EPA/620/R-95/007
September 1995
Quality Assurance Report
EMAP-Virginian Province
1990-1993
by
Charles J. Strobel
United States Environmental Protection Agency
Narragansett, Rl
and
Raymond M. Valente
Science Applications International Corporation
Narragansett, Rl
Darryl Keith
EMAP-Virginian Province Manager
United States Environmental Protection Agency
Narragansett, Rl
Kevin Summers
EMAP-Estuaries Technical Director
United States Environmental Protection Agency
Gulf Breeze, FL
Joseph LiVoIsi
Quality Assurance Officer
United States Environmental Protection Agency
Narragansett, Rl
United States Environmental Protection Agency
National Health and Environmental Effects Research Laboratory
Atlantic Ecology Division
Narragansett, Rl
-------
ABSTRACT
This report documents the results of Quality Assurance activities conducted in conjunction with sampling
performed by EPA's Environmental Monitoring and Assessment Program's Estuaries study (EMAP-Estuaries)
in the Virginian Province from 1990 through 1993. As part of the planning stage for each years activities, a QA
Plan was developed. All sampling and analytical activities were required to be conducted in accordance with
the prescribed methods, and following the standards stated in the QA Plan. This report discusses the results of
Quality Assurance activities by indicator, data qualifier flags, data quality, and, where appropriate, discusses
lessons learned and proposes changes or solutions to improve data quality.
Data collected in the Virginian Province from 1990 to 1993 were generally of high quality. A total of 446
Base Sampling Sites were scheduled for sampling over this period. Twenty one stations were eliminated due
to inadequate water depth or logistical concerns. With the exception of total suspended solids (samples for this
indicator were not collected in 1990), the success rate for all indicators exceeded 80% (percent of stations with
data passing QC), with most exceeding 85%.
Some significant problems were encountered in the chemical analysis of sediment samples resulting in the
deletion of some data from the database. The specific problems, and a discussion of the data deleted or qualified
are included in this report.
Page ii Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
DISCLAIMER
Mention of trade names or commercial products does not constitute endorsement by the Environmental Protection
Agency or recommendation for use.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page iii
-------
PREFACE
Contractor support for the preparation of this document was supplied via contract number 68-C1 -0005
to Science Applications International Corporation.
The appropriate citation for this report is:
Strobel, C. J. and R.M. Valente. 1995. Quality Assurance Report: EMAP-Virginian Province, 1990-1993. United
States Environmental Protection Agency, National Health and Environmental Effects Research Laboratory,
Atlantic Ecology Division, Narragansett, RI. September 1995. EPA/620/R-95/007.
This report is AED contribution # 1639.
Page iv Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
CONTENTS
ABSTRACT ii
DISCLAIMER Hi
PREFACE iv
CONTENTS v
TABLES vii
ABBREVIATIONS x
1 INTRODUCTION 1
2 FIELD CREW TRAINING AND AUDITS 2
2.1 1990 Results 2
2.2 1991 Results 4
2.3 1992 Results 5
2.4 1993 Results 5
3 OA RESULTS FOR CHEMICAL CONTAMINANT ANALYSES OF SEDIMENTS 7
3.1 Background 7
3.2 Data Qualifier Codes for Chemistry 8
3.3 Quality Assessment Results 9
3.3.1 Laboratory Audit 9
3.3.2 1990 OA Results 9
3.3.3 1991 OA Results 17
3.3.4 1992 OA Results 22
3.3.5 1993 OA Results 27
4 OA RESULTS FOR FISH CONTAMINANT ANALYSES 37
4.1 Background 37
4.2 1991 OA Results 37
5 OA RESULTS FOR PARTICLE SIZE ANALYSES 41
5.1 Background 41
5.2 Laboratory Audits 41
5.3 Qualifier Codes for Particle Size Data 41
5.4 1990 OA Results 41
5.5 1991 OA Results 42
5.6 1992 OA Results 42
5.7 1993 OA Results 42
6 OA RESULTS FOR SEDIMENT TOXICITY TESTING 43
6.1 Background 43
6.2 Data Qualifier Codes for Sediment Toxicitv 43
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page v
-------
Contents (continued)
6.3 Laboratory Audits 43
6.4 1990 OA Results 44
6.5 1991 OA Results 45
6.6 1992 OA Results 45
6.7 1993 OA Results 45
7 OA RESULTS FOR MACROBENTHIC COMMUNITY ASSESSMENTS 46
7.1 Background 46
7.2 Laboratory Audits 46
7.3 Data Qualifier Codes for Benthic Community Analyses 46
7.4 1990 OA Results 46
7.5 1991 OA Results 47
7.6 1992 OA Results 48
7.7 1993 OA Results 48
8 OA RESULTS FOR FISH COMMUNITY STRUCTURE AND PATHOLOGY 49
8.1 Background 49
8.2 Audits 49
8.3 Data Qualifier Codes for Fish Community Structure and Pathology 50
8.4 1990 OA Results 50
8.5 1991 OA Results 52
8.6 1992 OA Results 53
8.7 1993 OA Results 54
8.8 Lessons Learned 55
9 OA RESULTS FOR WATER QUALITY MEASUREMENTS 56
9.1 Background 56
9.2 1990 Calibration and Calibration Check Procedures 56
9.3 Data Qualifier Codes for Water Quality 58
9.4 1990 OA Results 63
9.5 1990 Lessons Learned/Changes for 1991 66
9.6 1991 OA Results 70
9.7 1992 OA Results 73
9.8 1993 OA Results 74
10 OA RESULTS FOR TOTAL SUSPENDED SOLIDS ANALYSES 77
10.1 Background 77
10.2 Data Qualifier Codes for Total Suspended Solids Data 77
10.3 Audits 78
10.4 1991 OA Results 78
10.5 1992 OA Results 78
10.6 1993 OA Results 78
10.7 Lessons Learned and Changes Suggested 79
11 SUMMARY OF DATA COLLECTION SUCCESS 80
12 REFERENCES 81
Page vi Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
TABLES
Table 3-1. Summary results for SRM 2704 (Buffalo River Sediment) used as a set control for the
1990 Virginian Province sediment inorganic analyses 10
Table 3-2. Results for SRM 1941 (Organics in Marine Sediment) used as the set control (Laboratory
Control Material) for the 1990 Virginian Province sediment PAH analyses 13
Table 3-3. Results for SRM 1941 (Organics in Marine Sediment) used as the set control (Laboratory
Control Material) for the 1990 Virginian Province sediment PCB/pesticide analyses .... 14
Table 3-4. Range in detection limits (in ng/g dry weight) reported for organic compounds in 1990
Virginian Province sediment samples 15
Table 3-5. Summary results for CRM BCSS-1 (Estuarine Sediment) used as a set control for the
1991 Virginian Province sediment inorganic analyses 18
Table 3-6. Results for SRM 1941 (Organics in Marine Sediment) used as the set control (Laboratory
Control Material) for the 1991 Virginian Province sediment PAH analyses 20
Table 3-7. Results for SRM 1941 (Organics in Marine Sediment) used as the set control (Laboratory
Control Material) for the 1991 Virginian Province sediment PCB/pesticide analyses .... 21
Table 3-8. Summary results for CRM BCSS-1 (Estuarine Sediment) used as a set control for the
1992 Virginian Province sediment inorganic analyses 23
Table 3-9. Results for SRM 1941 (Organics in Marine Sediment) used as the set control (Laboratory
Control Material) for the 1992 Virginian Province sediment PAH analyses 24
Table 3-10. Results for SRM 1941 (Organics in Marine Sediment) used as the set control (Laboratory
Control Material) for the 1992 Virginian Province sediment PCB/pesticide analyses .... 25
Table 3-11. Summary results for CRM BCSS-1 (Estuarine Sediment) used as a set control for the
1993 Virginian Province sediment inorganic analyses 28
Table 3-12. Results for SRM 194la (Organics in Marine Sediment) used as the set control
(Laboratory Control Material) for the 1993 Virginian Province sediment PAH analyses . 32
Table 3-13. Results for SRM 194la (Organics in Marine Sediment) used as the set control
(Laboratory Control Material) for the 1993 Virginian Province sediment PCB/pesticide
analyses 33
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page vii
-------
Tables (continued)
Table 3-14. Results of reanalysis of sediments from Station 725 34
Table 4-1. Summary results for CRMs DOLT and DORM (Dogfish liver and muscle tissue,
respectively) used as a set control for the 1991 Virginian Province fish tissue inorganic
analyses 38
Table 4-2. Performance evaluation results for analysis of organic contaminants in tissue. Average
reported values are based on 11 separate analyses of SRM 1974 (Organics in Mussel
Tissue) performed on different days 39
Table 4-3. Results of laboratory-fortified matrix spikes analyzed with each batch of fish tissue
organic samples analyzed 40
Table 6-1. QA Qualifier Codes associated with sediment toxicity data 44
Table 7-1. Results of recounts performed by the laboratory processing benthic infauna samples in
1990 47
Table 7-2. Results of recounts performed by the laboratory processing benthic infauna samples in
1991 47
Table 7-3. Results of recounts performed by the laboratory processing benthic infauna samples in
1992 48
Table 7-4. Results of recounts performed by the laboratory processing benthic infauna samples in
1993 48
Table 8-1. Qualifier codes for fish pathology data 51
Table 8-2. 1990 Pathology QA results based on laboratory examination offish crews believed to
have a pathology and reference, "pathology-free" fish 52
Table 8-3. 1991 Pathology QA results based on laboratory examination offish crews believed to
have a pathology and reference, "pathology-free" fish 53
Table 8-4. 1992 Pathology QA results based on laboratory examination offish crews believed to
have a pathology and reference, "pathology-free" fish 54
Table 8-5. 1993 Pathology QA results based on laboratory examination offish crews believed to
have a pathology and reference, "pathology-free" fish 55
Table 9-1. Summary of calibration procedures used for Virginian Province water quality instruments
in 1990 57
Table 9-2. Field calibration checks performed during the 1990 Virginian Province Demonstration
Project 57
Page viii Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Tables (continued)
Table 9-3. Data qualifier codes attached to 1990 - 1993 CTD water quality data 59
Table 9-4. Data qualifier codes attached to Hydrolab water quality data 63
Table 9-5. Results of CTD dissolved oxygen field QC checks used during the 1990 Demonstration
Project 64
Table 9-6. Results of 1990 post-sampling season CTD data review 65
Table 9-7. Results of calibration checks following retrieval of Hydrolab Datasonde 3 instruments for
1990 Virginian Province monitoring 66
Table 9-8. Summary of water quality instrument field calibration checks for 1991-93 Virginian
Province monitoring 67
Table 9-9. Summary of test in which all water quality instruments were placed in a well-mixed tank
of seawater prior to the 1991 field season 69
Table 9-10. Results of weekly calibration checks of water quality instruments used in the Virginian
Province, 1991 71
Table 9-11. Results of 1991 post-sampling season CTD data review 72
Table 9-12. Results of calibration checks following retrieval of Hydrolab Datasonde 3 instruments for
1991 Virginian Province monitoring 72
Table 9-13. Results of weekly calibration checks of water quality instruments used in the Virginian
Province, 1992 74
Table 9-14. Results of 1992 post-sampling season CTD data review 74
Table 9-15. Results of weekly calibration checks of water quality instruments used in the Virginian
Province, 1993 75
Table 9-16. Results of 1993 post-sampling season CTD data review 76
Table 10-1. Data Qualifier Codes for Total Suspended Solids Data 77
Table 11-1. Summary of collection and processing status of samples collected in 1990-1993 80
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page ix
-------
ABBREVIATIONS
AED Atlantic Ecology Division of NHEERL (formerly ERL-N)
AVS Acid Volatile Sulfide
BSS Base Sampling Site
CDF Cumulative Distribution Function
CTD Conductivity, Temperature, Depth datalogger
DBT Dibutyltin
DO Dissolved Oxygen
dry wt Dry weight
DS3 Hydrolab DataSondeS datalogger
EMAP Environmental Monitoring and Assessment Program
EMAP-E EMAP-Estuaries
ERL-N Environmental Research Laboratory, Narragansett (renamed AED)
MET Monobutyltin
mg/L milligrams per liter = parts per million (ppm)
mg/kg milligrams per kilogram = parts per million (ppm)
kg/m3 kilograms per cubic meter
NHEERL National Health and Environmental Effects Research Laboratory (U.S. EPA)
ND Not Detected
ng/g nanograms per gram = parts per billion (ppb)
PAH Polycyclic Aromatic Hydrocarbon
PCB Polychlorinated Biphenyl
QA Quality Assurance
QC Quality Control
SEM Simultaneously Extracted Metals
SQC Sediment Quality Criteria
TBT Tributyltin
Mg/g micrograms per gram = parts per million (ppm)
jU Micron
%o parts per thousand (ppt)
Pagex
Quality Assurance Report, EMAP-Virginian Province 1990 - 1993
-------
Section 1
Introduction
The Estuaries component of EPA's Environmental Monitoring and Assessment Program (EMAP-E)commenced
in 1990 with a Demonstration Project in the estuaries of the Virginian Biogeographic Province (mid-Atlantic
coast from Cape Cod, Massachusetts to Cape Henry, Virginia). Following the successful completion of this Demonstration
Project, EMAP-E monitoring in the Virginian Province (EMAP-VP) has continued on an annual basis through
1993. Complete descriptions of the EMAP-E monitoring approach and rationale, sampling design, indicator
strategy, logistics, and data assessment plan are provided in the Near Coastal Program Plan for 1990: Estuaries
(Holland 1990).
The EPA mandatory Quality Assurance (QA) Program requires that every environmental monitoring and
measurement project have a written and approved quality assurance project plan (QAPP). As such, a QAPP was
prepared for the 1990 Virginian Province Demonstration Project (Valente et al. 1990), and this plan has since
been revised in each subsequent year of monitoring in the Province (Valente and Schoenherr 1991; Valente et
al. 1992; Valente and Strobel 1993). The QAPP prepared each year describes the quality assurance and quality
control activities and measures that are implemented in the Province to ensure that the data meet certain established
criteria (i.e., measurement quality objectives).
The purpose of this report is to present and interpret the results of the various quality assurance activities
and quality control checks which have been performed over the first four years of monitoring in response to the
requirements of the Virginian Province QAPPs. As the various QA results are presented and discussed, an attempt
is made throughout the report to describe changes and "lessons learned" as the EMAP-VP QA Program has evolved
overthepastfouryears. In addition to this document, Quality Assurance Annual Report and Work Plans (QAARWPs)
have been prepared for each year of EMAP-E monitoring since 1990 (Valente 1991a; Valente 1991b; Latimer
1992; Summers 1993).
All field work was conducted by Science Applications International Corporation (SAIC), Versar Inc., or a
consortium of universities under the leadership of the University of Rhode Island (URI).
A table summarizing the percent of all data collected passing QC (i.e., data completeness) is presented in
Section 11.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 1
-------
Section 2
Field Crew Training and Audits
Monitoring for the EMAP-VP Program consists of intensive annual sampling by multiple fieldcrews operating
from small boats during a two month summer index period. EMAP-Ehas developed Standard Operating Procedures
(SOPs) for its field activities to insure the comparability of data collected by different teams operating across
wide geographic distances (i.e., both within and among provinces). EMAP-VP has instituted an annual cycle
involving rigorous field crew training and subsequent field performance reviews to insure uniform adherence
to Standard Operating Procedures.
Training sessions lasting from four to eight weeks typically occur immediately prior to the summer sampling
interval. Training involves a combination of both formal classroom instruction and "hands-on" practical experience
to impart necessary skills in everything from first-aid and seamanship to sample shipping and computer use.
As an essential aspect of the QA program, all fieldcrews must pass a final proficiency exam (i.e., "certification")
at the end of the training session before they are permitted to begin actual sampling. In addition, at least once
during the sampling interval, a formal field QA audit is conducted to ascertainthat SOPs continue to be followed.
In addition to the audit conducted by the QA Officer, a performance review of each crew is performed by senior
Program personnel. Written examinations and results of performance reviews are maintained as permanent record
by the Program.
The training certification exam and the subsequent field performance reviews typically are conducted by the
Province QA Coordinator. Formal procedures, involving checklists and grading systems, have been developed
to facilitate the certification/auditing process. Whenever deficiencies are noted, the field personnel are re-trained
immediately prior to resuming sampling activities.
Records documenting the results of the annual field crew certifications are maintained by the Province QA
Coordinator. In addition, following each field review, the QA Coordinator files a written report describing his/her
findings and any corrective actions undertaken. QA personnel also have performed periodic on-site evaluations
of laboratories responsible for processing samples. The purpose of these evaluations is to document that each
contract laboratory has adequate equipment, personnel and facilities to analyze samples in accordancewith prescribed
methods and QA requirements. Laboratory evaluation results also are documented in reports filed by the Province
QA Coordinator.
2.1 1990 Results
Formal training was held at the University of Rhode Island's (URI) Fisheries Center in Wickford, RI from
May 29 to June 15, 1990. All crew members were required to attend the entire course (however, crew chiefs
were periodically pulled from training for other activities). The development and conduct of the course was sub-contracted
by SAIC to the URI Marine Advisory Service and Fisheries Department. Instructors for the course were provided
by the URI Graduate School of Oceanography, URI Fisheries Department, NOAA (Milford, CT; Narragansett,
RI; and Woods Hole, MA), American Heart Association, Computer Sciences Corporation (CSC), and SAIC.
The class was generally divided into twogroups; oneclassroom and one practical (on-the-water). Most classroom
lessons were followed by practical training. Topics included boating safety, trailering, operation of sampling
equipment, navigation (including operation of the electronic instruments), data transfer, Quality Assurance/Quality
Control, fish and mollusc taxonomy, fish pathology, and CPR.
Page 2 Quality Assurance Report, EMAP-Virginian Province 1990 - 1993
-------
All participants were required to complete a Skills Evaluation Form on the first day of training. This information
was used to assign personnel to crews based on their skills, thereby assuring that each crew possessed the necessary
skill mix (computer and electronic instrument operation, fish taxonomy, bivalve taxonomy, sediment sampling,
etc.) for all aspects of sampling; and to help select those who would undergo additional training in a specialty
area (e.g., computer operation). Throughout training crewsworked together as a team during all hands-on activities.
At the end of training the composition of the crews was reviewed to assure that each crew had appropriate personnel
to complete all aspects of sampling. Based on the personal knowledge of crew members gained by the Crew Chiefs,
and information from the contract personnel managers, no changes to the crews were deemed necessary.
Trial runs, encompassing all components of sampling activities, were originally planned to be an important
component of training. This was notfully realized during formal training; no training in integrated sample processing,
packaging and shipping was provided. In addition, an evaluation of the crews by experienced EMAP-VP personnel
at the end of training revealed that some data collection methods were still not well understood or being followed
properly. As a result, it was decided at the end of the formal training period that the crews were not adequately
prepared for the Data Collection Phase. Therefore, the start of Interval 1 was delayed by lOdays. This reduction
left insufficient time for all stations to be sampled in that interval; therefore, the focus of Interval 1 activities
was changed from the collection of data to an extension of training. This was deemed necessary to ensure crews
were fully competent in all aspects of sample collection.
This extended training consisted of the actual collection of data and samples in the field at a limited number
of stations under the close supervision of senior EMAP-VP personnel familiar with the methods. This activity
served as more thanjust "dry runs", with some of thedata collected duringthis exercise being used in the characterization
of the Virginian Province.
During field operations each crew was visited by a senior EMAP staff member (Field Coordinator or QA
Coordinator). All aspects of sampling, from boat operations to shipping, were observed by the reviewer. Some
of the activities included confirming the presence/ absence of external pathologies, re-measuring fish and apparent
RPD (redox potential discontinuity) depth, assuring that all precautions were taken to avoid contamination of
the chemistry samples, assuring proper processing of benthic infauna samples, obsening data entry, and assuring
that all necessary safety precautions were observed. In 1990, no "field review check-off sheet" was utilized in
this review; however, a memo to the Province Manager was generated summarizing the review. Both reviewers
concluded that the crews demonstrated positive attitudes to QA issues, and that all sources of field-generated
error were in reasonable control.
Evaluation of Training and Lessons Learned
An evaluation of training, based on the overall results of the Data Collection Phase, indicated that the success
of training was mixed. URI provided an excellent facility and staff. Their contribution was mainly geared towards
boat operations and safety, areas in which they have extensive experience in providing classroom and hands-on
training to marine-related groups, such as commercial fishermen. The success can be measured by the absence
of any injuries during over 13,000 person-hours of field operations. Extramural instructors for fish and mollusc
taxonomy, and fish pathology provided excellent instruction; however, they were not expert in the goals of EMAP,
and had limited time to present their material. The material they presented was often too broad in scope, resulting
in inadequate instruction in the detailed areas pertinent to the Demonstration Project. It was suggested that, in
future years, such instruction should be more focused on Virginian Provinces issues, species, and conditions.
In-house instructors adequately presented instructions for the operation of gear; however, the science behind the
methods was not explained (e.g., the characteristics of a good dissolved oxygen profile). Severalareas were identified
that required more attention in subsequent courses, including packaging and shipping, and general maintenance
(lubricating trailer hubs, etc.).
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 3
-------
It was suggested that crew chiefs should be much more involved in training in future years. The proposal
for 1991 included extensive training for all crew chiefs prior to crew training. This training should provide them
with sufficient information to perform all sampling tasks. Important components should include the operation
of the field computer, understanding all Quality Assurance issues, any theory necessary for them to evaluate whether
or not sample or data collection must be repeated, and trouble shooting electronic sampling equipment. Crew
chiefs would then play an active role in training their crews.
2.2 1991 Results
Suggestions for improving training (detailed above) were incorporated into planning activities for the 1991
season. Crew chiefs underwent detailed training during the first two weeks of June, 1991. Training was limited
to two weeks because all but one of six crew chiefs were returnees from the previous year. Training was conducted
at the U. S. EPA Environmental Research Laboratory-Narragansett, RI (ERL-N) and focusedmainly on the sampling
methods, with emphasis placed on the electronic measurements and the computer system. Crew chief training
was conducted by SAIC and CSC personnel with oversight by EPA ERL-N staff.
Crew training was held from 17 June to 19 July 1991. Both safety and sampling methods were important
components of training. Crew training was broken into two phases: formal training which lasted for approximately
2!/2 weeks, and one week (per crew) of trial runs.
Trial runs consisted of four days in the field during which crews operated as they would during the sampling
season. They were assigned four stations to monitor for all parameters, including DataSonde deployment and
retrieval. Crews members stayed in motels, prepared samples for shipment, entered data into the field computer,
and electronically transmitted all data to the Field Operations Center (FOC) just as they would during actual field
operations. In addition, the Field Coordinator or the QA Coordinator visitedeach crew during trial runs, completing
a performance review sheet to determine the crew's overall grasp of the Program. All crews were deemed properly
prepared to begin sampling activities on 22 July, 1991.
Certification examinations for crew chiefs and field crew members were administered at the end of each course
and proved to be very useful. As a result of testing, two crew chiefs were identified as needing additional training.
Remedial coaching was provided and they were fully competent by the start of crew training. The examination
administered at the end of crew training suggested some areas, such as contingencies for moving stations, were
not adequately covered, so additional time was spent discussing these topics prior to trial runs.
In addition to the crew certification visits performed during dry runs, each crew was visited by a senior EMAP
staff member (Field Coordinator or QA Coordinator) during field operations. All aspects of sampling, from boat
operations to shipping, were observed by the reviewer. Some of the activities included confirming the presence/
absence of external pathologies, re-measuring fish and apparent RPD depth, assuring that all precautions were
taken to avoid contamination of the chemistry samples, assuring proper processing of benthic infauna samples,
observing data entry, and assuring that all necessary safety precautions were observed. The reviewer used a "field
review check-off sheet" to provide guidance during the review, and to document the crew's performance. Both
reviewers concluded thatthe crews were sufficiently concerned withallQA issues, and that all sources of field-generated
error were in reasonable control.
The only problem noted was the determination of the depth of the apparent RPD. This measurement was
determined to be too subjective, variable, and difficult to accurately measure based on a visual inspection of a
clear plexiglass core taken from a grab sample. Although reasonable measurements could be made in muddy
sands, the majority of the sediments encountered by field crews were fine grained muds where adhesion to the
plexiglass core creates too much smearing to allow for an accurate measurement. As a result of this observation,
Page 4 Quality Assurance Report, EMAP-Virginian Province 1990 - 1993
-------
RPD measurements were dropped from the sampling program, and all existing RPD datadeleted from the database.
2.3 1992 Results
Crew chiefs, who were all returnees from previous years, underwent a refresher training course during the
last week of May, 1992. This training was conducted at ERL-N and focused mainly on the sampling methods,
with emphasis placed on the electronic measurements and the computer system. Crew chief framing was conducted
by SAIC and CSC personnel with oversight by EPA ERL-N staff.
Crew training was held from 15 June to 17 July 1992. Both safety and sampling methods were important
components of training. Crew training was broken into two phases: formal training which lasted for approximately
3 weeks, and one week (per crew) of trial runs.
Trial runs consisted of five days in the field during which crews operated as they would during the sampling
season, monitoring practice stations for all parameters. Crew members stayed in motels, prepared samples for
shipment, entered data into the field computer, and electronically transmitted all data to the Field Operations
Center (FOC) just as they would during actual field operations. In addition, the Field Coordinator or the QA
Coordinator visited each crew during trial runs, completing a performance review sheet to determine the crew's
readiness. All crews were deemed properly prepared to begin sampling activities on 27 July, 1992.
