Nutrient loading - tropic state relationships in Florida lakes

MISSING IMAGE

Material Information

Title:
Nutrient loading - tropic state relationships in Florida lakes
Series Title:
Florida Water Resources Research Center Publication Number 56
Physical Description:
Book
Creator:
Baker, Lawrence A.
Brezonik, Patrick L.
Kratzer, Charles R.
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Notes

Abstract:
Quantitative relationships among important lake trophic state indicators and watershed enrichment factors were examined using a data base of 101 Florida lakes. Trophic state information was obtained from three previous surveys on Florida lakes, from which data were compiled into a uniform format. Watershed nutrient export data and lake nutrient loading rate data were obtained from a comprehensive review of the literature and from the Florida portion of the National Eutrophication Survey (NES) conducted by the U.S. EPA. This data base was used to determine relationships between non-point source (NPS) nutrient loading rates and land use characteristics of Florida watersheds; to evaluate interrelationships among trophic state indicators in Florida lakes; and to revise nutrient loading models and develop appropriate nutrient loading criteria for these lakes. The magnitude of NPS nutrient loadings was estimated from published export coefficients and by a statistical analysis of the NES watersheds. The literature-based approach produced a wide range of export coefficients for Florida watersheds. 0.2-0.7 kg P/ha-yr and 1.5-6.1 kg N/ha-yr for forests; 0.4-2.4 kg P/ha-yr and 2-50 kg N/ha-yr for cropland; 0.2-4.7 kg P/ha-yr and 1.5-7.4 kg N/ha-yr for residential areas; and 0.3-7.5 kg P/ha-yr and 3-10 kg N/ha-yr for urban areas. NPS nutrient loading (dependent variables) and land use characteristics (independent variables) for 41 NES watersheds were analyzed by stepwise multiple regression to improve the predictive capability of the land use-nutrient loading approach. For phosphorus, a model using three land terms (cropland, forest & rangeland) explained 72% of the variance in NPS loading (vs 21% for a model with drainage area as the sole independent variable). Models to predict NPS nitrogen loading and hydraulic flow had high levels of predictability using drainage area as the sole independent variable (r2 = 0.84 and 0.91, respectively); and inclusion of land use data resulted in little predictive improvement. Evaluation of the limnological characteristics of 101 Florida lakes indicated that most of these lakes are shallow and well-mixed; few have stable thermoclines or anoxic hypolimnia. The lakes are highly variable in alkalinity (0-16mg/Las CaCO3), pH (4.7- >10) and amount of color (2-54 CPU), reflecting differences in geological origin and watershed characteristics. The majority of the lakes in the data base are eutrophic, and unlike most temperate lakes, tend to be nitrogen-limited (46% had SIN:SRP ratios of <10:1). For a given level of total phosphorus, Florida lakes have less chlorophyll a than do temperate lakes; this is true even for phosphorus-limited Florida lakes. Carlson's trophic status index (TSI) was modified for application to Florida lakes by inclusion of a nitrogen index to reflect the importance of nitrogen as a limiting nutrient. A composite TS1 was developed by averaging the TS1's based on Secchi disk, chlorophyll a and nutrient concentration (the smaller of the nitrogen or phosphorus index) to produce an index that reflects the multidimensionality of the eutrophication concept. Various nutrient loading models were evaluated statistically for their ability to predict mean chlorophyll a, nitrogen and phosphorus concentrations in Florida lakes. The best predictions for nitrogen and phosphorus were made using a modified Dillon and Rigler-type model, while the best predictions of chlorophyll a were obtained using a Jones and Bachman-type model. Existing phosphorus loading criteria were evaluated for the Florida NES lakes. The 1975 Vollenweider criteria and the 1975 Dillon criteria were revised to improve their predictive capabilities; in both cases the revisions resulted in higher critical values (minimum mesotrophic and minimum eutrophic loading rates) than the original criteria. Both sets of criteria were equally successful in delineating eutrophic lakes from mesotrophic lakes in the NES data base. Nitrogen loading criteria were developed using loading terms analogous to those used for phosphorus. The most successful nitrogen criteria were based on a Dillon-type model. In evaluating the impact of a proposed management strategy it is suggested that both nitrogen and phosphorus loading criteria be used; the correct response of the lake will be obtained with the criterion that predicts the lower trophic status.

Record Information

Source Institution:
University of Florida Institutional Repository
Holding Location:
University of Florida
Rights Management:
All rights reserved by the source institution and holding location.
System ID:
AA00001534:00001


This item is only available as the following downloads:


Full Text



















Publication No. 56


NUTRIENT LOADING TROPIC STATE RELATIONSHIPS
IN FLORIDA LAKES

by

Lawrence A. Baker
Patrick L. Brezonik
&
Charles R. Kratzer










NUTRIENT LOADING TROPHIC STATE RELATIONSHIPS

IN FLORIDA LAKES







by


Lawrence A. Baker


Patrick L. Brezonik



Charles R. Kratzer



Publication No. 56



Florida Water Resources Research Center
Research Project Technical Completion Report


OWRT Project Number A-038-FLA


Annual Allotment Agreement Numbers
14-34-001-9010
14-34-001-0110



Report Submitted: May 21, 1981



The work upon which this report is based was supported
in part by funds provided by the United States Department of
the Interior, Office of Water Research and Technology as
authorized under the Water Resources Research Act of 1964 as
amended.










TABLE OF CONTENTS



Abstract iv
Acknowledgements vi

Chapter page

I. Introduction 1
Conceptual framework for predictions of lake trophic
status 1
Scope and objectives of this report 2

II. Data Sources and Methods 4
Data sources 4
Statistical methods 11

III. Predictions of Nutrient Loading for Florida Lakes 12
Introduction 12
Literature review: nutrient sources to Florida lakes 13
Point source loadings 13
Precipitation inputs 15
Non-point source loadings from Florida watersheds 15
Statistical analysis of nutrient export from the NES
watersheds 23
Equations to predict phosphorus loading (TPL) 27
Predictions of nitrogen loading (TNL) 30
Equations to predict flow (FLOW) 30
Application 31

IV Limnological Characteristics of Florida Lakes 40
Morphometric' characteristics 40
Chemical and physical characteristics 42
Relationship between pH and alkalinity 42
Factors affecting transparency 42
Biological characteristics 44
Phytoplankton communities 44
Fish populations 47
Nutrient limitation 47
The relationship between nutrients and chlorophyll a
standing crop 53
Analysis of seasonal trends 59
Development of a trophic state index 61

V. Application of Nutrient Loading Models to the Florida NES
Lakes 67
Introduction 67
Historical development 67










Chapter page


Phosphorus input/output models 67
Prediction of Rp 72
Prediction of chlorophyll a concentration 72
Nutrient loading criteria: graphical approaches 75
Application of nutrient loading models to Florida lakes 75
Total phosphorus concentration 75
Total nitrogen concentration 77
Phosphorus and nitrogen retention coefficients 80
Prediction of chlorophyll a 85
Nutrient loading criteria for Florida lakes 89
Phosphorus loading models 89
Nitrogen loading models 98
Application 107

Summary and Conclusions 112

References 114

Appendices 121










ABSTRACT


Quantitative relationships among important lake trophic state indicators
and watershed enrichment factors were examined using a data base of 101
Florida lakes. Trophic state information was obtained from three previous
surveys on Florida lakes, from which data were compiled into a uniform format.
Watershed nutrient export data and lake nutrient loading rate data were ob-
tained from a comprehensive review of the literature and from the Florida
portion of the National Eutrophication Survey (NES) conducted by the U.S. EPA.
This data base was used to determine relationships between non-point source
(NPS) nutrient loading rates and land use characteristics of Florida water-
sheds; to evaluate interrelationships among trophic state indicators in
Florida lakes; and to revise nutrient loading models and develop appropriate
nutrient loading criteria for these lakes.

The magnitude of NPS nutrient loadings was estimated from published ex-
port coefficients and by a statistical analysis of the NES watersheds. The
literature-based approach produced a wide range of export coefficients for
Florida watersheds. 0.2-0.7 kg P/ha-yr and 1.5-6.1 kg N/ha-yr for forests;
0.4-2.4 kg P/ha-yr and 2-50 kg N/ha-yr for cropland; 0.2-4.7 kg P/ha-yr and
1.5-7.4 kg N/ha-yr for residential areas; and 0.3-7.5 kg P/ha-yr and 3-10 kg
N/ha-yr for urban areas. NPS nutrient loading (dependent variables) and land
use characteristics (independent variables) for 41 NES watersheds were analyzed
by stepwise multiple regression to improve the predictive capability of the
land use-nutrient loading approach. For phosphorus, a model using three land
terms (cropland, forest & rangeland) explained 72% of the variance in NPS
loading (vs 21% for a model with drainage area as the sole independent vari-
able). Models to predict NPS nitrogen loading and hydraulic flow had high
levels of predictability using drainage area as the sole independent variable
(r2 = 0.84 and 0.91, respectively); and inclusion of land use data resulted in
little predictive improvement.

Evaluation of the limnological characteristics of 101 Florida lakes in-
dicated that most of these lakes are shallow and well-mixed; few have stable
thermoclines or anoxic hypolimnia. The lakes are highly variable in alkalinity
(0-16mg/Las CaCO3), pH (4.7 >10) and amount of color (2-54 CPU), reflecting
differences in geological origin and watershed characteristics. The majority
of the lakes in the data base are eutrophic, and unlike most temperate lakes,
tend to be nitrogen-limited (46% had SIN:SRP ratios of <10:1). For a given
level of total phosphorus, Florida lakes have less chlorophyll a than do tem-
perate lakes; this is true even for phosphorus-limited Florida Takes.

Carlson's trophic status index (TSI) was modified for application to
Florida lakes by inclusion of a nitrogen index to reflect the importance of
nitrogen as a limiting nutrient. A composite TSI was developed by averaging
the TSI's based on Secchi disk, chlorophyll a and nutrient concentration (the
smaller of the nitrogen or phosphorus index) to produce an index that reflects
the multidimensionality of the eutrophication concept.

Various nutrient loading models were evaluated statistically for their
ability to predict mean chlorophyll a, nitrogen and phosphorus concentrations
in Florida lakes. The best predictions for nitrogen and phosphorus were made










using a modified Dillon and Rigler-type model, while the best predictions
of chlorophyll a were obtained using a Jones and Bachman-type model.

Existing phosphorus loading criteria were evaluated for the Florida
NES lakes. The 1975 Vollenweider criteria and the 1975 Dillon criteria
were revised to improve their predictive capabilities; in both cases the re-
visions resulted in higher critical values (minimum mesotrophic and minimum
eutrophic loading rates) than the original criteria. Both sets of criteria
were equally successful in delineating eutrophic lakes from mesotrophic
lakes in the NES data base. Nitrogen loading criteria were developed using
loading terms analogous to those used for phosphorus. The most successful
nitrogen criteria were based on a Dillon-type model. In evaluating the im-
pact of a proposed management strategy it is suggested that both nitrogen
and phosphorus loading criteria be used; the correct response of the lake
will be obtained with the criterion that predicts the lower trophic status.










ACKNOWLEDGEBENTS


Appreciation is extended to Dr. Jack Gakstatter of the U.S. EPA Corvallis
laboratory for supplying NES data and aerial photos for analysis of watershed
land use. Ms. Janet Denger of the University of Florida Remote Sensing Labora-
tory performed the land-use analyses on the NES watersheds. Ms. Dorthy Murphy
assisted in compiling data for computer analysis. The statistical advice and
assistance provided by Dr. Wayne C. Huber and Mr. Gary Goforth of the Depart-
ment of Environmental Engineering Sciences and by Mr. Mike Conlon of the De-
partment of Statistics is gratefully acknowledged. Most of this report was
typed by Ms. Alicia Maxwell, and the task was completed by Ms. Patty Hersey.














CHAPTER I, INTRODUCTION


Inland lakes are an important natural resource for Florida, and they
are particularly valuable recreational assets. Unfortunately, the potential
beneficial use of many Florida lakes has been impaired by accelerated eutro-
phication. Although the process of eutrophication is natural, the addition
of plant nutrients from municipal sewage, septic tanks, urban and agricul-
tural runoff, livestock operations, and industrial effluents accelerates
the process by stimulating the growth of algae and macrophytes. Excessive
levels of aquatic production result in general impairment of water uses.
Under highly enriched conditions, blue-green algae tend to dominate the
algal flora and may form dense, unsightly surface blooms that impart un-
pleasant tastes and odors to the water. In lakes that stratify, the decom-
position of algae and macrophytes results in the depletion of dissolved
oxygen in the bottom waters, and this in turn limits the diversity of benthic
organisms and benthic feeding fish. The quality of fishing eventually
decreases, as populations of game fish, such as bass and sunfish, are re-
placed by populations of rough fish (bullheads, shad, carp). Recreational
boating may become restricted by thick beds of weeds. Contact sports are
diminished by the reduced transparency of the water, the growth of water
weeds, and the occurrence of infections in swimmers.

One of the most notable examples of cultural eutrophication in Florida
is Lake Apopka, a 12,000 hectare (ha) lake once recognized for its excep-
tional bass fishery. Since the late 1940's, water quality in the lake has
deteriorated as the result of discharges of municipal wastewater, citrus
processing plant wastes, and muck farm irrigation water to the lake, in
conjunction with ill-fated attempts to control the growth of water hyacinth
and rough fish. As the result of these perturbations, populations of impor-
tant game fish have been drastically reduced, and the dominant fish in the
lake today are two species of shad (USEPA 1978). Cultural eutrophication has
affected other lakes in the Oklawaha chain (Brezonik and Shannon 1971), Lake
Okeechobee (MacGillet al. 1976), and numerous smaller lakes throughout the
state. In order to manage Florida's lakes and control eutrophication, plan-
ners and regulatory agencies must be able to quantify nutrient loadings to
lakes from their watersheds and to predict the response of individual lakes
to changes in nutrient loading.

CONCEPTUAL FRAMEWORK FOR PREDICTIONS OF LAKE TROPHIC STATUS

The most widely used approach in predicting the trophic status of lakes
has involved the use of simplified input-output (I/O) models (Dillon and
Riglerl974a; Chapra and Tarapchak 1976; Vollenweider 1968, 1969, 1975, 1976;
Kratzer 1979). The I/O models are based on the assumption that a lake is a
continuously-stirred tank reactor (CSTR) in which all nutrient fluxes are at
steady state. Since phosphorus is the most common limiting nutrient in
temperate zone lakes, most of these models have been developed to predict
mean lake phosphorus concentrations in a lake. The utility of these models


- 1 -











is enhanced by their modest data requirements, which include only morphologic
and hydrologic parameters and areal phosphorus loading rates.

Phosphorus loading models have been further advanced by the recognition
of statistically significant relationships between concentrations of total
phosphorus and the concentration of chlorophyll a in lakes (Sakamoto 1965;
Dillon and Rigler 1974). This development has led to the formulation of I/O
models in which phosphorus loading can be used to predict the mean chlorophyll
a concentration directly (Vollenweider 1976; Chapra and Tarapchak 1976;
Kratzer 1979; Uttormark and Hutchins 1978). Finally, several authors have
found strong correlations between Secchi disk transparency and chlorophyll
a concentration (Carlson 1977, Brezonik 1978), enabling predictions of lake
transparency from chlorophyll a data. A conceptual framework for predic-
tion of lake trophic status based on I/O models is outlined in Figure I-1.

The need for nutrient loading data in the I/O models has prompted ef-
forts to predict the non-point source (NPS) loading of nutrients from tri-
butary watersheds based on watershed characteristics. The simplest approach
for estimating NPS nutrient loading is based on relationships between land
use and nutrient loading. Numerous studies have been conducted to determine
the areal loading of nutrients from agricultural, forested and urban water-
sheds. Omernik (1976, 1977) used a statistical approach to evaluate nutrient
export from watersheds as a function of land use. Reviews of areal nutrient
export rates for various land use categories have been assembled by Loehr
(1974), 1!tt.rrvark et al. (1974) and Reckhow et al. (1980).

SCOPE AND OBJECTIVES OF THIS REPORT

This report builds on and updates the earlier work of Brezonik and
Shannon (1971) in assessing nutrient loading-trophic response relationships
and developing critical loading rate that early report, a large volume of data has been collected on several as-
pects of eutrophication in Florida. Several studies of non-point source
nutrient loading have provided data on nutrient export rates for watersheds
in various land uses, and additional data are available on the trophic status,
nutrient budgets and water budgets for many Florida lakes.

During the period in which this additional data has become available
the I/O models for predicting lake trophic status have been developed. While
these models have been found to be quite useful for the temperate zone lakes
in which they were developed, the applicability of the models and the nutrient
loading criteria developed from them has not been evaluated for the warm
temperate and subtropical Florida lakes. Thus, the objectives of this study
are:

1) to examine the relationships between land use and nutrient export in
Florida watersheds;

2) to examine the general limnological characteristics of Florida lakes,
particularly with respect to relationships among trophic status in-
dicators and to compare these relationships to those found in tem-
perate lakes;


- 2 -









3) to assess the applicability of existing I/O models to Florida lakes,
and
4) to develop critical nutrient loading guidelines for Florida lakes.


DATA REQUIREMENTS
Watershed characteristics


PREDICTION
Export of nutrients
from tributary water-
sheds


I I


]


Nutrient inputs from P. Total loading of
direct precipitation nutrients to lake
and point source discharges



Morphologic and hydrologic : Average annual
characteristics of lake concentration of
nutrients in lake


I


Average annual
concentration of
chlorophyll a in lake



Secchi disk
transparency


Figure I-1.


Conceptual framework
status.


for predictions of lake trophic


- 3 -














CHAPTER II, DATA SOURCES AND METHODS


DATA SOURCES


In order to accomplish the objectives of this study, limnological data
from three synoptic studies were compiled into a uniform format for computer
analyses. The first of these studies (Brezonik and Shannon 1971) involved
an evaluation of trophic conditions of 55 lakes in north and central Florida.
Three groups of lakes were included in this study: 1) 16 primarily oligo-
trophic lakes in the Trail Ridge Region of Putnam and Clay counties, 2) 33
lakes and ponds throughout Alachua County and 3) six lakes of the Oklawaha
chain. Samples were collected four times over a one-year period from 36 of
the lakes, while the remaining 19 were sampled at two-month intervals to
obtain more detailed data on seasonal trends.

The second study from which lake data were obtained was the Florida
National Eutrophication Survey (NES), conducted by the EPA during the mid-
1970's. Lakes included in this survey were selected on the basis of three
criteria: 1) lakes impacted by one or more sewage treatment plant outfalls
within 40 km (25 miles) of the lake waters; 2) lakes having a surface area
greater than 40 ha (100 acres); and 3) lakes with residence times of greater
than 30 days. In addition to the lakes that met these criteria, five lakes
of special interest (South Lake, Lake Yale, Glenada Lake and Horseshoe Lake)
were included in the Florida NES, for a total of 40 lakes. Most of the
lakes are located in central Florida, particularly in Polk, Orange, Lake,
Osceola, and Seminole Counties. Each lake was sampled from two to four
times (usually three times) during 1973.

The third set of data is from a recent study of the effects of acid
rain on softwater lakes in Florida (Brezonik et al. 1931), Twelve of the
lakes included in this study are located in the Trail Ridge Region of north
Florida and eight are located on the Isotokpoka Ridge in Highlands County
(south-central Florida). Samples were collected every three months during
1979-80.

Since several lakes were included in more than one survey, only the
more complete data set was used in this report. The resulting data base
contains 101 lakes (Table II-1) whose locations are shown in Figure II-1.
The variables measured in each survey are listed in Table 11-2. In compil-
ing the limnological data set, we used mean values for the entire water
column. This approach is considered valid for Florida lakes, most of which
are generally shallow and do not exhibit seasonal thermal stratification.
Although the sampling methods differed slightly among the three studies from
which data were obtained, these differences are not likely to be important
with respect to the applications of the data in this report. The limnolog-
ical data (annual means) compiled for the study lakes are presented in Appen-
dix A-1; morphological data are presented in Appendix A-2.


- 4 -













Table II-1. Florida study lakes.


County


Lake
Code


Minneola
East Lake Tohopekaliga
Minnehaha
Weohyakapka
Tarpon
Istokpoga
Yale
Kissimmee
Jessie
Horseshoe
Haines
South
Okeechobee

Marion
Crescent
Poinsett
Doctors
Reedy
Gibson
Dora
Talquin
Apopka
Griffin
Glenada
Thonotosassa
Seminole
George
Tohopekaliga
Monroe
Hancock
Eloise
Howell
Banana
Jessup
Alligator
Trout
Lawne
Munson
Effie
Lulu
Santa Fe
Little Santa Fe
Hickory Pond
Altho (Alto)
Cooter Pond
Elizabeth
Clearwater
Hawthorne
Little Orange
Unnamed 10


Lake
Osceola
Orange
Polk
Pinellas
Highlands
Lake
Osceola
Polk
Seminole
Polk
Brevard
Glades, Hendry
Okeechobee
Polk
Flagler, Putnam
Brevard, Orange
Clay
Polk
Polk
Lake
Gadsden, Leon
Lake, Orange
Lake
Highlands
Hillsborough
Pinellas
Putnam, Volusia
Osceola
Seminole, Volusia
Polk
Polk
Orange, Seminole
Polk
Seminole
Columbia
Lake
Orange
Leon
Polk
Polk
Alachua
Alachua
Alachua
Alchua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua


Moss
Jeggord
Still Pond
Lochloosa
Orange
Palatka Pond
Newnan's
Mize
Calf Pond
Unnamed 20
Meta
Alice
Bivens Arm
Clear
Unnamed 25
Beville's Pond
Unnamed 27
Kanapaha
Watermelon Pond
Long Pond
Burnt Pond
Wauberg
Tuscawilla
Harris
Eustis
Weir
Kingsley
Sand Hill (Lowry)
Magnolia
Brooklyn
Geneva
Swan
Wall
Santa Rosa
Adaho
McCloud
Anderson Cue
Suggs
Long
Winnot
Cowpen
Gallilee
Annie
Clay
Francis
Johnson
Josephine
June
Letta
Placid
Sheeler


(1)Lake codes: 1-40 = NES lakes, 41-92 = 55 lakes study (Brezonik &
Shannon 1971), 93-101 = Brezonik et al. (1981).


- 5 -


Lake
Code


Lake Name


Lake Name


County


Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Alachua
Lake
Lake
Marion
Putnam
Clay
Clay
Clay
Clay
Putnam
Putnam
Putnam
Putnam
Putnam
Putnam
Putnam
Putnam
Putnam
Putnam
Putnam
Highlands
Highlands
Highlands
Clay
Highlands
Highlands
Highlands
Highlands
Clay





























































Fig. II-I. Location of 101 study lakes.














6 -








Table 11-2. Data collected in synoptic studies of Florida lakes.




