External Peer Review of EPA’s

“Proposed Methods and Approaches for Developing Numeric Nutrient
Criteria for Florida’s Inland Waters”

Peer Review Comments

Draft Final

EPA Contract No. EP-C-07-059

Work Assignment No. 1-16

Submitted to:

Jacques Oliver

Health and Ecological Criteria Division (4304T)

Office of Science and Technology

U.S. Environmental Protection Agency

1200 Pennsylvania Avenue, NW

Washington, DC  20460

Submitted by:

Eastern Research Group, Inc.

110 Hartwell Avenue

Lexington, MA  02421-3136

August 11, 2009

Printed on Recycled Paper

QUALITY NARRATIVE STATEMENT

ERG selected reviewers according to selection criteria provided by EPA.
EPA confirmed that the scientific credentials of the reviewers proposed
by ERG fulfilled EPA’s selection criteria. Reviewers conducted the
review according to a charge prepared by EPA and instructions prepared
by ERG. ERG checked the reviewers’ written comments to ensure that
each reviewer had provided a substantial response to each charge
question (or that the reviewer had indicated that any question[s] not
responded to was outside the reviewer’s area of expertise). Since this
is an independent external review, ERG did not edit the reviewers’
comments in any way, but rather transmitted them unaltered to EPA.

Table of Contents

Peer Reviewer Comments

  TOC \o "1-2" \h \z \u    HYPERLINK \l "_Toc238273712"  1.	Please
comment on the use of the approaches and methodologies presented in the
technical document, “Proposed Methods and Approaches for Developing
Numeric Nutrient Criteria for Florida’s Inland Waters,” for deriving
numeric nutrient criteria, given the available data, protective of the
Florida’s designated uses of Florida’s class III waters (i.e.,
“Recreation, Propagation and Maintenance of a Healthy, Well-Balanced
Population of Fish and Wildlife”).	  PAGEREF _Toc238273712 \h  3  

  HYPERLINK \l "_Toc238273713"  2.	Chapter 3.  Please comment on the use
of the SCI, LVI, and SDI as reliable indicators of the aquatic life
designated use attainment and whether these metrics appear to be
sensitive indicators to nutrient enrichment?	  PAGEREF _Toc238273713 \h 
11  

  HYPERLINK \l "_Toc238273714"  3.	Chapter 4. Please comment on the
approach used, given the available data, to derive the proposed
nitrate-nitrite criteria for Florida’s streams (< 40 pcu).	  PAGEREF
_Toc238273714 \h  15  

  HYPERLINK \l "_Toc238273715"  4.	Chapter 5. Please comment on the
regionalization of Florida’s streams in support of the nutrient
benchmark approach.	  PAGEREF _Toc238273715 \h  18  

  HYPERLINK \l "_Toc238273716"  5.	Chapter 6. Please evaluate the
stressor-response approaches for numeric nutrient criteria development.	
 PAGEREF _Toc238273716 \h  20  

  HYPERLINK \l "_Toc238273717"  6.	Chapter 7. Please evaluate the
Nutrient Benchmark Distributional Approach.	  PAGEREF _Toc238273717 \h 
24  

  HYPERLINK \l "_Toc238273718"  7.	Chapter 8. Please evaluate the
reliability of the Nutrient Longitudinal Study to protect the aquatic
life use downstream and whether this approach can be applied to other
estuaries or lakes in Florida?	  PAGEREF _Toc238273718 \h  27  

  HYPERLINK \l "_Toc238273719"  8.	Chapter 9. Please evaluate the
methods for developing chlorophyll a numeric nutrient criteria for
lakes.	  PAGEREF _Toc238273719 \h  30  

  HYPERLINK \l "_Toc238273720"  9.	Chapter 10. Please evaluate the
methods for developing chlorophyll a numeric nutrient criteria for lakes
based on stressor-response analyses.	  PAGEREF _Toc238273720 \h  34  

  HYPERLINK \l "_Toc238273721"  10.	Chapter 11. Please comment on the
“all streams” statistical distribution approach for deriving numeric
nutrient criteria	  PAGEREF _Toc238273721 \h  40  

  HYPERLINK \l "_Toc238273722"  11.	Chapter 12. Please comment on
extrapolating background nutrient concentrations as an approach to
deriving numeric nutrient criteria.	  PAGEREF _Toc238273722 \h  41  

  HYPERLINK \l "_Toc238273723"  12.	Chapter 13. Please comment on the
use of multiple regression and statistical distribution approaches to
deriving numeric nutrient criteria.	  PAGEREF _Toc238273723 \h  43  

  HYPERLINK \l "_Toc238273724"  13.	Chapter 14. Please comment on the
downstream protection approaches.	  PAGEREF _Toc238273724 \h  45  

  HYPERLINK \l "_Toc238273725"  Additional Comments/References	  PAGEREF
_Toc238273725 \h  51  

 



PEER REVIEWER COMMENTS

Specific Questions

Please comment on the use of the approaches and methodologies presented
in the technical document, “Proposed Methods and Approaches for
Developing Numeric Nutrient Criteria for Florida’s Inland Waters,”
for deriving numeric nutrient criteria, given the available data,
protective of the Florida’s designated uses of Florida’s class III
waters (i.e., “Recreation, Propagation and Maintenance of a Healthy,
Well-Balanced Population of Fish and Wildlife”).

What are the strengths and weaknesses of approaches and methods employed
given the available data?  

To what extent can each line of evidence be used independently to
support the derivation of numeric nutrient criteria for each of the
parameters EPA has recommended?

Given the data available, what additional technical considerations can
you recommend in the derivation of numeric nutrient criterion for each
parameter?

Reviewer 1	The driving force for creation of the document I am currently
reviewing comes from the Clean Water Act. The Clean Water Act (CWA)
establishes the basic structure for regulating discharges of pollutants
into the waters of the United States and regulating quality standards
for surface waters. The basis of the CWA was enacted in 1948 and was
called the Federal Water Pollution Control Act, but the Act was
significantly reorganized and expanded in 1972. "Clean Water Act" became
the Act's common name with amendments in 1977. Reading this whole act
quickly shows that the term fish is used approximately 68 time and
recreation approximately 72 times, leading to the general goal of
protecting waters so they are “Swimmable” and “Fishable” on into
the future.

This goal leads to an impossible management situation that the best
swimming lakes are generally clear unproductive lakes (oligotrophic) and
the best fishing lakes are generally green productive lakes (eutrophic).
A strength of the document is that it attempts at all costs to set
nutrient standards that are low, which would favor a swimming condition.
However, a great weakness of the document is that fish are not
considered in the document and if they were considered a low nutrient
standard would not favor “Fishable.” Additionally, the Florida Fish
and Wildlife Conservation Commission had no part in the creation of this
document, which I think is a great mistake and another weakness of the
document. Their insights as to how nutrients relate to fish and wildlife
should be part of this document. As the document and standards are set
up now, most of the excellent fishing and wildlife observing lakes in
Florida would be classified as impaired.

The strengths in the document entitled “, “Proposed Methods and
Approaches for Developing Numeric Nutrient Criteria for Florida’s
Inland Waters,” revolve around the general understanding that geology
and the Ecoregion concept is the master variable determining water
chemistry of aquatic systems. From my experience working on Florida
aquatic systems, the factor accounting for the largest amount of
variance in nutrient concentrations of rivers, lakes and estuaries is
the geology in which the system lies. This agrees with the basic
Ecoregion concept that the USEPA already uses. This is a great strength
of this document in Section 5 for streams and Section 14 for Estuaries
because both Sections show the importance of geology and/or Ecoregion.
Unfortunately, this same concept is not used for Florida lake nutrient
standards discussed in this document, another weakness.

Within an individual Ecoregion the next factors accounting for variance
in nutrient concentrations are all related to lake morphology and
hydrology. These factors can account for significant variance in
nutrient concentrations within an individual Ecoregion and for the most
part, these factors are ignored in the current document adding another
weakness. However, expanding the use of the Vollenwider Model concept
described in Section 14.2.1 could greatly strengthen the scientific
grounds for nutrient standards set in the document. This approach has
been extremely successful in lakes (Bachmann and Canfield 1981) and
shows great promise in Estuaries (Meeuwig et al. 1998; Meeuwig 1999;
Meeuwig et al. 2000).

Driving lake morphology and hydrology impacts on nutrient concentrations
and lake trophic state is the year-to-year variation in water inputs
from rainfall. I have tried to model impacts of rainfall to nutrient
concentrations in lakes with little success (Hoyer et al 2005). The
reason is that changes in lake water level can cause trophic state
parameters to increase (25% of the time), decrease (~25% of the time) or
have no impact (~50% of the time) depending on the dominant mechanisms
working within individual lakes. The importance of this concept for the
current document is that many lakes have tremendous year-to-year
variability and many of the approaches used in this document to define
nutrient standards do not have enough long-term data to define what true
background variance is. Thus, setting standards on a lake after a
100-year flood or draught could make it impossible for the lake to ever
reach the standard again. Some form of long-term variability needs to be
part of any nutrient standard.

The long-term variability of lake water chemistry add difficulty to the
strength of the document, which is understanding the important role that
color has on aquatic systems in Florida. However, lakes in Florida are
extremely dynamic and can go from a color of 20 PCU to 150 PCU after one
rain storm and maintain color over 40 PCC for months if not a year
before they clear up. Thus, using a fixed 40 PTU as a cut of for clear
and colored systems may not be the best approach for setting nutrient
standards. One month a lake many meet a standard and the next be in
violation.

The approaches to setting standards in this document should not be
independent of each other because all ecological processes are
connected. Unfortunately, the report entitled “Proposed Methods and
Approaches for Developing Numeric Nutrient Criteria for Florida’s
Inland Waters” shows that different groups were working independently
of each other and not communicating. For example, the rivers and estuary
groups followed an Ecoregion and water shed type approach and the lake
groups did not. All of the different Sections should follow the same
general game plan.

My recommendation for developing a nutrient standard for Florida is to
first use all available data and develop the most comprehensive
Ecoregions possible, following the lead of Griffith et al. 1997. The
upper 25% of the distribution of nutrient concentrations in lakes within
an Ecoregion should be examined to see if they meet use and those that
do not should be examined for restoration activities. Those that do
should be monitored to make sure they are not changing. To make this
determination, all long-term water chemistry data available should be
used to define normal background variance for nutrient concentrations in
lakes that could then be used as a standard classifying any lake
exceeding the defined normal background variance as impaired or not
meeting use. Once the remaining lakes have be classified as meets use,
monitoring lakes to see if they fall within normal back ground variance
would work to maintain lakes in a swimmable and fishable condition.
Finally, I would consult with the Florida Fish and Wildlife Conservation
Commission to define how nutrients impact the fish and wildlife
populations using aquatic systems and have them help define when
nutrients in a system cause harm to the fishable waters.

Canfield, Jr., D. E. and Bachmann, R.W. 1981. Prediction of total
phosphorus concentrations, chlorophyll a, and Secchi depths in natural
and artificial lakes. Can. J. Fish. Aquat. Sci. 38:414-423.

Griffith, G. E., D. E. Canfield, Jr., C. A. Horsburgh, and J. M.
Omernick. 1997. Lake regions of Florida. Report to the Florida
Department of Environmental Protection. U. S. Environmental Protection
Agency, Corvallis OR.

Hoyer, M. V., C. A. Horsburgh, D. E. Canfield, Jr., and R. W. Bachmann.
2005. Lake level and trophic state variables among a population of
shallow Florida lakes and within individual lakes. Canadian Journal of
Fisheries and Aquatic Sciences. 62: 1-10.

Meeuwig, J. J., J. B. Rasmussen, and R. H. Peters. 1998. Turbid waters
and clarifying mussels: their moderation of empirical chl:nutrient
relations in estuaries in Prince Edward Island, Canada. Mar. Ecol. Prog.
Ser. 171: 139-150.

Meeuwig, J. J. 1999. Predicting coastal eutrophication from land-use: an
empirical approach to small non-stratified estuaries. Mar. Ecol. Prog.
Ser. 176: 231-241.

Meeuwig, J. J., P. Kauppila, and H. Pitkanen. 2000. Predicting coastal
eutrophication in the Baltic: a limnological approach. Can. J. Fish.
Aquat. Sci. 57: 844-855.

Reviewer 2	

In my document Question 1 is  a "General Charge Question" with three
general considerations (i.e. strengths and weaknesses, validity of lines
of evidence, additional technical considerations for deriving criteria)
that I took into consideration throughout the document in reviewing the
individual chapters.  So, the general charges are addressed throughout
the review, of course depending upon the contents of the individual
chapters.



Reviewer 3	The State of Florida has developed multiple lines of evidence
in support of numeric nutrient criteria for inland waters.  Those lines
include effects-based relationships between biological communities and
nutrients, stressor-response relationships between nutrients and
periphyton indicators for streams and chlorophyll a for lakes and
estuaries, descriptive statistics, models linking downstream condition
to upstream nutrient loads, and heuristic arguments based on historical
comparisons, paleolimnology, and review of extant literature. 
Collectively, these lines of evidence are defensible and support numeric
criteria.  For some of the analyses, a few minor refinements are
suggested to help make the results less ambiguous, and in some cases,
more applicable.        

The biological indices, especially the Stream Condition Index (SCI)
based on macroinvertebrate community composition and structure, and the
Lake Vegetation Index (LVI), were convincingly demonstrated to be robust
measures of resource condition relative to a gradient of anthropogenic
disturbance.  The Stream Diatom Index was strongly influenced by
regional geology (however, component metrics were, apparently, sensitive
to a nutrient gradient).  Significant, though necessarily fractional,
amounts of variation in SCI and LVI scores were explained by a nutrient
gradient, particularly nitrogen.  Unfortunately, the small sample size
(N~75) used in the statistical analyses for both stream invertebrates
and, to a lesser extent, lake vegetation, appeared to limit the ability
to “read” the nutrient signal through the noise of confounding and
collinear variables.  The report was vague with respect to which SCI
data were used in the analyses.  For example, no mention is given as to
whether the data were culled to remove sites influenced by spurious
factors (e.g., sites known to be affected by toxic levels of pollution,
highly degraded physical habitat, desiccation, combined sewer overflows,
etc.) prior to analyses.  If this was not done, it is highly
recommended.  

The derivation of nitrate-nitrite criteria for clear, spring streams
employed historic comparisons, stratification by water color, laboratory
dose-response experiments, analysis of hydrologic and demographic data,
and change point analysis based on data collected over a nutrient
gradient.  As such, the methodology is particularly robust, and the
criteria defensible, especially in terms of recreational and aesthetic
uses.

The regionalization with respect to nutrients was straightforward,
methodologically sound, and in keeping with the results of others (i.e.,
collapsing multiple ecoregions into several nutrient regions - sensu
Robertson et al. 2006).       

The stressor-response analyses presented in Chapter 6 were conducted on
data retrospectively, but managed to avoid the tinge of data dredging by
acknowledging collinearity among stressor variables, and by noting that
a measured stressor variable can serve as a proxy for unmeasured
variables.  The analyses showed statistically significant relationships
between nutrients and biological metrics based on either
macroinvertebrates or diatoms, but given the aforementioned caveats,
were not considered sufficiently strong for derivation of criteria. 
Although the relationships were weak, and may not, in isolation, serve
as a basis for criteria, change points in the biological metrics in
relation to either TP or TN (or DIN, NOx, etc.) would serve as an
additional line of evidence for criteria.   The results from the rapid
periphyton assessment, however, appeared to provide the least ambiguous
results, yet were used only to support the general theme that nutrient
enrichment produces observable effects.  Again, change point analysis on
these data would augment the overall effort.    

The methods used to identify least-impacted benchmark sites in Chapter 7
were rigorous and resulted in a population of sites that yielded a
reference range for nutrient concentrations in the various nutrient
regions.  The report correctly cited an important caveat for using a
percentile approach, namely, that the approach fails to identify a
threshold concentration over which biological impairment is increasingly
likely.  The strength of the percentile approach hinges on the
supposition of the absence of evidence is evidence of absence (i.e., no
biological impairment observed at a few sites with high nutrient
concentrations, ergo, the criteria should be protective).  However,
before this can be concluded, mitigating factors, specifically those
that allow for good biological condition to be maintained in the
presence of known or suspected stressors, must first be ruled out, or at
least understood with an eye toward implementation.  The reference range
approach should be supplemented with an “all” sites approach (as in
Chapter 11), wherein distributions of nutrient concentrations are
examined by categorical level of SCI scores (e.g., poor, fair, good,
excellent ranges) for all sites within a region (or regions) where
biological data exist.  This would provide a contrast between degraded
sites, sites meeting expectations, and benchmark sites that may help
inferentially show the nutrient concentrations over which degradation is
likely.               