In addition to the crew certification visits performed during trial runs, each crew was visited by a senior EMAP
staff member (Field Coordinator or QA Coordinator) during field operations. All aspects of sampling, from boat
operationsto shipping, wereobservedbythereviewer. Some of the activities included confirming the presence/absence
of external pathologies, re-measuring fish, assuring that all precautions were taken to avoid contamination of
the chemistry samples, assuring proper processing of benthic infauna samples, obsening data entry, and assuring
that all necessary safety precautions were observed. The reviewer used a "field review check-off sheet" to provide
guidance during the review, and to document the crew's performance. Both reviewers concluded that the crews
were sufficiently concerned with all QA issues, and that all sources of field-generated error were in reasonable
control.
The EMAP-VP QA Officer participated in audits during both crew certification and field operations. During
these audits he evaluated both the crew and the QA Coordinator's ability to conduct a performance review. His
findings were then summarized in a memo to the ERL-N laboratory director and the Province Manager. Although
he disagreed with some of the methods employed, he was fully satisfied that crews were adhering to EMAP SOPs,
and that the QA Coordinator was competent at evaluating the remaining crews.
2.4 1993 Results
Crew chief training for 1993 was separated into pilot training and chief scientist training. Pilot training was
held at the University of Rhode Island's Graduate School of Oceanography from 17 May to 21 May, 1993. This
framing consisted of instruction in navigation, safety and boat handling. Chief scientist training was conducted
at URI from 14 June to 18 June, 1993. This training focused mainly on the sampling methods, with emphasis
placed on the electronic measurements and the computer system. Crew chief training was conducted by SAIC,
URI and ROW Sciences personnel, with oversight by EPA ERL-N staff. All chief scientists were returnees from
previous years.
Formal crew framing was held at URI from 21 June to 9 July, 1993. Both safety and sampling methods were
important components of training. Crew training was followed by one week (per crew) of trial runs.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 5
-------
Trial runs consisted of five days in the field during which crews operated as they would during the sampling
season, monitoring practice stations for all parameters. Crew members stayed in motels, prepared samples for
shipment, entered data into the field computer, and electronically transmitted all data to the Field Operations
Center (FOC) just as they would during actual field operations. In addition, the Field Coordinator or the QA
Coordinator visited each crew during trial runs, completing a performance review sheet to determine the crew's
readiness. All crews were deemed properly prepared to begin sampling activities on 26 July, 1993.
In addition to the crew certification visits performed during trial runs, each crew was visited by a senior EMAP
staff member (Field Coordinator or QA Coordinator) during field operations. All aspects of sampling, from boat
operations to shipping, were observedby the reviewer. Some of the activities included confirming the presence/absence
of external pathologies, re-measuring fish, assuring that all precautions were taken to avoid contamination of
the chemistry samples, assuring proper processing of benthic infauna samples, obsening data entry, and assuring
that all necessary safety precautions were observed. The reviewer used a "field review check-off sheet" to provide
guidance during the review, and to document the crew's performance. Both reviewers concluded that the crews
were sufficiently concerned with all QA issues, and that all sources of field-generated error were in reasonable
control.
The EMAP-VP QA Officer participated in audits during both crew certification and field operations. During
these audits he evaluated both the crew and the QA Coordinator's ability to conduct a performance review. His
findings were then summarized in a memo to the ERL-N laboratory director and the Province Manager. Although
he disagreed with some of the methods employed, he was fully satisfied that crews were adhering to EMAP SOPs,
and that the QA Coordinator was competent at evaluating the remaining crews.
Page 6 Quality Assurance Report, EMAP-Virginian Province 1990 - 1993
-------
Section 3
QA Results for Chemical Contaminant Analyses of Sediments
3.1 Background
Measurement Quality Objectives (MQOs) for the analysis of chemical contaminants in EMAP-E sediment
samples are specified in the annual Province Quality Assurance Project Plans. These plans variously require each
EMAP-E laboratory to analyze the following types of quality control (QC) samples along with every batch or
"set" of field chemistry samples: laboratory reagent blanks, calibration check standards, laboratory fortified sample
matrix (matrix spike), laboratory fortified sample matrix duplicate (matrix spike duplicate), laboratory duplicate,
and Laboratory Control Material (LCM). Results for these QC samples must fall within certain pre-established
control limits for the analysis of a batch of samples to be considered acceptable.
Standard or Certified Reference Materials (SRMs or CRMs) typically are used by EMAP-E laboratories as
their Laboratory Control Material (LCM). SRMs and CRMs have known or "certified" concentrations of the
analytes being measured and therefore are useful for assessing both accuracy and precision. The QA Project Plan
requires the laboratory's percent recovery (relative to the certified concentration in the reference material) to fall
within certain pre-established control limits to be considered acceptable. If the laboratory consistently fails to
meet these acceptability criteria for the CRM or SRM analysis, the values reported for the failed analytes are
considered to be suspect (biased) and are flagged in the database, as described in the following section.
In addition to the above QA requirements, each laboratory analyzing EMAP sediment chemistry samples
must participate in an intercomparison exercise conducted through NOAA's National Status and Trends (NS&T)
Program and coordinated by the National Institute of Standards and Technology (NIST).
Many of the goals and objectives of the EMAP-E program coincide with those of NOAA's NS&T program.
By interagency agreement, personnel from the two agencies have continued to coordinate their activities to ensure
that data produced by the two coastal monitoring programs are compatible. This applies in particular to measurements
of chemical contaminant concentrations intissue and sediment samples. Toachieve this goal, all EMAP-E laboratories
participated yearly in the NIST/NOAA intercomparison exercises. A brief description of this exercise follows.
The NIST/NOAA intercomparison exercise is akey element of the National Status and Trends Program and
the EMAP-E "performance-based" QA philosophy. In this continuing series of exercises, various materials are
distributed in common to all laboratories for blind analysis. These exercises are coordinated for EPA and NOAA
by NIST, which typically distributes a variety of materials including gravimetrically-prepared solutions, extracts
of environmental samples (tissue or sediment), or actual marine samples (tissue or sediment). A11EMAP-E laboratories
are required to participate in the NOAA/NIST intercomparison exercises in order to become "certified" prior
to analyzing actual samples, and as a means of assessing comparability on an on-going basis.
Each year the EMAP-E QA Coordinator j oined laboratory personnel from the EPA's Environmental Monitoring
Systems Laboratory (EMSL), Cincinnati, OH, in attending the NIST/NOAA intercomparison exercise annual
meeting, where the results of the intercomparison exercises were presented and discussed. The annual meetings
serve as an excellent forum for representatives of the various labs to identify common analytical problems and
discuss potential solutions.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 7
-------
3.2 Data Qualifier Codes for Chemistry
Four data qualifier codes or "flags" are used in EMAP-E's sediment chemistry datasets:
The "SC-A" code indicates that an analyte was not detected. When the "SC-A" code is used, the concentration
field is left blank and the detection limit for the analyte in that particular sample is reported under the variable
"MDL" (method detection limit).
It is sometimes possible for a laboratory to detect an analyte and report its concentration at a level which
is below the calculated method detection limit for the sample. In these situations, the analyst is confident that
the analyte was present in the sample, but there is a high degree of uncertainty in the reported concentration.
The "SC-B" code is used to flag reported values which are below the calculated method detection limit for the
sample. Such values are considered estimates only and should be used with discretion.
The "SC-C" code is applied in situations where the laboratory failed to meet required control limits for one
or more of the quality control samples analyzed along with each sample batch. In such situations, there is reason
to believe that the concentrations reported for an analyte or group of analytesmay not accurately reflect the actual
concentrations present in the samples. The "SC-C" code usually is applied when the Certified Reference Material
results indicate that a laboratory experienced a consistent bias in the analysis of a particular analyte or group
of analytes. The "SC-C" code is also applied whenever other QC sample results suggest a possible bias in the
reported values (e.g., sample contamination detected in the laboratory reagent blank). Values flagged with the
"SC-C" code therefore are considered estimates only and should be used with discretion.
Results of QC sample analyses are stored in the EMAP-E database and are available upon request. The "SC-C"
code used to flag suspect values is applied following a thorough QA review of the entire data package submitted
by the laboratory for a given year. Inmany instances, best professional judgement must be used to decide which
values should be qualified as estimates only. In the following sections, explanations are provided for the "SC-C"
codes which appear in the EMAP-E sediment chemistry datasets. Persons using these data may wish to perform
their own review of the QC sample results to determine the acceptability of these data for their purposes.
For the years 1991-1993 in the Virginian Province, the laboratory used gas chromatography/electronic capture
detection (GC/ECD) with dual column confirmation for the analysis of PCB congeners and chlorinated pesticides
in sediments. All values reported in the database for the PCBs and pesticides represent "confirmed" results (i.e.,
the analyte was detected and could be quantified on both the primary and secondary columns). In situations where
an analyte was detected on one column, but was not confirmed on the second column, the result was treated as
a "not detect" (i.e., the SC-A code is used to flag the result in the database).
Close inspection of the "confirmed" results for certain pesticides revealed a number of instances where there
was a significant discrepancy in the amount detected on the two GC/ECD columns (i.e., greater than a factor of
three difference). In these instances, it is difficult to ascertain which amount is more accurate (i.e., which is the
"right" answer). A decision was made to take a "conservative" approach and report the lower of the two values
in the database, and to flag these values using the "SC-D"code. The SC-Dcodehasthe following meaning: "Analyses
were conducted using GC/ECD with dual column confirmation. Quantitation on the two columns differed by
more than a factor of three, and the lower of the two results is reported."
Although this approach was deemed necessary, the user must be cautioned that the application of the "SC-D"
code may invalidate investigations of the ratios of compounds. For example, if the concentrations of p,p'-DDT
from the two columns were 6.1 and 2.0 ng/g respectively, the SC-D code would be applied and the lower value
of 2.0 ng/g reported. However, if the values for p,p'-DDE were 6.0 and 2.1 ng/g, the SC-D code would NOT
be applied and the original value of 6.0 ng/g would be reported. Most likely the ratio of these two compounds
Page 8 Quality Assurance Report, EMAP-Virginian Province 1990 - 1993
-------
is approximately 1, but the results as reported would indicate a ratio of about 3. Therefore, ratios of compounds
should only be used when either all or none of the compounds are flagged with the SC-D code.
Values which are notflagged with the SC-B, SC-C or SC-D codes are considered valid and useful for anticipated
assessment purposes.
3.3 Quality Assessment Results
In the following sections, results for chemistry QC samples are summarized, and data flags associated with
the 1990 to 1993 EMAP-VP chemistry datasets are explained.
3.3.1 Laboratory Audit
A technical systems audit was conducted on 29 and 30 April 1991 at EMSL in Cincinnati, OH. The audit
team was led by Mr. Raymond Valente, the QAO for the EMAP-E program. Mr. Valente was assisted by two
senior organic chemists from ERL-N: Dr. Richard Pruell and Mr. DonCobb. Mr. Robert Graves, the acting EMAP
QA Coordinator based at EMSL-Cincinnati, accompanied the audit team as an observer.
Major problems were uncovered in EMSL-Analytical's (the production laboratory arm of EMSL-Cincinnati)
adherence to QA protocols for the analysis of PCB and pesticides in sediments. These problems were uncovered
prior to and during the audit of the laboratory (see followingparagraph). As a result of the audit, the 1990 sediment
organic analyses were halted and a series of corrective actions were implemented to bring the process back into
control. Analyses resumed in late FY '91.
The primary purpose of the audit was to review the methodology being employed at EMSL-Analytical for
analysis of low-level organic compounds in estuarine sedimentsamples from the EMAP-E 1990 Virginian Province
Demonstration Project. This review was deemed necessary partly in response to delays in sample processing
and subsequent phone conversations with laboratory personnel which suggested technical difficulties with the
organic analyses had been encountered. The audit had to be scheduled with a minimum of advance notice (ca.
1 week) to include one of the principal EMSL-Analytical participants prior to her departure from the laboratory
on April 30. The audit team's specific goal was to determine the exact nature of any technical difficulties being
encountered and provide constructive assistance as appropriate. At the same time, the audit team evaluated the
adequacy of EMSL-Analytical's adherence to QA requirements in relation to the EMAP-E analyses.
The main deficiency noted was failure to adhere to QA specifications in performing sediment organic analyses
(PCBs and pesticides). The audit findings were documented in a report submitted to the EMAP-E Acting Technical
Director and appropriate EMSL-Analytical personnel. A series of corrective actions were implemented over the
course of the spring and summer 1991, and no further audits were conducted.
3.3.2 1990 QA Results
Major and trace element analyses (except mercury)
Two methodologies, inductively-coupled plasma atomic emission spectrophotometry (ICP-AES)
and graphite furnace atomic absorption (GFAA) spectrophotometry, were utilized for the analyses
of major and trace elements (metals) in sediment samples collected by EMAP. The results of QC samples
(e.g., calibration standards, laboratory reagent blanks, matrix spikes, and LCMs) run with each of
the 18 batches of 1990 VP sediment samples generally met the pre-established EMAP criteria for acceptability.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 9
-------
For the ICP-AES analyses, which included the metals Ag, Al, Cr, Cu, Fe, Mn, Ni, Pb, and Zn, a total of 18
analytical sets or "batches" of samples were analyzed. SRM 2704 (Buffalo River Sediment, issued by NIST)
was analyzed along with every batch as the Laboratory Control Material. The analysis of a LCM is a particularly
important component of EMAP's performance-based approach to QA/QC that provides assessments
of accuracy as well as precision. The 1990 QAPP required the laboratory's percent recovery (relative to the
certified concentration in the reference material) to fall within a range of 85% to 115% for each metal. Except
for silver, the average percent recovery of each metal (relative to the certified concentration in SRM 2704) was
within the acceptability range of 85% to 115% (Table 3-1), and no "SC-C" codes were applied.
Table 3-1. Summary results for SRM 2704 (Buffalo River Sediment) used as a set control for the
1990 Virginian Province sediment inorganic analyses.
ICP-AES METALS (n = 18 analysis sets or "batches"):
Element Average1 Stdv2 C.V.3 Min.4 Max.5
Ag na na na na na
Al 96 1.8 1.9 92 99
Cr 87 2.7 3.1 80 91
Cu 95 2.4 2.5 90 99
Fe 88 1.6 1.8 83 90
Mn 96 2.2 2.3 92 99
Ni 90 5.5 6.2 84 110
Pb 93 4.5 4.8 85 99
Zn 96 1.6 1.7 93 99
GFAA METALS (n = 18 analysis sets):
Element Average1 Stdv2 C.V.3 Min.4 Max.5
As
Cd
Sb
Se
Sn
78
100
79
97
80
4.1
7.0
11.9
12.4
30.0
5.3
7.0
15.1
12.8
37.5
70
87
51
70
29
89
111
99
119
144
1 Average percent recovery relative to the SRM certified value.
2 Standard deviation of the percent recovery values.
3 Coefficient of variation of the percent recovery values.
4 Minimum percent recovery for 18 analysis sets
5 Maximum percent recovery for 18 analysis sets
Silver was not detected in most of the 1990 Virginian Province samples; however, the laboratory's detection
limit of 1 ppm was well above the target detection limit of 0.01 ppm specified in the QA Plan. If the target detection
Page 10 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
limit had been achieved, silver probably would have been detected and quantified in a much higher number of
samples. Therefore, the 1990 results are not reliable for assessing silver concentrations in Virginian Province
sediments. This problem was corrected in 1991 by analyzing for silver using GFAA rather than ICP.
The GFAA analyses included the metals As, Cd, Sb, Se, and Sn; a total of 18 analytical sets or "batches"
of samples were analyzed. SRM 2704 was analyzed along with every sample batch as the Laboratory Control
Material. Average SRM percent recoveries fell outside the acceptability range of 85% to 115% for the following
metals: As (78%), Sb (79%) and Sn (80%) (Table 3-1). In addition, matrix spike recoveries for these metals
were highly variable. These low and variable recoveries are attributed to both the low concentrations of these
metals in SRM 2704 (i.e., close to the detection limit) and the less rigorous digestion procedure used (?.
-------
In general, results for reagent blanks and calibration check samples analyzed with each batch of field samples
fell within control limits and serve to verify that sample contamination did not occur and that all instruments were
calibrated properly throughout the analytical runs. However, the matrix spike results are of limited use in assessing
overall data quality because the spiking solutions used by the laboratory for the PAH andPCB/pesticide analyses
contained only a small subset of the analytes of interest and not the full suite as originally specified in the QA
Plan. Furthermore, it is difficult to evaluate laboratory performance solely on the basis of matrix spike results
because it is often equivocal whether low recoveries are due to flawed methodology, poor technique, or a true
matrix interference.
Results for laboratory duplicate samples, intended to serve as a check on precision, also are of limited value
in assessing the quality of the 1990 Virginian Province organics data because the laboratory usually failed to
detect the analytes of interest in the sample chosen at random for duplicate analysis (i.e., most of the analytes
in laboratory duplicate samples were reported as "not detected").
Given the above limitations on using the matrix spike and laboratory duplicate results to assess the overall
quality of the 1990 Virginian Province organics data, great emphasis was placed on the LCM results. For both
the PAH and PCB/pesticide analyses, SRM1941 (Organics in Marine Sediment, issued by NIST) was analyzed
as the LCM along with each batch of samples. The QAPP required the laboratory's percent recovery (relative
to the certified concentration in the reference material) to fall between 70% and 130% for each organic analyte.
For most of the individual PAH compounds and PCB congeners with 'known" concentrations in SRM 1941,
the average percent recovery achieved by the laboratory (based on n=20 batches for PAHs and n=22 batches for
PCB/pesticides) consistently fell within the control limit range of 70% to 130% (Tables 3-2 and 3-3). Very high
and variable SRM 1941 recovery rates were experienced for the pesticides heptachlor epoxide (231%), cis-chlordane
(322%),trans-nonachlor(412%),and4,4'-DDT (186%)(Table 3-3). In general, significantproblemswereexperienced
by EMSL in their analyses of samples for PCBs and pesticides. In addition to their poor performance for pesticides
in the intercomparison exercise, problems existed with their extraction and analysis of samples.
Due to problems with integrating the internal standard peak consistently between standards and samples,
EMSL-Analytical switched to an external standard quantitation. The EMAP QA Team was concerned about
this since the external calibration does not allow any accounting for errors introduced by different extract volumes
and injection volumes. The EMAP QA Team also felt that the internal standard chosen was not the best choice
since its chemical structure and properties are not the same as the analytes of interest. EMSL-Analytical was
strongly urged to begin using PCB 198 as the internal standard for quantification. This would eliminate the external
standard usage and allow analytes to be quantified directly from a similar compound.
Tetrachloro-m-xylene (TCMX) was used as an internal standard aswell in the 1990 analyses. In most cases,
peak area was used to quantify results; however, at times peak height was used due to interferences. The peak
height from the internal standard was used along with the ratio of the peak area/height from the 5 ppb standard
to calculate the peak areas in samples with interferences. It is not clear whether or not the ratio is constant with
increasing or decreasing concentrations. TCMX was recovered in excess of 200% in some samples and greater
than 100% in others.
EMSL-Analytical used two GC columns (RTX-50 and OV-5) in order to quantify and confirm PCBs and
pesticides. It was noted in a review of the 1990 raw data that use of a particular column for quantifying results
was dependent upon the TCMX recovery. It has become clear that EMSL-Analytical chose to report the best
result from each column. The EMAP Audit Team allowed the use of both columns to quantify results of SRMs
as long as the use is consistent. Weeks later it was found that EMSL-Analytical was still picking and choosing
the analytes instead of consistently measuring them on the same column. Given thepossible differences in results
(>3X) this practice is questionable at best.
Page 12 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
These analytical problems result in significant doubts regarding the quality of the 1990 PCB/pesticide
data. These findings, along with EMSL's acknowledgement of the problems, have resulted in the deletion
of all 1990 PCB and pesticide results from the EMAP database.
A major deficiency in the 1990 Virginian Province organics dataset is related to the laboratory's failure to
achieve the target detection limits originally specified in the QA Plan. These target detection limits were 10 ng/g
(dry weight) for each PAH compound and 0.25 ng/g for each PCB congener and pesticide. In general, the detection
limits achieved by the laboratory ranged from 1.5 to 30 times higher than the target value for PAH compounds
and up to 15 times higher than the target value for PCB congeners and pesticides (Table 3-4). In addition, the
detection limits varied widely because the laboratory analyzed a different amount (i.e., dry weight) of sediment
from each sample. As a result, the analytes of interest were not detected in a large number of samples, and the
"calculated" detection limit (i.e., the theoretical concentration of each analyte necessary for detection) differed
significantly from sample to sample (Table 3-4).
If the target detection limits had been achieved and consistent sample sizes had been used, the organic analytes
of interest probably would have been detected and quantified in most of the 1990 Virginian Province samples.
In reality, analytes of interest present in the samples at low concentrations were not detected and therefore not
reported. This limits the comparability of the 1990 Virginian Province organics data with other data sets for
which lower detection limits were achieved and limits data users' ability to make quantitative evaluations of sediment
contamination for these organic compounds in the Virginian Province. As a result of this problem, EMSL's poor
performance in the intercomparison exercise, and the results presented in Table 3-2, the "SC-C" code has been
applied to all 1990 PAH data. This QA code informs the user that problems were identified which question the
quality of the results. Therefore these results should be treated as estimates and used with caution.
Table 3-2. Results for SRM 1941 (Organics in Marine Sediment) used as the set control (Laboratory
Control Material) for the 1990 Virginian Province sediment PAH analyses (n = 20 analysis
sets or "batches").
Compound1 Average2 Stdv3 C.V.4 Min5 Max6
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benz[a]anthracene
Benzo[b+k]fluoranthene
Benzo[a]pyrene
Perylene
Benzo[ghi]perylene
lndeno[1 ,2,3-cd]pyrene
98.8
71.6
99.2
87.6
93.9
104.6
64.9
64.4
86.2
118.9
22.0
17.9
22.4
18.7
20.8
18.9
15.4
16.2
23.3
29.5
22.3
25.0
22.6
21.3
22.1
18.1
23.7
25.2
27.0
24.8
62
37
65
65
57
67
40
35
48
65
138
101
149
121
141
142
90
93
145
182
1 SRM 1941 has certified concentrations for only a subset of the PAH compounds analyzed by the laboratory
in 1990.
2 Average percent recovery relative to the SRM certified value.
3 Standard deviation of the percent recovery values.
4 Coefficient of variation of the percent recovery values.
5 Minimum percent recovery for 20 analysis sets
6 Maximum percent recovery for 20 analysis sets
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 13
-------
Table 3-3. Results for SRM 1941 (Organics in Marine Sediment) used as the set control (Laboratory
Control Material) for the 1990 Virginian Province sediment PCB/pesticide analyses (n = 22
analysis sets or "batches").
Compound1 Average2 Stdv3 C.V.4 Min5 Max6
PCB 18
PCB28
PCB 52
PCB 66
PCB 101
PCB 118
PCB 153
PCB 105
PCB 138
PCB 187
PCB 180
PCB 170
PCB 195*
PCB 206*
PCB 209
Heptachlor epoxide*
cis-Chlordane*
trans-Nonachlor*
4,4'-DDE
4,4'-DDD
4,4'-DDT*
79.4
54.8
101.5
67.7
73.9
99.2
94.5
96.3
77.1
82.7
97.0
82.3
147.0
100.3
93.9
231.0
322.0
411.9
104.8
92.3
185.8
17.1
9.2
23.5
9.7
17.1
14.4
15.1
17.9
16.3
18.6
19.5
20.5
39.0
27.9
21.5
91.7
81.5
710.7
32.0
21.4
135.4
21.5
16.8
23.1
14.3
23.1
14.5
16.0
18.6
21.1
22.5
20.1
24.9
26.5
27.8
23.0
39.7
25.3
172.5
30.5
23.2
72.9
23
34
60
47
48
65
60
67
53
58
66
57
80
61
61
109
87
86
65
33
63
101
76
146
80
105
116
121
130
105
122
132
143
213
176
134
448
450
2770
212
123
660
1 SRM 1941 only lists "non-certified" or informational values for this group of PCB congeners and
pesticides (* = concentration in the SRM is less than 10 times the target detection limit).
2 Average percent recovery relative to the SRM value.
3 Standard deviation of the percent recovery values.
4 Coefficient of variation of the percent recovery values.
5 Minimum percent recovery for 22 analysis sets
6 Maximum percent recovery for 22 analysis sets
Page 14 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 3-4. Range in detection limits (in ng/g dry weight) reported for organic compounds in 1990
Virginian Province sediment samples. The target detection limits were 10 ng/g for each
PAH compound and 0.5 ng/g for each PCB congener and pesticide.
Polycyclic Aromatic Hydrocarbons
Acenaphthene
Anthracene
Benz(a)anthracene
Benzo(a)pyrene
Benzo(e)pyrene
Biphenyl
Chrysene
Dibenz(a,Ji)anthracene
2,6-dimethylnaphthalene
Fluoranthene
Fluorene
2-methylnaphthalene
1-methylnaphthalene
1-methylphenanthrene
Naphthalene
Perylene
Phenanthrene
Pyrene
Benzo(b+k)fluoranthene
Acenaphthlylene
Benzo(g,h,i)perylene
ldeno(1 ,2,3-c,d)pyrene
2,3,5-trimethylnaphthalene
(PAHs)
Minimum
21
17
17
23
23
23
22
24
24
16
25
25
23
13
30
27
16
15
22
22
31
26
23
Maximum
207
121
72
151
153
150
72
252
156
114
176
162
150
86
54
189
44
39
145
212
325
249
219
Median
34
28
28
38
37
36
35
43
38
24
43
39
34
21
39
46
26
22
33
38
55
43
38
DDT and its metabolites
Minimum
Maximum
Median
2,4'-DDD
4,4'-DDD
2,4'-DDE
4,4'-DDE
2,4'-DDT
4,4'-DDT
0.13
0.12
0.10
0.04
0.12
0.18
1.93
6.10
1.11
0.45
1.26
3.22
0.24
0.20
0.18
0.07
0.22
0.58
continued
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
Page 15
-------
Table 3-4, continued.