Brezonik & Nation Eutro- Brezonik, et al.,
Shannon, 1971 phication Survey 1980 (1)
Watershed
Drainage area X X X
Land use characteristics X X

Hydrologic & morphologic
Water budget X
Volume X X X
Surface area X X X
Max. depth X X X
Mean depth X X X
Nutrient budget X X

Chemical & physical
Total nitrogen X X X
NH+ X X X
NH3
N04 X X X
NO X X X
Organic N X X X
Total phosphorus X X X
Orthophosphate X X X
COD X
pH X X X
Alkalinity X X X
Acidity X
Color X X
Dissolved oxygen X X X
Specific conductance
Major ions (Ca, Mg, K, Na,
Fe, S04, Cl, Si) X X
Total organic carbon X
Total inorganic carbon X
Trace metals X(2) X3
Secchi disk transparency X X X
Temperature X X X
Suspended solids X
Total solids X
Turbidity X X

Biological
Chlorophyll a X X X
Carotenoids X
Algal identification & counts X X X
Primary productivity X
Zooplankton identification & X
counts
Limiting nutrient bioassays X

Sediments
Sediment type (visual class-
fication) X
Benthic organisms X X
Chlorophyll derivatives X
Total carotenoids X
Volatile solids X
Organic nitrogen X
NH. X
Total phosphate X
Iron X
Manganese X


1Input of nutrients computed from land .use and population
characteristics.

Includes Mn, Cu, Zn, Fe and Sr

(3)Al only.


- 7 -










In addition to the limnological data, the EPA also computed nutrient
budgets for 34 of the 40 Florida NES lakes. Methods used to compute
nutrient budgets are described in NES Working Paper No. 175 (NES 1975). In
this study, non-point source (NPS) loadings of nitrogen and phosphorus for
each tributary were computed by subtracting the reported point source load-
ings from the total tributary loadings. Runoff for each tributary was com-
puted by dividing the total streamflow by the watershed area. Data on the
nutrient and hydrologic budgets for the Florida NES lakes are presented in
Appendix A-3.

Determination of land uses in the NES watersheds was made using Mark
Hurd photoquads and USGS 7.5 minute quadrangles (both 1:24,000). Photos
from the Agricultural Stabilization and Conservation Service and the State
of Florida CITRUS survey were used to provide additional resolution. The
land use classification scheme used was a modification of the system devel-
oped by Anderson et al. (1976). The modifications involve classification
of agricultural land, forests and barren land (Fig. 11-2). Agricultural
land was divided into two categories: "cropland and pasture" and "other
agriculture". The category "other agriculture" was formed to assess the
effects of different types of agricultural land use on NPS nutrient loading.
This category is comprised largely of citrus orchards in the NES watersheds.
The second modification was a combining of "deciduous forest land", "ever-
green forest land" and "mixed forest land" (categories 41 to 43 in Anderson's
Level II scheme) with "forested wetland" (category 61) into a composite
"forest" category. This was done to facilitate identification of forests
in the air photos. Finally, "salt flats", "barren land", "beaches", "other
sandy areas", "transitional areas" and "mixed barren land" (categories 71-
74 and 76-77) in Anderson's Level II scheme were grouped together with non-
residential urban land uses into an "other urban" land use category. This
modification was made because these barren land subgroups were largely com-
prised of "transitional areas" associated with urban areas.

The watersheds used in the evaluation of land use-nutrient loading
relationships are listed in Table 11-3. Several NES watersheds were not
included in the analyses because runoff from them was found to be abnormally
high (> 70% of total precipitation in the watershed). Such high values
could result from groundwater inflows from other watersheds or backpumping
in agricultural areas; or Alternatively, they may simply reflect errors in
flow measurement. Since it was impossible to establish the cause of high
runoff for individual watersheds in this study, all watersheds in which run-
off was > 70% of precipitation were excluded from the analyses. This re-
sulted in the deletion of watersheds 13B1, 18B1, 22C1, 37D1, 32B1 and 34G1.
The 70% criterion was considered a reasonable upper limit for natural runoff
from Florida watersheds (W. C. Huber, per. comm., 1980). Several other
watersheds were excluded because flow data were not available (31A1, 40B1)
or because the measured drainage basin area differed substantially from
that reported by the NES (17B1). Thus, data from 41 watersheds were used
in the analyses of land use-nutrient loading relationships. Land use data
and data for non-point source nitrogen and phosphorus loadings for these
watersheds are presented in Appendix A-4.


- 8 -












Anderson, et al. (1976)


Level II


11. Residential
12. Commercial and
Services
13. Industrial
14. Transportation,
communication &
utilities
15. Industrial & com-
commercial complexes
16. Mixed urban or built
up land
17. Other urban built up
land


la. Residential


-- lb.


Agricultural
land


21. Cropland & pasture --2a.
22. Orchards, vineyards,--
nurseries & ornamental ---2b.
23. Confined feeding
operations
24. Other agricultural
lands


Cropland & pasture

Other agriculture


31. Herbaceous --3. Rangeland
rangeland
32. Scrub & brush
rangeland
33. Mixed rangeland

41. Deciduous forest land --4. Forest----
42. Evergreen forest land
43. Mixed forest land

51. Streams & canals -- 5. Water
52. Lakes
53. Reservoirs
54. Bays & estuaries


61. Forested wetland
62. Nonforested wetland -_ 6.


Nonforested wetland


Barren land


Figure 1-2.


71. Dry salt flats
72. Beaches
73. Other sandy areas
74. Bare exposed rocks
75. Strip mines, quarries
& gravel pits ------7a.
76. Transitional lands
77. Mixed barren land____


Land-use classification system.
9 -


Level I

Urban or
built-up
land


Other urban -------


Forest


Water


Wetland


Strip mines


This study


















Table 11-3. NES watersheds used in analysis of land
use-nutrient loading relationship.


Tributary
Code


Tributary


Lake


Unnamed
Boggy Cr.
Unnamed
Unnamed
Tiger Cr.
South Cr.
Arbuckle Cr.
Jackson Canal
Unnamed
Unnamed
Taylor Creek
Lembin Creek
Indian Prarie
Canal
Harney Pond
Canal
Unnamed Cr.
Haw Creek
Unnamed
Swimming Cr.
Unnamed
Unnamed
Dora Canal
Unnamed
Ockawaha Cr.
Bear Cr.
Unnamed
Dead River
Baker Creek
Unnamed
Bayou Cr.
Shingle Cr.
Partin Canal
Bethyl Cr.
Saddle Cr.
Unnamed
Unnamed
Unnamed
Howell Cr.
Unnamed
Gee Cr.
Soldier Cr.
Unnamed
Howell Cr.
Salt Cr.
Sweetwater Cr.
Unnamed Cr.
Unnamed Cr.
Unnamed Cr.
Unnamed Canal
Unnamed Canal
Unnamed Canal
Unnamed
Unnamed


14 D1
15 Bl
15 Cl
17 Bl*
18 Bl*
19 Bl
20 Al
20 Bl
21 Bl
21 Dl
22 Cl*
23 Al
25 Al
26 Bl1
26 Cl
28 Al
28 B1
29 B1
30 A2
30 B1
30 Cl
31 Al*
32 Al
32 Bl*
34 Al
34 Bl
34 Cl
34 Dl
34 El
34 Gl*
35 Al
35 Bl
35 Cl
37 Bl
37 Cl
37 Dl*
38 B1
40 Bl*


Minneola
East L. Tohopekaliga

Minnehaha
Weohyakapka
Tarpon
Istokpoga
Kissimmee
Jessie
Okeechobee
Okeechobee
Okeechobee

Okeechobee

Okeechobee
Marion
Crescent
Crescent
Doctors
Reedy
Gibson
Dora
Dora
Talquin
Talquin
Apopka
Griffin
Thonotosassa
Seminole
Seminole
Tohopekaliga
Tohopekaliga
Monroe
Hancock
Hancock
Hancock
Eloise
Howell
Howell
Jessup
Jessup
Jessup
Jessup
Jessup
Jessup
Alligator
Alligator
Alligator
Lawne
Lawne
Lawne
Munson
Lulu


*Watershed dropped from analyses.


- 10 -


See text,











STATISTICAL METHODS


Statistical analyses were performed using the Statistical Analysis
System (SAS) (SAS Institute, 1979). Where multiple independent variables
were used in regression analyses, the STEPWISE (backward) procedure was
used to select the significant independent variables. The GLM procedure
was used to compute predicted values and confidence limits for final regres-
sion analyses. Confidence limits for the mean (CLM) were used when the
objective of the regression analysis was to predict the mean response of
the dependent variable. Confidence limits for individual predictions (CLI)
were used when the objective of the regression model was to predict values
of the dependent variables from individual values of the independent vari-
able (Snedecor and Cochran 1967).

In evaluating the nutrient loading models, several lakes were discarded
as the result of suspected errors in the data or on the basis of a statist-
ical outlier test. The test procedure used was the criterion Tn, defined as:

Tn = obs pred /s

where Yobs = observed y value,

Ypred = predicted y valueand

s = population standard deviation excluding the outlier.

If the resulting T value of a suspected outlier was greater than the
5% T value from a table of critical T values (Grubbs 1969), then the value
n
was considered an outlier. Although tle method is recommended for removing
only a single outlier, several outliers can be removed by reevaluating s
following the removal of each outlier. In this manner, up to three outliers
were removed from any one prediction equation. The removal of a value as
an outlier in one equation did not automatically result in its removal from
other equations.


- 11 -














CHAPTER III. PREDICTIONS OF NUTRIENT LOADING FOR FLORIDA LAKES



INTRODUCTION


In studies of lake eutrophication it is often necessary to know the
loadings of plant nutrients, particularly nitrogen and phosphorus, into a
lake. Ideally, these loadings are determined by obtaining data on water
fluxes and nutrient concentrations for all sources (tributary inflows, direct
point source discharges and direct precipitation). Unfortunately, the ac-
quisition of these data is expensive and time-consuming, usually requiring
at least one year of study. It is therefore desirable to be able to esti-
mate the loadings of nutrients from various sources indirectly.

The most difficult component of a lake's total nutrient loading to
determine is the non-point source loadings from the surrounding watershed.
In this chapter the major emphasis will be to develop a method of estimating
non-point source nutrient loadings from tributary watersheds using land use
data. The approach used, originally conceived by Lee et al. (1966), is based
on the assumption that nutrient export from a portion of a watershed can be
computed as the product of the drainage area and an export coefficient that
is determined by land use. Thus for a watershed composed of multiple land
uses:
n
L. = i i.. A. (3-1)


where L. = loading of constituent i, kg/yr,

K.. = export coefficient for constituent i from land use j,
kg/ha-yr, and

A. = area of watershed in land use j, ha.

Since this concept was first developed, dozens of studies have been
conducted to determine export coefficients for various land uses. These
have been compiled by Uttormark et al. (1974), Loehr (1974) and, most recently,
by Reckhow et al. (1980). Unfortunately, the range of reported export
coefficients for a given land use varies widely. This variability is ex-
pected since the approach does not take into account variations in precipi-
tation, soil type, length of the growing season and other parameters that
influence nutrient export. Furthermore, most of the studies included in
these compilations have been conducted in temperate climates, limiting the
validity of these coefficients for Florida watersheds.

In order to determine suitable land-use export coefficients for Florida,
two methods were used. First, data from studies of non-point source nutrient
loading conducted in Florida were compiled by land use. By examining only


- 12 -










studies that have been conducted in Florida, variability in export coeffi-
cients caused by differences in soil type, climate and terrain should be
reduced, resulting in a narrower range of export coefficients for each land
use. For studies in which export coefficients were not computed, the re-
ported loading data were used to perform the appropriate calculations. In a
few cases, data are included for several studies conducted in nearby states,
where land uses and other conditions (soils; climate) were judged to be re-
presentative of conditions in Florida.

The second method used to compute export coefficients was a statistical
approach similar to that used by Omernik (1976). In this approach, export
coefficients are determined by multiple regression techniques using data on
nutrient loading (dependent variable) and land uses (independent variables).
For this analysis, data on nutrient loading, nutrient concentration, runoff
and land use were compiled for 41 Florida NES watersheds (Chapter 2).


LITERATURE REVIEW: NUTRIENT SOURCES TO FLORIDA LAKES

Point Source Loadings. The results of a nationwide survey of nu-
trient loadings from 809 municipal wastewater treatment plants (Gakstatter
et al. 1978) may be used to estimate treatment plants loadin:- of these
nutrients for preliminary studies on the basis of population served and treat-
ment type (primary, trickling filter, activated sludge, stabilization pond).
As seen in Table III-1, the type of treatment has little effect on the phos-
phorus loading from wastewater treatment plants: median loadings ranged from
0.9 kg/cap-yr for stabilization ponds to 1.1 kg/cap-yr for primary treatment
facilities. These results compare favorably with those of Vollenweider (1968),
who computed a mean phosphorus loading of 0.8 kg/cap-yr for municipal waste-
water from the results of 15 studies reported in the literature (treated and
untreated wastewaters were included). Although the type of treatment had
little effect on phosphorus loadings, phosphorus loadings in the compilation
of Gakstatter et al. were significantly lower for a group of 33 plants that
included tertiary phosphorus removal processes (median loading = 0.4 kg/cap-
yr) and for a group of 25 convential treatment plants located in communities
having phosphorus detergent bans.

The results of Gakstatter et al. indicate that nitrogen loading from
wastewater treatment plants is affected by the type of treatment process used.
Loadings ranged from 2.0 kg/capita-yr for stabilization pond effluents to
4.2 kg/cap-yr for primary treatment plant effluents. In comparison, Vollen-
weider (1968) reported a nitrogen loading rate of 3.9 kg/cap-yr. For all four
treatment processes, the median effluent TN:TP ratios were less than 5:1,
indicating that sewage effluents typically are nitrogen limited.

Although the standard errors for loading estimates in Table III-1 indi-
cate that loadings from municipal wastewater treatment plants can usually be
estimated with reasonable accuracy, actual loadings for a given plant may
differ significantly from predicted values because of (1) modifications in
the design process, (2) hydraulic overloading resulting from stormwater in-
flows, (3) excessive infiltration or overuse, or (4) impairment of the treat-
ment process by toxic wastes or improper operation. Thus, while the values
presented in Table III-1 may be used for preliminary estimates of nutrient


- 13 -









Table III-1.


Median and mean phosphorus and nitrogen concentrations and
median loads in wastewater effluents following four conven-
tional treatment practices(1)


Treatment Type
Trickling Activated Stabilization
Primary Filter Sludge Pond

Number of Sampled Plants 55 244 244 119

Total Population Served 1,086,784 3,459,983 4,357,138 270,287

Ortho-P Conc. Median 3.5 + 0.29* 5.1 + 0.21 4.6 + 0.24 3.9 + 0.34
(mg/1) Mean 4.0 + 0.62 5.4 + 0.38 5.3 + 0.40 4.8 + 0.62

Total-P Conc. Median 6.6 + 0.66 6.9 + 0.28 5.8 + 0.29 5.2 + 0.45
(mg/1) Mean 7.7 + 1.19 7.2 + 0.50 6.8 + 0.51 6.6 + 0.81

Total-P Load Median 1.1 + 0.10 1.2 + 0.05 1.0 + 0.06 0.9 + 0.10
(kg/cap-y)

Inorganic-N Conc. Median 6.4 + 1.00 7.1 + 0.38 6.5 + 0.45 1.3 + 0.29
(mg/1) Mean 8.3 + 1.40 8.2 + 0.60 8.4 + 0.69 5.5 + 1.95

Total-N Conc. Median 22.4 + 1.30 16.4 + 0.54 13.6 + 0.62 11.5 + 0.84
(mg/1) Mean 23.8 + 3.48 17.9 + 1.23 15.8 + 1.16 17.1 + 3.59

Total-N Load Median 4.2 + 0.40 2.9 + 0.17 2.2 + 0.15 2.0 + 0.26

TN:TP Ratio Median 3.4 2.4 2.4 2.2

Per Capita Flow Median 473 + 72 439 + 19 394 + 26 378 + 38
(1/cap.d)


* Value + 1 standard error.

(1) From Gakstatter et al. (1978).










loadings from wastewater treatment plants, they should not be regarded as
substitutes for actual measurements when expensive management decisions are
made.

Precipitation Inputs. For many lakes, particularly seepage lakes with
long detention times, bulk precipitation (wetfall + dryfall) may be a major
source of nutrients. A recent study on the chemical composition of bulk
precipitation in Florida (Brezonik et al. 1981) includes data on nitrogen and
phosphorus loadings for 24 sites throughout Florida, providing a data base
that can be used to estimate precipitation loadings to lake surfaces (Table
III-2).

The mean deposition rate for nitrogen was 0.76 g N/m2-yr, but the depo-
sition rates at individual stations varied considerably. The lowest nitro-
gen deposition rate, 0.32 g N/m2-yr, occurred at Bahia Honda Key, while the
highest rate, 1.13 g N/m2-yr, occurred at Belle Glade in the intensively
cultivated Everglades Agricultural Area. When sampling stations were grouped
according to location and local land use, deposition rates were found to be
lowest at the coastal and non-agricultural sites and highest at the agri-
cultural sites (Table III-2). On a statewide basis, 69% of the total nitro-
gen deposited was in the form of inorganic species, and speciation generally
followed the sequence NH >NO3> organic N.

Phosphorus deposition was also site dependent and ranged from 17 mg
P/m2-yr at Bahia Honda to 111 mg P/m2-yr at Jasper, with a mean of 51 mg
P/m2-yr. As with nitrogen deposition, phosphorus deposition was generally
lowest at the coastal and non-agricultural rural sites and highest at agri-
cultural sites. Soluble reactive phosphorus was the dominant species, ac-
counting for 68% of the total phosphorus in bulk precipitation.

The significance of nutrient inputs from precipitation with respect to
lake eutrophication can be evaluated by comparing the magnitude of these
loadings with critical loading values. For example, using Vollenweider's
(1968) original critical loading criteria, a lake with a mean depth of 3 m
has a critical loading of 97 mg P/m2-yr and 1.45 g N/m2-yr. According to
these criteria, the mean precipitation loading for Florida corresponds to
53% of the critical loading for phosphorus and 52% of the critical loading
for nitrogen. Thus, precipitation inputs of nutrients may be a significant
component of the total nutrient budget for a lake.

_- -.int source loadings from. 1r-ida watersheds. Export coefficients
for nitrogen and phosphorus determined for Florida watersheds are presented
by land-use category (urban residential, agricultural, forest) in Tables III-3 to
III-6. For comparison, the results of a literature review of export co-
efficients for studies conducted throughout the U.S. and Canada (Rdckhow
et al. 1980) are presented.

Before discussing the magnitude of these export coefficients, several
comments are in order concerning the nature of these studies. First, the
methods used to determine nutrient loading are highly variable among investi-
gators, and some are based on rather limited data. For example, Lamonds
(1974) based his estimates of N and P export from a residential area in
Eustis, Florida, on data collected during only seven storms, while Wanielista


- 15 -














Table 111-2.


Atmospheric deposition (via bulk precipitation)
of total nitrogen and total phosphorus at sta-
tions grouped according to dominant land use in
the area. (1)


TN
kg/ha-yr


TP
kg/ha-yr


Coastal 5.8 0.31

Urban 7.6 0.50

Rural (non-agricultural) 6.2 0.27

Rural (agricultural) 8.8 0.66

State average 7.5 0.51


(1) From Brezonik et al. (1981)


- 16 -
















Table III-3. Nutrient export from urban areas.


Source


Location


Loading (kg/ha-yr)
Total Ortho Total N Org.


NO-N


NH, -N


Comments


P P N -

Burton and Nr. L. Jackson 7.49 0.19 0.37 0.18 0.17 80% residential &
Turner (1977) Fla. (25%) (inorg.) (0.02 NO) commercial
423 samples analyzed

Wanielista Orlando, Fla. 3.5 2.0 10 6 Commercial area.
(1977) 57% (TKN + NO3) Only two storms sam-
pled results ex-
trapolated.

Miller et al. Ft. Lauderdale, 0.26 0.11 2.88 2.12 0.55 0.10 97.9% impervious
(1979) Fla. 42% (74%) (.042 N02) (3.5%) area 31 storms sam-
(19.1%) pled. Loadings calc.
from raw data.






















Table 111-4. Nutrient export from residential areas.

Loading (kg/ha-yr)
Source Location Total P Ortho-P Total N Organic NO -N NH -N Comments
N

Burton and Vicinity of 4.74 0.09 0.58 0.16 Area is lightly de-
Turner (1977) L. Jackson, (1.9%) (.03 NO2) veloped but includes
Fla. 10% under highway
eenst. Stream re-
ceives package plant
eff. from school.

Wanielista et Near Orlando, 2.24 0.80 3.98 2.17 Loading est. extra-
al. (1977) Fla. (35.7%) (TKN + NO3) polated from data on
two storms.

Bedient et al. Houston, 0.745 0.29 Residential devel-
(1978) Texas opment included new
construction. Clayey
soils.

Mattraw & Broward Co., 0.21 1.48 Uniform single family
Sherwood Fla. dwellings. Only 5-
(1977) 10% of rainwater col-
lected as runoff.

Lamonds (1974) Eustis, Fla. 0.82 7.36 Flow estimated only
7 storms sampled.

















Table III-5. Nutrient export from agricultural areas.


Loading (kg/ha-yr)
Source Location Total P Ortho-P Total N Org. N NO3-N NH -N Comments

Campbell (1978) Alachua Co.
1975-76 1.34 1.21 6.36 5.30 -0.37 0.68 Land in intensive crop pro-
1976-77 0.86 0.63 2.10 1.92 0.09 0.09 duction w/some pasture near
stream. 3-8% slope, sandy
soil w/claypan at 1-2 m.

Burton & Turner Near L. 0.51 0.14 0.21 0.07 0.14 48% agricultural (52%
(1977) Jackson, Fla. ( inorg.) (.004 NO2) forested). 371 samples
analyzed.

Stewart et al. Upper Taylor
(1978) Creek 1% cropland 59" improved
W-3 1972 1.48 0.58 pasture" 30% range &
1974 2.80 0.44 forest 10% misc.
1975 0.65 0.16

W-5 1972 0.38 0.49 70% improved pasture
1974 1.61 0.42 20% range & forest
1975 0.53 0.20 10% misc.

W-13 1972 6.11 1.36 78% improved pasture
1974 12.73 3.16 21% dairy operations
1975 1.94 0.30

Ritter et al. Delaware
(1979) Coastal Plain 1359 ha. watershed 45% crop
Stockley 0.68 0.083 20.6 7.57 12.41 0.63 (soybean, corn, grain) 47%
Branch Forest 4% urban. Sandy
loam soils, low slope.

Blackwater 0.48 0.078 18.2 6.96 10.78 10.07 1456 ha. 57% cropland (as
Creek above). 37% forest, 2%
urban sandy loam soils.




















Table 111-5. Nutrient export


Source


Location


from agricultural areas (cont'd)

Loading (kg/ha-yr)
Total Ortho-P Total Org.
P N N


NO -N NH-N
3 4


Asmussen, Little River,
et al. (1979) Georgia
1975 0.141 0.24 Cropland = 36.8%. Forest, swamp, etc.
1976 0.145 0.14 %60%

CH2MHill Everglades All areas backpumped & irrigated.
(1979) Ag. Area Organic soils mean of 3 sites.
Sugarcane 0.65 27.18

Vegetable farm 2.37 38.86 Mean of 3 sites.

