
3.1.2.3 Hudson-Raritan
The Hudson-Raritan system comprises only 760 km2 or approximately 3% of the
Virginian Province. Impacted benthic resources made up 72±8% of the total area (Table
3-2). Individual resource class analyses show that 32±19% of the tidal river and almost
all of the small estuarine systems have impacted benthos. The Hudson-Raritan system
exhibits the most areally impacted benthic resources of the four major estuarine
systems.
3.1.2.4 Long Island Sound
Long Island Sound comprises 3,344 km2 or approximately 14% of the Virginian
Province. Benthic resources were impacted in only 28±11% of the total area (Table 3-
2). Individual resource class analyses show that 26±12% of the large estuarine class
and 51±12% of the small estuarine systems have impacted benthos. The absolute areal
extent of benthic degradation is much greater in large estuarine class despite a factor of
two larger percent area for small systems.
3.2 Dissolved Oxygen Condition Indicator
Dissolved oxygen (DO) is a fundamental requirement for the maintenance of
balanced indigenous populations offish, shellfish, and other aquatic biota. Most
estuarine populations can tolerate short exposures to low dissolved oxygen
22
-------
concentrations. However, prolonged exposures to less that 60% oxygen saturation
may result in altered behavior, reduced growth, adverse reproductive effects, and
mortality (Vernberg, 1972; Reish and Barnard, 1960). Exposure to less than 30%
saturation (~ 2 ppm, for seawater at summer temperatures) for 1 to 4 days causes
mortality to most biota, especially during summer months, when metabolic rates are
high. Stresses that can occur in conjunction with low dissolved oxygen (e.g., exposure
to hydrogen sulfide or ammonia) may cause as much, if not more, harm to aquatic biota
than exposure to low dissolved oxygen concentration alone (Brongersma-Sanders,
1957; Theede, 1973). In addition, aquatic populations exposed to low dissolved oxygen
concentration may be more susceptible to adverse effects of other stressors (e.g.,
disease, toxic substances).
Water column profiles for water quality parameters were collected at each station
using a SeaBird SBE-25 Sea Logger CTD. The unit was equipped with probes to
measure salinity, temperature, depth, pH, dissolved oxygen (DO), light transmission,
fluorescence, and photosynthetically active radiation (PAR).
For assessments with the 1990-93 Virginian Province data, benchmarks of 2 and
5 ppm are used for DO measured in the bottom waters. A concentration of
approximately 2 ppm often is used as a threshold for oxygen concentrations thought to
be extremely stressful to most estuarine biota. A threshold concentration of 5 ppm is
used by many states to set water quality standards. The USEPA, as of this writing, has
not established DO water quality criteria for estuarine and marine waters, but has
developed draft criteria for review and comment.
3.2.1 Dissolved Oxygen Condition
The condition of dissolved oxygen in bottom waters of the Virginian Province is
summarized in Table 3-3 and illustrated in Figure 3-5. The estuarine area with low
dissolved oxygen (DO < 2.0 ppm) in bottom waters was 1+1% in small estuarine
systems, 5+2% in large estuarine systems; and 14+6% in tidal rivers. The four-year
value for the Virginian Province as a whole was 5+2%. Annual values ranged from 2 to
8% in large systems, 4 to 7% in the province, < 1% in small estuarine systems, but was
highly variable (zero to 35%) in the tidal rivers. Percent area analyses of low dissolved
oxygen suggest that tidal river systems are potentially at risk while absolute areal
analyses indicate that large systems are at greatest risk from low dissolved oxygen.
This is due to the large areas associated with the large systems and because these
systems are most likely to stratify in the summer.
The area of bottom waters in the Virginian Province impacted by moderate to
severe hypoxia (< 5.0 ppm) was 17+5% in small estuarine systems, 27+4% in large
estuarine systems, and 18+7% in tidal rivers. The four-year estimate of percent area
impacted by moderate to severe DO for the Virginian Province was 24+3%. Annual
23
-------
Table 3-3. Areal Estimates for Dissolved Oxygen Condition of Bottom Waters in the Virginian
Province. Values Are Mean Estimate and 95% Confidence Interval (C.I.)- (a) Relative
Area (percent), (b) Absolute Area (km2).
(a)
dissolved oxygen <
1990
Province
Large Systems
Small Systems
Tidal Rivers
% Area
7
4
0
35
C.I.
3
6
0*
16
1991
% Area
4
2
1
21
1992
C.I.
3
4
2
20
% Area
6
8
0
0
C.I.
4
11
o*
ot
dissolved oxygen <
1990
Province
Large Systems
Small Systems
Tidal Rivers
(b)
% Area
20
22
7
37
C.I.
6
8
7
17
1991
% Area
17
15
21
21
C.I.
6
7
12
20
1992
% Area
30
38
12
13
C.i.
7
10
10
11
dissolved oxygen <
1990
Province
Large Systems
Small Systems
Tidal Rivers
Area
1,544
631
0
912
C.I.
771
947
ot
429
1991
Area
964
349
69
546
C.I.
705
700
108
512
1992
Area
1,306
1,305
0
0
C.I.
903
1,740
ot
ot
dissolved oxygen <
1990
Province
Large Systems
Small Systems
Tidal Rivers
Area
4,745
3,472
321
952
C.I.
1,431
1,360
323
444
1991
Area
4,024 1
2,450 1
1,029
546
C.I.
,414
,188
575
512
1992
Area
7,011
6,091
586
334
C.I.
1,631
1,603
490
298
2 ppm
1993
% Area
5
8
1
0
5 ppm
C.I.
4
8
3
ot
1993
% Area
27
32
23
3
2 ppm
C.I.
9
10
29
5
1993
Area
1,268
1,215
53
0
5 ppm
C.I.
875
1,214
158
ot
1993
Area
6,353
5,164
1,125
66
C.I.
2,122
1,544
1,392
132
1990-1993
% Area
5
5
1
14
C.I.
2
2
1
6
1990-1993
% Area
24
27
17
18
C.I.
3
4
5
7
1990-1993
Area
1,275
876
35
365
C.I.
408
372
65
167
1990-1993
Area
5,594
4,293
825
475
C.I.
740
718
263
188
: Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion
is either zero or 100%, then the C.I. is zero.
24
-------
Dissolved Oxygen Condition
o missing data
• severe hypoxia
* moderate hypoxia
« good '
Small
Systems
Tidal Rivers
Figure 3-5. Dissolved oxygen condition of bottom waters in Virginian Province for the period 1990-93.
25
-------
values were in the range of 17 to 30% in the province, 15 to 38% in large systems, 7 to
23% in small estuarine systems, and 3 to 37% in the tidal rivers. Analyses of moderate
to severe dissolved oxygen suggest that large estuarine systems are potentially at risk.
The spatial patterns of low dissolved oxygen (< 2.0 ppm) arid moderate hypoxia
(2 ppm < DO < 5 ppm) are illustrated in Figure 3-5. Most of the low dissolved oxygen is
found in the main stem of Chesapeake Bay and the mouths of the Potomac and
Rappahannock Rivers. Only two records of low dissolved oxygen were found outside
Chesapeake Bay. Moderate hypoxia, however, was more equally distributed
throughout the province including western Long Island Sound, the Hudson-Raritan
system, the upper Delaware River, and the central to lower portions of Chesapeake
Bay. Approximately one-half (48+12%) of Long Island Sound exhibits moderate to
severe hypoxia. The results of these estuarine system analyses are summarized in
Tables 3-4 and 3-5 and illustrated in Figure 3-6.
The results for dissolved oxygen conditions were based upon single point-in-time
values recorded during the visit to each station. The usefulness of the point-in-time
results depends upon the stability of the areal extent of low dissolved oxygen through
the index period. This was tested with province-wide data acquired during 1990 for two
successive sampling intervals (19 July to 31 August and 1 September to 23 September)
(Weisberg et a/., 1993). Comparisons of the cumulative distribution functions for bottom
dissolved oxygen concentrations for the two intervals demonstrated that the distribution
of dissolved oxygen was stable between the two sampling periods. Therefore, the point-
in-time measurements of dissolved oxygen were adequate for estimates of status
during the summer index period.
3.3 Sediment Toxicity Condition Indicator
Sediment toxicity tests are the most direct measure available for determining the
toxicity of contaminants in sediments. These tests provide information that is
independent of chemical characterizations and ecological surveys (Chapman, 1988).
They improve upon the direct measure of contaminants in sediments because many
contaminants are tightly bound to sediment particles or are chemically complexed and
are not biologically available (USEPA, 1989). However, sediment toxicity cannot be
used instead of the direct measurement of sediment contaminant concentrations, since
the latter are an important part of interpreting observed mortality in toxicity tests
(USEPA, 1994a).
Sediment toxicity testing has had many applications in both marine and
freshwater environments (Swartz, 1987; Chapman, 1988) and has become an integral
part of many benthic assessment programs (Swartz, 1989). A particularly important
application of sediment toxicity testing is in programs seeking to establish contaminant-
specific effects.
26
-------
Table 3-4. Areal Estimates for Dissolved Oxygen Condition (DO < 2 ppm) of Bottom Waters for Major
Estuarine Systems in the Virginian Province. Values Are Mean Estimate and 95%
Confidence Interval (C.I.). (a) Relative Area (percent), (b) Absolute Area (km2).
(a)
Chesapeake Bay
Potomac River
Rappahannock River
James River
Delaware Bay
Hudson-Raritan
Long Island Sound
(b)
Chesapeake Bay
Potomac River
Rappahannock River
James River
Delaware Bay
Hudson-Raritan
Long Island Sound
1990
% Area C.I.
13 7
*
* *
* *
0 Ot
o ot
o ot
1990
Area C.I.
1,533 760
* *
* it
* *
0 Ot
o ot
o ot
1991 1992
%Area C.I. %Area C.I.
5 4 12 8
* * * *
* * * *
* * * *
o ot o ot
o ot o ot
10 16 0 Ot
1991 1992
Area C.I. Area C.I.
540 439 1,313 859
* * * *
* * * *
* * * *
o ot o ot
o ot o ot
341 524 0 Ot
1993
% Area C.I.
10 7
* *
* *
* *
0 Ot
o ot
o ot
1993
Area C.I.
1,168 819
*
* *
* *
0 Ot
o ot
o ot
1990-1993
% Area C.I.
10
24
15
0
0
0
3
3
12
11
ot
ot
ot
4
1990-1993
Area C.I.
1,136
306
57
0
0
0
85
368
145
41
ot
ot
ot
131
* There can be a large uncertainty due to the small number of sampling sites for an individual year in a
specific tidal river. Therefore, estimates for the individual years are not presented.
t Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion is
either zero or 100%, then the C.I. is zero.
27
-------
Table 3-5. Areal Estimates for Dissolved Oxygen Condition (DO < 5 ppm) of Bottom Waters for Major
Estuarine Systems in the Virginian Province. Values Are Mean Estimate and 95% Confidence
Interval (C.I.)- (a) Relative Area (percent), (b) Absolute Area (km2).
(a) 1990 1991
%Area C.I. %Area C.I.
Chesapeake Bay 28 9 19 9
Potomac River * * * *
Rappahannock River * * * *
James River * * * *
Delaware Bay 3 60 Of
Hudson-Raritan 6978
Long Island Sound 51 25 44 25
(b) 1990 1991
Area C.I. Area C.I.
1992 1993 1990-1993
%Area C,l. %Area C.I. %Area C.I.
37 10 38 16 31
* * * * oc
* * * * 39
* * * * ^
2 7 4 8 3
47 50 12 37 17
51 25 42 25 48
5
12
20
<1
4
17
12
1992 1993 1990-1993
Area C,l. Area C.I. Area
C.I.
Chesapeake Bay 3,193 1,067 2,115 1,027 4,231 1,103 4,353 1,825 3,550 524
Potomac River
Rappahannock River
James River
Delaware Bay
Hudson-Raritan
309 145
145 76
26 <7
66
44
129 0
68 49
45
153
80
160 68
85
61 360 380 93 281 130 129
Long Island Sound 1.705 829 1,477 829 1,705 829 1,410 829 1,592 414
* There can be a large uncertainty due to the small number of sampling sites for an individual year in a
specific tidal river. Therefore, estimates for the individual years are not presented.
$ Due to assumptions in the estimation procedures, if the percent iarea of exceedence
of a criterion is either zero or 100%, then the C.I. is zero
28
-------
DO < 2
|25
Cheapeake Bay Potomac River Rappahannock James River
River
80 -
4)
< 60 -,
•4-> 1
c
o 40 .j
20 -
0 -
Pr
V
mm
ovin
. few;
V&-;# 3
_
- -j"./r j-
'! •' '
(•-
5
/
^""•""5,
ce Cheapeake Delaware Hudson- Long Island
Bay Bay Raritan Sound
Figure 3-6. Dissolved oxygen condition of bottom waters for major estuarine systems in'Virginian
Province for the period 1990-93.
29
-------
Data collected over the four years by EMAP in the Virginian Province measured
the acute toxicity of surficial sediments, i.e., top 2 cm. The sediments used for the
toxicity test were a subsample of the same composite from which sediment contaminant
concentrations and sediment physical/chemical properties were determined.
Sediment toxicity tests were performed on composite sediment samples
collected from each station using the standard 10-day acute test method (Swartz etal.,
1985; U.S. EPA, 19955, taken from U.S. EPA, 1994a) and the tube-dwelling amphipod
Ampelisca abdita. Amphipods were exposed to sediment from the site for 10 days
under static conditions. Ampelisca abdita has been shown to be both acutely and
chronically sensitive to contaminated sediments (Breteler et al., 1989; Scott and
Redmond, 1989). Because it is a tube dweller, Ampelisca is tolerant of a wider range of
sediment types than Rhepoxynius, the genus of amphipod that is commonly used in
sediment toxicity evaluations (Long and Buchman, 1989). Less than 80% survival
relative to control survival in the tested sediments was used as a benchmark for
determining toxic sediments, and less than 60% survival was used to define severely
toxic sediments.
3.3.1 Sediment Toxicity Condition
The toxic condition of bottom sediments in the Virginian Province is summarized
in Table 3-6 and illustrated in Figure 3-7. For the four-year period, sediment toxicity (<
80% survival) was observed in 9±2% of the bottom sediments of the Virginian Province.
Toxicity was observed in 4+4% of the tidal rivers, 9±3% of the large estuaries, and
12+6% of small estuarine system area. The range of annual values for sediment
toxicity in small estuarine systems was high (1-31%) and low in tidal rivers (0-8%).
Analysis of sediment toxicity data on percent area affected suggests that the small
estuarine systems are at greatest relative risk from toxic sediments (in the mean but not
statistically), while analysis of absolute area indicates that large estuaries exhibit over
twice the area of toxic sediments as do small estuarine systems. Toxicity appears to be
spread over the entire province (Figure 3-7).
Severe toxicity (< 60% survival) was generally low, occurring in only 1% of the
estuarine sediments of the Virginian Province, and primarily confined to the northern
extent of the province (Figure 3-7). Small estuarine systems exhibited the largest
percent area with severe toxicity (3±4%), although this distinction may not be
significant.
Distribution of sediment toxicity in the major estuarine systems for the Virginian
Province are illustrated in Figure 3-8 and summarized in Tables 3-7 and 3-8. Toxicity (<
80% survival) was exhibited in all of the major estuarine systems, ranging from a low of
2±2% in Delaware Bay to 15+14% in the Hudson-Raritan system. The toxicity was
observed across all of the classes of these estuarine systems except for Delaware Bay,
30
-------
Table 3-6. Areal Estimates for Bottom SedimentToxicity Condition, as Determined from Acute
Amphipod Bioassay, in the Virginian Province. Values Are Mean Estimate and 95%
Confidence Interval (C.I.)- (a) Relative Area (percent), (b) Absolute Area (km2).
(a)
amphipod survival < 60%
1990
Province
Large Systems
Small Systems
Tidal Rivers
% Area
3
2
6
3
C.I.
3
5
11
5
1991
% Area
1
0
4
0
C.I.
1
ot
6
ot
1992
% Area
<1
0
1
0
1993
C.I. % Area
<1
ot
2
ot
amphipod survival <
1990
Province
Large Systems
Small Systems
Tidal Rivers
(b)
% Area
8
2
31
3
C.I.
5
5
23
5
1991
% Area
21
24
17
8
C.I.
7
9
12
7
1992
% Area
6
8
1
3
C.I.
4
10
2
14
amphipod survival <
1990
Province
Large Systems
Small Systems
Tidal Rivers
Area
764
375
316
72
C.I.
707
749
518
129
1991
Area
212
0
213
0
C.I.
236
ot
287
ot
1992
Area
73
0
73
0
C.I.
73
ot
97
ot
amphipod survival <
1990
Province
Large Systems
Small Systems
Tidal Rivers
Area
1966
375
1519
72
C.I.
1179
749
1137
129
1991
Area
4875
3833
833
209
C.I.
1650
1408
581
195
1992
Area
1346
1207
73
66
C.I.
945
1610
97
366
2
2
4
0
80%
C.I.
3
4
8
0*
1993
% Area
3
2
6
0
60%
C.I.
3
4
9
ot
1993
Area
540
335
205
0
80%
C.I.
707
671
394
ot
1993
Area
620
335
285
0
C.I.
707
671
452
ot
1990-1993
% Area
1
1
3
<1
C.I.
1
2
4
<1
1990-1993
%Area
9
9
12
4
C.I.
2
3
6
4
1990-1993
Area
330
177
152
<26
C.I.
170
288
197
<26
1990-1993
Area
2105
1437
577
91
C.I.
462
447
270
110
t Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion
is either zero or 100%, then the C.I. is zero.
31
-------
Sediment Toxicity
o missing data
• severe toxicity
• moderate toxicity
» no toxicity
Small
Systems
Tidal Rivers
Figure 3-7. Toxicity condition of bottom sediments, as determined from acute amphipod bioassays, in
Virginian Province for the period 1990-93.
32
-------
H survival < 60% H 60% < survival < 80% n survival > 80%
100 -.
80 -
£
< 60 -
8 40 -
0>
Q.
20 -
0 -
,,
„ ,
. "• |
'
.
V
I -,
I'
•
I
HUM
Cheapeake Bay Potomac River Rappahannock James River
River
—
80 -
£
< 60 -
*•>
C
a>
o 40 -
0)
0.
20 -
0
X &$,''•%
,
SsfSSt
••
'
^^^_
':- ;.'
.-•.. ,v._
__:
-'*
'.,
••
,
^'.VV r;:-
•••>
:
r i
, ,„,
Province Cheapeake Delaware
Bay Bay
Hudson- Long Island
Raritan Sound
Figure 3-8. Toxicity condition of bottom sediments, as determined from acute amphipod bioassays,
for major estuarine systems in Virginian Province for the period 1990-93, amphipod
survival < 80%.
33
-------
Table 3-7. Areal Estimates for Bottom Sediment Toxicity Condition (Survival < 80%) for Major Estuarine
Systems in the Virginian Province. Values Are Mean Estimate and 95% Confidence Interval
(C.I.). (a) Relative Area (percent), (b) Absolute Area (km2).
(a)
Chesapeake Bay
Potomac River
Rappahannock River
James River
Delaware Bay
Hudson-Rarifan
Long Island Sound
(b)
Chesapeake Bay
Potomac River
Rappahannock River
James River
1990 1991 1992 11)93 1990-1993
%Area C.I. %Area C.I. %Area C.I. %Area C.I. %Area C.I.
8
6
9
<1
9
8 It
11 19 24 36 16 29 12 37 15 14
23
19
10
16
17
13
1990
Area C.L
909
684
1991 1992 1993 1990-1993
Area C.L Area CJ. Area C.I. Area C.I.
2,199 972 63 349 0 (
)* 695
342
13 <12
33 39
54 77
Delaware Bay 90 162 61 112 0 0$ 0 0$ 38 49
Hudson-Raritan 87 143 181 274 122 220 93 281 117 106
Long Island Sound 0
777 622 341 524 430 552 423 246
* There can be a large uncertainty due to the small number of sampling sites for an individual year in a
specific tidal river. Therefore, estimates for the individual years are not presented
$ Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion is
either zero or 100%, then the C.I. is zero.
34
-------
Table 3-8. Areal Estimates for Bottom Sediment Toxicity Condition (Survival < 60%) for Major Estuarine
Systems in the Virginian Province. Values Are Mean Estimate and 95% Confidence Interval
(C.I.). (a) Relative Area (percent), (b) Absolute Area (km2).
(a)
Chesapeake Bay
Potomac River
Rappahannock River
James River
Delaware Bay
Hudson-Raritan
Long Island Sound
(b)
Chesapeake Bay
Potomac River
Rappahannock River
James River
Delaware Bay
Hudson-Raritan
Long Island Sound
1990
% Area C.I.
0 Ot
* *
* *
* it
4 8
0 Ot
0 Ot
1990
Area C.I.
0 Ot
* *
* *
* *
90 162
0 Ot
0 Ot
1991
% Area C.I.
0 Ot
* *
* *
* *
o ot
5 11
3 6
1991
Area C.I.
0 Ot
* *
* *
* *
o ot
40 84
102 201
1992 1993 1990-1993
% Area C.I. % Area C.I. % Area C.I.
0 Ot 0 Ot 0
'*.*•* * Q
• * * * * 0
0
o ot o ot 1
16 29 11 35 7
0 Ot 13-17 5
ot
ot
ot
ot
2
10
4
1992 1993 1990-1993
Area C.I. Area C.I. Area C.I.
0 Ot 0 Ot 0
0
* * * * 0
* * * * 0
0 Ot 0 Ot 23
122 220 84 266 54
0 Ot 430 552 151
ot
ot
ot
ot
41
76
138
* There can be a large uncertainty due to the small number of sampling sites for an individual year in a
specific tidal river. Therefore, estimates for the individual years are not presented.
t Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion is
either zero or 100%, then the C.I. is zero.
35
-------
where toxicity was only observed in the tidal river. Severe toxicity (< 60% survival) was
observed only in Delaware Bay (1±2%), the Hudson-Raritan system (7+10%), and Long
Island Sound (5+4%). This toxicity was found primarily in small systems in the Hudson-
Raritan and Long Island Sound, but only in the tidal river portion of Delaware Bay. In
summary, sediment toxicity is often associated with the portions of rivers, bays, and
estuaries that are near areas of intense urbanization where a variety of anthropogenic
stressors from land-based activities impact estuarine ecosystems.
Sediment toxicity tests have repeatedly demonstrated a dose response
relationship with single and multiple contaminants under spiked sediment laboratory
conditions (e.g., Di Toro et a/., 1990; Swartz et a/., 1994). Field studies also have
established correlations between sediment toxicity and measures of sediment .
contamination (Long and Morgan 1990; Long et a/. 1995). However, concerns have
been expressed that the effects of factors other than presence of contaminants may
cause toxicity resulting in false positive results (Spies, 1989). The most often cited non-
contaminant factors that affect toxicity are particle size and pore water ammonia. The
amphipod used in the sediment toxicity test, Ampelisca abdita, inhabits fine-grained
sediment, and is one of the more abundant and widespread species found in benthic
communities of the Virginian Province, in water with salinity greater than 15 o/oo. The
potential effects of particle size would thus be operative in the coarser fraction of the
size range. The silt-clay content of the sediments tested over the four years ranged
from < 1 to 100%, yet there was no relationship, statistical or otherwise, between
amphipod survival and silt-clay content.
Estimates of pore water and overlying water concentrations of total and un-
ionized ammonia that are toxic to Ampelisca abdita were established in U.S. EPA
(1994a). No effect concentrations were estimated to be 30 ug/l and 0.4 ug/l for total and
un-ionized ammonia, respectively, at pH 7.7. Concentrations exceeding these levels
indicate the potential for some toxicity due to ammonia. Ammonia concentrations were
measured in pore water of sediment collected in 1993 prior to sediment toxicity testing.