Chlorinated pesticides
other than DDT
Aldrin
Alpha-Chlordane
Trans-Nonachlor
Dieldrin
Heptachlor
Heptachlor epoxide
Hexachlorobenzene
Lindane (gamma-BHC)
Mirex
18 PCB Congeners:
Minimum
0.10
0.09
0.04
0.04
0.10
0.08
0.03
0.16
0.03
Minimum
Maximum
1.78
1.16
0.87
0.52
1.47
1.85
7.23
27.5
1.93
Maximum
Median
0.27
0.19
0.07
0.08
0.19
0.19
0.09
0.64
0.08
Median
PCB 08
PCB 18
PCB 28
PCB 44
PCB 52
PCB 66
PCB 101
PCB 105
PCB 118
PCB 128
PCB 138
PCB 153
PCB 170
PCB 180
PCB 187
PCB 195
PCB 206
PCB 209
0.08
0.37
0.08
0.06
0.11
0.09
0.12
0.07
0.06
0.12
0.11
0.11
0.09
0.11
0.08
0.10
0.10
0.12
4.46
5.89
1.03
1.50
2.70
1.01
1.39
0.60
0.65
1.62
1.31
1.03
2.15
1.30
0.72
1.23
1.38
1.09
0.63
0.94
0.17
0.17
0.38
0.18
0.20
0.14
0.12
0.23
0.18
0.19
0.32
0.19
0.13
0.19
0.20
0.20
Total Organic Carbon analyses
All QC results for the analysis of total organic carbon in the 1990 Virginian Province sediment samples fell
within required control limits. The Certified Reference Material PACS-1 (issued by the National Research Council
of Canada) was utilized as the LCM. The certified concentration of total carbon in thisreference material is 3.69%
(percent dry weight). The average percent recovery for TOC in PACS-1 achieved by the laboratory for n = 18
batches of samples (i.e., 18 separate analyses of PACS-1) was 87.2%, with all values falling within the range
85% to 95%. Since the PACS-1 certified concentration includes both organic carbon and a very small fraction
of inorganic carbon, the laboratory's percent recovery values for organic carbon are expected to be below 100%.
Based on the good overall percent recovery of organic carbon in the Certified Reference Material, the 1990 Virginian
Province sediment TOC data were deemed acceptable for use without qualification.
Page 16
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Butyltin analyses
Data users are cautioned that there are deficiencies in the 1990 Virginian Province butyltin analyses which
might limit or preclude the use of these data. The main deficiency is related to the laboratory's failure to detect
the butyltin compounds of interest (TBT, DBT, MET) in the majority of samples analyzed. The method detection
limits established by the laboratory were 4 ng/g dry weight (as tin) for both TBT and DBT, and 10 ng/g dry weight
(as tin) for MET. It is possible that the butyltin compounds of interestwere present in many samples at concentrations
below these detection limits, and, therefore, the occurrence of butyltin compounds in Virginian Province sediments
may be more widespread than indicated by these data.
The Certified Reference Material PACS-1 (issued by the National Research Council of Canada) was utilized
as the LCM for these analyses. Average percent recoveries relative to the certified value for n = 14 analysis sets
were 73% for TBT, 57% for DBT and 394% for MET. These recoveries fall outside the QA Plan-specified accuracy
range of 85% to 115% and indicate that TBT and DBT were consistently under-recovered and MET was grossly
over-recovered in this reference material. Therefore, all values reported for TBT, DBT and MET in samples
where these compounds were detected are considered estimates (SC-C code) and should be used with discretion.
3.3.3 1991 QA Results
Major and trace element analyses (except mercury)
For the 1991 Virginian Province analysis of major and trace elements by ICP-AES and GFAA, the laboratory
generally met the pre-established acceptability criteria (control limits) for the QC samples (e.g., calibration check
samples, laboratory reagent blanks, matrix spikes, and Laboratory Control Materials). For the ICP-AES analyses,
which included the metals Al, Cr, Cu, Fe, Mn, Ni, Pb, and Zn, a total of 13 analytical sets or "batches" of samples
were analyzed. The Certified Reference Material (CRM) "BCSS-1" (Estuarine Sediment, issued by the National
Research Council of Canada) was analyzed along with every batch as the LCM. The 1991 QAPP required the
laboratory's percent recovery (relative to the certified concentration in the reference material) to fall within a range
of 80% to 120% for each metal. With the exception of Cr and Pb, the average percent recovery of each metal
was within this acceptability range (Table 3-5). The average percent recovery for Cr was slightly lower than
acceptable, and the average percent recovery for Pb was slightly higher than acceptable. These results suggest
that Cr may have been consistently "under-recovered" and Pb may have been consistently "over-recovered" in
the actual samples. Therefore, all reported values for these two metals were qualified with the SC-C code in the
database.
The GFAA analyses included the metals Ag, As, Cd, Sb, Se, and Sn; a total of 19 analytical sets or "batches"
of samples were analyzed. The CRM BCSS-1 also was analyzed along with every sample batch as the LCM.
Average percent recoveries for all metals fell within the acceptability range of 80% to 120% (Table 3-5), and
no results were flagged in the database. The CRM BCSS-1 does not have a "certified" value for silver, making
it difficult to assess laboratory accuracy and precision for this metal. However, the laboratory was able to achieve
a lower detection for this metal in 1991 compared to 1990, which resulted in silver being detected in a much higher
number of samples in 1991 compared to 1990.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 17
-------
Table 3-5. Summary results for CRM BCSS-1 (Estuarine Sediment) used as a set control for the 1991
Virginian Province sediment inorganic analyses.
ICP-AES METALS (n = 13 analysis sets or "batches"):
Element Average1 Stdv2 C.V.3 Min.4 Max.5
6.5 87 109
2.8 66 73
2.8 99 110
3.0 91 100
3.1 87 97
2.7 86 94
21.7 81 185
1.7 87 91
Element Average1 Stdv2 C.V.3 Min.4 Max.5
Al
Cr
Cu
Fe
Mn
Ni
Pb
Zn
GFAA METALS
95
70
105
95
93
91
122
89
(n = 19 analysis
6.2
2.0
3.0
2.9
2.9
2.4
26.5
1.5
sets'):
Ag
As
Cd
Sb
Se
Sn
na
94
91
98
111
111
na
9.0
23.6
15.4
32.5
14.9
na
9.6
26.1
15.6
29.3
13.4
na
76
39
78
50
66
na
114
157
137
189
135
1 Average percent recovery relative to the SRM certified value.
2 Standard deviation of the percent recovery values.
3 Coefficient of variation of the percent recovery values.
4 Minimum percent recovery for n analysis sets
5 Maximum percent recovery for n analysis sets
Mercury analyses
For the 1991 Virginian Province mercury analyses, the Certified Reference Material BEST-1 (issued by the
National Research Council of Canada) was analyzed along with every sample batch as the LCM (n = 9 sample
batches). The average percentrecovery of 92%for mercury in this reference material fell well within the acceptability
range of 80% to 120%. In addition, an average percent recovery of 104% was achieved for the matrix spike samples
analyzed in each batch. Overall, these results indicate acceptable accuracy for the mercury analyses, and no "SC-C"
codes were used to qualify the data. The 1991 Virginian Province mercury results were deemed acceptable for
use without qualification.
Organic analyses
In general, results for reagent blanks and calibration check samples analyzed with each batch of samples fell
within control limits and serve to verify that sample contamination did not occur and that all instruments were
calibrated properly throughout the analytical runs. Average recoveries of compounds in matrix spike samples
generally fell within control limits, although these recoveries tended to be highly variable between differentbatches.
This, in part, reflects the fact that the spiked samples were chosen at random and sometimes had high "background"
Page 18 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
concentrations of the spiked analytes. In these cases it was difficult for the laboratory to accurately recover the
spiked amount relative to the high background, resulting in zero percent recovery in some samples. Furthermore,
it is difficult to evaluate laboratory performance solely on the basis of matrix spike results because it is often
equivocal whether low recoveries are due to flawed methodology, poor technique, or a true matrix interference.
Given the above limitations on using the matrix spike results to assess the overall quality of the 1991 Virginian
Province organics data, great emphasis was placed on the LCM results. For both the PAH and PCB/pesticide
analyses, SRM 1941 (Organics in Marine Sediment, issued by NIST) was analyzed as the LCM along with each
batch of field samples. For most of the individual PAH compounds and PCB congenerswith "known" concentrations
in SRM 1941 (this includes both "certified" and "non-certified" values), the average percent recovery achieved
by the laboratory (based on n = 14 batches for PAHs and n = 15 batches for PCB/pesticides) generally fell within
the control limit range of 70% to 130% (Tables 3-6 and 3-7). Whenever the laboratory failed to achieve these
average recovery rates for a particular compound, all the results in the 1991 database for that compound were
flagged with the "SC-C" code to indicate the potential inaccuracy inferred from the SRM analysis. It is important
to note that the 70% to 130% recovery criteria only applies to compounds having SRM concentrations greater
than 10 times the laboratory's detection limit. When compounds occur at concentrations less than about 10 times
the detection limit, a greater amount of analytical uncertainty is expected and the normal control limit "acceptability"
criteria do not apply.
Based on the above, the results for the following organic compounds were flagged with the "SC-C" code in
the 1991 Virginian Province organics dataset: PCB 101, PCB 138, PCB 153, PCB 18, PCB 187, acenaphthylene,
chrysene,l-memylphenanthrene,andnaphmalene.maddition,almoughmeaveragepercentrecoveryfori^
was within limits (98%), all results for this compound were flagged with the SC-C code because the recoveries
between batches exhibited relatively high variability (i.e., 35% coefficient of variation). Although the average
SRMpercentrecoveriesforthecompoundsdieldrin,heptachlorepoxideandPCB195 also wereoutside the acceptability
range of 70% to 130% (Table 3-7), these compounds occur in the SRM at concentrations less than 10 times the
laboratory's detection limit. Therefore, the acceptability criteria do not apply.
Unlike the 1990 analyses, when the laboratory failed to achieve a consistent detection limit for the organic
compounds, in 1991 a consistent detection limit of 0.25 ng/g (dry weight) was achieved for each PCB congener
and pesticide and 10.0 ng/g (dry weight) was achieved for each PAH compound.
As previously indicated (see Section 3.2), the laboratory used gas chromatography/electronic capturedetection
(GC/ECD) with dual column confirmation for the analysis of PCB congeners and chlorinated pesticides in the
1991 Virginian Province sediment samples. All values reported in thedatabase for the PCBs and pesticides represent
"confirmed" results (i.e., the analyte was detected and could be quantified on both the primary and secondary
columns). In general, for all reported PCB congeners except PCB 195, the rate of confirmation was between
95% and 100% (PCB 195 rate of confirmation was 87%). The rate of confirmation exceeded 90% for all the
chlorinated pesticides except the following: heptachlor (59%), heptachlor epoxide (57%), mirex (82%), p,p DDT
(65%), and o,p DDT (72%). Whenever an analyte was detected on one column, but was not confirmed on the
second column, the result was treated as a "not detect" (i.e., the SC-A code is used to flag the result in the database).
Whenever there was a significant discrepancy in the amount detected on the two GC/ECD columns (i.e., greater
than a factor of three difference), the lower of the two values is reported in the database and flagged with the
"SC-D" code. Please note the warning associated with this code discussed in Section 3.2.
EMSL did not participate in the NOAA intercomparison exercise in 1991. They did participate in 1992 and
the results showed a continuing problem with pesticide analyses (see Section 3.3.4). Therefore, all pesticide data
(dieldrin, heptachlor epoxide, cis-chlordane, trans-nonachlor, all DDT-series compounds) are qualified with the
"SC-C" code to inform the user of potential problems with the data. Note that this qualification does not necessarily
apply to PCBs as well.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 19
-------
Table 3-6. Results for SRM 1941 (Organics in Marine Sediment) used as the set control (Laboratory
Control Material) for the 1991 Virginian Province sediment PAH analyses (n = 14 analysis
sets or "batches").
Compound1
Acenaphthene
Acenaphthlylene
Anthracene
Benz(a)anthracene
Benzo(a)pyrene
Benzo(e)pyrene
Benzo(b+k)fluoranthene
Benzo(g,h,i)perylene
Biphenyl
Chrysene
2,6-dimethylnaphthalene
Fluoranthene
Fluorene
ldeno(1 ,2,3-c,d)pyrene
1-methylnaphthalene
2-methylnaphthalene
1-methylphenanthrene
Naphthalene
Perylene
Phenanthrene
Pyrene
Average2
111
41
95
92
77
101
121
105
103
145
113
93
105
98
99
109
138
69
72
111
96
Stdv3
23.2
10.6
26.4
28.2
15.1
22.4
25.4
21.1
22.7
30.4
24.2
20.2
32.3
34.1
27.8
33.6
50.5
27.3
15.2
27.2
23.9
C.V.4
20.9
25.9
27.7
30.5
19.7
22.2
21.0
20.1
22.1
21.0
21.3
21.7
30.7
34.7
28.2
30.8
36.5
39.5
21.1
24.4
24.9
Min5
67
27
59
54
52
61
87
64
63
94
70
64
62
21
59
53
64
8
47
76
56
Max6
137
61
142
165
106
138
174
141
138
196
145
134
179
150
158
158
247
126
96
160
134
1 Listed compounds include those having both "certified" and "non-certified" concentrations in SRM 1941.
2 Average percent recovery relative to the SRM value.
3 Standard deviation of the percent recovery values.
4 Coefficient of variation of the percent recovery values.
5 Minimum percent recovery for 14 analysis sets
6 Maximum percent recovery for 14 analysis sets
Page 20 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 3-7. Results for SRM 1941 (Organics in Marine Sediment) used as the set control (Laboratory
Control Material) for the 1991 Virginian Province sediment PCB/pesticide analyses (n = 15
analysis sets or "batches").
Compound1
Average2
Stdv3
C.V.4
Min5
Max6
PCB 18
PCB28
PCB 52
PCB 66
PCB 101
PCB 118
PCB 153
PCB 105
PCB 138
PCB 187
PCB 180
PCB 170
PCB 195*
PCB 206*
PCB 209
Dieldrin*
Heptachlor epoxide*
cis-Chlordane*
trans-Nonachlor*
4,4'-DDE
4,4'-DDD
4,4'-DDT*
32
77
102
87
68
93
66
128
68
64
96
75
142
76
82
143
139
96
89
91
80
102
10.0
11.7
14.1
12.8
10.4
28.0
5.2
19.1
5.5
7.7
9.9
6.4
29.8
10.4
9.7
29.3
27.3
12.7
15.3
9.5
9.0
22.6
31.2
15.2
13.8
14.7
15.2
29.9
7.9
15.0
8.1
11.9
10.3
8.6
20.8
13.7
11.8
20.6
19.6
13.3
17.2
10.5
11.2
22.2
20
58
85
68
48
58
55
99
60
52
80
68
108
62
69
85
99
71
72
75
64
62
50
95
122
104
86
170
76
165
77
84
110
89
199
92
98
182
184
122
127
109
98
128
1 SRM 1941 only lists "non-certified" or informational values for this group of PCB congeners and
pesticides (* = concentration in the SRM is less than 10 times the target detection limit).
2 Average percent recovery relative to the SRM value.
3 Standard deviation of the percent recovery values.
4 Coefficient of variation of the percent recovery values.
5 Minimum percent recovery for 22 analysis sets
6 Maximum percent recovery for 22 analysis sets
Total Organic Carbon analyses
All QC results for the analysis of total organic carbon in the 1991 Virginian Province sediment samples fell
within required control limits. The Certified Reference Material PACS-1 (issued by the National Research Council
of Canada) was utilized as the LCM. The certified concentration of total carbon in thisreference material is 3.69%
(percent dry weight). The average percent recovery achieved by the laboratory for n = 11 batches of TOC samples
(i.e., 11 separate analyses of CRM PACS-1) was 94.0%, with all values falling within the range 88% to 99%.
Since the PACS-1 certified concentration includes both organic carbon and a very small fraction of inorganic
carbon, the laboratory's percent recovery values for organic carbon are expected to be below 100%. Based on
the good overall percent recovery of organic carbon in the Certified Reference Material, the 1991 Virginian Province
sediment TOC data were deemed acceptable for use without qualification.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
Page 21
-------
Butyltin analyses
Data users are cautioned that there are deficiencies in the 1991 Virginian Province butyltin analyses which
might limit or preclude the use of these data. The main deficiency is related to the laboratory's failure to detect
the butyltin compounds of interest (TBT, DBT, MET) in the majority of samples analyzed. The MDLs established
by the laboratory were 5 ng/g dry weight (as tin) for both TBT and DBT, and 12 ng/g dry weight (as tin) for MET.
It is possible that the butyltin compounds of interest were present in many samples at concentrations below these
detection limits, and, therefore, the occurrence of butyltin compounds in Virginian Province sediments may be
more widespread than indicated by these data.
The Certified Reference Material PACS-1 (issued by the National Research Council of Canada) was utilized
as the Laboratory Control Material for these analyses. Average percent recoveries relative to the certified value
for n = 12 analysis sets were 79% for TBT, 89% for DBT and 115% for MET. The percent recovery value for
TBT falls slightly outside the acceptable accuracy limits of 80% to 120% and indicates that TBT may have been
consistently under-recovered in this reference material. Therefore, all values reported for TBT in samples where
this compound was detected are considered estimates (SC-C code) and should be used with discretion.
3.3.4 1992 QA Results
Major and trace element analyses (except mercury)
For the 1992 Virginian Province analysis of major and trace elements by ICP-AES and GFAA, the laboratory
generally met the pre-established acceptability criteria (control limits) for the QC samples (e.g., calibration check
samples, laboratory reagent blanks, matrix spikes, and Laboratory Control Materials). For the ICP-AES analyses,
which included the metals Al, Cr, Cu, Fe, Mn, Ni, Pb, and Zn, a total of 16 analytical sets or "batches" of samples
were analyzed. The Certified Reference Material (CRM) "BCSS-1" (Estuarine Sediment, issued by the National
Research Council of Canada) was analyzed along with every batch as the LCM. With the exception of Cr, the
averagepercentrecovery ofeachmetal(relativetothecertifiedconcentrationinBCSS-l) was within theQAPP-specified
acceptability range of 80% to 120% (Table 3-8). The average percent recovery for Cr (71%) was slightly lower
than acceptable, suggesting that this metal may have been consistently "under-recovered" in the actual samples.
Therefore, all reported values for this metal were qualified with the SC-C code in the database.
The GFAA analyses included the metals Ag, As, Cd, Sb, Se, and Sn; a total of 16 analytical sets or "batches"
of samples were analyzed. The CRM BCSS-1 also was analyzed along with every sample batch as the Laboratory
Control Material. Average CRM percent recoveries for all metals fell within the acceptability range of 80% to
120% (Table 3-8), and no results were flagged in the database. The CRM BCSS-1 does not have a "certified"
value for silver, but the average recovery for this metal in laboratory spiked samples (matrix spikes) was within
quality control limits.
Mercury analyses
For the 1992 Virginian Province mercury analyses, the Certified Reference Material BEST-1 (issued by the
National Research Council of Canada) was analyzed along with every sample batch as the Laboratory Control
Material (n = 8 sample batches). The average percent recovery of 88% for mercury in this reference material
fell well within the acceptability range of 80% to 120%. In addition, an average percent recovery of 102% was
achieved for the matrix spike samples analyzed in each batch. Overall, these results indicate acceptable accuracy
for the mercury analyses, and no "SC-C" codes were used to qualify the data. The 1992 Virginian Province mercury
Page 22 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
results were deemed acceptable for use without qualification.
Table 3-8. Summary results for CRM BCSS-1 (Estuarine Sediment) used as a set control for the 1992
Virginian Province sediment inorganic analyses.
ICP-AES METALS (n = 16 analysis sets or "batches"):
Element Average1 Stdv2 C.V.3 Min.4 Max.5
4.5 78 93
4.7 66 80
3.7 94 107
4.2 80 92
3.5 83 96
3.3 79 90
14.9 72 137
4.1 80 92
Element Average1 Stdv2 C.V.3 Min.4 Max.5
Al
Cr
Cu
Fe
Mn
Ni
Pb
Zn
GFAA METALS Cn =
82
71
101
87
91
86
103
85
16 analysis
3.7
3.3
3.7
3.6
3.2
2.8
15.3
3.5
sets'):
Ag
As
Cd
Sb
Se
Sn
na
111
102
101
85
99
na
13.0
11.6
15.7
20.8
9.8
na
11.6
11.4
15.5
24.4
10.0
na
83
67
79
45
83
na
135
119
130
123
116
1 Average percent recovery relative to the SRM certified value.
2 Standard deviation of the percent recovery values.
3 Coefficient of variation of the percent recovery values.
4 Minimum percent recovery for n analysis sets
5 Maximum percent recovery for n analysis sets
Organic analyses
In general, results for reagent blanks and calibration check samples analyzed with each batch of samples fell
within control limits and serve to verify that sample contamination did not occur and that all instruments were
calibrated properly throughout the analytical runs. Average recoveries of compounds in matrix spike/matrix
spike duplicate samples generally fell within control limits, indicating acceptable analytical performance. However,
matrix spike samples are not the most ideal quality control samples because the analytes of interest are not truly
incorporated into the matrix in the same manner as an actual field sample. In addition, it can be difficult to evaluate
laboratory performance solely on the basis of matrix spike results because it is oftenequivocal whether low recoveries
are due to flawed methodology, poor technique, or a true matrix interference.
Given the above limitations related to the use of matrix spike samples to assess analytical performance, great
emphasis was placed on the LCM results. For both the PAH and PCB/pesticide analyses, SRM 1941 (Organics
in Marine Sediment, issued by NIST) was analyzed as the LCM along with eachbatch of field samples. For most
of the individual PAH compounds and PCB congeners with "known" concentrations in SRM 1941 (this includes
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 23
-------
both "certified" and "non-certified" values), the average percent recovery achieved by the laboratory (based on
n = 13 batches for PAHs and n = 13 batches for PCB/pesticides) generally fell within the control limit range of
70% to 130% (Tables 3-9 and 3-10). Whenever the laboratory failed to achieve these average recovery rates
for a particular compound, all the results in the 1992 Virginian Province organics dataset for that compound were
flagged with the "SC-C" code to indicate the potential inaccuracy inferred from the SRM analysis. It is important
to note that the 70% to 130% recovery criteria only applies to compounds having SRM concentrations greater
than 10 times the laboratory's detection limit. When compounds occur at concentrations less than about 10 times
the detection limit, a greater amount of analytical uncertainty is expected and the normal control limit "acceptability"
criteria do not apply.
Table 3-9. Results for SRM 1941 (Organics in Marine Sediment) used as the set control (Laboratory
Control Material) for the 1992 Virginian Province sediment PAH analyses (n = 13 analysis
sets or "batches").
Compound1 Average2 Stdv3 C.V.4 Min5 Max6
Acenaphthene
Acenaphthlylene
Anthracene
Benz(a)anthracene
Benzo(a)pyrene
Benzo(e)pyrene
Benzo(b+k)fluoranthene
Benzo(g,h,i)perylene
Biphenyl
Chrysene
2,6-dimethylnaphthalene
Fluoranthene
Fluorene
ldeno(1 ,2,3-c,d)pyrene
1-methylnaphthalene
2-methylnaphthalene
1-methylphenanthrene
Naphthalene
Perylene
Phenanthrene
Pyrene
127
57
93
88
69
95
105
111
118
152
128
100
126
114
115
126
130
77
71
127
113
22.4
12.5
27.8
14.5
10.4
19.9
17.4
15.2
29.9
25.6
32.0
21.5
21.4
16.2
29.3
40.5
44.6
35.9
9.2
28.7
17.0
17.7
21.9
29.8
16.4
15.1
21.0
16.6
13.8
25.3
16.8
25.0
21.5
17.0
14.2
25.5
32.0
34.2
46.6
13.1
22.7
15.0
98
38
59
69
52
63
79
95
56
118
66
70
91
84
62
52
69
8
58
75
92
167
79
145
109
86
132
137
146
153
198
177
143
176
140
153
190
239
131
89
162
156
1 Listed compounds include those having both "certified" and "non-certified" concentrations in SRM 1941.
2 Average percent recovery relative to the SRM value.
3 Standard deviation of the percent recovery values.
4 Coefficient of variation of the percent recovery values.
5 Minimum percent recovery for 13 analysis sets
6 Maximum percent recovery for 13 analysis sets
Page 24 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 3-10. Results for SRM 1941 (Organics in Marine Sediment) used as the set control (Laboratory
Control Material) for the 1992 Virginian Province sediment PCB/pesticide analyses (n = 13
analysis sets or "batches").
Compound1
Average2
Stdv3
C.V.4
Min5
Max6
PCB 18
PCB28
PCB 52
PCB 66
PCB 101
PCB 118
PCB 153
PCB 105
PCB 138
PCB 187
PCB 180
PCB 170
PCB 195*
PCB 206*
PCB 209
Dieldrin*
Heptachlor epoxide*
cis-Chlordane*
trans-Nonachlor*
4,4'-DDE
4,4'-DDD
4,4'-DDT*
47
74
121
94
69
78
71
146
70
67
100
80
176
82
86
125
160
108
120
91
84
102
9.8
12.0
21.2
15.5
10.4
11.2
6.1
24.1
6.6
5.4
10.1
7.3
27.5
11.5
15.3
52.2
72.6
24.9
28.1
17.9
16.4
36.0
20.9
16.1
17.5
16.4
15.0
14.2
8.6
16.5
9.4
8.1
10.1
9.2
15.7
14.0
17.7
41.9
45.3
23.0
23.4
19.6
19.6
35.2
30
49
90
74
56
64
61
102
61
58
88
69
133
54
57
63
58
74
75
69
65
41
57
94
157
122
90
101
80
183
81
77
118
89
222
95
104
255
281
150
160
125
110
167
1 SRM 1941 only lists "non-certified" or informational values for this group of PCB congeners and
pesticides (* = concentration in the SRM is less than 10 times the target detection limit).