Cattle ranch 0.55 12.36 Mean of 2 sites.

Lutz (1977) Everglades Ag. Note: drainage areas poorly defined
Area. in EAA
S-5A 0.77 50.1 88% agricultural (71.2% truck crops,
8.9% pasture)

S-6 0.58 27.0 93% agricultural (36% sugarcane,
15% truck farming, 45% pasture.

S-7 0.41 30.3 78% agricultural (69% sugarcane, 9%
pasture). 22% forest & wetland.

S-8 0.75 32.3 30.5% agricultural (26% sugarcane).
69.3% forest and wetlands.


Comments


























Table 111-6. Nutrient export from forested areas.

Loading (kg/ha-yr) +
Source Location Total Ortho Total Org. NO -N Ni -N Comments
P -P N N

Campbell (1978) Alachua Co. Native forest w/small amount of crop-
1975-76 0.33 0.30 1.43 1.21 0.12 0.11 land. Sandy soils w/claypan, 0-3%
1976-77 0.68 0.52 1.65 1.49 0.09 0.07 slope.

Bedient et al. Houston, 0.21 0.29 Heavily forested, clayey soils, 0.1%
(1978) Texas slope. 25 storms sampled.

Reikerk, et Bradford 0.4 0.2 6.1 5.5 0.1 0.5 Data collected from 3 coastal plain
al. (1978) Co., Fla. (inc. NO2) flatwoods areas for 1 water year.
Mean values shown here.

Duffy et Northern 0.30 0.029 Data collected for storm events only
al. (1978) Mississippi (no base flow between storms). Five
watersheds sampled. Pine (mix) forest;
loess soils.










et al. (1977) computed export coefficients for several watersheds in the
Orlando area using data collected during only two storms. In the most care-
ful studies (including Riekerk et al. 1978; Campbell 1979; Ritter et al.
1979; Burton et al. 1977) data on baseflow concentrations were collected on
a regular basis (usually weekly), and stormflow events were sampled at fre-
quent intervals using an automated sampler. Concentration data were then
combined with continuous flow measurement data to produce relatively accurate
estimates of loading. In applying export coefficients reported in the
literature, the reader should consider the quality of the methods used to
compute nutrient loadings.

A second weakness of many of these studies is that water budgets were
not constructed. Because of this, it cannot be certain that the streamflow
at the sampling station represents the sole outlet of water (and nutrients)
from the watershed, although this is generally assumed. Conversely, unless
a water budget is completed, it cannot be assumed that all of the water com-
ing from a watershed results from precipitation falling on that watershed
rather than from seeps and springs whose flow may originate from outside the
boundaries of the watershed in question. In order to develop reliable nu-
trient export coefficients as a function of land-use, it is important that
all inflows and outflows to the watershed are determined; this is particular-
ly true in Florida where the groundwater table is fairly shallow and springs
are common.

Finally, only a few of the studies reported here involved more than
one annual cycle, and therefore they did not evaluate annual variations in
loading. The data of Campbell (1978), Stewart et al. (1978) and Asmussen
et al. (1979) illustrate the extent of variation that may occur from year
to year (Table 111-5). Campbell observed decreases of 36% and 67% for total
P and total N loads, respectively, between the 1975-76 sampling period and
the 1976-77 sampling period. Most of the difference in loading between the
two annual periods occurred as the result of a decrease in streamflow rather
than changes in the concentrations of constituents. For the Upper Taylor
Creek watershed, the highest annual export rate of orthophosphate was 5.5
times the lowest export rate, and the highest export rate of NO3-N was 6.1
times the lowest export rate over a three year period (Stewart et al. 1978).
Changes in management practices, as well as variations in precipitation,
were considered to be responsible for the observed fluctuations in loading.
Asmussen et al. (1979) reported relatively constant orthophosphate loads but
a 2X variation in the nitrate load over a two year period for the Little
River watershed in southern Georgia. As can be seen from these examples,
the magnitude of the variations in loadings can be substantial, at least for
agricultural watersheds. Studies cited by Reckhow et al. (1980) indicate
that there are also substantial annual variations in the loadings from forested
watersheds. The extent of variations in nutrient loadings observed in
multiple-year studies suggests that for critical applications, calculations
of nutrient loadings should be based on data acquired over the period of
several years. Furthermore, it can be concluded that the variation in ex-
port coefficients reported for a given land use is attributed in part to
temporal variability.


- 22 -










The range of export coefficients for N and P reported in literature for
various land uses is shown in Figures III-1 and III-2, together with export
coefficients determined from the statistical analysis (to be discussed in
the following section). The range of phosphorus export coefficients reported
for Florida watersheds falls within the range of coefficients found by
Reckhow et-al. (1980) for.all three major land uses., although the range of
phosphorus export coefficients for urban areas in Florida is near the lower
end of the range reported by Reckhow and coworkers. The range of nitrogen
export coefficients found for Florida watersheds also falls within the range
of values reported by Reckhow et al. (1980) for all three major land uses.
Limited data on speciation of nitrogen in runoff from agricultural and
forested watersheds (Fig. III-3) indicates that while the majority of nitro-
gen in runoff from forested watersheds is in the organic form, both nitrate
and organic nitrogen are important fractions in the runoff from agricultural
wathersheds.

It is tempting to conclude that the narrower range of each land-use ex-
port coefficient for Florida watersheds compared to the range reported by
Reckhow et al. (1980) reflects the greater similarities in climate, soils,
and topography among the Florida watersheds. However, it is also possible
that the narrower ranges of export coefficients for Florida watersheds simply
reflects the fact that fewer studies have been conducted for Florida water-
sheds. Unfortunately, a statistical analysis of the variation in export
coefficients reported for Florida watersheds cannot be conducted because
only a few values have been reported for each land use.

The values of export coefficients reported for each land use in Florida
watersheds may vary by more than an order of magnitude for most land uses.
Thus, phosphorus export coefficients range from 0.26 to 7.49 kg/ha-yr for
urban areas, 0.21 to 4.74 kg/ha-yr for residential areas, 0.41 to 2.37 kg/ha-
yr for agricultural areas, and 0.21 to 0.68 kg/ha-yr for forested watersheds.
Nitrogen export coefficients exhibit similar variability: 0.37 to 2.88
kg/ha-yr for urban areas, 1.48 to 7.36 kg/ha-yr for residential area, 2.1 to
50.1 kg/ha-yr for agricultural areas and 1.43 to 6.1 kg/ha-yr for forested
areas. These broad ranges of values limit the accuracy of loading estimates
that can be made using a literature-base approach. A further disadvantage
of this approach is that the accuracy of predictions cannot be evaluated
using statistical analysis.


STATISTICAL ANALYSIS OF NUTRIENT EXPORT
FROM THE NES WATERSHEDS.

An alternative approach in predicting NPS nutrient export from water-
shed land-use characteristics is to use multiple regression techniques
whereby land-use characteristics (the independent variables) are used to
predict nutrient export (the dependent variable). In this study, data on
nutrient export, flow-weighted nutrient concentrations, flow, and land-use
characteristics were compiled for 41 NES watersheds, as described in Chapter
II. These data were used to develop statistically significant regression
relationships between land use and NPS nutrient loading, land use and flow-
weighted nutrient concentration, and land use and flow.


- 23














Urban
(3)
SI Urban
(5)
I ~1 Residential
(41)
URBAN i-- ALLURB


0.08o Mixed agricultural
Row crops
Non- row crops
'- Grazing and Pasture
S--1 Mixed agricultural
(41) 1 CPAST
AGRICULTURAL
0.02 MMW 1 ^ --- Reckhow et al.
(1980)
(5)
S (5 IL J Florida studies
H---- Mean 8 90% CI
for NES watersheds
Number in parenthesis indicates
FOREST number of watersheds
I I I i i


0.1 0.5 1.0 5 10
Phosphorus Export Kg/ha-yr
Figure III-1. Phosphorus export from urban, agricultural, and forested watersheds.


50 100










Urban
U (2)
R Urban I I
B
A (3)
N Residential I

A Mixed agriculture
R Row crops
I Non- row crops
U Grazing and Pasture
L Mixed agriculture I I

R PAST
E
F _.,_,_ _Forest__
0 Reckhowet ol. Forest
R (1980) (3)
E I Florida studies Forest
S I-- Mean & 90%
CI for NES watersheds


0.5


1.0 5 I
Nitrogen Export Kg/ha- yr


IUU


Figure III-2. Nitrogen export from urban, agricultural, and forested watersheds.


50 IC A


v


A













ORGANIC N


NO, + NO,
(3)

NHS
(3)


A ORGANIC N (4)
G
R
C NO + NOS
U (4)

T
U NHz
RAI
E I 1
i | Iil


30 40
Per Cent


50
of Total


70 80


Figure 111-3. Range of nitrogen fractions in runoff from forested and agricultural watersheds.










In order to determine which land uses were statistically significant
as independent variables in these regression equations, the backward elimi-
nation procedure (STEPWISE/B) of SAS was used. In this procedure, all po-
tentially significant independent variables (in this case, area in each
land use) are entered in the regression equation. The least significant
variables are deleted in sequential steps until all remaining independent
variables are found to be statistically significant in contributing to the
variability of the dependent variable. For these analyses, the 0.05 level
of significance was used as the criterion for selecting independent vari-
ables for the final regression equations. The 95% CLI for the final regres-
sion equations were determined using the GLM procedure of SAS.

Prior to performing the regression analyses, a correlation matrix of
the land use categories was computed, and it indicated that the categories
"other urban" (OURB) and "residential" (RES) were significantly correlated
(r2 = 0.25). In order to avoid the problem of multicollinearity among in-
dependent variables (Neter and Wasserman 1974), these two categories were
combined into an "all urban" (ALLURB) category. Thus, eight land use cate-
gories were initially included as independent variables in the regression
equations: ALLURB, "crops and pastureland" (CPAST), "other agriculture"
(OAG), "forest" (FOR), "rangeland" (RA), "non-forested wetland" (FOR),
"open water" (WA) and "strip mine" (SMINE) (see Figure II-2). The. land-use
characteristics of the 41 NES watersheds used in these analyses are shown
in Table III-7.

The utility of using land-use areas as independent variables in these
equations was evaluated by also using total drainage area (DA) as the sole
independent variable in equations to predict NPS nutrient loading, nutrient
concentration, and flow. Regression results are summarized in Table III-8.

Equations to Predict Phosphorus Loading (TPL). The best equation to
predict TPL, eq. 3-3, has an rZ of 0.71 (P >0.0001) and includes three land
use areas as statistically significant independent variables (ALLURB, CPAST,
and RA). The relationship between DA and TPL (eq. 3-2) was much weaker (r2 =
0.21), indicating that the use of individual land use areas results in much
better predictions of TPL than does the use of DA alone (Table 111-8).

The 0 value (phosphorus export coefficient) of ALLURB in eq. 3-3,
6.0 + 1.4 kg/ha-yr, is high compared with phosphorus export coefficients
reported for urban watersheds in other studies (Figure III-1). One likely
explanation for this is that ALLURB is a well-defined and fairly restrictive
land-use category (See Figure II-1), while the "urban" watersheds in other
studies often included areas of forest and other land uses that tend to de-
crease the overall phosphorus export. The 3 value for CPAST in eq. 3-3,
1.2 + 1.0 kg/ha-yr, falls within the range of phosphorus export coefficients
reported for other agricultural watersheds in Florida and throughout North
America (Figure III-1).

The B value for RA in eq. 3-3, (-1.4 kg/ha-yr) is a statistical anomoly
that reflects an inherent weakness of the regression approach in evaluating
nutrient export. However, since RA comprised little of the average water-
shed area (8%), the magnitude of its coefficient has relatively little ef-
fect on the predictions produced using eq. 3-3. In contrast, CPAST comprised


27 -











Table 111-7. Land use characteristics of study watersheds.


Land Use (%)(1)


Other urban

Residential

Total urban (2)

Crop and pasturaland

Other agriculture

Forest

Range

Non-forested wetland

Open water

Strip mine

Drainage Area (km 2)


Computer
Code

OURB

RES

ALLURB

CPAST

OAG

FOR

RA

NFWET

WA

SMINE

DA


Mean Minimum Maximum


8.2

18.5

26.7

19.6

11.2

19.2

8.3

7.6

5.7

0.9

135


0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.4


32.5

62.8

78.5

70.0

60.5

93.5

50.6

46.1

33.3

25.4

978


(1) Land uses defined in

(2) ALLURB = OURB + RES.


Chapter II.


- 28 -



















Table III-8. Regression equations
41 NES watersheds.

Equation Dependent Intercept


to predict nutrient loading, nutrient concentration and flow for


Independent


6


n r P > F


variable Estimate Std. Error variables Estimate Std. Error

3-2 TPL 7123 3663 DA 0.48 14.6 41 0.21 0.002
(kg/yr) (ha)

3-3 TPL 1705 2373 ALLURB 6.0 0.7 41 0.71 0.0001
(kg/yr) CPAST 1.2 0.5
RA -1.4 0.7
(ha) (kg/ha-yr)

3-4 TNL 8235 8897 DA 5.1 0.4 41 0.84 0.001
(kg/yr) (ha) (kg/ha-yr)

3-5 TNL 15145 8179 CPAST 4.2 1.2 41 0.87 0.001
(kg/yr) NFWET 16.1 1.8
WA 16.0 3.4
(ha) (kg/ha-yr)


0.36


0.06


ALLURB
RA
(ha)


1,314,000 3,794,000


2,765,000 2,978,000 ALLURB
FOR
RA
CPAST
OAG
(m2)


5.6 x 10-5 1.8 x 10-
-2.6 x 10-5 1.2 x 10-5


0.31
(m/yr)

0.25
0.55
0.49
0.15
0.75
(m/yr)


41 0.24 0.005


0.02 41 0.91 0.0001


0.09
0.04
0.08
0.06
0.09


41 0.96 0.0001


3-6


TOTAL P
(mg/L)



FLOW
(m3/yr)


FLOW
(m3/yr)










an average of 19.6% of the total watershed area, while ALLURB comprised an
average of 26.7% of the total watershed area.

The best predictive equation found to relate the concentration of total
phosphorus (TOTAL P) to land use is eq. 3-6, which includes only ALLURB and
RA as independent variables. The relatively low r2 for this equation (0.24,
P> .005) suggests that TOTAL P is not greatly affected by land use in these
watersheds. The strong correlation found between TPL and land use thus sug-
gests that variations in flow rather than concentration are responsible for
variations in TPL among watersheds.

The mean non-point source loading of phosphorus for all 41 watersheds
is 1.0 kg/ha-yr. Since precipitation contributes only 0.3 to 0.7 kg P/ha-yr
to Florida watersheds (Table III-2), it can reasonably be concluded that
there is a net addition of phosphorus to the runoff water from within the
watersheds.

Predictions of Nitrogen Loading (T'L). Unlike the situation for TPL,
there is a strong correlation between TNL and DA (r = 0.84, P> 0.001), in-
dicating that drainage area alone is a reasonably good predictor of non-
point source nitrogen loading (eq. 3-4). When the areas in individual land
uses were used as independent variables, three terms (CPAST, NFWET and WA)
were found to be significant at the 0.05 level, producing an equation (eq.
3-5) with an r2 of 0.87 (P> 0.001). Thus, the use of individual land use
areas as independent variables rather than total watershed area alone con-
tributes little toward improving predictions of TNL.

Two of the significant terms in eq. 3-5 are NFWET and WA, both of
which have nitrogen coefficients near 16 kg/ha-yr. The magnitude of these
export coefficients is high compared to the mean input of nitrogen from pre-
cipitation (7.5 kg/ha-yr). The high loadings for open water and non-forested
wetland may be caused by inputs from surrounding land uses, such as loadings
from lawn fertilization and septic tanks associated with shoreline develop-
ment. Alternatively, the high reported non-point source loadings for these
land uses may reflect inaccuracies in the computation of non-point source
loadings.

The only other significant term in eq. 3-5 is CPAST, which has a nitro-
gen export coefficient of 4.2 + 2.4 kg/ha-yr. It can be seen (Figure III-2)
that this 95% C.I. for nitrogen loading from agricultural areas is within
the lower end of the range of values reported for agricultural watersheds
in Florida (Table III-5) and throughout North America.

The mean non-point source loading for nitrogen in all 41 study water-
sheds is 5.7 kg/ha-yr., which is somewhat lower than the mean precipitation
loading of 7.5 kg/ha-yr reported by Hendry et al. (1981) and Brezonik et
al. (1981). Thus it appears that there is a net accumulation of nitrogen in
these watersheds.

Equations to Predict Flow (FLOW). As shown by eq. 3-7, FLOW is highly
correlated with DA (r2 = 0.91, P> 0.0001). The P value for DA in eq. 3-7,
0.31 m/yr, is about 24% of the mean annual precipitation for the NES water-
sheds. The use of individual land use areas rather than DA improves the


- 30 -










prediction of flow (eq. 3-8) only slightly (r = 0.96, P> 0.0001). The 3
values in eq. 3-8 are runoff coefficients (m/yr), which are significant at
the 0.05 level for five land uses.


APPLICATION

The accuracy of predictions fbr nutrient loading and flow can
be evaluated using the 90% and 95% CLI's shown in Figures III-4 to III-9.
The CLI is the confidence limit for individual predictions, and represents
the degree of confidence with which a regression equation can be used to
produce new predictions. Since the width of the confidence interval remains
approximately constant, predictions of high values are relatively more ac-
curate than are predictions of low values. For example, the 90% CLI for
predictions of TPL using equation 3-3 is approximately + 21,000 kg P/yr.
Thus, a watershed having a predicted phosphorus export of 13,600 kg P/yr
(the mean value) has a 90% CLI of + 21,000 kg P/yr, or + 154% of the pre-
dicted value. However, a watershed having a predicted phosphorus export of
40,000 kg P/yr also has a 90% CLI of around 21,000 kg P/yr., so the relative
accuracy is improved to + 53% of the predicted value. Predictions of TNL
and FLOW are relatively more accurate than are predictions of TPL. For pre-
dictions of TNL the 90% CLI is approximately + 80,000 kg N/yr, or + 104% of
the predicted value for a watershed having than mean value of nitrogen ex-
port (77,000 kg/yr), using either equation 3-5 (Figure III-6) or equation
3-6 (Fig. III-7). The 90% CLI for predictions of FLOW using equation 3-8,
is + 32.5 x 106 m3/yr, or + 75% of the predicted value for a watershed
having the mean flow of 43.1 x 106 m3/yr (Figure III-9). However, as seen
in Figure III-8, the confidence intervals for equation 3-7, in which DA is
the only independent variable, are comparable.

Since the use of individual land use areas rather than drainage area
alone does little to improve predictions of TNL and FLOW, it is suggested
that equations 3-5 and 3-7, in which DA is the sole independent variable,
be used for predictions of these parameters. For predictions of TPL, the
use of equation 3-3, in which land use areas are used as independent vari-
ables is recommended. In applying these equations, confidence intervals
should be used to assesses the degree of reliability associated with each
prediction.

In using these equations to predict nutrient loadings, nutrient concen-
tration or runoff in other Florida watersheds, several points should be em-
phasized:

1) These predictive equations were developed using a relatively small
group of watersheds that are not necessarily representative of all Florida
watersheds. Predictions made using these equations are valid for only
those watersheds that are from the same population as the test watersheds.
Thus, in evaluating the applicability of these equations for a new situa-
tion, the user is urged to compare the land use characteristics of the new
watershed(s) with those of the watersheds used in these analyses (See Table
III-7). For example, it would be inappropriate to apply the predictive equa-
tions developed for urbanized watersheds to a new watershed that is composed


- 31 -






























































0 1 2 3 4 5 6 7 8 9 10


Drainage Area,


104 ha


Figure III-4.


Non-point source phosphorus loading vs.
drainage area for 41 NES watersheds.


- 32 -































10 -







0

6

O

4




U* TPL = 1,705 +.6.0 ALLURB +
0 / 2 _1.2 CPAST 1.4 RA
4 r = 0.71
P 90% and 95% CLI shown

0 -

0 2 4 6 8 10 12
Measured NPS Phosphorus Loading, 10 kg/yr.


Figure III-5. Predicted vs. measured NPS phosphorus load-
ing using Equation 3-3.


- 33 -

























































0 1 2 3 4 5 6 7 8 9 10
Drainage Area, 104 ha


Figure III-6.


Non-point source nitrogen loading vs.
drainage area for 41 NES watersheds.


- 34 -
























60o -

55 -

50 -
o4545

b;
S40 -


o 35

S30 -

H 25 -

20 -

15



P 5 TNL = 15,145 + 4.2 CPAST +
5 0 16.1 NFWET + 16.0 WA
Sr2 = 0.87
0 90% and 95% CLI shown


0 5 10 15 20 25 30 35 40 45 50 55 60
Measured NPS Nitrogen Loading, 105 kg/yr


Figure III-7. Predicted vs. measured NPS nitrogen
loading using Equation 3-5.


- 35 -




























































0' 1 2 3 4 5 6 7 8 9 10

Drainage Area, 104 ha.


Figure III-8.


Non-point flow vs.
NES watersheds.


drainage area for 41


- 36 -































30



25 -








15

4J
H 10




5 FLOW = 2.77 x 10 + 0.25 ALLURB
-I -+ 0.55 FOR + 0.49 RA + 0.15
CPAST + 0.75 OAG
r2 = 0.96
0 90% and 95% CLI shown


0 5 10 15 20 25 30 35
Measured NPS Flow, 107 m /yr


Figure III-9. Predicted vs. measured NPS flow using Eq. 3-8.


- 37 -











of 90% ALLURB, since the largest fraction of any of the watersheds used in
developing these equations covered by ALLURB is 78.5% (Table III-7).

2) It would be inappropriate to use these predictive equations to esti-
mate the change in loading that would occur if a portion of a watershed is
converted from one land use to another (i.e., forest to residential). The
reason for this is that regression analysis does not necessarily imply a
causal relationship between the independent variables (i.e., nutrient load-
ing) and the dependent variables (land uses). Underlying factors, such as
soil type, drainage and slope may be related to both land use and nutrient
loading. Thus, the difference in loading between forest and urban areas
reflects, in part, differences in edaphic features of the landscape that
render certain areas suitable for urban development and others less so. How-
ever, as development pressure increases, there will be a tendency to urbanize
areas that are not currently considered suitable for urban development
(i.e. under the conditions in which these predictive equations were devel-
oped). Because of this, the loading of nutrients from urban areas devel-
oped in the future may be different from the loading from urban areas cur-
rently in existence. In a similar vein, it should be realized that changes
in management practices of existing land uses may alter the rate of nutrient
export from some areas. For example, in the past 10 years, the practice of
nitrogen fertilization in agricultural areas has changed considerably.
These trends include 1) decreased application rates, 2) the improvement of
tillage methods, 3) the use of fertilizers that are better retained in the
soil column (i.e., urea and ammonium instead of nitrate), and 4) the appli-
cation of nitrification inhibitors (Calvert 1975; Terman and Allen 1970).
These trends should act to conserve fertilizers and reduce the export of
nitrogen from agricultural areas.