There was no conclusive evidence that ammonia was causing toxicity; sediment with
elevated ammonia concentrations were as likely to be non-toxic as toxic.
A sediment toxicity test that is predictive of benthic community responses to
contaminants should have endpoints that correlate with community parameters of
interest. Comparison of amphipod survival in sediment toxicity test with abundance of
ampeliscid and non-ampeliscid amphipods is presented in Figure 3-9. Ampeliscid
abundance did not exceed 1140/m2 when survival in the toxicity test was less than
50%. At 80% survival, only six sediments contained ampeliscids at densities >
1140/m2. The remaining high densities of these ampeliscids were found at non-toxic
sites, the presence of ampeliscids at moderate to low abundances (0-1140/m2) in
sediments determined to be toxic is not unusual. When sediments are disturbed, as
they are in the process of collection and sample processing prior to toxicity testing (i.e.,
36
-------
a)
10000
o
c:
1 1 1000 -
:3 to
-Q 'c
< to
c E?
8° 100
2E o
^
S -*» 4 — * • •» • «•! 1 l*lt»«lll8»S •
b)
1000
I
CD
100
CO CO
.S2
"53
a.
E
cp
o
20
40 60 80
Amphipod Survival (%)
100
120
20
40 60 80
Amphipod Survival (%)
100
120
Figure 3-9. Relationships between benthic organism abundance and sediment toxicity bioassay. a)
Ampeliscid abundance vs. amphipod survival, b) Non-ampeliscid abundance vs.
amphipod survival.
37
-------
press sieved and homogenized), chemical equilibrium can be disrupted and
contaminants made available to cause toxicity. This effect on contaminant release has
been described for non-polar organic compounds (Word et a/., 1994), and may be
responsible for the toxicity observed in the six sediments with high ampeliscid
abundance and survival < 80%. It should be noted that these sediments exhibited
concentrations of at least one metal at or above ER-L values in conjunction with
elevated PAHs at some stations.
3.4 Sediment Contamination Condition Indicators
Metals, organic chemicals, and fine-grained sediments enter estuaries from
freshwater inflows, point sources of pollution and various nonpoint sources, including
atmospheric deposition. Contaminants generally are retained within estuaries and
accumulate within the sediments (Turekian, 1977; Forstner and Wittman, 1981; Nixon
et a/., 1986; Hinga, 1988; Schubel and Carter, 1984) because most have an affinity for
adsorption onto particles (Hinga, 1988; Honeyman and Santschi, 1988). Chemical and
microbial contaminants generally adsorb to fine-grained materials in the water and are
deposited on the bottom, accumulating at deposition sites, including regions of low
current velocity, deep basins, and the zone of maximum turbidity. The concentrations
of contaminants in sediments is dependent upon interactions between natural (e.g.,
physical sediment characteristics) and anthropogenic factors (e.g., type and volume of
contaminant loadings ) (Sharp et a/., 1984).
Composite, surficial sediment samples were collected using the same
procedures over the four-year period and analyzed for the NOAA National Status and
Trends suite of contaminants (NOAA, 1992) using subsamples from the homogenized
sediment samples. The NOAA suite includes chlorinated pesticides, polychlorinated
biphenyls (PCBs), polycyclic hydrocarbons (PAHs), major elements, and metals (Table
3-9).
Benchmarks of sediment contamination currently are expressed in a number of
ways and can be grouped into two ways that describe one's understanding of what
sediment contamination represents. The first, and most common, form of expression
simply presents contaminant concentrations as a cumulative distribution and a criterion
for contamination is selected based on the nature of the distribution.. Exceedence of
high percentile concentrations, that are sometimes normalized to the proportion of silt-
clay, are then denoted as contaminated. Examples of this approach are found in
O'Connor (1990) and Daskalakis and O'Connor (1994). The second way to establish
whether a site is contaminated is to compare the contaminant concentration against
those considered to be naturally occurring or at background levels. The mere presence
of synthetic organics causes a sediment to be considered contaminated. Regression
relationships of metal concentrations to those of conservative crustal elements, such as
aluminum, are used to determine the degree of enrichment for these compounds
38
-------
Table 3-9. Chemical Measurements Conducted for Sediments of the Virginian Province.
Polycyclic Aromatic Hydrocarbons (PAHs)
Acenaphthathene
Acenaphthlylene
Anthracene
Benz(a)anth racene
Benz(a)pyrene
Benzo(a)pyrene
Benzo(k)fluoranthene
Benzo(g,h,i)perylene
Benzo(e)pyrene
Benzo(b)fluoranthene
Biphenyl
Chrysene
Dibenz(a,b)anthracene
2,6-dimethylnaphthalene
Fluoranthene
Fluorene
1 -methylnaphthalene
2-methylnaphthalene
2-methylphenanthrene
Naphthalene
Perylene
Phenanthrene
DDT and its metabolites
o.p'-DDD
p,p'-DDD
o.p'-DDE
p,p'-DDE
o,p'-DDT
p,p'-DDT
Other chlorinated pesticides
Aldrin
Alpha-chlordane
Trans-Nonachlor
Dieldrin
Heptachlor
Heptachlor epoxide
Hexachlorrobenzene
Lindane (gamma-BHC)
Mirex
PCB congeners
Congener N umber Location of Cl's
Pyrene
Major elements
Al
Fe
Mn
Si
Aluminum
Iron
Manganese
Silicon
Trace elements
Sb
As
Cd
Cr
Cu
Pb
Hg
Ni
Se
Ag
Sn
Zn
Antimony
Arsenic
Cadmium
Chromium
Copper
Lead
Mercury
Nickel
Selenium
Silver
Tin
Zinc
8 24'
18 2 2'5
28 244'
44 2 2' 3 5'
52 2 2' 5 5'
66 2 3' 4 4'
101 2 2' 4 5 5'
105 2 3 3' 4 4'
118 • 2 3' 4 4' 5
128 2 2' 3 3' 4 4'
138 2 2' 3 4 4' 5'
153 2 2' 4 41 5 5'
170 2 2' 3 3' 4 41 5
180 2 21 3 4 4' 5 5'
187 2 2' 3 4' 5 51 6
195 2 2' 3 3' 4 4' 5 6
206 2 2' 3 3' 4 4' 5 5' 6
209 2 2' 3 3' 4 4' 5 5' 6 6'
Other measurements
Tributyltin
Acid volatile sulfides
Total organic carbon
39
-------
(Windom et a/., 1989; see Strobel et al., 1995, for the specific regressions used). These
two approaches are useful in cases where a knowledge of land-based or atmospheric
source inputs is important, and there is less regard for the relationship between the
chemical concentration and biological or ecological effects in the environment.
There are two classes of approaches that establish criteria for contamination that
relate the observed concentration to some biological effect. In the first approach, bulk
chemical concentrations are compared to concentrations known to either: (1) cause
biological effects in spiked-sediment or spiked-water laboratory experiments; or (2) are
associated with biological effects in field studies. Examples of these approaches are the
Puget Sound apparent effects thresholds (AETs), State of Washington screening level
concentrations (SLCs), and effects range median (ER-M) and effects range low (ER-L)
concentrations of Long and Morgan (1990), as updated in Long et a/. (1995). All of
these approaches benefit from the weight of evidence afforded by large data sets
associating bulk concentration with biological effect, but suffer from a failure to
incorporate the effects of multiple chemicals in complex mixtures.
The second approach to effects-based criteria relies on equilibrium partitioning
theory which predicts the concentration of chemical that elicits a biological effect
(bioavailable concentration) from a bulk sediment concentration. This approach
incorporates the role of sediment binding factors, such as organic carbon (OC) for
organic compounds, in defining chemical availability (Di Toro et a/., 1991). The EPA
sediment quality criteria for acenaphthene (USEPA, 1993a), phenanthrene (USEPA,
1993b), fluoranthene (USEPA, 1993c), and dieldrin (USEPA, 1993d) are based upon
this approach. A total PAH concentration of 200 ug/g-OC approximates the sediment
quality criteria values for the three PAHs in this list. An equilibrium partitioning-based
concentration of total DDTs of 100 ug/g-OC has been described by Swartz et al. (1994)
to cause biological effects.
Since the sediment contaminant data collected in the Virginian Province are
used primarily to interpret condition of the benthic community, the major emphasis will
be on contaminant criteria related to biological effects. This section presents
comparisons of chemical concentrations with: crustal levels (metals), ER-M and ER-L
values (Long et al., 1995), EPA sediment quality criteria, a criterion for organic carbon
normalized total PAHs, and the Swartz et al. (1994) organic carbon normalized total
DDT criterion.
3.4.1 Sediment Contamination Condition: Areal Patterns
3.4.1.1 Metals enrichment
Sediment metals enrichment above crustal levels for the province is summarized
in Table 3-10. The metals analyzed for enrichment include As, Cr, Fe, Hg, Mn, Ni, Sb,
40
-------
Table 3-10. Areal Estimates for Enriched Metal Concentrations in Bottom Sediments in the Virginian
Province. Values Are Mean Estimate and 95% Confidence Interval (C.I.)- (a) Relative Area
(percent), (b) Absolute Area (km2).
(a)
1990 1991 1992 1993 1990-1993
%Area C.I. %Area C.I. %Area C.I. %Area C.I. %Area C.I.
Province
Large Systems
Small Systems
Tidal Rivers
(b)
Province
Large Systems
Small Systems
Tidal Rivers
35
29
49
46
7
9
22
32
1990
Area
8,251
4,700
2,369
1,192
1
1
1
C.I.
,721
,497
,092
820
38
31
43
70
10
10
18
18
1991
Area
8,958
5,038
2,116
1,808
C.I.
2,240
1,529
853
458
57
54
67
59
1992
Area
13,461 1
8,676 1
3,271
1,525
7
10
15
20
64
52
83
100
7
10
23
ot
1993
C.I.
,721
,642
741
520
Area
15,017
8,387
4,022
2,602
C.I.
1,650
1,658
1,116
ot
49
42
64
69
4
5
•7
10
1990-1993
Area
11,622
6,696
3,135,
1,782
C.I.
849
789
317
268
t Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion is
either zero or 100%, then the C.I. is zero.
and Zn. An area is enriched if at least one of these metals exceeds the concentration
expected based on crustal weathering. For the entire province during the four-year
sampling interval, 49+4% of the area has at least one enriched metal. Areal extent of
enrichment ranged from 35+7% in 1990 to 64+7% in 1993. Sediments in large
estuarine systems exhibited enrichment in 42+5% of the area, with annual values
ranging from 29±9% in 1990 to 54+10% in 1992. Small systems and tidal rivers
exhibited larger percent area with metals enrichment, 64+7% and 69+10%,
respectively. These two classes also exhibited higher ranges in annual values in the
estimates of percent area of enrichment: 43+18% in 1991 to 83+23% in 1993 for small
systems and 46±32% in 1990 to 100% in 1993 for tidal rivers. These percent area
estimates indicate that small systems and tidal rivers are more enriched with metals
than the large systems.
Although sediments enriched with metals are observed in significant portions of
the sediments in the Virginian Province, no implication should be drawn concerning
potential ecological impact. For the metals levels to be of biological consequence, the
metals must be biologically available and above a level found to be toxic. The sediment
metals enrichment indicator only identifies those sediments that have concentrations
above expected crustal, or background, levels, and are potentially more useful in trends
rather than status assessments.
41
-------
Table 3-11. Areal Estimates for Bottom Sediment Contaminant Condition for Metals in the Virginian
Province. Values Are Mean Estimate and 95% Confidence Interval (C.I.). (a) Relative
Area (percent), (b) Absolute Area (km2).
(a)
Any ER-M Metal Exceedence
1990
Province
Large Systems
Small Systems
Tidal Rivers
% Area
5
2
18
3
C.I.
4
4
16
5
1991
% Area
5
2
11
11
C.I.
3
4
8
14
1992
% Area
7
8
6
5
C.I.
4
10
5
4
1993
% Area
1
0
3
8
C.I.
1
0*
4
8
1990-1993
% Area
5
3
9
7
C.I.
1
3
4
4
(b)
1990
Province
Large Systems
Small Systems
Tidal Rivers
Area
1,294
335
868
90
C.I.
943
671
786
138
1991
Area
1,148
335
532
282
C.I.
592
671
406
353
1992
Area
1,638
1,207
313
118
C.I.
879
1,610
234
117
1993
Area
349
0
140
209
C.I.
198
ot
216
197
1990-1993
Area
1,073
470
429
175
C.I.
297
404
172
111
: Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion
is either zero or 100%, then the C.I. is zero.
3.4.1.2 Metal ER-M exceedence
Sediment contaminant condition relative to ER-M exceedence for metals is
summarized in Table 3-11 and illustrated in Figure 3-10. Exceedence of the ER-M for at
least one metal for which ER-Ms exist was found in 5±1% of the area of the province
during the four-year sampling interval. The percent area of exceedence ranged from
1% in 1993 to 7% in 1992. Sediments in large estuarine systems exhibited exceedence
in 3+3% of the area with annual values ranging from zero in 1993 to 8% in 1992. Seven
percent of the area (±4%) in tidal rivers contained sediments with ER-M exceedence;
the percent area of exceedence was low in 1990 and 1992 (3-5%), and higher in 1991
and 1993 (11% and 8%, espectively), but these differences may not be significant.
Small estuarine systems exhibited the highest proportion of area, 9±4%, with ER-M
exceedence. Most of those high concentrations were found in 1990 and 1991 (18% and
11%, respectively). These ER-M metals exceedence suggest that small estuarine
systems are at greater risk than are large or tidal river systems. It is worth noting that
the annual variability for metal exceedence parallels that observed with sediment
toxicity, which was also greatest in small systems in 1990 and in tidal rivers in 1991.
These annual patterns in condition support the rationale of using the entire four-year
data set to minimize the uncertainty in the description of estuarine condition.
42
-------
Metal ERM Exceedance
(a no exceedence
metal ERM exceedence
n
<
Is
2
V
0.
100
80 -
60 -
40 -
20 -
0 -
-~
* '
™
'™*, "
•*' *
mmm
Province
Large Small Tidal Rivers
Systems Systems
Figure 3-10. Sediment contaminant condition of bottom sediments in Virginian Province for the period
1990-93, ER-M metals exceedence.
43
-------
The approach used for this indicator was developed by attempts to identify
concentrations of contaminants that are rarely associated with adverse biological
effects (ER-L, discussed in a later section) and those usually associated with effects
(ER-M), utilizing data from an extensive search of the literature (Long etal., 1995). The
underlying assumption of this approach was that, if enough data are accumulated, a
pattern of increasing incidence of biological effects should emerge with increasing
contaminant concentrations. It has been argued that sediment quality criteria are not
defensible if they do not account for factors that control bioavailability, such as acid
volatile sulfides for metals and TOG for organic compounds (Di Toro et ai, 1991).
However, Long et a/. (1995) argue that the approach is accurate for most chemicals
and that their results agree reasonably with other guidelines. They then conclude that
their guidelines are likely to be reliable tools in sediment quality assessments. The
debate does continue on the use of this approach as an indicator of sediment
contamination.
3.4.1.3 Organic ER-M exceedence
The percent area of ER-M exceedence for organics is found in Table 3-12 and
Figure 3-11. Note that because of Quality Assurance (QA) problems associated with
organic chemistry results from 1990 (Strobel and Valente, 1995), the organic chemistry
analyses for 1990 are not presented and are not used to produce the four-year
estimates. Exceedence of at least one ER-M value was found in 3±1 % of the area of
the Virginian Province for the four-year period. The proportional extent of organic
contamination was low in large and small estuarine systems (2%) and higher in tidal
rivers (14±2%). Much of this is due to the well-documented PCB contamination in the
Hudson River (Feng et a/., 1998). The areal extent of organic ER-M exceedence in
1991, 1992, and 1993 was 11%, 19%, and 12%, respectively. Tidal rivers appear to be
at the greatest risk to organic contamination. Based on the area affected, the extent of
contamination in tidal rivers is less than that in the large estuarine class.
Similar comments on the utility of this indicator, as discussed with the ER-M
metals results in section 3.4.1.2), apply here.
3.4.1.4 Organic carbon normalized total PAHs
The results for potentially toxic PAH sediment contamination based on organic
carbon normalization is shown in Table 3-12. Note that because of QA problems
associated with organic chemistry results from 1990 (Strobel and Valente, 1995), the
organic chemistry analyses for 1990 are not presented and are not used to produce the
four-year estimates. The extent of estuarine area with PAH contamination, as
represented by concentrations > 200 ug/g-OC, was low (5±1%) in the province for the
years 1991-1993. Variability amongst years also was low. The affected area in large
systems was lowest (2±2%), and that in the tidal river systems was highest (14±6%);
44
-------
Table 3-12. Areal Estimates for Bottom Sediment Contaminant Condition for Organics in the
Virginian Province. Values Are Mean Estimate and 95% Confidence Interval (C.I.)- (a)
Relative Area (percent), (b) Absolute Area (km2).
(a)
1990
Any ER-M Organic Exceedence
1991 1992 1993
% Area C.I.
Province
Large Systems
Small Systems
Tidal Rivers
tt
tt
n
tt
1990
tt
tt
tt
tt
%Area
4
2
5
11
OC
C.I.
2
4
6
6
% Area
4
3
1
19
C.I. % Area
2
5
2
8
Normalized PAHs > 200
1991
% Area C.I.
Province
Large Systems
Small Systems
Tidal Rivers
(b)
tt
tt
tt
tt
tt
tt
tt
tt
%-Area
6
2
11
10
C.I.
3
4
8
5
1992
% Area
5
3
11
13
3
2
2
12
C.I.
2
4
4
6
1990-1993
% Area
4
2
3
14
C.I.
1
3
2
4
mg/g-OC
1993
C.I. % Area
3
5
7
8
5
0
13
19
C.I.
4
ot
18
17
1990-1993
% Area
5
2
11
14
C.I.
2
2
7
6
Any ER-M Organic Exceedence
1990
Province
Large Systems
Small Systems
Tidal Rivers
Area
tt
tt
tt
tt
C.I.
tt
it
tt
tt
1990
Province
Large Systems
Small Systems
Tidal Rivers
Area
tt
tt
tt
tt
C.I.
tt
tt
tt
tt
1991
Area
849
335
222
291
OC
C.I.
500
671
268
164
1992
Area
959
402
73
484
C.I.
556
805
97
208 '
Normalized PAHs > 200
1991
Area
1,414
335
514
272
C.I.
707
671
401
131
1992
Area
1,247
402
518
327
C.I.
707
805
343
208
1993
Area
743
335
104
303
C.I.
497
671
203
158
1990-1993
Area
850
357
133
360
C.I.
299
415
117
103
mg/g-OC
1993
Area
1,132
0
641
490
C.I.
943
ot
897
441
1990-1993
Area
1,264
246
558
363
C.I.
458
349
347
168
t Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion
is either zero or 100%, then the C.I. is zero.
# Due to QA problems with the organic chemistry analyses for 1990 samples (Strobel and
Valente, 1995), results for 1990 organic chemistry analyses are not presented.
45
-------
Organic ERM Exceedance
* no exceedence
• organic ERM exceedence
(8 8° '
4s 60 -
<*->
| 40-
0)
Q. 20 -
0 -
F
_
m^
'rovinc
&
j
1
T t
e Large Smal
,
'•"'
Tidal Rivers
Systems Systems
Figure 3-11. Sediment contaminant condition of bottom sediments in Virginian Province for the period
1990-93, ER-M organics exceedence.
46
-------
the small estuarine systems exhibited an intermediate level of impact (11 ±4%). The
range for annual values in large and small estuarine systems was moderate; in tidal
river systems, the extent of degradation in 1991-1993 ranged from 10 to 19%. These
data suggest that the risk due to bioavailable PAHs on the province scale is low, and
more likely to occur in the small systems and tidal rivers.
3.4.1.5 Organic carbon normalized total DDT
There were no sample sites exceeding the criterion of total DDT concentrations >
100 ug/g-OC in the Virginian Province in 1991-1993.
3.4.1.6 Organic sediment quality criteria
The estuarine area of the Province that contained sediments exceeding any one
of the sediment quality criteria for the four criteria chemicals (acenaphthene - 230 ug/g-
OC; phenanthrene - 240 ug/g-OC, fluoranthene - 650 ug/g-OC; dieldrin - 20 ug/g-OC;
USEPA, 1993a-d) was < 1%. Those exceedence were restricted to small estuarine
systems, which exhibited < 1% of the area exceeding the criteria.
3.4.1.7 ER-L exceedence
The results for no ER-L exceedence (metals and organics) are summarized in
Table 3-13. Note that these results are depicted as areas for which there were no
ER-L exceedence observed. In contrast to the results presented for ER-M
exceedence which indicated potential for biological effects, the ER-L results are
presented indicating areas for which observed sediment contaminant concentrations
are below levels for which biological effects may be of concern. Half of the entire
province (50+4%) had sediment contaminant concentrations below ER-L levels. Annual
values ranged from 44±8% in 1993 to 53+7% in 1992. The large estuarine system
class exhibited the largest areal extent of sediments below ER-L values, 58+5%, with
small systems and tidal rivers exhibiting 35+6% and 25+12%, respectively. The range
in annual values was smallest for tidal rivers (from 21+30% in 1993 to 29+17% in 1991)
and largest for small systems (from 23+17% in 1990 to 51+39% in 1993). These results
indicate that sediments of the small systems and tidal rivers are more at risk to
sediment contaminant levels that have been observed to potentially elicit biological
responses.
3.4.2. Sediment Contamination Condition: Estuarine System Analysis
3.4.2.1 Metals enrichment
Sediment metals enrichment results for the major estuarine systems are
summarized in Table 3-14. Chesapeake and Delaware Bays exhibited the least
47
-------
Table 3-13. Areal Estimates for Bottom Sediment Contaminant Condition (No IER-L Exceedence) in the
Virginian Province. Values Are Mean Estimate and 95% Confidence Interval (C.I.). (a) Relative
Area (percent), (b) Absolute Area (km2).
(a)
Province
Large Systems
Small Systems
Tidal Rivers
1990 1991 1992 1993 1990-1993
% Area C.I. % Area C.I. % Area C.I. % Area C.I. % Area C.I.
49
60
23
25
7
10
17
24
50
60
29
29
7
10
18
17
53
65
28
24
7
10
13
26
44
46
51
21
8
10
39
30
50
58
35
25
4
5
6
12
(b)
Province
Large Systems
Small Systems
Tidal Rivers
1990
Area C.I.
11
9,
1,
,516
726
129
660
1,730
1,611
821
620
1991
Area C.I.
11,884
9,726
1,393
766
1,676
1,616
881
441
1992
Area C.I.
12,464
10,463
1,384
618
1,714
1,578
618
668
1993
Area C.I.
101,431
7,377
2,503
550
1,825
1,648
1,896
782
1990-1993
Area C.I.
11,768
9,407
1,712
648
865
805
270
320
enrichment of the major systems, 44+5% and 39+14%, respectively. However, within
these individual systems, the small systems and tidal river classes exhibited much
higher enrichment than that exhibited for the large system class, consistent with the
pattern shown for the entire province. The metals enrichment for the Hudson-Raritan
system and Long Island Sound were almost identical, 85+11% and 86±8%,
respectively. In contrast to the pattern seen across the province, Long Island Sound
large system class exhibited more enrichment (89±9%) than the small system class
(59+17%). These results for the major estuarine systems indicate that Hudson-Raritan
and Long Island Sound are almost entirely enriched with metals, while less than half of
Chesapeake and Delaware Bays exhibit metals enrichment. However, the small
systems and tidal rivers within all of these major estuarine systems exhibited significant
metals enrichment. The observed patterns of enrichment appear to be consistent with
the patterns of population distribution and point source discharges across the province,
i.e., higher levels of both in the New York City area.