The longitudinal study of the Steinhatchee and Waccasassa basins and
estuaries loosely demonstrated that nutrient concentrations found in
minimally disturbed watersheds are not associated with detrimental
effects, particularly loss of seagrass beds.  This study would have been
much more compelling if an enriched system, especially one enriched with
nitrogen, had been offered for comparison.

Chapter 9 details a basis for chlorophyll a thresholds for Florida lakes
based on empirical relationships between TP, TN and chlorophyll a (i.e.,
Trophic State Index), and augments that with several other lines of
evidence, specifically, paleolimnology, BPJ, literature review, user
perception, existing condition, and reference distribution.  The report
and appendices were vague with respect to which data were used (a subset
of those referred to in Chapter 10?) and the strength and form (i.e.,
linear vs. non-linear) of the relationships forming the TSI; however,
having a direct, and empirically derived, stressor-response model on
which to base criteria is clearly a strength.  The additional lines of
evidence roughly converge on the proposed threshold of 20 ug/l
chlorophyll a (i.e., meso-eutrophic boundary) for colored lakes, and the
lower threshold of 9 ug/l (oligo-mesotrophic boundary) for clear lakes
in the panhandle region.  The exercise in investigating the relationship
between cyanobacteria and chlorophyll a was offered as evidence that the
proposed chlorophyll threshold will protect beneficial uses; however,
the exercise was equivocal.  The real question is whether increasing
nutrient concentrations over time alter the composition of the algae
community in a given waterbody, thereby increasing the probability of a
harmful bloom.

The stressor-response analyses for lakes provided in Chapter 10
presented two lines of effects-based evidence, one based on an aquatic
vegetation index (LVI), the other on chlorophyll a.  The LVI was
statistically associated with mean nutrient concentrations averaged over
the preceding year, thus demonstrating a stressor-response relationship.
 The strength of association (for both lake types), as the report
acknowledges, was what one might have expected given that multiple
factors influence lake vegetation, but was deemed insufficiently robust
for derivation of nutrient criteria, except for TP in clear lakes. 
These data, however, should be used in change point analysis as an
additional line of evidence supporting numeric criteria.  Correcting for
the extraneous environmental variables by using the residuals from a
regression of the LVI on the ancillary variables in change point
analysis might help.  The statistical associations between chlorophyll a
and nutrients were much stronger, based on a large data set, and in
keeping with numerous published results.  These relationships were used
to derive criteria.          

 Compiling descriptive statistics on ambient water chemistry, as in
Chapter 11, is a necessary step toward gaining a broad understanding of
the system at a population level. However, the reference approach, in
the absence of biological data, provides the weakest line of evidence
for criteria development.  The all sites chemistry data should be, as
previously mentioned, paired with biological data (i.e., the SCI, SDI,
or a sensitive component metric) and reexamined within ranges of the
biological data.    

The effort toward extrapolating background nutrient concentrations
(Chapter 12) was equivocal, perhaps because the work was done ad hoc
with available data.  Sorrano et al. (2008) demonstrated the application
for lakes, but the applicability for streams is questionable, given that
streams are open systems and inherently more variable than lakes. 
Furthermore, the approach, with respect to streams, would seem to offer
little additional resolution beyond that given by the exhaustive
benchmark approach (Chapter 7).

The multiple regression models provided in Chapter 13 offer a more
refined way to derive criteria than the conventional linear models
appearing in Chapters 9 and 10, though the results from all three
chapters generally comport each other, adding strength to the overall
effort of defining criteria.  

The meat of Chapter 14 is section 14.1.14.  The description of model
development preceding it is highly ancillary for the exercise of
deriving nutrient criteria.  A very general and emphatically brief
summary of the TMDLs would suffice as a precursor, with details of the
models appearing as appendices.  To wit, simply stating and referencing
that approved TMDLs have been developed for the respective estuaries
identifying the target loads listed in Table 10 would suffice.  Given
that water quality models are abstractions, and have a good deal of
inherent uncertainty, the derived stream reach concentrations suggested
as criteria in Table 10 should be compared to reference ranges (i.e.,
which percentile for both the benchmark sites and the all sites do they
match).  Also, some expression of uncertainty around the estimates
should be provided.  That said, linking protection of downstream uses
with upstream concentrations is an important component of the criteria
derivation process, and the results presented in Table 10 appear to
generally comport with the other finding listed throughout the report,
thereby adding strength to the overall effort.

Reviewer 4	The document has remarkable breadth of coverage made possible
in part because Florida made the investment in building a massive data
record.  Considerable technical effort has been devoted to exploration
of different approaches for development of numeric nutrient criteria. 
Not all approaches will be found to have equal merit for developing
nutrient criteria in Florida, but the opportunity to compare and
contrast those approaches is invaluable.  After consideration of
relative merits, EPA and Florida should be well-positioned to move
forward with criteria development.

It was disappointing, although understandable, that there was no
synthesis component that might consider how the various approaches fit
together.  It does not seem likely that all approaches will have equal
merit in this case.  Will one approach play a primary role and others be
used as confirmation?  At some point, it will be necessary to reconcile
differences among the numbers produced by each method.  This step may
require additional work on some methods; the level of effort, especially
with regard to application of statistical tools was not uniform across
the approaches.  For example, more might be done to extract thresholds
from the “wedge-shaped” response relationships.

I would like to see more discussion about the interface between science
and policy.  Biological condition, for example, can be characterized
largely by technical evaluations, but the threshold consistent with use
attainment is mainly a policy decision.  That distinction is sometimes
obscured behind discussions about the precision of possible thresholds.

Reviewer 5	The document reviewed represents a compilation of methods and
approaches from several sources, namely Florida’s Department of
Environmental Protection (DEP), USEPA and their contractors, by which
EPA proposes to assist the State of Florida in determining nutrient
criteria for the protection of designated uses of inland waters as
mandated by the Clean Water Act.  The effort presented here represents a
wide-ranging assessment of techniques, from a number of perspectives, to
determine comprehensive nutrient criteria for both N and P for inland
waters.  A series of relatively thorough, though often high-level,
considerations of standard techniques is presented for establishing
numerical water quality criteria for lakes, streams, rivers, springs and
downstream waters. 

 The approaches cover a range of biological response variables including
invertebrates, periphyton, algae, Submersed Aquatic Vegetation (SAV),
chlorophyll a.  The controllable stressor variables of direct concern
are nitrogen and phosphorus, although mention is made of the importance
of other stressor variables, such as DO and color.  EPA recognizes that
a single nutrient criterion for all waters is unrealistic and will
result in inadequate protection of inland waters.  Therefore, EPA
recommends the states adopt criteria for two stressor parameters, in
this case TN and TP, and two response variables, for example chlorophyll
a and water clarity.  

The proposal does not explicitly include a second response parameter for
consideration and focuses on chlorophyll a as a response parameter but
does not treat water clarity at all.  Though many secondary response
variables are analyzed: invertebrates, lake vegetation, seagrass,
fishery, lake trophic state, only chlorophyll a is proposed as a
response parameter.  Nor does the document consider stressor parameters
other than TN and TP.  Other candidates, such as low DO, organic matter
load or sediment load, may be as or more appropriate stressors for some
waters, however are not formally incorporated into recommendations. 
This is likely a function of the lack of availability of data and a
desire to address (at least initially) the most common response
parameters being used throughout the country.

DEP and EPA clearly recognize the complex nature of the task at hand:
proposed criteria must have a high probability of being effective and of
eliciting the predicted results or the program will be undermined. 
Therefore, it is recommended that several elements be included in the
proposed approaches, to wit:

The approaches must be geographically based and be applicable statewide

There must be provision for differential application of criteria,
depending on conditions and geography

Downstream effects must be evaluated

The approaches should be ecosystem-oriented

They should target specific response variables

Approaches must be standardized and transportable

The strategy must incorporate research and monitoring

Temporal variability should be taken into consideration

Sampling protocol and schedules must be matched to the timescales of
relevant processes

Of necessity, many simplifying assumptions are made in the various
analyses and recommendations throughout the document, some of which are
innocuous and others should be further evaluated.  Evaluations of these
assumptions are elucidated in the individual chapter assessments.  Some
approaches presented are likely not to be effective for all waters in
Florida and some approaches are not sufficiently developed or expounded
here to make a thorough consideration of them.  Although the approaches
should be geographically comprehensive, south Florida estuaries are
noticeably absent from the discussion and analyses.  The proposed models
for downstream loading calculations are insufficient to assess whether
criteria are protective in areas where downstream waters do not receive
a coherent mass flow, such as in Florida Bay.

Biological response is related to a number of variables that obscure the
dose-response relationships.  A better understanding of the underlying
interactions and processes will allow a refinement of the basic
relationship.  Understanding of the physics of the system is essential
to understanding nutrient processing: depth, bathymetry, residence time,
tidal prism are important factors determinant of rates and interactions.
 Also lacking is a clear recommendation for a gap analysis of
information required be obtained for improvements in the establishment
of criteria.  The document lacks an outline of the kinds of data
required to implement a comprehensive nutrient strategy.  Such a plan
would be effective if components include: a research and monitoring
plan, or a reference to one, and an adaptive strategy for reviewing the
effectiveness of the criteria, on a schedule, and revising the criteria
as necessary.

A general caveat about the hysteresis effect is also worthwhile at this
point.  While the aim is to maintain nutrient concentrations below a
threshold in waters that are not impaired, for waters that are impaired,
rolling back concentrations to the threshold concentration may not
necessarily be restorative of impaired waters to a non-impaired
condition.  Every increment above an empirically derived threshold
produces an ecosystem that is in further imbalance, with biogeochemical
alterations whose effects are not easy to detect, account or predict. 
Applying reference conditions or even dose-response experiments does not
account for changes leading to these hysteresis effects.  The trajectory
of the ecosystem backwards along a nutrient gradient is unlikely to be
linear or to be the exact reverse of the trajectory forward along that
gradient.  Often, the dose-response relationship is far different, and
less favorable in the reverse direction.  This fact, in context of the
results of the statistical analyses, experiments and field observations
included in this document argue for a protective threshold somewhat more
conservative than the bare minimum required for prevention of a step
increase in a relatively pristine system.

In summary, there is a good foundation for establishing nutrient
criteria presented here.  Mass balance models and simple
process-oriented accounting for nutrient transformations and flows would
go a distance in supporting the nutrient dose-response relationships and
frequency distribution benchmarks.  The proposal should succeed with
incorporation of a plan for continued research and monitoring that will
provide the information necessary to supply these models.  



Chapter 3.  Please comment on the use of the SCI, LVI, and SDI as
reliable indicators of the aquatic life designated use attainment and
whether these metrics appear to be sensitive indicators to nutrient
enrichment?

Reviewer 1	I understand the desire to tag a nutrient standard to some
form of biological measure because the Florida Statue F.A.C. 62-302.400
defines a designated use for its water of “Recreation, propagation and
maintenance of a healthy, well balanced population of fish and
wildlife.” With this in mind, I examined and tested the ability of an
Index of Biological Integrity (IBI), using fish data, to account for
variability in anthropogenic impacts (Schultz et al. 1999; Schultz et
al. 2000). I found that the IBI could not account for the variability in
anthropogenic impacts. The problem was that there are naturally
occurring gradients in Florida that are related to individual metrics
used to calculate an IBI that overshadow anthropogenic impacts (trophic
state, pH, lake surface area and others). Developing models relating
metrics to these naturally occurring gradients then examining residuals
and anthropogenic measures may be a good way of fine tuning IBIs but I
have not yet attempted that analysis.

For this vary reason I believe that the SCI, LVI and SDI are the wrong
approach for setting nutrient standards for the State of Florida. My
strength is not with invertebrates but I know individual species of
aquatic plants and diatoms are found in lakes with different water
chemistry along gradients of pH, alkalinity and nutrient concentrations
(Hoyer et al 2004; Beal 1977). As mentioned in my general comments these
water chemistry parameters correspond to different geological factors
across Florida. Thus any biological index that has metrics that vary
with a naturally occurring gradient will be of little use in determining
the extent of anthropogenic impacts. This seems to be pointed out with
the SDI in the report showing the relation between pH and SDI in Section
3.7. As I mention, there may be a way to model the relations between
SCI, LVI, SDI and naturally occurring gradients related to geology and
then examine the relations between the residuals and anthropogenic
measures.

Another large shortcoming of the LVI is that there is no metric in it
accounting for the total abundance of aquatic plants in a lake system.
The presence or absence of individual plant species and/or plant
community types have little impact on whole lake species composition of
other fauna. However, the lake’s total percent area covered (PAC) with
aquatic vegetation and/or percent volume infested (PVI) with aquatic
plants are directly related to the species composition of other fauna in
lake systems. For example, both fish and aquatic bird communities have
individual species that increase as a percentage of the whole community
when PAC and PVI increase and individual species that decrease when PAC
and PVI decrease. Additionally, fish and bird communities have
opportunistic species that maintain a constant percent of the community
as aquatic plant abundance changes.

Beal, E. O. 1977. A manual of marsh and aquatic plants of North Carolina
with habitat data. Tech. Bul. No. 247. The North Carolina Agricultural
Experiment Station.

Hoyer, M. V., D. E. Canfield Jr., C. A. Horsburgh, and K. Brown. 1996.
Florida freshwater plants a handbook of common aquatic plants in Florida
lakes. SP 189. University of Florida/Institute of Food and Agricultural
Sciences. Gainesville, Florida.

Schulz, E. J., M. V. Hoyer and D. E. Canfield. 1999. An Index of biotic
integrity: a test using limnological and fish data from 60 Florida
lakes. Transactions of the American Fisheries Society 128: 564-577.

Schultz E. J., M. V. Hoyer, and D. E. Canfield Jr. 2001. Reply to
comment: An index of Biotic Integrity: A test with limnological and fish
data from sixty Florida lakes. Transactions of the American Fisheries
Society 130:172-173.

Reviewer 2	Chapter 3.

I commend South Carolina DEP on the development of the SCI.  Clearly
they spent a lot of time and effort to produce a solid, scientifically
defensible product.  It is an excellent tool for assessing benthic
health for general pollution impacts (organic, chemical stresses, and
sedimentation problems). However, its use to detect nutrient pollution
is likely to be only partially successful.  It would work where a stream
was chronically polluted by nutrients (downstream of a point source with
no nutrient limits, a stream polluted by septic system leachate, a
high-fertilizer agricultural field perhaps); these situations would
produce chronic algal blooms and consequent semi-continuous issues with
bottom hypoxia or food chain disruptions from excessive Cyanobacteria. 
However, in a situation where nutrient loading is acute (i.e. rain event
driven) like in a suburb for instance, blooms may be short lived. 
Benthic invertebrates are known (especially in the South) to colonize
rapidly and be multivoltine, thus by sampling benthos rarely (such as
every couple of years, or even twice a year) acute impacts will be
missed and the SCI will be misleading in these circumstances.

As to the LVI, invasive aquatic macrophytes will colonize and overgrow
even pristine waters.  While elevated nutrients will certainly
exacerbate such a situation, invasive nuisance species often do not
require excessive nutrients to proliferate and the index will not
reflect nutrient loading.

Regarding the SDI, as noted in the material this diatom index has issues
especially with the naturally-acidic pH values present in Florida (and
other Southern) waters.  Benthic diatom indices were developed by Ruth
Patrick and others working in northern streams in Pennsylvania and such
areas, where acid streams were un-natural (caused by acid mine drainage,
for instance) and devastating to many species that were not evolved
under such conditions.  Thus, the SDI at present remains more of a goal
than a product useful for nutrient loading assessment in Florida waters.

Reviewer 3	Biological indices have been widely and successfully
employed. The SCI and the LVI for Florida were clearly developed using
well-established and sound methodologies, were convincingly demonstrated
to be sensitive to environmental gradients and anthropogenic
disturbance, and are therefore good indicators of aquatic life use
attainment.  The periphyton based SDI was apparently still in a
developmental stage for Florida waters, though the SDI has been
demonstrated in other regions to be particularly sensitive to pollution
gradients.  Chapter 3 does not explicitly address whether or not the
indices are sensitive to nutrient gradients; chapter 6 does.

Reviewer 4	The SCI and the LVI appear to be sensitive to human
disturbance making them useful reaching conclusions about use
impairment.  Reliability was tested by having experts evaluate
biological condition without foreknowledge of the SCI or LVI scores. 
There was very good correspondence between the expert assessments and
the actual index scores.  Each index (SCI and LVI) was developed to be
responsive to human disturbance as defined by a human disturbance
gradient (HDG).  The HDG for the SCI included a water quality index that
incorporated several water quality measurements, while the HDG for the
LVI included ammonia concentration.