3) The predictive equations developed here are based on mean annual
loadings. As mentioned earlier, there may be considerable year-to-year
variation in the annual nutrient loading from non-point sources. For some
applications, such as the assessment of lake restoration techniques, it is
important to be able to determine actual annual loadings of nutrients into
a lake for a period of several years. For this type of study the predictive
equations developed here would not be suitable.

4) In the development of these predictive equations it was assumed that
all of the flow in a tributary other than that emanating from known point
sources (i.e., sewage treatment plants), was derived from precipitation fall-
ing on the watershed. Several watersheds were excluded from the analyses
because it was believed that the flows in the tributaries were too high to
be the result of natural precipitation. In applying these predictive equa-
tions, the user should also be reasonably certain that the flow in a partic-
ular tributary is derived from natural precipitation and runoff.

For some watersheds, the use of these equations would not be suitable,
either because of the limitations cited above or because their predictive
capability is not adequate for a particular purpose. In some cases, the
use of export coefficients obtained from the literature (Tables III-3 to
III-6) may be more appropriate in generating preliminary estimates of NPS
nutrient loading. This is particularly true for small watersheds in which
one land use is predominant. When the literature-based approach is used,


- 38 -











we suggest that original studies be consulted and that reported export
coefficients be used only when the watersheds in which they were determined
are similar, with respect to topography, soils and other characteristics,
to the watershed under investigation.

When more accurate results are required or when the effects of proposed
management strategies on nutrient loadings are being evaluated, the use of
simulation models should be considered. Examples of these include a number
of agricultural models (reviewed by Haith 1980) and the Storm Water Manage-
ment Model, designed for use in urban areas (Heaney et al. 1976).


- 39 -














CHAPTER IV. LIMNOLOGICAL CHARACTERISTICS OF FLORIDA LAKES


The data compiled in this study enable a broad characterization of the
limnological conditions of Florida's lakes. Data were collected for most
major limnological parameters in all three studies (Table 11-2), with the
exception of color and turbidity, which were not collected by the National
Eutrophication Survey. It should be noted that the sampling intensity for
most of the lakes in the three surveys was low; most lakes were sampled
only 3-4 times during an annual period. Considering the large variations
(often order of magnitude) that may occur in algal standing crop and in the
concentrations of major nutrient species during an annual period, computed
means for any single lake must be accepted with caution. However, consider-
ing the large number of lakes in the data base, use of the data to make
general inferences on limnological relationships in the lakes is justified.


MORPHOLOGICAL CHARACTERISTICS

Although Florida's lakes vary considerably in size and morphometry, most
are quite shallow. Of the 101 lakes included in this study, only ten have
maximum depths greater than 10 m, and only three (Annie, Kingsley and Mize)
are over 20 m deep. The study lakes have a considerable range in surface
area (Table IV-1). Many of the lakes in Alachua and Putnam counties have
surface areas of only a few hectares, but several of Florida's largest lakes
are included in this study. By far the largest is Lake Okeechobee (1890
km2), which after the Laurentian Great Lakes is the largest freshwater lake
(in surface area) entirely in the United States.

The morphometry of most Florida lakes has been affected by limestone
solution processes and many have been formed in sinkhole depressions. Some
of these lakes, like Lake Santa Rosa (Putnam County) are nearly circular,
while others are complex dolines that have been formed in adjoining solu-
tion basins (e.g. Cowpen Lake in Putnam County).

The smaller lakes are often hydraulically connected to perched water
tables that are separated from the main (Floridan) limestone aquifer by an
clayey aquiclude. Many lakes are in seepage basins and lack distinct in-
flows or outflows. The water level in these lakes may fluctuate by as much
as several meters between wet and dry periods. Water levels in the larger
lakes are usually structurally controlled to minimize natural variations.

Because of their shallowness and the mild climate, most Florida lakes
do not exhibit stable thermal stratification. Of the 55 lakes studied by
Brezonik and Shannon (1971), only eight developed stable thermal stratifi-
cation during the warm season. Approximately eight additional lakes showed
evidence of temporary thermal stratification lasting for periods of a few
weeks to a few months. The shallowness of Florida's lakes also encourages
resuspension of sediments to the overlying water, particularly in large
lakes and lakes with loose, flocculent sediments. Studies on Lake Apopka


- 40













Table IV-1. Geheral characteristics of study lakes.


Morphological Mean Minimum Maximum

Surface area, km2 23.0 0.01 1890.7

Volume, m3 x 106 61.4 0.03 2494.0

Mean depth (z), m 2.9 0.7 8.3

Maximum depth (z max), m 5.4 0.9 25.3


Chemical & Biological
(annual means)

Color, units 116 2 539

pH 7.1 4.7 10.4

Alkalinity, mg/L as CaCO3 32 0 163

Chlorophyll a (chl a), pg/L 29.1 0.9 276.6

Total nitrogen (N), mg/L 1.51 0.19 5.56

Total phosphorus (P), pg/L 231 7 2,750


- 41 -










(Pollman and Brezonik 1981) indicate that phosphorus is released to the
water column during wind events as the result of desorption from suspended
sediments. This phenomenon undoubtedly contributes to eutrophication prob-
lems in such lakes.


CHEMICAL AND PHYSICAL CHARACTERISTICS.

The majority of Florida's lakes are poorly buffered (mean alkalinity =
32 mg/L as CaCO3), even though they are underlain by limestone. This ap-
parently contradictory situation occurs because many lakes are not hydrau-
lically connected to the underlying limestone formations, but receive the
bulk of their water either directly from precipitation or by surface and
subsurface runoff from the sandy, low-calcareous soils. Many are also
highly colored; the mean color for the study lakes is 116 CPU (chloroplati-
nate units). The frequency and intensity of color reflects the abundance of
watersheds composed of pine forests and wetlands (swamps).

Relationship Between pH and Alkalinity. The pH of Florida lakewaters
is strongly correlated with alkalinity (Figure IV-1). The solid line in
the figure shows the relationship between pH and alkalinity in a water
system where the carbonate buffering system controls pH at atmospheric
pressure (PC02 = 10-3.5 atm). The data suggest that except for the most
acidic lakes the pH is near the equilibrium value defined by the C02-bi-
carbonate system. However, for many of the poorly buffered lakes in the
Trail Ridge region, the observed pH is considerably lower than that expected
for pure water in equilibrium with atmospheric CO2 (pH 5.7). Brezonik et
al. (1981) found that the pH of some of these lakes has decreased by as
much as 0.5 pH units over the past 20 years, apparently as the result of
inputs of acid precipitation. This trend is expected to continue as Florida
increases its production of electricity by coal-fired electric plants.

Factors Affecting Transparency. Seechi disk transparency is one of
the most commonly measured parameters in the study of lake eutrophication.
In addition to being of scientific interest, Secchi disk transparency is
readily comprehended by the public as a measure of water clarity. Several
investigators (Bachman and Jones 1974; Carlson 1977; Brezonik 1978) have
found a hyperbolic relationship between the concentration of chlorophyll a
and Seechi disk transparency. Brezonik (1978) reviewed the theoretical
relationship between light attenuation and Secchi disk transparency (SD)
and concluded that for many lakes the equation:

1/SD = a [color] + b [turbidity] (4-1)

can be used to describe the variation in Secchi disk transparency. Since
turbidity is often closely related to chlorophyll a, it was hypothesized
that

1/SD a [color] + b [chl a] (4-2)

The non-linearity between inverse Secchi disk and chlorophyll a observed
by Carlson (1977) and Brezonik (1978) may be explained by 1) an increase
in the amount of chlorophyll per cell in more eutrophic situations, 2) a


- 42 -

























































0 50 100 150


Mean alkalinity, mg/L as CaCO3


Figure IV-1.


Mean pH vs. mean alkalinity for 101 study lakes.
Solid line shows equilibrium relationship between
pH and alkalinity at pCO = 10-3.5 atm.


- 43 -











change in the size distribution of the seston and in the morphology of
algae with increasing eutrophication, or 3) light attentuation due to
color (Brezonik 1978).

A statistical analysis of the relationship between inverse Secchi disk
and color, turbidity, and chlorophyll a was conducted by Brezonik (1978)
using data from the 55 lake study. A similar analysis was conducted for
this report using the additional data from the two other lake surveys (NES
1977 and Brezonik et al. 1981). Results of the statistical analyses (Table
(IV-2) are similar to those of Brezonik (1978), although the coefficients
differ slightly, and the coefficients of determination (r2 values) are
slightly lower. The lower r2 values may result from the more diverse group
of lakes used in the present analysis. Differences in sampling and analyt-
ical procedures among the studies also may account for the slightly lower
r2 values.

When chlorophyll a is the only independent variable, a log-log model
yields better predictions of SD transparency than does an inverse relation-
ship (cf. eq. 4-2 and eq. 4-8). This result is consistent with the findings
of Carlson (1977), Bachman and Jones (1974), and Brezonik (1978). The co-
efficient of determination for the log SD vs. log (chl a) relationship is
lower for Florida lakes (r2 = 0.70) than reported by Carlson (1977) and
Bachman and Jones (1974) for temperate zone lakes (r2 0.86 and 0.90, re-
spectively) because color affects SD transparency more in Florida lakes than
in temperate zone lakes. As seen by equations 4-4 and 4-9, color alone is
reasonably good predictor of SD transparency.

The best equation for predicting 1/SD (eq. 4-6) uses turbidity and
color as independent variables (equation 4-6). Apparently, the use of
turbidity overcomes some of the problems mentioned above that are encounter-
ed when (chl a) is used to represent light attenuation. Despite the ap-
parent good fit (Figure IV-2), this equation does not predict SD transparency
accurately in some of the clear, oligotrophic lakes in the Trail Ridge
group. The intercept value of eq. 4-6(0.17nmF) corresponds to a SD of only
5.9 m at zero color and turbidity in the water column. This transparency
is far lower than that expected in a water column devoid of algae and color.
Hutchinson (1957) reported that the maximum SD transparency ever
measured in a lake is about 40 m, which would give an intercept term in eq.
4-6 of 0.025. The failure of this equation to predict high SD values accurate-
ly reflects the fact that high values of SD have very low inverse values
that do not affect the fit of the regression equation as much as do lower
SD values. Furthermore, the data base does not include any lakes with very
high transparency values; the maximum mean SD in the data set is 7.9 m
(1/SD = 0.013 m-1) for Lake Sheeler. However, eq. 4-6 does produce reason-
able estimates of SD transparency for predicted SD values less than 3 m.
This situation accounts for most of the lakes in the data set, and probably
most lakes in Florida. For lakes with a predicted SD >3m., eq. 4-6 under-
estimates the actual SD transparency.


BIOLOGICAL CHARACTERISTICS.

Phytoplankton Communities. The composition and standing crop of phyto-
plankton communities in the study lakes is highly variable. The most pris-


- 44 -











Table IV-2. Regression equations to predict Secchi disk
transparency. (1)


Inverse relationships


1/SD = 0.80
2
r =

1/SD 0.49
2
r =

1/SD = 0.70
2

1/SD = 0.37
r =

/SD= 0.17
/SD 2
r =


+ 0.01 (chla)
0.33 n = 100
+ 0.12 (T)
0.62 n = 63

+ 0.002 (C)
0.70 n = 63

+ 0.03 (chl a)
.62 n = 63
+ 0.11 (T)+ 0.
0.82 n = 63


+ 0.001 (C)


002 (C)


Log-log relationships


0.49 = 0.76
2
r = 0.55 n

0.55 0.47
2
r = 0.70 n

0.78 0.39
2
r = 0.53 n


log (T)
= 63


log (chi a)
= 100


log (C)
= 63


(1) SD = mean Secchi disk transparency, m; C = mean color, Pt
units; T = mean turbidity, FTU, (chl a) == mean chlorophyll
a concentration, pg/L.


- 45 -


(4-2)


(4-3)



(4-4)



(4-5)



(4-6)


log (SD)=



log (SD)=



log (SD)=


(4-7)



(4-8)



(4-9)





















w




C o





-e **
0 5

CO
















o' 12

Measured Secchi disk transparency, m



Figure IV-2. Predicted vs. measured Secchi disk transparency using Eq. 4-6.










tine are those in the Trail Ridge region. Most of these lakes (except
Altho, Geneva and Kingsley) have mean pH values of <5.6 and phytoplankton
communities typical of acidic, oligotrophic lakes (Schulze 1980). These
communities are dominated by Staurastrum sp., Scenedesmus sp., Ankistro-
desmus falcatus, Peridinium inconspicuum and several small green coccoids.
Blue-green algae occur in low abundance in these lakes and are represented
mainly by Oscillatoria limnetica and Anacystis incerta. At the other ex-
treme, many of the study lakes are highly eutrophic (e.g. Biven's Arm,
Newnan's, Wauberg, Apopka and Kanapaha). These lakes may exhibit dense
algal blooms dominated by blue green and green algae; in some of the most
fertile these blooms are virtually continuous.

Fish Populations. A recent survey of 22 Florida lakes describes the
changes that occur in fish communities with increasing eutrophication
(Kautz 1981). The oligotrophic lakes, characterized by well-developed com-
munities of littoral vegetation, limited planktonic production, and sandy
bottoms covered with a thin layer of detritus, have fish communities domi-
nated by populations of sport fishes (largemouth bass, bluegill and other
sunfish, striped bass, and pickerel) and forage fishes (Fig. IV-3). Popula-
tions of rough fishes (gar, gizzard shad, bowfin, and tilapia) and com-
mercial fishes (primarily catfishes) are limited and the total biomass and
species diversity of the fish communities is low.

The mesotrophic-eutrophic lakes have the best developed populations of
sport and forage fishes (Fig. IV-3). These lakes are characterized by well-
developed communities of littoral vegetation, large populations of benthic
invertebrates and planktonic organisms, and high habitat diversity. The
total density and species diversity of the fish communities was higher in
these lakes than in either the oligotrophic lakes or the hypereutrophic
lakes. Total fish biomass reached a maximum in the mesotrophic-eutrophic
lakes and fluctuated about this maximum in the hypereutrophic lakes.

The fish communities of the hypereutrophic lakes are dominated by
filter-feeding rough fish such as the gizzard shad. Populations of sport
and forage fishes are limited by the reduced littoral vegetation, blooms of
blue-green algae, accumulations of detritus on the bottom, and periods of
reduced oxygen concentration. Commercial fishes reached their highest levels
in the hypereutrophic lakes, accounting for 10% of the total biomass.

Nutrient Limitation. An assessment of the limiting nutrients) in a
lake is important because the control of inputs of a limiting nutrient to a
lake may be used as a means by which to control eutrophication. In the
majority of temperate zone lakes, phosphorus is the limiting nutrient for
algal growth, while nitrogen is limiting in most of the others. For exam-
ple, Miller et al. (1974) found that phosphorus was limiting algal produc-
tivity in 71% of the lakes and N was limiting in 16% of the 49 temperate
zone lakes they examined using the algal assay procedure. Other nutrients than
nitrogen or phosphorus were limiting in the remaining lakes. A variety of
other micronutrients, including iron, several other trace metals (Goldman
1972), and carbon (King 1972) have also been identified as limiting nutrients
in freshwater lakes.


- 47 -



























































Oligotrophic
lakes


Mesotrophic-
eutrophic
lakes


Hypereutrophic
lakes


Figure IV-3. Structure of fish communities in Florida
lakes. Data from Kautz (1981).


- 48 -










Of the methods used to evaluate nutrient limitation, the most common
are (1) enrichment bioassays and (2) the computation of nutrient ratios,
usually the ratio of total soluble inorganic N to soluble reactive phos-
phate (SIN:SRP). In enrichment bioassays, various nutrients are added to
aliquots of the test water. Either a test species of algal (usually
Selenastrum capricornutum) or an indigenous mixed culture, are added to
each treatment, and the limiting nutrient is determined by comparing the
growth of algae in the nutrient-spiked aliquots with a control. The pro-
cedure has become highly standardized in the Algal Assay Procedure: Bottle
Test (Miller et al. 1978) and is widely used. The second method involves
the measurement of nutrient concentrations and computation of the ratio SIN:SRP.
this ratio has been compared with the results of nutrient enrichment bio-
assays (cf. Chiaudini and Vighi 1974; Miller et al. 1975). The "critical
ratio" usually falls between 10:1 and 20:1. Porcella and Bishop (1975)
stated that a SIN:SRP ratio less than 10:1 clearly indicates N-limitation,
a ratio greater than 20:1 indicates P-limitation, and an intermediate ratio
indicates mixed nutrient limitation.

Algal bioassays (AAP:BT) conducted for 31 of the NES lakes (NES 1977)
indicated that 23 (74%) were nitrogen limited, seven (23%) were phosphorus
limited and one had mixed nutrient limitation. When the results of nutrient
bioassays are compared with the criteria of Porcella and Bishop for nutrient
limitation (Figure IV-4), few misclassifications occur. For the 27 lakes
having an SIN:SRP ratio of less than 10:1, 23 were found to be nitrogen
limited in the AAP:BT and four were found to be phosphorus limited or have
mixed nutrient limitation. All four lakes with SIN:SRP ratios >10:1 in
which biassays were conducted were phosphorus limited. Thus, the criteria
of Porcella and Bishop seem to be reasonably valid, although more data are
needed for lakes having high SIN:SRP ratios to make conclusive remarks con-
cerning the application of these criteria in Florida lakes.

A frequency distribution of mean SIN:SRP ratio for all 101 study lakes
(Figure IV-5) shows that 46% have SIN:SRP ratios <10:1, 21% have ratios
between 10:1 and 20:1, and 33% have ratios >20:1. Thus, if the proposed
criteria of nutrient limitation are valid, nearly half of the study lakes
are nitrogen-limited while only a third are phosphorus-limited.

A plot of the mean SIN:SRP ratio versus the mean chlorophyll a for each
lake (Figure IV-6) shows an inverse relationship. Lakes with high algal
standing crops (mean chl a >50 pg/L) have values of SIN:SRP <10, whereas lakes
with low algal standing crops show more variability but often have mean
SIN:SRP ratios >>20. This observation is consistent with that of Miller
et al. (1974), who concluded that highly productive lakes are more likely
to be nitrogen limited than are less productive lakes. Several explanations
may account for this observation. First, nutrient inputs to eutrophic lakes
are relatively more enriched with phosphorus than are the inputs to less
productive lakes. Sewage effluents, for example, have low ratios of both
SIN:SRP (<2:1) and total N: total P (<3:1) and are thus strongly nitrogen-
limited (Gakstatter et al. 1978). Furthermore, the presence of phosphate-
rich deposits and sandy soils with a low phosphorus retention capacity in
Florida may result in relatively high orthophosphate concentrations in many
streams. Odum (1953) reported that the concentration of orthophosphate in


- 49 -






































1.0 2.0


ORTHOPHOSPHATE


MG P/L


Figure IV-4.


Total inorganic nitrogen vs. orthophosphate concentrations (mean values)
in the Florida NES lakes. N and P next to data points indicate nutrient
found limiting in lake by Algal Assay Procedure.


0.10~



0.05~--
0.01































00



0


) 3 20 -

G4


X: 22.1

101-






0
0- 10.01- 2001- 30,01- 40.01- 50.01- 60.01- >70
10.00 20.00 3000 40.00 50.00 60.00 70.00

Soluble inorganic nitrogen: soluble reactive phosphorus

Figure IV-5. Frequency distribution of mean SIN:SRP ratios in the 101 study lakes.
















130

120

110

100

90

80

70

60

50

40

30

20

10

n


S
5..S 5


0 20 40 60 80 100 120 140 160 180 200 220 240 260

Mean chlorophyll a, pg/L


Figure IV-6.


Relationship between soluble nitrogen: soluble reactive phosphorus
(SIN:SRP) and mean chlorophyll a, pg/L.


0
S


C
-0

3m.
S
Is. *
C
S
S..
0 *

0


00
_ 0
0


280










Florida streams is related to their proximity to phosphate deposits, and an
analysis of the SIN:SRP ratios in the NES tributaries indicates that all
the streams in the vicinity of the phosphate deposits had ratios less than
5 (n = 6). However, about 30% of the streams not in the vicinity of these
deposits also had such low ratio, indicating that factors other than prox-
imity to phosphate deposits are responsible for the low SIN:SRP ratios in
many of Florida's streams.

In-lake mechanisms, particularly denitrification, also may result in a
decrease in the ratio of nitrogen to phosphorus with increasing eutrophica-
tion. In many lakes, loss of nitrogen via denitrification can represent a
large fraction of the input of nitrogen. For example, Messer et al. (1979)
concluded that up to 26% of the nitrogen input into Lake Okeechobee is re-
moved by denitrification; losses of over 10% have been reported in the liter-
ature for a number of other lakes. Although no direct correlation has been
demonstrated between trophic state and denitrification rates, conditions
existing in eutrophic lakes encourage denitrification. These conditions
include a supply of organic matter to supply energy, anoxic conditions near
the bottom, and high levels of nitrate.

The Relationship Between Nutrients and Chlorophyll a Standing Crops.
Several investigators (Dillon and Rigler 1974b Sakamato 1966; Jones and
Bachman 1976) have demonstrated a strong log-log relationship between the
mean concentration of phosphorus in the water column during turnover and
the mean epilimnetic chlorophyll a concentration during the summer (Table
IV-3). This relationship has become the basis for constructing input/out-
put (I/O) models that can be used to predict chlorophyll concentrations in
lakes using only data on phosphorus loading and hydraulic characteristics
(Chapter 5). However, these relationships were developed in temperate zone
lakes that are usually phosphorus limited (Miller et al. 1974). The lakes
in this study differ from these temperate zone lakes in that 1) a dimictic
pattern of stratification is usually not observed, and 2) most are N-limited
rather than P-limited.

Thus we examined the relationship between chlorophyll a and total N and
P for Florida lakes. Since there are no distinct periods of turnover and
stratification in these lakes, annual means usually were used, although the
relationship between mean spring phosphorus concentration (spring being de-
fined as March through May) and mean summer chlorophyll a (summer being de-
fined as June through September) also was evaluated.