3.4.2.2 Metal ER-M exceedence
The percent of estuarine area with exceedence of ER-M metals in geographic
areas are provided in Table 3-15 and Figure 3-12. The Hudson-Raritan Estuary is the
most contaminated major system with respect to this indicator; this system exhibited
27+9% of the area with sediments exceeding at least one ER-M value. Most of this
contamination was due to mercury, lead, and silver in the tidal portion of the Hudson
River (30% of the Hudson-Raritan area). The least contaminated system was Delaware
48
-------
Table 3-14. Areal Estimates for Enriched Metal Concentrations in Bottom Sediments for the Major
Estuarine Systems in the Virginian Province. Values Are Mean Estimate and 95% Confidence
Interval (C.I.)- (a) Relative Area (percent), (b) Absolute Area (km2).
(a)
Chesapeake Bay
Potomac River
1990 1991 1992 1993 1990-1993
%Area C.I. %Area C.I. %Area C.I. %Area C.I. %Area C.I.
30
32
54
11
59
44
53
14
Rappahannock River
James River
81 19
85 14
Delaware Bay
Hudson-Raritan
(b)
Chesapeake Bay
Potomac River
1990
Area C.I.
38 29 44 34 64
32
Long Island Sound 88 16 75
21
1991
Area C.I.
100 OJ
1992
Area C.I.
85
17
1993
Area C.I.
39 14
69 25 68 35 100 0$ 100 0$ 85 11
86
8
1990-1993
Area C.I.
3,365 1,004 3,650 992 6,205 1,198 6,673 1,038 5,042 525
663 176
Rappahannock River
James River
301 69
575 94
Delaware Bay
Hudson-Raritan
156 115 784 597 906 696 1,316 649 793 282
527 190 519 266 760 0$ 760 0$ 644 84
Long Island Sound 2,950 525 2,509 692 3,345 Oj: 2,843 552 2,880 258
* There can be a large uncertainty due to the small number of sampling sites for an individual year in a
specific tidal river. Therefore, estimates for the individual years are not presented.
$ Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion is either
zero or 100%, then the C.I. is zero.
49
-------
Table 3-15. Areal Estimates for Bottom Sediment Contaminant Condition (Any ER-M Metal Exceedence)
for Major Estuarine Systems in the Virginian Province. Values Are Mean Estimate and 95%
Confidence Interval (C.I.). (a) Relative Area (percent), (b) Absolute Area (km2).
(a) 1990
% Area C.I.
Chesapeake Bay 14 6
Potomac River * *
Rappahannock River * *
James River * *
Delaware Bay <1 6
Hudson-Raritan 34 28
Long Island Sound 7 1
(b) 1990
Area C.I.
Chesapeake Bay 1,592 684
Potomac River * *
Rappahannock River * *
James River * *
Delaware Bay 5 115
Hudson-Raritan 258 213
1991 1992 1993 1990-1993
%Area C.I. %Area C.I. %Area C.I. %Area C.I.
4596125
****** <^
****** Q
****** o
1 5 0 Ot 2 7 1
29 18 30 32 26 23 27
3 6 12 16 <1 1 4
3
<1
ot
10
3
9
4
1991 1992 11993 1990-1993
Area C.I. Area C.I. Area C.I. Area C.I.
449 573 1,020 708 68 228 594
* ' 3
0
52
20 112 0 Of 44 148 17
222 137 226 243 196 175 209
342
<12
ot
68
55
68
Long Island Sound 221 33 113 201 404 524 13 33 146 131
* There can be a large uncertainty due to the small number of sampling sites for an individual year in a
specific tidal river. Therefore, estimates for the individual years are not presented.
$ Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion is
either zero or 100%, then the C.I. is zero.
50
-------
B ERM metals exceedence n no exceedence 1
100 -.
80 -
£
< 60 -
+J
S 40
0)
0.
20 -
0 _
'
, , «>.*
*~
fe *
•f ^
.^•Sl-w.
,
"4 -
.r ~:
j ~*
* r?*
J
-
-------
Bay where only 1+3% of the area contained sediments exceeding ER-M values. These
exceedence were restricted to elevations of lead, nickel, and zinc in the Delaware
River. Chesapeake Bay and Long Island Sound were intermediate in level of
contamination with percent area of exceedence of 5±3% and 4+4%, respectively.
Elevations of nickel and zinc in the open waters of upper Chesapeake Bay and
bordering small estuarine systems accounted for exceedence in this system. Long
Island Sound contained the widest range of metal contamination with concentrations of
mercury, lead, silver, copper, and nickel exceeding ER-M values. These exceedence
were predominately found in small systems bordering the open waters of the western
Sound.
3.4.2.3 Organic ER-M exceedence
Table 3-16 and Figure 3-13 describe the percent area of ER-M organic
contaminant concentration exceedence. The Delaware Bay system was the least
contaminated of the four major systems. Less than 1% of its area contained sediments
exceeding organic ER-M values, and those sediments were restricted to the Delaware
River. Elevated concentrations of the chlorinated pesticides pp-DDE and total DDTs
accounted for these exceedence. A similarly low level of contamination was found in
Chesapeake Bay and Long Island Sound; 2-4% of their areas contained sediments with
organics exceeding the ER-M values. Elevations in concentration of a number of PAHs
resulted in 5+6% of the area of the open waters of Chesapeake Bay and 22+23% of the
Rappahannock River being contaminated relative to this indicator. High concentrations
of high and low molecular weight PAHs, total DDTs, and total PCBs in 20+27% of the
area in small estuarine systems bordering open waters of Long Island Sound resulted
in exceedence of the ER-M values. The greatest level of organic contamination was
found in the Hudson-Raritan system: 44+14% of the area of this system contained
sediments with organic concentrations that exceeded at least one ER-M value. Eighty-
seven percent (+23%) of the area in the tidal portion of the Hudson River, and 14+18%
of the area in the small systems exhibited ER-M exceedence. Small estuarine systems
were contaminated primarily with PAHs. The Hudson River was contaminated with
PCBs and, to a lesser extent, DDTs.
3.4.2.4 Organic carbon normalized total PAHs
The extent of estuarine area exceeding 200 ug/g-OC total PAH is detailed in
Table 3-17. The Hudson-Raritan Estuary was, by far, the most contaminated system
with respect to total PAH concentrations exceeding 200 ug/g-OC. Sixty-three percent
(+18%) of the area of this system contained sediments exceeding this criterion. The
affected area was proportionately greater in the Hudson River (74±17%) than it was in
the small systems (56±28%). The extent of OC-normalized PAH contamination was
uniformly low in Chesapeake Bay, Long Island Sound, and Delaware Bay (2 to 5%).
The greatest degree of contamination, other than found in the Hudson-Raritan, was in
52
-------
Table 3-16. Areal Estimates for Bottom Sediment Contaminant Condition (Any ER-M Organics Exceedence)
for Major Estuarine Systems in the Virginian Province. Values Are Mean Estimate and 95%
Confidence Interval (C.I.). (a) Relative Area (percent), (b) Absolute Area (km2).
(a)
Chesapeake Bay
Potomac River
Rappahannock River
James River
Delaware Bay
Hudson-Raritan
Long Island Sound
1990 1991 1992 1993 1990-1993
%Area C.I. %Area C.I. %Area C.I. % Area C.I. %Area C.I.
* * * * * * * * Q Q+
* * * * * * ** 22 23
*** * * * * * Q Q4-
it it 1 5 0 Ot 0 Ot <1 2
tt it 36 18 53 32 42 23 44 14
it it
0
Ot
(b)
Chesapeake Bay
Potomac River
Rappahannock River
James River
Delaware Bay
Hudson-Raritan
Long Island Sound
1990 1991 1992 1993 1990-1993
Area C.I. Area C.I. Area C.I. Area C.I. Area C.I.
tt it 350 479 574 543 292 439 405
********rv
******** 83
* * * * * . * * * f\
it tt 20 112 o ot o ot 7
ti it 276 137 400 243 322 175 333
tt tt 113 201 0 Ot 53 100 55
282
ot
84
ot
37
110
75
* There can be a large uncertainty due to the small number of sampling sites for an individual year in a
specific tidal river. Therefore, estimates for the individual years are not presented.
t Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion is either
zero or 100%, then the C.I. is zero.
tt Due to QA problems with the organic chemistry analyses for 1990 samples (Strobel and Valente, 1995),
results for 1990 organic chemistry analyses are not presented.
53
-------
B ERM organics exceedence Q no exceedence
100 ,
80 -
0)
< 60 -
o>
o 40 -
o>
Q.
20 -
0 .
•„• ,: ...
* "- :
m
::' ;':•'.!
•
} i
** j
/I *}
'! J t
/ ^ v
Cheapeake Bay Potomac River Rhapponnock James River
River
Province Cheapeake Delaware
Bay Bay
Hudson- Long Island
Raritan .Sound
Figure 3-13. Sediment contaminant condition of bottom sediments for major estuarine systems in
Virginian Province for the period 1990-93, ER-M organics exceedence.
54
-------
Table 3-17. Areal Estimates for Sediment Contamination Condition (Organic Carbon-Normalized PAHs >
200 Mg/g-OC) for Major Estuarine Systems in Virginian Province. Values Are Mean Estimate
and 95% Confidence Interval (C.I.). (a) Relative Area (percent), (b) Absolute Area (km2).
(a)
Chesapeake Bay
Potomac River
Rappahannock River
James River
Delaware Bay
Hudson-Raritan
Long Island Sound
(b)
Chesapeake Bay
Potomac River
Rappahannock River
James River
1990 1991 1992 1993 1990-1993
%Area C.I. %Area C.I. %Area C.I. %Area C.I. %Area C.I.
tt tt
8
o ot
22 23
2 17
tt tt
tt tt
tt tt
1990
Area C.I.
5 9
59 38
4 6
1991
Area C.I.
0 Ot
78 14
0 0$
1992
Area C.I.
8 10
53 37
3 5
1993
Area C.I.
4
63
2
5
18
3
1990-1993
Area
C.I.
396 480 820 543 348 913 521
388
83 84
16 112
Delaware Bay it tt 104 194 0 OJ 160 209 88 95
Hudson-Raritan ft ft 447 289 589 103 402 281 479 139
Long Island Sound
123 201
0
110 167
78
87
* There can be a large uncertainty due to the small number of sampling sites for an individual year in a
specific tidal river. Therefore, estimates for the individual years are not presented.
t Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion is
either zero or 100%, then the C.I. is zero.
tt Due to QA problems with the organic chemistry analyses for 1990 samples (Strobel and Valente, 1995),
results for 1990 organic chemistry analyses are not presented.
55
-------
the small systems bordering Long Island Sound (28±32% of the area), and in the tidal
portion of Delaware River (35±39% of the area). However, because of small sample
sizes for some of these estimates, uncertainties are large.
3.4.2.5 ER-L exceedence
The major estuarine system results for no ER-L exceedence (metals and
organics) are summarized in Table 3-18. Remember that these results are for areas
where no ER-L exceedences are observed. Almost none of the sediments in the
Hudson-Raritan system (1+3%) were observed to have contaminant concentrations
below ER-L levels. In other words, almost the entire Hudson-Raritan system has
sediments above levels that have been observed to potentially elicit biological
responses. In contrast, half or more of Chesapeake Bay (50+5%) and Delaware Bay
(62+14%) were observed to have sediment concentration levels below ER-Ls. Long
Island Sound exhibited 24+12% of the sediments to be below ER-L levels. Within
Chesapeake and Delaware Bays, the large system class exhibited more sediment area
below ER-L levels compared with the small systems and tidal rivers. In contrast, Long
Island Sound results indicated more percent area with sediments below ER-L levels in
small systems than in the large system class. These results for distribution of sediment
contamination across the classes within the major estuarine systems are consistent
with those observed for metals enrichment presented in an earlier section. The major
estuarine system results indicate that the Hudson-Raritan and Long Island Sound
systems are more at risk due to sediment contamination above potential biological
effect levels compared to Chesapeake and Delaware Bays.
3.5 Indicator Co-occurrence
One strength of the EMAP sampling design and indicator program is the co-
location of many types of measurements at each site. The synoptic nature of the data
facilitates identification of associations between condition indicators and habitat and/or
stressor indicators. Although these associations do not define cause and effect, they
can be used both to gauge how ecological condition indicators reflect habitat condition
or measures of stress and to formulate hypotheses concerning causal relationships.
3.5.1 Statistical Patterns for Benthic Impact and Habitat Condition
These analyses were conducted to determine which types of habitat
(environmental) conditions (e.g., silt-clay content, total organic carbon (TOC) content of
the sediments, and bottom water salinity) are more likely to be associated with
impacted benthic conditions. The distributions of each of the three habitat indicators
are first presented as they were observed across all of the sampling sites, then as they
are associated with impacted and unimpacted benthic areas.
56
-------
Table 3-18. Areal Estimates for Bottom Sediment Contaminant Condition (No ER-L Exceedence) for Major
Estuarine Systems in the Virginian Province. Values Are Mean Estimate and 95% Confidence
Interval (C.I.)- (a) Relative Area (percent), (b) Absolute Area (km2).
(a)
Chesapeake Bay
Potomac River
Rappahannock River
James River
Delaware Bay
Hudson-Raritan
Long Island Sound
(b)
Chesapeake Bay
1990 1991
%Area C.I. %Area C.I.
49 10 48 10
* * * *
* * * *
* * * *
92 6 50 32
5 13 0 Ot
32 24 45 25
1990 1991
Area C.I. Area C.I.
5,624 1,159 5,435 1,181
1992 1993
%Area C.I. %Area C.I.
54 10 44 12
* * * *
* * * *
* * * *
53 31 52 31
0 0$ 0 Ot
31 24 28 22
1992 1993
Area C.I. Area C.I.
6,142 1,129 5,045 1,339
1990-1993
%Area C.I.
50
36
0
23
62
1
24
5
20
ot
20
14
3
12
1990-1993
Area C.I.
5,700
602
Potomac River
Rappahannock River
James River
Delaware Bay
Hudson-Raritan
Long Island Sound
* . * * * * * * 453 253
* * * * * * * * A f\4-
* * * * * * * 153 138
1,887 114 1,020 654 1,100 648 1,070 648 1,269 283
40 101 0 Orj: 0 0$ 0 0$ 10 25
1,070 803 1,505 836 1,037 803 937 736 803 401
* There can be a large uncertainty due to the small number of sampling sites for an individual year in a
specific tidal river. Therefore, estimates for the individual years are not presented.
$ Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion is
either zero or 100%, then the C.I. is zero.
57
-------
Table 3-19. Areal Estimates for Association of Silt-clay Content of Sediments with Benthic Condition
for the Virginian Province.
silt-clay (%)
entire province
(% Area)
impacted benthic
communities
(% Area)
unimpacted benthic
communities
(% Area)
Province
Large Systems
Small Systems
Tidal Rivers
<20
20-80
>80
<20
20-80
>80
<20
20-80
>80
<20
20-80
>80
46
29
26
52
30
18
33
28
39
31
19
50
34
24
42
36
25
39
31
24
44
33
18
49
49
31
20
56
31
13
30
36
34
33
22
44
3.5.1.1 Silt-clay content of sediments
The distributions of silt-clay content in the sediments of the Virginian Province
are presented in Table 3-19. Muds are sediments with silt-clay content > 80%, and
sands have silt-clay content < 20%. Almost half of the Virginian Province (46±4%) has
sandy sediments. The large systems are primarily sandy environments (52+5%), the
small systems have almost equal distribution across the three silt-clay categories, and
tidal rivers are mostly mud (50+16%).
The associations for silt-clay content with impacted and unimpacted benthos
across the Virginian Province are presented in Table 3-19. Areal analyses indicate that
impacted benthic communities tend to be associated with muddy sediments,
irrespective of resource class (large, 39%; small, 44%; and tidal, 49%). Unimpacted
benthic communities are associated more closely with sandy environments in large
systems, while in large tidal rivers and small systems unimpacted benthos are
distributed evenly across silt-clay categories.
3.5.1.2 Total organic carbon content of sediments
The distributions of TOG content in the sediments across the Virginian Province
58
-------
Table 3-20. Area! Estimates for Association of Total Organic Carbon (TOC) Content of Sediments
with Benthic Condition for the Virginian Province.
TOC (%)
entire province
(% Area)
impacted benthic unimpacted benthic
communities communities
(%Area) (%Area)
Province
Large Systems
Small Systems
1 to 3
>3
1 to 3
>3
1 to 3
>3
54
39
7
63
31
6
37
54
10
29
51
20
27
50
23
32
52
16
51
46
4
60
37
3
36
58
6
Tidal Rivers
<1
1 to 3
>3
34
58
8
33
51
16
17
74
9
are presented in Table 3-20. Fifty-four percent (±4%) of the Virginian Province
sediments contain low TOC (TOC < 1%). The large estuarine systems are primarily low
TOC environments (63+5%). Small estuarine systems and tidal rivers are moderate
TOC environments; 54+4% and 58+5% of the area in these systems are in the 1-3%
TOC range, respectively.
Associations for TOC content with impacted and unimpacted benthos across the
Virginian Province are presented in Table 3-20. For the province as a whole and all of
its component resource classes, impacted benthic communities tend to be associated
with the moderate TOC content sediments (1-3% range). This contrasts with
unimpacted benthos across the entire province and for large systems where the
association is with low organic carbon environments, and for small and tidal river
systems where the association is with moderate organic environments. This pattern for
unimpacted benthic communities mimics the TOC distribution across the province
(Table 3-20).
3.5.1.3 Bottom water salinity
The areal estimates of bottom water salinity for the Virginian Province is
59
-------
Table 3-21. Areal Estimates for Association of Bottom Water Salinity with Benthic Condition for the
Virginian Province.
salinity (ppt)
entire province
(% Area)
impacted benthic unimpacted benthic
communities communities
(% Area) (% Area)
Province
Large Systems
Smalt Systems
Tidal Rivers
<5
5 to 18
>18
<5
5 to 18
>18
<5
5 to 18
>18
<5
5 to 18
>18
6
28
66
1
20
79
9
38
54
37
56
7
11
28
61
0 .
24
76
16
21
63
42
58
0
5
28
67
1
22
77
7
36
57
38
56
6
presented in Table 3-21. Almost two-thirds of the Virginian Province bottom waters
(66+3%) are polyhaline (salinity > 18 o/oo). The bottom waters of the large systems are
primarily polyhaline environments (79+14%). The areal distribution of bottom water
salinity in small systems closely follow the distribution of bottom water salinity
represented by the entire province. The tidal rivers have mostly low salinity bottom
water (37+10% and 56±4% for oligohaline (< 5 o/oo) and mesohaline (5-18 o/oo),
respectively).
The association of bottom water salinity with impacted and unimpacted benthos
across the Virginian Province are presented in Table 3-21. Impacted benthic
communities tend to associate with polyhaline bottom waters in large and small
systems in proportion to salinity distribution in these systems. Similarly, impacted
benthos in tidal rivers are proportionately associated with low and moderate salinity
environments (< 5 and 5-18 o/oo, respectively). Unimpacted benthos exhibit similar
distribution in proportion to salinity occurrence. The association analysis indicates that
there is no difference in associations for unimpacted and for impacted benthic
communities. This confirms that the benthic index as used for this analysis is
independent of salinity; in other words, the salinity normalization for the benthic index
metrics was effective (see Appendix B).
60
-------
3.5.1.4 Summary for habitat associations
The Virginian Province estuarine habitat is predominantly polyhaline and is
composed of mostly sandy, low TOC content sediments. Impacted benthic
communities are most frequently associated with muddy, moderate TOC content
sediments and polyhaline bottom waters. The exception is that impacted tidal rivers are
associated with the lower salinity (oligo- and mesohaline) bottom waters.
3.5.2 Statistical Pattern for Benthic Impact and Stressors
Three types of analyses were conducted to link impacted benthic communities
with stressors. The first examines associations among the individual stressors that co-
occur with the impacted benthos. The individual stressors include dissolved oxygen,
sediment toxicity, and sediment contaminants exceeding ER-M values. Metals
enrichment was not included in the analysis; there was no clear association of this
stressor with impacted benthic condition as compared to unimpacted benthic condition,
as indicated in Table 3-22. The second type of analysis examines associations with
each stressor separately, with combination of any two stressors, and finally
associations between impacted benthic communities and all three stressors. This
second approach provides a description of the potential severity of stressor
associations based upon the number of stressors that co-occur. The third approach is a
step-wise linear regression analysis of the benthic index (condition of benthic
communities) with the major habitat and stressor variables.
Table 3-22. Association of Benthic Condition with Enriched Metal Concentrations in Sediments for the
Virginian Province.*
percent area with any enriched metal concentrations
Province
Large Systems
Small Systems
Tidal Rivers
entire province
49
42
64
69
impacted benthic
communities
56*
57
56
56
unimpacted benthic
communities
46
36
66
76
* Table is read as follows: fifty-six percent of the area of the Virginian Province with impacted benthic
communities also experienced enriched metal concentrations. The remaining 44% of the area with
impacted benthic communities did not have enriched metal concentrations.
3.5.2.1 Association with low dissolved oxygen
Association between benthic communities and bottom water dissolved oxygen
conditions are presented in Table 3-23, using both 2 and 5 ppm as criteria for the
association. Twenty-two percent of the impacted benthic communities in the province
61
-------
Table 3-23. Association of Benthic Condition with Bottom Dissolved Oxygen for the
Virginian Province.*
percent area with dissolved oxygen < 2 ppm
entire province
impacted benthic
communities
unimpacted benthic
communities
Province
Large Systems
Small Systems
Tidal Rivers
Province
Large Systems
Small Systems
Tidal Rivers
5
5
1
14
percent area
entire province
24
27
17
18
22*
30
3
33
0
0
0
0
with dissolved oxygen < 5 ppm
impacted benthic
communities
44
58
19
49
unimpacted benthic
communities
19
22
14
0
* Table is read as follows: twenty-two percent of the area of the Virginian Province with
impacted benthic communities also experienced bottom dissolved oxygen < 2 ppm. The
remaining 78% of the area with impacted benthic communities had bottom dissolved oxygen
> 2 ppm.
are associated with DO < 2 ppm. In the large and tidal river systems, however, 30-33%
of impacted areas are associated with DO < 2 ppm. Association with low DO in the
small estuarine systems was low, 3%. None of the unimpacted benthic communities
are associated with DO < 2 ppm in any resource class.
When 5 ppm is used as the criterion for the association, almost half of the
impacted benthic communities for the province (44%) are associated with DO < 5 ppm.
Fifty-eight percent of the impacted benthic communities in the large system are
associated with DO < 5 ppm, 19% of the small systems, and 49% of the tidal rivers.
Approximately one-fifth of the unimpacted benthic communities in the province and
large estuarine systems are associated with DO < 5 ppm Only 14% of the unimpacted
benthic communities in small estuarine systems are associated with DO < 5 ppm. None
of the unimpacted tidal river systems are associated with DO < 5 ppm. These patterns
for unimpacted benthos are consistent with the distribution of bottom DO < 5 ppm
across the entire province, with the exception of the large tidal rivers.
3.5.2.2 Association with sediment toxicity
Associations between benthic communities and sediment toxicity are presented
in Table 3-24, using two criteria: survival < 60% and < 80%. Associations using survival
62
-------
Table 3-24. Areal Estimates for Association of Benthic Condition with Sediment Toxicity for
the Virginian Province.*
percent area with amphipod survival < 60%
Province
Large Systems
Small Systems
Tidal Rivers
entire province
1
1
3
0
impacted benthic
communities
3*
3
4
2
unimpacted benthic
communities
1
1
3
0
percent area with amphipod survival < 80%
Province
Large Systems
Small Systems
Tidal Rivers
entire province
9
9
12
4
impacted benthic
communities
16
12
25
7
unimpacted benthic
communities
7
8
6
2
* Table is read as follows: three percent of the area of the Virginian Province with impacted
benthic communities also experienced amphipod survival < 60%. The remaining 97% of the area
with impacted benthic communities had amphipod survival > 60%.
