United States
Environmental Protection
Agency
Science and Ecosystem
Support Division
Region 4 and
Office of Research &
Development
EPA-904-R-98-002C*
October, 1998
South Florida Ecosystem Assessment
Vol II. Final Technical Report Appendices
Monitoring for Adaptive Management:
Implications for Ecosystem Restoration
-------
APPENDIX A
Sampling Apparatus
-------
Figure A.l. Surface flange of soil cores.
A-l
-------
THREAD OUTSIDE OF PIPE APPROXIMATE 1"
AT 32 THREADS PER INCH. FINE
.125 WALL . n.D CLEAR ACRYUC
PIPE X 11 "LENGTH
USING A RLE. CUT EQUALLY SPACED
SERRATED EDGES AT 45*
J
SCALE: ,5"=1.0"
Figure A.2. Acrylic soil sampling tube.
A-2
-------
.235X15 DEEP GROVE
FOR 5.5"® 3/16" THICK 0 RING
SEE DETAIL A
SEE DETAIL B
3/32 DRILL.
3 PLACES
6.0 OD CLEAR ACRYLIC PIPE
W/ 5.0ID, .5 WALL THICKNESS
.1850 m
.235
3/32 DRILli.
3 PLACES !
TT
6.0
• SEE DETAIL B
Figure A.3. Main body of vacuum chamber for vacuum water sampler.
A-3
-------
.25(8 NOTCH
4 PLACES EQUAL
SPACED
4.83c
.25 DRILL
HOLES
2 PLACES
LENGTH OF TUBING
TO BE DETERMINED
BY YOUR SPECIFIC
NEEDS.
LENGTH OF TUBING
AND TEFLON PIPE/HANDLE
TO BE DETERMINED
BY YOUR SPECIFIC NEEDS.
HAND OPERATED VACUUM-
PUMP, W/ GUAGE
NALGE § 20940-00 OR
EQUAL
GELMAN # 13529-00 MAGNETIC
FILTER HOLDER W/ FUNNEL
USING A HACKSAW
CUT OFF THE FUNNEL
PORTION AND SHORTEN
THE TUBE ON THE
OTHER END.
Figure A.4. Details of parts for vacuum water sampler.
A-4
-------
2
3/8 DRILL
AND TAP FOR
3/8" MALE TO
MALE NYLON
COMPRESSION
FITTING
3/32 DRILL AND
TAP FOR 3/32
CAP SCREWS
3 PLACES
3/8" MALE TO
MALE NYLON
COMPRESSION
FITTING
Figure A.5. Details of parts for vacuum water sampler.
-------
DETAIL C
THREAD INSIDE OF
PIPE TO ACCEPT
THREADED PLUG
CUT A NOTCH IN SIDE
OF PIPE FOR 1CMX1CM
PLATES. SOLDER WIRE
TO CENTER OF PLATE
AND SEAL PLATE INTO
NOTCH BEING SURE
THAT IT IS WATER
TIGHT.
.5 DRILL
THREADED PLUG
W/ORING SEAL
i
DETAIL D
SOLID PVC PLUG
TAPERED AT 45'
GLUE IN PLACE.
. 1.5 ,
-SEE DETAIL C
16GA. PLATNUM
1CMX1CM 5 REQD.
END CAP
PLACEMFNT OF
lENSOR NOTCHFS PLACEMFNT
3F SENSORS
DETAIL D
Figure A.6. Details of soil redox probe with sensors.
A-6
-------
.125 DRILL <1
TAP FOR
SET SCREW
2.5 OD
LOCKING COLLAR
J3
L
-1X5.5X.25 THICK BAR
LOCKING
COLLAR
h> a
1X1X.25 THICK
ANGLE X 6"
LENGTH 2 REQD
.25 WELD
ALL AROUND
SIDE VIEW BASE PLATE ASSfMBl Y
BASE
PLATE
TOP VIEW BASE PLATE ASSEMBLY
S RR7S
5.6875X6.0
THICK PLATE
Figure A.7. Base plate assembly for redox soil probe.
A-7
-------
APPENDIX B
Data Quality Objectives
-------
SOUTH FLORIDA ECOSYSTEM ASSESSMENT PROJECT
DECISION-BASED DATA QUALITY OBJECTIVES
March 1997
-------
South Florida Ecosystem Assessment Project
Decision-Based Data Quality Objectives
March 1997
-------
March 1997
TABLE OF CONTENTS
Data Quality Objectives 1
Background 1
State the Problem ]
Identify the Decision 2
Identify the Inputs to the Decision 6
Define the Boundaries of the Study 6
Develop a Decision Rule 11
Specify Tolerable Limits on Decision Errors 12
Optimize the Design 19
APPENDIX
Appendix A Data Quality Objective Criteria
i
-------
March 1997
LIST OF TABLES
Table 1 Mercury Related Legislative and Regulatory Deadlines 3
Table 2 Policy-Relevant Questions Guiding the Project 4
Table 3 A Information Needs, Source and Method 7
Table 3 B Other Information Needs and Sources 8
Table 4 Geographic Area Boundaries 9
Table 5 Preliminary Decision Rules for South Florida 14
Table 6A Water Constituents Ranges in South Florida 16
Table 6B Soil/Sediment Constituents Ranges in South Florida 17
LIST OF FIGURES
Figure 1 Logic pathways for decisions on status and Trends and Diagnosis and
Management Questions. Pathways diagram information and analyses needed to
answer the seven policy-relevant questions 5
Figure 2 Study area boundaries, major management areas, and Everglades ecosystem
community types 10
Figure 3 Distribution of total phosphors and total mercury in water and total mercury in a
fish prey species, with 95% confidence intervals. Note: The existing action levels
are under protective because ecological effects are observed from eutrophication
and mercury contamination 13
Figure 4 Proposed mercury Action Level for predator prey species identifying The Gray
Regions and area for 10% Tolerable Decision 1 Error Rates 22
ii
-------
South Florida Ecosystem Assessment Project
Decision-Based Data Quality Objectives
March 1997
Data Quality Objectives
The Data Quality Objectives (DQOs) were prepared generally following the Guidance for the
Data Quality Objectives Process EPA QA/G-4 (EPA 1994). This US Environmental Protection
Agency (EPA) Guidance document, however, is not entirely appropriate for research projects.
The EPA Quality Assurance Management Staff are in the process of preparing DQO guidance for
research projects, but this guidance will not be available until late fall (John Warren, QAMS,
personal communication). The South Florida Ecosystem Assessment Project is a research project
that, in part, is developing risk-based criteria for decisions because the existing criteria are not
appropriate or no criteria exist. Therefore, two separate, but complementary, approaches were
used to develop DQOs. The first approach was to use the EPA QA/G-4 documentation as
guidance in developing decision-based DQOs, which are discussed in this document. This
document uses the EPA QA/6-4 report format. The second approach revised the DQOs
originally proposed in the REMAP Research Plan (Stober et al. 1993). These revised DQOs are
listed in Appendix A.
Background
In 1989, a Florida panther, an endangered species, died because of mercury toxicosis. Since then,
over 2 million acres in South Florida have been placed under fish consumption advisories
because of mercury contamination. The EPA Region 4 Science and Ecosystem Support Division
(SESD), therefore, was charged by the EPA Regional Administrator to develop an action plan to
evaluate the mercury issue and provide a scientific basis for evaluating options and strategies to
eliminate mercury contamination in the South Florida Everglades Ecosystem. Subsequently, the
Region 4 SESD prepared a research plan, had this plan peer-reviewed, and initiated the study as a
Regional Environmental Monitoring and Assessment Program (REMAP) Project. As the Project
planning and pilot Project proceeded, it became obvious that the environmental issues in South
Florida (eutrophication, mercury contamination habitat alteration, hydroperiod modification) are
highly interactive and need to be addressed through an integrated monitoring and assessment
program. Therefore, the REMAP Project was expanded to become the South Florida Ecosystem
Assessment Project addressing these multiple environmental issues. The variables being
measured in this Project will permit answers to questions on these multiple environmental issues.
A central goal of the Project, however, remains to answer assessment questions related to the
magnitude, extent, trends, and transformation processes in mercury contamination of the South
Florida Everglades Ecosystem.
State the Problem - a description of the problem(s) and specification of available resources and
relevant deadlines for the study.
1
-------
Ecosystem Assessment
Revised DQOs Mar 97
(1) Identify the members of the team - The team consists of the Region 4 Project Manager
(i.e., fisheries biologist), SESD; Assistant Project Manager (natural resources manager),
Water Division; Quality Assurance Officer (chemist); Southeast Environmental Research
Program manager (microbial ecologist and chemist), Florida International University;
spatial statistician, University of Georgia; and systems ecologist and data analysts, FTN
Associates, Ltd.
(2) Identify the primary decision maker(s) - The primary decision maker is the South Florida
Ecosystem Assessment Project Manager. Other decision makers include the Assistant
Project Manager, Division Directors for the Water Division and Science and Ecosystem
Support Division.
(3) Develop a concise description of the problem - Mercury contamination, nutrient loading,
hydropattern modification, and habitat alteration are impacting fish and wildlife in the
South Florida Everglades Ecosystem. The sources, causes, and interactions among many
of these environmental stressors are unknown. Environmentally-sound, cost-effective
restoration of the South Florida Everglades Ecosystem, however, depends on identifying
these sources, causes and interactions. Almost one billion dollars are estimated to be
spent on this restoration effort.
(4) Specify the available resources and relevant deadlines for the study - Approximately
$1 million dollars/year are needed and available to determine the magnitude, extent,
trends and possible causes of the mercury contamination, eutrophication, hydropattern
modification and habitat alteration problems. This represents less than 0.1% of the
proposed restoration expenditures. The relevant regulatory deadlines are listed in
Table 1. These regulatory deadlines extend through 2004, with a major milestone in 1999
when the EPA mercury report is due to the South Florida Ecosystem Restoration Task
Force.
Identify the Decision - a statement of the decision that will use environmental data and the
actions that could result from this decision.
(1) Identify the principal study questions - The principal study questions were identified as
part of the original proposal and specification of the DQOs. These seven policy-relevant
questions are listed in Table 2.
(2) Identify alternative actions that could result from resolution of the principal study
questions - The logical alternative actions and pathways that could result in answering
these seven questions were identified during the initial phases of the Project. These
pathways were incorporated into a Visual Basic computer program to show the logical
development of these alternative actions. The expanded logic pathways from this
computer program are shown in Figure 1. These logic pathways and alternative action
formulations are a major part of the Problem Formulation phase of the Ecological Risk
2
-------
Table 1. Mercury Related Legislative and Regulatory Deadlines.
Date
Federal
Florida
1995
NPDES Permit for the ENR project
(CWA)
1996
EIS for the Everglades Construction
Project (NEPA)
1996
404 Permit for the Everglades
Construction Project (CWA)
Oct 1997
404 Permit for STA-6 (CWA)
STA-6 NPDES Permit and 402
Certification
Sep 1998
USACOE Central & Southern Florida
Project Restudy Plan Draft Report &
Draft EIS (WRDA, NEPA)
Dec 1998
Evaluation of water quality standards
for the Everglades Protection Area &
EAA canals (EFA)
Jan 1999
STA-1W, 2, & 5 404 Permits (CWA)
STA-1W, 2, & 5 NPDES Permits,
402 Certification (CWA)
Jul 1999
Final Restudy Report and EIS due to
Congress (WRDA, NEPA)
Dec 1999
Report to Governor and Legislature
on status of EPA mercury study
(EFA)
Dec 2001
Phosphorus criterion rulemaking for
Everglades Protection Area and EAA
canals (EFA)
Oct 2003
STA-3 & 4 404 Permits (CWA)
STA-3 & 4 NPDES Permits and 404
Certification (CWA)
Dec 2003
Revised water quality standards for
the Everglades Protection Area &
EAA canals (EFA)
2004
Approval of water quality standards
for the Everglades Protection Area &
EAA canals (CWA)
WRDA: Federal Water Resources Development Act STA: Stormwater Treatment Area
EFA: Florida Everglades Forever Act EAA: Everglades Agricultural Area
CWA: Federal Clean Water Act NEPA: Federal National Environmental Policy Act
3
-------
Ecosystem Assessment
Revised DQOs Mar 97
Table 2. Policy-Relevant Questions Guiding the Project.
Status and Trends
1)
What is the magnitude of the mercury problem? What are the current levels of
mercury contamination in various species? What ecological resources of
interest are being adversely impacted by mercury?
2)
What is the extent of the mercury problem? (i.e., what is the geographic
distribution of the problem? Is it habitat specific?)
3)
Is the problem getting worse, better, or staying the same over time?
Diagnosis and Management
4)
What factors are associated with, or contributing to, methylmercury
accumulation in sensitive resources?
5)
What are the relative contributions and importance of mercury from different
sources (e.g., fossil fuel plants, waste incinerators, agricultural management
practices, geologic pools, natural peat deposits, global atmospheric background,
etc.)?
6)
What are the relative risks to different ecological systems and species from
mercury contamination?
7)
What management alternatives are available to ameliorate or eliminate the
mercury contamination problem?
Assessment Framework that forms the foundation of this study. Dichotomous trees were
formulated for each of the logic pathways developed during the initial Project phases.
These trees were developed prior to the initiation of the field sampling and were used to
assist in the formulation of the preliminary project DQOs.
(3) Combine the principal study questions and the alternative actions into a decision
statement - "Decide how the relative ecological risk from mercury contamination
compares with the risks from nutrient additions, hydropattern modification, habitat
alteration. Determine if controlling these other stressors will eliminate mercury
contamination; if not, determine procedures that can be used to eliminate mercury
contamination."
(4) Organize multiple decisions - Multi-decision pathways will be based on the outcomes
from the logic pathway analyses shown in Figure 1. These logic and decision pathways
will be refined as the Project proceeds and new information is collected and analyzed.
4
-------
Status and Trends
Eile yiew
Cuiient Statin
,Sp>atLal ¦
^£attein ¦'
I
Giatfisnts
Hot Spots'
N-S
Association
Analyses
. y.Statut "ardT. rends •
Historical Status
Historical'
\P(6Wm»;
('Recent '
iPiobieni
t
Trervdz
'>Rebo"spectiVe";>:
, :;N? -'
Regional'
5>'Change''
NoSpatial
.Random 3 High
T
Random
Varial.
— Low
nm.
No:PiobL:
Al'.Thi*
c'Scale
Ecporatory
Analyses
- Post-Stat
- Ouster rig
yes
ISnatler^ FoOqmhg
££'caje,« j^«iar>
Pr»W«&?:
-Tofpatial
Regional
"Change
Increased
UeHjf
Decreased
MeHo
t-fopl;,:.
^Attribute'
* '_ r •_ ¦>,
VSefect;
'fsuKpop-,
!*-*EnCre
£Po£\
Natural
Variation
, Early
Warning
¦To'Spatial
%Commdft\
NoXoaaon
r.
* •-1» j -
rptSpatial
No
Cttange'
B
Continue
: T 9
Monitof
Diagnosis and Management (Exploded
Back
Diagnosis
-¦;Oiaorew«;«rid M aiugeaent
Management
Sources
of Risk
Figure 1. Logic pathways for decisions on status and Trends and Diagnosis and Management
Questions. Pathways diagram information and analyses needed to answer the seven
policy-relevant questions
5
-------
Ecosystem Assessment
Revised DQOs Mar 97
Identify the Inputs to the Decision - a list of the environmental variables or characteristics that
will be measured and other information needed to resolve the decision statement.
(1) Identify the information that will be required to resolve the decision statement - The
information needed to resolve the decision statements is listed in Table 3.
(2) Determine the sources for each item of information identified - The South Florida
Ecosystem Assessment Project (SFEA) is the primary source of the information needed to
address the decision statements. The decision statements can not be resolved without this
Project. Additional sources of information also are identified in Table 3.
(3) Identify the information that is needed to establish the action level - The criteria that will
be used to establish the action level will be:
(a) Variability - ecological effects significantly different from natural variability
(b) Endpoints - reproduction, feeding efficiency, behavioral changes, and other
ecologically relevant processes, in addition to toxicity
(c) Temporal scale - chronic versus acute effects
(d) Spatial scale - small versus large scale effects
For most constituents, regulatory criteria or standards do not exist. The decision will be
made using risk-based action levels.
(4) Confirm that appropriate measurement methods exist to provide the necessary data - For
conventional pollutants, EPA approved methods are being used to measure environmental
variables with an approved QAPP. For some constituents, such as total phosphorus,
existing EPA methods do not have the resolution needed to detect low-level background
concentrations. For other constituents, such as methylmercury in water, soil, and
sediment, there are no approved measurement methods. Therefore, experimental
measurement methods are being developed for these constituents, with confirmatory
analyses being conducted by independent laboratories.
Define the Boundaries of the Study - a detailed description of the spatial and temporal
boundaries of the problem, characteristics that define the population of interest, and any practical
considerations for the study.
(1) Specify the characteristics that define the population of interest - The target population
or population of interest are all ecological resources in the South Florida study area. This
includes the freshwater wetlands, open water and canals found in the Everglades National
Park (ENP), Water Conservation Areas (WCAs), Big Cypress National Preserve (BiCY),
and Everglades Agricultural Areas (EAA). The media to be sampled include soil,
sediment, water, and biota. The emphasis is on mercury concentrations in biota,
especially fish tissue. However, one of the desired outcomes of the Project is better
estimates of the type and proportion of ecological resources and the impacts of other
stressors on these resources in South Florida.
6
-------
Table 3. Information Needs, Source and Method.
Measurement Variable
Source
Method
Physical Measurements
Site location
SFEA
Global Positioning System
Weather
SFEA, NOAA
Visual observation,
meteorological stations
Discharge, structure
SFWMD
Gage readings, pump capacity
Water depth
SFEA, SFWMD
Calibrated line, depth recorders
Temperature
SFEA
Thermistor
Peat depth
SFEA
Calibrated probe
Turbidity
SFEA
Turbidimeter
Bulk density
SFEA
Balance, weighing
% Mineral content
SFEA
Combustion furnace
Ash free dry weight
SFEA
Combustion Furnace
Chemical Measurements
Dissolved oxygen
SFEA
DO probe
Specific conductance
SFEA
Conductivity meter
PH
SFEA
pH meter
Total organic carbon
SFEA
Total carbon analyzer
Total phosphorus
SFEA
New method development
Sulfate
SFEA
New method development
Total mercury
SFEA
New method development
Methymercury
SFEA
New method development
Alkaline phosphatase
SFEA
New method development
Redox potential
SFEA
Volt meter
Biological Measurements
Resource class (canal,
sawgrass marsh, cattails, etc.)
SFEA
Visual inspection
Periphyton presence/absence
SFEA
Visual observation
Chlorophyll a
SFEA
New method development
Soil/Sediment total mercury
SFEA
New method development
Soil/Sediment methylmercury
SFEA
New method development
Fish total mercury
SFEA
New method development
7
-------
Ecosystem Assessment
Revised DQOs Mar 97
Table 3. Other Information Needs and Sources.
Information Needs
Sources
Water management operation records
SFWMD, COE
Atmospheric mercury deposition/evasion
FL DEP, EPA, FAMS,
SFWMD, UFL, FSU
Nutrient loading estimates
SFWMD
Habitat changes
FWS NWI, NPS
Simulated natural hydropatterns
SFWMD
Vegetation patterns and production
NPS, FWS, SFWMD
ENR Project results
SFWMD, FL DEP
Periphyton production - nutrient relationships
SFWMD, FL DEP, FIU, UWI
Organic carbon speciation
USGS
Sulfate reduction/loading
SFWMD, USGS, FIU, UWI
Mercury methylation/demethylation
USGS, SFWMD, UMD, FIU,
UFL, UWI
Fish and invertebrate impacts
FWS, NPS, FIU, UFL
Wading bird impact
FWS, NPS, UFL
Large mammal and reptile impacts
FWS, NPS, FIU, UFL, FSU,
UGA
(2) Define the spatial boundary of the decision statement
(a) Define the geographic area to which the decision statement applies. The
geographic area being studied, and for which decisions apply, is approximately
160 km long and 60 km wide, resulting in an area of about 9600 km2. The exact
boundaries are listed in Table 4 (next page) and shown in Figure 2.
8
-------
Ecosystem Assessment
Revised DQOs Mar 97
Table 4. Geographic Area Boundaries.
Boundary
Description
Northern
West from Canal L8 to its junction with Lake Okeechobee and across to the
Caloosachatchee River.
Western
Vertical line from the intersection of the Caloosahatchee River and Highway
833 south to the coast (the mangrove region is excluded from the target
population).
Southern
Edge of the western mangrove east to the intersection with Highway US 1.
Eastern
Highway US 1 north to its intersection with Highway 27, then along the eastern
boundaries of Water Conservation Areas to the Intersection with Canal L8.
(b) When appropriate, divide the population into strata that have relatively
homogeneous characteristics. Strata of interest were based on the decision
statement, rather than on homogeneity of variance. For example, there was less
interest in defining the characteristics of the Big Cypress National
Preserve (BiCY) than in other designated geographic areas. Therefore, BiCY was
sampled with a lower inclusion probability (approximately 1/3 the density of other
areas within the study boundaries). In addition, subsequent analyses have
indicated the areas north of Alligator Alley, between Alligator Alley and Tamiami
Trail, and south of Tamiami Trail have attributes that can influence management
and policy decisions.
(3) Define the temporal boundary
(a) Determine the timeframe to which the decision statement applies. The decision
statement applies from the time of the first data collection in April 1994 until at
least 2004. The mercury-related legislative and regulatory deadlines are defined
in Table 1. However, Project results are applicable to a longer timeframe because
the South Florida Ecosystem Restoration Task Force has legislative mandates for
hydropattern modification, habitat alteration and eutrophication deadlines beyond
2004 that can be addressed with results from this Project.
(b) Determine when to collect the data. Because time and space scales are inexorably
coupled, the synoptic sampling approach spatially dictates that the temporal
sampling frequency be seasonal. There are two distinct hydrologic seasons in
Florida. The dry season extends from November to April and the wet season
extends from June until September. May and October are transitional months.
Sampling during only one season could result in biased and flawed decisions on
9
-------
Everglades Ecosystem:
Wet Prairie -
Sawgrass Marsh
R-o C.O'ev* Nai'C.a1 c-ese've
vv-iG'-'e f.'anao<>'rioni Area
Cypress Forest
Figure i. Study area boundaries, major management areas, and Everglades ecosystem community
types
10
-------
Ecosystem Assessment
Revised DQOs Mar 97
management or regulatory issues, because of seasonal variability. Sampling, therefore,
needs to be done during both the dry and wet season. Decisions will be made over the
next decade, based, in part, on spatial and temporal trends in information. These
trends can not be defensibly determined with only one set (wet and dry season) of data
at the beginning and end of the decision time frame. Two reference periods define
change, not trends. Power analyses will be conducted to determine the number of
sampling intervals needed to detect statistically defensible trends and contribute to the
decision process.
(4) Define the scale of decision making - Decisions on mercury management and restoration
issues must be made for the entire South Florida ecosystem. The environmental issues
arose because of small-scale, piecemeal approaches to managing the system.
(5) Identify practical constraints on data collection - The large geographic area for sampling,
and the need to collect synoptic samples requires that sampling be conducted by multiple
teams using helicopters and airboats. The sampling period should be no longer than
10 days to minimize large scale changes in meteorology affecting water depth and quality
measurements. The number of samples and sample volume need to be minimized to
reduce weight and time for collection, but with sufficient volume to permit precision and
accuracy requirements to be achieved. Clean sampling procedures are required for the
mercury analyses, both in the field and in the laboratory. Low concentration nutrient
analyses also are required because of the ultraoligotrophic condition of the Everglades
wetlands.
Develop a Decision Rule - to define the parameter of interest, specify the action level and
integrate previous DQO outputs into a single statement that describes a logical basis for choosing
among alternative actions.
[NOTE: This DQO guidance statement is not compatible with the South Florida
Ecosystem Restoration goals and objectives. The issues in South Florida are not
independent; they are highly interactive. Multi-media decisions are required for multiple
issues. There is no single statement can be formulated that will permit decisions among
alternative actions. The greatest threat to the Everglades ecosystem is to assume these
issues are independent and derive one single statement to address all issues. The Project,
in part, will determine what the criteria should be for multiple issues such as phosphorus
loading, water depth, distribution and timing, methylmercury concentrations in
multi-media, and habitat types.]
(1) Specify the statistical parameter that characterizes the population of interest - REMAP is
an exploratory research program so no single statistical parameter has been selected to
characterize the population of interest. In addition, the emphasis is not on one single
constituent, such as a hazardous material that might exceed a regulatory standard. Rather,
11
-------
Ecosystem Assessment
Revised DQOs Mar 97
several statistical parameters are needed to characterize different population attributes,
including:
(a)
(b)
(c)
(d)
(e)
mean concentrations of selected constituents (see Table 3 for constituents)
cumulative distributions of constituents, by season, by area
distributional differences among constituents
spatial patterns of constituents, and
spatial/temporal associations among constituents.
(2) Specify the action level(s) for the study - Three action levels currently exist:
(a)
(b)
(c)
Phase I control target for total phosphorus of 50 /Ug/L (ppb);
Water total mercury criterion for protection of aquatic life of 12 ng/L
(ppt); and
Proposed predator protection level for mercury of 100 /^g/kg (ppb) for prey
species.
All three of these levels are underprotective. New risk-based action levels need to be
determined. Currently, 95% of the marsh has total phosphorus concentrations less than
50 ppb; 100% of the marsh has total mercury concentrations less than 12 ppt, and 68% of
the marsh has prey fish species with mercury concentrations greater than 100 ppb
(Figure 3). Developing appropriate risk-based action levels for total phosphorus and
mercury is one of the objectives of this Project. The detection and minimum quantitation
limits for all three of these constituents are less than the respective criterion. Because
risk-based action levels are needed, methods with increased sensitivity have been
developed and are being tested.
(3) Develop a decision rule (an "if...then " statement) - Decision rules express what the
decision maker ideally would like to resolve. The decision has been made that revised
criterion are needed, based on the information developed to date from the Project.
Preliminary decision rules, given this need, are listed in Table 5. Subsequent revisions of
the DQO document will expand and refine these decision rules as additional information
becomes available. Logic flow paths have been formulated (Figure 1) to increase the
probability future information will improve the efficacy of the decision rules.
Specify Tolerable Limits on Decision Errors - the decision maker's tolerable decision error
rates based on a consideration of the consequences of making a decision error.
(1) Determine the possible range of the parameter (s) of interest - The possible range of the
parameters of interest are listed in Table 6. These ranges are based on this Project and
other studies conducted in the South Florida Everglades ecosystem.
12
-------
Total Mercury, ng/L
Figure 3. Distribution of total phosphors and total mercury in water and total mercury in a fish
prey species, with 95% confidence intervals. Note: The existing action levels are under
protective because ecological effects are observed from eutrophication and mercury
contamination
13
-------
Table 5. Preliminary Decision Rules for South Florida.
Decision Rules
la
If the South Florida Everglades Ecosystem can not be managed to achieve all desired
ecological uses, then a comparative ecological risk assessment shall be conducted to
determine which stressors, and their interactions, are placing the system at greatest risk.
lb
If the South Florida Everglades Ecosystem can be managed to achieve all desired
ecological uses, then the management, regulatory and control practices shall be
maintained.
Based on the results of this comparative risk assessment,
the following decision rules might be used:
2a
If phosphorus concentrations exceed a risk-based action level, then nutrient loads will be
reduced until phosphorus concentrations are less than the action level.
2b
If phosphorus concentrations are less than a risk-based action level, then BMPs and other
nutrient control programs will be maintained.
3a
If hydropattern modification varies by more than 10% from the desired natural
hydropattern rule curve, then the hydropattern will be modified to match the desired
natural hydropattern rule curve while maintaining flood control and water supply.
3b
If the hydropattern modification is within 10% of the desired natural hydropattern rule
curve, and flood control and water supply purposes are satisfied, then the hydropattern
management and operational programs will be maintained.
4a
If hydropattern modification varies by more than 10% from the desired natural rule curve
and either flood control and/or water supply requirements can not be satisfied, then
alternative flood control and water supply options will be investigated to return the
hydropattern to within 10% of the desired natural rule curve.
4b
If hydropattern modification can not be returned to within 10% desired natural rule curve
and achieve water supply and/or flood control demands, then a risk-based benefit/cost
analysis will be performed to determine which alternatives have the lowest benefit/cost
ratio and that use eliminated.
5a
If habitat alteration exceeds risk-based landscape action level metrics, then habitat
alternation, a benefit/cost analysis will be done to determine if this habitat alteration
including urban development or agricultural production, should be banned and habitat
restoration under taken.
5b
If habitat alteration is less than risk-based landscape action level metrics, then habitat
alteration will be permitted until these values are within 5% of the lower limit of the
action level.
14
-------
Table 5. Preliminary Decision Rules for South Florida (Continued).
Decision Rules
6a
If mercury concentrations exceed a risk-based action level, then mercury sources will be
controlled until mercury concentrations are less than this risk-based level.
6b
If mercury concentrations are less than the risk-based action level, mercury sources will
be controlled to ensure the action level is not exceeded.
7a
If hydropattern modification greater than 10% from the risk-based desired natural rule
curve aggravates mercury contamination of fish and wildlife, then the hydroperiod shall
be modified to achieve the risk-based action level.
7b
If the hydropattern modification less than 10% of the risk-based desired natural rule
curve aggravates mercury contamination of fish and wildlife, then a comparative risk
assessment and risk-based benefit/cost analysis shall be conducted to determine which
stressor places that system at greatest risk and has the lowest benefit/cost ratio; that
stressor then will be reduced.
8a
If nutrient loading exceeds the nutrient risk-based action level and aggravates mercury
contamination of fish and wildlife, then nutrient loading shall be reduced to achieve the
risk-based action level.
8b
If nutrient loading is less than the nutrient risk-based action level and aggravates mercury
contamination of fish and wildlife, then a comparative risk assessment and risk-based
benefit/cost analysis shall be conducted to determine which stressor places that system at
greatest risk and has the lowest benefit/cost ratio; that stressor then will be reduced.
9a
If habitat alteration exceeds risk-based landscape action level metrics and aggravates
mercury contamination of fish and wildlife, then additional habitat alteration shall be
banned and habitat restoration under taken.
9b
If habitat alteration is within the risk-based landscape action level metrics and aggravates
mercury contamination of fish and wildlife, then a comparative risk assessment and risk-
based benefit/cost analysis shall be conducted to determine which stressor places that
system at greatest risk and has the lowest benefit/cost ratio; that stressor then will be
reduced.
15
-------
Table 6. Water Constituents Ranges in South Florida.
Measurement Variable
Range
Minimum
Maximum
Physical Measurements
Site location (deg.)
Latitude
25.30
Longitude
80.22
Latitude
26.93
Longitude
81.13
Weather
Discharge, structure (m3/s)
0
Water depth (ft)
Marsh
0
Canal
0.5
Marsh
8
Canal
25
Temperature (°C)
18
36
Turbidity (NTU)
0.1
80
Chemical Measurements
Dissolved oxygen (mg/L)
0
15
Specific conductance (//S)
10
2150
pH (s.u.)
5.5
8.8
Total organic carbon (mg/L)
5
80
Total phosphorus (mg/L)
0.001
0.500
Sulfate (mg/L)
1.0
850
Total mercury (ng/L)
0.02
12
Methymercury (ng/L)
0.03
1.5
Alkaline phosphatase
0.01
8.0
Biological Measurements
Resource class (canal, sawgrass
marsh, cattails, etc.) (Numeric rank)
1
7
Periphyton presence/absence (1,0)
0
1
Chlorophyll a (fxgfL)
0
100
Periphyton total mercury Oug/kg)
4
600
Periphyton methylmercury (//g/kg)
0.08
25
Fish total mercury (//g/kg)
5.0
1000
16
-------
Table 6B. Soil/Sediment Constituents Ranges in South Florida.