Log-log plots of (P)1 and (N)1 versus (chl a) (annual means) are shown
in Figures IV-7 and IV-8, together with the lines of the regression equations
that describe the best fit (equations 4-13 and 4-16). The regression line
describing the relationship between (P)1 and (chl a) (eq. 4-13) for the
study lakes has a much lower slope that do the regression lines of Dillon
and Rigler (1974b)or Jones and Bachman (1976) (see Figure IV-9). A second
regression line, determined using only spring total P and summer chlorophyll a
values, has a similar but slightly lower slope (eq. 4-15) than that for the
annual means. These two lines demonstrate that the amount of chlorophyll a
associated with a given level of total P is lower in Florida lakes than for
most temperate zone lakes. This is the situation that one would expect for
a group of lakes that are largely N-limited. In order to test the hypothesis


- 53 -








Table IV-3. Regression equations of total phosphorus & total nitrogen
vs. chlorophyll a.(1)


Phosphorus

Jones & Bachman (1976):
log (chl a)s = -1.09 + 1.46 log (P) (4-10)
2
r = 0.98 n = 143

Dillon and Rigler (1974b):
Using data from Sakamoto (1966):
log (chl a)s = -1.134 + 1.583 log (P)sp (4-11)
2
r = 0.95 n = 30

Using data from North American lakes:
log (chl a)s = -1.136 + 1.449 log (P)sp (4-12)
2
r = 0.90 n = 46


This study:

Entire data set, annual mean (P) vs. annual mean (chl a)
log (chl a) = -0.41 + 0.79 log (P) (4-13)
2
r = 0.72 n = 100

P-limited lakes only:
log (chl a) = -0.71 + 1.03 log (P) (4-14)
2
r = 0.53 n = 33

Entire data set, spring P vs. summer chl a:
log (chl a)s = -0.16 + 0.71 log (P)sp (4-15)
2
r = 0.57 n = 63


Nitrogen

Entire data set:
log (chl a) = 1.03 + 1.46 log (N) (4-16)
2
r = 0.77 n = 100

N-limited lakes only:
log (chl a) = 1.12 + 1.53 log (N) (4-17)
2
r = 0.77 n = 44
(1) chl a = mean chlorophyll a, ig/L., P = mean total phosphorus, pg/L.
N = mean total nitrogen, mg/L. Subscripts: s = summer mean, sp =
spring mean, none = annual mean.


- 54 -
















1000


100 *








10 **
P4* *
4 0 *
0%






1 1
**






Chl a = 0.392 [P]1 0.79
2
r = 0.72
95% = CLM shown

0.1 1 10 100 1000 10000
Mean total phosphorus, ig/L


Figure IV-7. Relationship between mean total phosphorus and mean
chlorophyll a for 101 study lakes.


- 55 -








































*/ .. 0
*


1.46
Chl a = 10.7 [N]1
2
r = 0.77


95% = CLM shown


1.0

Mean total nitrogen, mg/L


10.0


Figure IV-8.


Relationship between mean total nitrogen and mean
chlorophyll a in 101 study lakes.


- 56 -


1000


1 00 .






0
0

' 10







1
iQ)


0.1



















1/
I.
1/


//
I,
I.
i.! /


~1
/1
//
//
// .7
//
//
I,
I.
// .7,
//
I.
.7-


7,
7 I----"


.


II -I
10 100 1000
Total phosphorus, ig/L


Figure IV-9. Regression lines of total phosphorus vs. chlorophyll a
determined by several investigators.


- 57 -


100O


/


-JONES AND BACHMAN,


1976 (Eq. 4-10)
DILLON AND RIGLER,
1974 (Eq.4-11)
.THIS STUDY, P LIMITED
LAKES ONLY (Eq.4-14)
-*.*- ..THIS STUDY, SPRING P vs.
SUMMER CHL A(Eq.4-15)


I 1 |-I f% 1 1










that nitrogen limitation is actually the factor that results in the low
slope for the Florida lakes, a regression of (P)1 vs. (chl a) was determined
for a group of P-limited lakes (eq. 4-14). In this analysis, phosphorus
limitation was defined by a ratio of SIN:SRP of >20:1. The slope of this
line is steeper than that of the regression line determined for the entire
group of Florida lakes (Figure IV-9), but it still is lower than the slopes
of the lines determined for temperate zone lakes.

There are several factors that could account for the lower slope for
phosphorus-limited Florida lakes than for the temperate zone lakes. First,
it is possible that a ratio of 20:1 for SIN:SRP is not an accurate criterion
of phosphorus limitation; that is, some higher ratio would be more realistic.
This is unlikely, since a critical ratio of 20:1 is quite conservative.

Second, factors other than phosphorus concentration may limit algal
standing crop even in lakes that are considered phosphorus-limited on the
basis of nutrient ratios. For example, nutrients other than nitrogen or
phosphorus may limit algal productivity in some lakes, as may toxic sub-
stances. Miller et al. (1974) found that constituents other than N or P
were limiting to algal productivity in 6 of the 49 temperate zone lakes in-
cluded in their survey. Additional bioassay data are needed to determine
to what extent other nutrients or toxic constituents may be limiting for
algae in Florida lakes.

Third, biological interactions may limit the standing crop of algae in
Florida's lakes below the level in temperate zone lakes containing the same
level of phosphorus. Grazing by herbivorous grazing fish (e.q., shad) or
other structural differences in the food chain may control the algal stand-
ing crop more effectively in Florida's lakes than in temperate lakes, as may
interactions with macrophytes. A detailed discussion of ecological inter-
actions that affect the algal standing crops in lakes is presented by Shapiro
(1979).

Finally, the much longer growing season (essentially year-round in
Florida, compared to the compressed growing season in temperate lakes) may
also affect the chlorophyll a-total phosphorus relationship. The long
period of ice cover and dormancy in temperate lakes leads to a build-up of
inorganic nutrients, culminating in the "spring maximum". This in turn
leads to the spring and early summer pulses of algal blooms. In contrast,
inorganic nutrient levels exhibit less seasonal variations in warm temperate
lakes, and algal growth is more evenly distributed throughout the year (see
following section). Coupled with the biological interactions (e.g. grazing)
mentioned above, this may result in a lower standing crop of algae (hence
chlorophyll a) for a given total phosphorus level than is found in most
pulsed systems.

The relationship between (N)1 and (chl a) in the study lakes is shown
by Eq. 4-16. The r2 for this relationship (0.77) is slightly higher than
the relationship between (P)1 and (chl a) (r2 = 0.72), as one would expect
for a group of lakes that is largely nitrogen-limited. When only nitrogen-
limited lakes were considered (according to the criterion of SIN:SRP<10:1),
the regression equation was only slightly altered (eq. 4-17).


- 58 -










Analysis of Seasonal Trends. In temperate zone lakes, pronounced sea-
sonal variations occur in the standing crop of algae and in the concentra-
tions of major nutrients. Major nutrients tend to reach peak concentrations
during spring and fall turnover, and the algal standing crop tends to reach
maximum levels either following the turnover periods or during the summer
stratification period. In Florida, where most lakes do not undergo stable
thermal stratification and seasonal variations in temperature are less pro-
nounced than in the temperate zone, it is reasonable to hypothesize that
seasonal variations in algal standing crop and in the concentrations of
major nutrients will be minimal. To test this hypothesis for the study
lakes, normalized parameter values were computed for each lakes:

N.. = X.. (4-18)

X..


where N.. = the normalized value for variable i and lake j;
X.. = the variable value during a sampling period;
X.. = the annual mean value;
1J

An overall normalized mean can be calculated:
n N.
S= E (4-19)
j=l n


where n = the number of lakes in the study group.

Thus, if a study lake had a value for a variable during a particular
sampling period equal to the annual mean, Nij would be 1.0. Values greater
than 1.0 indicate a positive seasonal trend while values below 1.0 indicate
a negative seasonal trend.

Seasonal trends of chlorophyll a, total phosphorus and total nitrogen
were analyzed for the two geographical subgroups of lakes included in acid
lake study: the 13 Trail Ridge lakes in northern Florida and the seven High-
lands Ridge lakes in southern Florida. These data are particularly well-
suited for an analysis of seasonal trends for several reasons. First, these
two groups of lakes are located at opposite ends of the state and therefore
represent the extremes in climatic conditions in Florida. For the northern
group, the difference in mean daily air temperature between January and July
is 22C; for the southern group the'difference is only 160C. Second, most of
the lakes in both groups are relatively undisturbed by human activity and
pulses in algal productivity are not likely to reflect man's activities (e.g.,
nutrient-rich runoff during agricultural fertilization). Finally, since all
20 lakes were studied by one group of investigators using standarized methods,
differences between the two groups are not likely to be the result of varia-
tions in methodology.

A plot of the mean normalized values for chlorophyll a throughout the
year (Figure IV-10), shown with 95% confidence intervals, indicates that there
are no statistically significant seasonal trends in chlorophyll a values for


- 59 -






















2.5
-d T

S2.0 "I-



1.5 I






0.5 -





D J F M A M J J A S 0 N

Month


Figure IV-10. Seasonal trends in chlorophyll a concentration for 10 northern Florida lakes
and 10 southern Florida lakes. See text for calculations of mean normalized
parameter values.










either group of lakes. Similar results were obtained when total phosphorus
and total nitrogen data were analyzed.

Data compiled for Lake Apopka and the other Oklawaha lakes (Brezonik
et al. 1978; Tuschall et al. 1979; Pollman et al. 1980) also indicate that
while seasonal trends in chlorophyll a do occur, the trends are not consis-
tent among lakes within a given year or in any one lake throughout the three-
year study period. A plot of chlorophyll a data for Lake Apopka (Figure
IV-ll) shows that while there are major oscillations in the chlorophyll a
standing crop, the timing of periods of bloom and scenescence vary from year
to year. For example, in 1977 peak chlorophyll a concentrations occurred
in the fall, while in 1978 peaks occurred in May and August and in the first
half of 1979 a peak occurred in April. Periods of scenescence show a
similar lack of temporal regularity: minimum chlorophyll a values occurred
in March and May of 1977 and in February and June of 1978.

This analysis suggests that there are no (or only small) regular seasonal
trends in nutrient and chlorophyll a levels in Florida lakes. This conclus-
ion serves as a basis for using annual means for nutrient and chlorophyll a
concentrations, rather than seasonal means (e.g., mean spring total phos-
phorus; mean summer chlorophyll a) in the refinement of nutrient loading
models for Florida lakes (Chapter V).


DEVELOPMENT OF A TROPHIC STATE INDEX

A trophic state index (TSI) that allows the ranking of lakes along a
linear gradient is useful for several reasons: 1) a linear ranking system
facilitates comparisons of trophic state within a group of lakes, 2) use of
a TSI obviates the need to place a lake into a discrete trophic class (oligo-
trophic, mesotrophic, eutrophic), 3) a TSI can be used to quantify historical
changes in trophic state and thereby assess the impacts of cultural pertur-
bations, and 4) a numerical index can be comprehended by the public. Trophic
state indices have been developed based on both single measures of trophic
state (univariate indices) and on a composite of several trophic state indi-
cators (multivariate indices). Trophic state variables that have been used
in indices include dissolved oxygen, total and inorganic phosphorus and
nitrogen, Secchi disk transparency, chlorophyll a, primary production, and
the relative abundance of major ions). Concepts and applications of trophic
state indices and the relative merits of univariate versus multivariate in-
dices have been reviewed and discussed by Shapiro (1976), Brezonik (1976)
and Carlson (1977).

Trophic state indices that might be applied to Florida lakes include
those of Shannon and Brezonik (1972), the National Eutrophication Survey
(1975), and Carlson (1977). The National Eutrophication Survey developed
a "water quality index" (WQI) based on six parameters related to trophic
state. The WQI has two major shortcomings in ranking Florida's lakes: 1)
its use of a dissolved oxygen parameter and 2) its use of two phosphorus
parameters (median total phosphorus and median dissolved phosphorus). The
dissolved oxygen parameter (a constant value minus the dissolved oxygen con-
centration in the bottom waters) is inappropriate because most of Florida's


- 61 -


















150


125


100-


S75


S50


25



J F M AM J J AS 0 N D J F M A M J J AS 0 N D J F MA M J


Figure IV-11. Chlorophyll a concentrations in Lake Apopka, January, 1977 to
June, 1979.










lakes are not stratified. Rankings made using the D.O. parameter for the
Florida NES lakes were widely scattered and not well-correlated with other
trophic state indicators. The use of two phosphorus parameters in the WQI
overemphasizes the importance of phosphorus as a trophic state indicator,
particularly for Florida's many nitrogen-limited lakes.

The index developed by Brezonik and Shannon (1971) using data from 55
Florida lakes could reasonably be applied to other lakes in the present data
set except that data on primary production and cation composition required
for the computation of this TSI were not collected in the Florida NES and
the acid lake survey. Furthermore, values of chlorophyll a, total organic
nitrogen and specific conductivity for several lakes lie outside the range
of values in the original data base used to construct the TSI.

Carlson (1977) developed three separate univariate indices of trophic
state, based respectively on Secchi disk transparency, total phosphorus and
chlorophyll a. Carlson's index was initially based on Secchi disk trans-
parency, with values scaled so that the zero point corresponded to a Secchi
disk value greater than any value yet reported (64 m):

TSI(SD) = 10(6 ln(SD)/ln 2), (4-20)

where SD = Secchi disk transparency (m).

Trophic state indices for phosphorus and chlorophyll a were developed
using the relationships between these parameters and Secchi disk transparency.
Thus, for a group of 147 lakes the relationship between Secchi disk trans-
parency and chlorophyll a concentration was best expressed (Carlson 1977)
using the equation:

In SD = 2.04 0.68 In (chl a); r2 = 0.86. (4-21)

A TSI based on chlorophyll a was computed by combining equations 4-20 and
4-21:

TSI(CHA) = 10(6 -(2.04 0.68 ln(chl a)/ln 2)). (4-22)

A TSI based on phosphorus was based on an observed inverse relationship be-
tween total phosphorus and Secchi disk transparency:

SD = 64.9/(P)1. (4-23)

Combining eqs. 4-21 and 4-23 results in a TSI based on phosphorus:

TSI(TP) = 10(6 ln(48/(P)1)/ln 2). (4-24)

The indices developed by Carlson have several advantages, including
small data requirements, objectivity, and reliance upon commonly measured
and understood indicators of trophic state. However, linear regressions of
Carlson's TSI(TP) against both TSI(SD) and TSI(CHA) showed that the TSI(TP)
values were significantly different (at the 95% confidence level) from the
other two TSI values for the Florida NES lakes. This is not surprising,


- 63










considering the predominance of nitrogen-limited lakes in the NES data set.
Consequently, we developed a nitrogen-based TSI using a computation method
analogous to that used by Carlson in developing the phosphorus-based TSI
(Kratzer and Brezonik 1981). For the NES lakes, the relationship between
chlorophyll a and total nitrogen is expressed by the equation:
2
ln(chl a) = 2.44 + 1.6 ln(N) ; r = 0.89 (n = 39). (4-25)

Combining equations 4-21 and 4-25 yields:

SD = 1.46 (4-26)
(N)1

A TSI(TN) can be calculated by substituting eq. 4-26 into eq. 4-20:

TSI(TN) = 10(6 ln(l.46/(N) )/ln 2). (4-27)

Ideally, when both TSI(TP) and TSI(TN) are computed, the smaller of the
two TSI's should represent the limiting nutrient for any given lake. This
hypothesis is generally supported by the Florida NES data. The five lakes
(Yale, Kissimmee, Marion, Reedy and Apopka) with TSI(TP) values considerably
lower than TSI(TN) all were phosphorus-limited according to the NES algal
bioassay results.

For this study the lesser of TSI(TP) and TSI(TN) was averaged with the
corresponding TSI(SD) and TSI(CHA) to compute a TSI(AVE). The values of
the four univariate TSI's plus TSI(AVE) are listed for the 101 study lakes
in Table IV-4. The TSI(AVE) values correspond well with assessments made
by the NES limnologists (Lakes 1 to 40) and by Brezonik and Shannon (1971)
(lakes 41 to 92). In addition, the TSI(AVE) values for the NES lakes agree
reasonably well with the values of the EPA's Water Quality Index (r2 = 0.64).
The use of a TSI(AVE) has the advantage over the use individual TSI's in
that the use of a single index of trophic status can be more easily used for
comparative and management purposes. Furthermore, the compositing of
physical, biological and chemical components of trophic state into one in-
dex reflects the multidimensional nature of the eutrophication phenomenon,
since it is generally agreed that no single trophic indicator adequately
measures the underlying concept. Combining the major physical, chemical
and biological indicators of trophic state into a single index smooths out
the variability associated with individual indicators and provides a rea-
sonable composite measure of trophic conditions in a lake.


- 64 -











Table IV-4. Trophic Status Index for study lakes.

Lake TSI(CHL) TSI(TP) TSI(TN) TSI(SD) TSI(AVE) Assessment


101
84
89
93
78
77
91
79
82
86
92
81
80
87
90
96
53
1
100
47
99
65
41
98
5
62
76
61
95
43
94
42
83
44
2
3
52
88
51
66
4
50
85
21
70
49
73
6
67
46


- 65 -


U
U

U
U
U
U
U
U
U
U
U
U
M

U
0-M

0


0

M
H
0
M

0

0
0
0
M

0
0
0
M
M

U
E
U


M-E
M
M





Table IV-4. continued...


Lake TSICHL)


TSI(TP) TSI(TN) TSI(SD) TSI(AVE) Assessment


65
52
60
57
61
61
57
64
68
66
74
63
56
63
78
79
66
55
68
65
60
80
99
90
79
74
78
76
107
88
68
79
71
94
93
92
95
81
105
100
83
93
90
94
98
89
97
113
109
109
118


* U = ultraoligotrophic, 0 =
hyperutrophic. Assessment
1972 (lakes 41-92).


oligotrophic, M = mesotrophic, E = eutrophic, H =
by NES (lakes 1 to 40) or Shannon and Brezonik,


- 66 -
















CHAPTER V. APPLICATION OF NUTRIENT LOADING
MODELS TO THE FLORIDA NES LAKES

INTRODUCTION


In the past decade considerable efforts have been made to quantify the
relationships between nutrient loading and lake trophic status using simple
input-output (I/O) models. Nutrient loading models have been widely used
to predict the effects of changes in nutrient loading on lake trophic status
(Vollenweider 1969, 1975, 1976; Patalas and Salki 1973; Dillon and Rigler
1975; and many others) and to predict the trophic status of new reservoirs
(e.g. Bradford and Maiero 1978; Baker et al. 1978; Huber and Brezonik 1980).
The utility of these models is greatly enhanced by their modest data require-
ments and computational simplicity.

This chapter reviews the developments made in mass balance nutrient
models for lakes over the past decade and the resulting advances in our
ability to predict trophic conditions in lakes from information on phos-
phorus and hydraulic loadings and basic lake morphometry. Nearly all these
predictive models have been developed using data on temperate lakes. This
chapter describes the application of these models to warm-temperate and
subtropical lakes in Florida and evaluates their usefulness in predicting
trophic conditions in Florida lakes.


HISTORICAL DEVELOPMENT

Phosphorus Input/Output Models

The relationship between nutrient concentrations and algal productivity
in lakes has long been recognized (see review by Vollenweider 1968). Since
phosphorus has been identified as the most common limiting nutrient in tem-
perate lakes, the development of nutrient loading models has focused entire-
ly on phosphorus, although Vollenweider (1969) noted that the principles
involved could be applied to other nutrients.

In the development of his phosphorus loading model, Vollenweider (1969)
expressed the change in the mass of phosphorus in a simplified form:

dP/dt = J Lout S (5-1)

where dP/dt = rate of change of the mass of P in a lake.

J = flux of P into the lake,

L = flux of P from the lake via its outflow, and
out


- 67 -










S = the rate of loss of P via sedimentation and other mechanisms
other than loss through the outlet.

In formulating his model, Vollenweider made several simplifying as-
sumptions:

(1) A lake behaves like a continuously stirred tank reactor. That is,
any substance entering a lake becomes completely mixed as soon as it enters.

(2) The rate of sedimentation is proportional to the amount of phos-
phorus in the lake, i.e., S = aP P, where P = the mass of phosphorus in the
lake and a is a first order sedimentation coefficient with units of yr-1.
p
(3) The concentration of phosphorus in the outflow is equal to the con-
centration of phosphorus in the lake. Thus, L = p (P) where p is the
hydraulic flushing coefficient (Q/V) in yr- at(P) 1 = mean lake phosphorus
concentration.

(4) There is no seasonal fluctuation in loading.

Although none of these assumptions is entirely valid, they allow the
development of a simple expression to compute (P):

dP/dt = J a P p P, (5-2)
p w
If it is further assumed that the system is at steady state, i.e.,
dP/dt = 0:

J = pP + a P; orP= P (5-3)
w p a + p
p w

Dividing through by lake volume (V) yields

P J/V (5-4)
V p w+
w p

and since P/V = [P]1 and J/V = L (the volumetric loading rate),

L
[P] i v (5-5a)
+ p
p w

or

Lv= p [P]1 + Pw P]"1 (5-5b)


Phosphorus loading usually is expressed on an areal basis (in g/m 2-yr), as
L = L /z, where z = mean depth:
v. v

[P] = p_ L (5-6)
z (pw + ) z p + q
w p p s


- 68 -










where q = hydraulic loading (m/yr) = z p Since L z and p can readily
be measured, the major difficulty in using this model is the estimation of
the sedimentation coefficient, a Vollenweider (1975) found that for a
group of 25 temperate zone lakes a = 10/z Substituting this expression
into eq. (5-6) yields

[P1= LPj (5-7)
10 + q

This relationship is tantamount to stating that the apparent deposi-
tion or settling velocity for total phosphorus in lakes is constant. If
a = 10/z, then a z = constant; o .E has units of m/yr, which dimensionally
iA a velocity. This term can be interpreted as the settling velocity for
total P and is given the symbol v It should be noted that both Vp and a
are not subject to strict physical interpretation and measurement, since
there is more than a single mechanism (and a single form of phosphorus) in-
volved in deposition to sediments. Moreover, sediment deposition is not
the sole internal sink for phosphorus, although on a long term basis it is
by far the most important. Phosphorus also can be lost from the water
column via uptake by macrophytes or incorporation in fish biomass; neither
of these reservoirs is measured in typical phosphorus budgets. Thus the
basic mass balance model for phosphorus is a simplification of reality, and
the sedimentation term is a composite of all internal sink processes, in-
cluding deposition (settling) of detritus, adsorption of orthophosphate by
sediments, and uptake by macrophytes followed by direct incorporation of
dead macrophyte tissue into the sediments. Consequently, it is not possible
to measure Op or Vp directly, although sedimentation traps may provide good
approximations under certain limited conditions.

Jones and Bachman (1976) found that for a group of 16 Iowa lakes, the
best fit for eq. (5-6) was obtained using a constant value of 0.65 for a .
Thus for their data sedimentation rate was independent of depth, and eq.P
(5-6) becomes

[P] = 0.. (5-8)
q + 0.65

More recently, Vollenweider (1976) proposed that a 1/TJ7, where
T is the hydraulic retention time (T = p w-1; hence T w= Z/q ). Eq. 5-6
then becomes:

[P] =q 1 -2 (5-9)
qs(1 + T )

It is to be noted that in developing eq. 5-9, Vollenweider has gone beyond
the dimensionally- and theoretically-correct (albeit perhaps simplistic)
mass balance model and has interjected an element of empiricism into the
phosphorus predictive equation.

An alternative approach to that requiring a determination of a was
proposed by Dillon and Rigler (1974). These workers proposed that an easily
measured retention coefficient, R can be used to replace a in Vollenweider's
model: P


- 69 -










R = 1 out [P]out (5-10)
p Q. [P.
S Qin in

which is simply the fraction of the input phosphorus that is retained within
a lake. If the lake is at steady state, it follows that the phosphorus re-
tained must be added to the sediment. The sedimentation coefficient can be
expressed in terms of R Since L = p [P]1 + p[P]1 (eq. 5-5b),

L = u + p
v p w (5-11)
[P]
[e]1

From eq. 5-5b and the nature of steady state systems, it also is clear that
R L = a [P] Substituting this relationship into eq. 5-11, we obtain:

a pw (5-12)
S1 R '
p

and R = p (5-13)
p w

Substituting eq. 5-12 into eq. 5-6 and noting that q = z p we obtain
L (l R) = (l- R) (5-14)
[P = p p p(5-14)
w qs

Dillon and Rigler (1974a) developed loading criteria plots based on eq. 5-14.
Plots of Lp(l R )/p vs. E have a slope of equivalent to the steady state
concentration of totaY phosphorus ([P]1) in a lake and can be used to predict
trophic conditions (see following section).