< 60% as the criterion are minimal, with 3% for the province and ranging from 2% in
tidal river systems to 4% in small systems. There are no clear distinctions between the
associations for impacted and unimpacted benthic communities using survival < 60%
as the criterion for sediment toxicity.
Using the criterion of survival < 80%, 16% of the impacted benthic communities
in the province is associated with sediment toxicity, 25% in small systems, 7% in tidal
rivers, and 12% in large systems. Sediment toxicity (< 80% survival) co-occurs with
unimpacted benthos for 7% of the province, 6-8%of small and large estuarine systems,
and 2% in tidal rivers.
3.5.2.3 Association with ER-M exceedence
Association between benthic community condition and sediment contaminant
concentrations (ER-M exceedence for any metal or any organic compound) are
presented in Table 3-25. For ER-M metal exceedence, one-fifth of the impacted
benthic communities in small systems (20%) are associated with any sediment
concentration exceeding a metal ER-M value. The remainder of associations with ER-
M exceedence for metals are small. Values for large systems and tidal rivers are similar
across the entire province and for the unimpacted benthic communities.
63
-------
Table 3-25. Areal Estimates for Association of Benthic Condition with ER-M Exceedence
in Sediments for the Virginian Province.*
percent area with any ER-M metal exceedence
Province
Large Systems
Small Systems
Tidal Rivers
entire province
5
3
9
7
impacted benthic
communities
7*
0
20
3
unimpacted benthic
communities
3
3
2
6
percent area with any ER-M organics exceedence
Province
Large Systems
Small Systems
Tidal Rivers
entire province
3
2
2
10
impacted benthic
communities
5
0
5
14
unimpacted benthic
communities
3
3
1
9
* Table is read as follows: seven percent of the area of the Virginian Province with impacted
benthic communities also experienced ER-M metal exceedence. The remaining 93% of the
area with impacted benthic communities did not have any ER-M metal exceedence
There is little association between the condition of benthic communities and BR-
IM exceedence for organic compounds. While 14% of the tidal river area for impacted
benthic communities are associated with ER-M organics exceedence, this is not much
different from the values for the entire province and for the unimpacted benthos.
3.5.2.4 Association with multiple stressors
Analyses of associations between multiple stressors and impacted benthic
communities are summarized in Table 3-26 for the entire province. Note that because
of the restriction that valid data had to exist for all of the stressors at the sampling site
for associations to be conducted, slight differences exist between tables in this section
and Tables 3-23 to 3-25. Twenty-five percent of the Virginian Province has impacted
benthos, of which 39% co-occurs with low dissolved oxygen only (DO < 5 ppm), 6% co-
occurs with sediment toxicity only, 3% co-occurs with any ER-M exceedence only, 4%
with any two stressor indicators, 1% with all three, and 46% is not associated with any
of the stressors. These data indicate that low dissolved oxygen, sediment toxicity, and
ER-M exceedence together may account for 54% of the impacted benthic area. The
remaining 46% of impacted benthos is not associated with any of the three stressor
indicators. Similar analyses using dissolved oxygen < 2 ppm indicates that 60% of the
impacted benthic area is not associated with any of the three stressors (Table 3-26).
64
-------
Table 3-26. Co-occurrence of Stressors with Province-wide Impacted Benthic Communities (percent of
impacted area). Entries are for only one stressor, any two stressors, or all three stressors.
(a) Dissolved Oxygen Criterion of 5 ppm. (b) Dissolved Oxygen Criterion of 2 ppm.
__
only do only tox only erm do+tox do+erm tox+erm do+tox+erm other
39
<1
<1
1
46
do: DO < 5 ppm
tox: survival < 80%
erm: exceed any ERM
(b)
only do
only tox only erm do+tox
do+erm
tox+erm do+tox+erm other
25
4
<1
0
60
do: DO < 2 ppm
tox: survival!80%
erm: exceed any ERM
Nineteen percent of the total large estuarine area has impacted benthos, 56% of
which co-occurs with low dissolved oxygen only (DO < 5 ppm for the criteria), 7% co-
occurs with sediment toxicity only, no co-occurrence with any ER-M exceedence only,
4% co-occurs with two of the stressors, no co-occurrence with all three of the stressors,
and 33% is associated with none of the three stressors (Table 3-27). These data
suggest that low dissolved oxygen is the principal stressor of concern in large estuarine
systems and that sediment toxicity and ER-M exceedence are of less importance.
Small estuarine systems present a somewhat different picture with 37% of their
total area exhibiting impacted benthos, of which 6% co-occurs with low dissolved
oxygen (using DO < 5 ppm for the criteria), 5% co-occurs with sediment toxicity, 11%
with any ER-M exceedence, 5% co-occurs with any two, 4% with all three of the
stressors, and 69% is not associated with any of the three stressors (Table 3-27).
These data suggest that sediment toxicity and ER-M exceedence are the principal
stressors of concern in small estuarine systems, but there is a large unexplained
portion.
Tidal rivers have the highest percent area (38%) of impacted benthos of the
three resource classes of which 45% co-occurs with low dissolved oxygen and 6% co-
occurs with sediment toxicity. Unlike the small estuarine systems, none of the impacted
benthic area co-occurs with any ER-M exceedence, 2% co-occurs with any two of the
65
-------
Table 3-27. Cooccurrence of Stressors with Resource Class Impacted Benthic Areas (percent of impacted area).
Dissolved Oxygen Criterion of 5 ppm.
Large Systems
Small Systems
Tidal Rivers
only do
56
6
45
only tox
7
5
6
only erm
0
11
0
do+tox
4
4
0
do+erm
0
<1
2
tox+errn do+tox+erm other
0
<1
0
0
4
<1
33
69
46
do: DO < 5 ppm
tox: survival< 80%
erm: exceed any ERM
stressors, and < 1% co-occurs with all three (Table 3-27). The remaining 46% of the
impacted benthic area in the tidal rivers is not associated with any of the three stressor
indicators. Low dissolved oxygen is of major concern in the tidal river systems;
sediment toxicity and ER-M exceedence are less important.
Similar associations occur for the three resource classes using the dissolved
oxygen criteria of < 2 ppm (Table 3-28). The major exception is for small systems with
no co-occurrence of benthic impacts and DO < 2 ppm. This implies that very low DO is
of limited concern in small systems, and that a larger portion of the benthic impact is
associated with toxic contaminants. However, there is a large portion unexplained by
the three measured stressors.
3.5.2.5 Regression of the benthic index with individual variables
A step-wise linear regression analysis was conducted for the benthic index
against the major habitat and stressor variables for the all of Virginian Province
sampling sites (Table 3-29). These results indicate that 32% of the variance in the
benthic index can be explained with the following variables: TOG content of sediments,
Table 3-28. Co-occurrance of Stressors with Resource Class Impacted Benthic Areas (percent of impacted
area). Dissolved Oxygen Criterion of 2 ppm.
Large Systems
Small Systems
Tidal Rivers
only do
37
0
32
only tox
11
7
6
only erm
0
12
2
do+tox
0
1
0
do+erm
0
0
0
tox+erni
0
5
<1
do+tox+erm other
0
0
0
52
75
59
do: DO < 2 ppm
tox: survival< 80%
erm: exceed any ERM
66
-------
Table 3-29. Results of Stepwise Regression for Benthic Index Against Habitat and Stressor Indicators.
Grouping
Province
Large Systems
Significant Variables
TOO, DOSATB, BDO, NAPH,
PHENANTH, TPCB, TDDT,
STRATI, BSAL, CR, CD
CR, NAPH, TOG
Model r2
0.321
0.256
Small Systems
AS, CLARITY, TDDT, CD
0.401
Tidal Rivers
DOSATB, BDO, TPCB
0.269
dissolved oxygen saturation at the bottom, bottom dissolved oxygen, naphthalene,
phenanthene, total PCBs, total DDT, stratification, bottom water salinity, chromium, and
cadmium. Similar variance could be accounted for in large systems, small systems, and
tidal rivers (25%, 40%, and 27%, respectively); however, these variances were
explained with three or four variables each. These variables were chromium,
naphthalene, and TOC content of sediments in large systems, arsenic, water clarity,
total DDT, and cadmium in small systems, and dissolved oxygen saturation at the
bottom, bottom dissolved oxygen, and total PCBs in tidal rivers. Conducting a principal
component analysis on individual sediment contaminant concentrations, and then using
the results to represent sediment chemistry in the regression, did not explain additional
variance.
67
-------
SECTION 4: DISCUSSION
The prior section presented results from assessments of the ecological condition
indicators and of associations between stressors and benthic condition. These results
are discussed here in the context of answers to a series of environmental management
questions. The format is designed to aid managers in the interpretation of the results
presented in the prior section. The general questions were developed from intensive
interactions that have transpired between the EMAP-Estuaries staff and environmental
managers across the Virginian Province. These interactions included workshops
(USEPA 1991; Cochran 1991; Queen etal. 1992) and individual presentations for
environmental management groups, including each of the individual EPA National
Estuary Programs across the Virginian Province. These questions are:
1. Is there a problem in the estuarine waters of the Virginian Province?
2. If there is a problem in the estuarine waters of the Virginian Province, what is
the magnitude, extent, and distribution of the problem?
3. What factors are associated with the observed problems in the estuarine
waters of the Virginian Province?
4. Are the observed problems in the estuarine waters of the Virginian Province
consistent with existing knowledge?
5. What are the uncertainties associated with the conclusions?
6. How effective was the EMAP Virginian Province Demonstration Project in
meeting the program objectives?
4.1 Is there a problem?
The information presented in the prior chapter indicates that the estuarine waters
of the Virginian Province are having problems. Because no index of overall estuarine
quality exists, the suite of indicators reported identify problems with different
components of the estuarine systems. A summary of these problems follows.
A benthic index was developed and refined to determine the condition of the
benthic communities in the estuarine waters of the Virginian Province. Impacted
benthic communities were observed in 25+3% of the province area for the four-year
period (1990-93). Of the three resource classes, the large estuarine systems had the
smallest percent area with impacted benthic communities, 19±4% for the four-year
period. The percent area with impacted benthic resources was markedly larger in the
small estuarine systems and tidal rivers, 37+6% and 38±14%, respectively, for the four-
68
-------
year period.
The percent area of bottom waters in the Virginian Province with low dissolved
oxygen conditions (DO < 2 ppm) was 5+2% for the four-year period (1990-93). Small
estuarine systems had only 1+1%, large systems had 5±2%, and tidal rivers 14±6%.
These data suggest that the tidal rivers are most at risk from low dissolved oxygen.
Using moderate hypoxia as the criterion (DO < 5 ppm), impacted conditions were
observed in 24±3% of the province area, 17+5% of the small systems, 27+4% of the
large systems, and 18±7% of the tidal rivers. These analyses suggest that large
estuarine systems are potentially at risk from moderate reductions in dissolved oxygen.
The percent area of the Virginian Province sediments having moderate sediment
toxicity (survival < 80%) was 9±2% for the four-year period (1990-93). Examination of
individual resource classes showed that the percent area exhibiting sediment toxicity
was 4±4% of the tidal rivers, 9±3% of the large systems, and 12±6% of the small
systems. These analyses indicate that small estuarine systems are at greatest risk from
toxic sediments. Severe toxicity (survival < 60%) occurred in only 1 % of the estuarine
sediments of the province and was distributed across resource classes.
Using ER-M exceedence as a measure of sediment contamination, 5+1% of the
area of the province during the four year sampling interval had observed exceedence in
at least one sediment concentration for metals for which ER-Ms exist. Sediments in
large estuarine systems exhibited exceedence in 3±3% of the area. Seven percent of
the area (±4%) in tidal rivers contained sediments with ER-M exceedence. Small
estuarine systems exhibited the highest proportion of area, 9+4%, with ER-M
exceedence. These ER-M metal exceedence suggest that small estuarine systems are
at greater risk than are large or tidal river systems.
Sediment concentrations for organic contaminants indicated that 3+1% of the
area of the Virginian Province exhibited exceedence of at least one ER-M value. The
proportional extent of organic contamination was low in large and small estuarine
systems (2%) and higher in tidal rivers (10±2%). Tidal rivers appear to be at the
greatest risk due to organic contamination.
Using ER-L values for determining which observed sediment contaminant
concentrations are below levels for which biological effects may be of concern, half of
the entire province (50+4%) had sediment contaminant concentrations below ER-L
levels. The large estuarine system class exhibited the largest areal extent of sediments
below ER-Ls, 58+5%, with small systems and tidal rivers exhibiting 35+6% and
25±12%, respectively. These results indicate that sediments of the small systems and
tidal rivers are more at risk to sediment contaminant levels that have been observed to
potentially elicit biological responses.
69
-------
Overall, 52+5% of the estuarine waters of the Virginian Province were in good
condition, i.e., these waters exhibited unimpacted benthic conditions and bottom
dissolved oxygen > 5 ppm and sediment toxicity acute survival > 80% and no ER-M
exceedence for sediment contaminants. Small estuarine systems had 55+14% of the
area in good condition, tidal rivers had 52±13%, and large systems 51+5%.
4.2 What is the magnitude, extent, and distribution of the problem?
i
Chesapeake Bay, which exhibited benthic community impacts in 23+5% of the
area, accounted for 45% of the impacted benthic area within the Virginian Province.
The Potomac and Rappahannock Rivers have identical percent impacted areas,
44+22% and 44+33%, respectively, with the James River having a slightly smaller
percent area of benthic impact (19+23%). Delaware Bay and Long Island Sound
exhibited benthic impacts over 24+12% and 28±11% of their areas, respectively. The
percent areal extent of benthic impact in the Hudson-Raritan system, however, was the
greatest of the major estuarine systems examined (72+8%). Together, these four
estuarine systems account for 79% of the impacted benthic area within the Virginian
Province.
Of the four major estuarine systems, Chesapeake Bay was the most impacted
(10+3%) from low dissolved oxygen (DO < 2 ppm), with the Potomac and
Rappahannock Rivers particularly impacted (24±12% and 15+11%, respectively).
Using moderate hypoxia as the criterion (DO < 5 ppm), approximately 31% of
Chesapeake Bay and 48% of Long Island Sound areas have moderate hypoxia. The
moderate hypoxia in Long Island Sound was in the open water area of the western
Sound.
Moderate sediment toxicity (survival < 80%) was observed in 15+14% of the
Hudson-Raritan system, 13±7% of Long Island Sound, 6±3% of Chesapeake Bay, and
2+2% of Delaware Bay. Moderate toxicity was observed across all of the estuarine
classes of these estuarine systems, except Delaware Bay, where toxicity was only
observed in the Delaware River. Severe toxicity (survival < 60%) was observed in
Delaware Bay (1±2%), Hudson-Raritan system (7±10%), and Long Island Sound
(5+4%). This toxicity was found primarily in small systems in the Hudson-Raritan
system and Long Island Sound, and only in the tidal river portion of Delaware Bay.
Using ER-M exceedence as a measure of sediment contamination, the Hudson-
Raritan Estuary is the most contaminated major system, exhibiting 44+14% of the area
with sediments exceeding at least one ER-M value. Most of this contamination was due
to mercury, lead, and silver in the tidal portion of the Hudson River (30% of the area).
The least contaminated system was Delaware Bay where only 1+3% of the area
contained sediments exceeding ER-M values. Chesapeake Bay and Long Island Sound
were intermediate in level of contamination with percent area of exceedence of 5+2%
70
-------
and 4+4%, respectively. The exceedences in Long Island Sound were predominately
found in small systems bordering the open waters of the Sound.
Sediment concentrations for organic contaminants indicated that Delaware Bay
system was the least contaminated of the four major systems. Less than 1% of its area
contained sediments exceeding organic ER-M values, and those sediments were
restricted to the Delaware River. A similarly low level of contamination was found in
Chesapeake Bay and Long Island Sound; 3-4% of their areas contained sediments with
organics exceeding the ER-M values. The greatest level of organic contamination was
found in the Hudson-Raritan (mostly PCBs): 33+7% of the area of this system
contained sediments with organic concentrations that exceeded at least one ER-M
value. Sixty-six percent (±18%) of the area in the tidal portion of the Hudson River, and
11+5% of the area in the small systems exhibited ER-M exceedence.
Almost none of the sediments in the Hudson-Raritan system (1+3%) were
observed to have contaminant concentrations below ER-L levels. In other words,
almost the entire Hudson-Raritan system has sediments above levels that have been
observed to potentially elicit biological responses. In contrast, half or more of
Chesapeake Bay (50+5%) and Delaware Bay (62+14%) were observed to have
sediment concentration levels below ER-Ls. Long Island Sound exhibited 24+12% of
the sediments to be below ER-L levels. Within Chesapeake and Delaware Bays, the
large system class exhibited more sediment area below ER-L levels compared with the
small systems and tidal rivers. In contrast, Long Island Sound results indicated more
areal extent of sediments below ER-L levels in small systems than the large system
class.
4.3 What factors are associated with the observed problems?
Analysis of associations was conducted between the benthic community index
and both stressor and habitat indicators to provide possible explanations for the
observed condition of benthic resources. Analysis of benthic communities and habitat
indicators show that impacted benthic communities tend to be associated with muddy
(> 80% silt-clay), moderate TOC content (1-3%) sediments, and polyhaline bottom
waters (> 18 o/oo). In contrast, unimpacted benthic communities tend to associate with
sandy (< 20% silt-clay), low TOC content (< 1%) sediments, and with polyhaline bottom
waters (> 18 o/oo).
Association analysis of benthic communities with stressors for the entire province
indicates that moderate hypoxia (DO < 5 ppm), sediment toxicity, and sediment
contamination (ER-M exceedence) together account for 54% of the impacted benthic
area. The remaining 46% of impacted benthos is not associated with any of the three
stressor indicators. Low dissolved oxygen is the principal stressor of concern in large
systems and tidal rivers, being associated with 58% and 49% of the impacted benthos,
71
-------
respectively. However, in small estuarine systems, sediment toxicity and ER-M
exceedence are the principal stressors of concern, accounting for approximately 26% of
the observed impacted benthos. Similar analysis with low dissolved oxygen (< 2.0
ppm) indicates that 60% of the entire province was not associated with this suite of
stressors.
4.4 Are the observed problems consistent with existing knowledge?
The only province-wide consistent data set that could be used for comparison
with the EMAP Virginian Province results is the National Oceanographic and
Atmospheric Administration (NOAA) National Status and Trends (NS&T) Program for
Marine Environmental Quality (O'Connor, 1990; NOAA, 1992). NS&T collected
sediment samples at least one time at the 32 mussel watch sites in the Virginian
Province, and analyzed for the same suite of contaminants as was done for the EMAP
Virginian Province (Table 3-9). Restrictions on comparing EMAP data with the NS&T
data include (1) most of the sediments were collected prior to 1990 (but since 1986)
and (2) sampling sites were restricted to habitats suitable for oysters and mussels
(shallow water, salinity greater than 10 o/oo). Daskalakis and O'Connor (1994) report
on an inventory of coastal U.S. sediment contamination, compiled from various
electronic information systems, including NS&T and EMAP (1990-91 data from
Virginian Province and 1991-92 data from Louisianian Province). They observe that the
EMAP probability-based design produces a data set that contains the lowest proportion
of sites with "High" concentration for any chemical (defined as exceeding one standard
deviation above the mean of the NS&T sites), compared with the other data sets
inventoried. They reported that the frequency of "Highs" for the NS&T sites exceed
those for the EMAP sites because there is an urban bias in the location of NS&T sites
(Cantillo and O'Connor, 1992). Daskalakis and O'Connor (1994) further observe that all
data sets other than EMAP contain sites preselected (biased) for their likelihood to have
elevated chemical concentrations. They conclude that extremely high sites were
located near large cities, suggesting anthropogenic sources for the contaminants, and
that smaller water bodies with high human activity and high mean residence times bear
the majority of the pollution. These conclusions are consistent with the EMAP results
that the concentrations of ER-M exceedence occur around population centers (Figure
3-10 and 3-11) and that small systems and tidal rivers are at higher risk from chemical
contamination (Section 3.4.1).
Benthic community condition is available for comparison in Chesapeake Bay
system (USEPA, 1995a; Ranasinghe et a/., 1994), where the benthic information was
analyzed through use of habitat restoration goals (Ranasinghe et a/., 1993; Weisberg et
a/., 1997). This approach specifies habitat-specific measures that describe
characteristics of benthic assemblages expected to occur at sites with little evidence of
environmental stress or disturbance. It has a similar goal to the EMAP benthic index,
which is to aggregate the diversity of benthic information into a single quantifiable
72
-------
value. Figure 4-1 is reproduced from the 1995 State of Chesapeake Bay Report
(USEPA, 1995a) and is compared with Chesapeake Bay portion of Figure 3-2. The two
approaches for determining benthic condition depict the same general pattern for
benthic community impact. More detailed comparison of these approaches is currently
being undertaken.
The distribution of sediment contamination is available for three of the major
estuarine systems examined in this report. Figure 4-2 presents results of the analysis
on toxic stress for Chesapeake Bay (USEPA, 1995a), which could be compared with
Chesapeake Bay portion of Figures 3-10 and 3-11. The two programs show a similar
distribution of potential impact due to pollutants. Results for Delaware Bay are available
from the Comprehensive Conservation and Management Plan (USEPA, 1996). This
report indicates that the highest concentrations of toxic substances occur in the
urbanized area along the Delaware River. Elevated metals in bottom sediments are
associated with fine, organic-rich particles, particularly near municipalities and in the
central area of the Bay. Lead, zinc, cadmium, pesticides and some of the PAHs
exceed the Long and Morgan ER-M values. Acute sediment toxicity was concentrated
along industrialized portions of the Delaware River. These observations are consistent
with the results for the Delaware Bay discussed in Section 3.4.2 and depicted in
Figures 3-8, 3-10, and 3-11. The Long Island Sound Study reported that "most of the
Sound's sediments do not exhibit contamination levels of concern, problems have been
documented in some areas of the western Sound and in several, mostly urbanized,
harbors, rivers, and embayments" (USEPA, 1994b). Sediment contamination in Long
Island Sound (LIS) as observed by NOAA NS&T was summarized by Robertson et al.
(1991). They indicate that the NS&T sites in LIS (located around the perimeter of the
Sound) tend to have relatively high concentrations for both metals and organic
contaminants in sediments when compared to the reference set of NS&T sites.
However, in a national comparison, two areas (near New York City and Boston) clearly
exhibited highly ranked contaminant concentrations more frequently than did LIS.
These reported observations on sediment contamination in LIS are consistent with the
EMAP results (Section 3.4.2) indicating potential toxic impact localized in the western
end of the Sound and in the highly urbanized small systems around the Sound, and that
LIS ranks below the Hudson-Raritan system in contaminant impact.
Information is available for dissolved oxygen (DO) conditions in Chesapeake Bay
(Ranasinghe et al., 1994) and Long Island Sound (USEPA, 1994b). Results from 1995
State of Chesapeake Bay Report (USEPA, 1995a) are presented in Figure 4-3, and are
compared with Chesapeake Bay portion of Figure 3-5. The results from the two
programs indicate similar patterns for DO stress in the Bay. Chesapeake Bay Program
DO data were collected from routine cruises conducted twenty times per year (monthly
from November to February and twice monthly from March to October) at approximately
140 stations (Heasly et al., 1989), while the EMAP data resulted from 150 probability-
based sampling sites, each sampled once during the 1990-93 summer index period.
73
-------
Susquehanna
Legend
Q Meets Goals
• Degraded
H Severely De^aded
O Not Evaluated
Potomac
Rappahannock
Figure 4-1. Benthic community condition in Chesapeake Bay as reported by Chesapeake Bay
Program (from USEPA, 1995a). Ratings are based on average "Restoration Goals Index"
scores for summer benthic community sample taken between 1984 and 1990 (refer to
USEPA (1995a) for details).