Both indices – SCI and LVI – respond to changes in nutrients, but
they also respond to other facets of human disturbance.  [Chapter 7
presents a more compelling demonstration of the relationship between
disturbance and nutrients.]  The “wedge-shaped” relationships that
appear throughout the document are expected because of the multiple
pathways by which human disturbance can affect biological communities. 
Wedge-shaped relationships present a challenge in terms of defining
thresholds for nutrient effects, but the challenge is worth confronting
because the effect is real and important.

The distinction between reference and “impaired” sites is not so
clear that everyone will be convinced, the elegant statistical analyses
notwithstanding.  The underlying premise is that stressors produce a
gradient of biological response (Biological Condition Gradient).  When
experts evaluated sites for biological condition (and without
foreknowledge of SCI or LVI scores), they could agree about the
qualitative evaluation of sites at the ends of the spectrum (reference
vs. impaired), but did not agree about those in the middle of the
spectrum.

The results of the expert panel show clearly that reasonable people may
disagree about the biological condition of sites in the middle of the
stressor gradient.  Nevertheless, the analysis is used to support the
position that “not reference” equals impaired.  This position seems
to prejudge a policy decision regarding attainment, which is not
necessarily synonymous with reference.

Selecting the 2.5th percentile as the threshold for impairment would
seem to be the quantification of a policy decision regarding use
attainment.  Yet, it suggests that 2.5% of all reference sites would be
considered impaired.  Give the very rigorous screening procedures used
to select the benchmark nutrient reference sites, the threshold for use
attainment is set quite high.  A passage from the text (page 21)
suggests that the willingness to tolerate misclassification of impacted
sites may have been less than that for reference sites.

Therefore, the risk is low (virtually non-existent for the SCI) in
applying the biological assessment tool and falsely identifying impacted
sites as attaining an aquatic life use.

In contrast to the optimism I feel about use of the SCI and LVI, there
seems to be less to recommend the SDI (described as still under
development in other parts of the report).  The report concludes that it
is “an unreliable tool for assessing adverse human effects on stream
systems,” and I would not disagree.  The sensitivity of diatom
assemblages to pH is a strong confounding factor, but no one should be
surprised that diatoms are sensitive indicators of pH.  The extensive
literature on acidification is replete with examples.  Perhaps the
broader question concerns the merits of including the entire algal
community rather than restricting analysis to diatoms.

Reviewer 5	This chapter describes two existing index approaches
currently in use by DEP for assessing condition of streams and lakes:
the Stream Condition Index (SCI), and the Lake Vegetation Index (LVI). 
There is also a short discussion of the Stream Diatom Index (SDI) which
is currently under development by DEP.  It is not clear how the SDI
would be used in relation to the SCI.  All three techniques are tied
into the EPA’s Biological Condition Gradient (BCG) approach of rating
systems on a six-tier scale of habitat degradation, a standard
assessment methodology widely applied and validated throughout the
country.  The SCI was linked to the BCG via a calibration exercise and
the two systems cross-walked with a high degree of fidelity.

The SCI is based on macroinvertebrates and generally applied by
comparison to reference sites, since the pristine condition of most
sites sampled has long-since been altered by human impact.  Through
multiple lines of evidence, fundamental ecological concepts, expert
opinion, reference conditions, SCI ratings were established which
indicated impairment.  The technique must account for, and remove,
effects of natural variation and stress on community response variables.
 The methodology for establishing the rating scale and determining the
threshold for impairment appeared sound and well-founded.

Similar to the SCI, the Lake Vegetation Index incorporates four metrics
that together give an indication of the health of the water body
compared to a reference scale.   A calibration of the LVI to BCG similar
to that done with the SCI was performed with equally encouraging
results.  The LVI examines aquatic plant complexes as a measure of the
health status of the lake ecosystem.  The index is an integrative
approach that yields overall score of condition based on multiple
factors.  The application of the reference site approach is
well-justified and appears be based on a methodology that yields
statistically valid results.  The general interpretation of the highly
variable results in the low range of the Human Disturbance Gradient
(HDG) is reasonable.  Sampling guidance in order to reduce Type 1 errors
(false positives) is well constrained and reasonable.  Analysis of
variability at a site over time reflects the inherent temporal
variability that may be expected in the natural world when applying this
index.

Both the SCI and LVI approaches deliver a robust, standard way of
identifying impaired waters for which nutrient criteria should be
targeted and that are linked to the BCG used by EPA throughout the
country.  The utility of the SDI, based on diatom community structure,
is less clear at this point.  While the technique has been used in
Scandinavia (Soininen and Kononen, 2004) and Western Europe (Prygiel,
1994) for many years, the application to Florida waters is as yet
untested and will require further development.  If improvements are made
to the diatom index, it could provide a needed complement to the
information provided by the SCI for streams. 



Chapter 4. Please comment on the approach used, given the available
data, to derive the proposed nitrate-nitrite criteria for Florida’s
streams (< 40 pcu).

Reviewer 1	My major concern about the approach used to set
nitrate-nitrite standards for clear (<40 PCU) is that only concentration
is used and the total load of nitrogen is not accounted for in different
systems. This is important to me because I examined impacts of nutrients
to the growth of aquatic plants and periphyton in 17 central Florida
streams (Canfield and Hoyer 1988) and five coastal streams (Hoyer et al.
2004) and in each case concentration was not related to total biomass of
aquatic plants and/or periphyton. Both studies showed that physical
factors (canopy cover, water clarity, mean depth, stream velocity,
substrate type, and/or salinity) were more important to the growth of
plants and/or periphyton in the rivers than nutrient concentrations.
Additionally, each study showed that the percentage of nutrient
incorporated in the total biomass of aquatic plants and/or periphyton
was extremely small compared to the annual load suggesting that
regardless of concentration plenty of nutrients are available throughout
the year to support the growth of aquatic plants.

Canfield, D. E., Jr., and M. V. Hoyer. 1988. Influence of nutrient
enrichment and light availability on the abundance of aquatic
macrophytes in Florida streams. Canadian Journal of Fisheries and
Aquatic Sciences 45: 1467-1472.

Hoyer, M. V., T. K. Frazer, S. K. Notestein and D. E. Canfield, Jr.
2004. Vegetative characteristics of three low-lying Florida coastal
rivers in relation to flow, light, salinity and nutrients. Hydrobiologia
528:31-43.

Reviewer 2	Chapter 4.

I really liked the approach used in the derivation of a nitrate
benchmark for clear-water spring-sourced systems in Florida.  It was
based on multiple lines of evidence.  There were two types of bioassays
performed, and since I have utilized and published results of bioassays
on a number of occasions I was pleased by this.  I also liked the fact
that field data were included utilizing periphyton coverage and nutrient
data, with regressions performed.  In fact, strong statistical analysis
was a hallmark in this criterion derivation.

A couple of comments to improve the scientific validly of the approach
follow.  I noted that the great majority of the literature cited within
this chapter consisted of Florida DEP reports and other reports (USGS)
– which are “grey literature”, and not peer-reviewed by the
academic scientific community.  That weakens the arguments a bit
(although I agree that the proposed standard appears reasonable).  The
authors should get deeper into the peer-reviewed literature on this (if
there is literature available on the subject).  The second issue is that
of ammonium.  This is a form of inorganic nitrogen that is readily
utilized by algae, yet no data are provided on its prevalence in the
spring sites.  In North Carolina the inorganic N in groundwater can be
dominated either by nitrate or ammonium, depending upon sources (i.e.
nitrate often dominates downflow from golf courses, but ammonia
dominates where septic system leachate is a factor).  To strengthen
their arguments the authors need to provide data on ammonium in the
springs – if it is low then the proposed standard is valid, but if it
is high adjustments may need to be made.  In fact, ammonium is used in
Chapter 6 (see Fig. 6.1) so it should be taken into account here as
well.

Reviewer 3	Chapter 4 clearly articulates the beneficial uses of
Florida’s clear, spring fed streams by providing the historic
narrative and photographic comparisons. This approach is particularly
apt for the issue of nutrient enrichment, where concentrations tend to
rise slowly over time, attendant effects are subtle, and the changing
condition becomes culturally normative (at least until conditions become
egregious).  The empirical evidence presented in Figure 4-4 corroborated
the anecdotal historic accounts, and set the stage for the experimental
work.  The concentrations found to saturate algal growth in the
laboratory experiments dovetailed nicely with the change point
thresholds identified in the field surveys.  The TMDLs described in
section 4.7 pulled together multiple lines of evidence: comparison of
existing condition (i.e., nitrogen concentrations) with reference
condition, longitudinal field surveys, and change point analysis to
identify target concentrations.  Again, the thresholds found in the TMDL
work converged with the field surveys and laboratory studies.  In total,
Chapter 4 presents a highly defensible basis for nitrogen criteria
applicable to clear streams.

Reviewer 4	The proliferation of nuisance algae in Florida’s spring
ecosystems over the last 50 years is disturbing (and the photographic
documentation in the report is dramatic).  The attendant ecological
changes are significant, as are the concerns about impairment of
recreational use.  The scale of disturbance is astonishing, and there is
little doubt that it is of anthropogenic origin.  The application of
herbicides – intended to suppress non-native vegetation – apparently
created a disturbance that may have facilitated algal dominance.  It is
not clear how the disturbance may have complicate efforts to develop
nutrient thresholds.

The documentation of increasing nitrogen concentrations seems thorough. 
The connection to human population growth is logical, although the basis
for the proposed lag time should be strengthened.  It is asserted, but
not documented, that “nitrate-nitrite has been linked to many of the
observed detrimental impacts in spring systems.”  Nevertheless, I
would agree that lower nitrate concentrations are be desirable.

An issue not discussed is the appropriateness of deriving criteria based
on studies of a nuisance species (Lyngbya) in a disturbed ecosystem. 
The laboratory studies identify nitrate concentrations below which
Lyngbya biomass and growth rates are low.  I searched in vain for an
explanation of how this information would lead to conclusions about the
nitrate concentrations supportive of a healthy native plant community in
the spring ecosystems.  Even the field surveys contain no information
about nitrate concentrations in springs that still have healthy native
plant communities.

The rest of the chapter makes good use of field data to identify
thresholds based on periphyton abundance.  The case for the proposed
threshold of 0.35 mg/L needs to be stronger; visual inspection of the
figures might lead readers to conclude that a lower number would be more
representative.

Reviewer 5	This chapter described the derivation of criteria for TN and
TP for springs and clear streams in Florida.  The DEP recognizes the
unique importance of the clear stream class of Florida aquatic
ecosystems.  Pollution of groundwater and the reduction of groundwater
supply through consumptive use have increasingly become threats to
spring ecosystems, leading to replacement of native SAV by exotic
species, algal overgrowth and the development of thick benthic algal
mats.  Secondary problems include organic matter accumulation, hypoxia,
loss of water transparency and deleterious shifts in faunal composition,
including to toxic species.

It is proposed that the spring analysis and protocol for criteria
development be expanding to all clear stream systems.  This broadening
of target systems may be applicable but there are differences between
stream and a spring ecological systems which may present difficulty with
transportability.  These inequalities include differential source
waters, biogeochemistry and ionic composition and higher temperatures
that may affect biota differentially and promote uncertainty in
predictions deriving from nutrient enrichment.  Moreover, given the
nature of sources and nutrient inputs in the two types of flowing
waters, N inputs to stream sources may be more directly detectable and
effectively controllable than those to springs drawing from subsurface
inputs, arguing for a different set of analytical tools at least (and
possibly nutrient criteria) for the two types of systems.

By one criterion, namely the presence of nuisance macroalgae, a nutrient
criterion of xx is suggested. By a second criterion, “excessive”
algal growth, another is suggested.  On both lines of evidence, a TN
criterion of 0.35 mg/L is proposed to represent a threshold with a
significant margin of safety.  Based on the data presented though, both
criteria seem to suggest a threshold closer to 0.23 would more
effectively provide that margin.

Notes

p. 49: wouldn’t the conservative target concentration be 401 instead
of 420 µgN/L?

The inconsistent use of units (µg and mg) is distracting. 
Concentrations should all be expressed in one unit, preferably µg.

p. 51: why 0.23 is unnecessarily protective is not made clear- what is
the support for this statement?  The authors make a point to state that
the microcentrifuge experiment that produced the 0.23 result is likely
the more accurate of the two experiments- why disavow that statement
now?  Further if the range between 0.23 and 0.44 mg/L “clearly
demonstrate” an imbalance of aquatic flora, why is 0.23
“unnecessarily protective?”  

Figure 4-12: While there is clearly a step around 0.44 (consisting of
only two points), there is equally clearly an increase beyond about 0.24
mg/L.  Why disregard that?  It appears that the lower threshold is being
set at too high a concentration. There is a school of thought that the
first evidence of alteration of the ecosystem in response to a stressor
such as nutrients is already beyond the important threshold- that the
historic buffering and homeostatic properties of the system have been
exceeded and the system is in motion, possibly toward a change in state.
 At the least, autocatalytic properties may begin to create feedbacks
that amplify the negative effects of stressors (Odum, 2007).  



Chapter 5. Please comment on the regionalization of Florida’s streams
in support of the nutrient benchmark approach.

Reviewer 1	This could be the strongest section of the document and I
wish the approach was broadened to each section. I cannot overemphasize
the importance of the variable geology in Florida and its strong impact
on the water chemistry of aquatic systems. Griffith et al (1997) defined
47 Lake Regions in Florida based on water chemistry, hydrology and
physiography of over 1,000 Florida lakes. This may be a lot of regions
and some of them may be able to be combined but collapsing them to only
five nutrient regions dilutes some of the complicated geology present in
Florida. I think combining all available data in Florida and refining
Griffith et al. (1997) regions would be the best approach for setting
nutrient regions for the state of Florida.

Griffith, G. E., D. E. Canfield Jr., C. A. Horsburgh, and J. M.
Omernick. 1997. Lake regions of Florida. EPA/R-97/127 U.S. Environmental
Protection Agency, Corvalis, OR.

Reviewer 2	Chapter 5

Clearly there is strong regionalization in terms of TP for Florida’s
streams; it is well known that phosphate mining is a major industry and
that P releases from the Gulf Coast rivers impact algae blooms in
western Florida Bay.  The authors did a logical job of working through
several techniques to obtain their final product.  Kriging is well known
as a grouping technique of choice in these situations.  I have a few
comments for consideration:

1. Much is made of the “nutrient reference streams/sites” but these
are not really explained until later in the report (Chapter 7).  At
least some information on their criteria as presumably unimpacted
systems should be up front in Chapter 5.

2. It is not explained why the Everglades region is not addressed in
this chapter.

3. The TN stream nutrient regions deserve their own map, like the one
for TP on Figure 5-7.

Reviewer 3	Robertson et al. (2006) demonstrated that collapsing
ecoregions (and drawing ad hoc boundaries) into aggregated nutrient
regions was an effective way of setting regional expectations for
ambient nutrient concentrations.  The methods presented in Chapter 5
appear rigorous, and accomplished the objective of stratifying
meaningful population units.

Reviewer 4	Establishing a spatial framework is an important practical
component of criteria development.  EPA provided a good starting point
with the elaboration of the ecoregion approach.  States are well served
by the initial framework, but can use local knowledge to improve the
scheme.  Florida has taken a logical approach by adapting the ecoregion
boundaries to accommodate those geological considerations that affect
phosphorus concentrations in surface waters.

The number of regions seems manageable from a regulatory perspective
and, although the scheme was developed to recognize geologic factors
controlling phosphorus concentrations, it does not seem to be in
conflict with nitrogen distributions.  The result seems to be a suitable
companion to the nutrient benchmark approach.

Reviewer 5	Chapter 5 discusses a scheme for the regionalization of the
state in order to develop a spatial framework for better understanding
of nutrient effects on inland waters and in recognition that differences
in landscape or seascape settings can have an overarching effect on
nutrient dose-response relationships in the environment.  The narrative
describes the procedure undertaken for determining, verifying, refining
and classifying regions for determination of stream nutrient criteria
for P and N.  Regionalizations were established based first on the EPA
ecoregions, then refined based on taxonomy of biological distributions,
then finally by nutrient levels.  It is not clearly explained how the
ecoregion approach and the biological taxonomy approach were used in
support of the nutrient approach, if at all.  Biology may not reflect
underlying nutrient regime or biogeochemistry.