2 Average percent recovery relative to the SRM value.
3 Standard deviation of the percent recovery values.
4 Coefficient of variation of the percent recovery values.
5 Minimum percent recovery for 13 analysis sets
6 Maximum percent recovery for 13 analysis sets
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
Page 25
-------
Based on the above, the results for the following organic compounds were flagged with the "SC-C" code in
the 1992 database: PCB101,PCB105,PCB18,PCB 187, acenaphthylene,benz(a)pyrene,chrysene, and 1-methylphenanthrene.
In addition, although the average percent recovery for naphthalene was within limits (77%), all results for this
compound were flagged with the SC-C code because the recoveries between batches exhibited relatively high
variability (e.g., 47% coefficient of variation). Although the average SRM percent recoveries for the compounds
heptachlor epoxide and PCB 195 also were outside the acceptability range of 70% to 130% (Table 3-10), these
compounds occur in the SRM at concentrations less than 10 times the laboratory's detection limit. Therefore,
the acceptability criteria do not apply.
The results of the 1992 NIST/NOAA intercomparison exercise suggested that EMSL-Analytical was producing
acceptable data quality for PCB and PAH analyses. Pesticide analyses, however, were still questionable. EMSL
reported measurable concentrations for two pesticides (heptachlor and 2,4'-DDT) that were not detected by most
other laboratories. Concentrations reported for cis-chlordane and trans-nonachlor were considered too high by
NIST to be used in calculating the consensus value. EMSL's continuing problems in the analysis of samples
for pesticides resulted in all 1992 pesticide data being qualified with the SC-C code, indicating the quality of
the data is questionable.
Detection limits of 0.25 ng/g (dry weight) for each PCB congener and pesticide, and 10.0 ng/g (dry weight)
for each PAH compound, were achieved in the most of 1992 Virginian Province organics samples.
As previously indicated (see Section 3.2), the laboratory used gas chromatography/electronic capturedetection
(GC/ECD) with dual column confirmation for the analysis of PCB congeners and chlorinated pesticides in the
1992 Virginian Province sediment samples. Most values reported in the database for the PCBs and pesticides
represent "confirmed" results (i.e., the analyte was detected and could be quantified onboth the primary and secondary
columns). In general, the rate of second-column confirmation for all reported PCB congeners and chlorinated
pesticides was greater than 80%, with the following exceptions (confirmation rate in parenthesis): PCB 195 (75%),
heptachlor (26%), heptachlor epoxide (42%), lindane (35%), o,p DDT (77%), and p,p DDT (79%). Whenever
an analyte was detected on one column, but was not confirmed on the second column, the result was treated as
a "not detect" (i.e., the SC-A code is used to flag the result in the database). Whenever there was a significant
discrepancy in the amount detected on the two GC/ECD columns (/'. e., greater than a factor of 3 difference), the
lower of the two values is reported in the database and flagged with the "SC-D" code. Please note the warning
regarding use of this code discussed in Section 3.2.
Total Organic Carbon analyses
All QC results for the analysis of total organic carbon in the 1992 Virginian Province sediment samples fell
within required control limits. The Certified Reference Material PACS-1 (issued by the National Research Council
of Canada) was utilized as the LCM. The certified concentration of total carbon in thisreference material is 3.69%
(percent dry weight). The average percent recovery achieved by the laboratory for n = 8 batches of TOC samples
(i.e., eight separate analyses of CRM PACS-1) was 97.4%, with all values falling within the range 90% to 106%.
Since the PACS-1 certified concentration includes both organic carbon and a very small fraction of inorganic
carbon, the laboratory's percent recovery values for organic carbon generally are expected to be below 100%.
Based on the good overall percent recovery of organic carbon in the Certified Reference Material, the 1992 sediment
TOC data were deemed acceptable for use without qualification.
Page 26 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Butyltin analyses
Data users are cautioned that there are deficiencies in the 1992 Virginian Province sediment dataset for butyltin
compounds which might limit or preclude the use ofthese data. The laboratory detected dibutyltin (DBT) in only
18% and monobutyltin (MET) in only 3% of the samples analyzed in 1992, while tributyltin (TBT) was detected
in 73% of the samples. The MDLs established by the laboratory were 5 ng/g dry weight (as tin) for both TBT
and DBT, and 12 ng/g dry weight (as tin) for MET. It is possible that the butyltin compounds of interest were
present in many samples at concentrations below these detection limits, and, therefore, the occurrence of butyltin
compounds in Virginian Province sediments may be more widespread than indicated by these data.
The Certified Reference Material PACS-1 (issued by the National Research Council of Canada) was utilized
as the LCM for these analyses. Average percent recoveries relative to the certified value for n = 10 analysis sets
were 77% for TBT, 52% for DBT and 171% for MET. These values fall outside the acceptable accuracy control
limits of 80% to 120%; therefore, all values reported for TBT, DBT and MET in samples where these compounds
were detected are considered estimates (SC-C code) and should be used with discretion.
Acid volatile sulfides analyses
At present there are no Certified Reference Materials available for acid volatile sulfides. For the 1992 Virginian
Province samples, the laboratory utilized a laboratory fortified blank sample as the laboratory control material
(LCM). The average percent recovery of AVS for n = 68 laboratory fortified blank samples was 93%, suggesting
good overall analytical performance. Average percent recoveries for matrix spike samples were somewhat low
(55% for n = 9 matrix spike duplicate sets); these low recoveries were attributed to possible matrix effects. In
general, the 1992 AVS analyses were deemed acceptable, and no data qualifier codes were applied to these data.
3.3.5 1993 QA Results
A number of significant problems were uncovered during EMAP's QA review of the 1993 sediment chemistry
data. These included analytical problems, switched sample IDs, calculation errors, andtranscription errors. Several
samples were re-analyzed by either EMSL or ERL-N (after EMSL-Analytical's laboratory shutdown). All suspected
erroneous data have been qualified, corrected, or deleted. Specific discussions are found below.
Major and trace element analyses (except mercury)
For the analysis of major and trace elements by ICP-AES and GFAA, the laboratory generally met the pre-established
acceptability criteria (control limits) for the QC samples (e.g., calibration check samples, laboratory reagent blanks,
matrix spikes, and Laboratory Control Materials). For the ICP-AES analyses, which included the metals Al,
Cr, Cu, Fe, Mn, Ni, Pb, and Zn, a total of 18 analytical sets or "batches" of samples were analyzed. The Certified
Reference Material (CRM) "BCSS-1" (Estuarine Sediment, issued by the National Research Council of Canada)
was analyzed along with every batch as the Laboratory Control Material. With the exception of Cr, the average
percent recovery of each metal (relative to the certified concentration in BCSS-1) was within the acceptability
range of 80% to 120% (Table 3-11). The average percent recovery for Cr (73%) was slightly lower than acceptable,
suggesting that this metal may have been consistently "under-recovered" in the actual samples. Therefore, all
reported values for this metal were qualified with the SC-C code in the database.
The GFAA analyses included the metals Ag, As, Cd, Sb, Se, and Sn; a total of 18 analytical sets or "batches"
of samples were analyzed. The CRM BCSS-1 also was analyzed along with every sample batch as the Laboratory
Control Material. Average CRM percent recoveries for all metals fell within the acceptability range of 80% to
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 27
-------
120% (Table 3-11). The CRM BCSS-1 does not have a "certified" value for silver, but the average recovery
for this metal in laboratory spiked samples (matrix spikes) was within quality control limits. Although thepercent
recovery of all metals fell within the acceptable range, all values for As, Sb, and Se were qualified with the SC-C
code due to high variability of percent recovery in the matrix spiked samples for these metals.
Table 3-11. Summary results for CRM BCSS-1 (Estuarine Sediment) used as a set control for the 1993
EMAP-Estuaries sediment inorganic analyses.
ICP-AES METALS Cn =
Element
Al
Cr
Cu
Fe
Mn
Ni
Pb
Zn
Average1
91
73
101
92
97
84
101
87
GFAAMETALSm = 18
Element
Ag
As
Cd
Sb
Se
Sn
Average1
na
108
98
102
101
94
18 analysis sets or "batches"'):
Stdv2
5.1
1.8
2.5
1.9
1.5
2.6
18.9
2.4
analysis sets'):
Stdv2
na
10.9
12.6
21.2
26.6
10.3
C.V.3
5.6
2.5
2.5
2.1
1.5
3.1
18.7
2.8
C.V.3
na
10.0
12.9
20.7
26.3
11.0
Min.4
83
71
95
88
93
81
70
82
Min.4
na
84
71
67
66
77
Max.5
102
77
105
96
99
89
133
90
Max.5
na
123
123
139
143
116
1 Average percent recovery relative to the SRM certified value.
2 Standard deviation of the percent recovery values.
3 Coefficient of variation of the percent recovery values.
4 Minimum percent recovery for n analysis sets
5 Maximum percent recovery for n analysis sets
Page 28 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
During the QA review of the 1993 data, a problem was noted with the metals data associated with event 3145
(Station 188, chemID # 9303471). For Cu, Ni, Pb and Zn the ICP results did not agree well with data generated
from a second visit to that station, and the Simultaneously Extracted Metals (SEM) values for these metals were
significantly HIGHER than the bulk metals (1993 was the only year SEM metals were analyzed). EMSL checked
the ICP results andconfirmed them. They then re-extracted and re-analyzedthis sample and came up with significantly
higher results than for the first analysis; results similar to those from the duplicate sample. It appeared that they
may have switched samples, since the ratio of the difference between the original run and re-run was not consistent
among metals, ranging from a factor of 2 to a factor of 20. The results of the original analysis and re-analysis
are as follows:
Original value (ug/g) Reanalysis value (ug/g)
Cr 22 69
Cu 5.55 46
Mn 210 1,160
Fe 11,800 41,800
Ni 2.44 42
Pb 14.4 51
Zn 44.2 175
The reanalysis values agreed relatively well with the results from a second sample collected at that station
and from previous data from that station. The question was then raised "why were the original results so low?"
One possibility is that the sample was switched with another one from that batch. To determine if this was the
case, data on all samples in the batch were reviewed. Three sediments had similar values to those from the reanalysis
of 9303471, so were potential candidates. Since the EMSL chemistry laboratory was no longer active, these samples
were sent to ERL-N for analysis. The results of those analyses are as follows:
Reanalysis results (ug/g: EMSL's original results in parentheses)
Analyte Sample 9303328 Sample 9303400 Sample 9303477
Cr 23.2(59) 23.56(77.9) 9.13(68.6)
Cu 14.48(34) 10.30(38.8) 1.20(48.5)
Mn 299.75(1,120) 217.84(752) 80.89(1,220)
Fe 12,030(46,700) 8,834(36,000) 4,630(36,900)
Ni 11.77(20) 12.66(39.2) 2.53(39.1)
Pb 11.03(49) 9.74(70.8) 4.01(47.9)
Zn 50.09(123) 49.39(174) 13.48(181)
Several things are evident. For the first two samples listed, EMSL's values are consistently greater than those
determined by ERL-N by at least a factor of two. Theresults for sample 9303477 are very different. We believe
that in EMSLs original analysis, sample 9303477 was switched with 9303471. First, EMSLs reanalysis results
for sample 9303471 are very similar to those from 9303477 (which is why it was selected as one for reanalysis
at ERL-N). Second, ERL-N's reanalysis of sample 9303477 produced results much lower than the original results
for that sample.
Since EMSL reanalyzed 9303471 and came up with more "reasonable" values relative to a second sample
taken at that station, the original results were replaced with the reanalysis results. The results from 9393477
are suspected to be erroneous, and have been deleted from the database.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 29
-------
Mercury analyses
For the 1993 mercury analyses, the Certified Reference Material BEST-1 (issued by the National Research
Council of Canada) was analyzed along with every sample batch as the LCM (n = 7 sample batches). The average
percent recovery of 97% for mercury in this reference material fell well within the acceptability range of 80%
to 120%. In addition, an average percent recovery of 95% was achieved for the matrix spike samples analyzed
in each batch. Overall, these results indicate acceptable accuracy for the mercury analyses, and no "SC-C" codes
were used to qualify the data. The 1993 mercury results were deemed acceptable for use without qualification.
Organic analyses
In general, results for reagent blanks and calibration check samples analyzed with each batch of field samples
fell within control limits and serve to verify that sample contamination did not occur and that all instruments were
calibrated properly throughout the analytical runs. Average recoveries of compounds in matrix spike/matrix
spike duplicate samples generally fell within control limits, indicating acceptable analytical performance. However,
matrix spike samples are not the most ideal quality control samples because the analytes of interest are not truly
incorporated into the matrix in the same manner as an actual field sample. In addition, it can be difficult to evaluate
laboratory performance solely on the basis of matrix spike results because it is oftenequivocal whether low recoveries
are due to flawed methodology, poor technique, or a true matrix interference.
Given the above limitations related to the use of matrix spike samples to assess analytical performance, great
emphasis was placed on the LCM results. For both the PAH and PCB/pesticide analyses, SRM 1941 or 1941 a
(Organics in Marine Sediment, issued by NIST) was analyzed as the LCM along witheach batch of field samples.
For most of the individual PAH compounds and PCB congeners with "known" concentrations in SRM 194 la
(this includes both "certified" and "non-certified" values), the average percentrecovery achieved by the laboratory
(based on n = 10 batches for PAHs and n = 10 batches for PCB/pesticides) generally fell wilhin the control limit
range of 70% to 130% (Tables 3-12 and 3-13). Whenever the laboratory failed to achieve these average recovery
rates for a particular compound, all the results in the 1993 database for that compound were flagged with the
"SC-C" code to indicate the potential inaccuracy inferred from the SRM analysis. It is important to note that
the 70% to 130% recovery criteria only applies to compounds having SRM concentrations greater than 10 times
the laboratory's detection limit. When compounds occur at concentrations less than about 10 times the detection
limit, a greater amount of analytical uncertainty is expected and the normal control limit "acceptability" criteria
do not apply.
Based on the above, the results for the following organic compounds were flagged with the "SC-C" code in
the 1993 database: PCB 101, PCB 18 andchrysene. Although the concentration of PCB 18 in the SRM was less
than 1 Ox the detection limit, the very high mean percent recovery (343%) and high variability (range of recoveries
from 90% to 1,550% with a CV of 138%) resulted in the SC-C code being applied to all values for PCB 18. In
addition, although the average percent recovery for PCB 206 was within limits (110%), all results for this compound
were flagged with the SC-C code because the recoveries between batches exhibited relatively high variability
(i.e., 78% coefficient of variation).
The SC-C code was also applied to several specific samples forwhich the data were suspect (e. g., poor agreement
between field splits). As in previous years, all pesticide data were qualified with this code. The problems identified
with the pesticide data from station 725 (see below, Table 3-14) support this action.
Further review of the data illuminated additional concerns regarding the 1993 PCB results. One problem
noted concerned the formulation and use of control charts for assessing precision. Of the ten sample batches
submitted, the first two contain results for SRM 1941 while the remaining eight used SRM 194la (both SRMs
Page 30 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
are acceptable for use with marine sediment). The control limits for SRM1941a results were computed using
the SRM1941 (wrong SRM) results from the first two batches, the SRM194la results from two 1993 batches
(control limits being set by the same data on which they are to measure precision), and two additional SRM 194 la
samples prepared earlier. The associated control limits are so wide (24-162%RSD with average 44% for all
analytes) that in all of the eight batches to which they apply only once did an analyte exceed them (PCB 18 at
> 1500% recovery). These data may nothave exceeded control limits but they are not necessarily precise. Coefficients
of variation for SRMs calculated in previous reviews tend to suggest the 1993 PCB and pesticide data are the
most variable since the 1990 data set, which has the highest variability. As a result, all 1993 PCB data have
been assigned the "SC-C" code indicating potential problems with the data.
A detection limit of 0.25 ng/g (dry weight) generally was achieved for each PCB congener and pesticide and
a detection of 10.0 ng/g (dry weight) was achieved for each PAH compound in the majority of samples analyzed.
Some problems with specific samples were also noted. As part of a review of the field split data it was noticed
that the total of PAHs from 3069030 was 1,438 ng/g and from the split (3006030) all PAHs were non-detected.
The response from EMSL was "The pesticide data for these two field duplicates appear fairly high and agree
well, even though no PAHs were found in 9302940 [3006030], yethigh levels were reported in 9302943 [3069030].
Examination showed that the samples were correctly identified and appeared similar. However, examination
of the extracts showed them to be distinctly different. There was an extract in that same run 9302933 [3115030],
which exhibited similar results to 9302043. We are checking to see if we have enough sample left to re-extract
and verify our results."
EMSL did re-extract and run all three samples. Results showed that they did indeed mis-label the original
extracts for 3006030 and 3115030. The re-analysis of sample 3069030 showed poor precision relative to the
original run, with differences of concentrations between runs approaching a factor of 3 (e.g., 132 vs 329). The
original data have been replaced with the results of the re-analysis.
As part of the QA review of the data, ERL-N scientists noticed that the concentrations of dieldrin and p,p'
DDT from station 725 in the Providence River, RI appeared unreasonably high. This sample was shipped to ERL-N
and analyzed for pesticides and PCBs. The results can be found in Table 3-14. The dieldrin and DDT results
were found to be erroneous. Because of the uncertainty in pesticide results from this station, all 1993 pesticide
data from station 725 have been deleted from the database.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 31
-------
Table 3-12. ResultsforSRM 1941 and 1941 a (Organics in Marine Sediment) used as theset control (Laboratory
Control Material) for the 1993 sediment PAH analyses (n = 10 analysis sets or "batches"). Note
that since the results are presented as percent recovery, 1941 results and 1941 a results were
not separated.
Compound1
Acenaphthene
Acenaphthlylene
Anthracene
Benz(a)anthracene
Benzo(a)pyrene
Benzo(e)pyrene
Benzo(b+k)fluoranthene
Benzo(g,h,i)perylene
Biphenyl
Chrysene
2,6-dimethylnaphthalene
Fluoranthene
Fluorene
ldeno(1 ,2,3-c,d)pyrene
1-methylnaphthalene
2-methylnaphthalene
1-methylphenanthrene
Naphthalene
Perylene
Phenanthrene
Pyrene
Average2
118
73
86
119
93
104
129
99
73
136
74
97
109
107
70
88
92
88
84
102
93
Stdv3
16.6
17.8
15.2
11.4
22.7
22.4
23.5
22.4
17.9
9.7
20.1
6.8
8.7
20.1
17.2
18.0
17.4
25.3
21.5
10.7
9.5
C.V.4
14.1
24.4
17.8
9.6
24.4
21.5
18.2
22.6
24.7
7.2
27.0
7.1
8.0
18.8
24.4
20.5
19.0
28.8
25.6
10.5
10.2
Min5
100
61
57
101
68
80
97
74
60
120
55
88
102
82
47
57
69
49
63
88
82
Max6
133
94
110
137
137
147
176
146
112
153
116
110
119
143
106
120
120
120
131
120
109
1 Listed compounds include those having both "certified" and "non-certified" concentrations in SRM 1941 a.
2 Average percent recovery relative to the SRM value.
3 Standard deviation of the percent recovery values.
4 Coefficient of variation of the percent recovery values.
5 Minimum percent recovery for 10 analysis sets
6 Maximum percent recovery for 10 analysis sets
Page 32 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 3-13. ResultsforSRM 1941 and 1941 a (Organics in Marine Sediment) used as theset control (Laboratory
Control Material) forthe 1993 sediment PCB/pesticide analyses (n = 10 analysis sets or"batches").
Note that since the results are presented as percent recovery, 1941 results and 1941 a results
were not separated.
Compound1
Average2
Stdv3
C.V.4
Min5
Max6
PCB8*
PCB 18*
PCB28
PCB 44
PCB 52
PCB 66*
PCB 101
PCB 118
PCB 153
PCB 105
PCB 128*
PCB 138
PCB 187
PCB 180
PCB 170
PCB 206
PCB 209
Dieldrin*
cis-Chlordane*
trans-Nonachlor*
Hexachlorobenzene
2,4'-DDE*
4,4'-DDE
4,4'-DDD
4,4'-DDT*
106
343
77
84
93
119
69
76
72
84
77
91
91
121
96
110
81
136
164
89
76
250
110
99
137
51.6
473
13.4
26.0
22.7
58.9
14.9
11.7
16.6
21.1
24.9
21.9
23.6
24.0
31.5
86.2
13.8
94.8
66.9
25.8
21.2
123
28.3
21.0
168
48.5
138
17.3
30.8
24.4
49.6
21.3
15.4
23.1
25.1
32.4
23.9
26.1
19.8
32.7
78.4
17.0
69.6
40.7
29.0
27.9
49.1
25.7
21.3
122
39
90
61
59
68
84
53
64
51
44
40
66
65
86
38
50
64
49
97
43
48
112
67
72
12
218
1550
99
129
125
278
96
96
96
117
113
144
134
172
132
310
101
370
259
122
114
466
169
145
582
1 Listed compounds include those having both "certified" and "non-certified" concentrations in SRM 1941 a
(* = concentration in the SRM is less than 10 times the target detection limit).
2 Average percent recovery relative to the SRM value.
3 Standard deviation of the percent recovery values.
4 Coefficient of variation of the percent recovery values.
5 Minimum percent recovery for 10 analysis sets
6 Maximum percent recovery for 10 analysis sets
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
Page 33
-------
Table 3-14. Results of re-analysis of sediments from Station 725 (results in ng/g dry weight).
Analyte
PCB8
HCB
PCB18
PCB028
PCB52
PCB44
PCB66
PCB101
PPDDE
PCB118
PCB153
PCB105
PCB138
PCB187
PCB128
PCB180
PCB170
PCB195
PCB 206
PCB209
LINDANE
CISCHLORDANE
T-NONACHLOR
PPDDD
PPDDT
Dieldrin
ERL-N
Results
1.32
0.30
3.50
6.02
11.9
6.42
8.16
21.7
10.8
16.6
25.6
7.52
27.1
14.4
3.63
19.4
6.85
6.09
12.2
13.2
0.19
3.76
2.80
20.9674
1.37
Not
analyzed
ERL-N
Replicate
1.42
0.29
3.18
5.35
10.1
5.64
6.02
17.5625
10.4
14
21.7
5.56
20.8
10.7
2.91
15.9
5.62
4.53
9.57
10.3
0.69
3.51
2.63
18
4.02
8.92
EMSL Results
4.42
0.71
17.6
11.3
15.8
11.2
11.6
57.8
65.3
50.3
76.6
10.4
97
18.3
8.17
82
14.6
10.2
13.2
18.6
nd
17.2
3.06
62.4
251
170
EMSL Confirmation
column
2
16
3.6
5.4
29
87
58
Page 34
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Total Organic Carbon analyses
All QC results for the analysis of total organic carbon in the 1993 sediment samples fell within required control
limits. The Certified Reference Material PACS-1 (issued by the National Research Council of Canada) was utilized
as the LCM. The certified concentration of total carbon in this reference material is 3.69% (percent dry weight).
The average percent recovery achieved by the laboratory for n = 8 batches of TOC samples (i.e., 8 separate analyses
of CRM PACS-1) was 95.8%, with all values falling within the range 90% to 106%. Since the PACS-1 certified
concentration includes both organic carbon and a very small fraction of inorganic carbon, the laboratory's percent
recovery values for organic carbon generally are expected to be below 100%. Based on the good overall percent
recovery of organic carbon in the Certified Reference Material, the 1993 sediment TOC data were deemed acceptable
for use without qualification.
Butyltin analyses
Data users are cautioned that there are deficiencies in the 1993 sediment dataset for butyltin compounds which
might limit or preclude the use of these data. The MDLs established by the laboratory were 5 ng/g dry weight
(as tin) for both TBT and DBT, and 12 ng/g dry weight (as tin) for MET. It is possible that the butyltin compounds
of interest were present in many samples at concentrations below these detection limits, and, therefore, theoccurrence
of butyltin compounds in Virginian Province sediments may be more widespread than indicated by these data.
The Certified Reference Material PACS-1 (issued by the National Research Council of Canada) was utilized
as the LCM for these analyses. Average percent recoveries relative to the certified value for n = 11 analysis sets
were 74% for TBT, 74% for DBT and 188% for MET. These values fall outside the acceptable accuracy control
limits of 80% to 120%. Therefore, all values reported for TBT, DBT and MET in samples where these compounds
were detected are considered estimates (SC-C code) and should be used with discretion.
Acid volatile sulfides analyses
At present there are no Certified Reference Materials available for acid volatile sulfides. For the 1993 samples,
the laboratory utilized a laboratory fortified blank sample as the LCM. The average percent recovery of AVS
for n = 60 laboratory fortified blank samples was 94%, suggesting good overall analytical performance. With
the exception of two batches the 1993 AVS analyses were deemed acceptable.
The "SC-C" code was applied to all results from AVS run #3 84. This batch contained one set of blind field
splits. The AVS concentration reported for one of the splits was 22.6 mg/kg. The concentration reported for
the second was 1.77 mg/kg. The EMSL ran duplicate analyses (as part of routine QA) on the second sample and
determined a concentration of 7 mg/kg. All three of these concentrations should be the same. The differences
suggest a lack of precision for this batch; therefore all samples analyzed as part of this batch were flagged. It
should be noted that the precision of laboratory duplicates for AVS was generally good.