74
-------
Legend
Very Low Risk
Possible Risk
Significant Risk
Figure 4-2. Sediment contamination and risk to aquatic life in Chesapeake Bay as reported by
Chesapeake Bay Program (from USEPA, 1995a). Ranking of risk to aquatic biota from
sediment contamination was based on comparisons of the average sediment
concentrations of seven trace metals and eight PAHs to sediment quality guidelines (refer
to USEPA (1995a) for details).
75
-------
Bush
Gunpowder
Backt
Patapsco1
Elk: Run
7
Bohemia
Sassafras
Potomac
Rapp ahannodc
Status Key
• Sever ly Imp acted
EPoor
H Stressed
HFair
D Good
Ghoptsnk
Nanticoke
Wioomlco
Focomoke
Jaines
Figure 4-3. Bottom dissolved oxygen in Chesapeake Bay as reported by Chesapeake Bay Program
(from USEPA, 1995a). Dissolved oxygen levels in bottom waters were evaluated relative
to target concentrations required to support the growth, reproduction, survival of the Bayjs
fish, shellfish, and bottom dwelling organisms. A three level scale was used with "Good"
areas having among the highest bottom dissolved oxygen concentrations and "Severely
Impacted" areas having among the lowest dissolved oxygen concentrations (refer to
USEPA (1995a) for details).
76
-------
The Long Island Sound Study (LISS) reported that moderate DO stress (< 5
ppm) was observed during the summers of 1987-93 in about one-half to two-thirds of
the Sound. The distribution of low DO area encompassed the western portion of the
Sound to approximately New Haven (USEPA, 1994b). This is consistent with the EMAP
results for exposure to moderate DO stress areal extent (48+12% of the area during
1990-93) and distribution (Figure 3-5). The LISS reported numerous occurrences of
low DO stress (< 2 ppm) compared with only one observation at one station by EMAP.
This difference can be explained by the small observed percent area with DO < 2 ppm
in the extreme western end of the Sound (approximately 100 km2 or 3% of the Sound),
and the spatial density of EMAP probability-based sampling sites. Spatial enhancement
of the probability-based sampling sites in the western Sound would be necessary for
EMAP to better resolve the extent of low DO stress in LIS.
These comparisons of results from the EMAP Virginian Province Four-Year
Demonstration Project with published results from more intensively sampled programs
indicate that the EMAP approach can provide comparable accuracy for description of
the extent and distribution of estuarine impact. The EMAP approach was able to
quantify the magnitude of impacted conditions with estimates of uncertainty, which is
something other monitoring programs were not always capable of accomplishing.
4.5 What are the uncertainties?
All of the percent area values reported here have included an estimate of the
uncertainty of the value. In Section 3, reported values for individual years and the entire
four-year period were presented for the entire Province, individual estuarine classes,
and the four major estuarine system. Values for the three major tidal river systems
within Chesapeake Bay were only presented for the entire four-year period. Values for
the individual classes within the major estuarine systems and for the individual years in
the tidal rivers were not reported because of the small sample sizes (number of
sampling sites available to make estimates). A small sample size results in large
uncertainty about the mean value and can result in large year-to-year variability in the
mean value. This is illustrated in Table 4-1, which expands upon the entries provided in
Table 3-2 for impacted benthic communities. For example, uncertainty estimates for
estuarine classes within Chesapeake Bay for the entire four-year period are reasonably
small (less than ±10%, except for tidal rivers), but uncertainty for an individual year can
be much larger. The individual year mean estimates for the large system class within
Chesapeake Bay range from 5 to 33%, and the four-year uncertainty estimates for the
estuarine classes in Delaware Bay and Hudson-Raritan system are large (greater than
±16%).
This discussion on Table 4-1 points out the discretion that needs to be employed
for estimates of condition when sample sizes are small. This is not a problem with the
sampling design employed; rather it illustrates the practical limit imposed by trying to
77
-------
Table 4-1. Expansion of Table 3-2 to Include Estuarine Classes for Impacted Benthic Communities to
Illustrate Effect of Small Sample Size on the Mean and Uncertainly Estimates. Values Are
Mean Estimate and 95% Confidence Interval (C.I.)- N is Number of Sample Sites Available
for Four-year Estimates; Sample Sites for Individual Year Is Approximately N/4.
1990
% C.I.
Chesapeake Bay
Large Systems
Small Systems
Tidal Rivers
Area
33
14
69
57
10
27
19
33
1991
% C.I.
Area
20
5
43
44
11
10
28
44
1992
% C.I.
Area
22
24
25
12
9
24
14
15
1993
% C.I.
Area
27
33
3
32
11
19
8
38
1990-1993
% C.I. N
Area
23
19
26
36
5
6
8
17
195
85
52
58
Potomac River
Rappahannock River
James River
Delaware Bay
Large Systems
Small Systems
Tidal Rivers
Hudson-Raritan
Small Systems
Tidal Rivers
Long Island Sound
Large Systems
Small Systems
85
61
8
12
*
100
88
78
100
45
30
25
80
45
45
45
4
*
t
37
13
t
33
22
38
16
63
30
10
13
14
*
8
59
100
*
39
33
100
58
60
57
23
29
*
46
t
t
*
24
33
t
*
67
24
26
17
*
100
83
100
59
39
33
100
*
68
32
24
33
*
*
19
t
47
25
33
*
37
17
17
44
40
*
82
70
100
25
12
13
5
48
83
50
32
40
*
60
20
t
50
17
25
13
44
44
19
24
18
52
70
72
100
32
28
26
51
22
33
23
12
17
22
21
8
t
19
12
12
12
20
18
20
47
24
6
17
31
11
20
53
38
15
* There can be a large uncertainty due to the small number of sampling sites for an individual year in a
specific tidal river. Therefore, estimates for the individual years are not presented.
t Due to assumptions in the estimation procedures, if one resource class entirely exceeds the criterion
and other resource classes have no exceedences, the C.I. becomes zero.
t Due to assumptions in the estimation procedures, if the percent area of exceedence of a criterion is
either zero or 100%, then the C.I. is zero
make percent area estimates for which the sampling density (number of samples) is not
adequate for the question posed. If the sampling design were intensified for the areas
where the uncertainty estimates were unacceptable, then reduced uncertainties would
result. This is illustrated in Table 4-1 by the reduced uncertainty for the four-year
78
-------
estimates compared with the individual year estimates.
There are other sources of uncertainty in the conduct of assessments which can
not be mathematically expressed. These other sources of uncertainty are threefold: the
inherent randomness of events (stochasticity), imperfect or incomplete knowledge of
things that could be known (ignorance), and mistakes in the execution of assessment
activities (error). Sources of uncertainty discussed are those associated with the
sampling design and the indicators.
Advantages of the EMAP sampling design are: 1) it is systematic and probability
based; 2) it quantifies the areal extent associated with an indicator value; 3) it is
scalable to regions, watersheds, and local sites; 4) it permits estimation of uncertainties
for indicator values; and 5) it provides spatially explicit patterns and distributions of
ecological resources and associated habitat and stressor indicators. There are several
design issues, however, that contribute to the uncertainties in the assessments. First,
sampling is limited to a temporal index period (e.g., once per year) which does not
address intra-annual, seasonal, and episodic events that may have a long-term impact
on resources. For example, one possible contribution to the unexplained portion of an
association is the occurrence of a significant stress to the benthos that occurred outside
the index period, and therefore went undetected. Second, the techniques of
incorporating non-probabilistic extant data with the EMAP probabilistic sampling design
remain to be developed and tested. Overton et al. (1993) have proposed one approach
for this, but this has not been actually tested. Although this limits our ability to combine
extant with EMAP data, it does not prevent the use of extant data to support or
supplement EMAP data, such as was done in the prior section addressing how EMAP
results compare with existing knowledge. The ability to supplement EMAP indicator
data, as well as increase the total sample size of analyses, may contribute to
significantly reducing overall uncertainty in the analyses. Finally, there is uncertainty
associated with the sample density at different scales (number of sample points) and
the associated uncertainty related to the absolute number of samples, as has been
noted in Section 3 when confidence intervals are discussed.
Several contributions were made by EMAP to indicators for monitoring: (1) the
use of suites of stressor and habitat indicators that are simultaneously measured with
the condition indicators, (2) clear and unequivocal links between endpoints and their
metrics, and (3) habitat indicators that are directly related to and facilitate the
interpretation of both the condition and stressor indicator information. There are,
however, several indicator issues that contribute to the uncertainty in these analyses,
as illustrated by the large unexplained portion of the associations. First, there is the
assumption that the suite of stressor indicator used in these analyses are directly
related to and sufficient to discriminate between anthropogenic and natural stresses.
The absence of stressor information from indicators specific to nutrient or carbon
enrichment and physical stress can be postulated as a potential source of uncertainty.
79
-------
This is particularly applicable in small estuarine systems and tidal rivers that are likely to
experience both high nutrient and carbon loads as well are repeated physical
perturbation of the benthos through dredging and shipping.
There are also uncertainties associated with existing indicators. The benthic
index metric requires further testing and validation relative to specific habitat indicators
(e.g., salinity, grain size, etc.). Also, the only biotic condition indicator is for benthos.
The condition of the habitats such as the pelagic zone, marshes, and submerged
aquatic vegetation has not been addressed. There are a series of implicit assumptions
relative to the stressor indicators that need to be tested. First, sediment toxicity and
sediment chemistry should be re-examined from the perspective of bioavailability.
Second, the assumption that toxicity tests are an accurate surrogate for community
exposure, and that exposure is accurately coupled with relevant measures of
contaminant availability needs to be rigorously evaluated. The interpretation of
sediment contamination data and the combination of sediment toxicity, habitat
indicators, and sediment contaminant data into a complex exposure index must be
addressed. Currently there is uncertainty regarding the functional relationship between
condition and stressor indicators that may limit their predictive value.
Finally, uncertainties are associated with the indicator for dissolved oxygen.
Sampling uncertainty is related to the use of a single point estimate taken during the
index period as a representative measure of benthic exposure when the tidal and intra-
annual variability are not considered. Analyses of the 1990 data did evaluate dissolved
oxygen results for different sampling windows within the index period (Weisberg et a/.,
1993). It was observed that the province-wide cumulative distribution function for
bottom dissolved oxygen was stable across sampling windows. However, the capability
to characterize the DO condition for a site from a single point-in-time measurement was
limited. This becomes important when conducting associations. The ability to
discriminate natural from anthropogenic sources of low dissolved oxygen is another
source of uncertainty. This problem is important in the Virginian Province because
most of the hypoxic conditions occur in Chesapeake Bay where there is considerable
debate over the causes and sources of the problem (Officer ef a/., 1984).
One overall concern with the existing suite of indicators is the large unexplained
portion of impacted benthic conditions when associations are made .with stressors. This
large unexplained portion may be due to not including relevant indicators of important
stress on the benthic community. It may also be due to how and when we actually
measured the indicators; e.g., significant stress may have occurred outside the index
period and went undetected with the existing indicators. These are topics for future
research.
Although the interpretation of data on single indicators can be improved through
further refinement, taken together the suite of indicators used in the EMAP Virginian
80
-------
Province Demonstration Project does provide a qualitative weight-of-evidence approach
to understanding the status and condition of estuarine resources.
4.6 How effective was the Four-Year EMAP Virginian Province Demonstration
Project in meeting the program objectives?
As stated in Section 1.1, the four objectives of the original EMAP were: (1)
estimate the current status, trends, and changes in selected indicators of the condition
of the nation's ecological resources on a regional basis with known confidence; (2)
estimate the geographic coverage and extent of the nation's ecological resources with
known confidence; (3) seek associations among selected indicators of natural and
anthropogenic stress and indicators of ecological condition; and (4) provide annual
statistical summaries and periodic assessments of the nation's ecological resources.
The prior discussion in this section, answering the environmental management
questions, clearly indicates that the four-year Virginian Province Demonstration Project
met the first and third objectives. The second objective was clearly met for the Virginian
Province Demonstration Project, since NOAA maps were used for delineation of the
estuarine boundaries (Holland, 1990). The success in producing the annual reports
(Weisberg et a/., 1993; Schimmel etal., 1994; Strobel etal., 1994; Strobel et a/., 1995)
demonstrates that the fourth objective has been achieved.
The development of program data quality objectives (DQOs) (Olsen, 1992) has
been proposed as a quantitative measure of evaluating success in achieving the first
objective of EMAP. The EMAP Quality Assurance Management Plan (Kirkland, 1992)
defines DQOs as "statements identifying the anticipated use of environmental data
leading to actions to be taken by EMAP and defining the level of uncertainty a decision
maker is willing to accept in the data supporting the decision (and action), expressed in
quantitative, statistical terms." The target DQO proposed for status estimates with
condition indicators was:
For each indicator of condition and resource class, on a regional scale, estimate
the proportion of the resource in degraded [or impacted] condition within 10%
(absolute) of the value with 90% confidence based on four years of sampling.
Olsen (1992) also proposed a target DQO for indicators of condition for trend
detection:
Over a decade, for each indicator of condition and resource class, on a regional
scale, detect a linear trend of 2% (absolute) per year, i.e., a 20% change for a
decade, in the percent of the resource class in degraded [or impacted] condition.
The test for trend will have a maximum significance level of a = 0.2 and a
minimum power of 0.7 (3 = 0.3).
81
-------
Addressing the trends DQO is beyond the intent of this report. The application of the
trends DQO to the Four-Year Virginian Province data is addressed in a separate
document (Heimbuch et a/., 1995b). Results suggest that the design as used in the
Virginian Province is capable of detecting a 2% change per year over a 12-year period
with a power greater than 0.80 for dissolved oxygen and benthic condition.
The information for evaluating the ability of the Four-Year Virginian Province
Demonstration Project to meet the status DQO is presented in Table 4-2, which
summarizes information previously presented in this report. Note that this table, in
addition to providing province-wide (regional scale) status estimates, includes the
estuarine systems that status estimates were made for. Clearly, the EMAP target DQO
for status has been met for almost all of the indicators for the entire province and the
three resource classes. The major exceptions were for ER-L exceedence and benthic
index in the tidal river class. These two indicators marginally exceed the target. This
higher uncertainty in the tidal river class estimates was noted in analysis of the 1990
Virginian Province data (Weisberg et a/., 1993).
Even though the program DQOs were developed for the province scale, it is
encouraging that the target DQO for status could also be met for most indicators in the
major estuarine system analyses. The consistent exceptions were the tidal rivers in
Chesapeake Bay (Potomac, James, and Rappahannock Rivers). The larger
uncertainties in these estimates were primarily due to the small number of samples that
were taken in these systems.
4.7 Summary and Conclusions
The four-year assessment of the EMAP-Estuaries program illustrates several
important contributions to environmental monitoring in the areas of sampling design and
indicator research. In particular, the sampling design is both systematic and
probabilistic in nature; is extremely flexible; provides areal estimates, with confidence,
of the condition of all indicators; is spatially explicit allowing for a variety of spatial
analyses; accommodates post-stratification of data; can be scaled to the problem
setting without losing its defining characteristics; and, most importantly, has
comparability, which permits the direct comparison of results across differing spatial
scales as well as between different categories/populations of resources (e.g., large and
small systems and tidal rivers).
The four-year assessment of Virginian Province data affords the opportunity to
evaluate the applicability, sufficiency, and effectiveness of EMAP's indicator program.
The multi-indicator approach used by EMAP has proven both practical and necessary.
Traditional monitoring programs often measure only one type of indicator, either
exposure or ecological. EMAP, however, by including the patterns of both natural and
82
-------
I
I
Q.
S
a
1 8
IS
to J2*
LU .£
o E
= ra
5 CD
CO
CD CO
CO j£?
i
§1
1°
§ I
I!
>~
co .
CD (0
3
O CO
2
CO CO
o
'co"
O
O co
* 8
in OT
?0
§g
S'S
s^
S UJ
•sr
_o>
J2
CO
1
w c 8
3 n
3&Z.
3
i|lf
**»-
I p
2 m z
Q ""
o
E o °
° ii z
Q. ""'
X
^
Is?
a^s.
I
b.
0)
O > II
s* Q!! ^
—i
j£ w -~
c ™ S "
wldi
&||P
™ 3 ^> "
LJ OT 5.
v>
Sz
Q. ~
Q
01
d
d
(0
:
O
0)
*!
— :
o
TO
d
^*
d
•~? S>
" <
d
ro
%p d>
f?
•^
d
CO
d
d
(0
*^
T-
oo
CM
oo
CN
f-
CM
s
CM
CM
5
CO
CO
•sr
CO
CN
CD
T-
m
CO
CN
5
OO
CO
CD
CO
•sr
O)
CO
in
CM
Impacted Benthos
(benthic index < 0)
^.
CO
-H-
0
•H-
O
O
CM
T—
CN
•^«
'r~
m
o
o
CO
o
co
•sr
T-
-
CM
m
CM
m
0.
CM
VI
0
Q
CM
CO
V
h-
^
CO
CM
v-
m
CM
0
CM
05
CO
o
^r
in
CO
h-
oo
T —
in
V-
•sr
CM
CO
s
DO < 5 ppm
•sr
m
o
r-
CM
-
•H-
O
O
•H-
O
O
-H-
0
O
•H-
O
0
•H-
O
O
^
CO
CM
"-
-
x-
Sediment Toxicity
(survival < 60%)
^_
CO
?
m
CM
CM
^_
V
-
^~
'r~
CD
X—
OO
CO
CO
•sr
•sr
CD
CM
CO
CO
CM
CO
Sediment Toxicity
(survival < 80%)
oo
CD
oo
T—
m
oo
•sr
05
CO
•sr
CO
in
CO
CO
•sr
in
CO
in
•sr
^
o
CO
CD
^
•sr
CD
m
3
•sr
CD
•sr
Enriched Metals
(for any metal)
^.
•sr
CO
CM
CO
-
++
0
o
•H-
O
O
O
CO
CO
in
•sr
r-
^
CO
CO
CO
-
in
ER-M Metals
exceedence for any)
CM
CM
•sr
•sr
•sr
CM
T—
V
•H-
O
O
CO
CM
CM
CM
•H-
O
O
CM
•sT
•sT
•sr
CM
CO
CO
CM
-
^.
ER-M Organics
exceedence for any)
co
CM
oo
co
CO
in
*
•H-
O
O
CO
CM
CM
CM
£
CM
co
in
CO
-*
r^
^
CM
CM
CM
in
PAH Normalized for
roc (> 200 ug/g-oc)
CM
in
CO
CO
T~
?
CM
CD
O
CN
CD
CO
•H-
O
O
O
CM
CO
CM
CO
O
in
CM
in
CM
CD
m
CO
in
CO
in
^
o
in
elow All ER-L (metal.
and organics)
CD
S.
CD
N
CO
o
(D
c:
CD
*•*
O
o
0
Q
CD
N
CD
'CD
CO
c
o
•c
o
CO
o
8
CD
T3
CD
ft
M—
o
CD
E
CO
c
CD
Q.
CD
JC,
~
p
Q
a.
c
o
CO
CD
CD
g
CO
c
Due to assumptio
4+
.83
-------
anthropogenic stressors and habitat factors provides information critical for forming
hypotheses that explain the observed ecological observations. There are limitations,
however, to the indicator program in the Virginian Province as is illustrated by the
discussion on uncertainties. First, only one ecological indicator, benthic community
condition, was available to be used to assess the status of ecological condition in the
Province. Second, there were no explicit stressor indicators for enrichment or for
physical perturbation. This assessment re-iterates the need for additional research on
ecological and stressor indicators to reduce uncertainties in the assessments.
The value of the EMAP design and indicator program is illustrated by its ability to
successfully identify and quantify the major environmental problems in the estuarine
waters of the Virginian Province. When the EMAP characterization of the Virginian
Province was compared with analyses from other existing monitoring and
environmental data from the states and federal agencies, the general conclusions were
the same. The agreement between conclusions drawn from EMAP and those from
existing data could be viewed as an initial validation of the EMAP concept. This is very
important because it provides confidence that when the EMAP design and indicators
are applied to data-poor environmental areas, they are likely to capture successfully the
major ecological problems. The added value of the EMAP design over many other
studies, of course, is the quantification with degree of uncertainty (confidence limits)
that is provided with the results.
Although uncertainties certainly remain, the results of the four-year Virginian
Province assessment are encouraging. The Demonstration Project clearly indicated
that the EMAP objectives were not only reasonable but were achievable with available
indicators collected with a probability-based sampling design. It was demonstrated that
the EMAP design can be used to quantify with confidence the status and condition of
ecological resources. Reducing the uncertainties in the assessment should be
approached through a systematic program of directed research.
84
-------
SECTIONS: REFERENCES
Bailey, R.G. 1983. Delineation of ecosystem regions. Environmental Management,
7:365-373.
Boesch, D.F. 1973. Classification and community structure of macrobenthos in the
Hampton Roads area, Virginia. Marine Biology, 21:226-244.
Boesch, D.F. and R. Rosenberg. 1981. Response to stress in marine benthic
communities. In: Stress Effects on Natural Ecosystems, G.W. Barret and R.
Rosenberg, eds., John Wiley and Sons, New York, pp. 179-200.
Breteler, R.J., K.J. Scott and S.P. Shepurd. 1989. Application of a new sediment toxicity
test using the marine amphipod Ampelisca abdita to San Francisco Bay
sediments. In: Aquatic Toxicology and Hazard Assessment, Volume 12. U.M.
Cowgill, and L.R. Williams, eds., ASTM.
Brongersma-Sanders, M. 1957. Mass mortality in the sea. Pages 941-1010. In: Treatise
on Marine Ecology and Paleoecology, J.W. Hedgpeth, ed., Memoirs, Geological
Society of America, 67.
Broutman, M.A. and D.L. Leonard. 1988. National estuarine inventory: The quality of
the shellfish growing waters in the Gulf of Mexico. National Oceanic and
Atmospheric Administration, Strategic Assessment Branch, Rockville, MD, 43 pp.
Buhler, D.R. and D.E. Williams. 1988. The role of biotransformation in toxicity. Fish.
Aquatic Toxicology, 11:303-311.
Cantillo, A.Y. and T.P. O'Connor. 1992. Trace element concentrations in sediments
from the NOAA National Status and Trends Programme compared to data from
throughout the world. Chemistry and Ecology, 7:31 -50.
Capuzzo, J.M., M.N. Moore, and J. Widdows. 1988. Effects of toxic chemicals in the
marine environment: Predictions of impacts from laboratory studies. Aquatic
Toxicology, 11:19-28.
Chapman, P.M. 1988. Marine sediment toxicity tests. Pages 391-402. In: Chemical and
Biological Characterization of Sludges, Sediments, Dredge Spoils, and Drilling
Muds, J.J. Lichtenberg, F.A. Winter, C.I. Weber and L. Fredkin, eds., STP 976.
ASTM, Philadelphia, PA.
Cochran, J.K. 1991. Forging a partnership between federal monitoring programs and
the academic marine community. Summary of a workshop hosted by the Marine
85
-------
Sciences Research Center, State University of New York at Stony Brook, 29-30
April 1991. Special Report 93 of the Marine Sciences Research Center, State
University of New York at Stony Brook.
Connell, D.W. and G.J. Miller. 1984. Chemistry and Ecotoxicology of Pollution. John
Wiley and Sons, New York.
Cooley, W.W. and P.R. Lohnes. 1971. Multivariate Data Analysis, John Wiley & Sons,
Inc., New York.
Culliton, T.J., M.A. Warren, T.R. Goodspeed, D.G. Remer, C.M. Blackwell, and J.J.
McDonough. 1990. 50 years of population change along the nation's coasts,
1960-2010. Second Report of Coastal Trends Service. U.S. Department of
Commerce, NOAA, Strategic Assessment Branch, Rockville, MD, 41 pp.
Daskalakis, K.D. and T.P. O'Connor. 1994. Inventory of chemical concentrations in
coastal and estuarine sediments. NOAA Technical Memorandum NOS ORCA
76. National Oceanic and Atmospheric Administration, National Ocean Service,
Silver Spring, MD, 66 pp.