Measurement Variable
Range
Minimum
Maximum
Physical Measurements
Peat depth (m)
0
>4.25
Bulk density (g/cc)
0.05
1.4
% Mineral content (%)
3%
99%
Ash free dry weight (%)
1.0
96.0
Redox potential (mV)
-250
+600
Chemical Measurements
Soil/Sediment total mercury (/ig/kg)
3.0
500
Soil/Sediment methylmercury
(m g/kg)
0.01
50
Soil/Sediment total phosphorous
Og/kg)
10
9000
Soil/Sediment sulfate (^g/kg)
20
850
17
-------
Ecosystem Assessment
Revised DQOs Mar 97
(2) Identify the decision errors and choose the null hypotheses
(a) Define both types of decision errors and establish the true state of nature for each
decision error. By convention, a Type I (false positive) error is rejecting the null
hypothesis when it is true. A Type II (false negative) error is not rejecting the null
hypothesis when it is false. The two types of decision errors for the Project are (I)
deciding the risk-based action level is exceeded when it truly is not, and (II)
deciding the risk-based action level is not exceeded when it truly is.
The true state of nature for decision error (I) is that the null hypothesis is true.
The true state of nature for decision error (II) is that the null hypothesis is false.
(b) Specify and evaluate the potential consequences of each decision error. The
consequences of deciding the risk-based action levels are exceeded when they
truly are not (decision error I) means there will be increased control costs
associated with nutrient and mercury source reduction, restricted urban and
agricultural development, habitat restoration, and restricted hydropattern
modification around the natural hydropattern rule curve, which could result in
flood damage or water supply shortages.
The consequences of deciding the risk-based action levels are not exceeded when
they truly are (decision error II) means that ecological restoration of the South
Florida Everglades ecosystem will not be successful.
(c) Establish which decision error has more severe consequences near the action
level. Based on current laws and regulations related to the South Florida
Everglades ecosystem (e.g., Everglades Forever Act), the decision II error has the
more severe consequences near the action level because of the risk to both
ecological and human health and ecological restoration. However, this
consequence must be based on a comparative risk assessment and a risk-based
benefit/cost analysis of the risks and impacts. The economic consequences are in
the billion dollar category for both types of decision errors.
(d) Define the null hypothesis (baseline condition) and the alternative hypothesis and
assign the terms "false positive " and "false negative " to the appropriate decision
error. Null hypotheses for DOQs are not equivalent to experimental null
hypotheses for statistical testing. Null hypotheses for DQOs reflect the decision
error that has the most adverse potential consequences. The DQO null hypothesis
is equal to the true state of nature that exists when the more severe decision error
occurs. The null hypotheses for this Project, therefore, would be:
18
-------
Ecosystem Assessment
Revised DQOs Mar 97
H0 = The comparative ecological risk assessment indicates the interactions among
stressors puts the South Florida Everglades ecosystem at risk.
H0 = The risk-based action levels for nutrient concentrations are exceeded.
H0 = The risk-based action levels for mercury concentrations are exceeded.
H0 = The risk-based landscape action level metrics are exceeded.
H0 = The risk-based action levels for hydropattem modification exceed by X%
the natural hydropattem rule curve.
A "false positive" has the greatest consequences for each of these hypotheses.
(3) Specify a range ofpossible values of the parameter of interest where the consequences of
decision errors are relatively minor (gray region) - The purpose of this research project is
to determine the action level values. Until these action levels are defined, it is not
possible to specify actual numeric values to an area of minor importance. It is, however,
possible to indicate these areas of minor importance will be at the extremes of the
distribution. In this portion of the action level curve, there will be a low probability of
making either type of decision error.
(4) Assign probability values to points above and below the action level that reflect the
tolerable probability for the occurrence of decision errors. - The QA G-4 Guidance
manual indicates the gray region where greater tolerable errors are permitted are around
the action level, with lower tolerable errors around the extreme values. The planning
team disagrees with this concept. The greater tolerable errors are permitted at the
extemes of the distribution because it is unlikely that large errors in the metric would alter
the conclusion that the action level was either exceeded or not exceeded. However, near
the action level, particularly as values approach the lower limit of the action level,
decision errors can have significant consequences on subsequent actions (Figure 4).
Tolerable error around the action level in this region should be no more than 10%.
Optimize the Design - The REMAP monitoring design for South Florida was revised to provide
more resource-effective information at reduced cost without compromising the DQOs for the
marsh samples. The canal samples will be selected as subjective samples because the first four
sampling cycles indicated the marsh processes are more significantly affected by multiple
environmental stressors than canal processes in South Florida. Sufficient samples will be
collected in the canals so that temporal and spatial trends can be detected over time in loadings to
the marsh, but not with the same precision as marsh trends. This improved design will permit
answering the original seven policy-relevant questions guiding the Project and establishing the
action levels needed for decisions.
Appendix A contains statements for data representativeness, completeness, comparability,
precision and accuracy for each of the constituents measured in the EPA Region 4 South Florida
Ecosystem Assessment Program. These quantitative DQO criteria will be revised as additional
data become available to the program.
19
-------
90
80
70
60
50
40
30
20
10
0
4.
I I I I I I I i
100 200 300 400 500 600 700 800 900
Prey species Hg, ug/kg
ised mercury Action Level for predator prey species identifying The Gray Regions and area for 10%
Rates.
-------
Appendix A
Data Quality Objective Criteria
-------
Table A1A Water Data Quality Objective Criteria.
Measurement Variable
Representativeness
Completeness
Comparability
(Split Samples
SOPs, Std.
Units)
Field Precision
3 o
Duplicate Precision
% RSD @ 95%
Accuracy
(% Spike
Recovery)
Canals
Marsh
Canals
Marsh
Physical Measurements
Site location (dcg.)
Design-based
statistically
representative
90%
SOPs
NA
NA
Weather
<<
SOPs
NA
NA
NA
NA
NA
Discharge, structure (m3/s)
u
NA
NA
NA
NA
NA
Water depth (m)
a
SOPs
±15.0
±3.3
NA
NA
Temperature (°C)
SOPs
±5.06
±9.05
NA
NA
±0.15
Turbidity (NTU)
-------
Table A1 A. Water Data Quality Objective Criteria (Continued).
Measurement Variable
Representativeness
Completeness
Comparability
(Split Samples
SOPs, Std.
Units)
Field Precision
3 o
Duplicate Precision
% RSD @ 95%
Accuracy
(% Spike
Recovery)
Canals
Marsh
Canals
Marsh
Alkaline phosphatase
<<
(<
SOPs
±3 74
±3.48
±75.21
±61.18
70-130
Biological Measurements
Resource class (canal,
sawgrass marsh, cattails,
etc ) (Numeric rank)
(<
<(
SOPs, FWS,
NW1
NA
NA
NA
NA
NA
Periphyton presence/absence
(1,0)
a
<<
SOPs
NA
NA
NA
NA
NA
Chlorophyll a (yUg/L)
a
(«
SOPs
±22.55
NA
±191.48
NA
NA
Periphyton total mercury
0"g/kg)
ct
it
SOPs
NA
±487.87
NA
±225.95
70-130
Periphyton methlmcrcury
(Mg/kg)
<(
i<
SOPs
NA
±9 51
NA
±115.59
70-130
Fish total mercury (ug/kg)
n
(t
SOPs, Reg. 4
±257 42
±397.35
±159.36
±89.95
70-130
* Actual Units
RSD. Relative Standard Deviation
SOP' Standard Operation Procedures
-------
Table A1B Soil/Sediment Data Quality Objective Criteria.
Measurement Variable
Representativeness
Completeness
Comparability
(Split Samples
SOPs, Std. Units)
Field Precision
3o
Duplicate Precision
% RSD @ 95%
Accuracy
(% Spike
Recovery)
Canals
Marshes
Canals
Marshes
Physical Measurements
Peat depth (m)
••
¦¦
SOPs
NA
NA
NA
Bulk density (g/cc)
"
SOPs
NA
±0 77
NA
±35.56
NA
% Mineral content (%)
"
••
SOPs
NA
±93.82
NA
NA
NA
Ash free dry weight (%)
"
SOPs
±51 84
±93.82
±60.54
±52 79
NA
Redox Potential (mV)
••
-
SOPs
±386
±308
NA
NA
Chemical Measurements
Soil total mercury (/ig/kg)
••
¦¦
SOPs, USGS
±249 1
±196 9
±70 2
±89 5
70-130
Soil mcthylmcrcury (/^g/kg)
••
"
SOPs, USGS
±32 6
±3 87
±383 3
±255 8
70-130
Soil sulfate O^g/kg)
"
••
SOPs, USGS
NA
±1412
NA
±107
Total phosphorus in soil O^g/kg)
«•
"
SOPs, USGS
±3904.00
±741.64
±108.54
±72.98
75-125
RSD Relative Standard Deviation
SOP Standard Operation Procedures
NOTE: Use largest standard deviation when two measurements of a parameter-top and bottom, F1U & BATTELL, floating and soil periphyton
-------
APPENDIX C
Summary of Data Review Findings
-------
EXECUTIVE SUMMARY
The EPA Region 4 South Florida Everglades Ecosystem Assessment Project is a
pioneering research and monitoring effort to assess mercury contamination, eutrophication,
habitat alteration, and hydroperiod modifications in the South Florida Everglades ecosystem
using the ecological risk assessment process. Over a four year period, this Project made over
20,425 measurements on 20 constituents in 5 different media at about 200 canal stations,
550 marsh stations during the 9 field sampling events over the 10,000 km2 area in South Florida.
Included in these measurement were 7 structures sampled on a bi-weekly basis from February 94
thru February 97. The large scale patterns in total phosphorus and eutrophication, mercury
contamination, and habitat types, observed for the first time because of this Project are described
in the 1998 South Florida Ecosystem Assessment Technical Report, Volumes I and II (Stober
et al., 1998). To accommodate this large sample load, and to analyze ultra-trace level
concentrations of selected constituents, new methods also were developed as part of this Project.
In addition to the quality assurance reviews that occurred as part of the data analysis, a rigorous
quality assurance/quality control review was conducted on the data by scientists of the Region 4
Science and Ecosystem Support Division (SESD), Office of Quality Assurance (OQA).
OQA was requested to review all contractor data provided by laboratories supporting the
SESD Everglades ecosystem project. Raw data were obtained from each laboratory involved and
were reviewed against the four levels of criteria addressed in: 1. the Data Quality Objectives
(DQOs), 2. requirements stated in Section 2.5; Data Quality Requirements in the EPA Project
Research Plan, 3. the laboratories' respective Standard Operating Procedure (SOP) and 4. quality
control (QC) practices, and procedures typically applied for regulatory and compliance reviews
conducted by the OQA. The OQA goals were:
• to reconstruct the final reported data for the field, laboratory and QC activities,
• to determine if the contract laboratories' documentation was accurate and
provided adequate defensibility of the monitoring data, and
• to verify if the data were of acceptable quality based on the Project design.
Review findings are documented in this Summary of Data Review Findings for each constituent
by media and by cycle and applicable recommendations are included.
Quantitative and qualitative evaluations indicate the vast majority of the data were of
acceptable quality, and met the overall DQOs of the Project design. All data sets will have some
QA/QC deficiencies and certain QC procedures and documentation deficiencies also were noted
in the Project data. Out of the over 20,425 measurements made over the four year period, only
15 percent of the data were qualified. There were 1,325 points flagged "NR"- not reviewed
because raw data were not available from the laboratory, 1,783 assigned a "J" flag - analyte
identification was acceptable for use but certain QC criteria were not satisfied and 256 values
rejected and not reported in the database.
The data that have been qualified after this rigorous review are:
C-l
-------
J" - Soil/sediment methyl mercury for the secondary laboratory
NR." - Water total mercury for September 1993 Canal samples and April '94
transect data for primary laboratory
J" - Water total mercury, excluding September 1993 Canal, for primary
laboratory.
"J" - Periphyton total mercury for primary laboratory
"J" - Soil/sediment total mercury for May 1994, 1995 Canal cycles and May
1996 marsh samples for secondary laboratory
"NR." - Soil total mercury for September 1995 marsh split samples for secondary
laboratory.
"J" - Mosquitofish total mercury for very low mercury concentration fish for
primary laboratory
"J" - Total nitrogen for the secondary laboratory for Sept. '96 marsh samples
Reject - Water alkaline phosphatase activity for 123 samples for May 1996 marsh
cycle from the primary laboratory
Reject - Water sulfide for 123 samples for the September 1995 marsh cycle from
the primary laboratory.
The Study was designed to have a minimum of 10 percent of the samples split for certain
parameters with a second laboratory to provide between laboratory comparability and field
duplicates to provide a combined field and laboratory precision. Samples for certain parameters
were split at a higher frequency than 10 percent. The higher frequency of split sample data
provided strong analytical support for data analysis and interpretation even though
documentation or record retention may have been lacking in some cases.
The OQA finds that the data provided by all laboratories (EPA and supporting) permit
assessment of trends and the characterization of the Everglades on a spatial and temporal basis
and satisfy the Project objectives. Data qualified with a "J" code would not compromise the
characterization and trend analysis or Project objectives. "J" coded data should not be used solely
in the setting of environmental standards unless additional data are collected to substantiate these
original values that has all supportive QC and documentation. These data can be used as part of
the weight of evidence approach for establishing environmental standards.
This Summary of Data Review Findings report should be kept in perspective. The South
Florida Everglades Ecosystem Assessment Project was a research and monitoring effort, not a
regulatory compliance project. It developed and implemented new methods for rapidly analyzing
a large sample load with ultra-trace level constituent concentrations. The QA/QC review,
however, included rigorous QA/QC methods used in reviewing compliance monitoring data. Yet,
out of the over 20,425 measurements made over the four year period, only 15 percent of data
were qualified.. This data set should provide an important baseline for assessing the current
conditions of the Everglades ecosystem and for monitoring future trends in this condition. The
results from this QA/QC Review have already been, and will continue to be, used to refine the
QA/QC procedures in future Project monitoring efforts.
C-2
-------
INTRODUCTION
In 1993, EPA Region 4 Science and Ecosystem Support Division (SESD) initiated the
South Florida Everglades Ecosystem Assessment Project. This pioneering 10,000 km2 system
wide research and monitoring project was conducted to assess mercury contamination,
eutrophication, habitat alteration, and hydroperiod modification using an ecological risk
assessment process. A statistical survey design was used to select 200 canal sites and 500 marsh
sites to sample for 20 constituents measured in 5 different media during 4 different periods or
cycles. In addition, transect sampling was conducted to test the marsh sampling procedures,
constituents were measured on a bi-weekly basis at 7 structures in the South Florida system, and
new methods were developed to accommodate the large sampling load and measure ultra-trace
level constituent concentrations. Over a four year period, this project made over 20,425
measurements in the South Florida Everglades ecosystem.
Although QA/QC procedures were designed into the Project from its onset, a Quality of
Science Review was requested from the EPA Office of Research and Development (ORD)
National Exposure Research Laboratory (NERL) Environmental Sciences Division (ESD)-Las
Vegas, in 1995. ORD was, and continues to be, in the process of trying to develop a Data Quality
Objectives (DQO) process for research projects similar to the DQO process developed for
regulatory and compliance programs (EPA 1994). This has proven to be a formidable task. The
purposes of the NERL-ESD review were to (1) identify quality-related issues and provide
recommendations for correction or improvement, and (2) to provide the project participants with
the necessary tools to enable them to continue to monitor data quality. The Quality of Science
Review report (Chaloud, Heitmar, and Birch 1996) was provided to SESD in April 1997. The
Quality of Science Review report identified a number of strengths in the Everglades Mercury
Study. Strengths of this project include:
• Project Design
• Personnel
• Field Operations
• Sampling Procedures
• Low Crew Variability
• Project Management
• Sample Representativeness
• Laboratory QA/QC Programs
• Low Error Rate in data Base
• Spatial Analyses
Recommendations suggesting positive corrective actions were commenced or
implemented during the on-site audits in 1995 and continued during successive sampling events
and data evaluation by the OQA and the laboratories involved. Three areas were specifically
mentioned in the Review related to data quality. These were
• Data Package Issues
• Data Validation/Verification
• Total Mercury in Water Data
C-3
-------
The Review recommended:
• that all raw data provided by contract laboratories through the Interagency
Agreements with the U.S. Department of Energy and the U. S. National Park
Service should be obtained and maintained on file at EPA,
• validate or verify the quality of data generated,
• use flag fields to indicate validation/verification results, and
• document validation and verification criteria in metadata files or in a user's guide.
• recover additional sulfate data
As a result of the recommendations made in the Quality of Science Review, the SESD
Office of Quality Assurance (OQA) and Region 4's Environmental Service Assistance Team
(ESAT) conducted a data review of all available contractor acquired data for the Everglades
Assessment Project.
In addition to the above recommendations, the OQA also conducted selected reviews of
other raw data collected/analyzed by EPA staff and EPA's ESAT contractor. The goal of the
audit of data quality was to provide a quantitative and qualitative evaluation of the
documentation and procedures associated with environmental measurements to verify that the
resulting data were of acceptable quality.
Data Collection Activities
The gathering of data for review was conducted in two phases. First, the OQA and ESAT
Work Team conducted a preliminary review of the data submitted by the Florida International
University (FIU) Southeast Environment Research Program (SERP) laboratory for sampling
events conducted in 1993, 1994, 1995, and 1996. Randomly selected raw data packages were
obtained by the OQA and ESAT Team during a visit to the SERP laboratory on March 12-14,
1997. Total and organic mercury raw data packages were also requested and received in March
1997, from Battelle Marine Sciences Laboratory (Battelle), Sequin, WA. AmTest Analytical
Service, Inc., Seattle, WA, a subcontract laboratory for Battelle provided quality control (QC)
split sample data for total nitrogen and phosphorus for the Marsh Cycle 3 sampling event. Raw
data from AmTest were obtained in March 1998.
Since the Quality Assurance Project chemist participated in the ORD Quality of Science
Review, and was part of the data review Team, an additional on-site visit was not made to the
Battelle laboratory. The AmTest laboratory was not visited since only 26 samples were analyzed
during the last sampling event of 1996.
After an initial assessment of the raw data packages, the second phase of the OQA/ESAT
audit included two subsequent visits to the SERP on May 14-15, and October 14—15, 1997 to
obtain additional raw data for the nine sampling events. The visits were necessary to fulfill the
recommendation of Quality of Science Review Team that all raw data should be resident at
Region 4 SESD and to allow the SERP personnel time to gather the large volume of raw data. A
C-4
-------
third visit was made during June 1998 to review data stored in electronic format for methyl
mercury at the SERP laboratory. During each visit members of the data review Team discussed
the methodologies used and observed the SERP staff conducting the on-site analyses.
Data Review Guideline
Data were reviewed using data review practices employed by the OQA and the ESAT
Team for regulatory and compliance monitoring programs with standard methodology. These
practices included, but were not limited to, such items as sample collection, receipt, storage,
checks of holding time, preparation, analysis, reporting, and all associated QC. Preparation
checks included the use of blanks (reagent, digestion, lab and field blanks) for background
correction, use of calibration standards and application of linear regression analysis of calibration
curves for linearity, utilization of standard reference materials (SRMs) or performance evaluation
samples, use of matrix spikes for accuracy, analysis of both field and laboratory duplicates for
precision statements and determination of detection limits. Data packages also were reviewed
against criteria addressed in the Data Quality Objectives (DQOs) and quality assurance
requirements stated in Section 2.5; Data Quality Requirements in the EPA Project Research Plan;
against SERP's Standard Operating Procedures (SOP) for Total and Methyl Mercury analyses;
SERP's Comprehensive Quality Assurance Plan for the Mercury Laboratory dated 7/27/93, and
subsequent revisions dated 6/14/94 and 4/18/96; and SERP's Methods Validation for
Micromolar Concentrations of Total Nitrogen in Natural Water dated 4/25/96. The latter
included total phosphorus and total organic carbon (TOC) SOPs. Because method development
was a Project objective, the SERP staff expended considerable time in developing and refining
methods and modifying the SOP's. These modifications were necessary as procedures were
continually improved to enhance the quality of the data as the project moved from a
research-orientation to a research-production environment. The same data review procedures
were employed for the other participating contractor laboratories.
All data entries in the EPA database were checked against the raw data sheets from all
laboratories for completeness and correctness. Relatively few transcription errors were observed
considering the vast amount of data provided. Corrections were provided to the Project Leader.
Data Review Qualifiers
Reported data that did not satisfy all the criteria specified in the Data Quality Review
Guidelines were qualified with appropriate codes. Each parameter that has data qualifier codes
assigned has remarks added to assist the reader in understanding the code and the reason(s) for the
qualifier or data flag. A list of qualifier codes accompany each parameter or group of similar
parameters. The codes were:
C-5
-------
Code
Definition of analytical data qualifiers
NR A "NR." Qualifier Code indicates that these data were not reviewed, since raw data were
not available from the laboratory. In these cases, EPA data reviewers were not able to
reconstruct the QC and analytical activities.
"U" The analyte was analyzed for but not detected. The value preceding the "U" is the
"minimum quantitation limit (MQL)".
***Minimum Quantitation Limit (MQL) - Every sample has a concentration level below
which the variance of the results for a particular analyte (element or compound) exceeds
the acceptable quality control criteria. This level is the MQL and is reported as the value
preceding the "U". The MQL is determined from sample size, dilution required, and
instrument sensitivity. The value often varies from analyte to analyte within a sample.
Analytes are often detected at levels below the MQL and are reported as estimated values
(J). Generally, analytes identified below the MQL will only be reported if concentration is
greater than one tenth of the MQL.
"J" The identification of the analyte is acceptable for use, but certain QC criteria were not
satisfied.
***Estimated Value-Every sample analysis has quality control criteria associated with the
quantitative data which have been established based on similar analyses. When these
criteria are exceeded, the value for that analyte or similar analytes is flagged. Examples
are:
(1) calculated values were below or above an appropriate linear range
(2) calculated values were below the MDL or PQL of an analyte
(3) analytical holding times for analysis were exceeded
(4) surrogate recovery limits were exceeded
(5) some QC criteria were not documented or followed as specified in the laboratory's
SOP or requirements necessary for confirmation were not met.
"A" The analyte was analyzed in replicate. The value preceding the "A" is an "average value"
of the replicates.
~~~Average Value-Samples were often analyzed in replicate (usually in duplicate).
Aliquots of the same sample were analyzed and the values are averaged. Sometimes
replicate samples were analyzed and the values were reported as an average.
Specific data qualifier codes and explanations are provided with each constituent by media
and by cycle and the associated data.
C-6
-------
Summary of Data Review Findings
Summary results are presented for each constituent by media and by sampling cycle that
data qualifier codes were assigned. Applicable recommendations are included. If there were no
qualifying codes assigned, the data met the data quality review guidelines used to evaluate the
data. Primary and secondary laboratories were used for sample analysis. The primary laboratory
was the lead laboratory for the analysis with the secondary laboratory providing QA/QC on
duplicate and split samples. The goal was to have QA/QC samples for about 10 percent of the
total number of samples analyzed so the secondary laboratory had a significantly lower sample
load.
1. Total and Organic Mercury in Soil/Sediment and Water by Battelle
Battelle served as the primary laboratory for methyl mercury in water and provided
backup QC for total mercury in water splits/duplicates throughout the study and methyl mercury
in soil/sediment after the Sept. 93 Canal sampling. The Battelle total and methyl mercury data for
water and soils were reviewed for accuracy and compliance according to the same criteria used for
SERP total and methyl mercury analyses. Linear regression was checked using the standards from
the calibration curve. Quality control check samples were evaluated for recovery of the analyte
from spiked samples and relative percent difference of duplicate samples. The final results were
checked to ensure that the results were converted accurately from the instrument with the
appropriate units. For example, the results for soil samples are converted from pg/ml to ng/g. Less
than 5 percent of the methyl mercury samples for water were flagged "J" due to the quality control
checks being outside of the acceptable limits: matrix spikes outside DQR limits (70 percent to
130 percent), duplicate samples results relative percent difference greater than 30 percent, or the
calibration curve correlation coefficient less than 0.995.
After May 94, Battelle switched from an extraction to a steam distillation separation for
soil/sediment samples. Matrix spikes were conducted and recoveries were acceptable. Duplicate
splits with SERP were comparable with Battelle reporting values that tended to be higher than the
SERP data. It has been reported that steam distillation could produce a methyl mercury artifact in
samples containing elevated concentrations of total organic carbon (TOC) and total mercury
(Hintelmann 1997). However, another study (Bloom 1997) using Everglades water and peat
indicated this artifact would not be significant in Everglades samples. Methods for measuring
ultra-trace level total and methyl mercury concentrations in water and soil/sediment are in the
research phase. There currently is no standard EPA method for these constituents at these
concentrations. Therefore, Battelle QA/QC soil/sediment data after May 94 should not be used
solely in the establishing trends or environmental standards unless additional collaborative data
are available to substantiate these original values
2. Total Mercury in Water by SERP
SERP served as the primary laboratory for the approximately 900 total mercury samples in
water. These data were reviewed and validated. The September 1993 Canal and April '94 transect
C-7
-------
data for total mercury were flagged as "NR.," not reviewed. The reason was the raw data were
unavailable for review from the laboratory.
Data also were reviewed against QC checks EPA considered as good laboratory practices
for regulatory or compliance monitoring programs with standard EPA methods. An independent
stock standard prepared from another source to verify the initial calibration was not included in the
analysis, but it was not part of SERP's SOP. The lack of this independent calibration check would
not change the values obtained but would have strengthened the quality of the data. However, no
data were flagged because SERP followed its SOP and the independent stock standard was not part
of the initial QA plan reviewed by EPA. The independent stock solution will be added to future
sample analyses.
The SERP data prior to May 1996 did not include a matrix spike. The matrix spike
strengthens the data in that it shows that the analyte is actually being recovered during the digestion
of the samples. According to SERP CompQAP, the accuracy of the matrix spike should be
between 90-110 percent recovery or the sample analyses must be repeated. In May 1996, the data
includes a water spike at 2 ng/1. Six of sixteen data sets had recoveries ranging from
83-122 percent with an average for all sets of 103 percent. In September 1996, the data included a
sample (matrix) spike at 2 ng/1 with 9 of 17 data sets having recoveries of 49-138 percent. The
average percent recovery of this matrix spike for all 17 data sets was 94 percent recovery.
However, none of the samples analyzed with the spike recoveries outside of SERP's stated
performance limits were re-analyzed.
Although reagent blanks were analyzed, the sample results did not include a consistent
reagent blank correction prior to May 1996. Just prior to the May 1996 sampling event, the SERP
staff instituted a correction method that accounted for the background contribution from reagents
and glassware. Correction for any contribution by the reagents was made utilizing varying
concentrations of the reagents added to mercury free water and calculating the response due to the
reagents using linear regression. Data previous to May 1996, were corrected for any reagent
contribution by using an average value (0.5 ppt) for the reagent blank correction. The reagent blank
corrects for any contamination due to the reagents and could alter the result at the lower
concentrations if the contribution was significant. Since raw data for the first two sampling cycles
were not available for review, a statement could not be made about the quality of the data.
However, the SERP staff did have an in house policy of checking for background contamination
and if background contamination was determined to be significant, samples were to be re-analyzed.
Recalculation of the data using a four-point curve and subtracting the reagent blank showed
differences of as much as 1.0 ng/L in only 4 out of 552 cases prior to May 1996.
A detection limit for total mercury was established by SERP at 0.3 ng/L using the
recommended procedure for establishing detection limits given in 40 CFR 136 Appendix B. A
low-level 0.3 ppt standard was not analyzed to show this level could be achieved each day. To
provide a day-by-day assessment of the method stability near the MDL, a practical quantitation
limit (PQL) for total mercury was established by the OQA data review Team at 1 ppt., (three times
the standard deviation of 21 readings from the 2 ppt spike and reagent blanks). Results below
1 ppt. should be used with caution, especially since the lowest calibration standard reported as
C-8
-------
analyzed during the study was 2.5 ppt. The precision for field duplicates below 1 ppt ranged from
0-118 percent relative standard deviation (% RSD equals the standard deviation for duplicates
divided by the mean time 100) with an average of 28 percent. Based on the review findings, the
total mercury data in water by SERP were assigned a "J" flag except for the September 1993 canal
data assigned "NR.". The "J" flag indicates the data are acceptable for use, but did not satisfy
certain QA/QC criteria. The SERP data below 1.0 ppt should not be used solely for setting
regulatory limits based on field precision and blank correction.
QC data for total mercury in water from split sample QC data provided from the Battelle
laboratory showed a positive bias compared with data submitted by the SERP. However, the bias
between the laboratories diminished during the May and September 1996 sampling events. The
quality of the total mercury data were strengthened in two ways: 1) the observed collaborating
trends for split samples between the SERP and Battelle laboratories; and 2) consistent performance
by SERP and Battelle was observed on performance evaluation samples supplied as unknowns by
the Project Quality Assurance Chemist to each laboratory. (See Table 1). Continued performance
evaluation corroboration is expected to improve as mercury methods and analytical procedures
continue to be refined.
Table 1 Total Mercury in Water Performance Evaluation Data (ppt)
Date of Analysis
P.E. Value
SERP
Battelle
EPA
05/12/96 WP 30 CI
0.98
1.44
1.51
1U
05/13/96 WP 30 C2
2.10
1.73
2.30
2.20
05/14/96 WP 30 CI
6.23
5.71
6.14
6.20
09/15/96 WP 33 CI
5.81
5.02
5.73
5.60A
09/18/96 WP 33 CI
2.71
2.21
2.87
2.30
09/19/96 WP 33 CI
10.16
8.87
10.40
10.00
09/20/96 WP 33 CI
1.16
1.34
1.26
1.0UA
3. Total Mercury in Periphyton by SERP
SERP was the primary laboratory for the periphyton samples. QC check samples (SRM)
were analyzed by SERP with each set of total mercury in periphyton analyses to verify the accuracy
of the calibration standards and validate the digestion method from an independent source. SERP's
acceptable range for these check samples was 90-110 percent recovery of the true value of the
sample (60 ng/g). In 6 out of 16 data sets, this percent recovery was not achieved. The range of
recovery was from 68-135 percent. The average recovery was 101 percent for the 16 sets.
C-9
-------
A reagent blank was included in the analyses, but was not used to correct for reagent
contamination in the final calculation of the sample results. The reagent blank correction versus no
reagent blank correction showed a considerable difference (130 percent RPD) in sample results
where peak heights were less than two times the peak height of the reagent blank. This impacted
greater than 50 percent of the samples for April 1995 and May 1996, but only 10 percent of the
samples in September 1996. These sample values should be recalculated to remove the
contribution for the blank or the detection limits should be increased. A check standard of 500 ng/1
prepared from the same source was used for verification. Therefore, all total mercury data for
periphyton were given a "J" code. The data are acceptable for use, but did not satisfy certain
QA/QC criteria. The FL DEP recommends percent recovery ranges of 75-125 percent for
biological samples, rather than 90-110 percent. These changes would be implemented in future
sample analyses.
4. Total Mercury in Soil/Sediment by SERP
EPA-SESD served as the primary laboratory and SERP as the secondary or QC laboratory.
A soil SRM was analyzed by SERP with each set of total mercury in soil/sediment analyses to
verify the accuracy of the calibration standards and validate the digestion method from an
independent source. SERP's stated acceptable range for this check sample (60 ng/g) was
90-110 percent recovery of the true value. In the May of 1994 canal data, this check sample did not
meet SERP's acceptance limits. A reagent blank was included in the analyses to correct for reagent
contamination in the final calculation of the sample results in all the data except in the May 1995
canal data. The results for these data were "J" coded since the recoveries did not meet the SERP's
stated limits. The check standard was the high standard, at 500 ng/1. The September 1995 marsh
split samples were coded "NR" because raw data were not available for review. However, data
compared favorably between EPA-SESD and SERP values on split samples. The average
concentration reported for the splits for EPA was 117 ng/gm and for SERP 156 ng/gm with a
average difference of 39.6 ng/gm. Both labs had comparable %RSD of 13.7 and 15.2, respectively.