There is much description, a disproportionate amount of space in fact,
of the means by which a part of the western peninsula is split two
subregions: Peace River and Bone Valley, based on TP concentration.  The
discussion of regionalization based on TN is barely mentioned, although
TN is more heterogeneous and therefore more difficult to classify. 
Further, there is no discussion how the southern portion of Florida
would be subdivided into, perhaps, Everglades, rocky glades, marl
prairie, coastal plain and riverine environments, creating equally
heterogeneous landscapes.  As demonstrated in the application of
nutrient regionalization in later chapters, the step is a productive one
and does assist in the evaluation and quantification of effects of
nutrients on biological responses; further detail on state-wide
regionalization is required.

Notes

P 61: The statement that bioregions were not sufficiently homogeneous
with regard to nutrient concentration has no backup. What were the
criteria for homogeneity and why? The alternatives based on a slew of
variables are not defined well nor is a justification provided for their
use.

How will the disparate regionalizations between N and P driven be
reconciled to each other and to the biogeography?  Or will they?  Does
the biology reflect the nutrient regionalization(s)? TN or TP or
whatever is limiting?  This would have been an interesting analysis-
there is more to regionalization than classifying based on nutrient
levels as surely the authors understand.  

How will the regionalization be connected to the criteria formation? 
This is not described in this section although it may be referenced in
other parts of the document.  It would be useful to understand how the
regionalization will be incorporated into the overall process of
criteria development here as part of the chapter describing the
classification framework itself.

Chapter 6. Please evaluate the stressor-response approaches for numeric
nutrient criteria development.

Please provide any additional approaches or methods that could use the
available data to derive numeric nutrient criteria.

Can the stressor-response approach, as described, dependably correlate
in-stream nutrient concentration with measurement of biological
condition at the same place and time?  What else may need to be
considered?

What issues need to be considered when stressor-response based criteria
are applied across the regionalization scheme presented in Chapter 5 (or
other an alternative scheme)? 

Reviewer 1	As mentioned above, and the conclusions of Section 6, other
factors (pH, color, canopy cover, stream velocity, water color and
aquatic plant abundance) appear to be driving the biological responses
of rivers more than nutrient concentrations. However, fish and aquatic
bird’s abundances in a wide range of rivers are directly related to
nutrient concentrations, so the basic productivity of rivers is still
functioning similar to lakes. Therefore, I feel the best approach to
evaluate biological stressor-response is to make sure to separate
streams into ecoregions based on geology, hydrology and physiography
then develop the best models relating biological measures to physical
factors of the stream and compare the residuals of these models to
nutrient concentrations within ecoregions.

Hoyer, M. V., and D. E. Canfield Jr. 1991. A phosphorus-fish standing
crop relationship for streams. Lake and Reservoir Management 7: 25-32.

Hoyer, M. V., S. K. Notestein, T. K. Frazer, and D. E. Canfield, Jr.
2006. Bird density, biomass, and species richness on five Florida
coastal rivers, with comparisons to Florida lakes. Hydrobiologia. 567:
5-18.

Reviewer 2	Chapter 6

As I indicated earlier, using benthic taxa to assess nutrient impacts
can at best provide partial guidance.  However, I agree that there is a
general trend of negative responses to increasing TN and ammonium.  To
strengthen this point the authors need to add to Figure 6-1 the r2 and p
values, so the reader can see the individual statistics associated with
the relationships (and see if they are indeed statistically
significant).  

I also have a question regarding the first paragraph on page 73 – why
would % very tolerant taxa show an adverse response to increasing N?

To the point of alternative techniques, there are a couple of approaches
that I have found useful.  In addition to searching for nutrient
criteria, it is useful to determine nuisance levels of periphyton
biomass that can afflict streams.  Dodds et al. (1998) analyzed data
sets from several hundred streams for nutrients, benthic chlorophyll and
planktonic chlorophyll using a frequency distribution approach and
dividing the sets into lower, middle and upper thirds, with the dividing
line of the lower two thirds the border between oligotrophic and
eutrophic streams, and the middle-upper dividing line being the border
between mesotrophic and eutrophic streams.  Table 1 of that paper
provides the suggested boundaries for those parameters.  A follow-up
paper (Dodds et al. 2002) utilized worldwide data sets and both
parametric and non-parametric statistics to analyze stream data and
found significant relationships between N and P and mean and maximum
chlorophyll a (benthic).  They also provided breakpoints (Table 6 of
that paper).  While these papers utilized data from northern streams in
and outside the US, Florida appears to have a significantly large data
set to utilize these techniques (both as a whole and by individual
ecoregion) as a complement to the already-utilized approaches they have
taken.  By the way, those authors (as well as others such as Eugene
Welch) consider benthic chlorophyll a concentrations of 100-150 mg/m2 to
be nuisance levels of benthic algal biomass.

Dodds, W.K., J.R. Jones and E.B. Welch. 1998. Suggested classification
of stream trophic state: distributions of temperate stream types by
chlorophyll, total nitrogen, and phosphorus. Water Research
32:1455-1462.

Dodds, W.K., V.H. Smith and K. Lohman. 2002. Nitrogen and phosphorus
relationships to benthic algal biomass in temperate streams. Canadian
Journal of Fisheries and Aquatic Sciences 59:865-874.

Another approach that would certainly provide an additional basis for
determining stream nutrient criteria would be nutrient addition
bioassays conducted on stream water for various ecoregions.  We have
done so in North Carolina Coastal Plain streams that are heavily
colored, and have utilized both pristine and impacted systems (Mallin et
al. 2001; 2004).  We utilized response variables including chlorophyll a
(planktonic), bacteria counts, ATP, and biochemical oxygen demand (BOD5
and BOD20).  Our results found breakpoints for nitrogen (either nitrate,
ammonium, or urea) between 200 and 500 ppb, above which there was
significant phytoplankton growth and consequent significant BOD
increases.  We also found no significant algal response to phosphorus
inputs, but significant bacteria and ATP responses to P loading above
500 ppb (and significant BOD responses) – see Mallin et al. (2001;
2002 and 2004).

Thus, the response to N by phytoplankton chlorophyll a seen by the
Florida DEP statistical analysis in the more colored streams is not all
surprising to me.  It is important to note that phytoplankton can
utilize organic N as well as inorganic N; organic N (such as urea) is
characteristic of waste from cattle feedlots and grazing areas, which
are abundant in Florida.

An additional point is that there are heterotrophic responses to
nutrient loading that are not usually measured or even considered.  Our
streamwater bioassays found significant stimulation of BOD via the
indirect route of N stimulation of algal loom formation and subsequent
decay, as the Florida DEP also notes as a problem in this document. 
However, the direct impact of nutrients (especially phosphorus) on
bacterial growth and BOD increases needs consideration as well, as seen
in the publications listed below.

Mallin, M.A., L.B. Cahoon, D.C. Parsons and S.H. Ensign. 2001. Effect of
nitrogen and phosphorus loading on plankton in Coastal Plain blackwater
streams. Journal of Freshwater Ecology 16:455-466.

Mallin, M.A., L.B. Cahoon, M.R. McIver and S.H. Ensign. 2002. Seeking
science-based nutrient standards for coastal blackwater stream systems.
Report No. 341. Water Resources Research Institute of the University of
North Carolina, Raleigh, N.C.

Mallin, M.A., M.R. McIver, S.H. Ensign and L.B. Cahoon. 2004.
Photosynthetic and heterotrophic impacts of nutrient loading to
blackwater streams. Ecological Applications 14:823-838.

An additional paper that delves more into the heterotrophy aspect of
eutrophication is below:

Dodds, W.K. 2006. Eutrophication and trophic state in rivers and
streams. Limnology and Oceanography 51:671-680.

There was a question as to whether time and place issues essentially
throw a wrench in the nutrient loading and biological response paradigm.
 One must remember that streams are not totally “canopied” unless
they are in parks and reserves.  In reality streams wend their way
through forests (canopied), then open farm fields (full sunlight), and
road crossings (full sunlight).  Thus in a nutrient-impacted stream
there will be opportunities to form algal blooms even if much of it is
in a forest.  As to stream discharge, while elevated streamflow will
certainly suppress bloom formation, in Florida the slope of the land is
extremely gentle in most areas, and elevated flows will be seen only
after rain events.  Thus, there should be plenty of opportunities for
algal blooms to form in nutrient-enriched streams.  In North Carolina we
have documented very dense algal blooms in blackwater streams downstream
from nutrient sources, including streams that spend time in forested
areas (Mallin et al. 2004).

Reviewer 3	a)	Please provide any additional approaches or methods that
could use the available data to derive numeric nutrient criteria.

The regressions involving the SCI and SDI (and/or their component
metrics) appear confounded by multicollinearity.  For the purposes of
examining effects-based relationships, NOx and ammonia nitrogen should
not be combined into a single variable, as the two are measuring
different aspects of the pollution gradient.  Ammonia nitrogen is a
marker for organic enrichment and toxicity, whereas nitrate-nitrite
nitrogen is a good indicator of nutrient enrichment from properly
treated municipal effluent and agricultural pollution.  NOx should be
included in the regressions in lieu of other expressions of
cross-correlated nitrogen given the work presented in Chapter 4.  Change
point analysis along a NOx gradient should be done using the residual
variation in biological metrics following regression on conductivity,
color and ammonia nitrogen.  The wide confidence intervals resulting
from bootstrapping during change point analysis should not be construed
as a limitation.  Rather, the median change points found for each metric
should be considered in aggregate – do they generally fall within the
same ballpark?  If so, that shows a strong signal that can be used as a
line of evidence supporting criteria.          

Quantile regression applied to wedge-shaped data sets should be used to
identify a subset of data that is responding to a gradient of interest. 
The two data sets can then be compared to help understand the secondary
gradients, and to perform direct gradient analysis on the data that are
relatively less influenced by the secondary gradient(s).

Can the stressor-response approach, as described, dependably correlate
in-stream nutrient concentration with measurement of biological
condition at the same place and time?  What else may need to be
considered?

The report is vague with respect to which data are used for the SCI
regressions.  The sample size of ~ 75 appears rather small.  Were these
data collected prospectively along a nutrient gradient, randomly
selected, or the result of a data cull to remove spurious sites from
existing data?  As described, the results add another line of evidence
to the whole body of work.      

c)	What issues need to be considered when stressor-response based
criteria are applied across the regionalization scheme presented in
Chapter 5 (or other an alternative scheme)? 

The two issues are sample size within strata, and collecting data along
a defined gradient within strata to properly derive criteria specific to
a region.  Clearly, nitrogen criteria derived for a nitrogen limited
region cannot be applied to a region primarily phosphorus limited as the
criteria would be over-protective.  However, across regions, states, and
in disparate parts of the country, identified nutrient thresholds
generally fall in the same ballpark (~ 0.02-0.07 mg/l TP, and ~ 0.5-1.0
mg/l N), such that criteria derived for one region can be reasonably
applied to another if the two are relatively homogeneous with respect to
parent geology. 

Reviewer 4	There is a refreshingly candid recognition that conditions in
natural streams make it difficult to elucidate dose-response
relationships in as clean a manner as would be possible in laboratory
situations.  Under controlled conditions, the relationships between
nutrient concentrations and biological responses have been studied
exhaustively and can be defined quantitatively.  Certainly, the same
processes are in play in the natural environment, but the responses to
stressors are likely to be more variable because other factors are not
controlled.  The increased variability does not mean that the
fundamental relationships between nutrients and biological responses are
no longer operative, however.

Nutrient supplies set an upper bound on algal growth, for example, but
the observed amount of chlorophyll often will not reach the potential
defined by the nutrient supply.  Nutrients control growth, but not loss
factors (e.g., grazing or scouring).  In addition, growth may be
suppressed by toxics, like heavy metals.  This scenario, in which
nutrients set only the potential for growth, is fully consistent with
the wedge-shaped relationships that appear so often in the report.   At
high nutrient concentrations, algal response may be high or low
depending on the role that other factors, such as flow, may play in
regulating algal biomass.  Similarly, the SCI may be high or low at low
nutrient concentrations because biological condition can be suppressed
by other factors.  At the same time, however, when nutrients are high,
the SCI is unlikely to be high.

Deriving dose-response relationships from field data is a worthy goal,
despite the difficulties, because they would facilitate identification
of thresholds for nutrient impairment.  Do suitable applications exist
for dealing with the wedge-shaped relationships?  It would be
disappointing if none exists, although it would not preclude drawing on
these data for confirmation of thresholds derived by other methods.

Reviewer 5	Relationships between nutrient concentrations and various
biotic parameters in the ecosystem were developed in order to assess if
quantifiable dose-response algorithms could be used in establishing
nutrient criteria.  The analysis of dose-response characteristics of
nutrient enrichment on macroinvertebrates, periphyton abundance and
chlorophyll a concentrations yielded ambiguous results.  While all
analyses showed the adverse impact of nutrients on ecosystem components,
the high degree of variability in the relationships rendered the
establishment of well-constrained quantitative relationships untenable.

It is likely that the utility of applying analysis of
macro-invertebrates in relation to nutrients to determine dose-response
curves that can be used to establish quantitative criteria is of dubious
value.  While impairment of a stream will undoubtedly redound through
the trophic chain, looking at higher trophic levels as a first cut and
as a means to determine nutrient criteria is not likely to be effective
beyond its utility as a screening tool.  The instantaneous effect of
nutrient enrichment is felt in the microbial and plankton communities,
whose impairment then links to other ecosystem components such as
invertebrates.  THAT component will suffer directly through a
potentially long trophic chain that sometimes must pass through the
detrital stage, or via negative impacts on environmental conditions such
as reduced DO and changes in transparency and prey composition.  



Chapter 7. Please evaluate the Nutrient Benchmark Distributional
Approach.

What, if any, additional measures should be considered in the selection
of benchmark sites to ensure the selection of minimally disturbed sites?

Is there anything that can address or reduce the natural variability in
the benchmark distribution to enhance its representativeness for each
Florida Bioregion?

In Florida many surface waters have naturally low dissolved oxygen
levels.  Are there any issues to consider when including waters that
experience naturally low levels of DO as reference streams for
nutrients?

What issues should be considered if the distribution of a reference
condition parameter (e.g., total N) substantially overlaps with the
distribution of that parameter in all waters of a bioregion?  

Reviewer 1	Again, I like the idea of separating the rivers into groups
using an Ecoregion concept. However, I have always had a problem with
setting up reference benchmarks within regions due to the large amount
of within and among variance associated with any biological and/or water
chemistry measurements. As pointed out above, much of this variance is
due to physical factors (color, canopy cover, stream velocity, slope,
mean depth etc..) that have nothing to do with anthropogenic impacts.
The selection process Listed in Chapter 7 could cause the elimination of
lakes from the data base that are truly not impacted and it could also
cause the selection of rivers that are unique and not really
representative of the individual Ecoregion.

For this section, I would recommend an approach of modeling what factors
account for the most variance in biological measures and nutrients
concentrations within individual Ecoregions and the plot the residual
against measurements of potential of anthropogenic impacts (LDI). Any
relations or break points between the residuals and measures of
potential anthropogenic impacts may help point to a nutrient standard.

Reviewer 2	Chapter 7

Please define WBID up front in this chapter

I am extremely impressed by the way DEP has gone about choosing
reference sites for nutrient criteria.  It was a multifaceted approach
utilizing LDI, 303(d) list, the proposed N criteria, aerial photos, site
visits, statistical analysis, and expert opinion.  I can think of
nothing that would improve upon their selection process.

As to natural variability in a benchmark stream, one aspect that may not
have been considered is tidal action in freshwater streams.  In the
Coastal Plain, streams even 20 miles inland can move with the tides, so
sampling at a given tidal stage (say mid-tide outgoing) can reduce any
variability from that source.

The question about naturally low DO values is interesting.  The Coastal
Plain in NC, SC and GA has numerous blackwater streams that have
naturally low pH and low DO.  This is due to their draining extensive
swampy areas where the water is in contact with organic sediments and
bacterial action, which draws down the DO; thus the water entering such
streams already has low DO.  Additionally, these streams receive copious
inputs of DOM from the watershed, which is essentially food for
bacteria, which then impart a BOD on the water.  In unimpacted waters
there are healthy animal communities that are adapted/evolved to
situations of pH of 3-4, and DO of 4.0 (LMB fishing can be quite good!).
 Thus, these streams should not be excluded from use as benchmarks; in
fact, their inorganic nutrient levels are usually quite low (the
majority of their nutrient load is in the refractory organic form).

Reviewer 3	What, if any, additional measures should be considered in the
selection of benchmark sites to ensure the selection of minimally
disturbed sites?

The methods used to identify, screen, cull and eventually select
benchmark sites were rigorous and sound.

b)	Is there anything that can address or reduce the natural variability
in the benchmark distribution to enhance its representativeness for each
Florida Bioregion?