Di Toro, D.M., J.D.~Mahony, D.J. Hansen, K.J. Scott, M.B. Hicks, S.M. Mayer, and M.S.
Redmond. 1990. Toxicity of cadmium in sediments: the role of acid volatile
sulfide. Environmental Toxicology and Chemistry, 9:1487-1502.
Di Toro, D.M., C.S. Zarba, D.J. Hansen, W.J. Berry, R.C. Swartz, C.E. Cowan, S.P.
Pavlou, H.E. Allen, N.A. Thomas, and P.R. Paquin. 1991. Technical basis for
establishing sediment quality criteria for nonionic organic chemicals using
equilibrium partitioning. Environmental Toxicology and Chemistry, 10:1541-1582.
FDA (Food and Drug Administration). 1971. National shellfish register of classified
estuarine waters. Food and Drug Administration, Shellfish Sanitation Branch,
Northeast Technical Services Unit, Davisville, Rl.
FDA. 1985. National shellfish register of classified estuarine waters. National Oceanic
and Atmospheric Administration, Strategic Assessment Branch, Rockville, MD.
Fein, G.G., J.L Jacobson, S.W. Jacobson, P.W. Schwartz, and J.K, Dowler. 1984.
Prenatal exposure to polychlorinated biphenyls: Effects on birth size and
gestational age. Pediatrics, 105:315-320.
Feng, H., J.K. Cochran, H. Lwiza, B.J. Brownawell, and D. J. Hirschberg. 1998.
Distribution of heavy metal and PCB contaminants in the sediments of an urban
estuary: the Hudson River. Marine Environmental Research, 45(1): 69-88.
86
-------
Forstner, U. and G.T.W. Wittmann. 1981. Metal Pollution in the Aquatic Environment.
2nd Revised Edition. Springer-Verlag, New York.
Hargis, W.J., Jr. and D.S. Haven. 1988. Rehabilitation of the troubled oyster industry of
the lower Chesapeake Bay. Journal of Shellfish Research, 7:271-279.
Heasly, P., S. Pultz, and R. Batiuk. 1989. Chesapeake Bay basin monitoring program
atlas, Volume I, Water quality and other physicochemical monitoring programs.
CBP/TRS 34/89. U.S. Environmental Protection Agency, Chesapeake Bay
Program, Annapolis, Maryland.
Heimbuch, D.G., H.T. Wilson, D.S. Robson, and J.K. Summers. 1995a. Design-based
estimation of the proportion of area degraded and associated variance
estimators for EMAP-Estuaries sampling in the Louisianian and Virginian
Provinces: 1990-1994. Draft report prepared for U.S. Environmental Protection
Agency by Coastal Environmental Services, Linthicum, Maryland.
Heimbuch, D.G., H.T. Wilson, S.J. Schwager, and D.S. Robson. 1995b. Power
analysis for tests for long term trends in environmental conditions of estuaries.
Draft report prepared for U.S. Environmental Protection Agency by Coastal
Environmental Services, Linthicum, Maryland.
Hinga, K.R. 1988. Seasonal predictions for pollutant scavenging in two coastal
environments using a model calibration based upon thorium scavenging. Marine
Environmental Research, 26:97-112.
Holland, A.F. 1990. Near Coastal Program Plan for 1990: Estuaries. EPA/600/4-90-033.
U.S. Environmental Protection Agency, Environmental Research Laboratory,
Office of Research and Development, Narragansett, Rl.
Holland, A.F., AT. Shaughnessy, L.C. Scott, V.A. Dickens, J.A. Ranasinghe, and J.K.
Summers. 1988. Progress report: Long-term benthic monitoring and assessment
program for the Maryland portion of Chesapeake Bay (July 1986-October 1987).
PPRP-LTB/EST-88-1. Prepared for Maryland Power Plant Research Program
and Maryland Department of the Environment, Office of Environmental
Programs. Versar, Inc., ESM Operations, Columbia, MD.
Holland, A.F., A. Shaughnessy and M.H. Hiegel. 1987. Long-term variation in
mesohaline Chesapeake Bay macrobenthos: Spatial and temporal patterns.
Estuaries, 10:277-245.
87
-------
Honeyman, B.D. and P.H. Santschi. 1988. Metals in aquatic systems: Predicting their
scavenging residence times from laboratory data remains a challenge.
Environmental Science and Technology, 22:862-871.
Hunsaker, C.T. and D.E. Carpenter, eds. 1990. Environmental Monitoring and
Assessment Program: ecological indicators. EPA/600/3-90/060 U.S.
Environmental Protection Agency, Office of Research and Development,
Washington, DC.
Kirkland, L.L. 1992. EMAP quality assurance management plan. Office of Research
and Development, U.S. Environmental Protection Agency, Washington, DC.
Klecka, W.R. 1980. Discriminant Analysis. Sage University Paper Series on
Quantitative Applications in the Social Sciences, Series No. 07-019, Beverly
Sage Publications.
Leonard, D.L., M.A. Broutman, and K.E. Harkness. 1989. The quality of shellfish
growing areas on the east coast of the United States. National Oceanic and
Atmospheric Administration, Ocean Assessments Division, Rockville, MD.
Lippson, A. J., M.S. Haire, A.F. Holland, F. Jacobs, J. Jensen, R.L. Moran-Johnson,
T.T. Polgar, and W.R. Richkus. 1979. Environmental Atlas of the Potomac
Estuary. Prepared for Maryland Department of Natural Resources, Power Plant
Siting Program. Martin Marietta Corporation, Baltimore, MD.
Long, E.R., and M.F. Buchman. 1989. An evaluation of candidate measures of
biological effects for the National Status and Trends Program. National Oceanic
and Atmospheric Administration Technical Memorandum NOSOMA 45. Seattle,
WA, 105pp..
Long, E.R. and L.G. Morgan. 1990. The potential for biological effects of sediment-
sorbed contaminants tested in the National Status and Trends Program. National
Oceanic and Atmospheric Administration Technical Memorandum NOS OMA 52.
Seattle, WA, 175 pp.
Long, E.R., D.D. MacDonald, S.L. Smith, and F.D. Calder. 1995. Incidence of adverse
biological effects within ranges of chemical concentrations in marine and
estuarine sediments. Environmental Management, 19(1):81-97.
Malins, D.C. 1989. Sediment contamination and marine ecosystems: Potential risks to
humans. In: Contaminated Marine Sediments - Assessment and Remediation.
National Academy of Sciences. National Academy Press, Washington, DC, pp
155-164.
88
-------
Mayer, F.L., Jr., L.L. Marking, L.E. Pedigo, and J.A. Brecken. 1989. Physiochemical
factors affecting toxicity: pH, salinity, and temperature, Part 1. Literature review.
U.S. Environmental Protection Agency, Gulf Breeze, FL.
Morganthau, T. 1988. Don't go near the water: Is it too late to save our dying coasts?
Newsweek, August 1, 42-47.
Nixon, S.W., C.D. Hunt and B.L. Nowicki. 1986. The retention of nutrients (C,N,P),
heavy metals (Mn, Cd, Pb, Cu), and petroleum hydrocarbons in Narragansett
Bay. Pages 99-122. In: Biogeochemical Processes at the Land-sea Boundary, P.
Lasserre and J.M. Martin, eds., Elsevier, NY.
National Oceanic and Atmospheric Administration (NOAA). 1987. NOAA estuarine and
coastal ocean science framework. U.S. Department of Commerce, National
Oceanic and Atmospheric Administration, Estuarine Program Office,
Washington, DC.
NOAA. 1988. Federal plan for ocean pollution research development and monitoring:
Fiscal years 1988-1991. Prepared by the National Ocean Pollution Program
Office for the National Ocean Pollution Policy Board, Rockville, MD.
NOAA. 1992. The National Status and Trends Program, Marine Environmental Quality.
National Oceanic and Atmospheric Administration, National Ocean Service,
Rockville, Maryland.
National Research Council (NRC). 1989. Contaminated marine sediments -
assessment and remediation. National Academy Press, Washington, D.C. 493
PP-
NRC. 1990. Managing troubled waters: the role of marine environmental monitoring.
National Academy Press, Washington, DC.
O'Connor, J.S., J.J. Ziskowski, and R.A. Murchelano. 1987. Index of pollutant-induced
fish and shellfish disease. Special Report. National Oceanic and Atmospheric
Administration, NOS, Rockville, MD.
O'Connor, T.P. 1990. Coastal Environmental Quality in the United States, 1990,
Chemical Contamination in Sediments and Tissues. NOAA 20th Anniversary
Report, Rockville, MD, 34 pp.
Officer, C.B., R.B. Biggs, J.L. Taft, L.E. Cornin, M.A. Tyler, and W.R. Boynton. 1984.
Chesapeake Bay anoxia: Origin, development, and significance. Science,
223:22-27.
89
-------
Olsen.A.R. 1992. Program data quality objectives for EMAP status; and trends. Draft
report. Environmental Monitoring and Assessment Program, U.S. Environmental
Protection Agency, Washington, DC.
Olsen, S., D.D. Robadue, and V. Lee. 1980. An interpretive atlas of Narragansett Bay.
Coastal Resources Center, University of Rhode Island, Marine Bulletin 40.
OTA (Office of Technology Assessment). 1987. Wastes in Marine Environments.
Washington, DC.
Overton, J. McC., T.C. Young, and W.S. Overton. 1993. Using found data to augment
a probability sample: procedure and case study. Environmental Monitoring and
Assessment, 26:65-83.
Overton, W.S., D. White and D.L. Stevens. 1991. Design report of the Environmental
Monitoring and Assessment Program. EPA/600/3-91/053. U.S. Environmental
Protection Agency, Environmental Research Laboratory, Corvallis, OR.
Parker, C.A., J.E. O'Reilly, and R.B. Gerzoff. 1986. Oxygen depletion in Long Island
Sound. Historical Trends Assessment Program, National Oceanic and
Atmospheric Administration, Ocean Assessment Division, Rockville, MD.
Paul, J.F., K.J. Scott, A.F. Holland, S.B. Weisberg, J.K. Summers, and A. Robertson.
1992. The estuarine component of the US E.P.A.'s Environmental Monitoring
and Assessment Program. Chemistry and Ecology, 7:93-116.
Pearl, H. W. 1988. Nuisance phytoplankton blooms in coastal, estuarine, and inland
waters. Limnology and Oceanography, 33:823-847.
Pearson, T.H. and R. Rosenberg. 1978. Macrobenthic succession in relation to organic
enrichment and pollution of the marine environment. Oceanography and Marine
Biology Annual Review, 16:229-311.
Queen, W.H., B.J. Copeland, and J.R. Schubel. 1992. Forging a partnership between
federal monitoring programs and the academic marine community. A report of
the Board on Oceans and Atmosphere, National Association of State Universities
and Land-Grant Colleges, Washington, DC.
Rabalais, N.N., M.J. Dagg, and D.F. Boesch. 1985. Nationwide review of oxygen
depletion and eutrophication in estuarine and coastal waters: Gulf of Mexico.
Louisiana University's Marine Consortium.
90
-------
Ranasinghe, J.A., S.B. Weisberg, D.M. Dauer, L.C. Schaffner, R.J. Diaz, and J.B.
Frithsen. 1993. Chesapeake Bay benthic community restoration goals.
Prepared for the Chesapeake Bay Program Office and the Governor's Council
on Chesapeake Bay by Versar, Inc. Columbia, Maryland.
Ranasinghe, J.A., S.B. Weisberg, J. Gerritsen, and D.M. Dauer. 1994. Assessment of
Chesapeake Bay benthic macrobenthic resource condition in relation to water
quality and watershed stressors. Prepared for the Governor's Council on
Chesapeake Bay and the Chesapeake Bay Program by Versar, Inc. Columbia,
Maryland.
Reifsteck, D.M., C.J. Strobel, and D.J. Keith. 1993. EMAP-Estuaries 1993 Virginian
Province Field Operations and Safety Manual. U.S. Environmental Protection
Agency, Office of Research and Development, Environmental Research
Laboratory, Narragansett, Rl.
Reish, D.J., and J.L. Barnard. 1960. Field toxicity tests in marine water utilizing the
polychaetous annelid Capitella capitata (Fabricius). Pacific Naturalist, 1:1-8.
Rhoads, D.C., P.L. McCall and J.Y. Yingst. 1978. Disturbance and production on the
estuarine sea floor. American Scientist, 66:577-586.
Robertson, A., B.W. Gottholm, D.D. Turgeon, and D.A. Wolfe. 1991. A comparative
study of contaminant levels in Long Island Sound. Estuaries, 14(3):290-298.
Sanders, H.L. 1956. Oceanography of Long Island Sound, 1952-1954. X: The biology
of marine bottom communities. Bulletin of the Bingham Oceanographic
Collection, 15:345-414.
Sanders, H.L., 1960. Benthic studies in Buzzards Bay. III. The structure of the soft
bottom community. Limnology and Oceanography, 5:138-153.
Sanders, H.L., J.F. Grassle, G.R. Hampson, L.S. Morse, S. Garner-Price and C.C.
Jones. 1980. Anatomy of an oil spill: Long-term effects from the grounding of the
barge Florida off West Falmouth, Massachusetts. Journal of Marine Research,
38:265-380.
Sanders, J.E. 1989. PCB pollution in the upper Hudson River. In: Contaminated Marine
Sediments - Assessment and Remediation. National Academy Press,
Washington, DC, pp 365-400.
SAS. 1990a. SAS/STAT User Guide, Version 6, 4th Edition, Volume 1. SAS Institute,
Gary, North Carolina.
91
-------
SAS. 1990b. SAS/STAT User Guide, Version 6, 4th Edition, Volume 2. SAS Institute,
Gary, North Carolina.
Schubel, J.R., and H.H. Carter. 1984. The estuary as a filter for fine-grained suspended
sediment. Pages 81-104. In: The Estuary as a Filter, S. Kennedy, ed. Academic
Press, Orlando, FL.
Schimmel, S.C., B.D. Melzian, D.E. Campbell, S.J Benyi, J.S. Rosen, and H.W. Buffum.
1994. Statistical Summary: EMAP-Estuaries Virginian Province -1991.
EPA/620/R-94/005. U.S. Environmental Protection Agency, Office of Research
and Development, Environmental Research Laboratory,, Narragansett, Rhode
Island.
i ' i
Science Advisory Board (SAB). 1988. Future risk: Research strategies for the 1990s.
The report of the Research Strategies Committee. Science Advisory Board.
Washington, DC.
Scott, K.J. and Redmond, M.S. 1989. The effects of a contaminated dredged material
on laboratory populations of the tubificious amphipod, Ampelisca abdita. pp. 289-
303. In: Aquatic Toxicology and Hazard Assessment: 12th Volume, U.M. Cowgill
and LR. Williams, eds., ASTM STP 1027. Philadelphia, Pennsylvania.
Sharp, J.H., J.R. Pennock, T.M. Church, T.M. Tramontano, and L.A. Cifuentes. 1984.
The estuarine interaction of nutrients, organics, and metals: A case study in the
Delaware Estuary. Pages 241-258. In: The Estuary as a Filter, V.S. Kennedy,
ed., Academic Press, Orlando, FL.
Sinderman, C.J. 1979. Pollution-associated diseases and abnormalities offish and
shellfish: A review. Fish. Bull. 76:717-741.
Smart, T., E. Smith, T. Vogel, C. Brown, and K. Wolman. 1987. Troubled Waters.
Business Week, October 12, 88-104.
Smayda, T.J., and T.A. Villareal. 1989. The 1985 "brown-tide" and the open
phytoplankton niche in Narragansett Bay during summer. Pages 159-187. In:
Novel Phytoplankton Blooms, EM. Cosper, V.M. Bricelj, and E.J. Carpenter,
eds., Coastal and Estuarine Studies, Vol. 35. Springer-Verlag, NY.
Spies, R.B. 1989. Sediment bioassays, chemical contaminants and benthic ecology:
New insights or just muddy water? Marine Environmental Research, 27:73-75.
92
-------
Sprague, J.B. 1985. Factors that modify toxicity. In: Fundamentals of Aquatic
Toxicology, G.M. Rand and S.R. Petrocelli, eds., Hemisphere Publication Corp,
New York.
Strobel, C.J., H.W. Buffurn, S.J. Benyi, E.A. Petrocelli, D.R. Reifsteck, and D.J. Keith.
1994. Statistical Summary: EMAP-Estuaries Virginian Province -1992.
EPA/620/R-94/019. U.S. Environmental Protection Agency, Office of Research
and Development, Environmental Research Laboratory, Narragansett, Rl.
Strobel, C.J., H.W. Buffum, SJ. Benyi, E.A. Petrocelli, D.R. Reifsteck, and D.J. Keith.
1995. Statistical Summary: EMAP-Estuaries Virginian Province -1990-1993.
EPA/620/R-94/026. U.S. Environmental Protection Agency, Office of Research
and Development, National Health and Environmental Effects Research
Laboratory, Atlantic Ecology Division, Narragansett, Rl.
Strobel, C.J. and R.M. Valente. 1995. Quality Assurance Report: EMAP-Virginian
Province, 1990-1993. EPA/620/R-95/007. U.S. Environmental Protection
Agency, Office of Research and Development, National Health and
Environmental Effects Research Laboratory, Atlantic Ecology Division,
Narragansett, Rl.
Swartz, R.C. 1987. Toxicological methods for determining the effects of contaminated
sediment on marine organisms. Pages 183-198. In: Fate and Effects of
Sediment Bound Chemicals in Aquatic Systems, K.L. Dickson, A.W. Maki, and
W.A. Brungs, eds., Pergamon Press, New York.
Swartz, R.C. 1989. Marine sediment toxicity tests.. In: Contaminated sediments -
assessment and remediation. National Academy Press, Washington, DC pp
115-129.
Swartz, R.C., W.A. DeBen, J.K. Jones, J.O. Lamberson, and F.A. Cole. 1985.
Phoxocephalid amphipod bioassay for marine sediment toxicity. In: Aquatic
Toxicology and Hazard Assessment: Seventh Symposium, R.D. Cardwell, R.
Purdy, and R.C. Bahner, eds., American Society for Testing and Materials,
Philadelphia, PA. pp 284-307.
Swartz, R.C., F.A. Cole, J.O. Lamberson, S.P. Ferraro, D.W. Schults, W.A. DeBen, H.
Lee II, and R.J. Ozretich. 1994. Sediment toxicity, contamination and amphipod
abundance at a DDT- and dieldrin-contaminated site jn San Francisco Bay.
Environmental Toxicology and Chemistry, 13:949-962.
93
-------
Terrell, T.T. 1979. Physical regionalization of coastal ecosystems of the United States
and its territories. FWS/OBS-79/80. U.S. Fish and Wildlife Service, Office of
Biological Services.
The Conservation Foundation. 1988. Protecting America's Wetlands: An Action
Agenda. The Conservation Foundation, Washington, DC.
Theede, H. 1973. Comparative studies on the resistance of marine bottom
invertebrates to oxygen deficiency and hydrogen sulphide. Marine Biology 2:325-
337.
Toufexis, A. 1988. Our filthy seas: The world's oceans face a growing threat from
manmade pollution. Time, August 1.
Turekian, K.K. 1977. The fate of metals in the oceans. Geochimica et Cosmochimica
Acta, 41:1139-1144.
U.S. Environmental Protection Agency (USEPA). 1984. Chesapeake Bay: A framework
for action. Prepared for the U.S. Congress by the Environmental Protection
Agency, Region 3, Philadelphia, PA.
USEPA. 1989. Briefing Report to the EPA Science Advisory Board on the equilibrium
partitioning approach to generating sediment quality criteria. EPA/440/5-89-002.
USEPA, Criteria and Standards Division, Washington, DC.
USEPA. 1991. Summary, Environmental Monitoring and Assessment Program - Near
Coastal, Virginian Province User Network Exchange. A workshop held on 3-5
April 1991, Ocean City, Maryland.
USEPA. 1993a. Proposed sediment quality criteria for the protection of benthic
organisms: Acenaphthene. U.S. Environmental Protection Agency, Office of
Science and Technology Washington. In review.
USEPA. 1993b. Proposed sediment quality criteria for the protection of benthic
organisms: Phenanthrene. U.S. Environmental Protection Agency, Office of
Science and Technology Washington. In review.
USEPA. 1993c. Proposed sediment quality criteria for the protection of benthic
organisms: Fluoranthene. U.S. Environmental Protection Agency, Office of
Science and Technology Washington. In review.
94
-------
USEPA. 1993d. Proposed sediment quality criteria for the protection of benthic
organisms: Dieldrin. U.S. Environmental Protection Agency, Office of Science
and Technology Washington. In review.
USEPA. 1994a. Methods for assessing the toxicity of sediment-associated
contaminants with estuarine and marine amphipods. EPA 600/R-94/025. U.S.
Environmental Protection Agency, Office of Research and Development,
Narragansett, Rhode Island.
USEPA. 1994b. Summary of the Comprehensive Conservation and Management Plan.
EPA 8-42-S-94-001, Long Island Sound Office, U.S. Environmental Protection
Agency, Stamford, Connecticut.
USEPA. 1995a. The State of the Chesapeake Bay 1995. Chesapeake Bay Program,
U.S. Environmental Protection Agency, Annapolis, Maryland.
USEPA. 1995b. EMAP-Estuaries Laboratory Methods Manual, Volume I - Biological
and Physical Analyses. EPA/620/R-95/008. U.S. Environmental Protection
Agency, Office of Research and Development, National Health and
Environmental Effects Research Laboratory, Atlantic Ecology Division,
Narragansett, Rl.
USEPA. 1996. The Delaware Estuary. Discover its Secrets. A Management Plan for
the Delaware Estuary. The Comprehensive Conservation and Management
Plan for the Delaware Estuary, U.S. Environmental Protection Agency,
Delaware Estuary Program, Philadelphia, Pennsylvania.
Valente, R. and J. Schoenherr. 1991. EMAP-Estuaries Virginian Province Quality
Assurance Project Plan. U.S. Environmental Protection Agency, Office of
Research and Development, Environmental Research Laboratory, Narragansett,
Rl.
Valente, R. and C.J. Strobel. 1993. Environmental Monitoring and Assessment
Program-Estuaries: 1993 Virginian Province Quality Assurance Project Plan.
U.S. Environmental Protection Agency, Office of Research and Development,
Environmental Research Laboratory, Narragansett, Rl.
Valente, R., C.J. Strobel, J.E. Pollard, KM Peres, T.C. Chiang, and J. Rosen. 1990.
Quality Assurance Project Plan for EMAP Near Coastal: 1990 Demonstration
Project. U.S. Environmental Protection Agency, Office of Research and
Development, Environmental Research Laboratory, Narragansett, Rl.
95
-------
Valente, R., C.J. Strobel, and S.C. Schimmel. 1992. EMAP-Estuaries Virginian
Province 1992 Quality Assurance Project Plan. U.S. Environmental Protection
Agency, Office of Research and Development, Environmental Research
Laboratory, Narragansett, Rl.
Valette-Silver, N. 1992. Elemental Analyses in Marine Sediments and Biological
Tissues. Quality Assurance Program: Summary 1985-1990. NOAA Technical
Memorandum NOS ORCA 66. National Oceanic and Atmospheric
Administration, National Ocean Service, Office of Ocean Resources
Conservation and Assessment, Rockville, Maryland.
Vernberg, F.S. 1972. Dissolved gases, pp. 1491-1526.ln: Marine Ecology, Vol. I, Part 3,
O. Kinne, ed., Wiley-lnterscience, New York.
Washington, H.G. 1984. Diversity, biotic and similarity indices, a review with special
relevance to aquatic ecosystems. Water Resources, 19(6): 653-694.