Larsen and Mercier (1975) presented an alternative approach that is de-
rived directly from Vollenweider's model but emphasizes the importance of
the influent concentration of phosphorus, [P]., rather than the total loading.
Since L = Jii/A, where Ji is the total input of phosphorus (g/yr), and since
q = Q .A, it is clear that L /q = J./Q. = [P] .. Substitution of this rela-
tionship into the Dillon-Rigler relationship (eq. 5-14) results in:

[P] = [P] (1 R ). (5-15)

Finally, Chapra (1975) introduced the concept of an "apparent settling
velocity", Vp, where vp = a z, as discussed earlier. Substituting this term
into eq. 5-13 results in

R = v (5-16)
P v + q
p s

Combining equations 5-14 and 5-15 yields

[P] Lp (5-17)
S v + qs
p s


- 70 -










The apparent settling velocity can be determined by rearranging eq. 5-16 and
solving. If vp is assumed to be constant, R can then be calculated directly
knowing only qs. This relationship is conceptually sound in that when q =
0, R must be 0, regardless of the magnitude of v Chapra (1975) found a
value of 16 m/yr for v in a group of Ontario lakes, whereas for a somewhat
larger group of lakes PKirchner and Dillon (1975) found that vp was 13.3 m/yr
(using eq. 5-16 and measured values of Rp and qs). As noted earlier,
Vollenweider's (1969, 1975) formulation, Op = 10/2, corresponds to a vp of
10 m/yr.

In addition to the fundamental I/O approach to modeling lake phosphorus
dynamics, several empirical models have been developed. Two such models
recently were proposed as statistical improvements on the previous predictive
models for total phosphorus concentration (Reckhow 1977, Walker 1977). The
Reckhow (1977) model was developed by using a nonlinear regression to account
for those variables that produce a nonlinear response in the total phosphorus
concentration. The resulting equation, based on 33 north temperate lakes with
qs < 50 m/yr, was:

(P)1 = p (5-18)

10 + + 1.05 qs exp (0.012 qs)

Walker (1977) developed a model from a data base of 105 north temperate lakes
and came up with the predictive equation:
L
(P)1 = p (5-19)
Sq (1 + 0.824 T 0.454)
s w

Imboden (1974) and Snodgrass and O'Melia (1975) have developed more com-
plicated models of phosphorus that divide lakes vertically into two compart-
ments (the epilimnion or upper, mixed layer, and the hypolimnion, or lower
stagnant layer). Their models require the solution of four coupled differen-
tial equations for the summer (stratification) period (ortho-P and particulate-
P for the epilimnion and hypolimnion). Only two differential equations need
to be solved for the winter (circulation) period (ortho-P and particulate-P
for the entire lake). Data requirements for these models are more compli-
cated and include: (1) phosphorus exchange coefficients between the epilimn-
ion and the hypolimnion and between the sediment and overlying water; (2) rate
coefficients for photosynthesis and mineralization; (3) settling coefficients
for particulate phosphorus; (4) loading rates for both forms of phosphorus;
(5) water flow rates; and (6) depths of the epilimnion, thermocline, and
hypolimnion. These models produced one surprising result: deep lakes were
predicted to have higher phosphorus concentrations in their euphotic zones
than shallow lakes, a result that is contradictory to the observations of
several investigators. Snodgrass and O'Melia (1975) proposed to resolve
this discrepancy by assigning a depth dependence to the exchange coefficient
across the thermocline and to the effective settling velocity of particulate
phosphorus. Since most Florida lakes are thermally unstratified at all times,
the complications introduced by these two-compartment models are unnecessary.


- 71 -











Prediction of R
p

An advantage of the Dillon-Rigler and Larsen-Mercier models is that Rp
can be determined experimentally (eq. 5-10) from phosphorus budget measure-
ments. However, in some important applications, Rp is not known. For ex-
ample, in predictive studies on proposed reservoirs, Rp obviously cannot be
measured directly. Also, it is often desirable to predict Rp for an exist-
ing lake for which a complete phosphorus budget is not available or for
which proposed management activities may change hydraulic and or nutrient
loading characteristics. These needs have led to several empirical attempts
to predict Rp from other easily measured limnological variables, particular-
ly hydraulic parameters such as qs and pw. Kirchner and Dillon (1975) de-
rived a double exponential equation to predict Rp from qs using a data base
of 15 southern Ontario lakes:

R = 0.426 exp (-0.271 qs) + 0.574 exp (-0.00949 qs). (5-20)
p

The correlation coefficient for this relationship based on 15 lakes in
southern Ontario was 0.94; furthermore, the relationship is reasonable in
that it gives an R of 1.0 when qs = 0 (i.e. when there is no flow from the
lake).

Larsen and Mercier (1975) evaluated a variety of empirical formulations
to estimate Rp using a data base of 20 temperate lakes, for example:

R = 0.854 0.142 qs, (5-21)
P

and R = 0.482 0.112 Inp (5-22)
p w

Equation 5-21 yielded a correlation coefficient of 0.92 using all 20
lakes in the data set. When the two shallowest lakes were excluded, eq. 5-22
gave nearly as good a fit. Larsen and Mercier noted that these equations do
not provide theoretically correct predictions for lakes with extreme values
of qs or Pw, in which cases unreasonable predictions of Rp (>1 or <0) could
occur. A relationship that was more theoretically sound was obtained by
Larsen and Mercier from the finding that Op (as estimated by mass balance
models) was related to the flushing coefficient (p ) for the 20 lakes:

In a = 0.472 In p 0.273, (r = 0.84). (5-23)
p w
Substituting this term into eq. 5-13 and simplifying resulted in the
approximate relationship:
1
R = 1- (5-24)
P 1 + p 2
w

Prediction of Chlorophyll a Concentration

A major advance in the development of nutrient loading models came with
the recognition of a relationship between the concentration of phosphorus


- 72 -










during spring turnover (P)sp and the mean summer concentration of chlorophyll
a (Dillon and Rigler 1974a; Sakamoto 1966; Jones and Bachman 1976; see
Chapter 4). These correlations have led to the development of models that
can predict the chlorophyll a concentration directly from data on nutrient
loading, morphological features and hydraulic parameters (Vollenweider,
1976; Jones and Bachman 1976; Chapra and Tarapchak 1976). For example,
Vollenweider (1976) regressed the right side term of eq. 5-9 against mean
summer chlorophyll a (chl a)s to obtain the relationship:

[chl a] = 0.367 [Lp/qs ]0.91 (5-25)
s i + 71
w

In a similar manner, Chapra and Tarapchak (1976) combined Dillon and Rigler's
equation (5-14) and Chapra's equation (5-16), setting vp = 12.4, and regress-
ing the resulting term against [chl a] s to obtain

[chl a] = 1866 p 1.449 (5-26)
q + 12.4

Finally, Hand (1975) used an empirical approach to predict [chl a] for Florida
lakes using a shape factor (5) and the outflow phosphorus concentration (P /
out
Qout):

[chl a] = 114 (Nout/3 + Pout)/Qout 5. (5-27)

Nutrient Loading Criteria: Graphical Approaches

Parallelling the development of models to predict [P] ,and later (chl a)s
is the development of critical loading plots that depict critical loadings
(i.e., permissible loadings that maintain oligotrophic conditions and exces-
sive rates above which eutrophic conditions occur). These criteria are plot-
ted as functions of lake morphometry and/or hydraulic conditions. The first
such plot was developed by Vollenweider (1968) and was empirical in nature.
This graph (Figure V-1) depicts trophic status as a function of area phos-
phorus loading (Lp) and mean depth (2). Brezonik and Shannon (1971) develop-
ed a similar graph showing the relationships between nitrogen and phosphorus
loading and a trophic state index for Florida lakes. The permissible and
excessive loading rates they developed were higher than Vollenweider's cor-
responding loading criteria, implying that Florida lakes are capable of
assimilating more nutrients than are the temperate lakes used by Vollenweider.

Although Vollenweider's 1968 loading plots were widely used following
their introduction, Vollenweider and others realized that the failure of the
plots to account for hydraulic characteristics limited their usefulness.
Thus Vollenweider (1975) introduced a second critical loading plot based on
his phosphorus loading model (eq. 5-6). Critical phosphorus levels were con-
sidered to be 10 and 20 Pg/L, respectively, as lower limits for mesotrophic
and eutrophic conditions. The resulting plot of Lp vs. qs (Figure V-2) has
three distinct segments:

1) q < 4 m/yr. L constant (dependent only on a ).
s crit P


- 73 -














Cd

CIO
0



Cd

0




CL -


10

5,





1

0.5




0.1

0.)05




0.01


Figure V-1.





10


Cd
00
Cd
O'Cd
Ca 4
0






4~J
0


Cd


5




1

0.5





0.1


0.05




0.01


0.1 0.5 1 5 10,
Mean Depth () m

Vollenweider's phosphorus an
(1968), L versus z.


1- 00

50







5





S 1


~ 0.5
phic Zone

I
50 100


d nitrogen loading criteria


0.1 0.5 1 5 10 50 100
Hydraulic Loading Rate (qs), m/yr


Figure V-2. Vollenweider's phosphorus loading criteria (1975),
L versus qs.


- 74 -










2) 40 m/yr > qs > 4 m/yr. Lcrit proportional to qs (and a ).

3) qs > 40 m/yr. Lcrit proportional to qs and a qs being dominant.

This model is an improvement over Vollenweider's early depth-loading plots
in that the critical loading is recognized to be a function of the areal
water.

Dillon (1975) introduced a critical loading plot based on the model
developed by Dillon and Rigler (eq. 5-14), again using 10 and 20 pg/L for
critical levels of [P]1. In this case, Lp (1 Rp) Tw is plotted vs. z; the
slopes of the critical lines are 10 and 20 pg/L, respectively (Figure V-3).

Finally, Larsen and Mercier (1975) used eq. 5-15 to construct a plot
relating (P). to Rp (Figure V-4). This plot delineates zones of oligotro-
phic, mesotrophic, and eutrophic lakes on the basis of R and average in-
fluence concentrations of total phosphorus.



APPLICATION OF NUTRIENT LOADING MODELS TO FLORIDA LAKES

The nutrient loading models described above were developed using data
from temperate lakes, and their validity has not been evaluated for sub-
tropical lakes. As demonstrated in Chapter IV, Florida lakes differ con-
siderably from temperate zone lakes in their limnological characteristics.
Unlike temperate zone lakes, Florida lakes do not undergo a dimictic pattern
of stratification, as do most temperate zone lakes. Many are quite shallow
and harbor large beds of macrophytes. Overall, they are considerably more
colored than most temperate zone lakes. Finally, nitrogen tends to be the
limiting nutrient for many Florida lakes, raising the question of whether
loading models based on phosphorus are applicable.

Thus, the objective of this phase of the study was to evaluate the use
of I/O models using the Florida NES data base. Models to predict (P)I, (N)1,
(chl a) and phosphorus-and nitrogen-retained coefficients (Rp and RnY were
analyzed using regression analyses. The GLM procedure of SAS was used to
select the best models and to determine the 95% CLI for these models. Since
seasonal variability is less pronounced in Florida lakes compared to temper-
ate zone lakes (Chapter IV), annual mean values of (P)I, (N)1 and (chl a) were
used in these analyses. Available phosphorus loading plots (Vollenweider
1968, 1975; Dillon 1975; Larsen and Mercier 1975) were examined and modified
to fit Florida lakes. Analogous loading plots for nitrogen also were developed
since many Florida lakes are nitrogen-limited.



Total Phosphorus Concentration. Most of the predictive equations for
total phosphorus ([P]1) analyzed here are based on equations developed by
previous investigators. Coefficients for the equations were modified using
regression analysis to improve their predictive capability for the Florida
NES lakes. Log-log transformations of both (P)I and the predictive terms in
the equations were used because of the wide ranges of encountered values.


- 75

















1

0.5




0.1

0.05


0.01 1 -- I I 1
1 5 10 50 100

Mean Depth (2), m


Dillon's phosphorus
constant phosphorus
1000 I


500






100


50






10


5


loading criteria (1975), with lines of
concentration distinguishing trophic states.


0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
R


Figure V-4. The Larsen and Mercier phosphorus loading criteria (1975),


- 76 -


P-1




P.


Figure V-3.


500 1000








Equation TP5 is based on the concept of a "mean apparent settling velocity"
for phosphorus (Chapra 1975); Vp was calculated for each of the Florida NES
lakes by rearranging equation 5-16:

Vp = qs RP (5-28)
1 Rp

The mean value of Vp for the Florida NES lakes, 8.53 m/yr, was then substi-
tuted into equation 5-17 to produce TP5.
2
Eight equations were evaluated (Table V-1), and the r values ranged
from 0.77 to 0.91. The best predictive equation, TP2, is based on Dillon
and Rigler's (1975) model. Prediction of total phosphorus using TP2 re-
quires data on areal phosphorus loading, hydraulic loading, and the phos-
phorus retention coefficient (Rp). To obtain the best estimates of lake
phosphorus concentration the measured phosphorus retention coefficient
[Rp = (EPin EPout)/ZPinJ should be used in TP2. However, since the out-
flow phosphorus loading may not be known or economically determined, it may
be necessary to predict Rp from morphometric and hydrological parameters.
The ability to predict Rp in Florida lakes by equations involving such
parameters is limited (see p. 80), and estimates of (P)I made using predicted
values of Rp in TP2 will not be very accurate. Several other equations
that do not require data on outflow phosphorus loading (TP3 to TP6) have
r2 values between 0.82 and 0.84, but these equations all have C.V. values
higher than TP2. The 95% confidence limits for individual predictions shown
in Figure V-5 can be used to evaluate the accuracy of predictions made using
TP2. For example, a lake whose predicted phosphorus concentration is 0.100
mg/L has a 95% CLI of 0.057 to 0.116 mg/L.

Total Nitrogen Concentrations. The nitrogen balance in a lake can be
expressed by the following equation:

dN/dt = N. N + N N N (5-29)
in out fix sed den (
where dN/dt = change in the mass of nitrogen in a lake,

N. = flux of nitrogen into lake,
in
N out = flux of nitrogen throughout outflow,
Nfix = rate of nitrogen fixation,
Nden = rate of denitrification, and
Nsed = rate of nitrogen loss by sedimentation.

Since Nix and Nden were not determined for the NES lakes, these fluxes
are grouped together with Nsed as parts of a composite loss term, a N. The
nitrogen mass balance equation thus simplifies to:

dN/dt = N.in N NN, (5-30)

which is analogous to the phosphorus mass balance (eq. 5-2). At steady state:
dN
d = 0 = N. N(p + N) (5-31)

N.
in
or N = (5-32)
w N


77 -











Table V-1


Predictive equations for total phosphorus concentration


Original Equation Modified Equation

Predictive Equation* Investi- Eqn. no. 2 2
gator in text r n** r C.V.
TP1=0.682[L /q (l+vT)] 0.934 Vollenweider 5- 9 29 0.79 31.3
p w 862 (1976)
TP2=0.748[L (1-R )/q ] Dillon and 5-14 25 0.91 14.8
0.851 Rigler (1975)
TP3=0.984[Lp (1-R K+D)/qs] .851] Dillon and 5-14 28 0.84 23.6
09 0.964 Rigler (1975)
TP4=0.706[0.84L /(0.65z+q )] Jones and 5- 8 0.84 29 0.82 29.3
Sp 8 Bachman (1976)
TP5=0.952[L /(8.53+q )]W Chapra (1975) 5-17 29 0.83 28.1
L
TP6=0.885 P 0.968 Reckhow 5-18 0.88 29 0.82 29.1
18z -+1.05q exp(0.012q ) (1977)
10+z
L
TP7=0.643 [0 .454 932 Walker (1977) 5-19 0.91 29 0.78 32.6
qs (1+0.824T- )

TP8=0.416[L /q s]873 This study 26 0.77 30.0
ps


Equation]


where a and b


For each predictive equation (except TP10), TP. = a [Ori
** n is the number of lakes included in the regression.















10.0


1.0 -


0




U
0
8 *


O
0 0.i0- -
o 0.100




=8 0.748 (1-R /q 62

2
r = 0.91
95% CLI shown




0.01
0.01 0.10 1.0 10.0
Measured total phosphorus, mg P/L


Figure V-5. Predicted vs. measured total phosphorus concentration using
equation TP2.


- 79 -








Dividing by volume (V), we obtain a predictive equation for the steady-state
concentration of total nitrogen:
N. L
(N)1 = in n (5-33)
V(Pw + UN) = Z(P + ON)

Since there are no existing equations to predict total nitrogen concentration,
the equations evaluated here were based on modifications of phosphorus loading
equations. For example, TNI was derived by setting UN = 0.65z and substituting
Ln for L ; TN2 TN3 were derived by substituting Ln for L and R for R_.
Equation TN5 was derived by computing a mean apparent settling velocity or
nitrogen (vn) in Florida NES lakes in a manner analogous to that used above
to compute v-.-

The predictive equations (TN1-TN4) have r2 values ranging from 0.52 to
0.77 (Table V-2). The best predictive equation, TN2, is based on Dillon and
Rigler's (1974a) model, with Ln and Rn substituted for Lp and Rp. TN2 re-
quires data on areal nitrogen loading, hydraulic loading and the nitrogen
retention coefficient. The use of measured retention coefficients will pro-
duce the most accurate estimates of mean lake nitrogen concentration. How-
ever, in some cases measured values of Rn may not be available; it may there-
fore be necessary to use predicted values of Rn. Equations to predict Rn,
evaluated below, are unfortunately not highly accurate. Thus, when it is
necessary to use predicted values of Rn in TN2, the resulting predictions of
(N)1 will be of limited usefulness. Although the other three predictive
equations do not require data on the outflow nitrogen loading, their predic-
tive capacity is considerably lower than that of TN2. The small 95% confi-
dence limits for individual predictions shown in Figure V-6, indicates that
good accuracy can be achieved in predicting [N]1 using TN2.

Phosphorus and Nitrogen Retention Coefficients. Since the best equa-
tions to predict total phosphorus and total nitrogen require the use of re-
tention coefficients, it is desirable to be able to predict Rp and Rn using
data on the hydrologic and morphologic characteristics of a lake. Several
investigators (Kirchner and Dillon 1975; Larsen and Mercier 1975) have shown
that this approach can be successful in predicting R The seven Rp predic-
tive equati-ons-examined here--(Table-V-3) were based-on-modifications of -pre-
vious equations using the parameters z, TW, and qs. Unfortunately, none of
the equations is very successful in predicting Rp.
The r2 values for the equations ranged from 0.41 to 0.53. A plot of the
measured phosphorus retention Rp versus the values predicted by the best
predictive equation (RP6) shown in Figure V-7 illustrates the substantial
scatter in the relationship, and the 95% CLI for new predictions shows that
at mean value of Rp (0.48) the predicted values is approximately + 0.4 of
the actual value. The best predictive equation (RP6) was developed here by
combining mean depth (z) and water residence time (Tw) by multiplication,
rather than by division as is done to define the areal hydraulic loading
rate. The physical meaning of this factor (E.Tw) is uncertain, and the
difference in predictive capability between RP6 and the other equations is
too small to warrant conclusions about the relative importance of mean depth
and water residence time as predictive parameters of phosphorus retention.


- 80 -












Table V-2. Predictive equations for total nitrogen concentration


Predictive Equations* Original Basis n r C.V.
for Equation

TNl = 1.08 [L /(0.65z + q )] 0859 Jones and 27 0.53 69.7
Bachmann (1976)

TN2 = 0.899 [L /(l R )/q s] 0976 Dillon and 24 0.77 47.6
n n ~Rigler (1975)

TN3 = 0.841 [L /q ]0.877 This study 27 0.52 70.2
n s
TN4 = 1.29 [L/q s(1 + /-)] 0.858 Vollenweider 27 0.55 67.9
(1976)

0.341
TN5 = 1.69 [Ln/(5.49 + q s)] Chapra (1975) 29 0.19 91.5


* For each predictive equation (except TN3), the original equation was
phosphate predictive equation with L and Rp replaced by Ln and Rn.
TNi = a[Transformed Original Equatio ]b where a and b are constants
determined by the regression.


a total
Thus,
















10.0



















1.0


















0.1 -









Figure V-6.


0.1 1.0 10.0

Measured total nitrogen, mg N/L


Predicted vs. measured total nitrogen concentration using
equation TN2.


- 82 -











Table V-3. Predictive equations for phosphorus retention coefficient


Original Equation Modified Equation
Predictive Equation* Investigator Eqn. no. 2 2
in text r n r C.V.

RP1 = 0.767 0.367 log (qs) Larsen and 5-21 0.88 27 0.46 42.2
s Mercier (1975)
RP2 = 0.639 + 0.355 log (T ) Larsen and 5-22 0.86 27 0.46 41.9
Mercier (1975)
RP3 = -0.056 + 1.40/(1+/p ) Larsen and 5-24 0.88 27 0.47 41.6
w Mercier (1975)
RP4 = -0.009 + 8.17/(10+qs) Larsen and 0.86 27 0.46 41.9
Mercier (1975)
RP5 = -0.249 + 0.487 exp(-0.271 q ) Kirchner and 5-20 0.88 27 0.46 41.9
Dillon (1975)
+0.656 exp(-0.00949 q) Dillon (1975)

RP6 = 0.500 + 0.353 log (z T* w) This study 26 0.53 40.3

RP7 = -0.131 + 1.07 R(K+D) Hand (1975) 0.50 26 0.41 44.4
-0.172(N /Q )(R )V' -
out out (K+D) shape/z

RP8 = 0.734[8.53/(8.53 + q)] 0"91 Chapra (1975) 5-16 26 0.34 67.2


* For each predictive equation (except RP6) the variables used were the same as those


in the original equation, but the equation is altered by new
a and b, such that RP. = a + b [Origihal Variable].
1


regression constants
















1.0







0.8







Z 0.6
0


0

0.4







4J
.H

P-


0.0






-0.2


-0.2 0.0 0.2 0.4 0.6 0.8


Figure V-7.


Measured phosphorus



Predicted vs. measured
Equation RP6.


retention coefficient, R
P


phosphorus retention coefficients using


- 84 -


1.0











The predictive equations examined here to estimate Rn similarly were
based on hydrologic and morphologic characteristics (RN1 to RN5). These
equations (Table V-4), like those examined to predict Rp, also were not very
successful in predicting nitrogen retention coefficients for Florida's NES
lakes. The r2 values of the equations ranged from 0.12 to 0.51, and C.V.
values ranged from 67.0 to 89.6. The width of the 95% confidence intervals
shown for individual predictions in a plot of measured nitrogen retention
(Rn) vs. values predicted by the best equation (RN4) illustrates its poor
predictive ability (Figure V-8). Equation RN4 predicts Rn as a function of
the mean inflow nitrogen concentration (Ln/qs). The inclusion of the areal
nitrogen loading Ln in the predictive equation resulted in a vast improvement
over the other equations that were based entirely on hydrologic and morphologic
data.