5. Total Mercury in Fish by SERP
SERP was the primary laboratory for total mercury in fish analyses. Whole mosquitofish
(Gambusia) were digested and analyzed for total mercury. When the recovery QC criteria for the
SRM did not meet acceptance limits, the analysis could not be repeated because the entire fish was
digested.. SRMs were included within the analyses to verify the accuracy of the data during the
course of the analytical runs. For the September 1993 canal data, and one data set at the end of
September 1994, SRMs were not analyzed. SERP set acceptance limits of 90-110 percent recovery
of the true value for the SRM. The true value of the oyster tissue SRM was 64 ng/g and that of the
dorm standard (Dog fish) was 4640 ng/g. The average percent recovery of the SRM (oyster std.) in
September 1994 was 248 percent, and in April 1995, 287 percent. With the 64 ng/g oyster
standard, there was not an adequate detector response following dilution, which contributed to the
large variances for percent recovery. Four factors contributed to the wider range of recovery in the
samples. First, the 64 ng/g SRM used did not provide an adequate detector response following
dilution. Therefore, subsequent samples were run using the 4,640 ng/g SRM. Second, the inherent
variability in biological samples results in larger percent recovery ranges. The FL DEP
C-10
-------
recommends percent recovery ranges of 75-125 percent, rather than 90-110 percent. Third there
might have been either an interference (effervescing due to C02 production) and/or fourth some of
the SRM weights were documented incorrectly based on conversations with the analysts.
Corrections for the elevated recoveries were not warranted due to the stated problems. SERP
switched to the dorm standard because of sample handling problems. The average percent recovery
for the Marsh sampling events were 92 percent for the April 1995 (range 77-99 percent),
98 percent for May 1996 (range 45-129 percent), and 80 percent in September 1996 marsh (range
45-100 percent).
Reagent blanks were included in the analyses, but were not used to correct for reagent
contamination in the final calculation of the sample results. Reagent or glassware contribution in
the blank impacted 110 samples analyzed from September 1994 to 1996. These sample values
should be recalculated or the detection limits raised. Reagent blanks were not analyzed for
46 samples in September 1995 and 16 samples in September 1996. These data (172 of 704) were
"J" flagged for no reagent blank correction. Matrix spikes were not analyzed because of the
recognized difficulty to get a homogenous spike and the whole fish was used.
6. Organic Mercury in Soil/Sediment and Water by SERP
SERP served as the QA/QC laboratory for the Sept. '93 samples and as the primary
laboratory for organic mercury in soil/sediment after Sept. '93. Battelle served as the primary
laboratory for Sept. '93 but became the QC laboratory for solid/sediment duplicates/splits after this
date. Samples and matrix spikes were analyzed in duplicate and results were corrected for recovery
based on the matrix spike. SERP's policy in place during the time period of 1993-1996 was to
store the electronic copies of the methyl/ethyl mercury chromatograms. No hard copies of the
GC chromatograms were stored. Electronic chromatograms were only available for Sept. '96 for
review. Selected analytical runs (electronic copies) were reviewed by the Team during the June
1998 visit to SERP. The Team was able to confirm qualitative and quantitative steps taken by the
analysts for the selected data (soil/sediment) sets reviewed. Split water samples with the Battelle
laboratory showed acceptable agreement between the field replicates for water samples even
though the methods were different. The comparison for soil/sediment samples analyzed by both
laboratories showed SERP data having lower average concentration (0.63 ng/gm) from samples
collected at the same site than the Battelle average (0.96 ng/gm). Table 2 lists the comparison for
the methyl mercury data between the two laboratories. There were some Canal samples that SERP
had much higher concentrations than data reported by Battelle. This would indicate if there were a
artifact for the distillation procedure it was not a consistent phenomenon. None of these data were
qualified.
C-ll
-------
Table 2 Precision Data for Methyl Mercury
Parameter
Mean
%RSD
n
Methyl Mercury - (Battelle) H20 ppt
0.408"
14.68
68
(0.39)b
(14.9)
(21)
Methyl Mercury - (SERP) H20 ppt
0.2 lb
33.7
21
Methyl Mercury - (Battelle) Soil ng/gm
0.96c
-
62
Methyl Mercury - (SERP) Soil ng/gm
0.6C
-
62
a = mean of duplicate pairs
b = mean of duplicate pair from same distribution
c = average concentration for samples analyzed by the respective laboratory from the same location.
7. Total Phosphorus by SERP
The total phosphorus data in soil and water were reviewed for accuracy and compliance
with the SERP SOP. SERP followed the procedures as outlined in their SOP with some
documentation deviations. The laboratory did deviate from EPA's common laboratory practices in
preparing the calibration curve. SERP ran a five-point calibration curve and if the correlation
coefficient for this curve was greater than 0.995, indicating good linearity, the curve was
recalculated as a two-point curve using the high standard and the origin. Values were recalculated
by the review Team using all four points and there was no significant difference in the SERP
reported values and the EPA calculated results. SERP determined a MDL of 0.6 ^g/L using
Appendix B to Part 136 of 40 CFR. A MDL of 1.4 pig/L was calculated by the OQA/ESAT Team
from blanks and the lowest standard analyzed over a period of time, which corresponds well with
the SERP MDL. SERP's practical quantitation limit (PQL) was stated to be 1.8-3.0 ug/L or
3-5 times the MDL. There were nine water values that had concentrations between 1.8-3.0 (tg/L
and one value below the stated PQ1. All soil sample results were well above the calculated PQL 2.2
Aig/gm, and no detection limits were changed.
Total phosphorus in water precision data from field replicates collected during the study
from the canals and marsh show an average %RSD of 16.7 for the concentration range of
1.3-361 yag/L. Total phosphorus in soil/sediment precision data from field replicates from the
canals and marsh show an average %RSD of 13.5 for the concentration range of 44-2452 Mg/gm.
Split sample data between SERP and AmTest for the Sept. '96 sampling event showed
acceptable agreement between the two laboratories. AmTest reported a MDL of 5.0 as
compared to a lower MDL of 0.6 ngfL for SERP. No data were qualified.
8. Total Organic Carbon by SERP
Data obtained for TOC were reported from an instrument that operated in the concentration
only mode. The output of the instrument was in concentration units that were entered directly into
C-12
-------
the spreadsheet. It was not possible to check the actual calculations used to generate the results for
this parameter. Calibration curves and standards were checked and found to be in order. Therefore,
the TOC data were not qualified.
9. Total Nitrogen Data for Water by SERP and AmTest
SERF total nitrogen analyses (TN) were performed using an experimental method
developed with a ANTEK Model 7000N elemental analyzer by direct aqueous injection of the
samples. Results were calculated using a 2.0 mg/L KN03 standard. No SRM nor additional
standards such as glycine, urea or another stock source were analyzed during the analytical run.
Since this was a direct injection a reagent blank was not necessary. Raw data for quality control
split samples from AmTest were reviewed. Comparison of data between SERP and AmTest
showed SERP values to average 0.33 mg/L higher. However, the SERP method measured total
nitrogen and AmTest measured total Kjeldahl nitrogen (TKN). The results for AmTest would be
lower since nitrogen as nitrate-nitrite would not be determined but would be included in the SERP
data. The AmTest nitrogen data should be reported as TKN instead of TN. No SERP TN data were
qualified. The AmTest data were qualified based on reporting as TN instead of TKN.
10. Alkaline Phosphatase Data by ESAT
The Data Review Team has reviewed all the water alkaline phosphatase data. Comparison
of data for the May and September 1996, sampling events showed a difference in the fluorescence
readings of the standard curves for the 2.5, 5, and 10 micromolar (uM) methylfluoroscein
standards. The May standards appear to be approximately 1.5 times higher than the corresponding
standards analyzed in September. Examination of the raw data showed the calibration curves for
the analysis of both sampling events to be linear and recalculation of the curves by the data
reviewer produced linear curves with correlation coefficients all greater than 0.990. The data
reviewer was also able to reproduce the results for a selected random sampling of the samples for
these sampling events. This would indicate that the calibration procedure and actual analysis of the
data was performed correctly. It was not possible to compare samples analyzed in May and
September to determine if the
samples exhibited higher
fluorescence in May, since
there was no way to determine
which sites, if any, were the
same, or whether there should
be a comparison. The method
blanks, prepared daily from the
methylfluorosceinphosphate
substrate rather than the methyl
fluorescein, were observed to
see if the fluorescence readings
for them were comparable between the May and September sampling events. Except for the
method blank for the first analytical run in September, the fluorescence of the method blanks were
reasonably consistent during all analytical runs in both May and September and compared quite
Total Nitrogen Comparsion
11 00 13 00
Number of samples from Marsh
C-13
-------
well. The fluorescence reading for the method blank for the first analytical run in September is
approximately 3 to 4 times the fluorescence readings for the other analytical runs in both May and
September. An interview with the analyst was conducted. The age of the standard used to calibrate
the Shimadzu Spectrofluorophotometer was a possible reason for the elevated May results. The
only quality control standard analyzed, however, was the high standard (10 uM) from the
calibration curve, which always gave better than 90 percent recovery.
The difference in the fluorescence readings in May and those in September '96 could be
explained by the use of an old standard or more likely explained by possible changes in the
Shimadzu instrument. Raw data were obtained from SERF for alkaline phosphatase in other
similar samples analyzed just before, during and after the May 1996 sampling event. These data
showed similar fluorescence readings for the standards. There does not appear to be a practical way
to recalculate the curves from the May sampling event so these samples match the curves from
other sampling events, as stated above. It is not recommended that a factor be used to make the
curve fit other data. Such a procedure would not be scientifically defensible. Therefore, the Team
concluded that the May 1996 alkaline phosphatase data (123 samples) are highly questionable and
recommends that the data be rejected and not reported. EAB should delete the May 1996 data from
the database to avoid any misuse at a later date.
11. Sulfide in Water by ESAT
The EPA SESD/ESAT laboratory was the primary laboratory for total sulfide in water
analyses. EPA records show that sulfide samples were collected for September 1995 sampling
event. However, the method used for sulfide analysis during the sampling event was not followed
as written. Samples were not mixed properly prior to adding the color developing reagent. This
would cause the values to have a negative bias. The OQA Project QA chemist upon learning of the
sampling handling deviation after debriefing the analyst, recommended in writing to the EAB
Project Leader that the sulfide data be rejected. For this reason, 123 sulfide data points for the
September 1995 marsh cycle were rejected and not reported.
12. Sulfate in Water by EPA-SESD
EPA-SESD served as the primary laboratory for sulfate analyses. Analysis of water samples
were conducted using the turbidimetric method based on Method 375.4 during the period from
Sept. '93 - Sept 95. Starting in May '96 an ion chromatography (IC) Method 300 was used to
provide lower detection limits better recoveries for spiked water samples. EPA-SESD policy is to
report all results less than the lowest nonzero calibration standard as below detection (flag data
with a "U". The detection limit changed over the course of the sampling program. A 5.0 mg/L or
"5U" reported detection limit was used for the period of Sept. '93 - Sept. '94. In May 1995, the
detection limit was lowered to 2.0 mg/L or reported as "2U" to provide additional data except for a
couple of samples where the detection limit was 1.0 mg/L. A review of the %RSD showed that the
variability was generally less than 8 %RSD indicating excellent precision and did not show the
typical degradation of precision as the detection limit was approached. This indicated the
laboratory was producing valid results before the declared detection limits and data could be
reported at a lower level without compromise. The ion chromatograph method was commenced in
C-14
-------
May ) 996, which allowed for reporting data down to 0.5 mg/L. Since many Marsh samples have
concentrations below 0.5 mg/L a lower detection limit is needed for future studies to give added
information for spatial display of the sulfate data.
13. Conclusions
The South Florida Everglades Ecosystem Assessment Project data set was reviewed against
the Project DQOs, requirements defined in the Project Study Plan, each of the four laboratories
SOPs and QC practices, and additional OQA procedures typically used for regulatory and
compliance program reviews. Each individual datum was verified and validated. Over
20,425 measurements were made on samples analyzed by 4 laboratories for 20 constituents
sampled in 5 media at about 200 canal sites and 550 marsh sites during the 9 field sampling events
over the 10,000 km2 area in South Florida. Included in these measurement were 7 structures
sampled on a bi-weekly basis from February 94 thru February 97. The data that have been qualified
after this rigorous review are:
"J" - Soil/sediment methyl mercury (5%) for the secondary laboratory
"NR." - Water total mercury for September 1993 Canal samples and April '94
transect data for primary laboratory
"J" - Water total mercury, excluding September 1993 Canal, for primary
laboratory.
"J" - Periphyton total mercury for primary laboratory
"J" - Soil/sediment total mercury for May 1994, 1995 Canal cycles and May
1996 marsh samples for secondary laboratory
"NR" - Soil total mercury for September 1995 marsh split samples for secondary
laboratory
"J" - Mosquitofish total mercury for very low mercury concentration fish and
SRM for primary laboratory
"J" - Total nitrogen for the secondary laboratory for Sept. '96 marsh samples
Reject - Water alkaline phosphatase activity for 123 samples for May 1996 marsh
cycle from the primary laboratory
Reject - Water sulfide for 123 samples for the September 1995 marsh cycle from
the primary laboratory.
This Summary of Data Review Findings report should be kept in perspective. The South
Florida Ecosystem Assessment Project was a research and monitoring effort, not a regulatory
compliance project. It developed and implemented new methods for rapidly analyzing a large
sample load with ultra-trace level constituent concentrations. This rigorous review occurred over
about a 2.5 year period, being initiated in 1995 and completed in 1998. Out of the over
20,425 measurements made over the four year period, only 15 percent of data was qualified. There
were 1,325 points flagged "NR"- not reviewed because raw data were not available from the
laboratory, 1,783 assigned a "J" flag - analyte identification was acceptable for use but certain QC
criteria were not satisfied and 256 values rejected and not reported in the database.
C-15
-------
The OQA finds that the data set satisfies the Project objectives and permits the assessment
of trends and a characterization of status and extent of ecological condition in the South Florida
Everglades ecosystem. Data qualified with a "J" code would not compromise the characterization
and trend analysis or the Project objectives. "J" coded should not be used solely in the setting of
environmental standards unless additional data are collected to substantiate these original values
that have all supportive QC and documentation. These data can be used as part of the weight of
evidence approach for establishing environmental standards.
This data set should provide an important baseline for assessing the current conditions of
the Everglades ecosystem and for monitoring future trends in this condition. The results from this
QA/QC Review have already been, and will continue to be, used to refine the QA/QC procedures
in future Project monitoring efforts.
C-16
-------
APPENDIX D
Eastern Mosquitofish Studies
-------
APPENDIX D
1.0 Eastern Mosquitofish
A predator protection criteria of 0.1 mg/kg THg for prey species has been proposed by the
USFWS (Eisner et al. 1987). The eastern mosquitofish, Gambusia holbrooki whole body THg
concentrations were presented in the proceeding section. About 15% of the canal miles and
almost 70% of the marsh area have mosquitofish with mercury concentrations exceeding the
predator protection criteria of 0.1 mg/kg. Because the mosquitofish is a prey species for
piscivorous fish and birds and is an excellent indicator of Hg bioaccumulation, additional
analyses were conducted on the mosquitofish populations in the canals and marsh. The purpose
of these analyses were to determine if differences in population attributes or feeding habits
among subareas or among latitudes might contribute to mercury bioaccumulation. The results are
presented in the following section.
1.1 Descriptive Statistics
The descriptive statistics of the eastern mosquitofish were examined to determine the
nature of the sample population and the possible relationships with Hg bioaccumulation, size,
and condition factors. An additional test sample of mosquitofish was collected from near the
marsh Hg hotspot and the Everglades ENR in July 1997.
1.1.1 Canal Fish
A length frequency histogram of the mosquitofish in the combined canal sample showed
a normal distribution (Figure D.l). A total of 1,074 mosquitofish was analyzed with a median
total length of 24.0 mm (0.95 inches), ranging in size from 10 mm (0.4 inches) to 38 mm
(1.5 inches). Females made up 70.3% of the population sample and males made up 28.5% with
the remainder made up by juveniles. The median length for males was 22.9 mm (0.9 inches),
1.2 mm (0.05 inches) smaller than the females at 24.2 mm (0.95 inches). A box and whisker
analysis (Figure D.2) of the fish lengths by cycle indicated fish sizes m cycles I and 2 were
significantly larger and smaller, respectively, than cycles 0 and 3. A box plot analysis of fish
D-l
-------
length by geographic subarea showed significantly larger fish occurred in ENP, followed by
EAA, WCA, and BCNP. Length/weight distributions and growth curves for each cycle show
similar condition for each subpopulation (Figure D.3). A condition factor (CF) was calculated
with the formula CF = w/13, where w = individual weight and 1 = individual length (Tesch 1968).
The data combined by cycles showed significantly higher condition factors in fish from the EAA
and ENP and lower condition factors for fish in the WCA canals and BCNP (Figure D.4).
1.1.2 Transect Fish
A length frequency histogram of the mosquitofish in the combined transect sample
population is shown in Figure D.5. A total of 225 fish was analyzed, with 73.8% females,
25.3% males, and the remainder juveniles. The median total length was 25.6 mm (1.0 inch), and
the sample population ranged from 14.2 mm (0.56 inch) to 34.4 mm. The median total length for
males was 24.7 mm (0.74 inch), which was 1.9 mm (0.07 inch) less than females at 26.6 mm
(1.05 inch). A box and whisker analysis (Figure D.6) of fish length found a significant difference
between the LNWR and WCA3 in the fish sizes among the transects, however, all other
distributions were not significantly different. Length/weight distributions showed similar
condition among subareas (Figure D.7). The condition factors were significantly higher for
mosquitofish from WCA2 and ENP transects than those on LNWR and WCA3 transects
(Figure D.8).
1.1.3 Marsh Fish
A length frequency distribution of marsh fish included 2,158 individuals with a median
size of 23.0 mm (0.04 inch), ranging from 9 mm (0.35 inch) to 39 mm (1.54 inches)
(Figure D.9). The sample was normally distributed with 65.5% females, 25.1% males and 9.5%
juveniles. Males had a median length of 22.4 mm (0.88 inch), which was 1.2 mm (0.05 inch) less
than the females at 23.6 mm (0.93 inch). A box and whisker analysis (Figure D.10) of the data by
cycle indicated the sample population had significantly larger fish in cycle 0 (April 1995) and
significantly smaller fish in cycle 3 (September 1996) than the similar distributions in cycles 1
and 2. The September 1996 sample followed the driest dry period in spring 1996, which
D-2
-------
apparently killed a large number of these small fish due to dry down and predation, followed by
recruitment of young individuals during the following wet season. Length/weight distributions
were plotted by cycle with associated growth curves(Figure D.l 1). There were no apparent
differences in condition; however, in cycle 3 was somewhat less due to preponderance smaller
individuals. The condition factor plotted by subarea for the combined data, indicated the highest
condition factor occurred in fish from LNWR and BCNP and decreased in WCA2 and WCA3
with the lowest condition factor associated with fish in the ENP (Figure D.l2).
1.1.4 Fish in 1997 Test Sample
A mosquitofish sample was collected from two marsh sites in 1997, the ENR and WCA3
near where the Hg concentrations in mosquitofish were previously found to be the highest in the
marsh system. The ENR is an old agricultural field that has been converted into a prototype
wetland stormwater treatment area designed to remove TP from stormwater. Agricultural fields
were not sampled as a routine part of the REMAP study. A total of 153 fish was analyzed
(Figure D.l3). The median fish size at the ENR and WCA3 sites was 19.7 mm (0.78 inch) and
17.2 mm (0.68 inch), respectively. The fish ranged in size from 9.0 mm (0.35 inch) to 35.6 mm
(1.40 inches) at the ENR site and from 7.6 mm (0.3 inch) to 34.0 mm (1.34 inches) at the WCA3
site. A box and whisker plot shows the ENR fish were significantly larger than the WCA3 fish
(Figure D.l4). The condition factor was higher for ENR fish (Figure D.l5). The THg
concentrations in fish from the ENR were found to be the lowest measured in fish during this
study (Figure D.l6). Samples analyzed by three different laboratories found that THg
concentrations in fish from the ENR were less than 10 /Ug/kg while the THg in fish from the
WCA3 site averaged about 150 ^g/kg (Figure D.l6).
It is apparent from these data that a consistent sample of mosquitofish was collected from
both canal and marsh habitats throughout the course of this study. The sample collected is
representative of the naturally occurring mosquitofish population in the system. The average size
of the mosquitofish populations observed in the Everglades ecosystem are very small for the
species found anywhere in their range (Trexler personal communication). The Everglades
ecosystem is a food limited, oligotrophic system (Loftus personal communication) that reduces
D-3
-------
the size of this species. The usefulness of this species as an important indicator for monitoring
the bioaccumulation of Hg in the Everglades ecosystem is evident from the changes in THg
concentrations in this species across the TP, TS04 and TOC gradients in this system. The
consistent future sampling and analysis of mosquitofish can provide the information needed to
assess changes and trends in Hg contamination in the South Florida Everglades ecosystem.
1.2. Mosquitofish Gut Contents
Bioaccumulation and biomagnification of MeHg through the food web is influenced by
both the quantity and quality of the ingesta. An additional study was conducted to determine the
gut contents of mosquitofish collected during the September 1996 sampling cycle.
Many species are known to switch diets both during development and based on food
availability in their environment. Omnivorous species have the potential to yield important
influences on their prey by switching their diet choice with regard to its changing abundance,
targeting it in times of plenty, and ignoring it in times of scarcity. Furthermore, the ability to
switch foods, even going from herbivore to carnivore, may buffer a species from fluctuations in
food supply and permit it to sustain larger population densities. Ontogenetic changes in diet may
also reduce competition among age classes (Werner and Gilliam 1984). One possible outcome in
such a case is that individuals may begin to specialize in subsets of their potential prey
(Magurran 1993). Also, local populations of omnivores may diverge substantially in their feeding
biology, and role in local ecological communities, as a result of local environmental conditions
and food availability.
In spite of these possibilities, few studies have characterized the diet of an omnivorous
species over a large spatial scale, probably because of the inherent difficulties of collecting
adequate specimens to do so. The eastern mosquitofish (Gambusia holbrooki), and its western
congener, Gambusia affinis, are known to be aggressive omnivores (reviewed in Meffe and
Snelson 1989) with the potential to yield greater effects on the ecosystems where they live than
might be predicted based on their small size (Courtney and Meffe 1989). They have been widely
introduced world-wide for mosquito control and have been responsible for the extinction of
native fishes in some cases through consumption of larvae (Meffe et al. 1983). Experimental
D-4
-------
ecological studies have documented their potential to regulate the dynamics of other community
members through direct consumption and competition (Hurlburt et al. 1972, Hurlbert and Mulla
1981, Harris 1995, Belk and Lydeard 1994, Schaefer et al. 1994). Mosquitofish are also
cannibalistic (Krumhotz 1948, Meffe and Crump 1987). Mosquitofish are known to feed on plant
matter and detritus, in addition to these predatory predilections.
1.2.1 Methods
1.2.1.1 Data Gathering
Mosquitofish were collected by dipnet from 101 locations scattered across the Florida
Everglades (Figure D.18) from September 18 to 23,1996. The sites were selected by a
stratified-random procedure described earlier. Immediately following capture the fish were
placed in 10% formalin in the field to rapidly stop the digestion of the stomach contents. Twenty
fish were collected at each site of which 12 to 14 specimens were analyzed for stomach contents.
Duplicate samples were taken at 10 sites.
Individual mosquitofish were dissected and their gut contents removed and separated into
six categories: (1) plant matter (pooling algae, vascular plant, and detritus), (2) cladocera,
(3) aquatic mites, (4) chironomid larvae (midge larvae), (5) adult midges, and (6) other (primarily
spiders, ants, aquatic beetles, and fish). Counts of the number of items in all animal categories
were recorded for each mosquitofish, along with their sex and standard length. Males could be
identified readily by the presence of a gonopodium, and females were identified by presence of
mature ovaries or by standard length exceeding 18 mm (0.71 inch). Juveniles were all fish below
18 mm (0.71 inch) standard length lacking a gonopodium. The presence or absence of plant
matter was recorded for each specimen, and if no food was present this was also noted. All food
items for the fish from a single population sample were pooled and the mass of each food
category was determined. The sum of these masses provided an estimate of the total mass of food
consumed by that sample of fish.
D-5
-------
1.2.1.2 Statistical Analyses
The pattern of food choices by individual fishes were examined, followed by analyses of
the population samples. No mass estimates were available for the food items of individual
specimens, particularly for plant food, so the presence or absence of food types was examined.
Principle components analysis was conducted to investigate patterns in the food choices. The
covariance matrix was factored because all data were scored as present or absent rendering all
variables on a similar scale (Stevens 1986); varimax factor rotation was used. The effect of fish
size and sex on food choices were examined by logistic regression of presence/absence of food
types in individual fish gut contents (Trexler and Travis 1992). Fish with empty stomachs were
excluded from these analyses. The presence/absence of food in the gut of individual fish was
coded as a dependent variable and this was examined separately by logistic regression with fish
sex and standard length as independent variables.
The percentage of each food category in the diet of fishes from each population sample
was calculated from the mass data. These percentages were analyzed in analyses of covariance by
grouping populations into geographic regions of the study area using two schemes. First,
populations were grouped according to the water management region where they were found:
LNWR, WCA2, WCA3, ENP, and BCNP. There are general north to south gradients in
productivity across the Everglades following patterns of nutrient enrichment from agricultural
runoff (Davis 1994, Stober et al. 1996). The effects of this pattern by grouping the populations
into 6 regions by latitude from north to south were examined: (1) > 26.4°N; (2) 26.4°N to 26.2°N;
(3) 26.2°N to 25.9°N; (4) 25.9°N to 25.7°N; (5) 25.7°N to 25.5°N; and (6) <25.5°N. The average
standard length of fish from each collection was retained as a covariate in these analyses. In all
cases, data were examined for consistency with the assumptions of standard statistical procedures
such as normality, and transformations were applied as needed to fulfill the assumptions of
analyses (Zar 1984).
D-6
-------
1.2.2 Results
1.2.2.1 Overview
Over 1200 fish averaging 15.9 mm (0.63 inch) standard length, most of which were
juveniles were examined (Table D.l). Males were typically smaller than females by over 2 mm
(0.08 inch). A conversion between standard and total length for males, females, and juveniles
with an r2 over 0.9 is : Standard length = -1.337 + 0.886 (total length). Periphyton comprised
36% of the diet of mosquitofish based on biomass in gut contents, with insect, crustacean,
arachnid, and piscine prey accounting for the remaining 64%. Adult midges, undoubtedly
gleaned from the water surface, accounted for 34% of the biomass of the diet, and midge larvae,
probably taken from floating, epiphytic and benthic periphyton mats, accounted for an additional
9.6%. Two fish (both Heterandria formosa) and an assortment of spiders, ants, and beetles
account for 15% of the diet by biomass. About 50% of the individual fish had plant matter
present in their guts, and about 45% had adult midges (Table D.2). Chironomid larvae and
"other" prey were both found in about 10% of the fish, while mites were present in around 8.0%
and cladocerans in only 3.0% of the fish examined. Very few of the fish had empty stomachs (53
out of 1,265 fish examined).
1.2.2.2 Individual Fish
Both the size and sex of fish influenced the likelihood that they had empty stomachs.
Larger specimens were more likely to have empty stomachs than smaller ones, though even the
largest juveniles had less than a 5% chance to have empty stomachs because of their relatively
small size (always less than 18 mm (0.71 inch); Figure D.19). Females and juveniles did not
differ in the probability of having food present in their gut once size differences were accounted
for; however, males were more likely to have empty guts than females or juveniles at the same
size (Figure D.20). Still, the likelihood of not having fed prior to collection was low in the
sample; the model estimated that even the largest specimens had only a 20% likelihood of no gut
contents.
Principal components analysis indicated that patterns of consumption of the six food
categories did not overlap. No factor component loaded heavily on more than one food type,
D-7
-------
although the first two components explained over 60% of the total variance. A component
loading heavily on the incidence of plant matter explained 31.4% of the variance in gut contents,
one loading on adult dipterans explained a similar amount (30.9%), and two other components
loading on chironomid larvae and "other" each explained an additional 11.8%. The lack of
structure in the data revealed by this analysis led to the consideration of each food class
individually in subsequent analyses.
Size was a significant factor in the likelihood that mosquitofish had consumed plant
matter immediately preceding collection. The likelihood of finding periphyton in the gut
decreased with size (Table D.3) from approximately 60% for 10 mm (0.4 inch) juveniles to about
35% for a 25 mm (0.99 inch) adult female (Figure D.21). Males were less likely to have
consumed periphyton than females (adult fish only, model: Pr [periphyton in gut] = constant +
sex + size, t2,2 = 7.47, P = 0.032, size t5l5 = 2.28, P = 0.015, sex t5]5 = 2.14, P = 0.012;
Figure D.22).
Some categories of animal prey decreased in frequency in the diets of mosquitofish as
they got larger, others increased, and some were unaffected by size. The incidence of cladocerans
decreased with size, while adult dipterans and "other" prey increased in frequency as fish got
larger. Chironomids and mites were equally likely to appear in the diets of all size of fish
(Table D.4). In general, the diets of males and females did not differ regarding the incidence of
animal prey, with the exception of "other" prey. However, this difference was explained by the
size difference between the sexes (adult fish only, model: Pr("other" in gut) = constant + sex +
size +sex by size , t213 = 12.01, P = 0.009, size t5l5 = 2.26, P = 0.015, sex t5)5 = 1.63, P = 0.11, sex
by size t515 = 1.7, P = 0.09).
1.2.2.3 Geographic Variation
The average size of fishes examined differed among the five water management units of
the Everglades. Population samples of mosquitofish from LNWR and WCA3 contained
significantly smaller fish than average, and those from WCA2 and BCNP tended to be larger
(Table D.l). In light of the findings of diet changes with mosquitofish size, analyses comparing
these water management areas must adjust for the size of specimens in samples. And, as
D-8
-------
expected, the mass of food found in the guts of fishes is related to the average size of specimens
in the population samples (Table D.3). However, differences among management areas persist
after size variation is accounted for: WCA2 and WCA3 fishes tended to have more food in their
guts than average, while those from LNWR and BCNP had less than average (Table D.4). When
the data were sorted into six latitudinal categories, similar patterns were revealed.
Some food items appeared to vary across the water management units in their prevalence
in the diet of mosquitofish. The relative amount of plant matter in the diets displayed a tendency
to vary among regions (Table D.3; P=0.088, dropping to P=0.056 when length was excluded
from the model) with BCNP fishes displaying notably less plant matter than the fish from the
other four regions examined; this result was obscured when regions were created along a north to
south gradient (Table D.4). Chironomid larvae displayed a more marked variation among regions
in their inclusion in diets. They accounted for over 20% of the diets in LNWR and WCA2, but
dropped to below 10% in all other regions (Table D.4). This is also seen as a north to south
gradient when the data are grouped by latitude (Table D.4). "Other" prey items appeared most
frequently in the diet of fishes from BCNP, and least in data gathered from fish from LNWR,
though this result was obscured when samples were compared along the latitudinal gradient
(Table D.4). Adult midges and mites displayed no regional patterns in their incidence in
mosquitofish diets (Table D.3).
D-9
-------
Table D.l Descriptive statistics of size of the fish examined. Standard length (mm) and
95% confidence intervals are reported, with sample sizes below.
Data Grouping
All Fish
Males
Females
Juveniles
All Populations
15.93 ±0.20
1270
17.12 ±0.34
108
19.66 ±0.28
407
13.73 ±0.07
748
By Region
WCA1
15.04 ± 1.16
11
—
—
—
WCA2
17.20 ± 1.70
8
—
—
—
WCA3
15.20 ±0.26
40
—
—
—
ENP
16.18 ±0.75
33
—
—
—
BCNP
17.83 ± 1.54
9
—
—
—
D-10
-------
Table D.2 Relationship of diet to size in juvenile and female eastern mosquitofish. Results
from logistic regression of incidence (presence/absence) of food item in gut
contents on female size measured as standard length. In one case, adult dipterans,
a significantly better model fit was obtained by use of ln-transformed length.