Weisberg, S.B., J.B. Frithsen, A.F. Holland, J.F. Paul, K.J. Scott, J.K. Summers, H.T.
Wilson, D.G. Heimbuch, J. Gerritsen, S.C. Schimmel, and R.W. Latimer. 1993.
Virginian Province Demonstration Project Report, EMAP-Estuaries: 1990.
EPA/620/R-93/006. U.S. Environmental Protection Agency, Office of Research
and Development, Washington, DC.
Weisberg, S.B., J.A. Ranasinghe, D.M. Dauer, L.C. Schaffner, R.J. Diaz, and J.B.
Frithsen. 1997. An estuarine benthic index of biotic integrity (B-IBI) for
Chesapeake Bay. Estuaries, 20(1): 149-158.
Whitledge, T.E. 1985. Nationwide review of oxygen depletion and eutrophication in
estuarine and coastal waters: Executive Summary. Brookhaven National
Laboratory, Upton, NY.
Windom, H.L, S.J. Schropp, F.D. Calder, J.D. Ryan, R.G. Smith, L.C. Bumey, F.G.
Lewis, and C.H. Ralinson. 1989. Natural trace meta! concentrations in estuarine
and coastal marine sediments of the southeastern United States. Environmental
Science and Technology. 23: 314-320.
Word, J.W., N.P. Kohn, E.S. Barrows, and J.A. Ward. 1994. The influence of sediment
stability/disturbance on toxicity of non-polar organic compounds to marine
amphipods. Presentation at Workshop on Utility of AVS/EqP in Hazardous
Waste Site Evaluation, Woods Hole, Massachusetts.
96
-------
APPENDIX A: OVERVIEW OF SAMPLING AND ANALYTICAL METHODS
An important aspect of the EMAP Virginian Province Four-Year
Demonstration Project was that the data were collected and processed with consistent
methods. The field methods are documented in the Field Operations and Safety Manual
(Reifsteck etal., 1993). Laboratory methods are documented in the EMAP-Estuaries
Laboratory Methods Manual (U.S. EPA, 1995b). A performance-based approach was
used for chemical analyses, consistent with the approach used by the NOAA National
Status & Trends (NS&T) Program (Valette-Silver, 1992). A strict quality assurance
program was used from the initiation of field activities. All activities were conducted in
accordance with strict criteria described in a Quality Assurance Project Plan, which was
updated annually as needed (Valente and Schoenherr, 1991; Valente and Strobel,
1993; Valente et a/., 1990, 1992). An accounting of the results of the EMAP Virginian
Province QA Program is found in the 1990-1993 Quality Assurance Report (Strobel and
Valente, 1995). An overview of the methods is provided in this appendix, summarized
from Strobel ef, a/. (1995). All sampling was conducted from small (24') vessels, except
for fish from deep-water stations (> 25m), which were collected from larger vessels.
Water column profiles for water quality parameters were collected at each station
using a SeaBird SBE-25 Sea Logger CTD. The unit was equipped with probes to
measure salinity, temperature, depth, pH, dissolved oxygen (DO), light transmission,
fluorescence, and photosynthetically active radiation (PAR). Water quality
measurements were collected upon arrival at a sampling station; no effort was made to
standardize for the time of day or stage of tide. The CTD was equilibrated at the sea
surface, and then lowered through the water column until reaching a depth of one meter
above the bottom. There the CTD was allowed to equilibrate again. The unit was then
returned to the surface, where data were downloaded to an on-board computer for
review and storage. If the CTD cast appeared unusual or failed quality control criteria,
the cast was repeated. Beginning in 1991, a bottom water sample was collected, and
the dissolved oxygen concentration determined with a YSI Model 58 DO meter. This
measurement served as a check on the CTD probe as well as a back-up in case the
CTD failed.
Benthic samples for evaluation of invertebrate composition, abundance, and
biomass were collected at all sampling sites where a sample could be collected. Three
samples were collected at each site using a stainless steel, Young-modified van Veen
grab that samples a surface area of 440 cm2. A small core (2 cm diameter) was
collected from each grab for sediment characterization (grain size). The remaining
sample was sieved through a 0.5 mm screen using a backwash technique that
minimized damage to soft-bodied animals. Samples were preserved in 10% formalin-
rose bengal solution and stored for at least 30 days prior to processing to assure proper
fixation. In the laboratory, macrobenthic community samples were transferred from
formalin to an ethanol solution and sorted. Biomass was measured for key taxa and all
97
-------
other taxa were grouped according to taxonomic type (e.g., polychaetes, amphipods,
decapods).
An additional 6-10 sediment grabs at each station were obtained for sediment
chemistry and toxicity analyses. The top 2 cm of sediment was removed from each grab
using a stainless steel spoon and thoroughly homogenized in a stainless steel pot.
Sediment for chemistry analyses was placed in clean glass jars with Teflon liners or
polypropylene containers (for organic and metals analyses, respectively), shipped on
ice, and stored frozen in the laboratory prior to analysis for contaminants. Sediments
were analyzed for 24 polycyclic aromatic hydrocarbons (PAHs), 18 polychlorinated
biphenyl (PCB) congeners, DDTs, 11 chlorinated pesticides, tributyl tins, and 15 metals
(Table 3-9). The chemical analyte list is the same as used in the NOAA NS&T Program.
An additional aliquot was placed in a small polyethylene bag and refrigerated for grain
size analysis. The remainder of the composite sample (> 3,000 cm3) was placed in a
clean one gallon plastic jar for sediment toxicity testing.
Analysis of sediments for major and trace elements involved a total digestion
(i.e., complete dissolution) of the sediment matrix. For the metals Ag, Al, Cr, Cu, Fe,
Mn, Ni, Pb, and Zn, the total digestion was accomplished using HF/HNO3 in an open
beaker on a hot plate, followed by instrumental analyses using inductively-coupled
plasma-atomic emission spectrometry (ICP-AES). For metals As, Cd, Sb, Se, and Sn,
a microwave digestion using HNO3/HCL in a closed Teflon-lined pressure vessel was
followed by analysis using Zeeman-corrected, stabilized temperature graphite furnace
atomic absorption (GFAA). Mercury (Hg) was analyzed by cold vapor atomic absorption
spectrometry.
The analysis of organic contaminants in the sediment involved sample extraction
and cleanup followed by instrumental analysis. This included Soxhlet extraction, extract
drying using sodium sulfate, extract concentration using Kuderna-Danish apparatus,
removal of elemental sulfur with activated copper, and removal of organic interferents
with gel permeation chromatography (GPC) and/or alumina. Following extraction and
cleanup, PAH compounds were analyzed using gas chromatography/mass
spectrometry (GC/MS). The pesticides and PCB congeners were analyzed using gas
chromatography/electron capture detection (GC/ECD) with second column
confirmation.
Toxicity tests were performed on the composite sediment samples from each
station using the standard 10-day acute test method (U.S. EPA, 1995b, taken from U.S.
EPA, 1994a) and the tube-dwelling amphipod Ampelisca abdita. Arnphipods were
exposed to sediment from the site for 10 days Under static conditions in 1-L glass test
chambers. Five replicates per station were tested with 20 amphipods per replicate. A
performance control (i.e., treatment with uncontaminated sediment) was run with each
test, as was a water-only test using a reference toxicant (Cu or sodium dodecyl sulfate)
98
-------
to evaluate the condition of the test organisms. Eighty-five percent survival in the
sediment control was required for a test to be valid. To normalize for test conditions
and amphipod health, survival among treatments is expressed as percent of control
survival.
99
-------
APPENDIX B: REFINEMENTS TO BENTHIC INDEX
Measurements of the ecological condition of benthic communities are critical
components of the overall assessment of condition of estuarine waters; benthic
organisms are ideal integrators of water and sediment quality. Traditional analyses of
benthic communities yield a rich data set on species composition arid abundance, and
functional diversity and structure. The challenge of these analyses is to relate patterns
in taxonomic and functional structure of the benthos to stressor characteristics of
interest to assessments that also account for the range of habitat conditions
encountered in sampling estuarine waters. For the Virginian Province, important
stressors are related to toxic contaminants and hypoxia; affects of these stressors on
benthic organisms must be demonstrated across habitat conditions that range from
marine to fresh waters and silt to sand.
An index of estuarine benthic community condition was originally developed
using data from the 1990 Virginian Province Demonstration Project. The procedures
used to develop this benthic index were documented in the EMAP-Estuaries Virginian
Province 1990 Demonstration Project Report (Weisberg etal., 1993). The index was
updated using data collected during 1990-91 in the Virginian Province in Schimmel et
a/. (1994). The current index, discussed in this appendix and used to report on the
condition of benthic resources in this report, was developed from a subset of the data
collected over the entire four years of sampling in the Virginian Province (1990-93). This
effort represents EMAP's continued attempt to discover, among many individual
metrics, a single metric or combination of metrics that has a high level of discriminatory
power between good and poor ecological conditions. The index developed with the
1990-93 data is an improvement upon and a revision to the prior benthic indices; it has
always been the intent to continually revise the index as more data became available
(Weisberg etal., 1993).
The basic approach to develop the index is to determine the combination of
individual benthic metrics that best discriminates between good and poor benthic
conditions encountered during sampling. Discriminant analysis (Cooley and Lohnes,
1971; Kiecka, 1980) is the analytical tool used to accomplish this. The actual process
for developing the benthic index involves several discrete steps:
1. Identification of a set of candidate benthic metrics that include components of
faunal and functional diversity and structure.
2. Development of a test data set that contains relatively pristine sites and those
exhibiting toxic contamination, hypoxia, or both.
3. Identification of combinations of candidate metrics that discriminate between
impacted and unimpacted sites.
100
-------
4. Validation of the selected combination of metrics.
The test data set from 1990 sampling in the Virginian Province consisted of 19
impacted sites (4 with salinity < 5 ppt) based on high sediment contaminant
concentrations (combined with toxicity) or low near-bottom dissolved oxygen levels, and
14 sites (6 with salinity < 5 ppt) classified as unimpacted, or reference, based on low
sediment contaminant levels and absence of toxicity or low dissolved oxygen
conditions. The 1990 benthic index correctly classified 89% of the impacted sites and
86% of the unimpacted sites. Metrics in the 1990 index included percent expected
number of species (salinity normalized), number of amphipods, percent of total
abundance as bivalves, mean weight per individual polychaete, and number of
capitellids.
The process of validation for the 1990 benthic index involved an independent
data set to ensure that the multivariate solution was not specific to the original test data
set used in 1990. Using the same criteria applied in 1990 to define impacted and
unimpacted (i.e., reference) sites, an independent data set was established from the
1991 database (Schimrnel et a/.,1994). This validation data set consisted of 13 sites
classified as impacted and 46 sites classified as unimpacted.
Of the 46 sites from 1991 classified as unimpacted, the 1990 benthic index
correctly classified 39 (85%). Of the 13 sites from 1991 classified as impacted, the
1990 benthic index correctly classified 7 (54%). The relatively high overall rate of
misclassification, particularly for impacted sites, was deemed unacceptable, and a
decision was made to reconstruct a new benthic index using the combined 1990-91
data set.
The test data set from 1990-91 sampling in Virginian Province consisted of 31
impacted sites (5 with salinity < 5 ppt) and 51 unimpacted sites (9 with salinity < 5 ppt).
The 1990-91 benthic index correctly classified 84% of the impacted sites and 85% of
the unimpacted sites. Metrics in the index included mean abundance of opportunistic
species, biomass/abundance ratio for all species, and mean number of infaunal
species. Of particular note was that habitat normalization (e.g., salinity) was not
included in this index; habitat normalized metrics were considered for the index
development, but were eliminated during the analysis. (See Schimrnel et al. (1994) for
details.) The benthic index developed using the 1990-91 data set suffered from poor
representation of impacted and reference conditions in low salinity (< 5 ppt) in the test
data set. This benthic index applied to the entire 1990-93 data set indicated a strong
relationship with salinity (Figure B-1). It also appeared to misclassify good sites in the
oligohaline and impacted sites in the meso- and polyhaline (see Figure B-2)
with 90-91 BI applied to the 90-91 calibration data set.
Concerns with the 1990-91 benthic index were that no habitat normalization was
101
-------
35
30 1
| =5i
£?
1 20 ..
Z
1
I 15 4-
5 .
-6 -5
-3 -2-1 0 1 2
1990-91 Benthic Index
-5-
4
-4-
5
Figure B-1. The 1990-91 benthic index versus bottom water salinity for 1990-93 EMAP Virginian
Province data set. Negative benthic index values indicate impacted conditions
35
30 ..
M 20 ..
3
15 ..
5 ..
o
o
o
o
oo
o o
o
i unimpacted site
, impacted site
-2
-1 0 1
1990-91 Benthic Index
Figure B-2. The 1990-91 benthic index versus bottom water salinity for 1990-91 calibration data set
(Schimmel et a/., 1994). Negative benthic index values indicate impacted conditions.
102
-------
included in any of the individual metrics and that there was an apparent bias with
salinity. For these reasons, the benthic index was reconstructed using data from 1990-
93 addressing the following points:
1. Importance of habitat normalization of individual benthic metrics.
2. Inclusion of a wider selection of benthic metrics as candidates for the
discriminant analysis.
3. Balance of sites in the test data set for impacted and unimpacted categories
and across salinity zones.
RECONSTRUCTION OF THE BENTHIC INDEX USING 1990-93 DATA
Reconstruction of the benthic index using the combined 1990-93 data followed
the same basic steps described in the 1990 Demonstration Project Report (Weisberg
et. a/. 1993) and the 1991 Statistical Summary (Schimmel etal., 1994). Results and
discussion are presented in the following sections.
Stepl: Identify candidate benthic measures
As in 1990 and 1991, benthic abundance, biomass, and species composition
data were used to define descriptors of the major ecological attributes of the benthic
assemblages occurring at each sample site (Table B-1). Additional benthic diversity
metrics evaluated included Shannon's H, Simpson's D, Gleason's D, and Pielou's
evenness (Washington, 1984) based upon infauna, epifauna, and both infauna and
epifauna. These biodiversity measures are defined as:
Shannon's H:
H =
, ,
|08'°
Simpson's D: D =
Gleason's D: D =
Pielou's evenness: E =
s n.(n. -1)
1 = 1 N (N - 1) '
S
In N '
Shannon's H
where n.= number of individuals for species i, N = total number of individuals, and S =
number of species.
103
-------
Table B-1. Candidate Benthic Measures Used to Formulate the Benthic Index. T-tests Were Used to Test
Equality of the Means for Each Metric for Impacted Versus Unimpacted Sites in the Test Data Set.
Candidate benthic metrics
Measures of Biodiversity
Shannon's H based on total infauna
Shannon's H based on total epifauna
Shannon's H based on total infauna and epifauna
Simpson's D based on total infauna
Simpson's D based on total epifauna
Simpson's D based on total infauna and epifauna
Gleason's D based on total infauna
Gleason's D based on total epifauna
Gleason's D based on total infauna and epifauna
Pielou's evenness based on total infauna
Pielou's evenness based on total epifauna
Pielou's evenness based on total infauna and epifauna
Measures of Community Condition
Total benthic biomass per site
Mean biomass per grab
Mean infaunal abundance per grab
Mean epifaunal abundance per grab
Total number of infaunal species/site
Total number of epifaunal species/site
Mean number of infaunal species/grab
Mean number of epifaunal species/grab
Measures of Individual Health
Biomass/abundance ratio
Mean weight of individual bivalves
Mean weight of individual molluscs
Mean weight of individual polychaetes
Mean weight of all individual organism except molluscs
Measures of Functional Groups
Mean abundance of dominant species
Mean abundance of opportunistic species
Mean abundance of opportunistic species minus amphipods
Mean abundance of equilibrium species
Mean abundance of suspension feeding species
Mean abundance of deposit feeding species
Mean abundance of omnivore/carnivore species
Measures of Taxonomic Composition
Mean abundance of amphipods
Mean abundance of bivalves
Mean abundance of gastropods
Mean abundance of molluscs
Mean abundance of polychaetes
Mean abundance of capitellid polychaetes
Mean abundance of spionid polychaetes
Mean abundance of tubificid oligochaetes
Ratio of linholl and strebene to all infauna
t-test
(p-value)
< 0.001
0.01
< 0.001
0.003
0.02
0.006
< 0.001
0.007
< 0.001
0.004
0.11
0.007
0.76
0.76
0.11
0.07
< 0.001
0.006
< 0.001
0.007
0.31
0.49
0.46
0.39
0.36
0.08
0.03
0.01
0.1
0.48
0.04
0.003
0.05
0.3
0.03
0.44
0.68
0.17
0.14
0.06
0.58
Direction
(* = greater mean value at
unimpacted sites)
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
104
-------
Estuaries are characterized by large natural variations in certain physicochemical
conditions (e.g., salinity, sediment grain size) known to be major factors controlling the
diversity and abundance of resident biota. Such factors need to be identified and
controlled for before the responses of candidate benthic measures to environmental
stress can be characterized accurately. Pearson correlation coefficients were
calculated to determine relationships between the individual metrics listed in Table B-1
and various physical habitat variables such as sediment silt-clay content, bottom water
salinity, water depth, and latitude. Sediment total organic carbon (TOG) concentration
was not treated as an habitat variable since it can be viewed as a potential
environmental stress similar to low dissolved oxygen and sediment contamination.
Many of the candidate benthic measures were significantly (p < 0.05) correlated
with at least one of the habitat factors measured (Table B-2). However, only seven of
the correlations accounted for a significant proportion of the total variation, defined here
as more than 25% (r2 > .25). Four of these seven were measures of species richness
(total number of infaunal/epifaunal species and mean number of infaunal/epifaunal
species), which were positively correlated with bottom salinity. The other three were
diversity measures (Shannon's H based upon epifauna and Gleason's D based upon
infauna and both epifauna and infauna), which were positively correlated with bottom
salinity. Relationships between the rest of the candidate measures and the other
habitat factors (i.e., latitude, silt-clay content of sediments, bottom water temperature,
and water depth) occurred less frequently and did not account for as much of the total
variation as relationships with salinity (Table B-2).
Table B-2. Summary of Correlations Between Habitat Indicators and the Candidate Benthic
Measures for the Entire 1990-93 Data Set, Using Pearson Correlation Coefficients
(significance for p < 0.05).
Habitat Indicator
Bottom water salinity
Latitude
Bottom water temperature
Silt-clay content of sediments
Water depth
Number of
Significant
Correlations
37
30
32
27
24
Number of
Correlations
with r2 ^ 0.10
11
0
8
9
1
Number of
Correlations
with r2 ;> 0.25
7
0
0
0
0
105
-------
Polynomial regressions for these seven benthic metrics against salinity were
developed according to the procedure discussed in the Weisberg etal. (1993). These
regressions were established by fitting the 90th percentile of a 3 part per thousand
(salinity) moving average of the individual benthic metric versus salinity. Refer to Table
B-3 for the regression coefficients and correlation coefficients for the regressions.
Table B-3. Regression Coefficients and Correlation Coefficients (r2) for Bottom Water Salinity
Normalization Functions for Benthic Measures.
benthic metric
total number of infaunal species
total number of epifaunal species
mean number of infaunal species
mean number of epifaunal species
Shannon's H based upon epifauna
Gleason's D based upon infauna
Gleason's D based upon infauna and epifauna
polynomial for salinity normalization
21.795 - 2.704 sal + .292 sal2 - .0054 sal3
8.354 - 1.641 sal + .1496 sal2- 0.0026 sal3
13.015 - 1.353 sal + .1700 sal2 - .0032 sal3
7.146 - 1.230 sal + .0917 sal2 - .0014 sal3
.594 - .0191 sal + .00243 sal2 - .000044 sal3
3.394 - .366 sal + .0433 sal2 - .000871 sal3
4.283 - .4980 sal + .0542 sal2 - .00103 sal3
r2
•97
.77
.97
.63
.74
.95
.95
The salinity normalized metrics were added to the candidate list of measures.
The new metrics were rechecked for possible correlations with the habitat variables. No
significant correlations (r2 ;> .25) were observed. The complete candidate list of
measures for testing (48 total) included those identified in Table B-1 and the seven
salinity normalized metrics in Table B-3.
Step 2: Develop a test data set
Sites from the 1990-93 data set that had valid benthic assemblage, bottom
dissolved oxygen, bottom salinity, sediment toxicity, and sediment contaminant data
(539 sites) were used to select sites for reconstructing the benthic index. Note that
these sites include probability-based sampling sites as well as judgmental sites.
Continuous bottom dissolved oxygen measurements were only available for 1990-91,
so the point-in-time measurements for dissolved oxygen and salinity were used to
incorporate data from all four years. The criteria that were applied to the 539 sites were
modified from those used in Weisberg et a/. (1993) and Schimmel et al. (1994). The
criteria used here were more restrictive than those previously used (i.e., "better" good
sites and "worse" poor sites).
106
-------
Any of the following criteria needed to be met for a site to be classified as
impacted:
1. Sediment toxicity with survival < 80% and significantly different from controls,
or
2. Sediment contamination with at least one exceedence of Long et al. (1995) BR-
IM values or more than 10 exceedences of ER-L values,
or
3. Bottom dissolved oxygen concentrations < 2 mg/l.
All of the following criteria had to be met for the site to be classified as an
unimpacted site:
1. No sediment contaminant exceeded an ER-M value and no more than three
sediment contaminants exceeded ER-L values,
and
2. No sediment toxicity was observed (i.e., survival greater than 80% and not
significantly different from controls),
and
3. Bottom dissolved oxygen concentration > 7 mg/l.
The selected test data set from the 1990-93 data used for calibration contained
60 sites; 30 were categorized as impacted and 30 as unimpacted (Tables B-4 and B-5).
For each of these two categories, 10 sites were in each of the three salinity zones (< 5,
5-18, and > 18 ppt). These salinity zone categories are the same as used in Weisberg
et al. (1993) and Schimmel et al. (1994). Because we wanted an equal number of sites
in all 6 categories and the minimum number in any category was 10, each category was
trimmed to 10 sites. For those categories with more than 10 sites meeting the criteria,
the sites were selected that represented the best (for unimpacted sites) or worst (for
impacted sites) sites that maintained balance across sediment grain sizes. The
remaining sites (52) that were not used for calibration were set aside for use in the
validation step.
107
-------
1
I
E
(8
O
£
I
CO
TJ
1
O
CO
6>
UJ
Co o
Si CO
!!
11
CO
in
O L.
CO Jl
CM
m
o
CD CO
S
.£5 oo
•59
§^
g$
<>
"fr
CO
o
I?
P
0)
CO CM
CM o
T- en
••*• o5 co
ll
ll
b rr?
en
co
CD
S?
11
0 CM
co in
CO
C
O>
CO
co
CO
!~
«
.> Tf
a:
-i§
CM
50
S
IS
05 co
CO
•*•
CO
CM °
i CO
CO
&i
C ••*•
O O
E°?
11
>» o
CO CO
m co
i CM
CO o
CD CD
CO
CO
CD
Is
CO
8
co£
CD
s
CO o
9> co
O. CD
CD CO
j
O > o > O >
co
IP
CM
C3>
«
N-
DX9
|1
O
CO
O)
CO
en
O) O
in
o =
O T-
r- COO
II
OOo
oco
Co
CM
co oo
m co
CD
03 CO
Q. •*}•
%<3
CD
CM
C^SS
CQ CO
CD
1?
D.CD
»< ||
O > O >
f
fe1*
5
^1
O CD
'(O tf)
i
M in
s
CQ
S
CO Q-
-C S"
•i a
O CO
2T
m us.
CD 'P'
C O.
~ Q.
CO
1
LU
00
108
-------
•5
1
-
E
£
1/5
CO
CO
c
CO
te Includes Est
CO
s:
o
CO
Hi
.8
. Information
CD
TD
C .j
0 ®
'•B 2!