The "SC-C" code was also applied to all results from AVS run#387. As part of the QA protocol the laboratory
is required to run one sample in duplicate. The relative percent difference between duplicates must be less than
20%. The reported concentrations of the duplicate (LD1 and LD2) were 2100 and 2240 mg/kg with a RPD of
6.4% HOWEVER, the QA summary provided with that batch stated "The original analysis of the duplicate on
9/2/93 [the day all the samples from that batch were analyzed] did not meet the RPD acceptance criterion. The
duplicate analyses were performed again on 9/16/93, and the results reported here. LFM and CLE recoveries
associated with the reanalysis of the duplicates were reported in Run 383." The entire batch should have been
reanalyzed. This shows that the precision on the day the samples were run was poor. The original data were requested
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 35
-------
from EMSL. The original LD1 and LD2 were 1910 and 3950 mg/kg with an RPD of 70%. When EMSL provided
us the original numbers they also informed us of a transcription error in the moisture content of the sample. A
value of 62.9% was entered instead of 82.9%. Therefore the values reported in the original file of 2100 and 2240
mg/kg should be 4440 and 4740 mg/kg respectively. However, the moisture content reported in the electronic
file originally received from Cincinnati was, infact, 83% (rounded), not 63%. EMSL investigated this andinformed
us that despite the fact that our file lists 83% as the water content used to calculate the original numbers, and
83% is the correct value, a value of 63% was used in the calculation of the AVS concentrations. They could not
explain how this happened.
Page 36 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Section 4
QA Results for Fish Contaminant Analyses
4.1 Background
Measurement Quality Objectives for the analysis of chemical contaminants in EMAP-E tissue samples are
specified in the 1991 Province Quality Assurance Project Plan (Valente and Schoenherr 1991). This plan requires
each EMAP-E laboratory to analyze the following types of quality control samples along with every batch or
"set" of field chemistry samples: laboratory reagent blanks, calibration check standards, laboratory fortified sample
matrix (matrix spike), laboratory fortified sample matrix duplicate (matrix spike duplicate), laboratory duplicate,
and Laboratory Control Material. Results for these QC samples must fall wilhin certain pre-established control
limits for the analysis of a batch of samples to be considered acceptable.
Standard or Certified Reference Materials typically are used by EMAP-E laboratories as their Laboratory
Control Material. SRMs and CRMs have known or "certified" concentrations of the analytes being measured
and therefore are useful for assessing both accuracy and precision. The QA Project Plan requires the laboratory's
percent recovery (relative to the certified concentration in the reference material) to fall within certainpre-established
control limits to be considered acceptable. If the laboratory consistently fails to meet these acceptability criteria
for the CRM or SRM analysis, the values reported for the failed analytes are considered to be suspect (biased)
and are flagged in the database, as described in the following section.
Fish were collected from trawls conducted at each station. Individuals of "target" species were selected for
contaminant analysis. These individuals were tagged, wrapped in aluminum foil, and frozen. In the laboratory,
fish were cleaned, scaled, fileted, and composited by species. An analytical sample consisted of the edible flesh
from three to five individuals of a single species from a station.
Because of budget constraints and the poor distribution of target species across the Province, 1991 was the
only year in which fish were analyzed for contaminants.
4.2 1991 Results
Major and trace element analyses
For the 1991 Virginian Province analysis of major and trace elements, the laboratory generally met the pre-established
acceptability criteria (control limits) for the QC samples (e.g., calibration check samples, laboratory reagent blanks,
matrix spikes, and LCMs). The control limits for inorganic analytes is ± 20% of the CRM certified value. These
criteria were generally met (Table 4-1). The average percent recovery for Pb (DOLT) was slightly high; however,
the value for the DORM CRM was within the acceptable range and the confidence intervals around the DOLT
certified value were rather large.
A problem was noted by the laboratory in analysis of selected samples for mercury. The laboratory analyzed
84 composite samples and 40 individual fish. The analytical laboratory experienced a mercury-contamination
problem with their freeze-drier, resulting in contamination of all 40 individual-fish samples. As a result, these
data had to be deleted from the database. However, EMAP-VP's assessment was focused on the composite samples,
and none of these were contaminated.
With the removal of the above-mentioned Hg data from the database, the only flags applied are the "A" and
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 37
-------
"B" codes described in the sediment chemistry section (Section 3.2).
Table 4-1. Summary results for CRMs DOLT and DORM (Dogfish liver and muscle tissue,
respectively) used as a set control for the 1991 Virginian Province fish tissue inorganic
analyses. Average reported values are based on six separate analyses of the CRMs.
Element
As
Cd
Cr
Cu
Fe
Hg
Ni
Pb
Se
Zn
DOLT
DORM
DOLT
DORM
DOLT
DORM
DOLT
DORM
DOLT
DORM
DOLT
DORM
DOLT
DORM
DOLT
DORM
DOLT
DORM
DOLT
DORM
Average1
101.2
99.3
83.3
93.0
118.8
106.3
91.4
79.4
98.9
105.7
NA
91.9
103.8
89.0
130.4
89.5
102.1
102.8
101.7
100.7
Stdv2
2.5
2.5
8.2
10.4
15.2
8.8
9.1
5.3
1.8
11.9
NA
4.9
52.9
13.5
25.7
35.3
2.9
4.6
2.7
3.9
C.V.3
2.5
2.5
9.8
11.2
12.8
8.3
10.0
6.7
1.8
11.2
NA
5.3
50.9
15.2
19.7
39.5
2.8
4.4
2.7
3.8
Min.4
98.0
94.9
69.6
81.4
102.5
97.5
81.7
75.7
96.2
95.4
NA
86.5
50.0
76.7
92.7
55.0
99.7
96.9
98.0
96.2
Max.5
104.0
101.7
90.9
104.7
137.5
117.5
107.2
88.9
101.3
125.0
NA
100.3
188.5
111.7
164.7
142.5
107.4
108.0
105.7
106.6
1 Average percent recovery relative to the CRM certified value.
2 Standard deviation of the percent recovery values.
3 Coefficient of variation of the percent recovery values.
4 Minimum percent recovery for analysis sets
5 Maximum percent recovery for analysis sets
Organic analyses
Due to a miscommunication within the analytical laboratory, EMAP QA protocols were not followed during
the analysis of EMAP-VP 1991 fish tissue samples for organic analytes. However, sufficient data are available
for an evaluation of the quality of those samples. First, prior beginning processing of EMAP samples the laboratory
participated in a performance evaluation. Based on 11 separate analyses of SRM1974 (Organics in Mussel Tissue),
it was determined that the laboratory was sufficiently proficient to begin analyzing EMAP samples. The results
of this performance evaluation are listed in Table 4-2. Second, matrix spiked samples were analyzed with each
batch, and these results fell well within EMAP's control limits (Table 4-3). Third, during the same time period
when thelaboratory was processing EMAP samples, they were also processing samples for NOAA's NS&T Program.
SRM 1974 was used as the laboratory control material for those samples, and was analyzed with each analytical
batch. The laboratory has provided EMAP with those results, which fall within EMAP control limits. Fourth,
the QA protocols the lab followed for EMAP samples require the analysis of duplicate samples with each batch.
Those results were provided to EMAP and showed excellent precision, with a maximum Relative Percent Difference
for an analyte in a given set generally being less than 10%.
Page 38
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
The only flags applied are the "A" and "B" codes described in the sediment chemistry section.
Table 4-2. Performance evaluation results for analysis of organic contaminants in tissue. Average
reported values are based on 11 separate analyses of SRM 1974 (Organicsin Mussel
Tissue) performed on different days.
Analyte
alpha-chlordane
trans-nonachlor
Dieldrin
2,4'-DDE
4,4'-DDE
2,4'-DDD
4,4'-DDD
2,4'-DDT
4,4'-DDT
PCB18
PCB28
PCB44
PCB52
PCB66
PCB101
PCB105
PCB118
PCB128
PCB138
PCB153
PCB180
PCB187
Average
reported
value
21.2
17.7
11.3
M2
41.4
5.8
46.5
5.0
3.6
20.9
85.2
72.4
113.7
98.7
127.0
46.9
115.9
17.3
122.2
153.9
13.3
27.2
NIST non-
certified
value1
26 ± 1
21 ±5
8±4
5.8 ±0.6
48 ±2
20 ±7
68 ±3
4± 1
3±2
24 ±9
62 ±3
65 ±23
98 ±39
110±5
105± 11
45 ±3
110±5
15±2
110± 11
145 ±8
13± 1
30 ± 1
Percent
difference
-15%
0%
0%
NA
-10%
-55%
-28%
0%
0%
0%
31%
0%
0%
-6%
9%
-2%
1%
2%
1%
1%
0%
-6.2%
1 NIST non-certified values with 95% confidence intervals presented in the certificate of analysis for
SRM 1974. Reported values falling within these confidence intervals are listed as having a percent
difference of 0%.
2 Matrix interference, no peak was found for 2,4'-DDE
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 39
-------
Table 4-3. Results of laboratory-fortified matrix spikes analyzed with each batch of fish tissue organic
samples analyzed (n=10). Values are percent recovery of the spike.
Analyte Average1 Stdv2 C.V.3 Min.4 Max.5
aldrin 95.2 10.9 11.5 83 114
alpha-chlordane 100.2 10.9 10.9 82 112
trans-nonachlor 99.3 12.1 12.2 80 117
Dieldrin 95.2 14.0 14.6 71 118
2,4'-DDE 94.4 9.6 10.2 84 112
4,4'-DDE 99.1 10.5 10.6 86 118
2,4'-DDD 101.6 9.3 9.2 87 112
4,4'-DDD 98.7 12.7 12.9 76 118
2,4'-DDT 101.9 12.5 12.3 79 120
4,4'-DDT 100.5 15.2 15.1 74 118
Total PCBs 99.8 7.7 7.7 87 114
1 Average percent recovery relative to the concentration of the spike.
2 Standard deviation of the percent recovery values.
3 Coefficient of variation of the percent recovery values.
4 Minimum percent recovery for analysis sets
5 Maximum percent recovery for analysis sets
Page 40 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Section 5
QA Results for Particle Size Analyses
5.1 Background
At each station crews collected three samples for benthic infaunal analysis and a sediment homogenate which
was split for chemistry and toxicity analyses. Associated with each of these samples was an aliquot removed
for particle size analysis (percent silt/clay). The annual QA Plans require that approximately 10% of theseanalyses
be performed in duplicate and the maximum allowable percent difference for the predominant fraction (silt/clay
or sand) is 10%.
5.2 Laboratory Audits
In 1990 sediment particle size analyses were performed at Versar, Inc. in Columbia, MD. This facility was
audited by the EMAP-E QAO during the period 15-16 November 1990. No major problems were identified in
this audit. The main recommendation in the audit report was the need for minor revisions, mostly in the form
of clarifications, to Versar's methods manual and data forms. Versar met all measurement quality objectives
in performing the grain size analyses on 1990 samples.
In 1991 and 1993 particle size analyses were performed by SAIC, on-site at the EPA Environmental Research
Laboratory inNarragansett, RI. Although no formal auditswere performed, SAIC technicians were closely monitored
by the EMAP-VP QA Coordinator (located in the same facility). Prior to the start of analysis, the technician
was required to demonstrate proficiency through the analysis of sediments with a variety of grain sizes. Results
of these analyses met EMAP QA criteria and the technician was permitted to begin analysis of EMAP samples.
In 1992 particle size analyses were performed by EPA personnel at the EPA Environmental Research Laboratory
in Narragansett, RI. Although no formal audits were performed, EPA technicians were closely monitored by the
EMAP-VP QA Coordinator (located in the same facility). Prior to the start of analysis, the technician was required
to demonstrate proficiency through the analysis of sediments collected by EMAP in 1991 with a variety of grain
sizes. Results of these analyses met EMAP QA criteria and the technician was permitted to begin analysis of
1992 EMAP samples.
5.3 Qualifier Codes for Particle Size Data
No codes currently exist for particle size data, indicating all datameet QA criteria and are suitable for EMAP
assessment purposes.
5.4 1990 QA Results
All "sediment grain size" and "benthic grain size" samples collectedperstation were analyzedforthe determination
of percent silt/clay. Approximately 10% of these analyses were performed in duplicate and the percent difference
determined as per the EMAP-VP 1990 QA Project Plan. The maximum allowable percent difference for the predominant
fraction (silt/clay or sand) is 10%. The mean difference for the samples analyzed was 2.78%, with none exceeding
10% so no remedial action or retesting was required.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 41
-------
5.5 1991 QA Results
Because of budget constraints, not all benthic grain size samples were analyzed. However, at least one sample,
of the three collected per station, was analyzed. These data are solely used in the interpretation of benthic community
data, and all parties, including those conducting the benthic infaunal analyses agreed on its acceptability. Grain
size information presented in the Statistical Summaries is from "sediment grain size" analyses.
AH"sedimentgrainsize"andatleastone"benthicgrainsize" sample per station were analyzed forthe determination
of percent silt/clay. Approximately 10% of these analyses were performed in duplicate and the difference determined
as per the EMAP-VP 1991 QA Project Plan. The maximum allowable percent difference for the predominant
fraction (silt/clay or sand) is 10%. The mean difference for the samples analyzed was less than 1%, with none
exceeding 10% so no remedial action or retesting was required.
5.6 1992 QA Results
All "sediment grain size" and "benthic grain size" samples were analyzed for the determination of percent
silt/clay. Approximately 10% of these analyses were performed in duplicate and the difference determined as
per the EMAP-VP 1992 QA Project Plan. The maximum allowable percent difference for the predominant fraction
(silt/clay or sand) is 10%. The mean difference for the samples analyzed was less than 1%, with none exceeding
10% so no remedial action or retesting was required.
5.7 1993 QA Results
All "sediment grain size" and "benthic grain size" samples were analyzed for the determination of percent
silt/clay. Approximately 10% of these analyses were performed in duplicate and the difference determined as
per the EMAP-VP 1993 QA Project Plan. The maximum allowable percent difference for the predominant fraction
(silt/clay or sand) is 10%. The mean difference for the 50 samples analyzed in duplicate was 1.5%, with none
exceeding 5% so no remedial action or retesting was required (the control limit is 10%). In addition, the 1993
QA Project Plan states that if the relative standard deviation (RSD) among the three benthic grain size samples
collected from a single station exceeds 20%, the calculations should be checked by the laboratory. The RSD for
samples from three stations exceeded 20% and those samples were reanalyzed. The results did not change.
Page 42 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Section 6
QA Results for Sediment Toxicity Testing
6.1 Background
Ten-day laboratory toxicity tests, using the amphipodAmpelisca abdita, were performed on sediments collected
at each station in the Virginian Province. This test was employed during each year of monitoring of the Province.
The QA Plans describe certain requirements for this test to be acceptable, including minimum control survival,
physical characteristics (temperature, salinity, dissolved oxygen concentration), and the use of water-only reference
toxicant tests.
In addition to theAmpelisca test, in 1990 samples from low-salinity waters were tested using the freshwater
amphipod Hyalella azteca to test the response ofAmpelisca in low salinity habitats. As a result of this testing,
it was determined that the results of the Ampelisca test conducted on low-salinity sediments were representative,
and the Hyalella test was not utilized in subsequent years.
6.2 Data Qualifier Codes For Sediment Toxicity
Ten data qualifier codes, or "flags, exist to describe EMAP-VP sediment toxicity data (Table 6-1), based
on the criteria described in the Virginian Province Quality Assurance Plans. Data for some tests that failed QC
are included in the dataset because, under certain circumstances, they may be of use to non-EMAP users. These
data are flagged with the code describing why they are unacceptable (e.g., ST-D) and the ST-L code, indicating
that they were not used in EMAP's assessment of the ecological condition of the Province.
An example of why these data were kept in the database is as follows. Control survival in a particular test
was unacceptably low (e.g., 50%), and there was insufficient sediment available to repeat the test. All data associated
with this test would automatically receive the ST-D and ST-I flags. However, survival in some of the treatments
(/'. e., test chambers with sediment from individual stations) was high (e.g., 95%). The conclusion could be drawn
that, because of the high survival in the experimental treatment, the sediment fromthat station is not toxic. Users
interested in data from particular stations may find this information useful. However, when control survival is
low there is no means by which to classify sediment as toxic. Therefore, treatments associated with an unacceptable
control may be classified as non-toxic or unknown: they can never be classified as toxic. This results in an inherent
bias which makes these data unacceptable for use by EMAP; therefore they are flagged with the ST-L code.
6.3 Laboratory Audits
Sediment toxicity testing was conducted at the SAIC Environmental Testing Center. This facility was audited
during the period 20-23 August 1990 by a team consisting of the EMAP-E QAO, the EMAP QA Coordinator,
and technical representatives of the EMSL-Cincinnati laboratory. No major problems were identified; however,
the audit report noted that SOP's needed to be developed and in-house audits should be performed more frequently.
Corrective actions were implemented in response to the audit recommendations as described in a memo from
the EMAP-E QA Coordinator to the EMAP QA Coordinator dated 16 October 1990.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 43
-------
Table 6-1. QA Qualifier Codes associated with sediment toxicity data.
Code Description
ST-A No QA Comment
ST-C Fewer than 5 replicates were tested
ST-D Mean control survival < 85 %
ST-E Sample held for >30 days prior testing
ST-G No reference toxicant test was run
ST-H Hardness and alkalinity not measured (1990 only)
ST-I Control survival in one replicate <80%
ST-J Physical parameters out of bounds
ST-K <20 animals used per replicate
ST-L Not Used in Province Assessment
One of the findings of this audit was that the two week maximum holding time for samples created logistical
problems for the laboratory, and data demonstrate that extending this to fourweeks does not result in degradation
of the sample. As a result, the holding time for sediment toxicity samples was increased from two to four weeks.
A follow-up audit was conducted by the EMAP-VP and EMAP-E QA Coordinators on 5 September 1991.
Their findings showed the laboratory to have corrected any short-comings noted in the earlier audit, and they
were particularly impressed by the lab's in-house documentation and tracking programs. One recommendation
made by the auditors was that the representativeness of preserving amphipods in formalin at the completion of
a test, rather than picking them live, be assessed. Data showing the representativeness of this methodology were
provided by the laboratory and were satisfactory.
6.4 1990 QA Results
As per the QA Project Plan, the laboratory was required to maintain a control chart for toxicity testing using
a reference toxicant. The laboratory used cadmium chloride as their reference material, running a standard 96-hour
water-only toxicity test whenever EMAP samples were run. The control chart shows that the LC50 for cadmium
chloride ranged from approximately 0.4 to 1.2 mg/L, with all values falling within two standard deviations of
the mean as required in the QA Plan.
Of the 126 samples from base stations analyzed (including those duplicated for the Hyalella test), only one
was assigned the "ST-L" code.
Page 44 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
6.5 1991 QA Results
As per the QA Project Plan, the laboratory was required to maintain a control chart for toxicity testing using
a reference toxicant. This provides an indication of the "quality" of the test organisms relative to those previously
used. The laboratory used SDS (sodium dodecyl sulfate) as their reference material, running a standard 48-hour
water-only toxicity test with SDS whenever EMAP samples were run. The control chart shows that the LC50
for SDS ranged from 4.0 to 8.37 mg/L, with all values falling within two standard deviations of the mean as required
in the QA Plan.
Several tests failed to meet EMAP QA requirements for control organism survival. Field crews recollected
sediment from those stations included in the failed tests. Of the 19 tests run, three exhibited control organism
survival less than the required 85% (this was following repeating all tests that failed on the first attempt). These
tests were assigned the "ST-L" flag or were deleted from the database and were not included in the dataset utilized
in EMAP's assessment of the ecological condition of the Province. As a result of these failures, the volume of
sediment collected at each station was increased in 1992 to allow for retesting without the need to redeploy crews
for additional sediment collection.
6.6 1992 QA Results
As per the QA Project Plan, the laboratory was required to maintain a control chart for toxicity testing using
a reference toxicant. The laboratory used SDS (sodium dodecyl sulfate) as their reference material, running a
standard 48-hour water-only toxicity test with SDS whenever EMAP samples were run. The control chart shows
that the LC50 for SDS ranged from < 2.57 to 11.2 mg/L, with all but the lowest value falling within two standard
deviations of the mean as required in the QA Plan (one in 20 tests would be expected to fall outside of two standard
deviations). Results of the one reference toxicity test falling outside two standard deviations of the mean were
examined, as were all tests performed during the same time period. No anomalies in the tests were apparent and
no re-testing was performed.
No samples were assigned the "ST-L" code.
6.7 1993 QA Results
As per the QA Project Plan, the laboratory was required to maintain a control chart for toxicity testing using
a reference toxicant. The laboratory used SDS (sodium dodecyl sulfate) as their reference material, running a
standard 48-hour water-only toxicity test with SDS whenever EMAP samples were run. The control chart shows
that the LC50 for SDS ranged from 5.32 to 8.59 mg/L, with all the values falling within two standard deviations
of the mean as required in the QA Plan. Several treatments contained fewer than five replicates (ST-C code),
but no infractions were serious enough to warrant discarding data.
No samples were assigned the "ST-L" code.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 45
-------
Section 7
QA Results for Macrobenthic Community Assessments
7.1 Background
Three replicate sediment samples were collected by field crews at each station for macrobenthic community
assessments, including species composition, abundance, and biomass. Two QAsteps were required by the EMAP-VP
1990-1993 QA Project Plans: in-house QC checks (i.e., resorts, recounts, and ID confirmation) on 10% of each
technician's work, and verificationofspeciesidentification by anindependentlaboratory. Both laboratories performing
these analyses, as well as the experts contracted for the independent verification of species taxonomy, have a
long record of performing benthic infaunal analyses.
7.2 Laboratory Audits
Macrobenthic community assessments for freshwater samples were performed at Versar, Inc. in Columbia,
MD. Versar has subcontracted another laboratory (Cove Corporation in Lusby, MD) to perform the macrobenthic
community analyses on samples from saline environments. The two facilities were audited by the EMAP-E QAO
during the period 15-16 November 1990. No major problems were identified in this audit. The main recommendation
in the audit report was the need for minor revisions, mostly in the form of clarifications, to Versar's methods manual
and data forms. Both Versar and Cove Corporation met all measurement quality objectives in performing the
grain size and benthic community analyses on 1990 samples.
No subsequent audits were conducted; however, voucher specimens were evaluated by independent laboratories
as described in Sections 7.4 and 7.5.
7.3 Data Qualifier Codes for Benthic Community Analyses
No codes currently exist for benthic community analyses, indicating all data meet QA criteria and are suitable
for EMAP assessment purposes.
7.4 1990 QA Results
Two QA steps were required by the EMAP-VP 1990 QA Project Plan: 10% recounts and independent verification
of species identification. The recounts (multiple types - see Table 7-1) and preliminary species verification were
performed by the laboratory responsible for the analyses. These in-house QC measures met the requirements
established in the QA Plan. Definitive verification of species identification was performed by an independent
laboratory and the results are described below.
External reviews of the taxonomic reference collections (i.e., voucher specimens) maintained by both Versar
and Cove were completed in 1990. Taxonomic experts at SAIC's Woods Hole office performed the review of
the Cove Corporation reference collection of marine macroinvertebrates. This review disclosed that less than
5% of the total number of species had been misidentified. The species misidentifications subsequently were corrected
in the EMAP-E database and the taxonomic experts atCove Corp. used these results to improve their future accuracy
for the species in questions.
Page 46 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 7-1. Results of recounts performed by the laboratory processing benthic infauna samples in 1990.
Approximately 10% of all samples were processed in duplicate.
Measurement
Benthic sorting
Species identification and enumeration
Biomass
Weighing blanks for biomass
Mean Error
3.06%
1 .37%
0.23%
<0.0001g
Range of Error
0
0
0
0
- 1 0%
- 7.7%
- 1 .24%
-0.0013g
7.5 1991 QA Results
Two QA steps were required by the EMAP-VP 1991 QA Project Plan: in-house QC checks (i.e., resorts,
recounts, and ID confirmation) on 10% ofeach technician's work, and independent verification of species identification.
The recounts (multiple types - see Table 7-2) and preliminary species verification were performed by the laboratory
performing the analyses. Most of these met the requirements established in the QA Plan. Definitive verification
of species identification was performed by an independent laboratory and the results are described below.
A total of 137 specimens collected from oligohaline stations were sent to the Aquatic Resources Center in
Franklin, TN for independent taxonomic verification. Eleven (8%) were mis-identified, representing 8 species.
The identification of an additional 15 specimens could not be confirmed because of the condition of the specimen
(e.g., key taxonomic features missing or destroyed, or male needed for identification and only females sent).
The identification of many of these species is difficult. Misidentified species were closely related taxonomically
to the "true" species. In general, the report on species verification was "largely favorable" indicating the analytical
laboratory performed well. Suggestions were made regarding identification of tubificid oligochaetes and mollusks
prior to the next season.
Table 7-2. Results of recounts performed by the laboratory processing benthic infauna samples in 1991.
Approximately 10% of all samples were processed in duplicate.
Measurement
Benthic sorting
Species identification and enumeration
Biomass
Weighing blanks for biomass
Mean Error
4.5%
2.4%
0.13%
O.OOOIg
Range of Error
0
0
0
0
- 20.5%
-14%
- 1 .6%
-0.0023g
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 47
-------
7.6 1992 QA Results
Two QA steps were required by the EMAP-VP 1992 QA Project Plan: in-house QC checks (i.e., resorts,
recounts, and ID confirmation) on 10% ofeach technician's work, and independent verification of species identification.