Average Size, n-columns indicate the average size (mm) and sample size of fish
with each item absent or present. The Pr (Present) is the probability estimated at
the grand mean size. Slope is the probability that each food item will be present in
gut contents with a 1 -mm increase in standard length, except for adult dipterans
which were best fit on a log scale.
Food Type
X2
P
Item Absent
Average Size, n
Item Present
Average Size, n
Pr (Present) Slope Pr(Present) vs
Observed Predicted Standard Length
Periphyton
11.6
0.001
16.2, 580
15.5, 581
0.500
0.500
-0.014
Cladoceran
8.1
0.006
15.9, 1117
14.3,39
0.034
0.031
-0.004
Adult Dipteran
3.9
0.05
15.7, 645
16.0,511
0.442
0.442
0.127*
Chironomid Larvae
0.5
>0.4
15.8, 1042
15.6. 114
0.099
0.098
Mite
1.8
0.185
15.9, 1058
15.3,98
0.085
0.084
Other
17.9
<0.001
15.7,1040
17.2, 116
0.100
0.094
0.090
* indicates slope on a In (standard length) scale
D-ll
-------
Table D.3 Geographical analysis of the percentage of total mass attributable to each food
type from the gut contents of mosquitofish. Two geographical groupings were
used, Water Management Units refers to comparisons of population means among
management units, while Latitude refers to grouping populations by latitude. No
interactions were significant.
Water Management Units Latitude
Food Item
Effect
F
DF
P
F
DF
P
Plant Food
Length
0.004
1,95
0.843
0.060
1,94
0.807
Region
2.088
4,95
0.088
1.410
5,94
0.228
Cladocera
Length
0.115
1,95
0.736
0.541
1,94
0.464
Region
0.727
4,95
0.576
0.785
5,94
0.563
Adult Midges
Length
0.430
1,95
0.514
0.373
1,94
0.543
Region
1.033
4,95
0.395
1.558
5,94
0.180
Midge Larvae
Length
8.379
1,95
0.015
9.034
1,94
0.003
Region
4.130
4,95
0.014
5.042
5,94
0.001
Mites
Length
0.227
1,95
0.635
0.191
1,94
0.663
Region
1.175
4,95
0.327
0.725
5,94
0.606
Other Prey
Length
1.974
1,95
0.163
4.046
1,94
0.047
Region
1.900
4,95
0.117
1.224
5,94
0.304
All Food Mass
Length
7.373
1,95
0.008
6.85
1,95
0.010
Region
2.878
4,95
0.027
2.580
5,95
0.031
D-12
-------
Table D.4 Adjusted means from analyses of food items by geographical groupings. All
means are adjusted to the grand mean fish size of 15.93 mm standard length.
Grouping
Plant Food (%)
Midge Larvae (%)
Other (%)
All Food (mg)
Water Management Units
WCA1
27.3
25.1
6.1
0.973
WCA2
30.1
21.1
13.1
2.328
WCA3
42.7
7.9
16.5
2.352
ENP
36.5
6.0
13.1
1.348
BCNP
8.6
4.7
32.4
1.096
Latitude
>26.4
26.7
25.1
6.7
0.968
26.4 - 26.2
35.4
20.7
10.0
2.636
26.2 - 25.9
34.6
5.4
22.4
1.678
25.9-25.7
34.6
2.9
20.8
2.033
25.7 -25.5
50.3
5.7
14.2
1.863
<25.5
21.6
6.6
13.2
0.942
D-13
-------
90
a
<
60
a>
X3
6
£
30
0.
Canal Data for All Cycles
¦ Juvenile
~ Male
H Female
JL
*\i 111
6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Fish Length, mm
Figure D. 1 Length frequency histogram of mosquitofish collected in the canals.
-------
Canal Data
95%
CONFIDENCE
ON MEDIAN
*
*
>C
X
O > 75V* ~ ) MtDfcANGE
# > 75% ~ I 5 MIDRANGE
75% ~ 1 5 MIDRANGE
75%
MEDIAN
25%
i 25% • I 5 MIDRANGE
*
0
Sampling Cycle
Figure D.2 Notched box and whisker plot comparing mosquitofish length for each sampling cycle.
-------
0
1
Figure D.3 Canal mosquitofish length versus weight with derived growth curves for each sampling cycle.
-------
D
¦
-j
o
a
U*
g
o
"O
c
o
U
rt
•
CO
13
JO
6
o
Canal Data
o
o
t
95%
CONFIDENCE
ON MEDIAN
0> IV/. « > MID RANGE
« > 75V. ~ I 5 MIDRANGE
7SV. ~ I 5 MIDRANGE
75%
' MIDRANGE
, 25V.
I JSVt . 1 5 MIDRANGE
O
o
3 ^
o
o
o
o
o
o
0
EA A
WCA
ENP
BCNP
Figure D.4 Notched box and whisker plot comparing condition factors for canal mosquitofish in subareas.
-------
D
I
oo
30
20 -
-------
D
I
\o
40
30
B
B
00
G
0)
J 20
.2
C/3
3
rO
B
a
o
10
0
Transect Data
X
0> 75% ~ ) MID RANGE
* > 75% ~ \ I MlDRANGE
7J% ~ I 5 M1DRANGE
75%
95% j
CONFIDENCE \
ON MEDIAN ^
MEDIAN
25%
> MIDRANGE
2S% . I 5 MIDRANGE
LNWR
WCA2
WCA3
ENP
Figure D.6 Notched box and whisker plot comparing mosquitofish lengths in each of the marsh transects.
-------
FUh Length, mm
0
1
K)
O
FUb Length, mm
Flih Length, mm
Figure D.7 Mosquito fish length versus weight with derived growth curves for each marsh transects.
-------
Transect Data
95%
CONFIDENCE
ON MEDIAN
0> 75% ~ J MIDRANGE 1
* > 75% ~ 1 5 MIDRANGE
75% ~ I 3 MIDRANGE
75%
> MIDRANGE
25%
25% • I 5 MIDRANGE
o
*
*
T
*
*
JL
LNWR
WCA2
WCA3
ENP
Figure D.8 Notched box and whisker plot comparing mosquitofish condition factors for transects.
-------
200
hO
hJ
150
Ui
-------
40
T
T"
o
Marsh Data
30 -
bO
C
u
i—J 20
10
LZJ
0> 7J% ~ ) MID RANGE
* > 75% ~ 1 J MIDRANGE
f 7J% ~ I 5 MIDRANGE
95% J
CONFIDENCE <
ON MEDIAN I
75%
> MIDRANGE
1S%
25%. I 5 MIDRANGE
0
1
*
r>
3
Sampling Cycle
Figure D.10 Notched box and whisker plot comparing lengths of mosquitofish collected in the marsh for each sampling cycle.
-------
D
i
K>
¦U
0.6
05
Cycle 0
Y"»0.003'exp(0.14*X)+ep»
10.0 20 0 30 0
Piih Length, mm
•a
.3
0.6
0.5
0.4
03
£ 0.2
0.0
00
10.0 20 0 30 0
Filb Length, mm
40 0
1 1
Cycle 2
i ;
Y=0.005'exp(0.1 23*X)+ep»
-
1 /
• • • V
•
:l-ls
• 111 L
,
i
40.0
0.6
05
04
03
•a
£ 02
0.1
0.0
l I
Cycle 1
1 j
Y=0.004*exp(0.128*X)+eps
%/
I«
" •••
i$K • •
$i:-'
0.0
100 20.0 30.0
Fith Length, mm
J3
.9
0.6
05
0.4
03
.a
£ 0.2
0.1
00
Cycle 3
Y»0 002*cxp(0.16*X)+ep»
0.0
100 20.0 30.0
Flih Length, mm
40 0
40.0
Figure D.l 1 Length versus weight of mosquitofish collected in the marsh during each sampling cycle with derived growth curves.
-------
NJ
o
&
U*
G
O
T3
G
o
U
cj
•
GO
G
rO
E
o
0
Marsh Data
O > 75% ~ ) MIDRANGE
* > 75% ~ I 5 MIDRANGE
7)% ~ I J MIDRANGE^
75% —-x U
95% f
CONFIDENCE <
ON MEDIAN \
MEDIAN
25%
> MIDRANGE
25% - I 5 MIDRANGE
*
*
£3
WCAl
o
f
WCA2
o
o
o
o
o
o
>k.
t
o
o
WCA3
ENP
*
i
BCNP
Figure D.12 Notched box and whisker plot comparing condition factors for marsh mosquitofish collected in subareas.
-------
Fish Length, mm
Figure D.13 Length frequency distribution of mosquitofish in the 1997 test sample.
-------
40
30
£
a
v
to
-j
60
G
J 20
cd
»
C/D
G
X>
6
Ctf
O
10
0
EPA Data
%
*
*
95%
CONFIDENCE
ON MEDIAN
0> 73V. ~ ) MID RANGE
* > 7$% ~ I 5 MrDRANGE
73% ~ I 5 MTDRANGE
75%
1 25%
I 23% • I 5 MIDRANGE
ENR
WCA3
Figure D.14 Notched box and whisker plots comparing length of mosquitofish at the two marsh sites from the 1997 test sample.
-------
0.4
60 0.3
JS
to
2 0.3
u
0.1
0.0
0.0 10.0 20.0 30.0 40.0
Fish Length, mm
D. 15 Length versus weight of mosquitofish from the two sites in the 1997 test samples with
derived growth curves.
D-28
-------
0> 75% ~ 3 MLDRANCE
* > 75% ~ 1 5 M1DRANGE
t 75% + | S MIDRANGE
J 75%
> MIDRANGE
25%
25% - I 5 MIDRANGC
EPA
FDEP
FIU
500
60
M
I1 400
j? 300
6
a
o
a 200
3 ioo
o
S
WCA3
JL
EPA
_ I
FDEP
T
i/ \j
FIU
D. 16 Notched box and whisker plots comparing THg in mosquitofish analysis results from
3 labs.
D-29
-------
LONGITUDE, decimal degrees
Figure D. 17 Map of the study area indicating sites of sample collection.
D-30
-------
0.65
0.60^
0.551
3 0.501
C -
S 0.45-
Z
z 0.40^
•* ~
5 0.35-
6 1
£ 0.30E
0.25=
0.20
Cttta^JUVENILES
. FEMALES
0^^
&
e
$
e
0
0
12
18
24
30
36
Standard Length (mm)
Figure D. 18 Probability that an individual fish will have plant matter in its stomach relative to
standard length. Based on logistic regression. Results for juveniles, males, and
females are plotted separately.
1.02
0.98-
c0
0.94-
0.90-
0.86
0.82-
0.78
C®Q&7r£yENILES
ALES
MALES
\ %
0
0
0
-r
12
18 24
Standard Length (mm)
30
36
Figure D. 19 Probability that an individual fish will have food in its stomach relative to
standard length. Based on logistic regression. Results for juveniles, males, and
females are plotted separately.
D-31
-------
REFERENCES
Literature Cited
Belk, M. C., and C. Lydeard. 1994. Effects of Gambusia holbrooki on a similar sized synoptic
poeciliid, Heterandria formosa: competitor or predator? Copeia 1994:296-302.
Courtney, W. R., Jr., and G. K. Meffe. 1989. Small fishes in strange places: a review of
introduced poeciliids, pp. 319-331. In G. K. Meffe and F. F. Snelson, Jr., eds. Ecology
and Evolution of Livebearing Fishes (Poeciliidae). Prentice-Hall, Englewood Cliffs, NJ.
Harris, P. M. 1995. Are autecologically similar species also functionally similar? A test in pond
communities. Ecology 76:544-552.
Hurlbert, S. H., J. Zedler, and D. Fairbanks. 1972. Ecosystem alteration by mosquitofish
(Gambusia affinis) predation. Science 175:639-641.
Hurlbert, S. H., and M. S. Mulla. 1981. Impacts of mosquitofish (Gambusia affinis) predation on
plankton communities. Hydrobiologica 83:125-151.
Johnels, A. G., T. Westermark, W. Berg, P. I. Persson, and B. Sjostrand. 1967. Pike (Esox lucius
L.) and some other aquatic organisms as indicators of Hg contamination in the
environment. Oikos 18:323-333.
Krumholz, L. A. 1948. Reproduction in the western mosquitofish, Gambusia affinis affinis
(Baird & Girard) and its use in mosquito control. Ecological Monographs 18:1-43.
Magurran, A. E. 1986. Individual differences apd alternative behaviours, pp 441-477. In: T. J.
Pitcher, ed. Behaviour of Teleost Fishes, Second Edition. Chapman & Hall, London.
Meffe, G. K., and M. L. Crump. 1987. Possible growth and reproductive benefits of cannibalism
in the mosquitofish. American Naturalist 129:203-212.
Meffe, G. K., D. A. Hendrickson, W. L. Minckley, and J. N. Rinne. 1983. Factors resulting in
decline of the endangered Sonoran topminnow (Atheriniformes: Poeciliidae) in the
United States. Biological Conservation 25:135-159.
Meffe, G. K., and F. F. Snelson, Jr. 1989. An ecological overview of poeciliid fishes, pp 13-31.
In G. K. Meffe and F. F. Snelson, Jr., eds. Ecology and Evolution of Livebearing Fishes
(Poeciliidae). Prentice-Hall, Englewood Cliffs, NJ.
Polis, G. A., and D. Strong. 1996. Top-down and bottom-up forces in food webs: do plants have
primacy? American Naturalist 147:813-846.
Schaefer, J. F., S. T. Heulett, and T. M. Farrell. 1994. Interactions between two poeciliid fishes
[Gambusia holbrooki and Heterandria formosa) and their prey in a Florida marsh. Copeia
1994:516-520.
Stevens, J. 1986. Applied Multivariate Statistics for the Social Sciences. Lawrence Erlbaum
Associates, Hillsdale, NJ.
Stober, J., D. Scheidt, R. Jones, K. Thornton, R. Ambrose, and D. France. 1996. South Florida
ecosystem assessment. Monitoring for adaptive management: implications for ecosystem
restoration (interim report). EPA 904-R-96-008.
Trexler, J. C., and J. Travis. 1993. Nontraditional regression analyses. Ecology 74:1629-1637.
Werner, E. E., and J. F. Gilliam. 1984. The ontogenetic niche and species interactions in
size-structured populations. Annual Review of Ecology and Systematics 15:393-425.
Zar, J. H.. 1984. Biostatistical Analysis. Prentice-Hall, Englewood Cliffs, NJ.
D-32
-------
APPENDIX E
Response to Peer Review Comments
-------
Peer reviewers for the South Florida Ecosystem Assessment Final Technical Report.
EPA Reviewers
1) Tony Olson, EMAP, Corvallis**
2) John Stoddard, EMAP, Corvallis
3) Steve Paulsen, EMAP, Director
4) Rick Linthurst, NERL, RTP, Assoc. Director
5) Gil Veith, NEERL, RTP, Assoc. Director
6) Kevin Summers, Gulf Breeze Lab
7) Anthony Carlson, REMAP**
8) Rose Russo, NERL, Athens
9) Tudor Davies, OW,OST, Hdqs
10) Arnold Kuzmack, OW, Hg Task Force, Hdqs
External Reviewers
11) Jim Weiner, USGS, BRD, LaCross, WI*
12) Ronnie Best, USGS, BRD
13) Dave Krabbenhofit, USGS, WRJD
14) Bill Loftus, USGS, BRD
15) Aaron Higer, USGS, WRD
16) Anders Andren, U WI,
17) Jack Gentile, U Miami*
18) David Lean, Canada**
19) Bob Johnson, NPS, ENP
20) Sue Jewel, LNWR**
21) Wallace Hibbard, Big Cypress NP
22) Tom Fontaine, SFWMD
23) Tom Atkeson, FDEP (Don Axelrad)**
24) Jerry Keeler, U of Michigan
25) Larry Fink, SFWMD(combined)**
26) Lorraine Heisler, FGFWFC
27) Ted Lange, FGFWFC
28) Col. Joe Miller, USACE
29) Terrance "Rock" Salt, Restoration Task Force
30) Terry Rice, South Florida Task Force
31) Billy Cypress, Miccosukee Tribe
32) James Billie, Seminole Tribe(Craig Tepper)**
33) Ron Tipton, Everglades Coalition
34) Bubba Wade, US Sugar
35) Ed Barber, Sugar Cane Coop
36) Pete Rosandahl, FloSun**
37) Don Porcella, EPRI
38) Ed Zillioux, FPL
39) Scott Osborne, FL Power Coord. Group
40) Thomas Corcoran, Audobon Society
~Acknowledgment
**Review Comments
1
-------
41) John Rudd, FWI, Canada
42) Hans Hultberg, Goteborg, Sweden
43) Brian Rood, Mercer U., Macon
44) William Walker, for ENP**
45) Chris McVoy, SFWMD
46) Terry Haines, USGS, BRD
47) Ben McPherson, USGS
48) John Davis, ES&P**
49) Carol Kendall, USGS**
50) Steve Bartell, Oak Ridge
51) Gary Bigham, Exponent
52) Cynthia Gilmore
InHouse
53) Russ Wright
54) Antonio Quinones
55) Richard Harvey
56) Jerry Stober
57) Dan Scheidt
58) Ron Jones, FIU
59) Joel Trexler, FIU**
60) Steve Rathbun, UGA
61) Don Stevens, Dynamac
62) Mike Birch**
63) Linda Anderson-Carnahan and Stacy Howard**
""Acknowledgment
**Review Comments
2
-------
Department of
Environmental Protection
June 26, 1998
Dr. Jerry Stober
Science and Ecosystem Support Division
US EPA Region 4.
980 College Station Road
Athens, GA 30605-2700
Thank you for the opportunity to review your draft report. This work reported on therein clearly has been a massive
undertaking in the field and laboratory, and this work itself a massive data reduction effort. It is very informative and
useful-and in fact-it has already contributed to hypothesis generation for research we hope to fund on MeHg
bioconcentration by Everglades periphyton. Our only problem with your report, and one that I am sure you and your
colleagues have (and are) struggling with, is how to comprehend so much data!
I am returning the first volume of the report with many marginal notes. Please see my handwritten comments and
suggested edits in Chapters 8,9, and 10, as well as on previous pages marked by post-its. In some of my comments
I have been assisted by Tom Atkeson, and Mike Turtora, a USGS water-quality statistician working with Tom
Atkeson and me on mercury in lakes and streams outside the areas covered by your report. Anders Andren will also
be submitting comments within a few days. Specific comments are as follows:
Chapter 8 Mercury
There seems to be little direct exploration of relations between variables in this report. In addition to implying
relations by comparing trends in geographic medians among variables, direct regression or PCA of variables over
the individual sampling units should be considered as it may prove to be more informative. Multiple regression of
'independent' variables on Hg concentrations or regression of the first few principle components on Hg
concentrations would be valuable. In addition, maps could be prepared of principal component scores at each site.
Finally, maps of the regression residuals may reveal interesting patterns.
Response: Exploratory analyses were used extensively during data analysis. The primary variables,
DOC, Sulfate, TP, and methylmercury in water and total mercury in mosquitofish were
identified through PCA and factor analysis, as indicated on p 4-7. The large array of statistical
analyses performed on the data are presented in text and tables on pages 3-14 to 3-19.
Unfortunately, displays of exploratory analysis results, in many instances, are not very
illuminating if the reader is not familiar with the procedures. Therefore, these displays were
not included in the report even though the analyses were conducted and the results used to
identify multivariate relationships Information is presented using the most appropriate of the
many statistical analyses performed.
While median values appear on most plots, parametric ANOVA on log-transformed data is implied. The plots would
benefit by changing the vertical axis to a log scale and a nonparametric ANOVA would probably be more useful.
A nonparametric ANOVA would eliminate the need for outlier analysis and a log plot scale would produce more
interpretable graphics.
Response: Median values were chosen as a conservative analytical approach in which 50% of the data
fell above and below placing less weight on outliers in the data set since no data were
3
-------
removed. The data were log transformed for some analyses as indicated in Table 3.2. Some
log scale graphics were incorporated in the report.
Chapter 10 Synthesis and Integration
10.2 (p. 10.1)
There is much data presentation here. Consider placing these data in Chapter 8, leaving this chapter reserved for
"Synthesis and Integration."
Response: While there are data in Chapter 10, it represents a synthesis of information presented in
Chapter 8. Additional emphasis was placed on synthesis and integration in this chapter to
reduce introducing new information.
As you state, the LNWR as a rain driven marsh system is chemically distinct from other areas in the north to south
flowway. Much of the thrust of this work is looking for patterns among water chemistry variables as they change
along the gradient from northern WCA 2A down through ENP. As WCA 1 and Big Cypress are distinct from this
gradient, perhaps a number of the plots of data might be more informative if these two areas were treated separately.
It seems that one major thrust of your work is to describe the interrelationship between various water quality
variables as they change along this gradient. Omitting distracting or confounding comparisons as caused by data from
areas that don't fit along the gradient should improve the fidelity of the relationships you see.
Response: The Big Cypress information is not included in these plots. The WCA1 data is included
because it does not overlap with the other WCAs and does present a contrast with WCA2
and comparison for ENP.
10.3 Conceptual Models (p. 10-6)
It is confusing in going from 5 to 3 to 6 and back to 3 geographic areas for marsh data presentation and discussion.
Response. The different latitudinal divisions were established because there are different patterns that
emerge from these different divisions. However, it can be confusing for the reader. A more
logical approach for building these different perspectives was investigated and presented.
10.3.1 North of Alligator Alley
One factor you believe to be contributing to low MeHg concentrations in Gambusia here is high TS04 because it (or
it's product, sulfide) can complex Hg and reduce its availability for methylation. However, MeHg concentrations
are in fact high in water, periphyton and soil in this area, suggesting that high TS04 is not reducing methylation rate
sufficiently to explain the low MeHg in fish. Perhaps I am missing the nature of your argument here and it could be
explained more clearly.
A second factor you believe to be contributing to low MeHg concentrations in Gambusia is high TOC because it can
complex MeHg, making it 'unavailable for uptake. However, the fact that periphyton MeHg levels are high, indicates
that the MeHg is available for uptake.
Response: The presence of high methyl mercury in water in WCA-2 in the presence of high sulfate
indicates that MeHg should be available for biouptake, however, the fish remain low. We
think this is largely due to the biodilution effect of high to low (north to south) phosphorus
gradient and its stimulatory effects on primary level biomass production.
4
-------
As regards the low MeHg in Gambusia in the north, it seems to me that rather than TS04 or TOC explaining this,
either the food preferences of Gambusia in this eutrophic area, or low transfer efficiency of Hg from eutrophic
periphyton species to Gambusia, are more likely responsible. (Perhaps Gambusia can't digest the periphyton
occurring in eutrophic areas.) You have little discussion of the fish gut contents data from Appendix C here.
Response: The information on fish gut contents has only been briefly discussed because data are for a
single wet season. A comparison is needed with a dry season sample to allow a more
meaningful interpretation. The primary mechanisms controlling bioaccumulation operate at
the cellular level which has been suggested in Chapter 10, but which is beyond the scope of
this study.
Since most of the biomagniflcation of Hg in the food chain occurs by algae bioconcentrating Hg from water
(BCF = 10M06), and BCFs are affected by CI" ionic strength, and pH, it would be of great value to plot periphyton
BCF vs. pH and CI concentration similar to what you have done for geographic areas. We are planning to support
some experimental work following the lead of a paper by Mason et al. 1996 on this topic. It would be useful to see
if any real-world patterns of interactions between pH and CI' follow Mason's findings.
Response: The bioaccumulation of mercury in mosquitofish is also thought to be low in the north due to
changes in the food chain resulting from the stimulation of pollution tolerant species and a
less complex food web and lower transfer efficiency as you suggest. The pH gradient over
the marsh is quite small. The average pH is circumneutral throughout the marsh, except for
WCA1. We did not measure CI" so this plot is not possible.
Regrowth dilution, do calculations indicate that the increased Gambusia growth rate can account for much of the
lower MeHg levels in the Gambusia?
Response: No. We don't have growth rates for mosquitofish.
10.3.2 Alligator Alley to Tamiami Trail
Increased bioconcentration of MeHg by periphyton due to increased periphyton productivity that provides sites for
methylation is hypothesized. However it's not clear to me that MeHg concentrations in periphyton are higher here,
or that Gambusia are consuming more periphyton, or proportionally more periphyton relative to other dietary items
which could explain the increased MeHg in Gambusia in this area.
Response: Methyl mercury in floating periphyton, soil periphyton, Gambusia and Great Egret chicks
south of Alligator Alley shows remarkable consistency. The production of floating periphyton
also appears to be higher here. The changes in the food chain seem to be apparent from
north to south, however, although some of these changes in the food habits of Gambusia are
indicated to some extent in the one food habits analysis done so far the results are not yet
consistent enough to draw a conclusion. The food habits analysis needs to be repeated and
is planned for Phase II REMAP.
Didn't Gilmour find maximum methylation rates here, but were these in sediment or periphyton?
Response: Gilmour found maximum methylation rates in sediment cores collected from WCA2 and
WCA3A. However, the sediment periphyton were considered part of the sediment in her
studies. Sediment periphyton mercury concentrations were analyzed separately from
sediment mercury concentrations in the EPA study.
5
-------
10.3.3 South of Tamiami Trail
Doesn't Gilmour believe that S04concentration limits methylation rate here, which would affect MeHg in Gambusial
Response: Gilmour hypothesizes that low S04 concentration does limit methylation rates south of the
Trail. Data from the present study suggest that the combination of low S04, TP, and lower
DOC influence Hg methylation rates, but that the increase in food chain complexity also
results in increased BCF's in Gambusia.
10.4 Testable Hypotheses
Table 10.4 (p. 10-11)
For North of Alligator Alley, I suggest adding a testable hypothesis: Elevated TP concentrations alter periphyton
community composition favoring species that have low transfer efficiency of MeHg to fish.
A testable hypothesis relevant to all three geographic areas and worth adding is:
CI" and pH affect MeHg uptake by algae and consequently affect periphyton BCF and fish BAF.
Response: Hypotheses added.
Don Axelrad, Ph.D.
Mercury Program
6
-------
Jerry Stober, Ph.D.
U.S. Environmental Protection Agency Region 4
Science and Ecosystem Support Division
980 College Station Road
Athens, GA 30605-2720
RE: Peer Review of South Florida Ecosystem Assessment Vol 1. Technical Report Review Draft and Vol 2.
Appendices
Dear Dr. Stober:
Enclosed are our peer review comments from the Mercury Studies Program Staff on the subject peer review draft
documents. I apologize for the delay, but due to the scope of the documents and the number of comments generated,
it was not possible to meet your original deadline. I hope that you will still find these comments timely and helpful.
These comments originate with the Mercury Studies Program and therefore only represent our position [and do not
represent the position of the South Florida Water Management District].
The REMAP study was a monumental undertaking. Its conceptual, methodological, logistical, and financial
challenges cannot be overstated. We have also faced such challenges, albeit on a far smaller scale in the ENR Project,
in an attempt to achieve the optimum balance between what is scientifically rigorous and what is reasonably
achievable in practice with the resources available. The trade-offs between scientific rigor and practicability must
be made explicit, however. Whether such trade-offs should extend into the realm of quality assurance (QA) has also
been much discussed by the Mercury Studies Program staff in recent months. In our specific comments we have tried
to convey that unresolved quality control and methodological issues in the South Florida REMAP Study could
weaken some of its key conclusions. From our experience, the application of traditional QA criteria to ultra-trace
mercury sampling and analysis may not always be appropriate. Nevertheless, such changes cannot be made arbitrarily
or in a vacuum. We believe that all participants in the South Florida Mercury Science Program would benefit from
a thorough reevaluation of QA performance criteria for such specialized applications.
Such a reevaluation may best be carried out in the context of an overall reassessment of data to be used to define the
mercury baseline conditions in the Everglades. The reassessment is to be carried out over the next six months in
conjunction with the activities of the Everglades Mercury Baseline Oversight Committee created by the Technical
Oversight Committee Principals at its May 27, 1998, meeting in West Palm Beach.
Let me know if any of our comments require clarification. I can be reached at (561) 682-6749.
Sincerely,
Larry E. Fink, M.S.
Sr. Supervising Environmental Scientist
Mercury Studies Program Manager
7
-------
Comments on South Florida Ecosystem Assessment Vol. I Technical Report Peer Review Draft
By Larry Fink, Pete Rawlik, and Darren Rumbold, Resource Assessment Division, Water
Resources Evaluation Department, South Florida Water Management District
Preface
The REMAP study was a monumental undertaking. Its conceptual, methodological, logistical, and financial
challenges cannot be overstated. We have also faced such challenges, albeit on a far smaller scale in the ENR Project,
in an attempt to achieve the optimum balance between what is scientifically rigorous and what is reasonably
achievable in practice with the resources available. From our experience, the application of traditional quality
assurance criteria to ultra-trace mercury sampling and analysis may not be appropriate, and we believe that all
participants in the South Florida Mercury Science Program would benefit from a thorough reevaluation of quality
assurance performance criteria for such specialized applications. Such a reevaluation may best be carried out in the
context of the reevaluation of the mercury baseline data that affects District permit compliance over the next six
months in conjunction with the activities of the Mercury Baseline Oversight Committee created by the Technical
Oversight Committee Principals at its May 27, 1998, meeting in West Palm Beach.
It is in this context and spirit that we offer our review comments on the on South Florida Ecosystem Assessment
Vol. I. Technical Report Peer Review Draft.
Summary
• General
clarify audience and write consistently to that audience
add Background section
bring conceptual model/initial hypothesis discussion into Study Design chapter
summarize tesselated probabilistic sampling design in Study Design chapter and append the
technical details
document could benefit from more diligent editing for spelling, more complete citations of
key concepts and data not originating with the REMAP study, and more thorough checking
of calculations (e.g., check units on sediment concentrations: ug/cc should probably be
ng/cc)
add decision data quality objectives to performance data quality objectives
• Conceptual Model
reexamine conceptual model regarding the interrelationships of eutrophication processes,
reducing conditions, and methylmercury production in light of the sulfur cycle and potential
deficiencies in the redox measurements using the Eh probe
omission of sulfide from analytes list for surface and pore water severely compromises the
ability of the study design to discriminate the validity of hypotheses in which phosphorus
is the primary determinant of mercury speciation, fractionation, and transformation versus
those in which sulfide is the primary determinant, as mediated by surface and pore water
chemistry influenced by eutrophic conditions fostered by excess phosphorus
• Sampling Methods
justify and validate new methods with standard methods or accepted standards of practice
by competent practitioners, where standard methods do not exist sampling from helicopter
pontoon is not considered an accepted standard of practice in Everglades sampling by the
Everglades Technical Advisory Committee (ETAC)
8
-------
please provide results of study showing equivalence between REMAP and ETAC approved
helicopter sampling method
• Analytical Methods
no pore water chemistry
justify and validate new methods with standard methods or accepted standards of practice
by competent practitioners, where standard methods do not exist
Practical Quantitation Limits for THg and MeHg using these new methods?
What is significance of EtHg if no data are included or analyzed EtHg?
• Quality Assurance
Appendix B omits ORD quality assurance report. Cannot perform thorough evaluation of
data upon which report is based without a complete quality assurance data set and analysis
viz-a-viz performance criteria.
the use of non-standard quality assurance performance criteria should be avoided (e.g., 3
standard deviations of mean of entire data set to evaluate duplicate sample performance viz-
a-viz the use of ±25% RPD or RSD, which is based on paired duplicate sample results)
if non-standard quality assurance performance criteria must be used, include detailed
rationale
no results of paired splits correlation analysis of inter-laboratory comparisons
General Comments
By my recollection, the primary purpose of the REMAP study was to determine the nature, magnitude, extent, and
trends of mercury concentrations, exposures, and risks in South Florida that defined the mercury problem. A
secondary purpose was to determine whether the cause or causes of the mercury problem could be discerned from
an association between the meteorological, hydrological, chemical, and/or biological characteristics of the
environment and the spatial or temporal patterns of mercury speciation, fractionation, and accumulation in that
environment. A more complete presentation and analysis of the data collected under REMAP, including the quality
assurance data, would make it possible to determine the degree to which the REMAP Study has achieved its primary
and secondary purposes. The subject report contains an extensive analysis of mercury species concentration data,
but we could find no preliminary or scoping-level ecological risk assessment. This should be added. Absent any other
guidance, the Great Lakes Initiative Water Quality Criteria to protect wildlife could be used to quantify screening-
level hazards.