The real question is what steps can be taken to understand the
variability inherent in natural systems.  With respect to nutrients,
some of the benchmark sites have relatively high nutrient concentrations
co-occurring with good biological condition.  Yes, that is in keeping
with physically intact systems having a higher assimilative capacity
than degraded systems, or systems overly taxed.  A careful examination
of the sites where good biology and high nutrient levels intersect
should be done with an eye toward identifying mitigating factors like
canopy cover, local stream gradient (i.e., slope), substrate
composition, the form of nutrient resulting in the high concentration
(e.g., wetland-derived nitrogen vs. nitrate from groundwater), etc. 
That information would then aid in interpreting data from all sites.

c)	In Florida many surface waters have naturally low dissolved oxygen
levels.  Are there any issues to consider when including waters that
experience naturally low levels of DO as reference streams for
nutrients?

The daily range of DO is the key to understanding nutrient enrichment. 
For streams with naturally low DO (blackwater streams), the DO per se
does not define the reference condition.  In those cases, high DO
concentrations during the day may signal enrichment.       

d)	What issues should be considered if the distribution of a reference
condition parameter (e.g., total N) substantially overlaps with the
distribution of that parameter in all waters of a bioregion?   

I simply cannot emphasize strongly enough the need to examine the
distributions of nutrient concentrations within categorical ranges of
the SCI (or some other biological measure/metric) for benchmark sites
and for all sites by region.

Reviewer 4	Considerable effort has been applied to the task of screening
benchmark sites.  The steps seem reasonable in general.  Exclusion of
sites on the basis of a nitrate threshold seems like a clever basis for
identifying far-field effects transmitted via groundwater, although
setting the threshold at 0.35 mg/L presumes adoption of the proposed
criterion.  The exclusion of sites with SCI <40 reflects the assumption,
already mentioned above, that impaired equals “not reference.” 
Perhaps this is a trivial matter with regard to the number of sites
excluded.

I have reservations about equating “representativeness” with a
minimization of variability in the benchmark set.  It seems to suggest
that an “ideal” number exists for reference, rather than a range. 
Very strict criteria have been used to select those sites that are
minimally disturbed.  In addition, sites were excluded when nutrient
concentrations were “outliers.”  While it is true that these
procedures reduce variability, it is not a foregone conclusion that they
improve the representativeness of the remaining sites.

Exclusion of nutrient outliers is troubling because it seems contrary to
the following statement on page 85:

One disadvantage of using the benchmark approach is that it does not
identify the specific nutrient levels at which biological impairment
occurs.  For this reason, it cannot be concluded a priori that adverse
effects on aquatic life actually occur at concentrations above these
values.

The rigorousness of the benchmark screening process leads me to believe
that all nutrient concentrations observed at the benchmark are fully
protective of the use.  Consequently, it causes me to wonder why the
impairment should not be set to the concentration that represents an
outlier with respect to the benchmark distribution.

The final question on the Technical Charge is not a topic addressed
explicitly in the chapter, but it is an interesting issue (if I
understand it correctly).  When there is substantial overlap between
distributions of TN (for example) at reference sites and at all sites in
the bioregion, there are two obvious interpretations – most sites are
close to reference conditions, or undisturbed has given way largely to
minimally disturbed.  In either case, it undermines the statistical
distribution approach to characterization of reference conditions based
on a percentile of all sites.  In the present situation, given the
rigorousness of the process to select nutrient benchmark sites, it would
be logical to conclude that most sites in the region were in attainment.

Reviewer 5	The crux of the document for establishment of nutrient
criteria is embedded in this chapter, where DEP advocates an explicit
recommendation to utilize the benchmark approach in lieu of the
dose-response approach in the establishment of specific nutrient
criteria for Florida’s inland waters.  The dose-response analysis,
described in Chapter 6, was insufficiently robust to develop a
quantitative nutrient criterion for either N or P.  Consequently, an
analysis of the site distributional approach was undertaken.  An
elaborate screening protocol is described for winnowing available sites
to benchmark pool.  Though some assumptions can be questioned, the
overall impression is of a conservative approach, ensuring a valid list
of candidate datasets with which to establish minimally impacted
benchmark sites.  

The lowest SCI scores for the benchmark sites with respect to N and P
concentrations show no dose-response relationship between nutrients and
SCI.  Florida DEP considered using 75, 90, 95 and 99 percentiles of
benchmarks, with adjustment for nutrient region, for use in setting
numeric nutrient criteria for TN and TP, settling on the 90th
percentile.  The strengths of this recommendation include: General
applicability, standardization, speed.  Weaknesses include: Lack of
mechanistic understanding of processes underlying the patterns, Lack of
predictability, lack of inclusion of far field effects downstream as
(described in analysis of Chapter 8).  The proposed N and P criteria
selected are supported by a significant amount of underlying analysis
and are credible options as a first-cut.  Continued assessment of the
effectiveness of these criteria through a variety of standardized
performance measures is critical in determining their validity.  These
performance measures are not considered in this document and it is a
recommendation that a rigorous protocol of evaluation be implemented.



Chapter 8. Please evaluate the reliability of the Nutrient Longitudinal
Study to protect the aquatic life use downstream and whether this
approach can be applied to other estuaries or lakes in Florida?  

Reviewer 1	This nutrient longitudinal study focuses on only nutrient
concentrations and not nutrient loads. Upstream concentrations have
little to due with the whole load to Deadman Bay or Waccassassa Bay,
which could potentially change chlorophyll and thus water clarity in the
Bays potentially impacting the important grasses. Adding the data that
has already been collected to discharge values and using the
Vollenweider approach described in Section 14 would allow the prediction
of the change in the nutrient concentration of the river needed to a
significant change in the bays. As mentioned above, this approach has
been extremely successful in lakes (Bachmann and Canfield 1981) and
shows great promise in Estuaries (Meeuwig et al. 1998; Meeuwig 1999;
Meeuwig et al. 2000).

Canfield, Jr., D. E. and Bachmann, R.W. 1981. Prediction of total
phosphorus concentrations, chlorophyll a , and Secchi depths in natural
and artificial lakes. Can. J. Fish. Aquat. Sci. 38:414-423.

Meeuwig, J. J., J. B. Rasmussen, and R. H. Peters. 1998. Turbid waters
and clarifying mussels: their moderation of empirical chl:nutrient
relations in estuaries in Prince Edward Island, Canada. Mar. Ecol. Prog.
Ser. 171: 139-150.

Meeuwig, J. J. 1999. Predicting coastal eutrophication from land-use: an
empirical approach to small non-stratified estuaries. Mar. Ecol. Prog.
Ser. 176: 231-241.

Meeuwig, J. J., P. Kauppila, and H. Pitkanen. 2000. Predicting coastal
eutrophication in the Baltic: a limnological approach. Can. J. Fish.
Aquat. Sci. 57: 844-855.

Reviewer 2	Chapter 8

Florida DEP conducted the longitudinal study in naturally phosphate-rich
river basins to assess whether nutrient concentrations representative of
the upper portion of the benchmark site distribution were protective of
the designated uses of the downstream (estuarine) waters.  Samplings
were performed for a variety of standard physical, chemical, and
biological parameters, and they were performed at the sites during the
same day, which is an appropriate approach to reduce intraday
variability.  Whether it was winter or summer, the results showed no
algal blooms, low to moderate nitrate concentrations, low TSS, low
turbidity, normal Secchi  depths, and with a couple of exceptions,
generally good dissolved oxygen levels.  As noted the benthic
communities were in good shape, and periphyton coverage was no problem. 
TN was either lower than or average compared with moderate sized rivers
elsewhere.  Historical trend lines did not appear to show any concerns,
except for an apparent increase in TN in the Waccasassa River over time
– which did not translate into increased chlorophyll a.  The estuarine
chlorophyll a concentrations were not at levels that would be a problem
to estuaries in general, and the seagrass beds, while the seagrass
coverage appeared to be robust.  I would concur with the conclusion that
the estuaries are not being adversely affected by the (generally)
naturally high TP concentrations in the headwaters.  This should be an
appropriate tool to help assess baseline concentrations and affects in
other estuaries and lakes in Florida.

Reviewer 3	It is important to describe relatively intact systems as
basis for comparison with those more disturbed, and from that
standpoint, this chapter has value.  As a standalone, however, this
chapter offers little insight.  It would be really helpful to meld the
results of Chapter 8 with those of Chapter 14, including doing a model
of nitrogen loading for the Waccasassa and Steinhatchee estuaries for
comparison (and validation) to the others presented in Chapter 14. 
Similarly, a description of the condition of seagrass beds in the
estuaries listed in Table 10 in Chapter 14 should be compared to those
for the Waccasassa and Steinhatchee for context.

Reviewer 4	The longitudinal study seeks to demonstrate “proof of
concept” by showing no adverse effects in the estuary when tributary
streams meet benchmark nutrient conditions.  Although the longitudinal
studies showed that the two rivers and two receiving estuaries are
minimally disturbed, I would hesitate to conclude that they constitute
proof of concept.  Criteria for each water-body type would likely be
based on concentration thresholds, but the concentration in the stream
alone does not translate directly into the concentration expected in the
estuary (or lake).  Mass transport and the hydraulic properties of the
estuary (or lake) also would be needed to make such a prediction.

It does not seem practical on a statewide basis to derive stream
criteria that are uniformly protective of uses in every estuary.  Large
differences among estuaries in hydraulic properties, for example, would
preclude simple application of the same stream criteria everywhere
(least common denominator?).  Conversely, there is no reason to assume
that all streams are in attainment just because they terminate in an
estuary that is in attainment.

Criteria should be set separately for streams and estuaries (or lakes)
to be protective of uses in each system.  The TMDL process is
well-suited to the task of reconciling situations where the stream is in
attainment, but the estuary downstream is not.  There is no reason to
try to preempt that process.

Reviewer 5	Downstream reaches (streams and estuaries) are the ultimate
recipients of watershed nutrient inputs and are usually the most
susceptible waterbodies to loadings, despite the often attenuated nature
of the upstream loads of constituents.  They are also most complicated
due to geochemical interactions, the heterogeneous nature of physical
forcings, differences in water residence times, nutrient turnover,
nutrient mass balance characteristics, differences in biota and
biological processing.  This chapter describes what purports to be a
pilot study of the nature of upstream nutrient effects on downstream
estuaries.  The window on the subject is somewhat limited here as the
two estuaries selected are in the same geographic area, the same
bioregion, the same nutrient region, subjected to similar loadings, low
levels of nutrient impact (meaning little elucidation of nutrient
processing), generally similar processes and biology, and a limited
timeframe of analysis (two samplings in 2008 and 2009).  In fact, the
study is attempting to show that the presence of unspecified seagrass
and existence of an unspecified fishery of indeterminate province proves
that the upstream benchmarks are protective of downstream waters.  

The analysis is minimal in the extreme.  The supporting evidence of an
unimpacted fishery, its description or its history is non-existent. 
There is no exposition of the type of seagrass in the downstream
receiving waters, the SAV density, condition, its antecedent history, or
documentation of any change over time.  A fruitful dataset to mine for
additional analysis along these lines would be that of the Tampa Bay
seagrass restoration program (e.g. Greening and Janicki, 2006).  

Still, there is the fact that SAV found downstream of a benchmark
concentration is encouraging.  Such assessments might be considered more
a screening tool- if there is impairment in the downstream estuary,
upstream protective action must be taken; if there is no apparent
impairment, further assessment may still be required to truly comprehend
the effect of upstream impacts on downstream communities and ecosystems.

The conclusion that the benchmark is protective appears accurate as far
as it goes, but there are many factors, assumptions and questions that
remain untested by this pilot study.  The fact the chlorophyll a
increased at every station in the downstream direction is noteworthy. 
The high concentrations of organic compounds observed and hypothesized
as being derived from exports from surrounding marshes gives a sign that
nutrient transformations in the watershed are a potential source and
nutrient spiraling may generate unanticipated loads of transformed
species (organic N such as humics, urea, ammonium) with varying
bioavailability that can and will cascade through the ecosystem in
unpredictable ways.  

Marine end member nutrient subsidies provide additional inputs to
estuaries that also must be accounted when determining whether
impairment downstream might result from inputs upstream.  A nutrient
that may be limiting in a freshwater spring may not be in the downstream
receiving water body.  General understanding of estuarine circulation
and coastal water mass transport, if not outright circulation modeling,
will be required in many cases for true assessment of nutrient fate and
transport and downstream effects.  Furthermore, as with interpretation
of any of these studies, the non-linear response of downstream systems
to upstream inputs may result in rapid crossing of thresholds or tipping
points earlier in the nutrient loading trajectory than expected.   

Caution must be applied in interpreting the results of such
underspecified analyses in two of the most unimpacted streams in the
state.  Application of this analysis to lakes bears the same caveats. 
Each distinctive water body (stream, lake, estuary, spring, river)
should be assessed in its own context for response to nutrient criteria.



Chapter 9. Please evaluate the methods for developing chlorophyll a
numeric nutrient criteria for lakes.

Please provide any additional considerations which might provide support
for numeric nutrient criteria development.

Reviewer 1	The trophic state scale as originally defined is a gradient
without a real break point so I have always had difficulty with one
individual chlorophyll number as a standard. The standards that are
being proposed are reasonable and I appreciate the groupings by color
and specific conductance. As mentioned above, color can become a problem
with this proposed chlorophyll standards because with one rainfall color
can change from low to high with one rainfall shifting the chlorophyll
standard from 10 to 20 µg/l which is a huge difference. Using specific
conductance to get at differences based on geology is conceptually a
good idea but there are many lakes in Florida near the coast that have
high specific conductance because of saltwater interferences (e.g.,
Walton County Dune Lakes).

The proposed chlorophyll standards also do not take into account the
abundance of aquatic plants within individual lakes. Lake with abundant
aquatic macrophytes will maintain low chlorophyll levels. However, if
the macrophytes are controlled or die naturally that same lake will have
much higher chlorophyll levels. This needs to be accounted for in any
type of chlorophyll standard.

Getting back to the original purpose of setting standards, keeping
waters fishable and swimmable for the future, almost all of Florida’s
exceptional fishing lakes would have chlorophyll concentrations
exceeding 20µg/L making them impaired on paper. This is one reason why
I mentioned in my general comments that Florida Fish and Wildlife
Conservation Commission needs to be involved with setting nutrient
standards.

The approach I like the most that was reviewed in the document is the
one taken by Alabama of maintaining the existing condition. Several
studies (Knowlton et al 1984; Terrell et al. 2000) have calculated
normal background variance of chlorophyll in lakes and they are all
quite similar. Taking the abundant long-term data that is now available
and defining normal background variance would be a good way to setting a
chlorophyll standard. This would require constant monitoring of lakes
but any lake that has an annual mean exceeding the normal background
variance could be defined as not meeting use. The monitoring would be
expensive with state biologists but volunteers in Florida LAKEWATCH have
cost effectively been measuring chlorophylls for over 20 years so it
could be accomplished on a large scale.

Knowlton, M. F., M. V. Hoyer, and J. R. Jones. 1984. Sources of
variability in phosphorus and chlorophyll and their effects on use of
lake survey data. Water Resources Bulletin 20: 397-407.

Terrell, J. B., D. L. Watson, M. V. Hoyer, M. S. Allen, and D. E.
Canfield Jr. 2000. Temporal water chemistry trends (1967-1997) for a
population of Florida waterbodies. Lake and Reservoir Management 16:
177-194.

Reviewer 2	Chapter 9

I appreciate the use of the variety of approaches DEP has taken in
devising the proposed lake chlorophyll a thresholds.  They have adapted
the TSI to Florida’s rather unique situation of lakes to properly
account for colored waters interference with Secchi depth measurements,
and incorporated rather conservative N/P ratios to devise their TSIs. 
As to the use of paleolimnological studies – I am not a fan of their
use to set numeric criteria.  While they can provide some excellent
information on past histories of large systems such as Chesapeake Bay,
and provide strong evidence for pollution episodes (metals in cores) and
species changes, I do not feel their level of historic specificity is
clear enough for setting these criteria.  I do feel that expert opinion
supports the 20 µg/L criterion, though.

As to biological responses, there is something that DEP should consider
both as a potential reason for chlorophyll a use as a response variable,
and impacts of increasing chlorophyll a concentrations.  This would be
the impact of chlorophyll a on biochemical oxygen demand (BOD5 and
longer-term, perhaps BOD20).  In a selection of Coastal Plain waters
(urban lakes, urban streams, estuarine tidal creeks, and a large river)
Mallin et al. (2006) found strong correlations between chlorophyll a
concentrations and BOD levels.  I assume that Florida DEP has a rather
robust BOD data set and perhaps they could produce regression equations
for BOD as a response variable to chlorophyll a (or, for that matter,
examine the use of BOD as a response variable to N and P
concentrations).  BOD is more closely linked to hypoxia than is
chlorophyll.