The poor correlations between observed and predicted values of Rn and
R using predictive equations RN1-RN5 and RPl-RPB could be due to several
factors. First, it is likely there are substantial errors in many of the
NES water and nutrient budgets, particularly for many lakes where portions
of the nutrient and water budgets were estimated rather than measured direct-
ly. In particular, nutrient inputs were estimated for ungauged tributaries
and for some municipal wastewater treatment plants. It is also possible that
morphologic and hydrologic factors are simply not good predictors of nutrient
retention in Florida's lakes. Unlike temperate zone lakes, most of the
Florida NES lakes are shallow and do not undergo thermal stratification.
Nutrients lost via sedimentation thus may re-enter the water column more
readily in Florida lakes than in temperate zone lakes. Pollman (unpublished
data) has shown that resuspension of sediments in Lake Apopka during storms
causes a significant, temporary increase in the concentration of soluble
reactive phosphorus from sediment particles. Although the effect of this
phenomenon on the long-term retention of phosphorus is not known, the data
suggest that long-term retention may be affected. Biological processes also
may be more important in affecting nutrient retention in Florida lakes than
in temperate zone lakes because of the warm climate, long growing season and
generally nutrient-enriched conditions. For example, release of sediment-
derived phosphorus to the water column by rooted aquatic macrophytes has been
shown to be an important mechanism affecting the concentration of phosphorus
in several lakes (Smith 1978, Lie unpublished ms). It seems reasonable that
macrophytes may have important effects on nutrient retention in many Florida
lakes, although this has not been studied.

Predic- "' of chlorophyll a. The predictive models for chlorophyll %
evaluated here (Table V-5) generally were based on loading expressions devel-
oped by previous investigators, although regressions also were determined
for relationships between chlorophyll a levels and lake nutrient concentra-
tions. CHA12 and CHA13 are regression equations that describe the relation-
ship between chlorophyll a and the concentrations of phosphorus and nitrogen,
respectively, in the Florida NES lakes. As would be expected for a group of
primarily nitrogen-limited lakes, nitrogen is better correlated with chl a
(r2 = 0.79) than is phosphorus (r2 = 0.63). The coefficients for CHA13 are
nearly identical to those of eq. 4-17, which describes the relationship between
(chl a) and (N)1 for the 44 nitrogen-limited lakes-in the entire set of 101
study lakes.


- 85 -













Table V-4. Predictive equations for nitrogen retention coefficient


Predictive Equation* Original Basis n r C.V.
for Equation

RN1 = 0.445 0.189 log(q s) Larsen and 25 0.12 89.6
Mercier (1975)

RN2 = 0.391 + 0.210 log(T ) Larsen and 25 0.16 87.5
w Mercier (1975)

RN3 = 0.322 + 0.198 log(zT w) This study 25 0.18 86.6

RN4 = 0.010 + 0.597 log(L /q) This study 25 0.51 67.0

RN5 = 0.105 + 0.775 R (KD) Hand (1975) 24 0.20 88.3
0.124 (N /Q ) (R
out out (K+)shape/z
shape/z



* For RN1, RN2, and RN5 the variables used in the predictive equation were the same
as those used in equations RP1, RP2, and RP7, respectively.















1.0






0.8







0.6



0)
0)

0 0.4







4-J


0)
a-)








-0.2


-0.2 0.0 0.2 0.4 0.6 0.8 1.0

Measured nitrogen retention coefficient, R
n


Figure V-8. Predicted vs. measured nitrogen retention coefficients using
equation RN4.


- 87 -











Table V-5. Predictive equations for chlorophyll a concentration.


Original Equation Modified Equation
Eqn. no. 2 2
Predictive Equations* Investigator in text r n r C.V.
0 2


CHAl=81.3[L /q (1 + /w] ")
ps[ L ]0.54

CHA2=91.6 8.53+q 0.549


CHA3=21.9rL n 0.786
CHA 5.49+qs


CHA4=9.82[L n(1-R )/qs 1.58

CHA5=89.3[Lp (1-R )/q ] 0.604
p p s
CHA6=8.30[L /(0.65z+qs)] 1.71

CHA7=73.4[L /(0.65z+q)] 0.667

CHA8=97.5
18Z-+ +1.05q exp(0.012q
10+z
Lz 10.657
CHA9=78.7 LP 0.454 )
qs (l+0.824T w

CHA10=59.2[L /qs] 0.663

CHA11=5.62[Ln /qs] 1.61

CHA12=101(P)1 0640

CHA13=11.5(N) 1.60

CHA14=97.8[N /3+P )/ /Q ] 1.04
out out shape/z out


Vollenweider
(1976)

Vollenweider
(1975)


5-30d


5-34a


-Dillon and
Riler (1974a)
Dillon and
Rigler (1974a)
Jones and
Bachmann (1976)
Jones and
Bachmann (1976)
Reckhow (1977)


Walker (1977


This study

This study

This study

This study

Hand (1975)


0.75 29 0.59 21.1


29 0.52 22.8


29 0.32 27.1


- 26

- 27

- 26

- 29

- 29


0.60

0.52

0.69

0.60

0.61


19.0

22.7

19.2

20.9

20.5


- 29 0.59 21.1


- 29

- 26

- 40

- 39

25 0.94 26


0.57

0.58

0.63

0.79

0.59


21.7

22.5

20.0

14.7

20.1


b


* For CHA1 CHA9, and CHA14, CHA. = a [Original Equation] where a and b are constants determined
by the regression. However, in the case of CHA1, CHA2, and CHA14, a and b are the result of multi-
plication by the original regression constants also.










For predictive models based on loading terms, CHA6 gave the best re-
sults (r = 0.69). This equation is based on the equation of Jones and
Bachman (1976) with L substituted for Lp. A plot of predicted versus
measured values of chlorophyll a using CHA6 (Figure V-9) shows the 95% CLI
for this equation. Thus, for a lake having a predicted chlorophyll a con-
centration of 20 pg/L, the 95% CLI is 5 to 74 pg/L. Although this level of
predictibility is useful when considering the range of chlorophyll a values
in the entire data set (3 to 208 pg/L), further refinement of models to pre-
dict chlorophyll a concentrations from nutrient loading data is needed.


NUTRIENT LOADING CRITERIA FOR FLORIDA LAKES


One of the major objectives of the project this report summarizes was
to develop nutrient loading criteria for Florida lakes. In this section,
existing nutrient loading criteria (Vollenweider 1968; Shannon and Brezonik
1972; Vollenweider 1975; Dillon 1975) are analyzed for their ability to pre-
dict trophic status in the Florida NES lakes. The phosphorus loading cri-
teria based on mass balance models are then modified, using the predictive
equations developed earlier in the previous section, to improve their pre-
dictive ability for the Florida NES lakes. Since many of Florida's lakes
are nitrogen-limited, nitrogen loading criteria have been developed using
I/O models analogous to those used for the development of phosphorus loading
criteria.

The loading criteria are evaluated according to their ability to pre-
dict the trophic status of the Florida NES lakes. In addition, a trophicc
ratio", defined as the ratio of a lake's nutrient loading to the minimum
eutrophic loading for that lake, is used to evaluate the degree of eutrophi-
cation predicted by each model.

In earlier developments of nutrient loading criteria, the terms "ex-
cessive" and "permissible" have been used to describe the minimum loading
levels that result in eutrophic and mesotrophic conditions, respectively,
(Vollenweider 1968, 1975; Dillon 1975; Larsen and Mercier 1975). These
terms invoke a value judgement that many aquatic scientists now regard as
unnecessary and unjustified. Thus, in this report we have used "minimum
eutrophic loading" (MEL) to designate the minimum loading required to cause
eutrophic conditions and "minimum mesotrophic loading" (MML) to designate
the minimum loading required to cause mesotrophic conditions. These terms
correspond respectively to the "excessive" and "permissible" loadings of
earlier investigators.


Phosphorus Loading Models.

Vollenweider (1968) Loading Model. The Vollenweider (1968) loading
model delineates trophic state as a function of mean lake depth. The mini-
mum eutrophic loading and the minimum mesotrophic loadings are:

MELp = 0.05z0.6 (5-34a)

MELp = 0.025 0.6 (5-34b)


- 89 -














1000









100 *








10 ee *


*
0
- 10 -











r2 0.69
95% CLI shown





0
0 1 10 100 1000

Measured chlorophyll a, ig/L


Figure V-9. Predicted vs. measured chlorophyll a using equation CIIA6.


- 90 -










According to the Vollenweider (1968) criteria (Figure V-10), all of
the Florida NES lakes are eutrophic. Thus, these criteria are conservative
with respect to the classification of the mesotrophic lakes, since all of
the mesotrophic lakes lie above the minimum eutrophic loading line (MEP ).
The failure of the Vollenweider (1968) model is not surprising since this
model does not include any hydrologic variables. The Florida NES lakes are
extremely diverse with respect to hydrologic conditions (0.03 yr < T < 2.90
yr), and hydrologic conditions are an important factor affecting trophic
state in these lakes.

Shannon and Brezonik (1972) Model. Phosphorus loading criteria were
developed by Shannon and Brezonik (1972) for Florida lakes using a data
base of 55 lakes in the northern part of the state. Their criteria were
based on volumetric loading rates (0.22 and 0.12 g P/m3-yr, respectively,
for excessive and permissible loading rates), and like the original Vollen-
weider criteria, they ignore hydrologic conditions. Although these criteria
are more successful in predicting the trophic status of mesotrophic NES
lakes than are Vollenweider's criteria, the extent of eutrophication ex-
pressed is excessive for many lakes (Table V-6), particularly those with
high flushing rates. For the lakes in which Tw < 0.10 years (Monroe, LC
29; Howell LC 32; Banana, LC 33; Trout, LC 36; Lawne, LC 37 and Munson, LC
38) the degree of eutrophication expressed by the Shannon and Brezonik cri-
teria is greater than that expressed by the models that incorporate hydro-
logic variables.

Vollenweider (1975) Model. The criteria proposed by Vollenweider (1975)
are based on the equation:
Lp
[P]1 z + q (5-35a)
p 2 s

Vollenweider found o a 10 m/yr for his study lakes. He also considered
"permissible" (i.e., minimum mesotrophic) and "excessive" (i.e., minimum
eutrophic) levels of total phosphorus to be 0.01 mg/L and 0.02 mg/L, respec-
tively. Substituting these values into eq. 5-30a produces

MEL = 0.20 + 0.02 qs (5-35b)

MML = 0.10 + 0.01 qs (5-35c)

These criteria (Figure V-11) are an improvement over Vollenweider's 1968
criteria, although most of the mesotrophic lakes still appear in the eutro-
phic zone.

In order to improve this model for application to Florida lakes, two
modifications were made. First, the concentration criteria for phosphorus
were revised to account for the higher level of phosphorus associated with
a given concentration of chlorophyll a in Florida lakes than found in most
temperate lakes. For the phosphorus-limited lakes in the study set (n = 33),
the relationship between phosphorus and chlorophyll a is given by equation
4-14:

(chl a) = 0.195 [P]1 (4-14)'


- 91














*38 H


EUTROPHIC ZONE


* 33H


* 32E


.35H -25H
37E *29E


.34H
*30E


.21E
*28E


*9E *27E


20H
*26E
*23E


*15E 2M
I IE'19E*. 17E
2 **..18E
22H 3M
*8E*5M


*6M-E 13E
IOM


S10
E


-50
z




0

I I
C-
(n
0
I
n- 0.5


-J




I 0.1


OLIGOTROPHIC ZONE


MEAN DEPTH (1),m


Figure V-10. Trophic state delineation of the Florida NES lakes by the
Vollenweider (1968) phosphorus model.


- 92 -


*14E
w4M











trophic ratios and trophic state classifications.


LC EPA-NES TSI(AVG) Vollenweider Shannon and Vollenweider (1975) Dillon (1975)
Assessment Brezonik Original Modified Original Modified
(1972a)


(0-M)
(M)
(M)
(M-E)
(M)
(E)
(E)
(E)
(E)
(E)
(E)
(E)
(E)
(E)
(E)
(E)
(E)
(E)
(E)
(H)
(E)
(H)
(H)
(E)
(E)
(H)
(E)
(E)
(E)
(E)
(E)
(E)
(H)
(H)
(E-H)
(E-H)
(H)
(H)
(H)
(H)


. 98 (E)
7.93 (E)
.52 (E)
1.81 (E)
5.19 (E)
.22 (E)

.89 (E)
42.9 (E)

7.85 (E)


3.02
2.56


7.97
1.35
11.5
18.9
64.8
9.16


208
19.7
3 .8
5 .8
208
68.8
6- 8
675
987
83.1
24

284
1197


0.45 (0)
1.21 (E)
0.95 (E)
0.34 (0)
0.82 (M)
0.76 (M)

1.24 (E)
7.39 (E)

1.24 (E)


0
0
2


0.76 (M)
2.57 (E)
1.48 (E)
0.49 (0)
1.52 (E)
1.05 (E)

1.40 (E)


3.13 (E)


0.51 (M)
1.75 (E)
0.97 (M)
0.34 (0)
1.02 (E)
0.71 (M)

0.93 (M)


2.0 (E)


1.31 (E)
1.28 (E)
2.15 (E)
1.69 (E)
1.82 (E)
2.67 (E)

1.78 (E)


3.16 (E)


0.61 (M)
0.59 (M)
0.98 (M)
0.80 (M)
0.85 (M)
1.25 (E)

0.82 (M)


1.53 (E)


.45 (0) 1.25 (E) 0.88 (M) 1.29 (E) 0.61 (M)
.52 (M) 0.63 (M) 0.43 (0) 2.12 (E) 0.99 (E)
.16 (E) 3.22 (E) 2.17 (E) 5.00 (E) 2.32 (E)


1.17
0.90
1.32
2.77
9.76
1.68
2.21

32.7
3.01
5.09
9.26
37.3
15.6

106
224
17.5
47.4

62.1
231


3.28
2.66
3.88
6.12
6.82
2.94
3.80

27.0
5.06
5.81
16.2
9.77
12.5

72.4
95.4
14.2


35.5
63.8


2.28
1.84
2.73
4.13
4.28
2.08
2.51

17.11
3.35
3.72
10.91
6.04
8.48

45.58
61.39
9.60


23.18
39.64


5.02 (E:
1.62 (E:
9.25 (E'
5.58: (E:
4.55 (E'
21.2 (E:
4.79 (E:

36.8 (E:
9.87 (E:
8.07 (E:
8.15 (E:
12.8 (E:


66.4 (E
86.0 (E
22.5 (E'


20.5 (E)


2.35 (E)
0.75 (M)
4.33 (E)
2.61 (E)
2.18 (E)
9.73 (E)
2.23 (E)

16.68 (E)
4.62 (E)
3.72 (E)
3.80 (E)
5.42 (E)


31.72 (E)
41.60 (E)
10.66 (E)


9.37 (E)


Table V-6. Comparison of phosphor s models by




Full Text

PAGE 1

WATER IiRESOURCES researc center Publication No. 56 NUTRIENT LOADING TROPIC STATE RELATIONSHIPS IN FLORIDA LAKES by Lawrence A. Baker Patrick L. Brezonik & Charles R. Kratzer UNIVERSITY OF FLORIDA

PAGE 2

NlITRIENT LOADING TROPHIC STAlE RELATIONSHIPS IN FLORIDA LAI
PAGE 3

TABLE OF CONTENTS Abstract Acknowledgements Chapter I. II. III. IV V. Introduction Conceptual framework for predictions of lake trophic status Scope and objectives of this report Data Sources and Methods Data sources Statistical methods Predictions of Nutrient Loading for Florida Lakes Introduction Literature review: nutrient sources to Florida lakes Point source loadings Precipitation inputs Non-point source loadings from Florida watersheds Statistical analysis of nutrient export from the NES watersheds Equations to predict phosphorus loading (TPL) Predictions of nitrogen loading (TNL) Equations to predict flow (FLOW) Application timnological Characteristics of Florida Lakes Morphometric', characteristics Chemical and physical characteristics Relationship between pH and alkalinity Factors affecting transparency Biological characteristics Phytoplankton communities Fish populations Nutrient limitation The relationship between nutrients and chlorophyll a standing crop Analysis of seasonal trends Development of a trophic state index Application of Nutrient Loading Models to the Florida NES Lakes Introduction Historical development ii 1 1 2 4 4 11 12 12 13 13 15 15 23 27 30 30 31 40 40 42 42 42 44 44 47 47 53 59 61 67 67 67

PAGE 4

Chapter Phosphorus input/output models Prediction of Rp Prediction of chlorophyll concentration Nutrient loading criteria:/-graphical approaches Application of nutrient loading models to Florida lakes Total phosphorus concentration Total nitrogen concentration Phosphorus and nitrogen retention coefficients Prediction of chlorophyll a Nutrient loading criteria for Florida lakes Phosphorus loading models Nitrogen loading models Application "SUmmary and Conclusions References Appendices iii 67 72 72 75 75 75 77 80 85 89 89 98 107 114 121

PAGE 5

ABSTRACT Quantitative relationships among important lake trophic state indicators and watershed enrichment factors were examined using a data base of 101 Florida lakes. Trophic state information was obtained from three previous surveys on Florida lakes, from which data were compiled into a uniform format. Watershed nutrient export data and lake nutrient loading rate data were obtained from a comprehensive review of the literature and from the Florida portion of the National Eutrophication Survey (NES) conducted by the u.S. EPA. This data base was used to determine relationships between non-point source (NPS) nutrient loading rates and land use characteristics of Florida watersheds; to evaluate interrelationships among trophic state indicators in Florida lakes; and to revise nutrient loading models and develop appropriate nutrient loading criteria for these lakes. The magnitude of NPS nutrient loadings was estimated from published export coefficients and by a statistical analysis of the NES watersheds. The literature-based approach produced a wide range of export coefficients for Florida watersheds. 0.2-0.7 kg P/ha-yr and 1.5-6.1 kg N/ha-yr for forests; 0.4-2.4 kg P/ha-yr and 2-50 kg N/ha-yr for cropland; 0.2-4.7 kg P/ha-yr and 1.5-7.4 kg N/ha-yr for residential areas; and 0.3-7.5 kg P/ha-yr and 3-10 kg N/ha-yr for urban areas. NPS nutrient loading (dependent variables) and land use characteristics (independent variables) for 41 NES watersheds were analyzed by stepwise multiple regression to improve the predictive capability of the land use-nutrient loading approach. For phosphorus, a model using three land terms (cropland, forest & rangeland) explained 72% of the variance in NPS loading (vs 21% for a model with drainage area as the sole independent variable). Models to predict NPS nitrogen loading and hydraulic flow had high levels of predictability using drainage area as the sole independent variable (r2 = 0.84 and 0.91, respectively); and inclusion of land use data resulted in little predictive improvement. Evaluation of the limnological characteristics of 101 Florida lakes indicated that most of these lakes are shallow and well-mixed; few have stable thermoclines or anoxic hypolimnia. The lakes are highly variable in alkalinity (0-16mg/Las CaC03), pH (4.7>10) and amount of color (2-54 CPU), reflecting differences in geological origin and watershed characteristics. The majority of the lakes in the data base are eutrophic, and unlike most temperate lakes, tend to be nitrogen-limited (46% had SIN:SRP ratios of <10:1). For a given level of total phosphorus, Florida lakes have less chlorophyll a than do temperate lakes; this is true even for phosphorus-limited Florida lakes. Carlson's trophic status index (TS1) was modified for application to Florida lakes by inclusion of a nitrogen index to reflect the importance of nitrogen as a limiting nutrient. A composite TS1 was developed by averaging the TS1's based on Secchi disk, chlorophyll a and nutrient concentration (the smaller of the nitrogen or phosphorus index) to produce an index that reflects the multidimensionality of the eutrophication concept. Various nutrient loading models were evaluated statistically for their ability to predict mean chlorophyll a, nitrogen and phosphorus concentrations in Florida lakes. The best predictions for nitrogen and phosphorus were made iv

PAGE 6

using a modified Dillon and Rigler-type model, while the best predictions of chlorophyll a were obtained using a Jones and Bachman-type model. Existing phosphorus loading criteria were evaluated for the Florida NES lakes. The 1975 Vollenweider criteria and the 1975 Dillon criteria were revised to improve their predictive capabilities; in both cases the revisions resulted in higher critical values (minimum mesotrophic and minimum eutrophic loading rates) than the original criteria. Both sets of criteria were equally successful in delineating eutrophic lakes from mesotrophic lakes in the NES data base. Nitrogen loading criteria were developed using loading terms analogous to those used for phosphorus. The most successful nitrogen criteria were based on a Dillon-type model. In evaluating the impact of a proposed management strategy it is suggested that both nitrogen and phosphorus loading criteria be used; the correct response of the lake will be obtained with the criterion that predicts the lower trophic status. v

PAGE 7

ACKNOWLDGEr-BnS Appreciation is extended to Dr. Jack Gakstatter of the U.S. EPA Corvallis laboratory for supplying NES data and aerial photos for analysis of watershed land use. Ms. Janet Denger of the University of Florida Remote Sensing tory performed the land-use analyses on the NES watersheds. Ms. Dorthy Murphy assisted in compiling data for computer analysis. The statistical advice and assistance provided by Dr. Wayne C. Huber and Mr. Gary Goforth of the Department of Environmental Engineering Sciences and by Mr. Mike Conlon of the De partment of Statistics is gratefully acknowledged. Most of this report was typed by Ms. Alicia Maxwell, and the task was completed by Ms. Patty Hersey. vi