Response: The ecorisk assessment planned from the outset by this study has been deferred due to the
need to analyze and interpret the large amount of data collected. Additional sampling is
planned in 1999 to complete baseline assessment of key plant indicators and to initiate trend
monitoring. An initial ecological risk assessment will be based on the interpretation of the
1993-96 information followed by an update when the 1999 data becomes available in addition
to input from studies from other agencies.
An ancillary benefit of the REMAP study was the generation of calibration data sets for the hydrodynamic, water
quality, and landscape ecology models that are being used by agency scientists and engineers to organize, synthesize,
filter, and integrate all of the data gathered to date under the various monitoring and research projects to support
management-relevant predictions about the responses of the Everglades to various stressors under various water
management scenarios. Had REMAP provided no other benefit, it would have justified its cost.
Please clarify the audience for this report: general reader? ... educated layman? ... legislative aide with interest in
environmental issues? ... environmental manager? ... scientist with no specific knowledge of Everglades
9
-------
environmental issues? The content and tone of the Introduction suggests a target audience of an educated layman,
while the following chapter on Study Design requires advanced academic training in statistics to understand. The
other chapters tend to fall between these two extremes. I think the target audience should be an environmental
manager, whose decisions you intend to guide with the information contained in the report. An educated layman's
version can be prepared for a subsequent 4-page, 3-color fact sheet. The detailed statistics lesson in Chapter 2 might
better be appended to the document, reducing the discussion in Chapter 2 to a management-level summary of the
approach and its benefits over other approaches, with a reference to the appended scholarly discussion.
Response: The target audience is the environmental manager as well as the scientist. The detailed
statistical discussion up front is to provide the information on the statistical survey design
which is frequently misunderstood. A summary of the main statistical design points will be
added to the text.
Please summarize key background discussion in one paragraph for the Introduction chapter and add a new
Background chapter that summarizes what was known about mercury sources, disposition, fate, bioaccumulation,
and effects at the time of study design. This will allow the highlighting of the contributions of the REMAP study to
our knowledge base in Chapter 10.
Response: Chapter 1 has been rewritten to better summarize key background issues. A background
section on mercury has been added to Chapter 8. This avoids having to revise all subsequent
table and figure numbers, while providing a reference of REMAP contributions in Chapter 10.
Citations should be provided to support characterizations of the system using the results of studies other than this
one. There are a number of instances where such citations have been omitted (e.g, meteorological, hydrological,
geological, and habitat data).
Response: Additional citations have been added.
Specific Comments
1.0 Introduction
1.4.1(1-7). Purpose of this Report. Move to beginning of this chapter to assist reader in appreciating why the
report is being written sooner than later.
Response: Purpose moved.
1.3.1 "Clearly, the greatest change in the Everglades ecosystem is the change in the natural hydropattern..."
Debatable. One could make a stronger case that greatest change is the loss of between 25 and 50% of the
Everglades acreage.
Response: Text has been modified.
1.3.2 (1-3). Florida Bay advisory was issued in October 1995, not in 1994.
Response: Date changed.
1.3.2(1-3). The statement of the relative contributions of the various exposure pathways mercury exposures
mixes mercury species. There is virtually no MeHg in air or drinking water. Nearly 100% of MeHg
10
-------
exposure occurs via food, and more specifically, fish. For elemental mercury, most exposure occurs
via air, not water or food. For Hg(II), most exposure occurs via air associated with particles and
drinking water. If mercury is treated as one entity, the statement is probably accurate. However, it
is not accurate to treat all mercury species as one entity, as their sources, fates, BAFs, exposure
pathways, and toxicities are very different.
Response: Text was modified to reflect different exposure pathways for different mercury species.
1.3.2( 1 -4). What vast areas of Hg contamination? Hg in water and sediments is within normal ranges for aquatic
ecosystems. Its MeHg in water and more importantly in biota that is higher than background over
a vast area of South Florida. Please clarify
Response: Text was clarified to reflect the appropriate mercury species and media.
"While Everglades Hg contamination ... since 1989, the sources, etc., remain unknown." This
statement seems out of date. Perhaps it was written in 1992 for the original proposal or work plan
for the REMAP study and was overlooked in the early editing. It ignores the important contributions
made by Delfino et al. (1993) and more recently Robbins and Keeler (1997) on the rate of deposition
of mercury as a function of time, the Florida Atmospheric Mercury Study (FAMS) (see for example
Pollman et al. 1995; Landing et al., 1996; and Guentzel 1997) to our understanding of the Hg(II)
deposition via bulk rainfall, and the screening-level calculations of Landing (1996) on the potential
role of reactive gaseous mercury deposition in making up the difference between the measured
deposition rate of Delfino et al. and Robbins and Keeler (1997) and the bulk rainfall rate. Keeler and
his students are now making important contributions to our ability to quantify local source-derived
wet and dry deposition to the Everglades. It also ignores the contributions of the USGS ACME team
to our understanding of the factors that are controlling net methylmercury production,
bioavailability, and bioaccumulation, especially the role of excess sulfate derived from EAA runoff
and perhaps acid rain deposition in determining the location and magnitude of the sulfide
concentration gradient in the Everglades, the modeling of Hg(II) fractionation, and support to
USEPA/ORD's efforts to refine an existing mercury cycling model for application to the Everglades.
Response. The text was clarified to indicate that atmospheric emissions, transport and deposition
represents a major source of the mercury, and, perhaps, the source of mercury, but the
cycling of mercury through the system and the fate of mercury in peat is still an unknown.
There has been up to 2 m of peat lost over the past 50 years. It is also unknown what
proportion of the atmospherically deposited Hg is methylated and accumulated through the
food chain.
Although REMAP made limited contributions to our understanding of the fate, bioaccumulation, and effects of
mercury in the Everglades, it did make important contributions to our understanding of: (1) the semi-annual spatial
dispositions of THg and MeHg in water, sediment, periphyton, and mosquitofish in the Everglades that reflect its
distinct ecophysiological regions, (2) the present-day insignificance of THg and MeHg in EAA runoff to the new
annual loads to the Everglades, (3) the inverse relationship between the peat accretion rate and the THg concentration
in peat, (4) the absence of an inverse relationship between phosphorus in water and mercury in mosquitofish in the
southern Everglades, and (5) the strong correlation between mosquitofish hotspots and wading bird hotspots in
WCA-3A.
Response: Contributions incorporated in the text in Chapter 10.
11
-------
I would modify the above identified problematic statement to say that "While ..., various degrees of uncertainty
still remain in our quantitative understanding of sources, etc. that limit our ability to make fully informed
management decisions at this time. These uncertainties are expected to be reduced substantially over the next two
years as the monitoring, research, modeling, and assessment data collected over the last five years are further
analyzed, synthesized, and integrated into a more complete, accurate, and precise predictive quantitative model. The
REMAP studies are expected to make an important contribution to this effort."
Response: This statement was substituted for the problematic statement in Section 1.3.2.
1.3.3 (1-5): "A combination of agricultural. . . is being implemented in an attempt..." I would use "natural"
rather than "historic" levels.
Response: Natural was substituted for historic.
Strategy for Study Design
In a fixed or static study design, samples of specified type and volume are collected at specified locations and times
and characterized and analyzed for the constituents of interest. The study is carried on for a specified period of time.
The data are analyzed, synthesized, and integrated into an understanding of the problem, and from that understanding
a preferred solution emerges that is eventually adopted and implemented. In an adaptive or dynamic study design,
the study evolves over time as the data collected in the early or reconnaissance and scoping phases is analyzed,
synthesized, and integrated into a more complete, accurate, and reliably conceptual model of mercury sources,
cycling, and exposures in the Everglades. This more robust conceptual model then forms the basis for a more refined
set of management questions to be answered by the study and a more refined study design to answer those questions,
and so forth, until the questions are answered to the satisfaction of the resource managers. The REMAP study appears
to have followed the static study design strategy. This should be stated explicitly.
Response: The study design is clearly described in Chapter 3. There is no need to characterize the
design as "static" versus "adaptive". In particular, the study was designed around
management questions, not around specific hypothesis regarding Hg. The design was
intentionally focused on achieving broad population coverage to provide some power for
addressing a variety of hypotheses.
Conceptual Basis for Study Design
Hypotheses to be Tested: Rather than rely solely on management questions to introduce the
reader to the study design, I think it would be more appropriate to bring the initial hypotheses to be tested from
Table 8.1 (8-22) into Chapter 2.0.
Response: The study was not designed around the null hypothesis. It was designed around the
policy/management questions.
Conceptual Model: Perhaps most notably absent from Chapter 2 is a conceptual model that includes the transport,
disposition, transformation, accumulation, and effects of mercury species in the Everglades that warrant study). I
think it would be beneficial to bring the initial conceptual model displayed in Figure 8.2 (8-23) into Chapter 2.
Response: The conceptual model is directed specifically at the mercury issue. The project and the report
is directed at an assessment of the ecosystem. The conceptual model is most appropriate
in chapter 8.
12
-------
Data Quality Objectives: Not defined until much later in the document and not complete and definitive. What are
the acceptable limits of quantitation and detection of each factor? How were these determined? With what resolving
power, probability, and confidence level must the study design be able to discriminate true differences in a factor
between any two cells? any two times? For adjacent cells or times, this ability to resolve true spatial or temporal
differences will determine the minimum gradient or trend that can be quantified accurately, precisely, and reliably
at the sampling grid scale and time step adopted.
Response: Data Quality Objectives are documented in a separate report, which is referenced in this
report. Additional information can be obtained from that document. In addition, an analysis
of variance was conducted for each variable collected as part of the REMAP program to
assess the within site versus among site variance estimates. The EMAP goal was to have
within site variance estimates about 10% of the among site variance estimates For almost
all the variables except fish, this goal was achieved. For Gambusia mercury concentration,
the within site variance was 14% of the among site variance. Future sampling efforts will
collect and analyze 7 individual fish per site rather than the 5 fish collected previously. This
will reduce the within site variance to 9% of the among site variance and satisfy the EMAP
goal. Finally, EPA Office of Research and Development is in the process of preparing DQO
guidance for research programs such as the REMAP project because the DQO guidance
documents under which the REMAP DQOs were developed are not applicable for research
programs.
Grid Scale: Reason for selection not defined. The unbiased statistical sampling model was adopted as the basis for
the design of this study. It assumes no a priori knowledge of the spatial distributions or interrelationships of the
factors to be studied. Therefore, one cannot determine a priori the appropriate grid scale to overlay upon the system
of interest to ensure that nothing significant is missed in terms of differences between sites. As an alternative, one
could have made the grid scale a random variable to test the hypothesis that the selected grid scale was inappropriate
for each factor of interest. This would have required random subsampling at randomly selected finer and broader grid
scales, until the desired resolution emerged for each factor of interest. This, of course, presupposes that well-defined
data quality objectives had been adopted with which to discriminate appropriate from inappropriate grid scales. This
was not the case. This also presupposes that there was enough time and money to conduct the required scaling
scoping studies to achieve this objective. This was also not the case.
Response: The approach suggested by the reviewer is not feasible The time and resources required
to mount the study suggested by the reviewer are enormous. Furthermore, the reviewer
assumes that an "optimum" design exists, which is not the case. A general rule is that a study
cannot be simultaneously optimized for more than a single objective. This was definitely a
multiple-objective study, so the best we can do is a design that addresses all our objectives
with an acceptable degree of uncertainty.
The design relies on several fundamental principles to achieve "acceptableness":
(1) Design-based inference and variance estimates must be possible This is achieved by the
choice of the RTS design.
(2) We assume that the phenomena being studied all have spatial pattern, in the very general
sense that two locations near one another are more likely to resemble one another that two
locations far apart.
(3) Major results will be reported as a distribution quantile, e.g., "x% of the resource population
is in classification y".
13
-------
(4) More samples are always better than fewer, but 50 is enough to give acceptable confidence
intervals for resource condition statements as in (3). With a sample size of 50, the worst case
(estimating the 50th percentile of a population with no spatial pattern) confidence interval will
be no more that + 12 percentage points.
(5) Given a spatial pattern as in (2), there is a result in spatial statistics that implies that regular
point spacing is better than a purely random sample. Some of the relevant literature is cited
in Section 2.1.1. The result is model-based and not strictly applicable to design-based
estimation. Moreover, strict-systematic spacing does not provide a design-based variance
estimate. (In the terminology of classical statistics, there is only one degree of freedom, so
no variance estimator is possible.) The RTS does provide a variance estimator, and it can
be shown, again using model-based methods, that it captures most of the efficiency of a strict
systematic design. These results are in the references cited in Section 2.1.1.
As there was insufficient time and money to carry out such studies, the choice of grid scale was inherently arbitrary,
based on what could reasonably be expected to be sampled and analyzed with the available transport, staff, and
analytical resources over a period of time considered short relative to the response times of key transport, fate, and
bioaccumulation processes. This is touched on the introductory discussion in Chapter 2. That's good. It could be
made even more explicit, however.
Time Scale: Semi-annual sampling corresponded to the late spring and early fall, or the late dry season and the late
wet season in subtropical South Florida. The timing allowed one to detect differences in periods of rapid plant growth
in the late spring but not the rapid onset of senescence in the late fall or early winter. The maximum external loadings
of new mercury to the system occur in the mid-summer and are minimal in winter. In this context, perhaps quarterly
sampling at half as many sites may have provided greater insight into the temporal dynamics without losing
significant information about spatial distribution.
Response: We were not attempting to establish or study temporal dynamics. The two sampling times
were picked to examine the system at two extreme hydrologic regimes. The design does
provide limited information about temporal dynamics, but that comes as a side-benefit, not
as a factor that controlled the design.
Splitting the total effort into 4 quarterly cycles per year, as suggested by the reviewer, is not
desirable. It changes the objective of the study, and substantially dilutes the spatial resolution
of the sample. Estimates for a subpopulation that are feasible with 30-50 samples are not
with only 15-25 samples.
What were key transport, fate, and bioaccumulation processes and their response times with which to determine the
appropriate time scale for sampling? Perhaps a table with the ranges of response times of the various components
of the system would help guide the reader in evaluating the logic of a 10-day period for synoptic sampling.
Response: The synoptic sampling period was a function of how rapidly sampling could physically be
accomplished over the 9600 km2, rather than based on time and length scale analyses of
processes and transport times. The 10 day period was the maximum time to sample the
marsh from the mangrove fringes to Lake Okeechobee (i.e., south to north). The last
sampling cycles were conducted within 7 days. It might be useful at a later date to consider
time-length scale analyses, but this information was not available for inclusion in the report.
Process and transport studies relate to different study objectives than those of the REMAP
project.
Factors of Interest: The conceptual model should have guided the choice of factors of interest. It is not possible to
determine if this was the case from the organization and content ofthis report. Perhaps the choice of these key factors
14
-------
derived from the literature review prepared by ORD-Athens. If so, 1 would recommend that it be distilled down and
added to the Background chapter in the report or append it for ready reference.
Response: The conceptual model did guide the choice of factors of interest. The conceptual model was
developed by Dr. Ron Jones prior to the REMAP proposal which preceded the ORD literature
review by a couple of years.
In addition to pointing out what was included in the study, it is also important to point out what was excluded. For
example, sediment pore water was not sampled and analyzed for THg and MeHg, not because such information was
not thought important in defining the disposition of factors that determine the susceptibility of an ecosystem to
methylmercury production, transport, and bioaccumulation but rather because the laboratory used by USEPA for this
work had no such capability at the time of the study. Also excluded was an estimate of the flux of THg and MeHg
from the sediment to the overlying water, because sediment strata were not analyzed separately. Only a sediment
composite was analyzed. While this may have been sufficient for calculating the magnitude of storage compartments
and saved significant cost, it lost significant information. Perhaps the random selection of a subset of cores for such
more detailed analysis might have added to the conceptual power of the study without increasing its cost
dramatically.
Response: Sediment pore water was not sampled for total and methylmercury because a reliable
sampling method which would produce the required volume of uncontaminated pore water
in less than one hour could not be designed. The sampling requirement remains a challenge
with the above constraints.
The original study plan proposed sampling soil at the surface, 25 and 45 cm depths. This
strategy was tested on the transects with frequent loss of one or both of the deep samples
along with complications resulting from compression of the deep cores. Since only the top
10 cm of soil could reliably be collected a composite sample with the periphyton layer
removed was used as a standard sample. In hindsight it might have lost some information
on the strata, however, at the time there was no spatial information describing the mercury
species in soil.
2.1.4 Variable Probability Estimation
There are a number of other probabilistic models for distributing sampling resources than the tesselated probabilistic
sampling design. Why were these rejected?
Response: The comment does not pertain to Section 2.1.4, rather, it questions the choice of the RTS
design as opposed to some other design. The rationale for selecting the RTS design is
detailed in Section 2.1 Design Rationale and Section 2.3 Design Summary. The response to
the question on conceptual basis for the design also addresses this comment.
The tesselated probabilistic sampling design is only one component of the statistical basis for study design. Without
statistical DQOs for hypothesis testing, it is not possible to determine whether the study design was adequate to
resolve statistically significant differences in a factor between sites or times. The ability to discriminate significant
differences between sites is important in locating hotspots, identifying and quantifying gradients, and spatial
correlation analysis relating changing magnitudes of factors to changing states and vice versa. The ability to
discriminate significant differences between times is important in quantifying transport velocities and response times,
in identifying and quantifying lag correlations between states, factors, or both, and in identifying and quantifying
temporal trends.
15
-------
Response: The study was not specifically designed to test hypotheses, but rather address management
questions. As stated earlier, there was an EMAP guideline to have within site variances
i 10% of the among site variances. This was achieved for almost every constituent. The data
permit hotspot identification, gradient quantification, and spatial correlation analyses. The
systematic selection of probability sites is a design feature that contributes directly to these
capabilities.
3.0 Materials and Methods
Figure 3-1(3-9) was there a need to develop new methods for THg or MeHg in any matrix, when such capability
existed at Frontier Geosciences, Brooks Rand, and Battelle Northwest? The PQL and MDL for THg
and MeHg in water for the Bloom method (Bloom and Fitzgerald, 1988) were 0.2 and 0.05 ng/Land
0.02 and 0.005 ng/L, respectively, while those adopted in this study were unspecified and 0.2 ng/L
and unspecified and 0.01 ng/L (sulfhydryl cotton method), respectively.
Response: At the time of initiation of this project there was no approved USEPA standard method for
ultra trace level analysis of either THg or MeHg analysis in any matrix. F1U-SERP was under
contract by the USDI, NPS, Everglades National Park to develop an ultra trace level
analytical capability. USEPA Region 4 was planning a large REMAP project which was going
to generate a large volume of samples from intensive field surveys in multiple matrices and
it was clear that in order to meet holding times additional laboratory capability would be
needed preferably located in South Florida. In an effort to meet the high sample through-put
required by the project, emphasis was placed on flow-through methodology. Dr. Ron Jones
(FIU-SERP) was approached as a primary cooperator both because of his detailed
knowledge of the Everglades ecosystem and his ability to develop new analytical methods
on a wide variety of matrices and parameters. However, due to the lack of USEPA standard
methods a three laboratory design was developed for the project for QA/QC purposes and
Battelle Marine Sciences Laboratory was included. During the initial pilot studies it was found
that some existing methods were not sensitive enough to provide the required unqualified
concentration values for MeHg in soil/sediment and additional analytical methods
development was carried out by FIU-SERP. The PQL and MDL for Thg and MeHg in water
for the methods used in this study were 1 ng/L (USEPA OQA) and 0.3 ng/L and 0.008 ng/L,
respectively.
3.2 (3-11) No discussion of why a new method was needed for the determination of MeHg in water samples.
If it is only to allow for the identification and quantitation of EtHg, one must ask whether there is
any significance to the presence of EtHg in the environment? Are there other organo-mercury
compounds that are markers of the key processes that transform mercury species by microbially
mediated processes? Questions like this might have been answered with the development of a
conceptual model of mercury cycling in the Everglades. EtHg is nowhere subsequently discussed
in the report.
Response: A new method was not needed for the determination of MeHg in water samples because it
was the primary project responsibility of the Battelle Marine Science Laboratory to analyse
MeHg in water using the Bloom method throughout this project. Only after the need for a
more sensitive analytical method for the analysis of organic mercury in soils/sediments
became evident was a new method developed which lead to the subsequent discovery of
EtHg. The discovery of EtHg was an unexpected benefit from this project, however, it has
been reported in several European mercury studies and the Bloom method cannot detect
EtHg. EtHg has not been discussed in this report because we have left this aspect to the
academicians who have published numerous articles (e.g., Alii et al., 1994, Jones et al.,
16
-------
1996; Cai et al., 1996; Cai et at., 1997; Cai et a!., 1998) on analytical methodology and the
presence of EtHg in the Everglades ecosystem. Phase II REMAP will include analyses of
additional plant materials in the Everglades ecosystem which may develop a more complete
understanding of EtHg in this system. Once a new organic mercury method was developed
it was applied to water and found to compare very well with the results for MeHg obtained by
Battelle MSL.
The helicopter sampling method chosen is inconsistent with that adopted by the SFWMD and the Everglades
Technical Advisory Committee (ETAC). While there is a time and cost rationale for this difference, it should be
made explicit.
Response: The Everglades Technical Advisory Committee (ETAC) served as a technical forum for
development of a standard method for field sampling shallow Everglades surface water.
FDEP subsequently published the outcome as a recommended method, which SFWMD
adopted for their field sampling efforts. The method involves walking away from the
helicopter to dip a marsh surface water sample. These samples were to be analyzed
primarily for phosphorus. This field protocol is not an appropriate clean sampling method
for trace level determination of mercury. During development of the marsh sampling protocol
the dip method was tried both in the immediate vicinity and away from the helicopter. It
became apparent that the dip method had several intrinsic problems: 1) It was impossible to
minimize the particulate matter in the sample 2) it was not possible to fill the bottle with zero
head space if the water depth was less than 6 inches due to the diameter of the 2 L teflon
bottle; 3) neither of these problems facilitated a clean sampling protocol for mercury; and 4)
samples collected under the helicopter when compared to those collected away from the
helicopter showed no difference in mercury concentrations nor did the field blanks show any
mercury contamination. However, a formal study to compare the two methods would have
been confounded by the high particle loading in the dip samples. It was clear that a clean
sampling device designed to correct these problems as well as allow sampling very shallow
water (approx. 2.5 cm deep) was needed. The vacuum sampling device with a 105 urn mesh
screen was developed and used in this study. The time and cost rationale was of secondary
importance With the above significant limitations the wade and dip method was not a useful
method consistent with obtaining clean water samples from the marsh unimpacted by the
presence of the sampling effort
3.1 "Water samples collected at various distances from the helicopter as well as the field blanks indicated no
contamination of the samples was evident." Include these study results.
Response: The dip method of sampling was used during the first canal (pilot) cycle and the transect
(marsh pilot) study which fully demonstrated the limitations of the method. The vacuum
sampler was used for all subsequent sample cycles in the canals and the marsh. A
comparison of the pilot study data with the subsequent vacuum samples showed no
significant changes in the mean and median concentrations, however, removal of large
particles may have reduced total mercury in some samples.
Concerning soil slurry measurements taken in the laboratory. Measurements of temperature, pH and Eh on
samples transported in coolers, in helicopters, exposed to the atmosphere, mixed, cleaned of large debris and
then mixed with deionized water are not representative.
Response: These measurements were part of the laboratory protocol in the preparation of soil samples.
These measurements were not taken to describe the environment from which the samples
were taken but to develop an internally consistent set of measurements in the laboratory
which could detect potential differences during processing.
17
-------
"Soil pH and Eh along the marsh transect were measured on site by inserting an Eh electrode into an intact
soil core." This is not a standard method. Compression of the sample by the soil corer will destroy existing
stratifications. Furthermore, even though the core is encased in the corer diffusion of air into the sediments
is rapid and will result in oxic measurements that are dependent on exposure time.
Response: This method was only used in the transect pilot study.
"The in situ Eh probe was inserted into the marsh soil..." Contradictory to previous statements. Was the
Eh probe inserted into the marsh soil or the marsh soil core?
Response: During the marsh transect (pilot study), an Eh probe was inserted into an intact soil core This
procedure was eliminated in subsequent marsh sampling in favor of the in situ Eh probe. The
in situ Eh probe was inserted into the marsh soil following the pilot study. A design drawing
of the probe system is presented in the appendix.
"This probe was allowed to equilibrate for 15 minutes. . ." This is insufficient time for equilibration. See
Koch-Rose et al. (1994).
Response: The probe was tested in the marsh over time and little change was found after 15 minutes
had elapsed.
"Water samples were collected byfilling bottles underwater at each site. "Unclear. Is this is a deviation from
canal operations? If not, it is inconsistent with previous statements. If this is a deviation for canal operations
this means that the canal samples were not screened for large particulates.
Response: As described above the dip sampling method was used on the first canal and transect pilot
sampling efforts. The inclusion of large particulate matter especially in the marsh water
samples resulted in great difficulty in obtaining comparable sample splits. The vacuum
sampler was developed with a 105 pdm mesh intake screen to remove the large particles in
the water while allowing the considerable fine particulate load to be included in the sample.
"SoilpHandEh were measured by inserting the respective electrodes into an intact core collected at the site.
Soil cores were sampled to a depth of 45 cm along the transects." Unclear. This description comes after the
deviations statement and therefore could be describing the canal sediment samples but the ponar dredge
would not bring up soil cores. The use of the word "transects" implies that these may be marsh samples
rather than canal samples. The description of inserting the probe into the core is consistent with some
previous statements but contradictory to others.
Response: The appropriate changes in Section 3.1.4 Sampling Routine were made to alleviate any
confusion.
Pg. 3-4. Value of Eh readings of soil mixed in a pan with DI water under laboratory conditions. What does
this measure?
Response: It was an effort to determine if any changes could be detected at this level of sample
processing. None were and the in situ data were used exclusively in the analysis.
18
-------
2.1 Laboratory Analyses
The section presents several problems. The MDLs are stated up front as 0.3 ng/L for THg and 0.02 ng/L for
MeHg. Later on in the text the description of the THg method notes the limit as 10 ng/L without
preconcentration. The next statements are presumably concerning the process of preconcentration, but no
final MDL is stated. Later on in the text the MDL for MeHg is given as 0.01 ng/L for the sulfhydryl cotton
method. Maybe a table with method name, performance ranges, and date at which it was brought on-line
would be helpful here.
Response: Text in the Laboratory Analyses section was clarified to alleviate this confusion.
2.2 QAJQC
A summary of the lab and field QA protocols would be helpful here, especially as regards bottle blanks, trip blanks
and spikes, field blanks, equipment blanks, and field duplicates.
Response: The text in the QA/QC section indicates that the QA/QC evaluation of the database is
attached as an appendix to this report. The QA/QC samples which were analyzed are
included in the database which will be uploaded to the Internet.
Appendix B does not define decision DQOs, only field and laboratory performance DQOs.
Response: Correct. Decision DQOs were not defined for the project.
The absence of the ORD QA Report from Appendix B presents severe problems for the reviewer. With what
frequencies were accuracy, precision, and completeness criteria violated in each study cycle? If the frequencies were
unacceptable, what corrective actions were taken? What were the new frequencies of violation in the remaining study
cycles after the corrective actions were implemented? If samples were split with an outside lab, what was the
correlation relationship between the study and reference lab? If the correlation was unacceptable, what corrective
actions were taken?
Response: The ORD report recommended that the EPA Region 4 Office of Quality Assurance (OQA)
conduct a review of all data obtained from contract laboratories. The Region 4 OQA report
was in preparation at the time the draft Technical Report was distributed for review and is the
reason it was not included at that time. The QA/QC report is now final and has been included
in Appendix B The QA/QC report also will be uploaded to the Internet along with the data set
as a metadata file.
4.0 General Characteristics of the Water Regime
There were no multivariate analyses performed.
There was no lag correlation analysis performed.
Response: The purpose of this section was not to conducted extensive analyses on the water regime,
but rather to provide the reader with some perspective on the general flow patterns and wet
and dry season characteristics during the sampling period. The SFWMD and US Army Corps
of Engineers, as well as others, have, and are, conducting exhaustive analyses of the water
regime. These studies should be consulted if detailed information is desired on hydropatterns
within the water conservation areas or the Park.
19
-------
Pg. 4-1.
Table 4.1 is out of order. Pg 4-7.
Response: Corrected.
Flow path analysis should be expanded to include information on topography, vegetation types and densities, as well
as water chemistry of tracers like chloride, sulfate, and conductivity. Here it would be very helpful to add District
maps of the types and distributions of vegetation and USGS maps of Everglades topography to enhance our
understanding of the relationships between water depth, vegetation communities, and water flow path. Overlay of
the most probable flow path of each associated variable should provide a probabilistic contour of most probable flow
path.
Response: The purpose of the general characteristics of the flow regime section was to provide the
reader with perspective, rather than an exhaustive review and analysis of the flow regime.
While the suggested information would be useful, it is beyond the scope of this report to
include.
4.2.4 This section compares surface DO measurements with bottom DO measurements. However the methods
refer only to a single-depth measurement of DO (page 34).
Response: The text has been modified to indicate the mid-depth sampling was conducted in the marsh
and DO profiles were taken in the canals. Surface and bottom DO concentrations were
compared at canal sites.
2.3.1 Temperature
Based on the timing and duration of temperature measurements at a given site, the change in water
temperature from a minimum just before dawn and a maximum in the afternoon is missed. It is not valid to
compare temperature results from different sites collected at different times of day. Furthermore, the diurnal
temperature change is mediated by vegetation type and density, so any site intercomparison must take into
account these differences, as well.
Response: Diel temperatures, indeed, change significantly at a specific site. However, the temperature
comparisons in the report focused on differences in median temperatures between the wet
and dry season and not differences among individual sites. The focus was on seasonal
differences among water conservation areas.
4.3.4 This section describes the dependence of DO on a diel cycle, but does not properly caveat its conclusions
concerning low DO measurements in the marsh. Furthermore the statement.. it is improbable that many
native Everglades species are tolerant of prolonged oxygen depletion ..is open for debate. This statement
should be limited solely to fish, not reptiles or amphibians, and further caveated with the observation that
several species of Everglades fish species are surface gulpers, which enhances their ability to respire under
low DO conditions. Thus, fish like the gar may be at a competitive advantage over fish species less tolerant
of low DO conditions.
Response: The referenced statement was taken from the McCormick et. al. (1997) citation, as noted in
the text. A sentence was added to caveat this conclusion for fish only.
4.3.5 The conclusion that turbidity is inversely related to water depth could be more an artifact of the way the
samples were collected (by helicopter within a 15-minute period) than a true characteristic of the system.
20
-------
Response:
The paragraph following the paragraph referenced above states that the elevated turbidity
at the shallow water sites might have been disturbance due to the intake probe when
sampling very shallow waters.
Pg. 4-33. Conductivity in WCA2-B is shown to be high, but consistent with surrounding areas. This is
noteworthy given the fact that this area is isolated and its hydrology is rain and seepage driven.
Response: So noted.
5.0 Habitat
Here it would be very helpful to add District maps of the types and distributions of vegetation and USGS maps of
Everglades topography to enhance our understanding of the relationships between water depth, vegetation
communities, and mercury contamination, if any.
Response: This chapter describes the habitat characterization conducted as part of the REMAP project.
The purpose was to describe population attributes rather than individual site characteristics.
5.1 ". . . . cattail (Typha domingensis), a highly invasive species that outcompetes native species..." Provide
citation. Change to (Typha spp.) To account for latifolia and hybrids. Change "native" to "more slowly
growing species adapted to the low-nutrient environment of the unimpacted Everglades"
Response: Changes made.