Mallin, M.A., V.L. Johnson, S.H. Ensign and T.A. MacPherson. 2006.
Factors contributing to hypoxia in rivers, lakes and streams. Limnology
and Oceanography 51:690-701.

Reviewer 3	Chapter 9 is essentially a heuristic exercise, but a
presentation of how the equations listed on page 126 were derived
(scatter plots, statistical output, etc.) would have been helpful. 
Presumably these are based on regressions using data from the 313 lakes
referred to on the same page.  Otherwise, the information present has
value in that it shows convergence across regions and studies, thereby
strengthening the target criterion of 20 ug/l chlorophyll a.

The section relating cyanobacteria abundance to chlorophyll a levels
offered little towards understanding effects of nutrient enrichment.  It
simply argues that maintaining existing conditions, will, in general,
prevent HABs.  A better approach would be, where data are available, to
look at trends in cyanobacteria abundance over time for selected lakes
where nutrient concentrations have increased.      

a)	Please provide any additional considerations which might provide
support for numeric nutrient criteria development.

The results from Chapter 9 should be melded with those from Chapter 10. 
Specifically, for lakes where the LVI was scored, the TSI should be
calculated, and the scores regressed against each other.  That might
show a significant change point in the LVI in relation to the TSI, thus
bolstering both indices as effects-based outcomes.

Reviewer 4	Florida has already developed a proposal for chlorophyll
criteria in lakes, and the chapter summarizes multiple lines of evidence
in support of those criteria.  The development of lake criteria benefits
from long experience is assessing lake water quality.  Past assessments
have focused largely on trophic state using a Trophic State Index (TSI).
 Carlson’s original concept for a TSI was modified by substituting
nitrogen for Secchi depth and by averaging the three component indices. 
Although Carlson did not recommend averaging component indices, it is
not without precedent.  The use of a TSI for criteria development is
practical, especially when there is a strong interest in supporting
fishery yield (to the extent that fishery productivity is related to
trophic state).

Six lines of evidence were examined in the chapter.  The conclusion is
drawn that they “converge to support the Florida IWR TAC’s original
recommendation that 20 µg/L of chlorophyll a in colored lakes is
protective of designated uses.”  All approaches are not necessarily
mutually compatible, however.

a.)  Florida is fortunate in dealing principally with natural lakes
because paleolimnological studies can be used to good advantage.  These
studies suggest that at least some Florida lakes have always been
relatively productive.  If some lakes were productive prior to human
disturbance, it argues for caution in the application of a reference
distribution approach, wherein low productivity is presumed to reflect
minimally disturbed conditions.

b.)  It is not clear what role expert opinion (i.e., what other states
have done) should play in view of the other options available, unless it
is a matter of seeking reassurance in similar conclusions.

c.)  Biological response is clearly a desirable approach, and several
states have presented ideas.  Minnesota has probably done the most in
this regard.  [Note that, the information attributed to Colorado is
mistaken because no proposals have been made yet for nutrient criteria.]
 Criteria developed for fishery support may not be uniformly compatible
with all uses in other states.

d.)  User perceptions would seem to be a good basis for defining
thresholds for aesthetics in particular, but work in Texas suggests that
“acclimation” of the observer may affect perceptions.  One of the
difficulties in considering criteria simultaneously for fishery support
(likely to be on the high side of chlorophyll) and user perception
(likely to want clearer water) is the potential for objectives to be in
conflict.  What approach would Florida take to balance fishery and
aesthetic considerations if these were found to be in conflict?

e.)  Maintenance of existing conditions can be a strong approach for
setting thresholds in situations where there has been a relatively long
history of data collection and conditions have remained relatively
stable.  Generally, this would be more applicable to site-specific
criteria development for a single lake where it might be practical to
assume that uses are being met currently.  It is difficult to see how it
would address use protection on the statewide level.

f.)  The reference approach is mentioned only briefly in the chapter. 
Supporting materials show that much effort was invested in this
approach, which also produced recommendations for nutrients and
transparency.  However, as mentioned above, development of criteria with
this approach would have to be reconciled with the paleolimnological
information.

The connection between chlorophyll and toxins produced by cyanobacteria
is examined in some detail because of the proposed magnitude of the
chlorophyll criterion.  An annual average chlorophyll concentration of
20 ug/L creates a measurable risk that one or more instantaneous values
might exceed the World Health Organization’s moderate risk threshold
of 50 ug/L.  Previous studies have shown a tendency for cyanobacteria to
become the dominant phytoplankton at higher chlorophyll levels.  Under
the circumstances, it was prudent to address the issue in the criteria
development process.  Evaluation of data from a large set of lakes
dispelled this concern.

Reviewer 5	This chapter discusses the history and background of the
derivation of trophic state indices for lakes that have been used in
Florida and several lines of evidence for the established chlorophyll a
standard for lakes.  Paleolimnological studies indicate that some lakes
in Florida have had naturally high chlorophyll a levels even prior to
human habitation, around the currently established standard threshold of
20 µg/L.  This is consistent with paleologic evidence from the region
arguing for establishment of a chlorophyll a threshold around that
range.  Certain colored lakes had even much greater concentrations
several times higher which supports the application of site specific
alternative criteria (SSAC) for lakes that warrant a deviation from a
lower criterion.  Biological responses of macroinvertebrates and
fisheries have been investigated but so far have not proved useful as
ecological attributes for specifying nutrient criteria.

Based on these and several additional lines of evidence including user
perceptions and threshold for maintaining existing conditions, DEP has
settled on a chlorophyll a threshold of 9 µg/L for clear lakes and 20
µg/L for high conductivity lakes.  These are reasonable thresholds
given the lines of evidence presented.  In the absence of sufficient
data, DEP has proposed to use the 90th percentile of reference
conditions for nutrient criteria.  This is also a reasonable alternative
but the evidence for its use is not presented in this chapter.

Notes

p. 124: Note misleading sentence structure: Florida lakes do not have
cool-water fisheries (true) but do have oxygenated hypolimnia (contrary
to the implication).  



Chapter 10. Please evaluate the methods for developing chlorophyll a
numeric nutrient criteria for lakes based on stressor-response analyses.

Please provide any additional considerations which might provide support
for numeric nutrient criteria development.

Please evaluate the proposed classification and sub-classification
approach for lakes.

Reviewer 1	I mentioned above problems with the LVI and that naturally
occurring gradient like pH and nutrient concentrations that are impacted
by geology are related to the LVI. Therefore, I am glad to see the
Section 10 suggests that nutrient relations with LVI are not robust
enough to be used here to determine nutrient standards.

If chlorophyll standards are set at 9 µg/L for clear waters and 20
µg/L for colored and highly conductivity clear lakes them the modeling
conducted in Section 10 is done well and correctly. The corresponding
nutrient standards would reflect the set chlorophyll standards.

I am concerned and not sure we have the ability to easily classify lakes
into the color and/or conductivity groups that are being recommended.
For sure, using color data Lake Okeechobee as any kind of a model for
other lakes in the state is not appropriate because there are no other
lakes in the state like Okeechobee. Using 40 PCU as the cut off for
colored and clear lakes can be a difficult boundary to really define.
For example, Grasshopper Lake in Lake County Florida would be classified
as colored for 4 years and clear for 4 years between 2000 and 2008
(Figure 1). Florida LAKEWATCH has color data on multiple lakes that show
the same kind of difficulty. Additionally, 18 coastal dune lakes in
Walton County Florida (Hoyer and Canfield 2008) show data where the
lakes would be classified as low conductivity and/or high conductivity
depending on the year data are collected (Figure 2). I feel much more
work needs to be done with long-term data to better define the lake
groups based on color and conductivity.

Figure 1. Relationship between color and date in Grasshopper Lake, Lake
County Florida between 2000 and 2008 (Florida LAKEWATCH, unpublished
data).

Figure 2. Relationship between salinity and date in Alligator Lake,
Walton County Florida between 2003 and 2008 (Florida LAKEWATCH,
unpublished data).

Hoyer M. V. and D. E. Canfield, Jr. 2008. Findings from workshops on
citizen’s concerns regarding the future management of Walton County
Coastal Dune lakes. Report to Chactawhatchee Basin Alliance and Coastal
Dune lakes Advisory Board. Florida LAKEWATCH, Department of Fisheries
and Aquatic Sciences, University of Florida/Institute of Food and
Agricultural Sciences. Gainesville, FL.

Reviewer 2	Chapter 10

Macrophyte analysis – the lack of a strong correlation between
nutrients and macrophyte coverage is not at all surprising to me.  Even
at log scale there is a remarkable amount of scatter in the plots.  As
alluded to earlier in this review (Chapter 3) invasive macrophyte
species do not need high nutrients to overtake a water body; just a lack
of natural grazers and competitors.  Additionally, as pointed out in
this chapter, emergent or otherwise rooted macrophytes draw mainly upon
soil nutrients to sustain the, which would not be reflected in the water
column nutrient concentrations.  One further note on this: on page 135,
line 5; it states that lake vegetation exhibited a significant adverse
response to nutrients.  Actually it was a positive response (though
weak).

Chlorophyll a analysis – The DEP has analyzed a robust data set to
draw its chlorophyll-nutrient responses, and I agree that yearly
averages are appropriate.  Unlike in northern states, blooms can occur
is spring and fall in Florida lakes (they can do so as well even in the
Carolinas).  The response relationships are strong and little scatter is
present on the regression graphs.  I am also pleased with the way DEP
has broken out the colored lakes by degree of color.  It is evident that
in highly colored systems there will be a dilution of the regression
strength with less available light.  That having been said, highly
colored lakes can and do support algal blooms, especially the shallow
lakes that characterize much of Florida.  While applying the criteria
for the moderately-colored lakes to the highly colored lakes may be
considered by some a bit overprotective, I support this approach, having
seen dense blooms in colored lakes in the Carolinas.

I am pleased with the use of 20 ug/L of chlorophyll a for colored lakes
and for the clear but high conductivity lakes.  However, I feel that the
9 µg/L response criterion for the clear, low conductivity lakes is too
high.  These systems are really gems that need stricter protections.  I
have been in them and, while their biotic purposes may be protected by 9
µg/L, their aesthetic ones may not be.  For a large-scale example of
what I am getting at we can look to Lake Tahoe, once one of the clearest
lakes in the world but now inexorably becoming more eutrophic due to
land development along the shoes and airborne N.  I also wish to note
that, on a worldwide scale, Wetzel (2001) Table 13-18, uses a mean of
1.7 µg/L with a range up to 4.5 µg/L for his definition of
oligotrophic lakes.  So I would favor a chlorophyll a response criteria
of 5 µg/L for clear, low conductivity Florida lakes.

Wetzel, R.G. 2001. Limnology: Lake and River Ecosystems, 3rd Ed. San
Diego, Academic Press, 1006 p.

The attempt to determine a relationship between chlorophyll a and the
percent of cyanobacteria dominance is interesting.  I surmise that it
was not particularly successful in that cyanobacteria respond strongly
to physical factors as well as nutrients – by that I mean that
cyanobacteria are adapted to the warm waters and high sunlight that
characterizes so many Florida lakes.  I would like to recommend DEP have
a look at some work out of JoAnn Burkholder’s group at North Carolina
State University, in which they found that TP and chlorophyll a were
both significant predictors of total cyanobacterial abundance (see
Touchette et al. 2007).  They also found the presence of cyanotoxins in
drinking water reservoirs, and I wonder if DEP has a program to do so,
and what nutrient factors these toxins may be related to.

Touchette, B.W., J.M. Burkholder, E.H. Hannon, J.L. Alexander, C.A.
Kinder, C. Brownie, J. James and C.H. Britton. 2007. Eutrophication and
cyanobacterial blooms in run-of-river impoundments in North Carolina,
USA. Lake and Reservoir Management 23:179-192.

Reviewer 3	a)	Please provide any additional considerations which might
provide support for numeric nutrient criteria development.

As noted for the SCI and streams, the stressor-response relationship
between the LVI and nutrients appears to offer an additional line of
evidence for identifying thresholds for setting criteria.  Again, using
the residuals of the LVI following regression on color and conductivity
in change point analysis over a gradient of TN or TP should be tried.   
 

b)	Please evaluate the proposed classification and sub-classification
approach for lakes.

A separate slopes regression model would help here.  Coding the three
color/conductivity strata  –1, 0, and 1, would allow one to directly
compare differences between the strata in unit response of chlorophyll
to nutrients, and help to evaluate the appropriateness of the
chlorophyll a targets.

Reviewer 4	The derivation of criteria by this stressor-response analysis
appears to aim at a tiered approach for nutrients.  If a lake is
“biologically healthy” and chlorophyll does not exceed the
appropriate threshold, some latitude is granted for phosphorus and
nitrogen concentrations.  This is a very practical way of acknowledging
uncertainty in the dose-response relationship.  In the present case,
however, it is not clear what action would be taken if chlorophyll is
within bounds, but one or both nutrients exceed the upper threshold
(lower prediction interval).

There seems to be a strong rationale in the report and in supporting
documents for partitioning lakes into clear and colored categories.  It
is not apparent if all lakes can easily be segregated into those two
groups, or if there is an unbroken continuum of lakes.  It may be an
important practical matter if two similar lakes having 39 and 41 pcu
respectively might be assigned standards that differ by a factor of two.

Less information is presented regarding the merits of conductance as an
additional classification variable, although there is little doubt that
higher nutrient concentrations in general would be associated with
increasing solute concentrations.  It might be interesting to learn more
about the TAC’s reasoning regarding the 100 umho/cm dividing line, but
I don’t have any particular reservations about accepting their
conclusion.  Using a rolling average for color seems like a very sound,
if somewhat data intensive, approach.

Florida presently has independent approaches for development of criteria
based on chlorophyll or lake vegetation.  Are the two approaches drawing
on the same set of lakes?  Lakes in my area (and I believe it is true
elsewhere) that have an abundance of macrophytes typically have very
sparse phytoplankton populations even when nutrient concentrations in
the water column would normally support more phytoplankton.  Conversely,
lakes with moderate to high levels of suspended algae rarely develop any
substantial growth of macrophytes.  If this observation applies to
Florida lakes, it might have implications for the criteria development
strategy.  A plot of chlorophyll vs. LVI might be helpful.

Reviewer 5	The preferred means of determining a nutrient criterion for
inland waterbodies is use of a stressor-response relationship.  Florida
DEP addressed the determination of nutrient criteria for lakes based on
two response variables: lake vegetation and chlorophyll a.  Both groups
of primary producers are expected to respond to nutrient enrichment, but
in different ways: macrophyte vegetation, generally benthic vegetation
would largely be expected to decrease in response to enrichment, as a
function of reduced light in the water column.  This light attenuation
is attributed to the increase in phytoplankton chlorophyll a that occurs
with increased nutrient availability.  As such, the macrophyte response
is indirect- nutrients are not affecting the plants themselves, in fact
would tend to promote increases in macrophyte growth, all else being
equal.  But given that most macrophyte species are generally unable to
grow to the lake surface, they will be primarily impacted by the more
direct response of phytoplankton growth to nutrient enrichment.  As
such, because it is at least the second link in this chain, any relation
between nutrients and macrophytes would be expected to be looser, with
the degree of “play” dependent on many other factors.  

In fact, DEP investigated the relationship between macrophytes and
nutrients and found exactly that to be the case- using the lake
vegetation index (LVI) a weak, though significant correlation with
nutrient concentration was found.  An adverse (i.e. inverse) response of
vegetation to nutrients was apparent but insufficiently robust to be
applicable as a basis for establishing nutrient criteria.  A more
effective relationship between nutrients and phytoplankton chlorophyll a
was found, and produced a useful basis for establishing nutrient
criteria as discussed in the previous chapter (Chapter 9).   

Authors recommend that the criteria be performance-based.  This is a
reasonable and acceptable way to proceed- provided that it works both
ways.  The recommendation of the chapter’s authors is that if nutrient
criteria are exceeded but there is no exceedance in the designated
response variable (chlorophyll a), then no nutrient reductions are
needed.  This makes sense ecologically and from a management point of
view.  However and conversely, when data show that nutrients are within
in compliance of established criteria, yet chlorophyll a criteria are
exceeded, then one would expect that nutrient reductions would have to
be undertaken.  This reflects on a much larger issue that permeates this
document: every one of the analyses and criteria presented are
statistically-based, usually based on a lot of long-term monitoring data
and a very few variables.  This results in the emergence of general
patterns that are accompanied by high variability.  Fifty-to-sixty
percent of explained variance is considered exceptional.  