The recounts (multiple types - see Table 7-3) and preliminary species verification were performed by the laboratory
performing the analyses. Most of these met the requirements established in the QA Plan. Both of the laboratories
performing these analyzes were evaluated by independent laboratories in 1990 or 1991; therefore, the use of such
an independent evaluation in 1992 was deemed unnecessary.
Table 7-3. Results of recounts performed by the laboratory processing benthic infauna samples in 1992.
Approximately 10% of all samples were processed in duplicate.
Measurement
Benthic sorting
Species identification and enumeration
Biomass
Weighing blanks for biomass
Mean Error
1 .7%
1 .8%
1 .2%
7 x 1 0-5 g
Range of Error
0
0
0
0
- 1 8%
-12%
- 1 .4%
-7x 10-4
g
7.7 1993 QA Results
Two QA steps were required by the EMAP-VP 1993 QA Project Plan: in-house QC checks (i.e., resorts,
recounts, and ID confirmation) on 10% ofeach technician's work, and independent verification of species identification.
The recounts (multiple types - see Table 7-4) and preliminary species verification were performed by the laboratory
performing the analyses. Most of these met the requirements established in the QA Plan. Both of the laboratories
performing these analyzes were evaluated by independent laboratories in 1990 or 1991; therefore, the use of such
an independent evaluation in 1993 was deemed unnecessary.
Table 7-4. Results of recounts performed by the laboratory processing benthic infauna samples in 1993.
Approximately 10% of all samples were processed in duplicate.
Measurement
Benthic sorting
Species identification and enumeration
Biomass
Weighing blanks for biomass
Mean Error
2.9%
0.75%
0.07%
1.1 x 10'4g
Range of Error
0
0
0
0
- 8.9%
- 6.7%
- 0.8%
-9x ID'4
g
Page 48 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Section 8
QA Results for Fish Community Structure and Pathology
8.1 Background
At each Base Sampling Site crews conducted a standard trawl (10 ± 2 minutes at 2-3 knots speed over bottom)
to collect fish for community structure analysis. Fish were identified, measured, counted, examined for the presence
of selected external pathologies, and selected individuals of a set of 10 target species processed for chemical residue
analysis. As part of EMAP-VP's QA program, in 1990 and 1991 the first individual of every species collected
by each crew was preserved in formalin and sent into the laboratory for verification by an expert taxonomist.
In 1992 and 1993 crews were instructed to save the first two individuals of each species collected. Fisheries experts
within EMAP and the National Marine Fisheries Service (NMFS) were employed in 1990 to 1993, and on two
occasions when identification was difficult, specimens were sent to outside experts. Preserved fish were archived
for use during training in subsequent years..
To verify each crew's ability to properly identify pathologies, fish identified as having an external pathology
by the field crews were shipped to ERL-Gulf Breeze (1990) or ERL-Narragansett (1991, 1992, and 1993) for
verification by the laboratory's pathologist. It is important to note that this verification in 1990 to 1992 was "blind"
(i.e., the pathologist did not know which fish the field crews believed to have a pathology). This provided an
estimate of the percentage of "false positives". In addition, in order to develop an estimate of the rate of "false
negatives" (i.e., number of pathologies missed, therefore never sent in for verification), crews collected andshipped
up to 25 individuals of each target species and 10 from any other species (which they determined to be free from
external pathologies) caught at Indicator Testing and Evaluation stations. These steps were necessary because
in 1990 through 1992 fish were also collected for chemical residue analysis, which took priority over pathology
QA. (Note: only fish collected in 1991 were actually analyzed for residues). Because of this, a fish observed
by the crew to have a pathology may have been sent in for chemical analysis rather than pathology verification.
Therefore, the assessment produced by EMAP on the prevalence of gross external pathologies in fish is based
on field observations, not laboratory observations. An error rate is then associated with these data based on the
results of the QA review. Because of poor agreement between field and laboratory examinations, this protocol
changed in 1993 (described in Section 8.7).
Following a review of the 1990 and 1991 pathology QA data, and in consultation with experts from NMFS,
EMAP-VP elected to condense field observations for fish pathologies to four basic categories: lumps, growths,
ulcerations, and fin erosion. It was hoped that by making the examination more simple the success rate (i.e., proper
identification) would increase.
8.2 Audits
No laboratory audits were conducted for these indicators. Field performance reviews and audits were conducted
as described in Section 2. The QA Coordinator or Field Coordinator visited each crew both during trial runs and
the field season. One activity observed by thereviewer was the measurementprocess, with the reviewer remeasuring
selected fish. The reviewer also observed and checked the examination for pathologies conducted by the crew.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 49
-------
8.3 Data Qualifier Codes for Fish Community Structure and Pathology Data
No QA codes currently exist for fish community structure data, indicating all data meet QA criteria and are
suitable for EMAP assessment purposes. Codes do exist for pathology data. These codes pertain to whether the
pathology was verified by an expert pathologist, and are listed in Table 8-1. These codes were used to provide
estimates of the percentage of false positives (crew identified a pathology which was not verified in the laboratory)
and false negatives (pathologist identified a pathology on a "reference" fish which, by definition, the crew believed
to be pathology-free). These codes only pertain to the four "EMAP" pathologies crews focused on beginning
in 1992: lumps, growths, ulcers and fin erosion. These are a subset of the list of pathologies targeted in 1990
and 1991; therefore, the codes were applied to 1990/1991 data as well.
8.4 1990 QA Results
To verify each crew's ability to correctly identify fish species for the community structure indicator, the first
individual of each species collected by each crew was shipped to ERL-N or Versar for verification by an expert
taxonomist.
Three types of errors were detected: misspelled or incomplete species names (in the database), misidentifications,
and fish that could not be identified in the field. Errors falling into the first category were easily detected, corrected
in the database, and documented. An example of this type of error can be found looking at the "Atlantic tomcod".
Records were received from the field for "Atlantic tomcod", "tomcod", and "torn cod" (two words). Each was
listed by the computer as separate species.
The second type of error was mis-identifications. Of the 136 fish sent in for taxonomic verification, nine
were misidentified, representing seven species. In all cases the crew identified a closely-related species, such
as longspine porgy instead of scup, brown bullhead catfish instead of the yellow bullhead, and lizardfish instead
of inshore lizardfish. An additional 16 individuals (12 species) were sent in as unknowns or partial unknowns
(e.g., herring uncl).
All errors were corrected in the database. If a QA fish was misidentified by the crew, all other fish in the
same size class of that species from the same trawl were changed to the correct ID.
Results of laboratory pathology examinations reveal that the crews were generally conservative, classifying
"borderline" conditions as pathologies so the fish would be examined by an expert rather than being discarded.
Table 8-2 presents results of the laboratory review for the four final pathology categories EMAP-VP selected
for continued use.
Page 50 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 8-1. Qualifier codes for fish pathology data
Code Description
FP-A LUMPS NOT Observed in Field / NOT Observed by Quality Assurance Laboratory
FP-B LUMPS NOT Observed in Field but was Observed by Quality Assurance Laboratory
FP-C LUMPS Observed in Field but NOT Observed by Quality Assurance Laboratory
FP-D LUMPS Observed in Field and Confirmed by Quality Assurance Laboratory
FP-E LUMPS NOT Observed in Field but NOT Looked for by Quality Assurance Lab.
FP-F LUMPS Observed in Field but NOT Looked for by Quality Assurance Laboratory
FP-G GROWTHS NOT Observed in Field / NOT Observed by Quality Assurance Laboratory
FP-H GROWTHS NOT Observed in Field but was Observed by Quality Assurance Laboratory
FP-I GROWTHS Observed in Field but NOT Observed by Quality Assurance Laboratory
FP-J GROWTHS Observed in Field and Confirmed by Quality Assurance Laboratory
FP-K GROWTHS NOT Observed in Field but NOT Looked for by Quality Assurance Lab.
FP-L GROWTHS Observed in Field but NOT Looked for by Quality Assurance Laboratory
FP-M ULCERS NOT Observed in Field / NOT Observed by Quality Assurance Laboratory
FP-N ULCERS NOT Observed in Field but was Observed by Quality Assurance Laboratory
FP-O ULCERS Observed in Field but NOT Observed by Quality Assurance Laboratory
FP-P ULCERS Observed in Field and Confirmed by Quality Assurance Laboratory
FP-Q ULCERS NOT Observed in Field but NOT Looked for by Quality Assurance Lab.
FP-R ULCERS Observed in Field but NOT Looked for by Quality Assurance Laboratory
FP-S FINROT NOT Observed in Field / NOT Observed by Quality Assurance Laboratory
FP-T FINROT NOT Observed in Field but was Observed by Quality Assurance Laboratory
FP-U FINROT Observed in Field but NOT Observed by Quality Assurance Laboratory
FP-V FINROT Observed in Field and Confirmed by Quality Assurance Laboratory
FP-W FINROT NOT Observed in Field but NOT Looked for by Quality Assurance Lab.
FP-X FINROT Observed in Field but NOT Looked for by Quality Assurance Laboratory
FP-Y Fish Not Examined for Gross External Pathology in the Field
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 51
-------
Table 8-2. 1990 Pathology QA results based on laboratory examination offish crews believed to have a
pathology and reference, "pathology-free" fish (n=769).
Pathology Type False Positives1 False Negatives2
Body Ulcerations 9/20 (45.0%) 8/749 (1.1%)
Body Lumps/Growths 3/12(25.0%) 26/757(3.4%)
Fin Erosion 8/17(47.1%) 16/752(2.1%)
1 False Positives: The denominator in this column is the total number of fish identified by the field crews
as having a given pathology. The numerator is the number of these fish for which the pathology was
not confirmed by the pathologist.
2 False Negatives: The denominator in this column is the total number of fish identified by the field
crews as not having a given pathology. The numerator is the number of these fish for which the
pathology was observed by the pathologist.
8.5 1991 QA Results
To verify each crew's ability to correctly identify fish species for the community structure indicator, the first
individual of each species collected by each crew was shipped to ERL-N or Versar for verification by an expert
taxonomist. Threetypes of errors weredetected: misspelled orincompletespeciesnames(inthedatabase),misidentifications,
and fish that could not be identified in the field. Errors falling into the first category were easily detected, corrected
in the database, and documented. An example of this type of error can be found looking at the "Atlantic tomcod".
Records were received from the field for "Atlantic tomcod", "tomcod", and "torn cod" (two words). Each was
listed by the computer as separate species.
Of the 187 fish sent in for taxonomic verification, 14 were misidentified, representing nine species. In all
cases the crew identified a closely-related species, such as longspine porgy instead of scup, brown bullhead catfish
instead of the yellow bullhead, and lizardfish instead of inshore lizardfish. An additional 14 individuals (five
species) were sent in as unknowns or partial unknowns (e.g., herring uncl.).
The total of 28 incomplete identifications or misidentifications represent 51 fish records in thedatabase (including
other fish of the same species caught in the same trawl). A total of 7,134 fish were collected in standard trawls
during the 1991 field season representing 69 species. The percentage of errors detected was therefore less than
one percent. All errors were corrected in the database. If a QA fish was misidentified by the crew, all other fish
in the same size class of that species from the same trawl were changed to the correct ID.
Results of laboratory pathology examinations reveal that the crews were generally conservative, classifying
"borderline" conditions as pathologies so the fish would be examined by an expert rather than being discarded.
Of the six fish sent in for verification of a pathology (four additional fishwere not shipped), only three were verified
by the pathologist. Of the "reference" fish shipped, the pathologist determined that none had a pathology. Fin
erosion was not included in these statistics as damage was incurred due to the method of shipping fish (packaged
in mesh onion bags) prohibiting accurate examinations by the laboratory staff. These results are for all types
of pathologies. Table 8-3 presents results of the laboratory review for only the four final pathology categories
EMAP-VP selected for continued use.
Page 52 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 8-3. 1991 Pathology QA results based on laboratory examination offish crews believed to have a
pathology and reference, "pathology-free" fish (n=195).
Pathology Type False Positives1 False Negatives2
Body Ulcerations 2/5 (40.0%) 0/190 (0.0%)
Body Lumps/Growths 1/1(100.0%) 0/194(0.0%)
Fin Erosion not available not available
1 False Positives: The denominator in this column is the total number of fish identified by the field crews
as having a given pathology. The numerator is the number of these fish for which the pathology was
not confirmed by the pathologist.
2 False Negatives: The denominator in this column is the total number of fish identified by the field
crews as not having a given pathology. The numerator is the number of these fish for which the
pathology was observed by the pathologist.
8.6 1992 QA Results
To verify each crew's ability to correctly identify fish species for the community structure indicator, the first
two individuals of each species collected by eachcrew was shipped to ERL-N for verification by an expert taxonomist.
Three types of errors were detected: misspelled or incomplete species names (in the database), misidentifications,
and fish that could not be identified in the field. Errors falling into the first category were easily detected, corrected
in the database, and documented.
Of the 397 fish sent in for taxonomic verification, 36 were misidentified. In all cases the crew identified a
closely-related species, such as longspine porgy instead of scup, or brown bullhead catfish instead of the yellow
bullhead. An additional eight individuals were sent in as unknowns or partial unknowns (e.g., herring uncl.).
Most mis-identified or partially identified individuals were juveniles.
The total of 44 incomplete identifications or misidentifications represent 116 fish records in the database
(including other fish of the same species caught in the same trawl). A total of 14,704 fish were collected in all
trawls (both standard and non-standard) from all station types during the 1992 field season representing 78 species.
The percentage of errors detected was therefore less than one percent. All errors were corrected in the database.
If a QA fish was misidentified by the crew, all other fish in the same size class of that species from the same
trawl were changed to the correct ID.
Results of the pathologist's review offish collected by field crews in 1992 are illustrated in Table 8-4. Crews
appeared to be overly conservative, classifying fish as having a pathology when, in fact, they did not in almost
all cases. It is also possible that by requiring the pathologist to blindly examine hundreds offish, some of the
few with a pathology might be missed. Onion bags were no longer used for containing fish starting in 1992.
This was the cause of the damage in 1991 which prevented verification of fin erosion; therefore, QA data on fin
erosion are included in Table 8-4.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 53
-------
Table 8-4. 1992 Pathology QA results based on laboratory examination offish crews believed to have
a pathology and reference, "pathology-free" fish (n=427).
Pathology Type False Positives1 False Negatives2
Body Ulcerations 9/9 (100.0%) 1/418 (0.2%)
Body Lumps/Growths 3/4 (75.0%) 0/423 (0.0%)
Fin Erosion 5/5(100.0%) 1/422(0.2%)
1 False Positives: The denominator in this column is the total number of fish identified by the field crews
as having a given pathology. The numerator is the number of these fish for which the pathology was
not confirmed by the pathologist.
2 False Negatives: The denominator in this column is the total number of fish identified by the field
crews as not having a given pathology. The numerator is the number of these fish for which the
pathology was observed by the pathologist.
8.7 1993 QA Results
As a result of the 1990-1992 data, and the fact that chemistry fish were no longer to be collected, the QA
process for pathology data changed after the 1992 field season. Starting in 1993, the results on the prevalence
of pathologies in fish of the Virginian Province are based on the laboratory examination, NOT the field exam.
Crews were instructed to examine all fish and ship every one suspected of having a pathology to the laboratory
for confirmation. In 1993, the examination by the pathologist was no longer "blind". Fish received at the laboratory
were coded as "pathology" or "reference" fish. If the pathologist disagreed with the crew's observation (i.e., he
felt a pathology fish did not have a pathology or a reference fish was found to have one), a second pathologist
was consulted and their collective decision entered into the database. Although data from 1990 through 1992
show the crews to be efficient at not missingmany pathologies (/'. e., low incidence of false negatives), the pathologist's
review of reference fish continued. The results of the laboratory examinations are presented in Table 8-5. The
high rate of "false positives" is likely the result of the crews being overly conservative following instruction to
ship any fish SUSPECTED of having a pathology.
Page 54 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 8-5. 1993 Pathology QA results based on laboratory examination offish crews believed to have
a pathology and reference, "pathology-free" fish (n=620).
Pathology Type False Positives1 False Negatives2
Body Ulcerations 10/12 (83.3%) 1/608 (0.2%)
Body Lumps 5/5 (100.0%) 0/615 (0.0%)
Body Growths 4/11(36.4%) 2/609(0.3%)
Fin Erosion 1/4(25.0%) 0/616(0.0%)
1 False Positives: The denominator in this column is the total number of fish identified by the field crews
as having a given pathology. The numerator is the number of these fish for which the pathology was
not confirmed by the pathologist.
2 False Negatives: The denominator in this column is the total number of fish identified by the field
crews as not having a given pathology. The numerator is the number of these fish for which the
pathology was observed by the pathologist.
8.8 Lessons Learned
In 1990-1992 samples were sent in "blind". In general, samples had been stockpiled until the end of the
field season and then examined. Because only a few fish with pathologies were inter-mixed with hundreds of
reference fish, it is possible that some true pathologies identified by the crews may have been missed in the laboratory
examination. In 1993 fish were no longer sent in to the laboratory as blind samples. Each fish was identified
to the analyst as being a fish which the crew believed to have a pathology, or as a pathology-free fish.
The incidence of gross external pathologies reported for 1990 to 1992 is based solely on field operations,
with the rate qualified by the reported rates of false positives and false negatives (see Sections8.4 to 8.6). Because
the potential existed for a fish with a pathology to be saved for chemical residue analysis instead of laboratory
verification of the pathology, the incidence rate could not be reported based on the laboratory results. Therefore,
the incidence rates are based on the QA codes beginning with "observed in the field" regardless of whether or
not the pathology was observed in the laboratory. The incidence reported for 1993 was based on the laboratory
results and need not be qualified.
It should be noted that the incidence rates for all four years are similar when using uncorrected results (i.e.,
1990 to 1992 results are not adjusted based on the rates of false positives or negatives).
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 55
-------
Section 9
QA Results for Water Quality Measurements
9.1 Background
During the four years of EMAP monitoring in the Virginian Province two different approaches have been
utilized to characterize water quality: 1) point-in-time water column profiling using aCTD(Conductivity, Temperature,
Depth logger), and 2) continuous, long-term near-bottom measurements using a moored datalogger. The Seabird
SEE 25 Sealogger CTD has been used to obtain vertical profiles of temperature, salinity, dissolved oxygen (DO),
pH, light transmission, fluorescence and photosynthetically active radiation (PAR). The Hydrolab DataSondeS
datalogger has been used to record long-term time series of temperature, salinity, dissolved oxygen and pH in
the near-bottom waters. During the 1990 Demonstration Project, CTD casts were conducted at all station classes
once per sampling interval, while DataSondeS instruments were repeatedly deployed for approximately 10-day
durations at 23 long-term dissolved oxygen(LTDO) stations throughout sampling intervals 1 and 2. The Seabird
CTDs have continued to be used in each subsequent year of sampling in the Virginian Province. In 1991, the
deployment interval for the DataSondeS's was shortened to one to three days, and units were deployed at all stations.
Prior to the 1992 season, the decision was made to cease deploying these instruments.
The Field Methods Manuals prepared for each year of Virginian Province sampling provide detailed descriptions
of the field protocols for use of the various water quality instruments. The following sections describe the QA/QC
protocols that have evolved and the subsequent QA results achieved for each year of Virginian Province water
quality monitoring over the period 1990-1993.
9.2 1990 Calibration and Calibration Check Procedures
Seabird CTD
In 1990, the Seabird CTDs were calibrated prior to sampling and throughout the field season asneeded (Table
9-1). The dissolved oxygen calibrations were checked against Winkler titrations and/or saturation table values
and the pH calibrations checked against standard pH buffer solutions. Field QC checks of the CTD temperature,
conductivity (salinity), dissolved oxygen, and pH sensors were conducted daily. For these checks, real-time CTD
readings from just below the surface were compared to sample measurements taken with a mercury thermometer,
refractometer, and Winkler titrations from a water sample collected with a Go-Flo water sampling bottle. It should
be noted that the use of a refractometer for verifying the CTD's salinity sensor simply serves as a crude check
to determine if the sensor has suffered an electronic problem resulting in gross errors. The pH readings were
checked using a pH 7 standard buffer solution. If any of the parameters did not fall within the acceptable QC
limits (Table 9-2), the instrument was checked and, if necessary, recalibrated.
For deployment of the CTD, the instrument was first turned on while on deck then lowered to just below the
surface and allowed to equilibrate for two minutes. The unit was then lowered slowly to one meter above the
bottom and again allowed to equilibrate for two minutes. The vessels were not equipped with a meter wheel;
therefore, at times too much cable was paid out and the CTD came in contact with the seafloor. When this occurred,
the CTD was immediately brought up to one meter off the bottom. After the two-minute bottom soak, the unit
was hauled back on deck. Electronic files containing the CTD cast data were usually downloaded to the on-board
computer while the vessel was anchored on station. The field crew quickly reviewed the temperature, salinity
and dissolved oxygen data. On a number of occasions problems with the on-board computer prevented the field
crews from downloading and reviewing the CTD cast data.
Page 56 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 9-1. Summary of calibration procedures used for Virginian Province water quality instruments in
1990.
Instrument
Sensor
Calibration Procedure
Seabird SEE
25CTD
Temperature
Conductivity
DO
pH
LightTrans.
Fluorescence
PAR
Pressure
Calibrated by manufacturer prior to sampling season
Calibrated by manufacturer prior to sampling season
Two point (zero & air-saturated water)
Three point (pH 4, 7 & 10 std. solns.)
Calibrated by manufacturer prior to sampling season
Calibrated by manufacturer prior to sampling season
Calibrated by manufacturer prior to sampling season
Calibrated by manufacturer prior to sampling season
Hydrolab
DataSonde 3
Temperature
Salinity
DO
pH
Depth
Calibrated by manufacturer prior to sampling season
0.5 M potassium chloride (KCI) solution
Water-saturated air
Two point (pH 7 & 10 std. solns.)
Zeroed at water's surface (sealevel)
Table 9-2. Field calibration checks performed during the 1990 Virginian Province Demonstration
Project.
Instrument
Seabird
SBE
25CTD
Hydrolab
Datasonde
3
Frequency of
check
Daily
Pre- and post-
deployment
(each use)
Parameter
Temperature
Salinity
DO
PH
Temperature
Conductivity
DO
PH
Checked against
Thermometer
Refractometer
Winkler titration
pH 7 std. solution
Thermometer
0.5M KCI std.
Water-saturated air
pH 7 std. solution
Maximum acceptable
difference
±2 °C
±2ppt
± 1 .0 mg/L
± 0.5 pH units
±2 °C
± 5 mS/cm
± 12.5% saturation
± 0.5 units
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
Page 57
-------
Hydrolab DataSonde3
TheDataSondeSunitswerecalibratedpriortoeachdeploymentusingthemanufacturer'srecommended procedures
(Table 9-1). QC checks were conducted at the dock on the morning that the instruments were to be deployed
or onboard the vessel just prior to deployment. The QCchecks procedures were similar to the calibrations: dissolved
oxygen percent saturation was compared to expected readings of 102.5% in water-saturated air (102.5% is used
insteadoflOO%samrationbecauseHydrolab'sLo-Flomembranewasinstalledontheinstruments),specificconductivity
was compared to a standard reading using aO.5 M KC1 solution, pH was compared to pH 7 standard buffer solution
and temperature was compared to thermometer readings. If any of the parameters did not fall within acceptable
QC limits (Table 9-2), the crews re-calibrated the sensor prior to deployment.
Individual units, housed inside a protective PVC casing were moored approximately one meter above the
bottom. They were programmed to record data internally at 3 0-minute intervals throughout their ten-day deployments.
Upon retrieval, the units were examined for evidence of biological fouling of the probes. They underwent a post-deployment
QC check that was identical to the pre-deployment calibration check. The data files were downloaded either on
board the vessel or in the mobile laboratory. The raw data files were quickly reviewed, paying particular attention
to the dissolved oxygen values. If the dissolved oxygen dropped to zero atany time during the record, no replacement
unit was deployed at that station for fear of "poisoning" the DO probe.
9.3 Data Qualifier Codes for Water Quality
Because of the number of parameters monitored and the overall complexity of the water quality datasets,
a rather large number of data qualifier codes, or "flags", are incorporated into these datasets. These codes are
listed in Tables 9-3 and 9-4. In order for the codes to make sense, it is important to understand the data manipulations
employed in the analysis stage.
CTD Qualifier Codes
The first step in assessing the quality of CTD profiles was for the crew chief to examine the profile on the
on-board field computer as soon as the data were collected. As described in Section 9.4, problems were encountered
in 1990 with this procedure. Upon receipt of the electronic file in the Information Management Center, the first
step is verification. An analyst examined each cast to determine if it was associated with the correct station.
The CTD depth was compared to the depth recorded from the boat's fathometer, and individual measurements
were compared to those expected (e.g., low salinity would not be expected from a station in Long Island Sound)
or measured via other mechanisms (e.g., from the Hydrolab or ambient checks).
The next step in the data assessment process is validation. Each CTD file consists of an entire profile of
measurements made through the water column from the surface to the bottom and back up to the surface. For
ease of analysis, each CTD file was split into separate components which were stored as individual files. Upon
submersion, the CTD was allowed to sit approximately one meter below the surface for several minutes to allow
it to come to thermal equilibrium after being on the hot deck. The section of the profile from the point of immersion
until the unit is lowered through the water column is the "surface soak". Data from this file were not used other
than to ensure that the crew allowed sufficient time for equilibration. The section of the profile starting when
the unit is lowered and ending when it reaches the bottom (actually one meter off the bottom) is the "downcast".
The unit was then allowed to sit at the bottom and record data for several minutes. This is the "bottom" file.
Finally, the point from the start of the unit's ascent until it reaches the surface is the "upcast".