Pg. 5-4. The L67 canal is discussed in third paragraph (other canals are ref. in text elsewhere), but canals are
not labeled in any of the maps. Please label canals for quick reference.
Response: The L67 canal and other major canals have been labeled.
6.0 Soils
6.1 Soil redox (Eh) was measured along the transcets [typo]. Comparison of soil redox data among transects,
however; is not appropriate because the sampling methods used differed from transect to transect. "Why
did sampling methods differ from transect to transect? If methods were not comparable, one might consider
removing Eh from data reporting and analysis.
Response: Transect sampling methods did not differ from transect to transect (see page 3-4). The soil
redox method for the transects did differ from the marsh grid soil redox sampling method.
Pg. 6-5. The report should define Everglades Protection Area for general audience.
Response: Agreed. The area is defined.
Pg. 6^5. With regard to percent organic matters and bulk density: is this data for top 10 cm?
Response: Yes.
Pg. 6-5. Be consistent with landmark references, i.e., switching from Alligator Alley to 1-75.
Response: Noted.
21
-------
Pg. 6-6. The redox condition in soils is of course critical to the biogeochemistry of Hg as well as other
parameters. Reducing conditions are defined relative to the hydrogen electrode, such that negative
values are reducing and positive values are oxidizing. The activity of the hydrogen electrode is a
function of pH, so it is important that Eh is standardized for pH-E7. In addition, general statements
such as" the presence of Eh less than 100 mV indicates anoxic reducing conditions are occurring..."
need to be supported with citations. We know of no such citations. Sulfate is generally reduced by
desulfovibrio bacteria under reducing conditions in the range -30 to -100 mv (ref?).
Response: The point at which no available molecular oxygen occurs in the soil is around 100 mv When
oxygen is depleted those microorganisms thriving under anaerobic conditions will thrive,
further reducing the Eh.
7.0 Nutrient Conditions
7.1 ... TP has been hypothesized as being one or several variables that influences mercury methylation
processes in the Everglades ecosystem.,' What kind of influences are being referred to here? Direct agnosim
or antgonism or indirect influences via DO, biomass turnover, and the production of DOM, colloids, and
particles? Also, it would help to refer reader to the section of the report where this hypothesis is evaluated.
Response: The hypotheses are that TP might have an indirect influence by increasing DOM production,
stimulating periphyton production (internal environment for methylation), stimulating microbial
decomposition and creating an environment suitable for methylating bacteria, and other
indirect influences. This hypothesis was not evaluated as part of this report. Intensive, site-
specific studies, such as those being conducted by the USGS ACME team are needed to test
these hypotheses. REMAP can assist in identifying where these studies might be conducted.
'An extensive characterization of nutrients... was initiated." The Methods section contains no mention of
analytical methods for nutrients, chlorophyll or APA. Please add.
Response: Citations to publications and methods have been added to the methods section in Chapter 3.
8.0 Mercury
8.1 "The sources, extent and magnitude, transport, transformation, and pathways ofHg through thew Everglades
ecosystem are poorly known." Is this intended to apply to the time prior to study design or still. If the latter,
see papers by Landing, Gill, Krabbenhoft, Hurley, Cleckner, Gilmour.
Response: The statement was intended to apply to the time before the comprehensive mercury studies
began.
"Among the possible Hg sources in south Florida are. . . and agricultural operations." How were these
possible sources identified? For example, were there high mercury concentrations in EAA soils? ... high
transport rates of EAA soil particles? What are the background concentrations in Lake Okeechobee water
upstream of the EAA? Study design was deficient in this regard.
Response: This is a general statement intended to indicate possible sources of mercury to the system.
There is mercury in Everglades soil. We don't think the study design was deficient by not
including Lake Okeechobee water. The study monitored the four main structures biweekly
for three years to determine the seasonal concentrations of total and methyl mercury in the
22
-------
water leaving the EAA. There is also no evidence of a mercury problem in fish in Lake
Okeechobee.
"Various hypotheses have been put forward to account for the apparent susceptibility of the Everglades to
Hg impacts . . Reference SFMSP Science Plan.
Response: This reference has been included along with USEPA, 1993.
"For fish hg contamination to reach concentrations that have ecological and human health consequences,
three factors must exist: (1) Hg source; (2) unique combination of environmental conditions; and (3)
bioaccumulation and biomagnification through the food chain." Split thought and deal with human health
and ecological health in separate sentences. For ecological health: Capitalize Hg. Change "and" to "or".
Change "factors" to "conditions". (1) might be interpreted to imply anthropogenic "Hg source". Natural
mercury deposits together with favorable conditions of natural or anthropogenic origin can create an aquatic
mercury problem. Change (I) to "The presence of mercury in locations, forms, and concentrations accessible
to natural aquatic bacteria." (2) is not accurate. Might better delete "unique" and add after "conditions" the
phrase" favorable to a high rate of net methylmercury production and bioavailability." Also, there should
probably be a fourth and a fifth factor for consideration: (4) There must be a significant rate of exposure to
the contaminant by consumption of contaminated food. (5) One or more species of sensitive wildlife
consuming contaminated food at normal rates accumulate methylmercury to toxic levels.
Response: Text changed to reflect this suggestion.
8.2 Initial Conceptual Mercury Cycling Model
A summary of this discussion should appear in Chapter 2.0 Study Design.
Response: A summary has been included in Chapter 2.
Page 8-4. "Organic soils and sediments, such as those found throughout Everglades wetlands, would be
expected under appropriate conditions to sequester Hg. The natural processes, by which this occurs,
however, have been altered by water management and other anthropogenic activities in South
Florida..." What is meant by sequester? ... bind irreversibly such that it is not available to
methylating bacteria? This implies that the efficiency of sequestration has decreased over time and
that this is why there is more Hg available for methylation. I don't think anyone knows what the
sequestered fraction was under historical conditions vs today. However, if the increase in sulfur in
the Everglades environment has increased the incorporation of sulfur in plant and microbial proteins,
or sulfide concentrations in the soil pore waters, then the ability of present day soils to sequester Hg
may be greater than historically. Eutrophication has increase the peat production rate, in which case
the naturally deposited mercury is being biodiluted. However, associated with this higher production
rate are higher concentrations of DOM, that can prevent the formation of or resolubilize insoluble
mercury sulfides. This would decrease sequestration efficiency of soils. To support this simple
statement you have to get into the heart of the thermochemistry and kinetics of the mercury cycle,
which is well beyond~ the scope of this study and won't be sorted out for several more years.
Haven't mercury loadings also increased during this period? This has somewhat offset the effect of
biodilution. Due to the complexity of the underlying logic of this statement, I would delete as either
unnecessary or unprovable. Or perhaps could rephrase as a hypothesis to be tested in process studies
to be conducted by others.
23
-------
Response: The use of sequester in this context means burial of mercury below the soil recycling zone.
The remainder of the comment goes well beyond the general intent of the statement.
Pg. 8-4. The interaction of the sulfur cycle and sulfate-reducing bacteria with the chemistry of Hg present
numerous questions for understanding the processes affecting Hg contamination in the Everglades.
"Why stop here? Present some of the questions and discuss how the sulfur gradient runs parallel to
the phosphorus gradient.
Response: The interactions between sulfur and mercury intimately involve the process aspects of
mercury cycling which was beyond the scope of this study A closer look at the results in the
report indicate that the sulfur and phosphorus gradients are not parallel suggesting the
interactions are spatially limited to WCA-2 where most of the sulfate is reduced to sulfide.
Pg. 8-4. "Under anoxic conditions, inorganic Hg is converted to MeHg by sulfur bacteria and
bioaccumulated in the food chain"; How can you reconcile the magnitude and disposition of the
methylmercury problem in the Everglades with the statement made on Pg. 6-6 that "most of the areas
had oxic soil conditions."
Response: It is possible that mercury methylation occurs in the diverse marsh system at sites other than
soil. REMAP data indicate that the most probable methylation sites are not found in the soil
except in the highly reduced soils which are primarily found in areas with high phosphorus
concentrations. The methylation of mercury can also apparently occur in thick periphyton
(floating, epiphytic or soil mat) which may be anoxic during the dark phase. The statements
can therefore be reconciled due to the shifting sites of methylation. Given the diversity of
microhabitats habitats and conditions throughout the vast marsh system one would
reasonably expect microhabitats where it is anoxic in one area, and oxic in another.
Pg. 8-6. "Thus despite the presence of flooded soils, Hg methylation may be expected to be limited in the
absence of eutrophic conditions. "Is the implication here that the naturally oligotrophic Everglades
soils are oxic and therefore there will be only inconsequential methylmercury production? Then
what is producing the high methylmercury concentrations in the Park waters and fish? I haven't done
the calculations, but decomposing plant litter has to exert a BOD in the water column that exceeds
the physical reaeration rate of the Everglades in the absence of photosynthetic activity, so nighttime
DO must drop to near zero throughout the water column, except on exceptionally windy evenings.
Same for sediment. However, since REMAP was a daytime study, would have missed this effect.
During the day, at and below the sediment/water interface, decomposing surface peat must exert an
oxygen demand that exceeds the rate of oxygen production by epibenthic periphyton. So, with the
exception of the vicinity of cattail roots, which passively and actively pump air into their roots
during night and day, respectively, Everglades sediment cannot be oxic, unless sawgrass litter and
peat are so refractory that the BOD and SBOD are low compared with the daytime physical
reaeration and oxygen production rates and oxygen readily diffuses into the litter and peat to
significant depth. I don't think so. Have the Eh measurements been verified with DO measurements
in the sediment pore water?
This statement also implies that we can ignore sulfur and increases in the mercury deposition Hg
flux as potential causes of increased rates of methylmercury production. I don't think so.
Response: Yes. Methylmercury in water is the lowest in the Park when compared to the rest of the
system and lowest in the southern half of ENP which is the area farthest from the phosphorus
source. The concentration of mercury in mosquitofish is high in northern ENP and equivalent
to that found in the hotspot south of Alligator Alley, however, the mercury concentration in
24
-------
mosquitofish is lowest in the southern ENP where the phosphorus is also the lowest. The
bioaccumlation factor (BAF) in mosquitofish increases consistently from north to south and
is highest in the southern ENP indicating the uptake is strongly affected by a food chain
which is very likely recovering (least affected) from the impacts of excess phosphorus from
the north. Eh was sampled at over 500 sites across the system many of which occurred in
sweet brown fibrous peat soils. It should be expected that for peat soil to occur and
accumulate, decomposition must be minimal and these conditions are found in ultra
oligotrophic marsh systems. It should not be inferred that sulfur or mercury deposition can
be ignored. The key processes are likely to change from one part of the system to another
due to the strong gradients from agricultural runoff that affect the interactions of multiple
processes throughout the diverse marsh ecosystem.
8.3 Results
8.3.1 Hg load determinations require structure flow measurements. How were calculations carried out? Not shown
in Chapter 9. Please add and cross-reference.
Response: Chapter 9 is not a mass balance; it is mass estimates. The intent was NOT to compute a
mass balance where inputs and outputs would have to be estimated, but rather to compute
instantaneous mass estimates for the sampling cycle only, so that a relative comparison
could be made of the mass of mercury in various media (water, soil, periphyton, fish). This
does not include flux, turnover, etc., but does provide a relative indication of how the mass
distribution changes by season and among years. This will be made clearer in Chapter 9.0.
8.3.2 Water Quality Patterns
Is there a way to map out the most probable water flow paths in the interior marshes by season using something in
addition to conductivity (p. 4-7), such as a combination of water depth, conductivity, and/or a conservative tracer
(e.g., chloride). When this is done, what are the ranges of travel times between stations?
Add graphs of box/whisker plots of ratios of THg in fish vs MeHg in water, sediment, benthic periphyton, and
floating periphyton at a specified latitude by season vs longitude to quantify north-south bioaccumulation factor
(BAFs) gradients?... THg and MeHg in floating periphyton vs THg and MeHg in water and sediment.... THg and
MeHg in epibenthic periphyton vs THg and MeHg in water and sediment with and without normalization to organic
fraction.
Add multivariate analysis of water BAFs vs water constituents by season and sediment BAFs vs sediment
constituents by season for each distinct ecosphysiographic region and the whole system. (Can one do the same thing
using latitude and longitude as variables in such an analysis?)
Perform a pairwise station similarity analysis for THg, MeHg, and MeHg/THg in water, sediment, periphyton, and
fish. Same for water, sediment, and periphyton BAFs for fish and for water and sediment BAFs for periphyton.
Which stations are the most similar and most different?
Plot minimum and maximum rates of change of each of the water quality variables and reduced variables (e.g.,
BAFs) between adjacent stations as a function of space by season. Plot kriging contours.
Building on the work in Appendix C, what is the relationship between the mosquitofish size distribution vs. MeHg
concentration in each distinct ecosphysiographic region by season? I'm especially interested in THg/unit length vs
water depth per season and THg/unit length vs. water or sediment TP per season
25
-------
Response: The primary purpose of the Technical Report review was to ensure the scientific analyses
and conclusions were sound and supported the conclusions. Additional analyses will be
performed on the data in the future, but the most important current objective is to publish the
report and the data set so the information is available to the scientific community and other
users of the data. Conducting the proposed analyses would delay publishing the report. The
data set will be uploaded to the Internet so the reviewer will be able to perform these
analyses, if desired.
Misnumbered sections.
Response: Corrected.
8.3.2 "The methylation of inorganic Hg to its bioaccumulated form (i.e., MeHg) occurs under a unique set of
environmental conditions." Substitute "favorable" for "unique". As noted before the methods section
provides no analytical methods for nutrients.
Response: Substitution made.
Pg. 8-9. Is THgF - THg filtered samples or THg in fish? Ambiguous notation.
Response: THgF is total mercury in fish. The notation will be made clearer.
Pg. 8-10. Arguments of inverse relationship between Hg and nutrients fall away when looking at gradients
south of Alligator Alley - this should be underscored in text.
Response: This is part of the conceptual model, which also reinforces the threshold concept between
TP in water and methylation of mercury. When TP concentrations decrease below a certain
threshold, MeHg concentrations in the water and sediment and fish Hg concentrations
significantly decrease. The latitudinal gradients indicate these inverse relationships nicely,
until the TP threshold is reached. Below the threshold, the inverse relationship is no longer
apparent.
8.3.3 Soil Parameters
One of the most frustrating things about this section is the choice of units of ug/cc as the only way to
quantify mercury species in peat soil. This has value, but nobody works solely in these units. Please include
traditional wt/dry wt units, as well, to make it possible to calculate fish and periphyton BAFs relative to
sediment. Also, if one includes such units, include the bulk density data used to calculate these values, so
that the ambitious reader can perform the conversion calculations for himself or herself.
Response: These data will be uploaded with the report. Bulk density and soil mercury in weight units are
part of this data so the reviewer will be able to perform these analyses.
See recommendations in 8.3.2 for additional data analyses.
8.3.4 Marsh Characteristics
See recommendations in 8.3.2 for additional data analyses.
26
-------
9.0 Mercury Mass Estimates
What about equations for calculating flowing water loads and rainfall deposition loads?
Omits THg and MeHg mass stored in submerged and emergent macrophytes and discussion of significance
of same. (I know you didn't sample macrophytes, but you could use emergent vegetation coverages and
densities and plant/water or plant/sediment ratios from ENR Project studies to estimate range of plant
mercury concentrations and bound the values of this storage reservoir.)
Response: As indicated above, mercury mass balances or budgets are NOT being computed in this
section. This section provides a relative perspective of mercury among media for snapshots
in time, not mass balances or budgets over time. Loads and fluxes are not needed to
estimate relative mass units at a single instant in time.
Is this only for mosquitofish? If so, then use of generic "fish" is misleading. How were "fish" densities obtained? No
references. Were these densities correlated with season, water depth, TP, or other water quality constituents?
Response: Estimates are only for mosquitofish so the generic "fish" will be replaced by mosquitofish.
Density estimates were obtained from J. Trexler and do not reflect seasonal differences.
Proper periphyton mass storage equation includes a coverage factor. See Bob Ambrose for his estimates of same.
How were periphyton densities obtained? Were these densities correlated with season, water depth, TP, or other water
quality constituents? How much of difference between seasonal mass estimates was due to difference in densities
or coverages vs difference in THg or MeHg concentration?
Response. Periphyton densities were estimated from sample weights for each station. Seasonal
differences due to densities, TP, etc. were not estimated.
What is the ratio of the new mercury load from EAA runoff and rainfall deposition to the stored mercury load by
compartment for both total mercury and methylmercury. If floating and epibenthic periphyton MeHg is newly
produced MeHg, what is the production rate per unit area required to produce the stored MeH load?
Response: These estimates were not computed.
10. Synthesis and Integration
I think the additional analyses recommended in Chapters 8 and 9 will help formulate and test existing and new
hypotheses here.
10.3(10-6) Conceptual Models
Authors propose three revised conceptual models based on the segmentation of the system: above Alligator Alley,
between AA and Tamiami Trail (TT). Testable hypotheses are proposed for each segment. This is good. While such
a segmentation has conceptual value, it misses other relationships that transcends such segmentation. For example,
a cursory review of Table 10.3 (10-10) suggests that the bioaccumulation factors in mosquitofish increase
monotonically from the northern to southern Everglades. What is the hypothesis to be tested that accommodates this
observation? Is this evidence of biodilution, as total phosphorus concentrations fall (nearly) monotonically from north
to south? ... rainfall dilution of the factors that limit MeHg production by diluting out sulfate, shifting sulfide/sulfate
to favor the formation of the bioavailable neutral sulfide species? ... slower growth rates in fish from north to south
27
-------
due to lower average stage-duration in south than north? ... slower growth rates in fish from north to south due to
lower average mass and energy throughput due to nutrient limitation? combinations of the above?
Response: A hypothesis to be tested which most likely would explain the increasing BAF from north to
south would be: what fundamental alterations in the structure of the food chain (e.g., pollution
tolerant species to species native to the original ultra oligotrophy marsh) have been brought
about by the phosphorus gradient? Biodilution is very likely a part of the explanation with the
proliferation of pollution tolerant biomass in the north. Rainfall dilution and sulfur interactions,
and growth rates in fish are probably less likely to be large factors in explaining the rapid
change in the BAF.
10.3.1(10-7) North of Alligator Alley
Reference is made to increased growth and condition of mosquitoflsh in the ENR Project that results in
biodilution, which explains the low MeHg concentrations in these mosquitofish. May be confusing growth
dilution associated with rapidly growing fish with biodilution related to total biomass per unit area or
volume. Fish biomass densities in ENR Project are actually lower than expected based on extrapolations
from interior marsh studies of similar habitats in lower phosphorus concentration regime (Jordan, 1996).
Growth rates of mosquitofish have not been measured. Peak concentrations correspond to or lag by one
quarter the onset of the summer rainy season when the deposition rate of soluble, reactive mercury is at a
maximum. MeHg concentrations in the ENR Project interior water are in the concentration range of 0.04-
0.13 ng/L, with the-mean 0.07 ng/L. This is about one-eighth the concentration in LNWR and AA-N.
Mosquitofish concentrations are in the range of 1 -30 ng/g, with a mean of around 7 ng/g, which yields a BAF
of about 1 x 10s. This is about double that reported for the LNWR, but a little less than half that in AA-N.
The more likely explanation for the lower mean MeHg concentration in ENR Project water is the much
higher sulfate concentrations, which, in the presence of eutrophic and anoxic conditions, produces an
abundance of sulfide that poisons the methylation process. This can be seen in ENR sediment MeHg
concentrations, which are about an order of magnitude lower than those in WCA-3A.(Gilmouret al., 1998).
The more likely explanation for the peak in MeHg concentrations in the interior water in the summer is that
with the onset of summer rainfall, not only is more soluble Ilg(ll) delivered to the system, but both sulfate
and phosphate are diluted, reducing the rate of production of sulfide, shifting conditions to favor the
formation of the neutral sulfide complex that is readily taken up by methylating bacteria.
Response: Growth dilution does not appear to be a factor that occurs to any significant degree.
Therefore there is no confusion with the concept of biodilution. If growth dilution were a
factor there should be a relationship between Hg concentration and fish size. Hg uptake in
small fish is apparently very rapid and the concentration does not decline with size. The
biodilution process begins at the base of the food chain and is much more likely to be
influenced by TP than other variables in this system.
11.0 Missing
Response:
12.0 Missing
Response:
Section has been added.
Section has been added.
28
-------
13.0 References
Complete incomplete citations.
Response: Complete citations added.
Appendix A. Sampling Equipment
Appendix B. DQOs
Appendix B has an Appendix A. Perhaps better Appendix I.
Response: Change made.
Appendix B does not define decision DQOs, only field and laboratory performance DQOs.
Response: As explained above decision DQOs were not established due to the research nature of this
project.
Three standard deviations does not define outliers per se.
Three standard deviations not applicable to evaluate performance relative to %RSD of %RPD.
Response: The guidance found in USEPA QA/G-4 (USEPA 1994) which is not entirely appropriate for
research projects was followed.
Table 3A This table lists the methods used in the study. A large number of these methods are listed as new
method development. Provide citations for peer reviewed method comparisons. If not available,
provide data for method validation studies.
Response: The peer reviewed citations are included.
Table 6A The maximum value for water depth in the marsh is listed as 6 meters, this seems excessive.
Probably 6 feet or 2 meters.
Response: Units corrected.
Table A1A The lack of text associated with this table makes it difficult to interpret. It is apparently the data
quality objective targets for the project. If this is the case, then the precision targets for duplicate
samples were 3s which is significantly higher than the more typical 2s. Were these criteria relaxed
because of the research or scoping nature of the study? If so, make explicit. Furthermore, a
secondary criterion of %RSD @95% is listed. This is difficult to interpret and appears to state that
duplicate data will be accepted up to the 95th percentile confidence levels of the RSD. This is also
very different from the standard of+1-25%. Same question regarding relaxation of QC criteria for
research or scoping study?
Response: This table is the result of the application of the guidance provided in the document cited
above for Appendix B.
29
-------
Table B. 1 .A The lack of text associated with this table makes it difficult to interpret. It appears to be a summary
of the field data quality results. If this is the case then the precision targets for duplicate samples
were 3s which is significantly higher than the standard 2s. Furthermore, these targets appear to be
used to identify outliers for removal from the data. If this study followed standard procedures than
duplicate analyses were performed on about 10% of all samples and each duplicate pair is meant to
represent a certain associated subset of data. Thus each pair represents a minimum of around ten data
points. It is from these duplicate pairs that the calculation of precision is made. These values are not
compared to the mean of the data but to each other. Similarly, it is inappropriate to qualify one value
as failing the QA standard, rather it takes both values. So single points in the outliers list are not
appropriate. Furthermore, if a duplicate pair fails QA the pair is not considered an outlier, rather the
associated data set is flagged as suspect.
Response: This table follows guidance cited for Appendix B above in assessing the database as
prescribed.
Appendix C. Eastern Mosquito Fish Studies Mosquitofish is one word.
Nice work as far as it goes. Would benefit from an analysis of the correlation of fish size distribution in each
distinct ecophysiographic region vs mean water depth, season, presence or absence of periphyton,
concentration of TP.
Response: This is the contribution of Dr. Joel Trexler and students who want at least another season of
data (preferable dry season) before carrying the analyses further.
30
-------
Author: Mike Birch at REGION4
Date: 6/23/1998 2:15 PM
Priority: Normal
TO: Jerry Stober
Subject: comments on report
Comments on South Florida Ecosystem Report
1. Page i Executive Summary "total sulfate" should be just sulfate. Since the method only gives ionic sulfate
and would not measure sulfate bound in particulate matter.
Response: Agreed.
2. Page iii list of Acronyms TS04 total sulfate should be S04 ionic sulfate. The acronym for ThgF is listed in
the report but not explained here.
Response: Total ionic sulfate has been added to list of acronyms ThgF definition has been added to the
acronyms.
3. Page xxi People left off list. There is no mention of Doug Winters, Jim chandler, Sandra Allen, Charlie
Appleby of the ESAT team, or EPA EMSL-LV ORD Deb Chaloud or Ed Heithmer.
Response: These individuals were added to the list.
4. Page 2-14 Table 2.3 there were some Total phosphorus values for the media periphyton.
Response: Total phosphorus was analyzed in periphyton samples from the April 1994 marsh transects.
Only water, soil and sediment were analyzed for TP for the marsh grid samples.
5. Page 3-3 "lexan" does not have a trademark symbol.
Response: Trademark symbol added.
6. Page 3-7 Last sentence of first paragraph, "this unit is equipped with a stirrer for accurate DO
determinations." The stirrer maintains a constant flow of water across the membrane to prevent oxygen sag,
but the stirrer does not increase the accuracy.
Response: Sentence was revised.
7. Page 3-10 third sentence should be rewritten. For the analysis of THg in soil, sediment, and fish, the tissue
samples were placed in glass ampules with concentrated nitric acid, sealed and autoclave to facilitate
digestion.
Response: So noted.
8. Page 3-11 last sentence, the reference should be "Bloom".
Response: Reference added.
31
-------
9. Page 3-13 The correction was not made in the standard curve but in making an appropriate correction for
the reagents contained in the samples, blank and standards.
Response: So noted.
10. Page 3-17 Table 3.1 Under the column for Battelle for Total nitrogen and Total phosphorus there were splits
and duplicates.
Response: Addition made.
11. Page 4-10 Section 4.3.6 Is it possible that the decrease in alkalinity is due to nitrification?
Response: The data do not permit this type of analysis.
12. Page 8-B what is THgF? Not defined in List of Acronyms
Response: ThgF is total mercury in Gambusia. It has been added to the acronym list.
Volume II comments
13. Page 16 Table 6A specific conductance. Conductivity accordingto Standard Methods iscustomarily reported
in micromhos per centimeter (^mho/cm). In the International system of Units (/uS) is used. Is everything
else reported in the International system?
Response: Yes.
14. Page 16 Table 6A Does alkaline phosphatase have units?
Response: The units are ^Mol/hr.
15. Page 8-17 where the data compromised by the detection levels reported or was the ability to define the
significance compromised. The term "high" detection limit is a misnomer. The method is only suppose to
go down to 1 mg/L. Both statement tends to say the data were "inaccurate" or lead to that impression.
Response: The sulfate analysis conducted by USEPA Region 4 SESD had MDLs of 5, 2 and 0.5 mg/L
throughout the course of the study which resulted in a left censored data set These MDLs
were certainly adequate to identify the major sulfate gradients across the system, however,
it is unknown whether a detection level below 0.5 mg/L would have been able to further
elucidate gradients in the southern part of the system.
16. Table A1A. Total Mercury and Methyl mercury under the column comparability (split samples SOPs, Std.
Units) SOPs Battelle hab???
Response: This column in this table simply indicates with whom the comparability was made.
17. Table B.l.A. Canal data field precision. This table is not reproducibility between replicates, but how well a
distribution agrees with it self. The title needs to be changed.
Response: Agree. Title has been changed. The table does indicate strong consistency within the
sizable database.
32
-------
18. Table B.l .A No statement about ethyl mercury but ethyl mercury is talked about in the report.
Response. Ethyl mercury data is not presented in the report, however, the method developed detects
ethyl and methyl mercury. Ethyl mercury was not found consistently in all soil or periphyton
analyses.
33
-------
FLO-SUN INCORPORATED
JUNE 15, 1998
Jerry Stober, Ph. D.
Project Leader
Everglades Ecosystem Assessment
USEPA, Region 4
Science and Ecosystem Support Division
980 College Station Road
Athens, GA 30605-2720
Dear Jerry.
Thank you for providing an opportunity to comment on the South Florida Ecosystem Assessment Peer Review Draft
report. Given the unavailability of the raw data, our comments are based solely on the content of the report. In
addition, due to the short review time, do not construe that a failure to comment on a specific section implies that
we concur with conclusions contained in that section.
My comments follow:
Section 1.4 & 7.2.4
Pgs. 1-7 & 7-9
Although this study collected a large amount of information, it was conducted over a short period under a
narrow range of hydrologic conditions. The report states that the data "was collected over two very different
years with respect to precipitation." This conclusion is unsubstantiated by the rainfall data presented. Based
on Table 4.2, the report should have concluded that studies were conducted during a normal period. No data
was collected during dry periods. In addition, since large volumes of regulatory releases were made from
Lake Okeechobee during the study period, historical long-term comparisons of surface inflows need to be
made. Dry and wet season comparisons also need to be provided.
Response: Table 4-2 does not reference either a dry or a wet period, but states the flows in 1994-1996.
Since the flow through 12D is strongly influenced by water management decisions, the total
Shark River Slough flow (S12ABCD+S333) was substituted for a frame of reference. With
this frame of reference, all three years were wet with 1995 being exceptionally wet. The
Shark River Slough flow was 4 times the average flow experienced in WV1979-1993 and 1.6
times the maximum flow. A table with these computations was added to the report. We agree
long-term comparisons need to be made. We also agree that wet and dry year comparisons
should be made. Therefore, it is proposed that the REMAP program continue to monitor
conditions in the marsh and canals so these comparisons will be possible in the future.
The scope of this study provides little insight into two of your policy questions - trend and cause. It would
be speculative to base conclusions of long-term trends on this short-term data collected under a narrow range
of hydrologic conditions. In addition, inferring causal relationships using synoptic data is speculative.
Response: There was no attempt to base any conclusions in the report on temporal trends, but spatial
trends are especially applicable for this synoptic sampling approach. Because the data were
collected using a statistical survey design, it is possible to provide population estimates of
resource condition with known confidence. The design also permits spatial analyses and
34
-------
associations that provide insight into functional relationships. Without such a design (e.g.,
judgmental samples selected at a single site), it is speculative to make conclusions about
conditions or relationships anywhere else, but at that site.
Section 3.3
The Florida Department of Environmental Protection recently released an audit of the FIU-SERP laboratory.
The audit covered the lab analysis of both the phosphorus and mercury. DEP concluded "it remains unclear
whether past data can be defined as legally defensible and of know quality". In addition, EPA has conducted
a quality control audit of the FIU lab and it is generally understood that the audit has identified major
shortcomings. The results of these audits should figure clearly and prominently in the QA/QC section of your
report Conclusions based on the FIU-SERP data should be flagged for the reader. Furthermore, EPA should
reach no specific conclusions based upon this data.
Response: The audit by USEPA, Region 4, SESD, Office of Quality Assurance of the South Florida
Ecosystem Assessment Project database found that of the over 20,425 measurements made
during the four year REMAP study period, only 15 percent of the data were qualified after
being reviewed using rigorous QA/QC compliance monitoring methods. Most of the data
within the 15 percent remained usable in the assessment and characterization of the system.
The results of the EPA audit are included as an appendix to the final report. In addition, the
data set and qualifiers will be available on a world wide web site.
Section 1.3.3
Pg. 1-4
The report states that Nutrient loading from the EAA and urban areas has significantly increased nutrient
concentrations, particularly phosphorus, in the downstream WCAs and ENP". The references cited in support
of this statement are outdated. Recent analysis of concentrations entering ENP clearly demonstrate that
phosphorus concentrations have declined since 1986 and are at the same level as in 1978 (Walker 1997,
MacVicar, Federico & Lamb Inc., 1996).
Response: The marsh phosphorus data collected during this study show lower concentrations in 1996.
However, whether the observed concentrations are due to the initiation of phosphorus control
efforts, water management practices, the varying hydrologic conditions that occurred during
a particular sampling event, or some combination of these factors, is unclear.
Pg. 1-5
FDEP has not concluded that eutrophication of the Everglades results in violations of water quality standards.
The report referenced was and still is a draft that has never been adopted by the Department. It is misleading,
therefore, to state the FDEP has reached the conclusions contained in the report
Response: Disagree. The report cited is a draft FDEP publication. In addition, FDEP did clearly stipulate
these violations in the 1991 federal settlement agreement concerning eutrophication of the
Everglades.
Section 7.3
Pg.7-11
It is unclear how the data collected during this short-term study provide any insight into the effectiveness
STAs.