The performance-based approach proposed by DEP for lake chlorophyll a
begs the question of what is responsible for the large variability
associated with all nutrient-based dose-response relationships.  The
variability is a function of lake and stream physical and biological
processing of nutrient inputs.  The large variability of bathymetry,
hydrology, physico-chemistry, flushing, tidal signal, and biology makes
one marvel that nutrient-chlorophyll relationships are as robust as they
are.  Yet the refinement and true understanding of those relationships
will only be achieved by an understanding of the processes operating
within the ecosystem, their rates and the timescales of these processes
with respect to the nutrient supply.  This so-called impedance matching
(Ulanowicz and Hannon, 1987) is an ecosystem principle that reflects how
well-tuned a system is to its cross-boundary flows, and that determines,
in time and space, its ability to process these flows, both as inputs
and export.  The remainder of this sum is left within the system for a
time as a stock or pool with which the system can push off in a variety
of trajectories, depending again on internal processing.  

Therefore, while accepting and endorsing DEP’s call for
performance-based approach to establishment of criteria, this is a
recommendation for investment in additional research into ecological
relationships and evaluation of systems or classes of systems and their
driving processes.  Within a modicum of understanding of these
mechanisms, managers will be able to make informed decisions about
whether and how to shape the Site Specific alternatives for managing
nutrients, beyond what the useful and necessary screening of blind
statistical approaches provides.

Chapter 11. Please comment on the “all streams” statistical
distribution approach for deriving numeric nutrient criteria

Reviewer 1	Again using the Ecoregion concept to form the base for
analyses is a good start that needs to be used throughout the document.
The second paragraph in Section 11 points out that “Multiple factors
can strongly influence the expression of biological responses to
nutrient across water bodies, such as water velocity, residence time,
availability of other nutrients, presence of grazers, availability of
light (due to tree canopy cover and/or water transparency) and
availability of suitable habitat. I have mentioned above, that in my
experience physical factors dominate most aspects of stream biology and
overshadow impacts of nutrients. For this reason I have a hard time with
this approach to setting standards. No matter what percentage of the
data distribution you use for a standard, some streams that are meeting
use and functioning fine will be considered impaired.

Reviewer 2	Chapter 11

Florida DEP certainly has a robust “All Streams” data set with which
to work.  The outlier removal process could use a bit of tweaking, in my
opinion.  I liked that they did not use the removal of the upper
adjacent value – that seems rather confusing to me.  What does concern
me is using TN of 100 mg/L and TP of 15 mg/L as screening thresholds. 
Levels of nutrients found in raw human sewage include TN of 35 mg/L,
ammonium of 15 mg/L and TP of 10 mg/L (Clark et al. 1977).  Therefore,
unless DEP is sampling raw sewage ditches it would be reasonable to
remove outliers more in line with these concentrations that the ones DEP
utilized!

Clark, J.W., W. Viessman, Jr., and M.J. Hammer. 1977. Water Supply and
Pollution Control, 3rd edition. IEP-A Dun-Donnelly Publisher.

I do have another issue with the statistical approach.  DEP kicked out
data when there were 4 or less observations within a year.  Ho9wever, a
lot of variability can occur with only 5 or 6 samples as well, both
seasonal and event-driven.  I suggest drawing the line at 8 observations
or less as year as a cutoff point for data sets.  Natural log
transformation is proper for nutrients and chlorophyll – my own
laboratory regularly does this for our publications.  Finally, the
comparisons between the empirical and theoretical 25th percentiles
produced tight agreement in general.  Thus, with the exception of the
upper outliers and number of observations I am in support of this
approach.

Reviewer 3	To reiterate what I noted in the general comments,
descriptive statistics are necessary for a broad understanding of
systems at the population level. However, the reference approach, in the
absence of biological data, provides the weakest line of evidence for
criteria development.  The all sites chemistry data should,
emphatically, be paired with biological data (i.e., the SCI, SDI, or a
sensitive component metric) and reexamined within ranges of the
biological data.

Reviewer 4	The “all streams” approach has been touted by EPA
ostensibly because it can be applied to situations where dose-response
relationships are not available and reference conditions do not exist or
are very scarce.  However, because reference (or minimally disturbed)
sites seem readily available in Florida, there is little motivation to
use the all streams approach.  It is not reassuring to know that “the
25th percentile from the All Streams data was less than the 75th
percentile … from the Benchmark data set.”  Some reference sites
still will be classified as impaired.

Reviewer 5	The All Streams approach to developing water quality criteria
is an alternative suggested by EPA for areas where data insufficiency
precludes the development of specific numerical nutrient criteria for a
region.  For the Florida nutrient database, this is an interesting data
exercise that presents an alternative benchmark for a nutrient region. 
The lower 25 percentile nutrient benchmarks for all streams, the TN and
TP criteria proposed by EPA as the upper nutrient threshold for this
analysis, represents a much lower concentration for both nutrients than
the 90th percentile reference benchmark as calculated in Chapter 7.

This is a spare chapter that presents data without significant analysis.
 The information presented here is lacking recommendation or
interpretation.  It would have been an improvement to present actual
shapes of nutrient concentration frequency distributions graphically,
and specifically with the two alternative approaches overlain for each
nutrient for each nutrient region.

The All Streams approach is presented as a fail-safe option to criteria
determination and, in the absence of more sophisticated analyses or data
sufficiency; it presents a reasonable, quite conservative stop-gap
provision.  The next step that would have been anticipated was not taken
here- that is relating the nutrient thresholds as calculated using All
Streams to various estimates of stream condition as was done for the
other alternatives, dose-response and nutrient benchmark distributional
approach.  It was not explained why that was not undertaken, but
obviously the All Streams statistical distribution approach is not being
considered as an option for establishing numerical nutrient criteria for
Florida (nor can its worthiness as an option be thoroughly assessed
given the minimal analysis presented).



Chapter 12. Please comment on extrapolating background nutrient
concentrations as an approach to deriving numeric nutrient criteria.

Reviewer 1	As the conclusion of Section 12 states, the data used in this
section are too cloudy to be of any significant use in determining
nutrient standards.

Reviewer 2	Chapter 12

I agree that candidate nutrient criteria cannot be reliably derived
using the Dodds and Oakes (2004) method.  A few comments on Chapter 12:

1) My Chapter 12 citations of tables and figures did not match the ones
supplied.  I assume Table 1 is actually Table 12-5, Figure 1 is Figure
12-3, Figure 2 is 12-2, and Figure 4 is 12-4.

2) Regarding the 9% increase between 1995 to 2004, I think we can safely
assume it is due to increased landscape development (been to Gainesville
lately? I couldn’t recognize the place when I drove through last
winter....).

3) The directions of the correlations certainly are similar to what we
see in the Coastal Plain of the Carolinas.  TN and TP are negatively
correlated with forest, TIN positively correlated with urban land and
negative with wetlands, and so on.  We also find fecal coliforms and TSS
negatively correlated with wetlands, by the way.

Reviewer 3	It is a moot point.  Having rigorously identified benchmark
sites that are good representations of regional background conditions
obviates the need for extrapolating historic background conditions.

Reviewer 4	It seems unfair to say that the Dodds and Oakes method
didn’t work (“candidate nutrient criteria could not be reliably
derived using the Dodds and Oakes (2004) methods as applied here with
the available data”); it was not given an adequate test because, as
stated in the report, the required data were not available.  Instead,
LDI scores were used in a simple regression model, from which only weak
patterns were evident.

The effort to derive background conditions based on the LDI relationship
seemed perfunctory in view of the very promising results presented
previously in the report.  In Chapter 7 (Figure 7-2), nutrient export
(kg/ha/y) was strongly related to LDI for the St Mark’s watershed, and
background yields could have been estimated.  Perhaps the results for
one watershed are not replicated elsewhere, but at least the effort
should have been made to explore the possibility. 

Reviewer 5	This chapter presents an interesting concept that describes
modeling that enables the back-calculation of natural nutrient levels in
unimpacted hydroscapes by extrapolating the relationship between some
index of land use and water-borne nutrient levels to the y-intercept
where anthropogenic land use purportedly equaled zero.  The authors
chose the LDI (Landscape Development Intensity) index of human
disturbance calculated from the area-weighted value of land use type,
because it is a standard calculation of DEP and exists for many sites in
Florida.  The results are predictably scattered- there is no evidence of
a clean relationship with low variance with such an exercise using an
integrative parameter such as the LDI index.  Nonetheless the results
are intriguing and not altogether worthless.

The study design basically assumes that inorganic N and TP are tracers
for human activity, on the assumption that these constituents,
particularly nitrate, nitrite and ammonium would become more part of the
landscape with increasing human activity.  Because organic N is produced
and released in great quantities by natural forested lands and wetland
communities, it is predicted to be poorly correlated with LDI.  The
results do show significant relationships between nutrients and LDI, and
no significance between organic N and LDI.  

The non-linearity of N vs. LDI is to be expected and renders the
hoped-for linear trajectory to an intercept on the nutrient axis an
unattainable goal.  Nutrient releases and transformation in the
landscape resulting from human impact are fraught with variability,
thresholds, breakpoints, non-linearities such that a simple linear
back-projection is not realistic.  The simple fact that human impact
alters the system’s hydrology in unpredictable ways will create a
non-linear nutrient response.  Ultimately though there are more
effective ways of deriving benchmarks for nutrients in unimpacted waters
that are presented in Chapters 6 and 7.



Chapter 13. Please comment on the use of multiple regression and
statistical distribution approaches to deriving numeric nutrient
criteria.

Reviewer 1	For reasons mentioned above, I am not comfortable with
setting standards based on a percentage of a distribution of data. There
are too many lakes that are functioning well and meeting uses that will
be considered impaired. Similarly, without more work on long-term color
data in lakes, I am uncomfortable with the separation of lakes into
colored and clear lakes based on a measured color of 40 PCU.

Reviewer 2	Chapter 13

I am in agreement that large variability in chlorophyll a and nutrient
concentrations renders a linear regression inefficient for this task,
and that the non-parametric model is likely to yield more relevant
results.

Editing remark – page 175, middle, Figure 2 should read Figure 13-3 I
believe.

Page 182 – there are some issues with the Outliers and Invalid Data. 
Nitrate is kicked out if it is above 75 mg/L, yet TN is kicked out if it
is above 20 mg/L – a disconnect here.  The highest nitrate I have seen
in a stream is 25 mg/L (below a failing wastewater treatment plant) thus
DEP needs to bring this expulsion criterion down to 20 or 25 mg/L.  As
to chlorophyll a, the highest level I have seen in a bloom is 400 µg/L,
while DEP is using 1,000 µg/L as a criterion. I suggest cutting that
way down to at least < 400 µg/L.  As to Secchi depth, I am not aware of
Florida lakes that are 50 m deep, so why use a Secchi of 50 m to kick
out an outlier?  How about 10 m?

Another concern I have is determining annual averages using all
available data for a given station in a year.  What if it is only 3 data
points?  Many programs are skewed toward “growing season” sampling
(admittedly Florida has a long growing season) but this would tend to
skew the data for some lakes toward elevated chlorophyll in the means. 
I suggest using a lower limit of 8 samples in a year to create a mean,
with samples taken seasonally to collect lower-impact periods as well.

Reviewer 3	The multiple regression models provided in Chapter 13 offer a
more refined way to derive criteria than the conventional linear models
appearing in Chapters 9 and 10, though the results from all three
chapters generally comport each other.  More importantly, however, the
results suggest that once static numeric criteria are derived for either
nitrogen or phosphorus, those criteria should be interpreted as having
uncertainty when applied to a specific waterbody, and therefore not
applied with the traditional independent application reserved for
toxicants.  Figures 13-5 and 13-7 clearly show the range of potential
criteria derived from a population of lakes.  

As with Chapter 11, the distributional analyses of data are important
for understanding the system at a population level, but are not
particularly helpful for deriving criteria.  Again, melding in the LVI
and examining the chemistry data within categorical levels of the LVI
would help to lend context to the percentile ranges.

Reviewer 4	I am not familiar with multiple logistic regression, but it
seems like an innovative analytical approach, and the graphs are
informative.  Placement of the diagonal line (now set to the “average
ratio between ln (TN) and ln(TP)”) deserves more attention because it
makes an important statement about the nutrient regime.  The ratio used
in the graph (0.588/0.013 = 45:1) is very high and indicative of a
phosphorus-limited situation.  Placement of the diagonal line might be
adjusted according to expectations for a particular region.  For
example, in supporting references for Chapter 9, an N:P ratio of 16:1
was assumed to have relevance for distinguishing between limitation by N
or P.

The output of the within-lake model (Figure 13-2) shows the risk
inherent in basing assessments solely on nutrient concentrations. 
Basing decisions on nitrogen and phosphorus relationships from the
entire data set would place criteria well below anything observed in the
lake, even though most chlorophyll measurements in the lake are in
attainment of the 5 ug/L threshold.  The analysis seems to add weight to
the notion that a tiered approach would be useful.

Why are the distributional statistics for lakes determined on a
statewide basis when there are identifiable geographic patterns for
phosphorus and nitrogen in streams?

Reviewer 5	Chapter 13 provides an intriguing analysis of the use of
non-parametric statistics and Bayesian modeling techniques in
establishing nutrient criteria for lakes.  The analysis presents a
proof-of-concept exercise that demonstrates how a set of TN and TP
criteria might be developed using multiple regression models.  The
results of the analysis supports an earlier recommendation in this
review that within-lake models of TN-TP and chlorophyll a will yield the
most robust results and the most likely statistical approximation of a
mechanistically based cause and effect relationship.  

Various aspects of the balance between site specificity, model
robustness and the need for a large enough sample size to support three
degrees of freedom leads to the expansion of the sample size across
similarly classed lakes such that a large pool of lakes (93) can be fit
to the model.  It is curious that the entire pooled dataset in figure
13-4 shows individual lakes within the pool seem to trend to a lower
slope than the slope of all pooled observations.  This is not discussed
in the text.  Overall, the nonparametric multiple regression approach to
setting nutrient criteria protective of a specific chlorophyll threshold
for lakes seems to hold promise and should be further explored.

It is of passing note that here as in other chapters, the
distinctiveness of colored lakes is brought up and significant attention
is devoted to incorporating terms in the model that will account for
color.  There are a number of colored lakes that show a negative
relationship between color and chlorophyll a.  While this most obviously
may be related to light attenuation, there are other possibilities. 
Often the constituents of colored dissolved organic matter may be
inhibitory to phytoplankton growth, and they may often act as binding or
chelating agents for nutrients, reducing bioavailability.  Therefore,
the complication of dose-response mechanisms introduced by colored
waters continues to plague efforts to quantify relationships with simple
models.



Chapter 14. Please comment on the downstream protection approaches.

Does the modeling approach provide a sound, scientific basis for
determining numeric nutrient loadings from upstream waters to estuaries?

Could these approaches be used to determine nutrient loadings for other
waterbody types (i.e., lakes)?

Given the available information, do these approaches account for spatial
variability, distance from downstream water, and estimated assimilation?

Do you know of any alternative approaches to capturing the relationship
between upstream and downstream water quality in a WQS.

Reviewer 1	The downstream protection approaches described in Section 14
are sound and consistent with the conventional approach used for decades
to manage eutrophication in lakes (Bachmann and Canfield 1981). While it
has not been used extensively, Vollenwider type models have successfully
estimated actual total phosphorus concentrations in estuaries and have
tied nutrient concentrations to different land uses (Meewig et al 1998;
Meewig 1999). Breaking the estuaries into watershed and measuring the
actual loads from different sources is the only way to really manage
aquatic systems. Using some form of Vollenwider loading model will allow
managers to pinpoint significant loads that may be causing unwanted
eutrophication.

One significant problem I have with Section 14 is that there is an
assumption that nitrogen is the only important nutrient controlling
trophic state in estuaries. In 2000 I sampled near shore coastal waters
from three stations off of every Florida County that touched the ocean.
These data showed that total phosphorus accounted for significantly more
variance in measured chlorophyll than total nitrogen, strongly
suggesting that phosphorus in the primary limiting nutrient in near
shore coastal waters of Florida (Hoyer et al. 2002). Tampa Bay is an
exception to this and nitrogen is more important there, but that is
because Tampa Bay receives water from phosphorus rich soils. Therefore,
the modeling efforts accomplished in Section 14 should be done using
phosphorus instead of nitrogen.