Page 58 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 9-3. Data qualifier codes attached to 1990 -1993 CTD water quality data. Only C-A through C-
H were applied to the 1990 data.
CODE DESCRIPTION
C-A Reject entire CTD cast (all parameters).
C-B Accept entire CTD cast (all parameters).
C-C Bottom file acceptable; downcast file rejected; no surface values; reported bottom values are the
average of all bottom records.
C-D Downcast file acceptable; bottom file rejected; first and last records of downcast file used for
reported surface and bottom values, respectively.
C-E Downcast and bottom files rejected; however, first and last records of downcast file appeared
reasonable and were used for surface and bottom values, respectively.
C-F Downcast file acceptable; bottom file rejected; reported surface values are the first record of the
downcast file; bottom values are the first record of the bottom file (appeared acceptable).
C-G Downcast and bottom files rejected; reported surface values are the first record of the downcast
file (appeared reasonable). No bottom values reported.
C-H Downcast and bottom files rejected; bottom values are the last record of downcast file (appeared
reasonable). No surface values reported.
C-IA Reject surface values (all parameters)
C-IB Reject pre-deploy. soak, accept post-deployment soak (all parameters)
C-IC Reject entire bottom soak, no bottom values available (all parameters)
C-ID Reject entire downcast file (all parameters)
C-IE Reject bottom soak, use last value of downcast (all parameters)
C-IF Reject average of bottom soak but accept last value (all parameters)
C-IG Shallow station with pre-deployment soak and bottom soak only (no profile)
C-IH Shallow station: surface and bottom values equal. Bottom file used for both.
C-ll Depth values questionable
C-IJ Reject surface dissolved oxygen (pre and post)
(continued)
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 59
-------
Table 9-3. continued.
CODE DESCRIPTION
C-IK Reject pre cast dissolved oxygen but accept post cast dissolved oxygen
C-IL Reject downcast dissolved oxygen
C-IM Reject bottom dissolved oxygen
C-IN Reject bottom soak dissolved oxygen but use last value of downcast
C-IO Reject average bottom dissolved oxygen but use last value of bottom file
C-IP Reject surface salinity (pre and post)
C-IQ Reject pre cast salinity but accept post cast salinity
C-IR Reject downcast salinity
C-IS Reject bottom salinity
C-IT Reject bottom soak salinity but use last value of downcast
C-IU Reject average bottom salinity but use last record of bottom file
C-IV Reject surface temperature (pre and post cast)
C-IW Reject pre cast temperature but accept post cast temperature
C-IX Reject downcast temperature
C-IY Reject bottom temperature
C-IZ Reject bottom soak temperature but use last value of downcast
C-JA Reject average bottom temperature but use last value of bottom file
C-JB Reject surface pH (pre and post)
C-JC Reject pre cast pH but accept post cast pH
C-JD Reject downcast pH
C-JE Reject bottom pH
C-JF Reject bottom soak pH but use last value of downcast file
(continued)
Page 60 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 9-3. continued.
CODE DESCRIPTION
C-JG Reject average bottom pH but use last value of bottom file
C-JH Reject surface PAR (pre and post soak)
C-JI Reject pre cast PAR but accept post cast PAR
C-JJ Reject downcast PAR
C-JK Reject bottom PAR
C-JL Reject bottom soak PAR but use last value of downcast
C-JM Reject average bottom PAR but use last value of bottom file
C-JN Reject surface transmissometry (pre and post)
C-JO Reject pre cast transmissometry but accept post cast transmissometry
C-JP Reject downcast transmissometry
C-JQ Reject bottom transmissometry
C-JR Reject bottom soak transmissometry but use last value of downcast
C-JS Reject average bottom transmissometry but use last value of bottom file
C-JT Reject surface fluorescence (pre and post)
C-JU Reject pre cast fluorescence but accept post cast fluorescence
C-JV Reject downcast fluorescence
C-JW Reject bottom fluorescence
C-JX Reject bottom soak fluorescence but use last value of downcast
C-JY Reject average bottom fluorescence but use last value of bottom file
C-JZ Fluorescence off-scale
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 61
-------
Each component of the profile (i.e., each file) was examined to determine if it was reasonable. In 1990 only
the dissolved oxygen (DO) records were examined. Files were classified as acceptable or not solely based on
the DO record. In subsequent years each parameter was assessed independently, resulting in a significant increase
in the number of codes. Components of the profile could be classified as unacceptable for a number of reasons.
For example, the downcast could contain unexplained spikes (it should be relatively smooth), or the shape of
the bottom soak might suggest the unit impacted the bottom and mud was sucked up into the pumping system,
clogging it. Also, the upcast may not match the downcast.
As part of EMAP's data assessment activities, "surface" and "bottom" values are reported for key parameters
such as DO and salinity. These values were extracted from the profile. In general, the surface value is the first
record of the downcast, and the bottom value is the average of all values in the bottom file. As shown in the CTD
qualifier codes listed in Table 9-3, when a section of the profile is determined to be unacceptable, other alternatives
are employed. If the bottom soak is determined to be unacceptable, the last record of the downcast is generally
used as the reported bottom value. If the appearance of the bottom section of the profile suggests mud clogged
the pump, and this clog cleared itself during the bottom soak; or if there was a significant near-bottom oxycline
resulting in a change in DO between the start and end of the bottom soak, the last value of the bottom file is used
in place of the average. Similarly, the downcast file might be deemed unacceptable because of severe spiking
in the middle of the downcast, suggesting a temporary clog or an intermittent electronic problem. However, both
the beginning and end of this file may appear reasonable. In such cases the file may be classified as unacceptable
but the first record is still used for the surface value, and, if the bottom file is unacceptable, the last record may
be used for the bottom value.
These codes generally are not of interest to users requesting the summary data, /'. e., reported surface and bottom
values. The flags simply point to documentation on how those values were determined. The codes are likely
of greater importance to users requesting the actual profiles as they point to potential problems with those data.
Hydrolab Qualifier Codes
Codes describing Hydrolab files are listed in Table 9-4. As described in Section 9.2, QC checks of individual
units were performed both before deployment and following retrieval. The qualifier codes indicate the results
of those checks. For example the code "H-K" means that the DO at the end of the file may underestimate the
actual DO concentration. This would indicate that the Hydrolab unit failed QC upon retrieval, likely due to fouling
of the DO sensor. Fouling is a gradual process, making it difficult to determine at what point during thedeployment
the readings become unreliable.
The code "H-H" requires explanation. Prior to deployment each unit is set to log for a certain period of time
at a selected interval. In 1990 when the units were deployed for 10-day periods at selected stations, the units
were set to log at 30-minute intervals. In 1991 when they were deployed at every station for one to three days
they were set to log at 15-minute intervals. "Autolog" is a back-up which automatically logs data every hour
regardless of how the logging run is set up. In the event that the crew accidentally set the unit incorrectly (e.g.,
set it to start logging on 4/5/96 instead of 4/5/91), Autolog would still log data hourly. As a result, data are collected
for the duration of the deployment; however, the logging interval would be different from other files collected
that year.
Page 62 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 9-4. Data qualifier codes attached to Hydrolab water quality data.
CODE DESCRIPTION
H-A No file available.
H-B Acceptable file.
H-C Dissolved oxygen not acceptable.
H-D Salinity not acceptable.
H-E Temperature not acceptable.
H-F pH not acceptable.
H-G Discontinuous record due to power loss.
H-H Autolog file.
H-l Total record less than 24 hours.
H-J DO at start of file may overestimate actual ambient DO concentration.
H-K DO at end of file may underestimate actual ambient DO concentration.
H-L DO at start of file may underestimate actual ambient DO concentration.
H-M DO at end of file may overestimate actual ambient DO concentration.
H-N Data not available for entire deployment.
H-O Depth not acceptable.
H-P Salinity at file end may underestimate actual ambient salinity concentration.
H-Q pH at start of file may underestimate actual ambient pH value.
H-R Battery Power not acceptable.
9.4 1990 QA Results
Seabird CTD
The CTD data were affected by several procedural problems that came to light during and after the sampling
season. First, the QC checks for both dissolvedoxygen and pH did not perform satisfactorily during the Demonstration
Project. The field crews were not prepared to identify unacceptable CTD casts in the field, and as a result, many
of the casts were later flagged for containing unacceptable data. In addition, many CTD data files were lost in
the beginning of the field season as a result of a computer software problem.
Performance of the dissolved oxygenprobe was checked by comparing the CTDsensor's reading to that calculated
using a digital titrator that was part of a HACK Winkler titration field kit. The results of the Winkler titrations
were not as reliable as initially expected: the difference between two replicate dissolved oxygen water samples
exceeded 0.5 mg/L in over 11% of the field QC checks conducted throughout the sampling season (Table 9-5).
This large amount of variability between replicates made it difficult to assess whether the QC checks were reliable
enough to evaluate the performance of the dissolved oxygen sensor. It was unknown whether the 60 CTD QC
checks that exhibited differences between the dissolved oxygen sensor and the Winkler titrations in excess of
1 mg/L were a result of faulty sensors or poor QC check (i.e., Winkler titration) procedures.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 63
-------
The field QC check of the pH sensor involved comparing the CTD sensor's reading to a pH 7 standard buffer
solution. This procedure was implemented to insure that the sensor's calibration did not drift; however, it proved
to be an inappropriate check. Post-sampling-season review of the CTD casts revealed that one of the pH sensors
was malfunctioningformostofthe summer, aproblemthatwasnever detected in the field because upon malfunctioning,
the reading defaulted to a constant value of pH 7. This was not detected because the pH 7 buffer solution was
used for the check, and crews generally did not review the pH data collected with each profile.
A total of 480 CTD casts were conducted during the 1990 Demonstration Project. Data from 9% of these
casts were lost and are notincluded in the database. The remaining 437casts were carefully reviewed for acceptability
based solely on the performance of the dissolved oxygen sensor (Table 9-6). Of the reviewed casts, 68% had
acceptable dissolved oxygen profiles (see Section 9-3) and 23% yielded unacceptable dissolved oxygen profiles,
although individual surface and/or bottom values were accepted in some of the casts where the profile itself was
rejected. The profiles of other parameters were used to assess the validity of the dissolved oxygen data (e.g.,
high fluorescence would be expected in areas of supersaturated dissolved oxygen); however, the acceptability
of the data recorded by these other sensors was not determined.
Table 9-6 also reports QA results specific to Base Sampling Sites sampled during Interval 2. These are the
data that are currently being used in the assessment of the ecological condition of the Virginian Province. Many
stations were visited on more than one occasion; however, only one dissolved oxygen value is reported per station.
The table shows that for 92% of the stations used in this assessment, acceptable bottom DO values were obtained
in 1990.
Table 9-5. Results of CTD dissolved oxygen field QC checks used during the 1990 Demonstration
Project. Dissolved oxygen readings from the CTD sensor were compared to HACH
Winkler titrations.
# of CTD field QC checks
# of Winkler replicates w/ DO differences > 0.5 mg/L
# of CTD/Winkler QC checks w/ DO differences > 1 .0 mg/L1
Range of CTD/Winkler DO differences (mg/L)
174
20 (1 1 %)
60 (34%)
-4. 7 to +4.0
It was unknown whether the 60 CTD QC checks that exhibited differences between the dissolved
oxygen sensor and the Winkler titrations in excess of 1 mg/L were a result of faulty sensors or poor QC
check (i.e., Winkler titration) procedures.
Page 64 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 9-6. Results of 1990 post-sampling season CTD data review. (Percentages are based upon #
of total casts attempted and total number Base Stations sampled in Interval 2).
Acceptability of casts in 1990 was based solely upon performance of the DO sensor
whereas performance of all sensors were considered in subsequent years.
Total Casts Base Stations
# Total casts/Base Stations
# Events w/ lost files
# Casts accepted for all parameters (C-B)
# Casts rejected for all parameters (includes lost casts)
# Casts w/ acceptable surface DO
# Casts w/ acceptable bottom DO
480
43
298 (62%)
142 (30%)
315 (66%)
337 (70%)
111
3
98 (88%)
11 (10%)
99 (89%)
1 02 (92%)
Hydrolab DataSonde3
Evaluation of the Hydrolab QA activities during the Demonstration Project revealed several concerns regarding
field procedures. The calibration of the conductivity cell, for measuring sdinity, produced unsatisfactory results.
In addition, the dissolved oxygen QC check was deemed to be inappropriate. The ten-day deployment period
often resulted in extensive biological fouling of the probes and unacceptable dissolved oxygen records.
Calibration of the specific conductivity parameter resulted in inaccurate salinity measurements. It wasdetermined
that the 0.5 M KC1 solution was not standardized properly and thus did not have the assumed standard reading
of 58.64 mS/cm. The actual conductivity of the standard utilized was determined using an Auto-Salinometer
calibratedtoCopenhagenseawater. Alloftherawdatafileswerethenre-processedusingasub-routine that compensated
for the incorrect calibrations.
The dissolved oxygen QC check using water-saturated air was not appropriate since the sensor membrane
had to be wiped dry during the process. This may have had a large effect on the retrieval QC checks since this
procedure removes biological and physical fouling that may have altered the sensor's performance during the
deployment. A substantial amount of biological fouling appeared on the instrument casing and probes during
the ten-day deployments. This fouling may have been responsible for underestimating dissolved oxygen values
(towards the end of the datafile) in over 60% of the DataSondeS records (Table 9-7).
DataSondeS units were deployed a total of 123 times during the 1990 Demonstration Project. Data from
18% of these deployments are notpart of the database because of lost units, incorrect datalogging setups or missing
datafiles (H-A code). The results of post-retrieval QC checks are summarized in Table 9-7. Of the 104 files
which were reviewed, only 19% were totally acceptable for all parameters throughout the entire record. While
failures of the conductivity, temperature and pH sensors occurred sporadically, this low overall percentage is
due mainly to fouling which resulted in poor performance of the dissolved oxygen sensor during the later half
of the records, as discussed earlier.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 65
-------
Table 9-7. Results of calibration checks following retrieval of Hydrolab Datasonde 3 instruments for
1990 Virginian Province monitoring. Percentages indicate the number of times that
acceptance criteria were met.
Parameter
Temperature
Salinity
DO
PH
Acceptance Criteria
± 1 °C
±2ppt
± 0.5 mg/L
± 0.5 units
Percent Accepted
99% (1 03/1 04)
96% (1 00/1 04)
38% (40/1 04)
97% (101/104)
9.5 1990 Lessons Learned/Changes for 1991
Seabird CTD
The post-season evaluation of CTD casts ledto a restructuring of the quality control criteria for these instruments.
During the 1990 Demonstration Project, the field crews were expected to re-calibrate the CTD sensors in the
field. This proved to be a difficult task, particularly for the dissolved oxygen sensor calibration which requires
the CTD to be placed in a large tank of air-saturated water. The field crews were constantly on the move and
rarely had the opportunity to set up a proper calibration tank. In short, the 1990 experience served to demonstrate
that accurate calibration of the dissolved oxygen sensor requires a controlled environment and experienced personnel.
Procedural changes implemented in 1991 required the field crews to send back to the field operations center any
CTD that failed a calibration QC check. At the field operation centers, trained technicians were on hand to perform
a more complete evaluation and, if necessary, re-calibration of any malfunctioning instruments under controlled
laboratory conditions.
The daily QC checks conducted during the 1990 Demonstration Project helped to identify sensor drift and
the need for re-calibration; however, they could not be used to determine if the instrument performed properly
during a specific CTD cast. Field QC checks were changed to include two components: QC checks on the sensor
calibrations and QC checks on each deployment. Much investigation went into determining the most appropriate
methods for conducting these tests, and the most important findings are summarized below.
Very little sensor calibration drift was observed in the CTDs throughout the 1990 field season; therefore,
it was determined that weekly calibration checks (outlined in Table 9-8) would be sufficient for the 1991 field
season. The criterion for acceptance of DO data was re-evaluated and the acceptable difference during QA checks
was reduced from 1.0 to 0.5 mg/L. The dissolved oxygen and pH QC checks used during the Demonstration Project
yielded unsatisfactory results and had to be modified for the 1991 field season.
The dissolved oxygen sensor needed to be compared to a reliable dissolved oxygen value. Winkler titrations
were the first choice; however, they had produced unacceptable results in 1990. The performance of the HACK
kits were evaluated in the laboratory. These tests identified three faulty titrators that showed excessive variability
in amount of titrant delivered when running replicate samples. Further testing demonstrated that the Hach kits
could accurately measure dissolved oxygen concentrations under the following conditions: use of properlycalibrated
titrator, daily standardization of sodium thiosulfate prior to titrating samples, and properly trained technicians
who were familiar with conducting Winkler titrations.
Page 66
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 9-8. Summary of water quality instrument field calibration checks for 1991-93 Virginian Province
monitoring.
Instrument
Seabird SEE
25CTD
Seabird SEE
25CTD
Hydrolab
Datasonde 3
YSI Model 57
DO meter
Frequency of check
Each station
Once each week (in
concert with YSI
check)
Pre- and post-
deployment (each
use)
Once each week
Parameter
Temperature
Salinity
DO
Temperature
Salinity
DO
PH
Temperature
Salinity
DO
PH
Temperature
DO
Checked against
Thermometer
Refractometer
YSI DO meter
Thermometer
Refractometer
YSI DO meter
pH buffer solution
Thermometer
Refractometer
YSI DO meter
pH buffer solution
Thermometer
Winkler titration
Maximum
acceptable
difference
±2 °C
±2ppt
± 0.5 mg/L
±2 °C
±2ppt
± 0.5 mg/L
± 0.5 units
± 1 °C
±2ppt
± 0.5 mg/L
± 0.5 units
± 1 °C
± 0.5 mg/L
This last point, that Winkler titrations require experienced personnel in order to produce accurate results,
was a great concern. The field crews are only exposed to a minimal amount of training in many different topics
prior to sampling and once in the field, have many demands placed upon them. It was unrealistic to depend upon
all of the crew members having the needed experience and available time to conduct daily titrations. Assorted
instruments were evaluated and it was determined that the hand-held dissolved oxygen meter manufactured by
Yellow Springs Instruments (YSI) provided reliable DO measurements that could be used in the QC checks.
However, weekly titrations were still performed, but only by selected individuals who were provided with additional
training. Also, beginning in 1991, field crews utilized a potassium iodide/iodate solution to determine the true
normality of the thiosulfate solution prior to the weekly QC check of the YSI instrument.
A post-season check of all CTDs revealed that a faulty pH sensor went undetected throughout much of the
field season. Consultation with Seabird electronics identified the fact that broken pH sensors will default to a
reading of 7; therefore, the field QC check didnotdetectthe broken sensor. Since the crews only reviewed temperature,
salinity and dissolved oxygen data in the field, they did not identify the faulty sensor during their normal sampling
routine. The field QC check was modified for the 1991 season to require comparing the pH sensor's reading to
a standard pH 10 solution. Additionally, the field computer system was modified to include vertical profiles of
all parameters, including pH, to be reviewed when the data were downloaded.
Review of the CTD casts obtained during the Demonstration Project revealed several deployment problems
that affected the performance of the dissolved oxygen sensor. The most commonly encountered problems were:
1) air bubbles trapped in the dissolved oxygen plumbing loop, 2) mud being sucked through the conductivity
cell and into the plumbing loop upon instrument contact with the bottom, and 3) insufficient thermal equilibration
time of the dissolved oxygen sensor. Research scientists at Seabird Electronics Inc. were extremely helpful in
assessing the CTD datafiles from field and tank tests and in identifying these deployment problems (Report on
Dissolved Oxygen Data by Nordeen Larson, March 1991).
CTD deployment procedures were modified for the 1991 field season in hopes of minimizing these problems
(Strobel and Schimmel 1991). The instruments were not turned on until just prior to entering the water, to allow
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
Page 67
-------
all the air to purge from the plumbing loop during the two-minute pumpdelay. In order to allow the units to thermally
equilibrate, the CTDs soaked in the surface waters for aminimum of three minutes prior to being lowered through
the water column. The crews were instructed to keep the instruments from coming in contact with the seafloor.
This was accomplished through a buoy/counter weight system or a well marked pay-out cable. The CTDs remained
at depth (ca. one meter off the bottom) for at least two minutes. The units were hauled back to just below the
surface, held there for a one minute surface soak, then brought back on board.
Additional emphasis was placed upon deployment QC checks of CTD casts. The hand-held YSI meter was
used to measure dissolved oxygen concentration in water collected in a Go-Flo bottle from approximately one
meter off the bottom at each station. This measurement was taken at approximately the same time as the CTD
cast and provided a check on the operation of the CTD dissolved oxygen sensor during deployment. It also provided
redundant data in case the data were lost or deemed unacceptable during the post-season review.
The CTD component of the field computer system was modified so the crews could view vertical profiles
of each parameter along with the raw data file. Each CTD cast data file was reviewed in the field for evidence
of deployment problems. A standard check on the data file was comparison of the downcast versus the upcast
for all parameters, with particular attention to dissolved oxygen, salinity and light transmission. The dissolved
oxygen profile was further evaluated by comparing the surface dissolved oxygen values at the beginning and end
of the cast, and by comparing the bottom dissolved oxygen value to that recorded by the hand-held YSI meter.
If either of these dissolved oxygen differences exceededO.5 mg/L, the field crews recalibrated the YSI and redeployed
the CTD to obtain a second profile.
It was suggested that, as part of the pre-season calibration, all units be tested side-by-side in a controlled
tank test. The results of that test are shown in Table 9-9.
Hydrolab DataSonde3
Hydrolab DataSondeS deployments conducted during the 1990 Demonstration Project along with post-sampling
season tank tests revealed the need for modification of certain calibration, QC check and deployment procedures.
DataSonde 3 evaluation tests resulted in a new salinity calibration procedure for the 1991 field season. It
was found that salinity should be calibrated using a seawater standard rather than calibrating specific conductivity
which is converted to salinity units. Tanks tests showed that it was better to calibrate salinity using a 3 0 ppt standard
and deploy the instrument in nearly freshwater than to calibrate salinity with a 15 ppt standard and deploy the
unit in a high salinity environment. In 1991, the conductivity cell was calibrated using a secondary seawater standard,
the salinity of which was determined using a Guidline laboratory salinometer calibrated with Copenhagen seawater.
It was decided to use a single standard throughout the entire Province rather than using assorted calibrationstandards
for deployment in different salinity waters; therefore, a secondary seawater standard of approximately 30 ppt
was used throughout the field season. The salinity of the standard was measured with the laboratory salinometer
prior to being sent out in the field, throughout the summer and at the end of the sampling season. In all cases,
the salinity drifted by less than 0.1 ppt over the three-month period.
Calibration and retrieval QC check procedures were modified to include immersing the DataSonde3 unit in
a bucket of local seawater or freshwater, and comparing its temperature, salinity and dissolved oxygen readings
to those recorded by a thermometer, refractometer and YSI dissolved oxygen meter, respectively. This appeared
to be a better field check, because it eliminated the problem of wiping the membrane dry and possibly removing
some of the biological fouling that may have affected the dissolved oxygen probe's performance.
Page 68 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Table 9-9. Summary of test in which all water quality instruments were placed in a well-mixed tank of
seawater prior to the 1991 field season. Values are means ± 95% confidence limits for the
simultaneously-recorded readings from "n" number of instruments of each type. Testing
was conducted over several hours, during which time dissolved oxygen was varied to give
high, medium and low concentrations; the other parameters did not vary significantly.
Hydrolab
Datasonde 3
(n = 34a)
Seabird CTD
(n = 4)
YSI meter
(n = 3)
Winkler
titration
(n = 16)
Dissolved
Oxygen
High
(mg/L)
7. 3 ±0.4
7.1 ± 0.2b
7.1 ±0.1
7.4 ±0.4
Medium
(mg/L)
4.4 ± 0.5
4.4 ±0.7
4.2 ±0.1
4.4 ±0.2
Low
(mg/L)
2.1 ±0.5
2.1 ±0.4
2.0 ± 0.2
2.2 ± 0.2
Salinity
(ppt)
24.6 ± 0.5
24.4 ±0.1
NA
NA
Temperature
°C
21 .9 ± 0.1
21.9±0.1
21.8±0.1
NA
PH
8.6 ± 0.2
8.6 ±0.1
NA
NA
a Dissolved oxygen values for Hydrolab Datasonde 3's are based on n = 31 instruments; DO sensors on
three instruments failed due to improperly installed membranes.
b The value from one instrument was omitted as an outlier in calculating this mean.
The most critical lesson learned regarding the DataSondeS instruments was that ten-day deployments are
not appropriate for the waters encountered throughout the Virginian Province. The general impression was the
dissolved oxygen sensor produced reliable readings for the first five days of deployments, but then underestimated
the dissolved oxygen concentration towards the end of the records. It was impossible to determine what sections
of the records were acceptable and when the sensor became too fouled to produce accurate readings. A major
change for the 1991 sampling season was that DataSondeS units were deployed at all base stations for a single
deployment rather than long-term servicings at a selected group of stations. The deployment period for these
continuous near-bottom records were reduced from ten to three days during the 1991 field season.
The field computer system was modified to standardize the format of the data files being recorded and to
streamline calibration and QC check procedures. In addition, the software included a more detailed data review
routine, including time series plots for all parameters. Unfortunately, there were problems interfacing the Hydrolab
software module with the boat computer system; therefore, all communication with the DataSondeS instruments
had to be done in the mobile laboratory rather than onboard the vessel. Specifics of the Hydrolab software module
are documented and on file with the EMAP-VP data management group.