35
-------
Response: The paragraph refers to the synoptic sampling providing a baseline for comparison of
conditions in the future following the implementation of the STAs. We agree without continued
monitoring it will be difficult to distinguish the effects of hydrometeorological variability from
STA performance. Text has been modified.
Section 7.2.4
Pg. 7-8
This section relies on logistic regression models to describe the relationship between phosphorus and
cattail/periphyton abundance. However, the referenced figures (Figures 7.22 & 7.23) neither display the raw
data used in the regressions nor the regression statistics. Without, such information the conclusions reached
are unsupported and should be deleted.
Response: The results from the logistic regression models were highly significant, but the proportion of
variance explained was not judged to be satisfactory by the Project Team. Therefore, these
results and conclusions were removed.
Section 10.2
Pg. 10.5
The application of the data collected under this study to the determination or inference of a phosphorus
threshold concentration is an inappropriate extrapolation. The report associates low ranges of phosphorus
concentration to greater periphyton biomass. However, the Ecosystem Assessment study did not collect
periphyton biomass data (only presence/absence was noted). Therefore, the inferences were based on causal
observations at best. The determination of a phosphorus threshold concentration is too critical an issue to
be based on such casual relationships. Government agencies and the EAA Environmental Protection District
are spending millions of dollars on the experimental determination of phosphorus threshold concentrations
The report should delete all references to phosphorus threshold concentrations.
Response: The association of plant responses with total phosphorus concentrations along the "river of
grass" is too obvious to ignore. The inference of a phosphorus threshold will more clearly be
stated as a hypothesis.
We would appreciate receiving all the raw data in electronic form in order to conduct independent analysis. Please
consider this a formal request for the data. If this is not sufficient, please let me know what is necessary in order to
receive an electronic copy of the data upon which this report is based.
Sincerely,
Peter C. Rosendahl, Ph.D.
Vice President of Environmental Relations
36
-------
William W. Walker, Jr., Ph.D.
Environmental Engineer
1127 Lowell Road
Concord, Massachusetts 01742-5522
Tel:978-369-8061
Fax: 978-369A230
e-mail: wwwalker@shore.net
http://www.shore.net/~wwwalker
MEMO
To: Jerry Stober, U.S. Environmental Protection Agency, Region 4
Copies: Mike Zimmerman, Everglades National Park
Su Jewell, Loxahatchee National Wildlife Refuge
Subject: Comments on "South Florida Ecosystem Assessment, Vol. 1, Technical Report, Peer Review Draft"
Date: June 18,1998
I appreciate the opportunity to provide comments on the above draft report. Your letter of May 11 indicated that if
you did not receive my review by June 15, you would assume that 1 "fully concur with the substance of the report."
The following comments reflect the level of effort that I have able to commit to the review within this time frame.
Important items may have been overlooked. Because of time limitations, I did not review Chapters 8-9 and some of
Chapter 10. Please do not assume that I fully concur with these or other parts of the report that are not discussed.
To some extent, my review is handicapped by the fact that 1 have not been able to get a copy of the data set, though
1 understand the need for quality control. It is difficult to form opinions on spatial and temporal patterns and
relationships discussed in the report without working with the data. One important aspect is consistency with other
regional data sets (in particular, those derived from SFMWD's long-term monitoring programs). Overlaying the EPA
data sets with others may provide important insights. Given the spatial intensity, the water-column and soils data
could be extremely useful for modeling and other purposes. I hope that you will be able to provide a complete copy
if the data set in the near future.
p. 1-5 "The effectiveness of these controls in reducing nutrient concentrations to near historical levels, however,
is not yet known". Considerable information is available on these controls. In ~3 years of full BMP implementation
(1995-1997), EAA phosphorus loads were reduced by an average of 51 %, relative to 1979-1988 levels and adjusted
for variations in rainfall (SFWMD Everglades Regulation Department). The Everglades Nutrient Removal Project
(small-scale STA) removed an average of more than 75% of its influent phosphorus load between August 1994
(startup) and November 1996 (SFMWD, Everglades Nutrient Removal Project, 1996 Monitoring Report, March
1997).
Response: Controls have reduced phosphorus concentrations below the 1979-1989 levels, but although
the phosphorus concentrations in the ENR effluent have been consistently less than 30 ^g/L,
they still exceed the Everglades marsh background concentration of 10 ^g/L. The text will
be clarified so that historical refers to natural phosphorus concentrations of 10 ^g/L or less.
The effectiveness of Phase I controls in achieving these phosphorus concentrations over the
long-term remains unknown.
37
-------
Chapters 2/3, also p.5-2. It would be helpful if more detail could be provided on the criteria used to classify sites
based upon vegetation. If a single cattail (or periphyton mat) was visible from the sampling location was this enough
to indicate "presence"? Were any density criteria applied? Did the vegetation have to be within a certain distance of
the sampling location in order to be considered "present"?
Response: The simple presence of a cattail or periphyton mat was enough to indicate existence when
visible in two routine 35mm slide photographs taken at each sampling site: 1) a ground level
shot showing the typical plant types and 2) an oblique shot from an altitude of about 100 feet
of the sampling site. General habitat types were identified by field personnel when on station,
however, after all photographs were in hand and catalogued the same two people viewed all
photos and classified the presence of selected plant types which were used in this analysis.
No density criteria were used. No distance criteria were used except that the plant type had
to be identifiable in the photo.
Chapter 3. It would be appropriate to document in greater detail the method used for selecting the precise spot for
collecting a water column and soil sample. Once the helicopter lands at a site, the sampling crew has to decide
precisely where to collect the sample Was there a tendency to collect samples in open waters (vs densely vegetated
areas)? Or was this aspect also randomized?
Response: The GPS used was estimated to be accurate to within about 10 m. The sampling site was
located with the GPS and the helicopter was landed as close to the point as possible. An
effort was made to collect all media at each site whether or not in wet prairies or dense
vegetation.
P 3-3. Collection of Soil Cores. The text does not define the sampled "soil". Marsh bottoms are often coated with
periphyton and loose floc/detritus/soupy stuff. Was this material included or excluded from the 10-cm "soil" cores?
The interface between the soup and the soil can be difficult to determine. Were plant roots picked out of the cores
before analysis? These aspects of the sampling protocol could be very important for interpreting the soil data and
comparing them with other data.
Response: Some of these questions are answered on pages 3-7 and 3-8. The intact soil periphyton was
removed from the soil core and analyzed separately as soil periphyton. The overlying water
or soup was not sampled but allowed to run off the top of the core. The large plant roots were
picked out of the soil and the remaining soil was the sample.
P 4-2, Table 4-2. The flow through S12D is listed as a reference for hydrologic conditions. Since this flow is strongly
influenced by water management decisions, the total Shark Slough flow (S12ABCD+S333) would provide a better
frame of reference. It should be noted that all 3 years were wet and 1995 was extremely wet. The Shark Slough flow
was 4 times the average flow experienced in WY 1979-1993 and 1.6 times maximum flow. A row with long-term-
average flows could be included in Table 4.2. have recomputed the flows as follows:
WYear
55A
S6
S7
S8
S12s+S333
1994
13.4
13.8
9.9
11.0
33.5
1995
16.2
23.9
18.0
25.0
97.5
1996
11.5
12.7
9.7
14.0
58.0
1979-1993
9.1
6.4
8.8
11.4
24.5
Flows in cubic meters per second, Water Years ending Sept 30. values differ somewhat from values cited in the draft
table. Note that the S5A values differ somewhat from values cited in the draft table.
38
-------
I was curious about stage values. Stage is generally (temporally) correlated with water column P concentrations in
the Everglades, particularly at more enriched sites. The following figure shows monthly mean stages for 1980-1996
at station P33 in ENP. As a consequence of the high stages during marsh sampling events, I would expect that the
measured distribution of phosphorus concentrations at marsh stations is below the long-term average. Since water
levels were higher during marsh sampling than during canal sampling, some of the apparent differences in P
concentration between marsh sites and canal sites may be related to hydrologic variations.
Response: Thank you for the analyses. We agree and will cite this in the text with reference to Dr.
Walker.
pp.4-3,7. Tables 4.3 & 4.4 are useful summaries. A reference should be given for the formula to compute confidence
intervals for the median. The meaning of "confidence interval', is not clear (see comments on Tables 10.2 & 10.3
below).
Response: Reference added.
Figure 4.19,4.28. Since turbidity values are highly skewed, the Box plots would be more informative if the y axes
were logarithmic. Likewise for all total P Box plots & frequency distribution plots (Figures 7.2, 7.3, 7.9.7.10, 7.12)
Response: Comment noted.
Figure 4.23. The legend should provide more detail on the frequency curves. Are these derived from the long-term
stage record or from the stage record during the period of monitoring? The figure for P33 suggests that the sampling
regime is representative (points are spread over curve). This is misleading in light of the above figure. Stages and
water depths were well above normal in 3 out of the 4 marsh sampling dates.
Response: The legend will be expanded and the analyses will be reviewed. The frequency analyses
were for the period of record, not just the sampling period.
Tables 5.1 & 5.2. This information is extremely interesting & useful. Did the habitat types change from one sampling
event to another? Is there any way of computing the uncertainty associated with the habitat distribution estimates?
Response. Each sampling event was a new set of randomized stations. The plant species analyzed
were the same for each event and did not change. Due to the relatively short time span of
the study the plant responses were not expected to change and therefore were pooled over
the entire system combining about 500 observations. Uncertainty estimates were not
calculated.
Table 6.1. p.6-3. Are the bulk density & %organic matter values for the entire soil column or for 0-10 cm only? Are
the statistics based upon pooling data from 4 sampling rounds and all stations within each Subarea? i.e. do the
standard deviations reflect spatial variations, temporal variations, or a mixture? Were soil properties measured in
each sampling round?
Response: Bulk density and % organic matter were based on samples from the 0-10 cm soil profile. The
statistics, including standard deviations, represent pooled data for all four cycles. Soil
properties were measured in each sampling cycle.
Table 7.1 p.7-3. Standard errors of the geometric means seem appropriate here.
Response: Standard errors of the geometric mean do not transform and therefore are not provided with
the means.
39
-------
p.7-7, top. What is meaning of "marsh water" in the EAA? Holeyland/Rottenberger?
Response: The EAA was not sampled, however, the Holeyland/Rottenberger area was included in marsh
sampling. The text should have read "were highest in WCA2, WCA3, and LNWR and lowest
in ENP and BCNP regardless of season."
p.7-7 bottom. Where did total N data come from? This is not on the measured parameter list (Tables 2.1,2.2). Was
total N sampled at the same frequency and locations as total P? If not, average N/P ratios should be computed based
upon paired values only.
Response: Total N was measured only in the last two cycles. It has been added to the parameter list.
N/P ratios were not calculated, but would be done on a paired basis.
Figures 7.22 & 7.23. Some measure of goodness of fit should be provided (classification error, etc.). At least show
the data points upon which these curves are based. If the curves integrate over all regions and data, they may be
misleading because the relationships may vary with habitat type, region, and hydroperiod.
Development of similar curves within regions with similar hydroperiod would be more informative and may give
very different results. These curves may be very misleading and could be misused. In any case, much more statistical
detail is warranted if they are to be included in the report.
Response: The results from the logistic regression models were highly significant but the proportion of
explained variance was not judged satisfactory by the Project Team. Therefore, these results
were removed.
p.10-2. last paragraph. "This plot shows an inverse relationship between TP in water and THg in mosquitofish.."
The plot does not show a relationship, only a spatial correlation. The causal "relationship" for THg could be with
another spatially varying factor. The report should be careful about assigning or implying causality based upon spatial
correlations.
Response: So noted.
p.10-7. Section 10.3.2. "In the marsh, TP concentrations are between 10 to 20 fxg/L, which is a level currently within
the assimilative capacity of the system and that stimulates periphyton production and increased periphyton biomass."
Where in the report has the "assimilative capacity" been determined? This sweeping statement does not seem to be
supported in the report. The last part of the sentence is in direct contradiction with results of ongoing threshold
research by SFWMD which, I believe, demonstrates experimentally that native periphyton mats tend to decrease at
P levels above 10 £
-------
Response: The meaning of the 95% confidence interval about the median is given in Section 2. The
median estimator and its CI are design-based, so that the only lack of statistical
independence that needs to be accounted for is that introduced by the spatial restriction of
the sample design. As noted in Section 2, under the condition of spatial pattern, the CI is
conservative (in Walker's terms, corresponding to under-estimating the degrees of freedom,
not over-estimating as suggested by Walker).
In particular, the value estimated is the median of the population that existed at the four
sampling times. It is not a model-based estimate of the "real" median of some hypothetical
population subject to stochastic spatial and temporal perturbations. So-called "temporal
correlation" has no bearing on the estimator.
Table A1A - Water Data Quality Objective Criteria. The meanings of the terms should be clearly defined. Some
of the numbers do not seem to make much sense. For example, Duplicate Precision (%RSD) value for Turbidity in
canals is 182.9. Does this mean that duplicates can be expected to differ by as much as 183%? Likewise, duplicates
for total phosphorus and mercury, by 101% and 230%, respectively? These seem like situations to be avoided, not
OBJECTIVES. Does "Field Precision" reflect variance across samples taken at the same location and time? What
program generated this table? How many replicates, duplicates, spikes etc.?
Response: The terms are more clearly defined in the table and the reader is referred to EPA DQO and
QA documents cited in the reference section. The information in the table was generated by
EPA QAMS based on QA programs which they developed. The number of replicates,
duplicates, spikes, etc. can be found in the field and laboratory methods manuals which are
cited in the reference section.
Table B.l.A. What methods were used to detect "Outliers"? Were outliers included or excluded from the analyses
conducted in the report? It seems like one of these columns should contain a value for relative standard deviation
(coefficient of variation) across replicate samples. It is not at all clear what the numbers mean or how they were
generated.
Response: Outliers were detected if values were greater than ± 3o or were outside range checks for
Everglades constituents. However, all data, including outliers, were used in all statistical
analyses and are included in the results presented.
This concludes my comments. I hope that you find them useful. The report and data sets will contribute substantially
to Everglades research and management. Please contact me if you have any questions.
41
-------
Author:
ckendall@usgs.gov at FN
Date:
6/14/1998 5:11 PM
Priority:
Normal
BCC:
jerry stober at PEGI0N4
TO:
STOBER.JERRY at IN
CC:
ahiger@usgs.gov at IN, dpkrabbe@usgs.gov at IN, ckendall@usgs.gov at IN
Subject:
review of technical report
Hi Response:
Thanks for sending me a copy of your report after the recent meeting. Here are a variety of comments based on
spending a day reading it (alas, the first blue-sky weekend day we have seemingly had in months!). 1 only read Joel
Trexler's part of the appendix, not the rest of it. It was a shear delight getting to see the rest of your data, after so
long. In particular, I have been really wondering about your vegetation data and what that might tell me about the
spatial patterns in dl3C of the periphyton and Gambusia. When I get a spare moment, I will do all sorts of
comparisons to see if the changes in relation dominance of the plants might explain some of the isotope variability.
For example, is there a difference in the Gambusia in places with and without periphyton mats?
I also look forward to working with you and Joel to make sense of the spatial variability in Gambusia and to
rationalize Joel's gut contents data with the isotope data. If neither of you has time or interest in taking the lead on
such a comparison in the near future, 1 might see if Cecily Chang, the biologist in our group who has a moderately
good statistical background, might want to make a start on the comparisons. What do you think? I will send a note
to Joel to see what he thinks, too. I am sure that there is room for more than one paper relating these massive datasets.
This is an impressive document, sure to give readers lots of things to think about and hypotheses to test with other
datasets. I hope that you will now have time to take chunks of these data and publish them in the white literature. You
have LOTS of neat stuff here.
You have made me a believer in box and whisker plots. I finally see why you divided the system into latitudinal
bands and used these plots to show broad regional changes. I like the idea of viewing the system as a "wide shallow
stream", and I guess it is OK that the divisions are not perpendicular to the flow lines. Did you try it the other way?
I understand that there are some fancy 3-D visualization packages that let you rotate the data in space to see in what
plane the patterns are strongest, cool! These box plots work pretty well for showing big patterns, and there are not
many readily available tools for comparing spatial patterns, but I am still hoping to find some better tools to deal with
the smaller patterns.
There is a large amount of relevant ACME project work that should be cited in your report, as well as recent work
by other USGS FL groups such as Bill Orem's sediment group. One paper by Gilmour is the only one I saw from
these groups. There have been lots of published abstracts about our work in the proceedings of the annual USGS
meeting and other international meetings, several papers have been published, and 1 know that a lot of papers were
submitted to a Hg conference in Hamburg, and these certainly could have been cited as in press before (but now I
think they are published). It seems like some of your Hg-related speculations and hypotheses are partially based on
some ACME work or could be partially validated by reference to some ACME work. The FL missions of the two
agencies are nicely complementary. I realize that this is an EPA report and you naturally want to emphasize your data
and their patterns, but when you start interpreting the data, these USGS studies are as much a part of the public
literature as the many academic or SFWMD papers that you cite extensively.
Below are miscellaneous comments I scribbled as I read through your report:
42
-------
Executive summary: you need to explain what biodilution and BAF mean. I don't think BAF was defined anywhere
in the text, nor was anything said about its units (if any). The whole concept of biodilution seems weird to me- just
because a fish grows large, its MeHg is considered diluted?
Response: The concept of biodilution is used in reference to eutrophic systems. In eutrophic systems
there is high nutrient availability which usually results in high primary plant production
(periphyton or phytoplankton). As a result there are more cells per unit volume each with a
portion of the available mercury in the system thereby diluting the mercury broadly across
the base of the food chain. The grazers cannot utilize all of the available plant biomass and
the efficiency of the bioaccumulation is lower at all higher trophic levels.
Response: The BAF is the bioaccumulation factor. The BAF for methylmercury is defined as the
concentration of methylmercury in whole fish divided by the concentration of dissolved MeHg
in the water. In this case it has only been used for the mosquitofish prey species which is
about a trophic level 3 species. A BAF can be calculated for each trophic level up to the top
predator.
I had not realized that you collected both soil and floating periphyton. Are all my samples floating ones or are they
a mixture? If the latter, please tell me which is which. If the former, do you or Jones still have any soil periphyton
archived?
Response: All periphyton sent to you was floating. We may have a limited number of soil periphyton
samples archived.
I note that sometimes the tables (especially the larger ones) are at the ends of chapters, with the figures, but
sometimes they (especially the smaller ones) are in the text. I found this confusing.
Response: Report has been revised with all tables embedded in the text. Because the figures are so
numerous, they have been provided at the end of each chapter.
Page 4-8 — 1 think the high conductivity of the EAA discharge is probably due to the leaching of highly weatherable
materials from the rapidly oxidizing soils, plus agricultural amendments.
Response: So noted.
Chapter 5: you should define what you mean by wet prairie and what a Muhlenbergia is. Do you have some table
where you list the dominant and secondary communities at each site? 1 would like this.
Response: Wet prairies include a group of low stature, graminoid marshes. They are found over peat
and marl, and each soil type supports distinct communities. Wet prairies over peat occur in
the central, wetter areas of the Everglades. Those over marl are found in the southern
Everglades on higher, drier sites (Craighead, 1971; Davis, 1943 and Gunderson, 1994).
Muhlenbergia occurs in the latter. See Tables 5.1 and 5 2 for listings.
You need to better explain figure 7.2 — I spent a lot of time wondering what you meant by "% canal length" --1 kept
figuring you were trying to actually show how the concentrations changed from N to S.
Response: % canal length refers to the population estimates where each sample represents a specific
portion of the population. This is the strength of the EMAP design-you can estimate
population attributes from the samples. In this example, the population is defined as the total
length of the canals sampled. Explanation provided in revised report.
43
-------
I think Fig 7.4 (and others like them) were especially effective, even more so that the box plots.
I found your ANOVA tables a bit confusing (eg Tables 8.2, 8.3). When you star the averages that are statistically
significantly different — it is not so clear what they are different from -- both of the unstarred values one of them
(which one?). Sure, I could mostly figure them out after some thought, but I don't think this was a very clear way
to present the data.
Response: Table revised and differences made explicit.
If you look at the spatial plot of % calcite I sent you, you will note that the area where periphyton "% calcite is
especially low corresponds pretty well with your WCA3-C area, and also matches up with the large E-W band of
marsh where I showed that the dl3C of Gambusia was inconsistent with bulk periphyton (or something of comparable
d 13C) being a major food source (I think I sent you this plot too). Perhaps the non-calcite-forming periphyton is not
yummy? Or is the lack of calcite a response to some environmental condition that favors the production of different
food sources? Interesting questions!
Response: It looks like there might be something there which could relate to the type of periphyton. In
order to interpret your analyses and understand how they relate to the community
composition requires some level of taxonomic identification of the samples.
I think you should revise and replace Figures 10.3 to 10.5. These are your final summary figures and they should be
dazzlers - you want the patterns to leap from them! These figures are hard to decipher because they are rather busy
and the lines are not sufficiently distinct.
Response: Agreed. Figures revised.
I found section 10.2 in the synthesis difficult to read, partly because of all the descriptions of patterns. This would
read much better if you subdivided it into mini-sections, each with a different heading and purpose.
Response: Section was revised for greater readability.
Page 10-5: you state that the MeHg in the canals comes from the marshes - I missed how you determined this
important point.
Response: The MeHg in water is consistently higher in the marsh than in the canals, indicating
methylation is higher in the marsh system than in the canal system. This does not
necessarily mean that the marsh is the source of canal MeHg.
Appendix C: It would be useful if there was a bit more discussion of what Joel's "Pr models" were and how they
work. He explained it to me once, and I understood for only a brief moment!
Response: These are logistic regressions indicating the probability that various sizes of juvenile, male
and female mosquitofish are likely to have plant matter or food in its stomach based on the
relative standard length.
Comments on your "five questions" on the cover letter:
1) Yes.
44
-------
2) I wish you had done more to try to relate changes in water chemistry to habitat, soil characteristics, or
hydrologic data - or the habitats to the other parameters. I think that with such a huge dataset there might have been
more gleaned from the spatial patterns with fancy statistics. Your division of the marshes into 7 bands is a good start,
but there is a lot high-powered GIS type might have more sophisticated tools available for comparing these strong
spatial patterns. Do you have such a person on your team or could you contract with one? 1 am facing the same
problem with my isotope data - how do I express, in some statistically meaningful fashion, the fine-scale spatial
similarities in patterns?
3) Yes. But as mentioned in #2 above, I think you might have done more to compare spatial patterns among
plant, environment, and chemistry - which could have lead to more testable hypotheses. I feel that correlations that
are only apparent after lots of fancy statistics are not very satisfying (which is perhaps why you didn't do this), but
they are useful for developing testable hypotheses.
4) My copy of the report was missing Chapter 11 - so I couldn't comment on the conclusions. Was this an
oversight?
Response: No. This chapter and chapter 12 were not released for peer review since they were still under
internal review.
5) My copy was missing was missing Chapter 12 - so I couldn't comment on the recommendations. Was this
an oversight?
Response: No. See response above.
I hope theses comments are useful and I look forward to working on papers with you in the near future - and in
continuing our isotope collaborations in REMAP-II. When this gets funded, we can discuss the suggestions 1 listed
above to see what you can accommodate and how to make things as easy for the field team as possible.
—Carol
Carol Kendall
U.S. Geological Survey
345 Middlefield Road, MS 434
Menlo Park, CA, 94025 USA
Phone: +1 650 329 4576
Fax: 41650 329 5590
Internet: ckendall@usgs.gov
45
-------
SEMINOLE TRIBE OF FLORIDA
Tribal Officers:
JAMES E. BILLIE
Chairman
MITCHELL CYPRESS
Vice Chairman
WATER RESOURCE MANAGEMENT
6300 STIRLING ROAD
HOLLYWOOD, FLORIDA 33024
(954) 967-3402
June 15,1998
Jerry Stober, PhD
Project Leader, Everglades Ecosystem Assessment
Science and Ecosystem Support Division
U.S. Environmental Protection Agency, R4
980 College Station Road
Athens, Georgia 30605-2720
RE: South Florida Ecosystem Assessment:Peer Review Draft
Dear Dr Stober:
Thank you for the opportunity to review the above referenced documents Be advised that though sent to
Chairman Billie's office in a timely fashion, these documents were not received by our Department until June 5.
While we appreciate the opportunity to review the materials provided and to comment on the approach used in the
study, time constraints and limited resources preclude us from providing formal comments on the submitted materials
at this time
While we wish to reserve providing you with any substantive comments until a later date, some preliminary
observations/questions are as follows:
1. The appendices lacked the referenced materials/raw data addressed by the document.
Response: We hope the important references have been added and the raw data will be released on the
Internet with this report serving as the metadata.
2. What Quality Assurance/Quality Control protocols were obeyed in the collection of the data?
Response: The QA/QC protocols used in the field and laboratory have been included as an Appendix
to the report.
46
-------
3. What are the qualifications of the principal scientists peer reviewing the papers submitted in this report?
Response: A list of the peer reviewers is presented in the response to comments.
Should you have any questions at all regarding this matter please do not hesitate to call me.
Sincerely,
Craig D. Tepper
Director
47
-------
Author:
trexlerj@fiu.edu at IN
Date:
6/18/1998 5:00 PM
Pricrity:
Normal
BCC:
jerry stober at REGION
TO:
stober.jerry at IN
Subject:
mercury report
Hi Jerry,
Sorry this is a couple of days late, I have been going crazy with some summer syposium talks I have to give. I
examined the vol II issue because it is the section related to my kind of work (I don't know much about mercury
chemistry) . I have two suggestions.
1) page C-3. This isn't my section, but it is not correct to claim length weight curves indicate growth rate. The slope
of those curves is simply condition factor. Without some data on age at a given size, you can not address growth rate.
This is mentioned twice on this page Sorry, but I don't buy it.
Response: Text corrected accordingly.
2) There are two typos on my section, page C-S. Both appear in grey, and should be easy to find. They are both where
the computer couldn't decipher my chi-square symbols. On line 12 from the top, it should read chi-square sub 2 =
7.47, and on line 21 it should be chi-squre sub 3 = 12.01. Maybe you could use a capital X instead of the greek chi
symbol if your wordprocessor can't make greek letters.
Response: Done.
Hope this is of some help. The rest looks great. I will be in Vancouver until next Wednesday, in case you need to
reach me.
Joel
48
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
REGION 4
ATLANTA FEDERAL CENTER
61 FORSYTH STREET
ATLANTA, GEORGIA 30 3 03-8960
SUBJECT: South Florida Ecosystem Assessment Volume 1, Technical Report, Peer Review Draft
FROM: Linda Anderson-Carnahan, Chief
Air Planning Branch
TO: Jerry Stober, Project Leader
Everglades Ecosystem Assessment
Attached please find comments on the subject report. If you have any questions on the accompanying comments,
please call me at (404) 562-9074 or Ms. Stacy Gent-Howard at (404) 562-9063. Thank you for the opportunity to
review this research report.
Attachment (1)
1. Air Planning Branch Comments
49
-------
Comments/Recommendations on the South Florida Ecosystem Assessment
Volume 1 Technical Report
Peer Review Draft
1. Under the "APTMD" section, on the list of participants in the Everglades Assessment Project, add the name
Stacy Gent-Howard
Response: Done.
2. Move the listing of acronyms on page iii to a location after the Table of Contents and identify it within the
Table of Contents. Also, all acronyms used within the report should be listed on the acronym page; i.e. some
acronyms such as "ENP-N, AA-N, and WCA-3S" are not included.
Response: Done.
3. Add a figure to the Executive Summary depicting the changes in mercury (Hg) in mosquito fish with total
phosphorus (TP) (ref: para 6, Executive Summary) and a map of the study area indicating where each sample point
is located, if possible.
Response: The Executive Summary is intended to be one or two pages. While the mosquitofish mercury-
total phosphorus relationship is one part of the story, there are other factors that also would
have to be added as figures, increasing the length of the Summary. This figure, therefore,
was not added.
4. Place all graphs and figures within the text of the report near where they are referenced. This provides
continuity for the reader.
Response: Done.
5. Move the "Purpose" section, section 1.4, of Chapter 1 to the beginning of the Introduction, ahead of the
Background section. This section sets the tone for the reader and will assist them in understanding the need for the
research.
Response: Done.
6. Given the current national attention on mercury, add a table or other figure to Section 1.3, that provides the
reader with some perspective on how high/low the Hg levels in fish are in the study area as compared to other regions
where mercury is a concern.
Response: The text provides the reader with the concentrations for Hg in bowfin and states that these
are the highest recorded nationwide. The EPA Mercury Report to Congress was added to
refer the reader to Hg levels in other regions.
7. Provide a frame of reference in Section 1.3.2, at the end of paragraph 1 where the discussion on human health
effects of mercury is located to provide the reader with a basis for determining if the mercury levels noted are a
concern, and to what extent.
Response- Reference to EPA Mercury Report to Congress was added to provide the reader with a
citation for further information relative to human health issues.
50
-------
8. In Section 1.3.3, the second paragraph mentions the natural water phosphorus concentration of 10 ppb. The
significance of this number should be discussed in the text; i.e. what events are triggered/retarded at levels above this
concentration?
Response: A reference for this phosphorus concentration has been added.
9. Section 1.3.5 mentions that non-indigenous plant and animal species within the Everglades need to be
controlled, but provides no justification for this statement.
Response: Some justification has been added. The intent was to introduce issues, not provide a detailed
discussion.
10. Section 1.3.6 states that previously mentioned issues are interdependent, but no discussion of this
interdependence and its ramifications are provided. Further discussion of this topic should be provided.
Response: Again, the intent was to introduce the issues, not provide detailed discussion. This particular
topic is discussed frequently throughout the report. Some additional discussion has been
added to Section 1.3.6.
11. Section 1.4, paragraph 1 mentions the South Florida Interagency Ecosystem Restoration Task Force for the
first time. A background and history discussion for this group and its relevance to this study should be provided
within the Background section, Section 1.1.
Response: This information is provided in Section 1.2.
12. The last paragraph of Section 1.4 states that this study focused on a subset of hypotheses," but does not
identify these hypotheses, nor are they identified elsewhere in the report.
Response: Done.
13. The purpose of this report, as defined in Section 1.4.1 is to provide preliminary answers to the four policy
relevant assessment questions of 1) magnitude, 2) extent, 3) trend, and 4) cause. The text of the report currently does
not explain how the hypotheses the study focused on assisted in answering each policy relevant assessment question.
Additionally, an explanation of how the chapters following the purpose statement relate to and answer these questions
needs to be discussed.
Response: Chapter 11 on Management Implications discussed these questions and the answers
obtained by the study as well as the their use in making appropriate management decisions
now and in the future. However, this chapter was not available for peer review since it was
still under internal review.
14. Chapter 8, the Mercury Chapter, does not fit into the report; i.e. no relationship between mercury, the
preceding information, the hypotheses tested, and the policy relevant questions is established in the prior chapters.
A relationship needs to be established prior to presenting the mercury related information.
Response: The interface between mercury and these other issues was initiated in Section 1.3.6, and in
the initial conceptual model discussion. The report now references these earlier sections so
the relationships are more evident.
51
-------
15. The report provides a great deal of data and information, but does not establish any relationship between the
data and the conclusions drawn, leaving the reader to ponder the relevance of what is presented.
Response: The Synthesis and Integration Chapter brings the interrelationships of the data together and
formulates a number of hypotheses to be tested based on the associations indicated by the
data. Management implications of the study are stated in bullet form in Chapter 11 which was
omitted in the peer review draft.
16. Figure 8., TS04, is an excellent visual aid for providing an understanding of where reported concentrations
of compounds are found. More figures like this should be used in the text to aid the reader in visualizing the study
area and drawing conclusions from the data.
Response: Various figures have been modified to improve visual clarity.
17. Sections 8.3.2.1 and 8.3.2.2 discuss canal water quality and mention the compounds TOCand TS04, but are
located in the mercury section. No relationship between these compounds and mercury has been established. The
correlation (if any) between these two compounds and the behavior or presence of mercury should be established
prior to this discussion.