Canfield, Jr., D. E. and Bachmann, R.W. 1981. Prediction of total
phosphorus concentrations, chlorophyll a , and Secchi depths in natural
and artificial lakes. Can. J. Fish. Aquat. Sci. 38:414-423.

Hoyer, M. V., T. K. Frazer, S. K. Notestein and D. E. Canfield, Jr.
2002. Nutrient, chlorophyll, and water clarity relationships in
Florida’s nearshore coastal waters with comparisons to freshwater
lakes. Canadian Journal of Fisheries and Aquatic Sciences 59: 1-8.

Meeuwig, J. J., J. B. Rasmussen, and R. H. Peters. 1998. Turbid waters
and clarifying mussels: their moderation of empirical chl:nutrient
relations in estuaries in Prince Edward Island, Canada. Mar. Ecol. Prog.
Ser. 171: 139-150.

Meeuwig, J. J. 1999. Predicting coastal eutrophication from land-use: an
empirical approach to small non-stratified estuaries. Mar. Ecol. Prog.
Ser. 176: 231-241.

Meeuwig, J. J., P. Kauppila, and H. Pitkanen. 2000. Predicting coastal
eutrophication in the Baltic: a limnological approach. Can. J. Fish.
Aquat. Sci. 57: 844-855.

Reviewer 2	Chapter 14

14.1.5 – I am concerned about one of the statements on page 188.  It
states “in the absence of more specific information it may be
appropriate to reduce the anthropogenic contribution of the TN load to
estuaries by a constant fraction.”  This is not appropriate.  As is
pointed out within this very chapter, Florida estuaries vary widely in
their watershed/water ratios, their percent loading by individual
sources, their total loads, their size, depth, flushing, and certainly
their response variables such as seagrass coverage and fisheries.  So,
the answer here is that if necessary specific information is not
available – DEP needs to get the funding to obtain that information. 
There is no “one size fits all” solution for ecological problems
such as this.

14.1.18 - Another concern I have is that of the contribution of septic
systems to the nutrient load in estuarine waters.  While DEP certainly
accounts for direct point source contributions of nutrients to estuaries
in their models, it is known from research in the Carolinas that
nutrients from septic systems directly enter estuarine waters as well
(Cahoon et al. 2006).  Especially on the west coast of Florida there are
many septic systems with direct or near-direct access to estuaries.  In
studies in Sarasota Bay and Charlotte Harbor it has been found that
fecal microbes from septic systems enter the surface waters on the
outgoing tide (Lipp et al. 1999; 2001); thus, presumably ammonium and TP
will as well.  This should be factored in for estuaries where septic
system use is prevalent.

Cahoon, L.B., J.C. Hales, E.S. Carey, S. Loucaides, K.R. Rowland and
J.E. Nearhoof. (2006). Shellfish closures in southwest Brunswick County,
North Carolina: Septic tanks vs. storm-water runoff as fecal coliform
sources. Journal of Coastal Research 22:319-327. 

Lipp, E.K., S.A. Farrah and J.B. Rose. (2001). Assessment and impact of
fecal pollution and human enteric pathogens in a coastal community.
Marine Pollution Bulletin 42:286-293. 

Lipp, E.K., J.B. Rose, R. Vincent, R.C. Kurz and C. Rodriquez-Palacios.
(1999). Diel variability of microbial indicators of fecal pollution in a
tidally influenced canal: Charlotte Harbor, Florida. Southwest Florida
Water Management District, Technical Report.

The tide is an important consideration in obtaining accurate measures of
nutrient loading to downstream waters.  The most accurate way of
accomplishing this is by use of boat mounted acoustic Doppler current
profiling (Reed et al. 2004).  This process has been utilized to obtain
the most comprehensive modeling of nutrient loading, trends over time
and patterns in the Neuse River Estuary to date (Burkholder et al.
2006).  Additionally, in terms of spatial variability, the sampling
scheme supporting that effort was able to conclusively show that
collecting data only mid-stream leads to incorrect loading and impact
estimates.  By that I mean that near-shore sampling should be
accomplished as well as mid-stream, because due to wind, current and
tidal patters some of the largest algal blooms are found in shallower
areas near-shore.

Reed, R.E., H.B. Glasgow, J.M. Burkholder and C. Brownie. 2004. Seasonal
physical-chemical structure and acoustic Doppler current profiler flow
patterns over multiple years in a shallow, stratified estuary, with
implications for lateral variability. Estuarine, Coastal and Shelf
Science 60:549-566.

Burkholder, J.M., D.A. Dickey, C. Kinder, R.E. Reed, M.A. Mallin, G.
Melia, M.R. McIver, L.B. Cahoon, C. Brownie, N. Deamer, J. Springer,
H.B. Glasgow, D. Toms and J. Smith. 2006. Comprehensive trend analysis
of nutrients and related variables in a large eutrophic estuary: A
decadal study of anthropogenic and climatic influences. Limnology and
Oceanography 51:463-487.

Finally, I recommend that the modelers also have a look at the following
paper:

Ensign, S.H., J.N. Halls and M.A. Mallin. 2004. Application of digital
bathymetry data in an analysis of flushing times of two North Carolina
estuaries. Computers and Geosciences 30:501-511.

This paper provides an approach that determines both flushing time of
the estuary and assimilation of N within that estuary, and consequent
export to sea.

Reviewer 3	a)	Does the modeling approach provide a sound, scientific
basis for determining numeric nutrient loadings from upstream waters to
estuaries?

The concentrations identified in Table 10 of Chapter 14 are simply an
additional line of evidence that support identification of criteria. 
Models are abstractions and depend on the quality of inputs.  The TMDLs
presented appear to have been developed by adhering to good methods, and
were well-calibrated using empirical data.  The approach would be more
compelling if the examples for intact systems (as previously noted) were
included for comparisons.    

b)	Could these approaches be used to determine nutrient loadings for
other waterbody types (i.e., lakes)?

Yes, it would have been nice to see the Vollenweider example worked out
for a few lakes.

c)	Given the available information, do these approaches account for
spatial variability, distance from downstream water, and estimated
assimilation?

I am not sufficiently familiar with the vagaries of the SPARROW model to
comment on this point.

d)	Do you know of any alternative approaches to capturing the
relationship between upstream and downstream water quality in a WQS.

Short answer, no.  The load-based targets for a receiving body are the
most straightforward, and defensible on an ad hoc basis.  In other
words, criteria derived to protect the use of a given class of
waterbodies should stand unless superseded by another need.

Reviewer 4	The SPARROW model was designed for estimating nutrient loads,
and it has typically been applied to large regions (e.g., multi-state). 
While there is no conceptual reason why the model could not be applied
to lakes as well as estuaries, there may be a practical limitation to
the lower bound on geographic scale.  As stated in a recent USGS fact
sheet:

In general, SPARROW models illustrate the broad spatial patterns of
water quality well, but may be less accurate for a single stream because
of limitations in the underlying data sets.

I am not sufficiently familiar with SPARROW to know if it incorporates
process (e.g., denitrification) that would be relevant to assimilative
capacity.

I think the larger issue concerns the extent to which this approach
provides information useful to the development of nutrient criteria for
streams.  Some of the concerns are the same as those I cited for the
Nutrient Longitudinal Study (Chapter 8).  There is a fundamental
question about linking attainment in estuaries with attainment in
tributary streams.

SPARROW provides nitrogen load estimates that are used to drive models
(not presented) that predict concentrations in each estuary.  It is not
clear if the linkage includes contributions of nitrogen generated by the
marsh grasses within the estuary.  Based on existing thresholds,
conclusions are reached about total loads (and thus a flow-weighted
concentration) consistent with attainment in estuaries.  In the context
of the TMDL process, it looks like a reasonable approach.

It remains a mystery how this information relates to the criteria that
are appropriate for attainment in streams.  Given the differences among
estuaries in hydraulic conditions, for example, it is hard to imagine
how generalizations can be made about the allowable load, much less the
appropriate concentrations in tributary streams.  Perhaps that was the
motivation for the comment on page 211:

It is important to note that values derived from this approach are not
applicable to protection of instream conditions or use attainment.

The section on lake modeling is so brief that it is hard to evaluate. 
Just like estuaries, there is a fundamental issue about the linkage
between concentration-based standards in lakes and in the tributary
streams.  Criteria are defined to protect uses in lakes; if they
aren’t being met, a TMDL should be developed.

Reviewer 5	Two potential methods are presented for assessing the
“protectiveness” of watershed nutrient criteria to downstream
waters- estuaries and lakes respectively.  The bulk of the chapter deals
with a description of application of the Spatially Referenced Regression
on Watershed Attributes (SPARROW) model, developed by USGS.  This model
is used to estimate TN loading from the watershed to the receiving
estuary for 13 estuaries in the state.  While the model shows promise as
a tool for estimating N loads to many Florida estuaries, it is important
to note that several Florida estuaries would be inappropriate as targets
for SPARROW modeling.  Specifically, several estuaries of the southern
peninsula, such as Florida Bay, Whitewater Bay and Rookery Bay, are not
amenable to analysis by the SPARROW watershed model.  Large interaction
of surface flows with groundwater flows, the lack of mass flow or
well-constrained point-source input, and multiple ungauged inputs would
render such modeling ineffective.

There is a need also to evaluate the spatial heterogeneity of receiving
waters, particularly in estuaries.  The gradients that characterize
estuarine systems demand that spatial analysis be performed, or at least
considered, at the sub-basin or sub-estuary scale (Madden et al. 2008). 
Direct loadings from point sources in the upper estuaries render those
areas different in the extreme from the lower estuary, or even
mid-estuary.  Spatially complex receiving waters create complex
hydrodynamic circulations, nutrient regimes and biological responses. 
Nutrient spiraling, airshed loading and other uncertainties will combine
create complex dose-response relationships.  Adaptation of SPARROW
analysis to the sub-estuary scale is warranted.  Temporal scales are
also exceedingly important to estuarine processing of nutrients- the
timestep of the sampling protocol and of the analysis is critical and
these need to be evaluated in the establishment of criteria. 
Seasonality, pulses, and low frequency oscillations, such as ENSO, will
impart variability in physical and biological processing of nutrient
inputs.  

The section describing application of the Vollenweider model to
downstream lakes represents a start to the issue of protectiveness of
upstream criteria inland receiving waters.  There is minimal discussion
of implementation presented.  The model is valid for the application
described- calculation of P loading and concentration in lakes as
functions of discharge, residence time, lake volume and flow-weighted
nutrient concentration.  Yet some aspects of watershed management are
not adequately addressed.  The total loading rate can be calculated, but
the lake would be receiving loading from multiple sources, both point
and non-point.  Any action taken would be dependent on the source(s) of
the nutrient and a balancing of the source reductions necessary to
achieve compliance.  How that would be apportioned or even determined
among the sources is left for later exposition.  

The widening of the application from single lakes to regions of lake
should be undertaken with caution.  The uncertainties of the model when
applied to a single lake will be amplified exponentially if broadened to
groups of lakes, with the attendant issues of assigning and apportioning
reduction becoming even more intractable.  The protectiveness of the
upstream nutrient criteria should be approached with more rigor with
respect to lakes, with individual modeling efforts per lake and strict
determination of the terms in the model, including sources, atmospheric
deposition and processing.

Furthermore, it is probable that the suggestion to address lake regions
by averaging of terms across groups of lakes would generate additional
uncertainty to the point where models would break down.  An
understanding of lake mean depth and morphoedaphic factors is necessary
to completely assess the fate of P entering a lacustrine system. 
Biogeochemical fate and transformation of nutrients, in this case P,
will be dependent on the nature of the soils and sediments, presence of
groundwater seeps and nutrient inputs, presence of phosphatic rock,
fluctuations in lake depth, and biological variables such as whether or
not benthic vegetation is present, the presence and types of
phytoplankton communities, predation and trophic interactions within the
system, types of fauna present.  

P dynamics will also be dependent on status of other nutrients - N
dynamics and cycling, N limitation of primary production, nitrification-
denitrification coupling (Jenkins and Kemp 1994), and C limitation of
primary production.  In short, simple models will work well as a primary
screening tool as applied to individual lakes with well-constrained
datasets and boundary conditions but there is a large uncertainty
associated with demanding too much predictive capability from them,
especially when used to “batch process” the information for groups
of lakes. 



Additional Comments/References

Reviewer 1	Preface

It is obvious that individual scientists have spent several years
creating this document. This is truly a large, technical and complicated
document, especially with all of the appendices, to review in a short
two-week time. I could spend several months following lines of analyses
and testing them with data I have available or finding other available
data to test them. With that said, I will give my best general and
specific comments on this document with the caveat that I could do a
much better job with much more time. Additionally, in the future of such
endeavors, giving reviewers much more time and even chances to question
the original authors would lead to a much stronger final document.

Reviewer 2	My Peer Review Charge has no "Specific Question 1". 
"Specific Charge Questions" begin with 2. (Chapter 3).  

In my document Question 1 is a "General Charge Question" with three
general considerations (i.e. strengths and weaknesses, validity of lines
of evidence, additional technical considerations for deriving criteria)
that I took into consideration throughout the document in reviewing the
individual chapters.  So, the general charges are addressed throughout
the review, of course depending upon the contents of the individual
chapters.

Overall, I am pleased that Florida DEP has spent a lot of time and
resources attempting to obtain nutrient criteria.  I am supportive of
the majority of the techniques utilized within the document.  There are
a few approaches I took issue with as noted within.  I have also
supplied some alternatives (some which can be used in parallel) that DEP
may wish to try (I have supplied references) to provide greater
scientific validity to their efforts in this very important process.

Reviewer 3

	Reviewer 4

	Reviewer 5	Conclusions and overall recommendations

This effort represents a good foundation for the development of nutrient
criteria for inland waters in the State of Florida.   However, overall,
the document is a mixed result due to several methodologies in the
process of being developed; lack of statewide applicability and the most
effective current methodology for streams the SCI as a good functional
baseline from which to start.  The Dose-Response analyses proved
unsatisfactory yet should, with additional research and information
development, hold promise.  Methods to determine whether the upstream
criteria are protective of downstream waters are insufficiently
evaluated in this document.  The SPARROW model should be applied
cautiously and only to appropriate systems in assessing downstream
effects, namely those with well-defined channel flows, and well-
especially to receiving reaches and coastal waters.  The Nutrient
Distributional Approach represents a good backstop where data are
insufficient to develop specific numeric criteria.  

What the document mostly lacks is an overall orientation and strategy, a
set of clear recommendations, and an implementation plan for bringing
these various approaches to bear.  A determination of the boundaries of
analysis, of what protocol takes precedence, when each would be applied,
where each is applicable, how data and results are to be synthesized to
arrive at criteria is wanting.  Recognizing that this is a work in
progress, there is the expectation that a coherent strategy for
combining the information in these individual chapters will be
developed.  It is expected that with additional input and review, and
development of the document and well-rounded approach, such a strategy
for implementation will be forthcoming.



Literature Cited

Greening, H. and A. Janicki. 2006.  Toward reversal of eutrophic
conditions in a subtropical estuary: water quality and seagrass response
to nitrogen loading reduction in Tampa Bay, FL USA.  Environ. Management
38:163-178.

Madden, C. J., R. Smith, E. Dettmann, N. Detenbeck, J. Kurtz, J.
Latimer, J. Lehrter, W. Nelson, S. Bricker.  2008.  A typology of
estuaries supporting the development of a national nutrient criteria
framework for estuarine systems.  USEPA report of the National Nutrient
Criteria Development Workgroup.  In EPA review.

Odum, H. T. 2007. Environment, Power and Society for the 21st Century:
The hierarchy of energy. 2nd Edition. Columbia University Press. 426 pp.


Prygiel, J. 1994. Comparaison de six indices diatomiques et et deux
indices invertebres pour l’estimation de la qualite de l’eau de la
riviere Senese (France). Ecologia Mediterranea, 20(1/2):121-133.

Soininen, J. and K. Kononen. 2004. Comparative Study of monitoring
South-Finnish rivers and streams using macroinvertebrates and benthic
diatom community structure.  Aquatic Ecology. 38:63-75.

Ulanowicz, R. and B. Hannon. 1987. Life and the production of Entropy. 
Proc. Royal Soc. London. 232:181-192.



 Environmental Management (2006) 37:209-229

   HYPERLINK
"http://www.twca.org/downloads/report/report_no_appendix.pdf" 
http://www.twca.org/downloads/report/report_no_appendix.pdf ; see
Conclusions.

 SPARROW fact sheet –   HYPERLINK "http://pubs.usgs.gov/fs/2009/3019/"
\t "_blank"  http://pubs.usgs.gov/fs/2009/3019/ 

  PAGE  i 

  PAGE  1 

  PAGE  51 

  PAGE  52 