A series of controlled experiments were conducted to answer questions regarding the performance of the Hydrolab
DataSondeS instruments. These performance evaluations included an experimentthat was conducted during 1991
crew chief training where 34 DataSondeS units, 4 CTD instruments, 4 YSI meters and individual Winkler titrations
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
Page 69
-------
were used to measure the concentration of dissolved oxygen in a 5 00 gallon test tank. Results of this experiment
are summarized in Table 9-9 and highlighted below.
-Hydrolab dissolved oxygen measurements are normally precise to within ±0.5 mg/L of the mean value; however,
they are less reliable when exposed to low dissolved oxygen concentrations. This could be due to a longer
response time of the Lo-Flo membranes to low dissolved oxygen levels.
-Approximately 10% of the instruments deployed experienced some sort of sensor malfunction; this was
mostly due to a faulty calibration (e.g., insufficient stabilization time prior to calibration, air bubbles beneath
Lo-Flo membrane, etc.) rather than a malfunctioning sensor.
These experiments helped to better understand the performance and expectations of the instruments. Many
practical lessons were learned that were passed on to the 1991 field crews during their training sessions.
YSI Dissolved Oxygen Meter
Incorporation of the use of the YSI dissolved oxygen meter required an additional quality control check on
its performance. The YSI meters were calibrated immediately prior to use at each station using the water-saturated
air calibration procedure recommended by the manufacturer. Calibration QC checks were conducted at weekly
intervals in the mobile laboratories. Following calibration, the YSI probe was immersed into a bucket ofair-saturated
water and allowed to stabilize. The dissolved oxygen of the water bath was determined by Winkler titration and
compared to the YSI reading. If the dissolved oxygen difference exceeded 0.5 mg/L (Table 9-8), the instrument
was checked thoroughly and the probe was either recalibrated or replaced. Because the unit was air-calibrated
prior to use at each station, this served as a check on the overall performance of the unit and on the air-calibration
method.
9.6 1991 QA Results
The 1991 sampling season yielded more reliable water column measurements than the 1990 Demonstration
Project. Many lessons were learned during the Demonstration Project that led to improved calibration, field quality
control check and deployment procedures. These changes, along with improved training of the field crews and
more elaborate data review protocols, resulted in a significant increase in acceptable water column profiles and
continuous near-bottom records.
One of the most significant improvements in 1991 was the addition of the YSI dissolved oxygen meter. The
YSI meter was used for comparisons in the CTD and DataSondeS field QC checks of dissolved oxygen. It was
also used to measure bottom dissolved oxygen at all stations, which resulted in three separate bottom dissolved
oxygen measurements (CTD, DataSondeS and YSI) for most stations. These values were compared during the
post-sampling season data reviews and helped to identify acceptable versus unacceptableCTD casts and DataSondeS
records.
Page 70 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Seabird CTD
Procedural problems regarding CTD calibrations, QC checks and deployments were minimal during the 1991
sampling season.
All calibrations were conducted at the Virginian Province instrumentation facility in Narragansett, RI. A
calibration tank with air-saturated freshwater was always on hand to perform dissolved oxygen calibrations.
A downfall of this system is that the faulty CTD had to be shipped to the testing facility and a replacement unit
sent to the field crew. This often created a hiatus in the collection of CTD data for that field crew. On one occasion,
the CTD unit was damaged during shipment, which resulted in further loss of CTD data.
Weekly calibration QC checks were an appropriate method for evaluating the performance of the sensors
and recognizing any calibration drifts. The side-by-side comparisons with the YSI dissolved oxygen meter were
a simple check that produced reliable results (Table 9-10).
Modified deployment procedures and more elaborate QC checks helped to increase the quality of data being
collected. The on-station comparison of the CTD sensor's bottom dissolved oxygen value with the YSI bottom
dissolved oxygen measurements proved to be a valuable sensor performance check. The improved data review
procedures, using the updated CTD software module, allowed the field crews to recognize unacceptable casts
while they were anchored on-station, providing them the opportunity to conduct another cast when needed. These
checks improved the number of acceptable CTD casts (see Table 9-11). In 1991, 80% of the casts had acceptable
bottom dissolved oxygen values, compared to only 70% in 1990. Acceptable bottom DO values from the CTD
were measured at 91% of the stations used in EMAP's assessment of the ecological condition of the Province
(i.e., Base Stations); and, because redundant measurements were taken with the YSI meter, bottom dissolved
oxygen concentration data are available for those stations where the CTD failed to pass QC.
The CTD software on the field computer system was a great improvement overthatused in the 1990 Demonstration
Project. The biggest improvement was the data archiving system, which resulted in cast data being lost from
only 11 events compared to 43 in 1990. The CTD units still experienced intermittent problems of hanging up
which prevented on-station downloading of data to the field computer. When this occurred, the casts could not
be reviewed and the risk of unacceptable data increased, along with field crew frustration levels.
Table 9-10. Results of weekly calibration checks of water quality instruments used in the Virginian
Province, 1991.
Instrument
YSI meter
Seabird
CTD
Parameter
Temperature
DO
Temperature
Salinity
DO
PH
Checked
against
Thermometer
Winklertitration
Thermometer
Refractometer
YSI meter
Standard buffer
Acceptance
Criteria
±2 °C
± 0.5 mg/L
±2 °C
±2ppt
± 0.5 mg/L
± 0.5 units
Percent
Accepted
1 00% (27/27)
89% (24/27)
1 00% (27/27)
1 00% (27/27)
89% (24/27)
1 00% (27/27)
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
Page 71
-------
Table 9-11. Results of 1991 post-sampling season CTD data review. (Percentages are based upon #
of reviewed casts or number of Base Stations). Note: different criteria were used for
accepting and rejecting CTD cast data in 1990 vs. 1991. Acceptability of casts in 1990 was
based solely upon performance of the DO sensor, the performance of all sensors were
considered in 1991.
Total Casts
Base Stations
Total casts/Base Stations
# Casts accepted unqualified for all parameters (C-B code)
# Casts rejected for all parameters (includes lost casts)
# Casts w/ acceptable surface DO
# Casts w/ acceptable bottom DO
291
166(57%)
40(14%)
236 (81%)
233 (80%)
101
80 (79%)
5 (5%)
94 (93%)
92(91%)
Hydrolab DataSonde3
Continuous long-term near-bottom records collected from the Hydrolab DataSondeSunits were greatly improved
in 1991. This was a direct result of shorter deployment periods and improved quality control procedures.
The changes in pre- and post-deployment QC checks resulted in improved field checks. Theunits were immersed
into a bucket of local seawater and real-time readings compared to those from the YSI(DO), refractometer (salinity),
and thermometer (temperature). This provided amore realistic assessment of the DataSondeS dissolved oxygen
probe's performance than the saturated air method employed in 1990 by eliminating the problem of potentially
removing biological fouling that may have affected the dissolved oxygen records.
DataSondeS units were deployed for a single three-day or less period at 113 stations throughout the Province
(includes other than base sampling sites). This decreased deployment period resulted in a significant increase
in acceptable records, particularly dissolved oxygen. In 1991, 87% of the datafiles were accepted in their entirety,
compared to only 19% in 1990. A total of 94% of the 1991 retrieval QC checks for dissolved oxygen met the
acceptance criteria (see Table 9-12).
The modified Hydrolab software module in the field computer system helped to decrease the number of lost
data files and standardized the datafile format. The field crews were able to review time series plots for all parameters
and identify any malfunctioning units in the field. On several occasions, the crews were unable to establish communications
and download data from a retrieved unit. These units were returned to the field operations center and usually
the data were retrieved; however, there was not a post-deployment QC check for these records.
Table 9-12. Results of calibration checks following retrieval of Hydrolab Datasonde 3 instruments for
1991 Virginian Province monitoring(base sampling sites only). Percentages indicate the
number of times that acceptance criteria were met.
Parameter
Temperature
Salinity
DO
PH
Acceptance Criteria
±1 °C
±2ppt
± 0.5 mg/L
± 0.5 units
Percent Accepted
1 00% (1 06/1 06)
99% (105/106)
94% (1 00/1 06)
99% (1 05/1 06)
Page 72
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
YSI DO Meter
The YSI dissolved oxygen meter provided a useful QC comparison for theCTD and DataSondeS instruments
and an additional point-in-time measurement of bottom dissolved oxygen at most stations.
The YSI probes were calibrated prior to use at each station and calibration QC checks were conducted weekly.
The water-saturated air calibration method appeared to yield acceptable results. During the weekly calibration
QC checks, the YSI meter measured slightly lower dissolved oxygen levels than Winkler titrations and expected
saturation table values, but the 0.5 mg/L acceptance criteria was met 89% of the time (Table 9-10). The YSI
dissolved oxygen values agreed closely with the CTD dissolved oxygen measurements during their side-by-side
checks (Table 9-10).
The results of the HACK Winkler titrations were greatly improved in 1991. In no cases did the dissolved
oxygen measured in two replicate water samples exceed 0.5 mg/L; in fact the maximum difference was only 0.3
mg/L.
9.7 1992 QA Results
SeaBird CTD
All calibrations were conducted at the instrumentation facility in Narragansett, RI. A calibration tank with
air-saturated freshwater was always on hand to perform dissolved oxygen calibrations. A downfall of this system
is that the faulty CTD had to be shipped to the testing facility and a replacement unit sent to the field crew. This
often created a hiatus in the collection of CTD data for that field crew.
Weekly calibration QC checks were an appropriate method for evaluating the performance of the sensors
and recognizing any calibration drifts. The side-by-side comparisons with the YSI dissolved oxygen meter were
a simple check that produced reliable results (Table 9-13).
Results of the review of 1992 CTD files are presented in Table 9-14. Acceptable bottom DO values from
the CTD were measured at 92% of the stations used in EMAP's assessment of the ecological condition of the
Province (i.e., Base Stations). And, because redundant measurements were taken with the YSI meter, bottom
dissolved oxygen concentration data are available for those stations where the CTD failed to pass QC.
YSI DO Meter
The YSI dissolved oxygen meter provided a useful QC comparison for theCTD and an additional point-in-time
measurement of bottom dissolved oxygen at most stations. In addition, surface YSI values were also collected
beginning in 1992. This provided a better check on the CTD than bottom measurements because the crew could
better assure that both measurements were made at the exact same depth.
The YSI probes were calibrated prior to use at each station and calibration QC checks were conducted weekly.
The water-saturated air calibration method appeared to yield acceptable results. During the weekly calibration
QC checks, the YSI meter measured slightly lower dissolved oxygen levels than Winkler titrations and expected
saturation table values, but the 0.5 mg/L acceptance criteria was met 100% of the time (Table 9-13). The YSI
dissolved oxygen values agreed closely with the CTD dissolved oxygen measurements during their side-by-side
checks (Table 9-13).
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 73
-------
Table 9-13. Results of weekly calibration checks of water quality instruments used in the Virginian
Province, 1992.
Instrument
YSI meter
Seabird
CTD
Parameter
Temperature
DO
Temperature
Salinity
DO
PH
Checked
against
Thermometer
Winklertitration
Thermometer
Refractometer
YSI meter
Standard buffer
Acceptance
Criteria
±2 °C
± 0.5 mg/L
±2 °C
±2ppt
± 0.5 mg/L
± 0.5 units
Percent
Accepted
100% (16/16)
100% (16/16)a
1 00% (1 7/1 7)
1 00% (1 7/1 7)
100% (17/17)b
1 00% (1 7/1 7)
One check barely passed with a difference of 0.5 mg/L.
Two tests barely passed with a difference of 0.5 mg/L
Table 9-14. Results of 1992 post-sampling season CTD data review. (Percentages are based upon #
of reviewed casts or number of Base Stations). Note: different criteria were used for
accepting and rejecting CTD cast data in 1990 vs. 1991 and 1992. Acceptability of casts in
1990 was based solely upon performance of the DO sensor whereas performance of all
sensors were considered in 1992.
Total Casts
Base Stations
Total casts/Base Stations
# Casts accepted unqualified for all parameters (C-B code)
# Casts rejected for all parameters (includes lost casts)
# Casts w/ acceptable surface DO
# Casts w/ acceptable bottom DO
144
58 (40%)
3 (2%)
126(88%)
123(85%)
103
57 (55%)
0 (0%)
98 (95%)
95 (92%)
9.8 1993 QA Results
SeaBird CTD
All calibrations were conducted at the Virginian Province instrumentation facility in Narragansett, RI. A
calibration tank with air-saturated freshwater was always on hand to perform dissolved oxygen calibrations.
A downfall of this system is that the faulty CTD had to be shipped to the testing facility and a replacement unit
sent to the field crew. This often created a hiatus in the collection of CTD data for that field crew.
Weekly calibration QC checks were an appropriate method for evaluating the performance of the sensors
and recognizing any calibration drifts. The side-by-side comparisons with the YSI dissolved oxygen meter were
a simple check that produced reliable results (Table 9-15).
Page 74
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Results of the review of 1993 CTD files are presented in Table 9-16. Acceptable bottom DO values from
the CTD were measured at 97% of the stations used in EMAP's assessment of the ecological condition of the
Province (i.e., Base Stations). And, because redundant measurements were taken with the YSI meter, bottom
dissolved oxygen concentration data are available for those stations where the CTD failed to pass QC.
YSI DO Meter
The YSI dissolved oxygen meter provided a useful QC comparison for theCTD and an additional point-in-time
measurement of surface and bottom dissolved oxygen at most stations.
The YSI probes were calibrated prior to use at each station and calibration QC checks were conducted weekly.
The water-saturated air calibration method appeared to yield acceptable results. During the weekly calibration
QC checks, the YSI meter measured slightly lower dissolved oxygen levels than Winkler titrations and expected
saturation table values, but the 0.5 mg/L acceptance criteria was met 100% of the time (Table 9-15). The YSI
dissolved oxygen values agreed fairly closely with the CTD dissolved oxygen measurements during their side-by-side
checks (Table 9-15).
Table 9-15. Results of weekly calibration checks of water quality instruments used in the
Virginian Province, 1993.
Instrument
YSI meter
Seabird
CTD
Parameter
Temperature
DO
Temperature
Salinity
DO
PH
Checked
against
Thermometer
Winkler titration
Thermometer
Refractometer
YSI meter
Standard buffer
Acceptance
Criteria
±2 °C
± 0.5 mg/L
±2 °C
±2ppt
± 0.5 mg/L
± 0.5 units
Percent
Accepted
1 00% (26/26)
100%(26/26)a
1 00% (25/25)
92% (23/25)
96% (24/25)b
1 00% (25/25)
a Two checks barely passed with a difference of 0.5 mg/L.
b One test barely passed with a difference of 0.5 mg/L
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
Page 75
-------
Table 9-16. Results of 1993 post-sampling season CTD data review. (Percentages are based upon # of
reviewed casts or number of Base Stations). Note: different criteria were used for accepting
and rejecting CTD cast data in 1990 vs. 1991 -1993. Acceptability of casts in 1990 was based
solely upon performance of the DO sensor whereas performance of all sensors were considered
in 1993.
Total Casts Base Stations
Total casts/Base Stations 147 111
# Casts accepted unqualified for all parameters (C-B code) 105 (71 %) 84 (76%)
# Casts rejected for all parameters (includes lost casts) 0 (0%) 0 (0%)
# Casts w/ acceptable surface DO 139 (95%) 105 (95%)
# Casts w/ acceptable bottom DO 143 (97%) 108 (97%)
Page 76 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Section 10
QA Results for Total Suspended Solids Analyses
10.1 Background
In 1990 water samples for total suspended solids (TSS) analysis were collected only at Indicator Testing
and Evaluation Sites. The intent was to use these data to evaluate transmissometer data collected from those
stations. Because these data are not used in the assessment of the ecological condition of the Province, no QA
results for 1990 TSS analyses are presented in this document.
Surface water samples were collected at all Base Sampling Sites beginning in 1991. A 250cc plastic bottle
was filled with water, refrigerated, and shipped to the laboratory for filtering and analysis according to standard
EPA methods.
A problem with the QC process for TSS samples was discovered during the preparation of this report. No
QA criteria were in place in 1991 by which the data could be evaluated against. A retrospective evaluation of
the 1991 data using the criteria set in place in 1992 show that large percentage of the samples should have been
flagged as failing QC when, in fact, they were not. However, in reviewing data from all three years it was discovered
that the QC requirements in the QA Plans were unrealistic and frequently not met. This is discussed below in
Sections 10.4 and 10.7.
10.2 Data Qualifier Codes for Total Suspended Solids Data
Data qualifier codes for the suspended solids dataset are listed in Table 10-1. The SOP called for filtering
a large enough volume of water to ensure the residue weight was atleast one milligram. The SS-C code was applied
to those samples with low TSS concentrations for which samples were refiltered using a larger volume of water,
and the weight of the residue was still less than one milligram.
Table 10-1. Data qualifier codes for total suspended solids data (NOTE: These codes may change - see
Section 10.7).
Code
Description
SS-A Sample failed to meet EMAP-Estuaries QA requirements. Relative percent difference between
duplicates exceeded 10%. Data should be used with caution.
SS-B No QC samples were run on the day this sample was analyzed. These data can not be evaluated
relative to EMAP-Estuaries QA standards.
SS-C Residue weight was less than 1.0 mg even with larger volume filtered. Value reported was associated
with the largest volume filtered.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
Page 77
-------
10.3 Audits
A11TSS analyses conducted in 1991 and 1992 were performed by SAIC's Environmental Testing Center (ETC).
As described in Section 6.3, this laboratory was audited in 1990 and 1991. The results of the 1991 audit included
TSS analysis, and were generally favorable, with no QA infractions noted. Samples collected in 1993 were analyzed
by the Marine Ecosystem Research Laboratory (MERL) of the University of Rhode Island. This laboratory has
extensive experience in TSS analyses; therefore, no audits were deemed necessary.
10.4 1991 QA Results
TSS samples were introduced as a research indicator in 1991, and, as such, no QA requirements were included
in the QA Plan. However, as part of routine analysis approximately 10% of the samples were reanalyzed. Subsequent
QA Plans required that at least 10% of all samples analyzed for TSS concentration be analyzed in duplicate.
To pass QA, the RPD between the duplicates must be less than 10%. If it exceeds 10%, all samples analyzed
since the last successful QC check must be repeated.
The mean RPD for the 14 sets of duplicates was 10.4%, with a maximum of 32.6%. Six of the fourteen sets
exceeded 10%; however, none of the data were assigned QA codes because control limits were not in place at
the time of the review. See Section 10.7 for additional discussion.
10.5 1992 QA Results
The QA Plan required that at least 10% of all samples analyzed for TSS concentration be analyzed in duplicate.
The RPD between the duplicates was then calculated. To pass QA, this value must be less than 10%. If it exceeds
10%, all samples analyzed since the last successful QC check must be repeated.
Due to an apparent mis-communication at the analytical laboratory, the first group of samples did not have
the appropriate QA samples run. Therefore, the quality of the resultant data cannot be evaluated and are "flagged"
in the EMAP database with the SS-B code. A sufficient number of duplicate analyses were performed with the
remainder of the samples; however, several failed QA, with the RPD exceeding 10%. Unfortunately this was
not discovered until several months after the analyses were completed, and the original samples (degradable)
had been discarded. As a result, approximately 44.4% of the data have been flagged as being of questionable
quality (SS-A or SS-B).
10.6 1993 QA Results
The QA Plan required that at least 10% of all samples analyzed for TSS concentration be analyzed in duplicate.
The RPD between the duplicates was then calculated. To pass QA, this value must be less than 10%. If it exceeds
10%, all samples analyzed since the last successful QC check must be repeated.
The analytical laboratory chose to analyze all of the samples in duplicate. The RPD for these analyses ranged
from 0 to 35.7% with a mean of 10.5%. The median RPD was 8.5%. Because duplicate analyses were available
for nearly all samples (duplicate data for 17 of the samples were not available due to analytical problems), and
the mean RPD slightly exceeded EMAP control limits, we chose to report the mean of the duplicates and not to
assign QA qualifier codes to any of the results. See Section 10.7 for additional discussion.
Page 78 Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
10.7 Lessons Learned and Changes Suggested
Results for total suspended solids generated in 1991 to 1993 suggest a problem with the stated QA process.
The RPD for nearly half of the 1991 and 1992 duplicate pairs fell outside of the control limits. Evaluation of
the 1993 results, which were generated by MERL (an academic laboratory with extensive experience inTS S analyses),
showed that half of the samples had an RPD greater than 8.5%, with a mean RPD of 10.5%.
In the analysis of TSS samples, water is filtered and small masses of sediment weighed. The relatively large
tare weight of the pans when compared to the small weight of the samples likely results in the errors shown. The
1993 results suggest that a better method would be to analyze ALL samples in duplicate and report the mean
of the measurements. We recommend that in the future this methodology be employed for all TSS samples.
In the interim we suggest that a new QA Qualifier Code be applied to all 1991 and 1992 samples which simply
states thatthevalue reported represents results from a single measurement rather than themean of two measurements.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 79
-------
Section 11
Summary of Data Collection Success
Data completeness goals are provided in the annual Quality Assurance Project Plans. Generally a minimum
completeness goal of 90% is listed for each indicator. Table 11-1 provides summary information regarding data
completeness. Ofthe446 Base Sampling Sites originally selected, 21 were deemedunsampleable due to inaccessibility,
obstructions, or water depth and could not be moved in accordance with the design. The completeness rate for
most indicators was above or close to the 90% mark. The notable exception is suspended solids; however, the
collection of samples for TSS analyses at all Base Sampling Sites did not begin until 1991.
Table 11-1. Summary of collection and processing status of samples collected in 1990-1993 (Base Sampling
Sites only).
Sample Type
Dissolved Oxygen
Light Attenuation Coefficient (CTD cast)
Suspended Solids
Sediment Chemistryd
Organics
Metals
Sediment Toxicity
Sediment Grain Size
Benthic Infauna
Fish Community Data (successful trawl)
# Stations
Expected to
be Sampled3
446
446
446
446
446
446
446
446
446
# Stations Sampled
With Data Passing
Final QCb (%)
420
408
298
397
394
373
394
404
390
(94%)
(91%)
(67%)c
(89%)
(88%)
(84%)
(88%)
(91%)
(87%)
A total of 446 Base Sampling Sites were originally selected for sampling. Of these, 21 were found to be
unsampleable due to obstructions or inadequate water depth prior to the sampling season.
This value takes into account samples not collected, damaged or lost during shipping or processing, or failing to
pass final Quality Control checks.
Samples for TSS analyses were not collected in 1990. Note that QA Criteria did not exist for 1991 samples.
The success rate denotes percent of stations with some valid data. However, as discussed in Section 3, not all
stations successfully sampled have valid data for all analytes.
Page 80
Quality Assurance Report, EMAP-Virginian Province 1990 -1993
-------
Section 12
References
Holland, A.F., ed. 1990. Near Coastal Program Plan for 1990: Estuaries. EPA 600/4-900/033. Narragansett,
RI: U.S. Environmental Protection Agency, Environmental Research Laboratory, Office of Research and
Development.
Latimer, R.W. 1992. FY 1992 Annual Quality Assurance Report and FY 1993 Work Plan for the Environmental
Monitoring and AssessmentProgram Estuaries Resource Group. Narragansett,RI:U. S. Environmental Protection
Agency, Office of Research and Development, Environmental Research Laboratory. September 1992.
Schimmel, S.C., B.D. Melzian, D.E. Campbell, C.J. Strobel, S.J. Benyi, J.S. Rosen, andH.W. Buffum. 1994.
Statistical Summary: EMAP-Estuaries Virginian Province - 1991. EPA/620/R-94/005 Narragansett, RI:
U.S.EnvironmentalProtectionAgency,EnvironmentalResearchLaboratory,OfficeofResearchandDevelopment.
Strobel, C.J. and S.C. Schimmel. 1991. EMAP-Estuaries 1991 Virginian Province Field Operations and Safety
Manual. Narragansett, RI: U.S. Environmental Protection Agency, Office of Research and Development,
June 1991.
Summers. 1993. FY 1993 Annual Quality Assurance Report and FY 1992 Work Plan for the Environmental
Monitoring and Assessment Program Near Coastal Resource Group. Gulf Breeze, FL: U.S. Environmental
Protection Agency, Office of Research and Development, Environmental Research Laboratory. September
1993.
Valente, R.M. 1991a. FY 1990 Annual Quality Assurance Report and FY 1991 Work Plan for the Environmental
Monitoring and AssessmentProgram Near Coastal Component. Narragansett,RI:U.S. Environmental Protection
Agency, Office of Research and Development, Environmental Research Laboratory. February 1991.
Valente, R.M. 199 Ib. FY 1991 Annual Quality Assurance Report and FY 1992 Work Plan for the Environmental
Monitoring and Assessment Program Near Coastal Resource Group. Narragansett, RI: U.S. Environmental
Protection Agency, Office of Research and Development, Environmental Research Laboratory. September
1991.
Valente, R. and J. Schoenherr. 1991. EMAP-Estuaries Virginian Province Quality Assurance Project Plan.
Narragansett, RI: U.S. Environmental Protection Agency, Office of Research and Development, Environmental
Research Laboratory. July 1991.
Valente, R., C.J. Strobel J.E. Pollard, K.M. Peres, T.C. Chiang and J. Rosen. 1990. Quality Assurance Project
Plan for EMAP'Near Coastal: 1990 Demonstration Project. Narragansett, RI: U.S. Environmental Protection
Agency, Office of Research and Development, Environmental Research Laboratory. July 1990.
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, RI. May 1993.
Valente, R., C.J. Strobel and S.C. Schimmel. 1992. EMAP-Estuaries Virginian Province 1992 Quality Assurance
Project Plan. Narragansett, RI: U.S. Environmental Protection Agency, Office of Research and Development,
Environmental Research Laboratory. July 1992.
Quality Assurance Report, EMAP-Virginian Province 1990 -1993 Page 81
------- |