Response: The statistical tests which identified the associations were mentioned in the first paragraph,
however, additional discussion will be added to indicate why these parameters have
important interactions with mercury
18. Place Section 8.3 at the beginning of Chapter 8 to provide some perspective and lead-in to the rest of the
chapter.
Response: Section 8.3 is the Results Section. It is in the appropriate place.
19. Section 8.3 cites complex relationships among the water depth, TOC, sulfate, and total phosphorous
concentrations, but does not justify this claim or elaborate on what these relationships might be. Similarly, this
Section states that canals play a major role in the transport of total phosphorous, but does not explain what this role
is or its relevance to the policy relevant assessment questions.
Response: See question 17 above.
20. In Section 10.1, the third sentence begins "Other sections discuss the conceptual model for Hg..." It is
suggested that this sentence be reworded to refer to "Previous sections..." or 'Chapter 8..." and follow with "...new
models are offered later in this Chapter" to more accurately identify the context of the conceptual models in this
report.
Response: Sentence reworded as suggested.
21. Data presented on page 10-3, Section 10.2, paragraphs 3,4, and 5 would be better presented in a table with
surrounding text providing perspective on the data and discussing its relevance and meaning.
Response: The data are in tables, Tables 10.2 and 10.3. The text has been modified to make it more
readable.
52
-------
22. Page 10-5, Section 10.2, paragraph 7 indicates there are important management implications of a threshold
TP concentration, but does not elaborate upon this claim, or what the implications might be.
Response. We are stating what is indicated by the data in the TP gradient and the apparent associated
plant responses as one proceeds downstream from north to south. A threshold TP
concentration must exist and it will be stated more clearly as an hypothesis.
23. Page 10-5, Section 10.2, paragraph 9 remarks upon the small maximum sizes of mosquito fish in the
Everglades and refers to sizes of mosquito fish found in other parts of the mosquito fish's "natural range." Provide
additional information on what the "natural range" is and to what magnitude Everglades mosquito fish are deficient.
Response: The two authors cited have a paper in preparation.
24. Section 10.4 refers to "testable hypotheses" and "conceptual models"; Some distinction should be made as
to whether these are the hypotheses/models developed prior to this study or subsequent to this study and upon which
future work should be based.
Response: Distinctions have been made.
25. Chapters 11 and 12 are missing from the document provided for review. The "Summary and Conclusions"
and "Recommendations and Future Directions" Chapters are very important and should be circulated for peer review
as well.
Response: These chapters will be circulated for review at a later date.
53
-------
United States Department of the Interior
FISH AND WILDLIFE SERViCE
Arthur R. Marshall
Loxahatchee National Wildlife Refuge
10216 Lee Road
Boynton Beach, Florida 334374796
June 12, 1998
Dr. Jerry Stober
Project Leader
Everglades Ecosystem Assessment
U.S. Environmental Protection Agency
980 College Station Road
Athens, GA 30605-2720
Dear Dr. Stober:
I have received the "South Florida Ecosystem Assessment" peer review draft that you sent to me.
Unfortunately, 1 do not have sufficient time to review it as you requested. I have "skimmed" it and have found one
item that I found confusing. In Volume II (Appendices), Table 1 shows the deadlines for permits. However, I know
at least one deadline has not been met (NPDES for the ENR). The table makes it appear as if the deadline has been
met, since it doesn't say otherwise.
Response: Missed permit deadlines are a regulatory issue beyond the scope of peer-review of this
scientific report.
I have another issue that was brought to my attention independently recently and that is of concern to me.
I have heard that there are various methods of analyzing mercury samples, just as there are for phosphorus. We have
found that the results between labs can vary for phosphorus for split samples, as demonstrated by the Everglades
Technical Advisory Committee's Round Robin sampling events. I have heard that the mercury results can vary
significantly. Could this be a factor in your results for this document? Are you investigating this? I would like to have
your input, because, aside from the effect on results such as in this report, I must deal with mercury due to its
relationship with phosphorus in the Everglades as part of my responsibilities here at the refuge.
Response. The interagency Mercury Program Management Committee has initiated laboratory
intercomparisons and a round-robin exercise in order to determine the comparability of data
across the various mercury laboratories and Florida mercury research efforts. In addition,
numerous sample splits have occurred throughout this study across the three analytical
laboratories. The potential mercury-phosphorus relationship issue is included within
Chapter 8.
Response: Because of different analytical methods for various mercury species, we had both primary
and secondary laboratories designated for all analyses with the secondary laboratory
providing QA/QC analyses. The mercury methods used in the study have all been published
in the peer-reviewed scientific literature. In addition, the project participated in several
mercury round robin analyses.
Thank you for soliciting my review. I hope you can shed some light on the Hg analysis problem.
Sincerely,
Susan D. Jewell
Senior Biologist
54
-------
Environmental Services & Permitting, Inc.
June 15,1998
File: 295-94-01
Jerry Stober, Ph.D.
Project Leader
Everglades Ecosystem Assessment
U.S. Environmental Protection Agency, Region 4
Science and Ecosystem Support Division
980 College Station Road
Athens, GA 30605-2720
Re: EPA PEER REVIEW DRAFT REPORT "EPA-904-R-98-XXX" SOUTH FLORIDA ECOSYSTEM
ASSESSMENT
Dear Dr. Stober,
We appreciate receiving the draft report for review. The report contains a lot of data. We would like to have
the raw data in electronic format so that we could review it and conduct our own analyses. This would enable us to
make more meaningful comments. We also need more time to review the data and the report. We only received the
report May 26 and have not been able to drop all our other projects to review this one. Given the importance of this
issue we feel more time is needed and the raw data need to be provided.
We are also concerned that EPA may have the proverbial cart before the horse. Much of the data in the report
was apparently collected by FIU. The Florida Department of Environmental Protection (FDEP) Quality Assurance
section has recently audited the FIU mercury laboratory and found several major deficiencies that caused them to
express some serious concerns about the quality and reliability of the mercury data from the FIU laboratory. We
understand EPA is aware of these concerns and has been looking into them. We feel these issues need to be addressed
prior to using the data in a report. We do not want to spend a lot of time working with the data or reviewing the report
prior to these issues being definitively resolved. It would be inappropriate to use questionable data in statistical
analyses and then draw conclusions about the results. Once the EPA QA\QC reviews are completed and made public,
the report needs to be revised to include the QA\QC assessment and any questionable data need to be removed from
the data sets and the analyses rerun. If none of the data are found to be questionable, then we need additional time
to review the raw data and conduct our own analyses.
Response: The data have undergone extensive review and the QA/QC report has been included as an
appendix. In addition, the database with all qualifiers will be available on a world wide web
site.
We are also concerned about the synoptic nature of the data. The data clearly need to be collected over a
much broader range of hydrologic and seasonal conditions.
Response: We agree and propose the project be continued so a greater range of hydrologic conditions
can be monitored.
55
-------
Jerry Stober, Ph.D.
June 15,1998
Page 2
The report needs to be updated to use the most recent nutrient data and not rely on outdated and draft reports.
The impacts of the Everglades Construction Project on future nutrient loads and their impact, if any, on the mercury
issues needs to be addressed.
Response: The nutrient data reported in this study were collected from 1993 through 1996. The purpose
of this report is to summarize this data set only. The text has been modified to include other
published data on TP concentrations and results of ENP and BMPs, however, a detailed
discussion of the ECP is not warranted in this report.
Please keep us on the list to receive all drafts and data distributions relative to Everglades issues. If you
would like to discuss any of our comments, please call.
56
-------
Author: federico@mfl-inc.com at IN
Date: 6/15/1998 4:47 PM
Priority: Normal
BCC: jerry stober at REGI0N4
TO: stober.jerry at IN
Subject: Pete Rosendahls' comments on Ecosystem Assessment Report
Pete Rosendahl requested that I e-mail you his comments on the Ecosystem Assessment Report. He will be sending
a hard-copy in the mail.
57
-------
Author:
tolsen@mail.cor.epa.gov at IN
Date:
6/15/1998 11:08 AM
Priority:
Normal
BCC:
jerry stober at REGION4
TO:
stober.jerry at IN
Subject:
South Florida Assessment report
Jerry,
] looked through the two volume South Florida Ecosystem Assessment report. It appears to be a comprehensive and
well-written technical report. The sections describing the statistical design and analysis are well written and
communicate the essential information. The statistical analyses use exploratory data summaries, design-based
estimates for populations and model-based (Kriging) estimates. Each has role to play in understanding the data and
the techniques appear to be used correctly. In several instances additional information could have been given
concerning the uncertainty of some of the estimates. It may be sufficient to present such information for several
typical situations and then state that similar levels of uncertainty accompany other similar summaries. For example,
a 4-cycle and 1 -cycle variance contour map from kriging estimates may be used to illustrate the uncertainty in other
maps (assuming that such is the case!)
Overall the project provides extensive data on a critical problem
Tony
Anthony R. Olsen
US EPA National Health and Environmental Effects Research Laboratory
Western Ecology Division
200 SW 35th Street
Corvallis, Oregon 97333
Voice: 541 -754-4790 Fax: 541 -754-4716
email: tolsen@mail.corepa.gov
58
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
NATIONAL HEALTH AND ENVIRONMENTAL EFFECTS
RESEARCH LABORATORY
MID-CONTINENT ECOLOGY DIVISION
6201 CONGOON BOULEVARD. DULUTH, MINNESOTA 55804-2595
OFFICE OF
RESEARCH AND DEVELOPMENT
June 3,1998
MEMORANDUM
SUBJECT:
FROM:
TO:
South Florida Ecosystem
Assessment Draft Report
Anthony R. Carlson
Research Aquatic Biologist
Jerry Stober
Project Leader
Everglades Ecosystem Assessment
This report clearly provides answers to policy relevant questions related to the magnitudes, extent, trends,
and causes of current ecological conditions in the Everglades. The statistical survey design sampling data and its
synthesis provides the baseline and conceptual models for use in assessing future trends in ecological condition
related to mercury contamination, eutrophication, habitat alteration and hydropattern modification issues. I agree with
the report conclusions and recommendations for future studies. I believe this technical report is an excellent example
of the EMAP approach and recommend that it be published.
59
-------
Dr. Jerry Stober
U.S. Environmental Protection Agency, Region 4
Science and Ecosystem Support Division
Athens, Georgia
Response:
In response to your written request of May 11,1 regret that I cannot provide a review of the South Florida Ecosystem
Assessment, Volumes 1 & 2.
Last fall, our Center Director (USGS Upper Mississippi Science Center) transferred to a position in our Reston
National Office, and I spent months serving as Acting Center Director. As a result of the associated heavy
administrative and managerial workload, I have fallen behind schedule in several areas and for the foreseeable future,
am declining all requests for reviews. On the positive side, 1 am very pleased to report that our new center director,
Dr. Leslie Holland-Bartels, will arrive here in June.
My sincere regrets, Jerry, to you and your coauthors, but it's evident that good intentions don't get reviews done in
such busy times. Please let me know if you would like me to return the subject documents to you.
Regards.
James G. Wiener
U.S. Geological Survey
Biological Resources Division
Upper Mississippi Science Center
2630 Fanta Reed Road (Street address for courier)
P.O. Box 818
La Crosse, WI 54602-0818 (zip code for courier is 54603)
phone 608-783-224 (voice mail)
fax 608-783-066
E-mail james-wiener@usgs.gov
60
-------
Author: jgentile@rsmas.miami.edu at IN
Date: 06/03/98 10:26 AM
Priority: Normal
ECO: jerry stober at REGION4
TO: stober jerry at IN
Subject: Review: South Florida Ecosystem Assessment
Jerry,
Thanks for the opportunity to review the SF Ecosystem Assessment. Unfortunately I will be unable to review it
within the time frame you needed. So I can't say I concur with the conclusions and didn't want you to assume that
I did if I didn't respond. I'm sure I will get to it but not at the moment.
Regards,
Jack
John H. Gentile, Ph.D. Senior Scientist
Rosenstiel School of Marine and Atmospheric Science
University of Miami,
Miami, FL 33149
[305] 361-4152 (Phone)
[305] 361-4C77 (Fax)
jgentile@rsmas.miami.edu
61
-------
6 May 1998
United States Environmental Protection Agency
Region 4
Science and Ecosystem Support Division
980 College Station Road
Athens, Georgia 30605-2720
Attention: Dr. Jerry Stober, PhD, Project Leader
Everglades Ecosystem Assessment
stoberjerry@epamail.epa. .gov
Dear Jerry,
Your report did not arrive until 22 May and it was delivered in rather poor condition. It must have got a rough ride
from Athens to Ottawa. Nevertheless, it provided excellent reading during my trip to and from Athens. I am very
impressed with the care and quality of the work. It was a monumental task and will be a classic study that will be
scrutinized by scientists for years to come. This means that we need to make it as tight as we can.
A hard copy will also be sent by courier.
1 think that the problem that is discussed is not restricted to the Everglades. Here in Ontario, bogs and wetlands were
drained from 1850 to the present and farmers have been diverting water into them and adding nutrients as well.
Consequently, I think that the mobilization of mercury is widespread and your study will have rather broad
application.
Please send me a copy of the Stober et al. 1996 report so that I can see what was discussed earlier (in case I get asked
questions about it).
Attached are my reviews and comments on the Technical Report.
Answers to your questions are given below:
1. Is the purpose of the report clear and does the report achieve this purpose? Yes.
2. Are the report sections appropriate? Yes but there is some redundancy.
Are there additional hypotheses that might be appropriate to address with this monitoring design? Yes, see below.
I think that the color of the water has inhibited the photoreduction processes which would remove the mercury which
comes in the precipitation.
3. Are the conceptual models reasonable? Yes but photoreduction must be quantified. Are there additional
hypotheses that might be appropriate to address with this monitoring design? Yes. The role of DOC and other
ligands in changing the BAFs in the upper reaches.
4. Are the report conclusions supported by the information collected both from this program and from other
on-going or completed projects? Are there other sources of information that support or refute the
conclusions? I think that the proportion of MeHg to Thg is very high. Other literature values should be given.
62
-------
5. Are the recommendations for future studies reasonable? Yes
Are there additional studies that should be considered and why? Yes, but include the part about the interaction with
DOC both as it influences BAFs, photoreduction of Hg2' and photodecomposition of MeHg.
Sincerely,
David Lean
Department of Biology
University of Ottawa
P.O. Box 450 Station A
Ottawa, KIN 6N5
63
-------
Review of Technical Report by Stober et al. "Monitoring for Adaptive Management: Implications for Ecosystem
Restoration"
One of the strong aspects of the work is the spatial coverage and the selection of the stations. Too often in the past
people went to easily accessible locations that would bias the analyses (because they were already impacted). The
strength of the present study is the extent of the survey. This should be emphasized more.
Response: The spatial extent and the ability to make population estimates has been emphasized more
in the revised report.
Executive Summary: covers the topic but ends without any forceful conclusion. As discussed below there are some
conclusions that are unequivocal, Phosphorus and DOC inputs to the Everglades have done what? The principal
component analyses was not sufficiently developed. 1 know what was found but some additional help to the reader
in the interpretation would add to the weight of the discovery.
Response: Conclusions have been added to the Executive Summary.
Chapter 2. Study Design This report provides a lot of information and to get to the heart of the subject effectively,
I would move supporting information to appendices. For example, 1 would suggest taking the essential information
from pages 2-5 to 2.10 and moving it to an appendix. We don't want the reader to get bogged down with too much
technical stuff At the same time, this is vital information that should be available.
Response: While the information is technical, this provides the support and documentation for one of the
report's strengths (see comment above) - the spatial extent and population estimates. The
information will be retained in the chapter.
Table 2.1 is especially useful. The only part that 1 question is the statement that low pH promotes methylation.
Perhaps consider rewording this to say methyl mercury is often found in low pH waters. There must be a sufficiently
low pH were methylating bacteria are no longer active.
Response: The sentence will be modified. However, there can be abiotic methylation occurring at low
pH (e.g., pH < 5.0).
Chapter 3. Materials and Methods 1 thought this was especially well done. There was enough information for the
reader to appreciate what was done but a reference was provided for the detailed procedure used (usually in first rate
journals)
Chapter 4. General Characteristics of the Water Regime I find the oxygen levels in the canals and the marshes
especially alarming. Some reference to the very poor light penetration would be useful. This would also support the
observation that the chlorophyll levels seem to be lower than that expected from the total phosphorus concentration.
It appears to be a light limited system and some comments on euphotic zone depth of light penetration to total depth
would be appropriate. It is my opinion that the dark color of the water is a principal factor in mercury contamination
of the ecosystem (see below). Since turbidity values are provided and light penetration is related to turbidity a
correlation at selected sights between extinction coefficient for photosynthetically active radiation or the 1% light
level and turbidity could be generated and used to provide the depth of euphotic zone. In the marshes, the DO values
are higher. Somewhere, the percent saturation at 26-35 degrees should be provided. In ENP for example, the oxygen
values must be near saturation. The implication of the second paragraph on page 4-9 is significant. Diel patterns of
oxygen can influence microbial activity. For example, at night are levels sufficiently low to cause a diel pattern in
methyl mercury production?
64
-------
Response: We did not measure Secchi depth or light penetration. A comment has been added that,
based on general observation, the canals are dark water systems. The percent saturation of
DO at 26 to 35° C was also added to the text.
In the second paragraph of the Synthesis section (page 4-10) it reads "specific conductivity patterns provide an
indication of..add and DOC after conductivity. As will be discussed below the impact of EAA in adding DOC
to the system is contributing to the mercury problem and the patterns of DOC can be traced throughout the everglades
system.
Response: The sentence has been modified.
Prior to recent human influences on the Everglades ecosystem, water was supplied mostly from rainwater (and the
buffering capacity and conductivity would be low). During these conditions the DOC levels were also likely quite
low. With the changes in hydrology and water diversion, water with much higher calcium and buffering capacity was
injected into the system. This may be significant.
In the synthesis section some discussion of the % saturation of oxygen at these high temperatures would be
useful. The standard of 5 mg/L is interesting. What is the % saturation at 35 degrees? Most people work with water
in the 20's not the 30's.
Response: % DO saturation has been added to the text.
The statement "that biogeochemical reaction rates might be expected to be relatively rapid in these systems"
is not clear.
Response: The sentence has been modified to state "that because reaction rates are temperature
dependent, biogeochemical reaction rates in the warm tropical Florida systems are expected
to be more rapid than in temperate systems."
Chapter 5. Habitat. 1 have no comments.
Chapter 6. Soils
Page 6-2 (top of page) Other factors which contribute to subsidence of soils include the influence of adding
water with substantial calcium in it. This well buffered water can alter the pH of the system (see above). For example,
a bog has a very low pH but if well hard water is added or the water level is reduced the pH can be altered and
microbial activity (even in the absence of added nutrients) can be turned on with the net result of a substantial loss
in soil substrate. DO WE HAVE ANY RELIABLE HISTORICAL INFORMATION ON THE pH OF
EVERGLADES WETLANDS? I think that this is vital information. Could it be that the pH in the peat regions was
in fact much lower during the 1930-1950 period? This may also be a factor in the loss of sediment from the marshes
discussed at the end of this chapter.
Response: The investigators are unaware of any such information.
On page 6-4 near the top it states that "maximum soil loss calculate (added a d for past tense) over the last
50 years indicate (add s) that the portion of WCA3 north of Alligator Alley has lost over 0.9 m (3 feet) of peat. I find
this astounding and this must be a factor in the overall degradation of the Everglades.
Response: Agreed.
65
-------
On page 6-7 pH is discussed but with the exception of the 5.8 value in WCA1 all are near or above 7. Clearly,
some historical values would shed light on the discussion above.
Response: Agreed, but the investigators are unaware of any such historical information.
The heading for Figure 6.6 is not clear. Is this a plot of Log Bulk Density or Bulk Density plotted on a log
scale. I think it is the former so the heading should be log bulk density.
Response: This is Log (Bulk Density). The heading has been clarified.
Chapter 7. Nutrients
Knowing that (line 3,4) that the marsh hydrology was precipitation driven is a key part of the story. During
this time the pH was likely less well buffered and was likely lower than it is today. Certainly, the water was clearer
and this has a profound influence on the mercury problem. This will be discussed more fully below
I would suggest replacing the word "drastically" (page 7-3 near the middle) with something like "rapidly".
Response: Change made.
On page 7-5 near the middle of the page, there is a statement that the TP decreases exponentially". Be careful
that someone does not fit the data and find that it is not an exponential. It does decrease rapidly.
Response: The decrease is exponential, but the sentence will be modified to state that there is a rapid
decline in TP, approaching an exponential decline.
On page 7-6 the TP "hot spots" are discussed. These are especially prevalent during the dry season. Could
it be that the soils become sufficiently to cause internal loading?
Response: We do not have the data to determine this, but it is an interesting hypothesis that could be
tested.
The total nitrogen (TN) concentrations in Table 7.4 are no doubt correct but the ratio of TN.TP to infer
phosphorus deficiency should be viewed with caution (Page 8-5). The ratio of DON:DOC is about 0.1. This means
that much of the DON in some areas would be dissolved organic nitrogen which is quite refractory and is not utilized
by plants or microorganisms to any great extent.
Response: So noted.
Chapter 8. Mercury
At the bottom of page 8-1 perhaps there should be some discussion of the influence of drainage of peat soils.
Historically, the rain fell on flooded soils and would run off without moving through the soils. Later, when soils were
drained, rain would run through the soils and carry easily dissolved forms.
Response: This idea will be added to the text.
On page 8-3, it states "Hg being washed out by rainfall". It is the Hg2+ form that is soluble in rainwater. The
statement about gaseous Hg coming into contact with the water requires that it is being oxidized after it equilibrates
with the water phase. We have some evidence that this might occur in some areas but is there evidence that it occurs
here?
66
-------
Response: See Krabbenhoft et al. 1998.
I divided the reported deposition rates of total mercury of 40-50 kg/yr, measured by Landing et al., by the
area (7600 sq km) and got a deposition rate of 6 /^g m'2 y'1 or by 5000 sq km and got a rate 10 /ug m'2 y1. These values
are consistent with other reliable values and some reference to other deposition rates should be made. It indicates
there is nothing unusual about the mercury deposition rates to the Everglades. In fact, it is just like in our remote
lakes in Ontario.
Please note the following. We have conducted experiments in a wide range of waters from Florida to the
high Arctic and found that photoreduction of mercury (usually Hg2^) results in the formation of Hg°or elemental
mercury which due to its high Henry's Law Constant and low water solubility will volatilize from the water back
into the air (one paper by Amyot et al enclosed). This rate has been estimated to be at least as high as the atmospheric
deposition rate in clear waters. Here the water is dark there is little or no photoreduction. It may be that the color of
the water in the WCAs inhibits photoreduction and the transport of Hg° back into the atmosphere. There are a
number of references that color and mercury loading are correlated. In addition, the most significant predictor of fish
mercury contamination in Ontario lakes is color. The bottom line is that in colored waters, more comes in and less
gets out. This almost certainly is one of the critical factors which has caused the mercury contamination to occur in
the Everglades.
Response: The variation of color in Everglades water is large. For the most part the canals have more
highly colored water than the marsh which is mostly clear. Although there is no doubt that this
flux is occurring it is difficult to determine how much color may be affecting the mercury flux
across the marsh.
Page 8-8 at the top. The "exploratory analyses' should be better defined.
Response: A discussion of the statistical analyses conducted to determine the critical variables will be
expanded in the text.
Pages 8-12 and further. There is some redundance from previous sections e.g., Fig. 7.3.
Response: Redundancy will be minimized. However, because of the report length, it is anticipated some
readers will only read selected sections. Some redundancy is retained to summarize results
from previous sections.
On page 8-13 the percent mercury as MeHg is give as 45%. This is extremely high. Without looking it up,
I think that values are generally near 1 % so what we have here is remarkable. Note that they axis of Fig. 8.20 is given
as the Ratio. This should be percent.
Response: The percentage of MeHg is probably the highest ever report for any system. Fig. 8.20 will be
corrected.
Synthesis on page 8-21. While mercury loading was dominated by atmospheric deposition the rate expressed
in ^g/m2/yr was not unusual. Furthermore, we have seen photochemical reduction rates at levels which would send
this amount back into the atmosphere again (see above). This likely cannot occur in the Everglades due to the dark
color.
Response: It most likely does occur since most of the marsh water is clear, and relatively shallow.
67
-------
The patterns with latitude are especially interesting. It seems that there is partitioning of methyl mercury
between the DOC perhaps or other dissolved ligands such as some of the sulfide or thiol compounds that causes a
reduction in the bioaccumulation factor. As such, it is there but not completely available to the organisms. Perhaps
there is also a biodilution and this illustrates the importance of the diet studies that were so well done. Alternatively,
the MeHg partitioned to the DOC becomes available later on. These questions illustrate the need for more studies
on the interrelationships between the forms of mercury and DOC. It is also known that MeHg can also be
photodegraded. This almost certainly is inhibited in brown waters.
Response: These thoughts will be used to expand on the discussion in the text. Other than the canals
and parts of the marsh with excessive phosphorus loading the water is extremely clear.
Volume II. Appendices. I have no comments
Comment of proposed solution
From the comments made above, I think that there should be studies to find ways to further reduce phosphorus
leaving EAA. In addition, after reading this report there must be controls on DOC export from EAA. If the waters
were clean, atmospheric deposition of mercury could easily be balanced by photoreduction (especially under the
strong Florida sun). We would be pleased to make some of these measurements. It would seem that if the high DOC
could be eliminated by some trick (which may even reduce phosphorus) the problems would not be as severe.
Response: We agree with your comments regarding phosphorus and DOC, however, photoreduction
doesn't seem to have the effect observed in other systems. Without reducing the atmospheric
wet deposition of mercury from local sources balancing the mercury flux in the system will
remain problematic.
68
-------
SFWMD
FAX MEMORANDUM
To: Dan Scheidt, US EPA
From: Christopher McVoy, HSM, SFWMD
Subject: Review of So. FL Ecosystem Assessment, Vol. I Technical Report
Date: 9 July, 1998
CC: Jerry Stober
Dear Dan:
Thank you for the opportunity to review this presentation of your project. This synoptic, system-wide study
is a fine piece of work. The soil change information is striking and clearly very important. Nicely backed up with
the organic matter and bulk density data.
I've made a few suggestions, most minor. Overall looks good. I would strongly encourage you to publish
the graphs and maybe even the raw data in a scientific journal.
Christopher McVoy, Ph.D.
P. 6-1 "High rockland . ." Rockland was present only from about half way between New River and Miami River
on south; to the north was all sands . . .
Response: Agreed.
• Stephens (1956): should probably read "once contained the largest body," rather than "contains"
Response: Agreed.
• "depth up to 6 m probably mean elevation of 20-21 feet; the pre-drainage peat thickness was about
3-4 m max.
Response: Agreed.
P. 6-2 Stephens (1984): Stephens would very likely suggest emphasizing process (6) oxidation, and (5) burning.
These are irreversible, and are the dominant processes.
Response: Agreed.
• Add following refs: Shih, S.F., J. W. Mishoe, J. W. Jones, and D.L. Myers. D1979. Subsidence related
to land use in Everglades agricultural areas. Trans. Amer. Soc. Agric. Eng. 22(3):560-563.
Shih, S.F., E.H. Stewart, L.H. Alleo, Jr., and J. W. Hilliard. 1979. Variability of depth to bedrock in
Everglades organic soil. Soil Crop Sci Soc ???. 38:66-71.
69
-------
• "public Everglades" suggest using different name to emphasize not only ownership, but the land use
difference form EAA, i.e., even the so-called wildlife habitat areas have lost soil under govt/public
stewardship. Strongly agree on "poorly documented"!! Your work fills a key gap.
Response: Agreed.
• "unimpeded surf.." Suggest emphasizing here the combined effect: (1) eliminated northern inflow
from Lake 0. overflow; (2) lowered water depths due to major canals; and (3) lack of impoundments
to keep water from "just running out." Combination is important to emphasize as the pre-drainage
system also had "unimpeded flow," but because of sufficient northern input, and no drainage canals,
was able to keep peats wet.
Response: Agreed.
P. 6-3 "A geostat rep . . ." (1) make more specific - kriging, nearest neighbor, etc. If kriging was used, might not
be bad idea to include a map of the estimated uncertainty/error. (2) suggest one sentence recapping method
of measurement here; mention includes both organic and mineral soils.
Response: Agreed.
• Table 6.1 Very nice summary. Suggest 2 signif. digits only on % O.M.; use leading zero on s.d.'s
for b.d.
Response: Agreed.
• "diff in peat thick . . ." need to give methodology here - how the difference map was
calculated/created.
Response: Agreed.
• "Soil volumes . .." this is separate topic - suggest moving to end of 6.2.1 section.
Response: Agreed.
• "Davis . . reports . ." If memory serves, Davis did give some methods - might even quote him.
Should also emphasize that he had no. of years on foot, boat, in field in So Fl, and wrote whole
monograph on peat deposits. In other words, he is a credible and highly qualified source . ..
Then, as separate issue, take up point of how to deal with his 2 foot intervals.
Response' Agreed.
Pg. 6-4 "hi or lo extreme" replace "extreme" with "threshold"
Response: Agreed.
• "Max soil loss calc" "calculated"
Response: Agreed.
70
-------
• "Up to 1.5 m (5 ft)" replace "3 to 5 feet" (there is a slight difference).
Response: Agreed.
• Question An alternative approach to the min max business would be to take the mean value and add
a range: Ex: 3 to 5 ft class becomes 4 plus or minus 1 ft. This might avoid the lightning rod for
criticism set out to be discussing the maximum change in thickness; nobody can argue with a mean.
Response: Noted.
• northeast S. Slough Capitalize (or not) consistently.
Response: Agreed.
• "53% of volume..." Suggest presenting loss percentages first as depth %s, rather than volume %s --
more direct. In later paragraph, come back to volume.
Response: Agreed.
• "system soil volume..." "system" is not clear; make more specific.
Response: Agreed.
• Table 6-2 (1) Soil volumes - max of 2 signific. figures; three not justified by methods. (2) suggest
1 signific. fig. For Min/max change (%) cols. Shouldn't plus signs be neg.? --loss of peat? Didn't
understand first row: "-3.0 (4%)" - why different than other rows?
Response: (1) Agreed. (2) Some calculations do show an increase in soil thickness.
• Percent O. M. Need to state sample methods here — how deep were cores? Maybe how % OM
determined (or refer on this to Methods Chapter). Make "geostat. represent." more specific.
Response: Agreed.
-5 "0.76% to 97.5%" replace: "less than 1 % to 98%"
Response: Agreed.
• "Are highly inorganic" replace: "primarily mineral".
Response: Agreed.
• "Mean of 92.0 ±6.8%" Replace: "92 ±7%" (similar on subsequent %s)
Response. Agreed.
• Bulk Density again, more specific on "geostat. display", maybe "kriged map."
Response: Agreed.
71
-------
on correlation w/OM suggest restricting sample to the >40% (39%) portion of your sample, which
would cover the areas known as the organic soils, then recalculating the regression, might try linear-
linear too.
Response: Noted. Linear regression was performed. Plot was much more scattered.
P. 6-6. Soil redox: can't comment knowledgeably on this.
Response: Noted.
• Transects: could map with location of transects be included? perhaps also some comment on the
graphs and their relation to what was seen in the field? (Particularly for thickness and O.M.)
Response: Map of transects is included as Figure 2.1.
Figs:
• Fig. 6.1 legend recommend clarifying that the 1946 is thickness of peat soils only; 1996 is soil
thickness, whether peat or mineral or mixed...
Response: Agreed.
• Fig. 6.2 (1) may be misleading to separate off WCA3A N of A. Alley — this was only a partial
hydrologic separation - 3 A N generally seems to be too dry rather than too wet... (2) What happened
to ENP along Tamiami Trail - funny gap?
Response: (1) Agreed. The purpose was to divide the system into geographic areas for data analysis
and interpretation. The figure is not intended to infer a hydrologic separation.
(2) Agreed. Gap is inadvertent, map will be corrected.
72
------- |