S"S
CD o
£3
"^ o
•§'•=
CO Q.
i E
= OJ
co o
0 §
o 0
*r^
1990-93 Used
i Number, anc
S-2
O CO
"- co
l£
col
•§m
o en
co .£
P
C CO
nco
CO
CD
£3
£
o
1
J5
ites for Calil
CO
•a
CO
t>
8.
E
'E
^
w
w
CO
0
3
!5
n
X
= co !b
„. CM tf CO
¥- ^ co 5f
i Is- CM CM
co in ° °
o o CD in Ti- co
.i CD CM b- o Is-
co •"- 06 oo
9 = 9 =
i_ ?- 2 § CM $5
CD in ^ %. co 5,3
.2 *4* CO •*— .
rv* o o o
~; co co co
•g CO CO CO
O i_ i-
i-r- ScM >CO
g^co E o Q> CM
Di > co > il >
= = £
in T- co
3f En T- r~
„ CM T- O •*
CO T— ._ o i o
CM o CD (x. OO CD
fi t ^ ^ 9^
0 ^ 00 CO
CM 9 = *" =
Si = co co co
05 CO CO CM Tf
Is- OO m CO
CO T- ~ CM
o o TO o
00 Is- f£ CO
CO •* ^ CD •% (Q
E CM qj CO => CO
"o m ^ 3 *^
Q- ^* -^ ^ CO -"*
_
T— £
in = ^- "^ c^ =
0 g CM W ^ CM
.2 T- .2 t- CM CO
rv; o rv' o o o
CO CO 1^ CM
-g CO -g CO CO __ ^
O O i- CO
C C CD >
CO cCO >O « •'fr
caC3 coco fy^ "CM
X:CM JZCD L'-CM c"*
CO (A CO fA W /A O ~i-
II II 11 11
>n
™ x—*.
~ a.
CO Q.
CO m
CD ^
CM ^ =
"7 J~ T- m 'V T-
o SP • m o co
r^- § ^ ti **"
».. co. °3.
00 CO if) ^f
in !* co
in r^ CM
CO £- OO CO Is-
OD CO »_ CO rh CO
0}
CD > CO
y <-~ \x
co •*- Cd co co ^
CD O) _ CO 0) IO
Q. CN .s= CD o O
CO T— . ~5£ co tO Q
li li li
= =
00 •* =
m co in
23 & or S g
i o _1 CD iJ o
oo CD § ^- o co -
« > i^ i^ i^ i^
CO CO CM
CO CO CO
o o o
OO CO CO
%: CO -a CO -o CO
.1 § §
Qi CD ,9 m ,9 co
w CO CO •sf CO CO
CO CM • — CO — ^~
«S CO COCO D) CO
H •> 1— ->
_
- ^— = js^
o| oP co§ o;5
CO ° t~ in CM ?— 00 CD
CM ^" O O O
OT . CO ^_ CM = CO =
in c° TJ- c° "3 m
3*" CO CM
Is- T- 1^ Tj-
CQCO CDCO jj5C'5 CQCO
CD g .> CD
Q.O 0s? CO"^ o O
[§CM ._yco ECM {go
~ ^*
O CO
CO T-
m 10,
= =
,- |D M- S 50 ^:
i Cj |* ^_ i^ c>
cb CN ^ t*^ tQ S
Oh- CZ) i "^
8 - " 8 v g &,
CM S <°
?t CO T-
CO CM CO
O >•» O *T3 O
•<- CO T- C T-
co ^— o
CD « CO
a? Co OT T— *- r*-
D> ^~ CO O "o O
•z.> •z.> •z.>
CD :
m = in
CM ._,. Is- T •
F 98 cog
Sj^ OCM S^-
ir- •* in m co
CO O5 •* "O
CM "0 CD CO
o Co o
CO 3 T- T-
CO O ^1" ^*
co •§
>,co "c Is- •£ m
•*-* ^~ ~" CO CM
CO CO OJC3J CD CO
(D < C < •* <
o •"" J3 ^ 5 •->
CO ?- =
co in Is-
T- CD CM S" S
T- f CM >- !>
§s gg ^g g?
CO O CO 9
CO CM CD 9 = 2 =
•* OJ •<*• T- 0 °> CO
co co co ^f jr
^- in T- co oo
O o o o o
S Is- Is- h- CO
g CO CO CO CO
00 ro ^ &
CUO3 mLO mf^ ^"
-gCO ^CD gCD g>0
li fi |i ||
CD C-
c a.
•c a.
i?
w ,
LLJ -^
109
-------
Steps 3 and 4: Identify combinations of candidate benthic measures that
discriminate between unimpacted and impacted sites, and validate the
combination of metrics
A series of discriminant analyses were run in succession to identify the benthic
measures from Tables B-1 and B-3 which best discriminated between the impacted and
unimpacted sites in the test data set and validated with the validation data set (Table B-
6 and B-7). The SAS procedure for stepwise discriminant analysis, STEPDISC (SAS,
1990b), was used to select the combination of metrics for further analysis. The SAS
procedure DISCRIM (SAS, 1990a) was then used to determine the discriminant
function from the parameters selected by STEPDISC. Targets of 90% correct
classification for calibration (using the calibration data set, Tables B-4 and B-5) and
80% for validation (using the validation data set, Tables B-6 and B-7) were set as goals
for selection of metrics in the benthic index. The cross-validation classification feature of
DISCRIM was used to determine the sensitivity of the index to individual sites in the
calibration data set.
The results of the various discriminant analyses are summarized in Table B-8. Six
variables were included in the model generated by the first stepwise discriminant
analysis (Index 1): 1) salinity normalized Gleason's D based upon infauna, 2) tubificid
abundance, 3) epifaunal abundance, 4) polychaete abundance, 5) Pielou's evenness
based upon infauna, and 6) Shannon's H based upon infauna. This combination of
metrics correptly classified 93% of the impacted sites and 83% of the unimpacted sites
(Table B-8). The cross-validation feature of the discriminant analysis procedure showed
that this index had correct classification of 90% for impacted and 77% for unimpacted
sites (misclassification mostly for low salinity sites). The canonical r2, which
approximates the total variance explained by the analysis, was 0.65. This index missed
the target goal for correct classification, had problem with correctly classifying
unimpacted sites for cross-validation, and met the validation target goals. Pielou's
evenness based upon infauna entered the discriminant function with negative
coefficient, in contrast to how benthic communities are expected to respond with this
diversity metric. Exploratory analysis indicated that salinity normalization for the tubificid
abundance might be required to correct for misclassification of low salinity unimpacted
sites.
The second index was developed incorporating a salinity normalized tubificid
abundance metric. The salinity normalization for tubificid abundance required a
different procedure than that used to normalize the other benthic metrics. Tubificids are
only observed for low salinity water, with some occurrence being normal for unimpacted
sites. Impacted sites would be characterized by large tubificid abundances. The salinity
normalization selected was:
110
-------
U)
w"
CD
CD
"55
HI
_3
O
i
CO
o
co
HI
o
fc CO
t (0
II
gi
T3 C3
T} 2
£•§.
§ &
m g>
CD CD
0) T3
1 s
4P CD
OT C
Z) 0
o>
T™~
E
Impacted Sib
Date of Sam|
CO
CQ
_CD
JD
i5
dation
mpacted Sites
CO
co
co
o
^ O5
9> co
5
CO
CD <
a :>
CD CN
CN o
CN
"5
_
CD CO
CN
II
co
in
03
55
!
F-
CM
°> in
CN o
if
0> &5
i ^d~
oo T-
CQ co
CD
NX
co fS
|S
CO
CD
0
co
in
CM o
»" co
CN
CM
v-
o
t_ co
CD C*3
£ CM
I"
il
0. >
00
<0f*.
O TJ-
CD
CO
D
CD f
_>
o oo
co V
£ o
CO
Chesapeake Bay
VA91-431 38°0'
m ^.
T in
CO CO
O 0
o3 r-
> rj-
O
=
CD
CO
a: >
o
9£
CO o
9 s:
CO h~
O)
co
co
>,co
to °
I
co
o
CD
"ro£
is
OCM
>CM
in
Is
O.LO
8 CN
a>
CO ?-I
o
o co
co
CM
I
m
in
m
CD co
CN=
cnco
co
CN
9 |v~
co
o
o> £
CM
m
00
co
CO
II
CM
in
co
°
o
—
—
Jo CM
g.N-
i
0=
CM
!"§
^ c\i
coo
1
CO
co
F--
O) =
o
CD
>
11
O3
cpco
$
>•
DQl°
CM
^
" =
O CO
gcS
CN
in '*
O
N-
co
i- CO
o cb
•2 co
CO o
11
CN
co
o
?N
co
CD
"
^0
i 6>
O>
CN
CN
CO
in
Q.CD
TOCO
m05
2
i
CO
£°
_T-
5^
r-
05 18
O5
Is
-go
^6
roo>
cs
O
la
co
s= Q-
CO O.
CO LO
CD Z5~
c a.
•c a.
o co
E V
m m
111
-------
_0
4^
(0
;o
1
S
I
CO
1
u
§.
"E
z>
U)
V
o
ID
oo
co
ro
o
OO
CO
3
O
CO
t_
.92
co
o
IO
>.?
mS
-^m
gs
0.0
II
6 >
CO 10
O o
J. CD
in
cb
8
c
co
ro S
ce >
CD
CO CM
o
CO
co
o
_ oo
^ CO
3
CO co
,«
05 .
«? «
oo ^-
o o
cb
0
,5 co
CQ CO
(U
is •<-
5n co
S.CD
co
o >
OO
O
co
m
CO
CO
co po
S.5?
o
|v. CO
CM
oo
CO
ro
o
CO
co
3
O
CO
10 °
CM fe>
ro m
CM
oo
CO
>*co
co oo
m -a-
0 CM
'E 3
o <
:s >
—.co
^ m
i CD
OO T—
o o
.c
o
co
o
m
in
co
CO
*--^
CUCM
o^T
3CO
1
CM
CMCO
Oo
co
°> co
co
in
co
CQ co
<5 ^
oco
CO
8
o
<35
CO
COO
CM
CM
ro
O)
in
CM
CM
CO
*--iv.
cT"
CO
if
ro 3
® ^
Q >
CO
I"
CO
T3 CO
3 o7"
CO 5
o
'
S-.-CO
§
CO
CM
com
CO .
<"£
CJ>
coo
5o
rob)
cMin
>«co
CO
fi*
59
0)
<0
f
CD t--
CM
CJ> =
m
ro
co
•o
"3 ro
o 8
co "?
TO O)
2.o
TOCO
.11
co>
"o
co-r-
CM
en"
o^.
OO h-
CO
•a F-
co •*
•a
JO o
"3 co
15 ro
o <
V£°
o>h-
-a £P
COO
co
ro
CQ co CQ co
(1) Q)
^ ^
co ^3" co ^*
(U in
f^co
N!_O
ss
o)
co
CM
co CD
CM 60
§fe
o
ro .
S-"
^N.
-------
CO
§
o
O5
T—
0)
•a
o
O
o
CD
m
M—
O
X
0)
•
CO
CO
§
CO
CO
0)
o
CD
CO
3
0)
15
o
O
T3
I
•
O
0
c
_o
as
•o
15
>
c
,o
ffl
>
1
18
8
u
c
.0
2
.Q
"15
u
^•a
'55 t3 $
to re a*
en d o)
o E
•D
tls
JS E '»
0 c
3
CO IB -2
(Q D. u)
o E
•O
S« g
1 E '5
"5 'E
3
til
«f "
"g
f §j
•§ — *"
3
n
u
ST.
c
n
1
0)
'5
o
o
CO
£
3
U
m
CD
Selected
_co
"55
" ^H
n
^s
ro
sP
&s
in
oo
0
O)
g°
is-
^
CO
Cf>
CO
ro
in
CO
CD
•* ro co f- °i CM
•«- en ;.- ro in o
CN CO CO •* f^~ CD
X— 1 1 I
co
cr
i
c
o
CL
ZJ
-a co
CD ^
CO ^
co ,J5
•Q c: co
« | &
c. £j- c
o ^ —
• salinity normalized Gleas
• tubificid abundance
• epifauna abundance
• polychaete abundance
• Pielou's evenness based
• Shannon's H based upon
,-
X
0)
•o
OS
ro
o
°
ro
CO
o>
s^
ro
^
j^
co
o
o>
CD
CD
gte|gsg3
CM r — ^^ f1^ ^- QQ
•^ i- o' r r •
co
c.
1 8
C =
o 45
§• '°-
— > (i\
-0 CD CO ^
O5 d 3 C
CO CO .CO CO
ol| 1
O ~T
m •** o CO
""•S 1- =
o ^ 3 —
• salinity normalized Gleasi
• salinity normalized tubifici
• epifauna abundance
• Pielou's evenness based
• polychaete abundance
• Shannon's H based upon
CN
X
o
•o
•£ 113
OS
in
CO
o
T—
ro
co
o>
go
O
O)
^
CO
CD
O
CD
CO
O
iss
co
c:
I
c
o
CL
-o v
% =
5|
Q §
-S-8
n T3
• salinity normalized Gleasi
• salinity normalized tubifici
• epifauna abundance
M
X
0>
•o
c
5s
ro
o
ro
ro
gS
ro
go
ro
0*
o
CD
^
ro
o
co
CD
CD r- in
OO 10 is.
CO CO CO
T— 1 1
CO
c
1
c
o
Q.
-0 • JD
% =
S •§
Q § ni
.ǥ Ǥ
0 T3 CO
• salinity normalized Gleasi
and epifauna
• salinity normalized tubifici
• spionid polychaete abund
,,
X
0>
_c
5s
ro
-
ro
ro
o**
f**^
ro
5s1
ro
o
CD
O
O)
O
CD
CD
CN m h~ ^
S •* CD ^i.
CN CD OO ^1
t— 11*
CO
c
13
c
o
CL
T3 CD
CD £2
CO CO
-Q ^3 co
i-» "O CO —
• salinity normalized Gleas<
and epifauna
• salinity normalized tubifici
• spionid polychaete abund
• Shannon's H based upon
IO
X
S
•o
5s
T—
co
0
°
oo
5?
CO
a>
5^
co
co
'0
»,^
o>
o
• co
S
CD
g in CO fO if)
o co (Q in co
T— 1 1
CO
CD
'o
g.
CO
"co
1 §
'*- co
° = 2
Q) ^ C
•i ro J
rs "O C
• salinity normalized total m
• salinity normalized tubifici
• Gleason's D based upon i
• polychaete abundance
• epifauna abundance
co
^
fl)
•o
< §
CO '>
M-2
b •£
CD 2
o; LLI
CO
DO
-------
salinity normalized tubificid abundance =
tubificid abundance - 500 * exp(-15* bottom salinity),
where exp(...) denotes the exponential function. Note that salinity normalized tubificid
abundance is negative for unimpacted sites and positive for impacted sites.
The six variables for Index 2 were (Table B-8): 1) salinity normalized Gleason's
D based upon infauna, 2) salinity normalized tubificid abundance, 3) epifaunal
abundance, 4) Pielou's evenness based upon infauna, 5) polychaete abundance, and
6) Shannon's H based upon infauna and epifauna. This combination of metrics
correctly classified 97% of the impacted sites and 90% of the unimpacted sites. The
cross-validation feature of the discriminant analysis procedure showed that this index
had correct classification of 93% for impacted and 87% for unimpacted sites, correcting
for the problem with Index 1. The canonical r2 was 0.67. This index met the target goal
for calibration, had no problem with cross-validation, and met the validation target goal.
However, Pielou's evenness based upon infauna and polychaete abundance both
entered the discriminant function with negative coefficients, in contrast to how benthic
communities are expected to respond with these metrics.
Index 3 was developed similarly to Index 2, but only selecting the first three
metrics identified by the stepwise discriminant analysis. This eliminated the variables
that contributed to the discriminant function in an unreasonable ecological way. The
three variables for this index were (Table B-8): 1) salinity normalized Gleason's D
based upon infauna, 2) salinity normalized tubificid abundance, and 3) epifaunal
abundance. This combination of metrics correctly classified 93% of the impacted sites
and 90% of the unimpacted sites. Cross-validation showed that this index had correct
classification of 93% for impacted and 90% for unimpacted. The canonical r2 was 0.61.
This index met the target goal for correct classification, had no problem cross-
validation, but missed the validation target goal for the impacted sites.
Index 4 was developed by removing salinity normalized Gleason's D based upon
infauna (the first metric to enter for Index 3) from the stepwise discriminant analysis and
keeping the first three metrics identified. The three variables for this index were (Table
B-8): 1) salinity normalized Gleason's D based upon infauna and epifauna, 2) salinity
normalized tubificid abundance, and 3) spionid abundance. This combination of
metrics correctly classified 87% of the impacted sites and 90% of the unimpacted sites.
Cross-validation showed that this index had correct classification of 87% for impacted
and 87% for unimpacted. The canonical r2 was 0.60. This index just missed the target
goal for calibration, had good cross-validation, and met the validation target goal.
Index 5 was similar to Index 4 but kept all of the metrics identified by the
stepwise discriminant analysis. The four variables for this index were (Table B-8): 1)
114
-------
salinity normalized Gleason's D based upon infauna and epifauna, 2) salinity
normalized tubificid abundance, 3) spionid abundance, and 4) Shannon's H based upon
infauna. This combination of metrics correctly classified 90% of the impacted sites and
unimpacted sites. Cross-validation showed that this index had correct classification of
90% for impacted and 87% for unimpacted. The canonical r2 was 0.60. This index met
the target goal for calibration, had good cross-validation, and met the validation target
goal. This index provided marginal improvement over Index 4 by including a metric with
a low coefficient in the discriminant function
Index 6 was developed similarly to Index 4, but by removing the salinity
normalized Gleason's D based upon infauna and epifauna (the first metric to enter for
Index 4) from the stepwise discriminant analysis. The five variables for this index were
(Table B-8): 1) salinity normalized total number of infaunal species, 2) salinity
normalized tubificid abundance, 3) Gleason's D based upon infauna, 4) polychaete
abundance, and 5) epifaunal abundance. This combination of metrics correctly
classified 90% of the impacted sites and 97% of the unimpacted sites (Table B-8).
Cross-validation showed that this index had correct classification of 83% for impacted
and 93% for unimpacted. The canonical r2 was 0.64. This index met the target goal for
correct classification, had a slight problem with cross-validation, and met the validation
target goal. Polychaete abundance entered the discriminant function with negative
coefficient, in contrast to how benthic communities are expected to respond with this
metric.
Selection of Index for Use in Four-Year Assessments
Index 4 was selected as the benthic index to use for the four-year assessments
in this report. This index just missed the target for calibration for unimpacted sites, but
correct classification of one additional site would have allowed this index to meet the
target goal. Marginal improvement of calibration classification was achieved with Index
5, but it was decided that inclusion of a metric with a low coefficient in the discriminant
function was not significant enough to warrant use, A spatial pattern analysis using all
of the 1990-93 data was conducted with both of these indices to determine if Index 5
provided any spatial difference in identification of impacted areas. No directly
observable difference in patterns could be discerned between use of Index 4 and Index
5.
Cross plots of Index 4 against salinity and grain size for the calibration and
validation data sets are show in Figures B-3 to B-6. No discernible bias with respect to
these habitat variables are apparent in these figures.
Since a balance in number of sites between impacted and unimpacted
categories for the test data set was used, the demarcation in the discriminant function
score between impacted and unimpacted sites was zero. No scaling of the benthic
115
-------
35
30 ...
U 25 ..
CL
m 20..
w
j 4
1 1° --
5 ..
0 impacted site i
| unimpacted site {
-* •! »
-3
-2
-1 0 1
1990-93 Benthic Index
B-3. The 1990-93 benthic index versus bottom water salinity for 1990-93 calibration data set.
Negative benthic index values indicate impacted conditions.
100
90 ..
H 80 ..
I 70..
•*-*
J 60 ..
I 5°-
I 40"
| 30 ..
I o
I 20 -
W 10..c
0 ._
0
0
8
o
°
o
BO
0 impacted site ]
m unimpacted site'
-f-
H-
* •
-3
-2
-1 0 1
1990-93 Benthic Index
Figure B-4. The 1 990-93 benthic index versus sediment silt-clay content for 1 990-93 calibration data
set. Negative benthic index value indicate impacted conditions.
116
-------
oo -
30 -
S 25 -
a.
£•
| 20 -
n
CO
^
1 15-
g
E
| 10 '
5 -
o
Figure B-5.
120
c- 100
I
Q_
t 80
C
O
0 6C
re
a
co 4C
•s
0)
E
1 2C
CO
c
7
.
.
.
I .
) .
)
" •-*"•""* "
0
o « • •
0 0
o
o impacted site $ •
i o .0
aunimpacted site) ° <> • * *
^
o o
* «
o
-5-3 -1-1 3 5
1990-93 Benthic Index
The 1990-93 benthic index versus bottom water salinity for 1990-93 validation data set.
Negative benthic index values indicate impacted conditions.
0 0
0 o o
A °
- o impacted site
m unimpacted site 0 o 0
4 ' "
o u
o
o
0° "
O O ,. B
• 0 0
o •
o • m •
»°* B11-.— B
1 1 IB mfm m 1
7 -5 -3 -1 1 3 £
1990-93 Benthic Index
Figure B-6. The 1990-93 benthic index versus sediment silt-clay content 1990-93 validation data set.
Negative benthic index values indicate impacted conditions.
117
-------
35
30 ..
25-
1 20 4-
ra
CO
15 ..
10 ..
5 ..
0
o
o o
o impacted site
B unimpacted site
8
o
o
.o o
*•*
..V'
o
-I*-
-7
-5
-3 -1 1
1990-91 Benthic Index
Figure B-7. The 1990-91 benthic index versus bottom salinity for combined four-year calibration and
validation data sets. Negative benthic index values indicate impacted conditions.
index was required as was necessary in prior versions of the benthic index (Weisberg et
a/., 1993; Schimmel et a/., 1994).
As a check to see if there was indeed improvement with the 1990-93 benthic
index over the 1990-91 index, the latter index was applied to the four-year calibration
and validation data sets. For the calibration data set, the 1990-91 index correctly
classified 77% of the unimpacted sites and 90% of the impacted sites. For the for the
validation data set, the 1990-91 index correctly classified 92% of the unimpacted sites
and 65% of the impacted sites. The overall efficiency of the 1990-91 index was 85% for
unimpacted sites and 78% for impacted sites. The results are displayed in Figure B-7. A
salinity bias exists for the unimpacted sites. The overall efficiency of the 1990-93 index
combined data sets was 86% for unimpacted sites and 86% for impacted sites.
Calculating the 1990-93 Benthic Index
The three benthic metrics in the index are: salinity normalized Gleason's D based
upon infauna and epifauna, salinity normalized tubificid abundance, and abundance of
spionids. The diversity measure is associated with unimpacted conditions (positive
contribution). The latter two measures are associated with impacted conditions (negative
118
-------
contribution), with tubificid abundance important in low salinity waters and spionid
abundance important in higher salinity waters. The discriminant function calculation
normalizes the individual metrics using the mean and standard deviation for the metric in
the test data set used for calibration.
The formula for the EMAP Virginian Province 1990-93 benthic index is:
1389 * (salinity normalized Gleason's D based upon infauna and epifauna - 51.5) 728.4
-0.651 * (salinity normalized tubificid abundance - 28.2) /119.5
- 0.375 * (spionid abundance - 20.0) / 45.4,
where
salinity normalized Gleason's D based upon infauna and epifauna =
Gleason's D / (4.283 - 0.498 * bottom salinity
+ 0.0542 * bottom salinity2
- 0.00103 * bottom salinity3) * 100
and
salinity normalized tubificid abundance =
tubificid abundance - 500 * exp(-15 * bottom salinity),
and
exp(...) denotes the exponential function.
119 T&U.S. GOVERNMENT PRINTING OFFICE: 1999 - 550-101/20012
-------
-------
-------
-------
-------
3 <
"O
m
en
------- |