PAGE 8

CHAPTER I I INTRODUCTION Inland lakes are an important natural resource for Florida, and they are particularly valuable recreational assets. Unfortunately, the potential beneficial use of many Florida lakes has by accelerated eutrophication. Although the process of eutrophication is natural, the addition of plant nutrients from municipal sewage, septic tanks, urban and agricultural runoff, livestock operations, and industrial effluents accelerates the process by stimulating the growth of algae and macrophytes. Excessive levels of aquatic production result in general impairment of water uses. Under highly enriched conditions, blue-green algae tend to dominate the algal flora and may form dense, unsightly surface blooms that impart unpleasant tastes and odors to the water. In lakes that stratify, the decomposition of algae and macrophytes results in the depletion of dissolved oxygen in the bottom waters, and this in turn limits the diversity of benthic organisms and benthic feeding fish. The quality of fishing eventually decreases, as populations of game fish, such as bass and sunfish, are replaced by populations of rough fish (bullheads, shad, carp). Recreational boating may become restricted by thick beds of weeds. sports are diminished by the reduced transparency of the water, the growth of water weeds, and the occurrence of infections in swimmers. One of the most notable examples of cultural eutrophication in Florida is Lake Apopka, a 12,000 hectare (ha) lake once recognized for its exceptional bass fishery. Since the late 1940's, water quality in the lake has deteriorated as the result of discharges of municipal wastewater, citrus processing plant wastes, and muck farm irrigation water to the lake, in conjunction with ill-fated attempts tocoutrolthe growth of water hyacinth and rough fish. As the result of these perturbations, populations of important game fish have been drastically reduced, and the dominant fish in the lake today are two species of shad (USEPA 1978). Cultural eutrophication has affected other lakes in the Oklawaha chain (Brezonik and Shannon 1971), Lake Okeechobee (MacGill et al. 1976), and numerous smaller lakes throughout the state. In order to manage Florida's lakes and control eutrophication, planners and regulatory agencies must be able to quantify nutrient loadings to lakes from their watersheds and to predict the response of individual lakes to changes in nutrient loading. CONCEPTUAL FRAMEWORK FOR PREDICTIONS OF LAKE TROPHIC STATUS The most widely used approach in predicting the trophic status of lakes has involved the use of simplified input-output (I/O) models (Dillon and Rigler 1974a; Chapri and Tarapchak 1976; Vollenweider 1968, 1969, 1975, 1976; Kratzer 1979). The I/O models are based on the assumption that a lake is a continuously-stirred tank reactor (CSTR) in which all nutrient fluxes are at steady state. Since phosphorus is the most common limiting nutrient in temperate zone lakes, most of these models have been developed to predict mean lake phosphorus concentrations in a lake. The utility of these models -1

PAGE 9

is enhanced by their modest data requirements, which include only morphologic and hydrologic parameters and areal phosphorus loading rates. Phosphorus loading models have been further advanced by the recognition of statistically significant relationships between concentrations of total phosphorus and the concentration of chlorophyll a in lakes (Sakamoto 1965; Dillon and Rigler 1974). This development has led to theiolrmailrl3.;t'itrn:;uf I/O models in which phosphorus loading can be used to predict the mean chlorophyll concentration directly (Vollenweider 1976; Chapra and Tarapchak 1976; Kratzer 1979; Uttormark and Hutchins 1978). Finally, several authors have found strong correlations between Secchi disk transparency and C:hlorophyll a concentration (Carlson 1977, Brezonik 1978), enabling predictions of lake transparency from chlorophyll a data. A conceptual framework for predic-tion of lake trophic status based on I/O models is outlined in Figure I-I. The need for nutrient loading data in the I/O models has prompted efforts to predict the non-point source (NPS) loading of nutrients from tributary watersheds based on watershed characteristics. The simplest approach for estimating NPS nutrient loading is based on relationships between land use and nutrient loading. Numerous studies have been conducted to determine the areal loading of nutrients from agricultural, forested and urban watersheds. Omernik (1976, 1977) used a statistical approach to evaluate nutrient export from watersheds as a function of land use. Reviews of areal nutrient export rates for various land use categories have been assembled by Loehr (1974) ,'Uttormarket ale (1974)c-and Reckhowet al. (1980). SCOPE AND OBJECTIVES OF THIS REPORT This report builds on and updates the earlier work of Brezonik and Shannon (1971) in assessing nutrient loading-trophic response relationships and developing critical loading rate:guidelines for Florida lakes. Since that early report, a large volume of data has been collected on several aspects of eutrophication in Florida. Several studies of non-PQint source nutrient loading have provided data on nutrient export rates for watersheds in various land uses, and additional data are available on the trophic status, nutrient budgets and water budgets for many Florida lakes. During the period in which this additional data has become available the I/O models for predicting lake trophic status have been developed. While these models have been found to be quite useful for the temperate zone lakes in which they were developed, the applicability of the models and the nutrient loading criteria developed from them has not been evaluated for the warm temperate and subtropical Florida lakes. Thus, the objectives of this study are: 1) to examine the relationships between land use and nutrient export in Florida watersheds; 2) to examine the general limnological characteristics of Florida lakes, particularly with respect to relationships among trophic status indicators and to compare these relationships to those found in temperate lakes; -2

PAGE 10

3) to assess the applicability of existing I/O models to Florida lakes, and 4) to develop critical nutrient loading guidelines for Florida lakes. DATA REQUIREMENTS Watershed characterist Nutrient inputs from direct precipitation and point source disch Morphologic and hydrol characteristics of lak ics arges ogic e PREDICTION Export of nutrients from tributary water-sheds Total loading of nutrients to lake Average annual concentration of nutrients in lake Average annual concentration of chlorophyll a in lake Secchi disk transparency Figure 1-1. Conceptual framework for predictions of lake trophic status. 3

PAGE 12

CHAPTER I I. DATA SOURCES AND METHODS DATA SOURCES In order to accomplish the objectives of this study, limnological data from three synoptic studies were compiled into a uniform format for computer analyses. The first of these studies (Brezonik and Shannon 1971) involved an evaluation of trophic conditions of 55 lakes in north and central Florida. Three groups of lakes were included in this study: 1) 16 primarily oligotrophic lakes in the Trail Ridge Region of Putnam and Clay counties, 2) 33 lakes and ponds throughout Alachua County and 3) six lakes of the Oklawaha chain. Samples were collected four times over a one-year period from 36 of the lakes, while the remaining 19 were sampled at two-month intervals to obtain more detailed data on seasonal trends. The second study from which lake data were obtained was the Florida National Eutrophication Survey (NES), conducted by the EPA during the mid1970's. Lakes included in this survey were selected on the basis of three criteria: 1) lakes impacted by one or more sewage treatment plant outfalls within 40 km (25 miles) of the lake waters; 2) lakes having a surface area greater than 40 ha (100 acres); and 3) lakes with residence times of greater than 30 days. In addition to the lakes that met these criteria, five lakes of special interest (South Lake, Lake Yale, Glenada Lake and Horseshoe Lake) were included in the Florida NES, for a total of 40 lakes. Most of the lakes are located in central Florida, particularly in Polk, Orange, Lake, Osceola, and Seminole Counties. Each lake was sampled from two to four times (usually three times) during 1973. The third set of data is from a recent study of the effects of acid rain on softwater lakes in Florida (Brezonik et al. '[.'981) Twelve of the lakes included in this study are located in the Trail Ridge Region of north Florida and eight are located on the Ridge in Highlands County (south-central Florida). Samples were collected every three months during 1979-80. Since several lakes were included in more than one survey, only the more complete data set was used in this report. The resulting data base contains 101 lakes whose locations are shown in Figure II-I. The variables measured in each survey are listed in Table II-2. In compiling the limnological set, we used mean values for the entire water column. This approach is considered valid for Florida lakes, most of which are generally shallow and do not exhibit seasonal thermal stratification. Although the sampling methods differed slightly among the three studies from which data were obtained, these differences are not likely to be important with respect to the applications of the data in this report. The limnological data (annual means) compiled for the study lakes are presented .in Appendix A-I; morphological data are presented in Appendix A-2. -4

PAGE 13

Table II-I. Florida study lakes. Lake Lake Code Lake Name Count! Code Lake Name 1 Minneola Lake 51 Moss 2 East Lake Tohopekaliga Osceola 52 Jeggord 3 Minnehaha Orange 53 Still Pond 4 Weohyakapka Polk 54 Lochloosa 5 Tarpon Pinellas 55 Orange 6 Istokpoga Highlands 56 Palatka Pond 7 Yale Lake 57 Newnan's 8 Kissimmee Osceola 58 Mize 9 Jessie Polk 59 Calf Pond 10 'Horseshoe Seminole 60 Unnamed 20 11 Haines Polk 61 Meta 12 South Brevard 62 Alice 13 Okeechobee Glades, Hendry 63 Bivens Arm Okeechobee 64 Clear 14 Marion Polk 65 Unnamed 25 15 Crescent Flagler, Pl.\tnam 66 Beville's Pond 16 Poinsett Brevard, Orange Unnamed 27 17 Doctors Clay 68 Kanapaha 18 Reedy Polk 69 Watermelon Pond 19 Gibson Polk 70 Long Pond 20 Dora Lake 71 Burnt Pond 21 Talquin Gadsden, Leon 72 Wauberg 22 Apopka Lake, Orange 73 Tuscawilla 23 Griffin Lake 74 Harris 24 Glenada Highlands 75 Eustis 25 Thonotosassa Hillsborough 76 Weir 26 Seminole Pinellas 77 Kingsley 27 George Putnam, Volusia 78 Sand Hill (Lowry) 28 Tohopekaliga Osceola 79 Magnolia 29 Monroe Seminole, Volusia 80 Brooklyn 30 Hancock Polk 81 Geneva 31 Eloise Polk 82 Swan 32 Howell Orange, Seminole 83 Wall 33 Banana Polk 84 Santa Rosa 34 Jessup Seminole 85 Adaho 35 Alligator Columbia 86 McCloud 36 Trout Lake 87 Anderson Cue 37 Lawne Orange 88 Suggs 38 Munson Leon 89 Long 39 Effie Polk 90 Winnot 40 Lulu Polk 91 Cowpen 41 Santa Fe Alachua 92 Gallilee 42 Little Santa Fe Alachua 93 Annie 43 Hickory Pond Alachua 94 Clay 44 Altho (Alto) Alchua 95 Francis 45 Cooter Pond Alachua 96 Johnson 46 Elizabeth Alachua 97 Josephine 47 Clearwater Alachua 98 June 48 Hawthorne Alachua 99 Letta 49 Little Orange Alachua 100 Placid 50 Unnamed 10 Alachua 101 Sheeler (l)Lake codes: 1-40 = NES lakes, 41-92 = 55 lakes study (Brezonik & Shannon 1971), 93-101 = Brezonik et al. (1981). -5 -Count! Alachua Alachua Alach,ua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Alachua Lake Lake Marion Putnam Clay Clay Clay Clay Putnam Putnam Putnam Putnam Putnam Putnam Putnam Putnam Putnam Putnam Putnam Highlands Highlands Highlands Clay Highlands Highlands Highlands Highlands Clt!Y

PAGE 14

I I -.i Fig. II-I. Location of 101 study lakes. -6

PAGE 15

Table 11-2. Data collected in synoptic_studies of Florida lakes. Watershed Drainage area Brezonik & Shannon, 1971 Nation EutroBrezonik, et al., phication Survey 1980 (1) Land use characteristics Hydrologic & morphologic Water budget Volume Surface area Max. depth Mean depth Nutrient budget Chemical & physical Total nitrogen NH+ N03 4 N03 Organic N Total phosphorus Orthophosphate COD pH Alkalinity Acidity Color Dissolved oxygen Specific conductance Major ions (Ca, Mg, K, Na, Fe, S04' Cl, Si) Total organic carbon Total inorganic carbon Trace metals Secchi disk transparency Temperature Suspended solids Total solids Turbidity Biological Chlorophyll a Carotenoids -Algal identification & counts Primary productivity Zooplankton identification & counts Limiting nutrient bioassays Sediments Sediment type (visual c1assfication) Benthic organisms Chlorophyll derivatives 'Total carotenoids Volatile solids nitrogen NH Toia1 phosphate Iron Manganese x X X X X X(l) X X X X X X X X X X X X X X X X(2) X X X X X X X X X X X X X X X X X X X X X X X X X X X x X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X (3) X X X X X X X X (1) Input of nutrients computed from' land ,use and ,pop,jlat.1on (2)Includes Mn, Cu, Zn, Fe and Sr (3) Al only. -7

PAGE 16

In addition to the limnological data, the EPA also computed nutrient budgets for 34 of the 40 Florida NES lakes. Methods used to compute nutrient budgets are described in NES Working Paper No. 175 (NES 1975). In this study, non-point source (NPS) loadings of nitrogen and phosphorus for each tributary were computed by subtracting the reported point source loadings from the total tributary loadings. Runoff for each tributary was computed by dividing the total streamflow by the watershed area. Data on the nutrient and hydrologic budgets for the Florida NES lakes are presented in Appendix A-3. Determination of land uses in the NES watersheds was made using Mark Hurd photoquads and USGS 7.5 minute quadrangles (both 1:24,000). Photos from the Agricultural Stabilization and Conservation Service and the;State of Florida CITRUS survey were used to provide additional resolution. The land use classification scheme used was a modification of the system developed by Anderson et al. (1976). The modifications involve classification of agricultural land, forests and barren land (Fig. 11-2). Agricultural land was divided into two categories: "cropland and pasture" and "other agriculture". The category "other agriculture" was formed to assess the effects of different types of agricultural land use on NPS nutrient loading. This category is comprised largely of citrus orchards in the NES watersheds. The second modification was a combining of "deciduous forest land", "evergreen forest land" and "mixed forest land" (categories 41 to 43 in Anderson's Level II scheme) with "forested wetland" (category 61) into a composite "forest" category. This was done to facilitate identification of forests in the air photos. Finally, "salt flats", "barren land", "beaches", "other sandy areas", "transitional areas" and "mixed barren land" (categories 71-74 and 76-77) in Anderson's Level II scheme were grouped together with nonresidential urban land uses into an "other urban" land use category. This modification was made because these barren land subgroups were largely comprised of "transitional areas" associated with urban areas. The watersheds used in the evaluation of land use-nutrient loading relationships are listed in Table 11-3. Several NES watersheds were not included in the analyses because runoff from them was found to be abnormally high (> 70% of total precipitation in the watershed). Such high values could result from groundwater inflows from other watersheds or backpumping in agricultural areas; they may simply reflect errors in flow measurement. Since it was impossible to establish the cause of high runoff for individual watersheds in this study, all watersheds in which runoff was 70% of precipitation were excluded from the analyses. This resulted in the deletion of watersheds l3Bl, l8Bl, 22Cl, 37Dl, 32Bl and 34Gl. The 70% criterion was considered a reasonable upper limit for natural runoff from Florida watersheds (W. C. Huber, per. corom., 1980). Several other watersheds were excluded because flow data were not available (3lAl, 40Bl) or because the measured drainage basin area dif"fered imhst'ant:i:ally from that reported by the NES (17Bl). Thus, data from 41 watersheds were used in the analyses of land use-nutrient loading relationships. Land use data and data for non-point source nitrogen and phosphorus loadings for these watersheds are presented in Appendix A-4. -8

PAGE 17

Anderson,. et aI. (1976) This study Level I Urban or built-up land Agricultural land Rangeland Forest "\:
PAGE 18

Table 11-3. NES watersheds used in analysis of land use-nutrient loading relationship. Tributary Code Tributarz Lake 01 Bl Unnamed Minneola 02 Al Boggy Cr. East L. Tohopekaliga 02 Bl Unnamed 03 Bl Unnamed Minnehaha 04 Al Tiger Cr. Weohyakapka 05 Al South Cr. Tarpon 06 Bl Arbuckle Cr. Istokpoga 08 Dl Jackson Canal Kissimmee 09 Al Unnamed Jessie 13 Bl* Unnamed Okeechobee 13 Cl Taylor Creek Okeechobee 13 Dl Lembin Creek Okeechobee 13 F1 Indian Prarie Canal Okeechobee 13 Gl Pond Canal Okeechobee 14 Dl Unnam,ed Cr. Marion 15 Bl Haw Creek Crescent 15 Cl Unnamed Crescent 17 Bl* Swimming Cr. Doctors 18 Bl* Unnamed Reedy 19 Bl Unnamed Gibson 20 Al Dora Canal Dora 20 Bl Unnamed Dora 21 Bl Ockawaha Cr. Talquin 21 Dl Bear Cr. Talquin 22 Cl* Unnamed Apopka 23 Al Dead River Griffin 25 Al Baker Creek Thonotosassa 26 Bl Unnamed Seminole 26 Cl Bayou Cr. Seminole 28 Al Shingle Cr. Tohopekaliga 28 Bl Partin Canal Tohopekaliga 29 Bl Bethyl Cr. Monroe 30 A2 Saddle Cr. Hancock 30 Bl Unnamed Hancock 30 Cl Unnamed Hancock 31 Al* Unnamed Eloise 32 Al Howell Cr. Howell 32 Bl* Unnamed Howell 34 Al Gee Cr. Jessup 34 Bl Soldier Cr. Jessup 34 Cl Unnamed Jessup 34 Dl Howell Cr. Jessup 34 El Salt Cr. Jessup 34 Gl* Sweetwater Cr. Jessup 35 Al Unnamed Cr. Alligator 35 Bl Unnamed Cr. Alligator 35 Cl Unriamed Cr. Alligator 37 Bl Unnamed Canal Lawne 37 Cl Unnamed Canal Lawne 37 Dl* Unnamed Canal Lawne 38 Bl Unnamed Munson 40 Bl* Unnamed Lulu *Hatershed dropped from analyses. See text. 10

PAGE 19

STATISTICAL METHODS Statistical analyses were performed using the Statistical Analysis System (SAS) (SAS Institute, 1979). Where multiple independent variables were used in regression analyses, the STEPWISE (backward) procedure was used to select the significant independent variables. The GLM procedure was used to compute predicted values and confidence limits for final regression analyses. Confidence limits for the mean (CLM) were used when the objective of the regression analysis was to predict the mean response of the dependent variable. Confidence limits for individual predictions (CLI) were used when the objective of the regression model was to predict values of the dependent variables from individual values of the independent variable (Snedecor arid Cochran 1967). In evaluating the nutrient loading models, several lakes were discarded as the result of suspected errors in the data or on the basis of a ical outlier test. The test procedure used was the criterion T defined as: n where Y b o s observed y Y d = predicted y value,artd pre s = population standard deviation excluding the outlier. If the resulting T value of a suspected outlier was greater than the 5% T value from a of critical T values (Grubbs 1969), then the value was B:onsidered an outlier. Although dl:e method is recommended for removing only a single outlier, several outliers can be removed by reevaluating s following the removal of each outlier. In this manner, up to three outliers were removed from anyone prediction equation. The removal of a value as an outlier in one equation did not automatically result in its removal from other equations. 11

PAGE 20

CHAPTER I I I I PREDICTIONS OF NUTRIENT LOADING FOR FLORIDA LAKES INTRODUCTION In studies of lake eutrophication it is often necessary to know the loadings of plant nutrients, particularly nitrogen and phosphorus, into a lake. Ideally, these loadings are determined by obtaining data on water fluxes and nutrient concentrations for all sources (tributary inflows, direct point source discharges and direct precipitation). Unfortunately, the acquisition of these data is expensive and time-consuming, usually requiring at least one year of study. It is therefore desirable to be able to estimate the loadings of nutrients from various sources indirectly. The most difficult component of a lake's total nutrient loading to determine is the non-point source loadings from the surrounding watershed. In this chapter the major emphasis will be to develop a method of estimating non-point source nutrient loadings from tributary watersheds using land use data. The approach used, originally conceived by Lee et a1. (1966), is based on the assumption that nutrient export from a portion of a watershed can be computed as the product of the drainage area and an export coefficient that is determined by land use. Thus for a watershed composed of multiple land uses: n L. = L: S .. A. 1 j =1 1J J where L. 1 S .. 1J = (3-1) loading of constituent i, kg/yr, export coefficient for constituent i from land use j, kg/ha-yr, and A. area of watershed in land use j, ha. J Since this concept was first developed, dozens of studies have been conducted to determine export coefficients for various land uses. These have been compiled by Uttormark et a1. (1974), Loehr (1974) and, most recently, by Reckhow et a1. (1980). Unfortunately, range of reported export coefficients for a given land use varies widely. This variability is expected since the approach does not take into account variations in precipitation, soil type, length of the growing season and other parameters that influence nutrient export. Furthermore, most of the studies included in these compilations have been conducted in temperate climates, limiting the validity of these coefficients for Florida watersheds. In order to determine suitable land-use export coefficients for Florida, two methods were used. First, data from studies of non-point source nutrient loading conducted in Florida were compiled by land use. By examining only -12

PAGE 21

studies that have been conducted in Florida, variability in export coefficients caused by differences in soil type, climate and terrain should be reduced, resulting in a narrower range of export coefficients for each land use. For studies in which export coefficients were not computed, the reported loading data were used to perform the appropriate calculations. In a few cases, data are included for several studies conducted in nearby states, where land uses and other conditions (soils; climate) were judged to be representative of conditions in Florida. The second method used to compute export coefficients was a statistical approach similar to that used by Omernik (1976). In this approach, export coefficients are determined by multiple regression techniques using data on nutrient loading (dependent variable) and land uses (independent variables). For this analysis, data on nutrient loading, nutrient concentration, runoff and land use were compiled for 41 Florida NES watersheds (Chapter 2). LITERATURE REVIEW: NUTRIENT SOURCES TO FLORIDA LAKES Point Source Loadings. The results of a nationwide survey of nu from 809 municipal wastewater treatment plants (Gakstatter et al. 1978) may be used to estimate treatment plants lOlldillgs of these nutrients for preliminary studies on the basis of population served and treatment type (primary, trickling filter, activated sludge, stabilization pond). As seen in Table III-I, the type of treatment has little effect on the phosphorus loading from wastewater treatment plants: median loadings ranged from 0.9 kg/cap-yr for stabilization ponds to 1.1 kg/cap-yr for primary treatment facilities. These results compare favorably with those of Vollenweider (1968), who computed a mean phosphorus loading of 0.8 kg/cap-yr for municipal wastewater from the results of 15 studies reported in the literature (treated and untreated wastewaters were included). Although the type of treatment had little effect on phosphorus loadings, phosphorus loadings in the compilation of Gakstatter et al. were significantly lower for a group of 33 plants that included tertiary phosphorus removal processes (median loading = 0.4 kg/capyr) and for a group of 25 convential treatment plants located in communities having phosphorus detergent bans. The results of Gakstatter et al. indicate that nitrogen loading from wastewater treatment plants is affected by the type of treatment process used. Loadings ranged from 2.0 kg/capita-yr for stabilization pond effluents to 4.2 kg/cap-yr for primary treatment plant effluents. In comparison, Vollenweider (1968) reported a nitrogen loading rate of 3.9 kg/cap-yr. For all four treatment processes, the median effluent TN:TP ratios were less than 5:1, indicating that sewage effluents typically are nitrogen limited. Although the standard errors for loading estimates in Table 111-1 indicate that loadings from municipal wastewater treatment plants can usually be estimated with reasonable accuracy, actual loadings for a given plant may differ significantly from predicted values because of (1) modifications in the design process, (2) hydraulic overloading resulting from stormwater inflows, (3) excessive infiltration or overuse, or (4) impairment of the treatment process by toxic wastes or improper operation. Thus, while the values presented in Table 111-1 may be used for preliminary estimates of nutrient -13

PAGE 22

t-"'" +:Table III-I. Median and mean phosphorus and nitrogen concentrations and median loads in wastewater effluents following four conventional treatment practices(l) Treatment T:lpe Trickling Activated Stabilization Primary Filter Sludge Pond -------------_ ... _-------Number of Sampled Plants 55 244 244 119 Total Population Served 1,086,784 3,459,983 4,357,138 2V10,287 Ortho-P Conc. Median 3.5 + 0.29* 5.1 + 0.21 4.6 + 0.24 3.9 + 0.34 (mg/l) Mean 4.0 + 0.62 5.4 + 0.38 5.3 + 0.40 4.8 + 0.62 Total-P Conc. Median 6.6 + 0.66 6.9 + 0.28 5.8 + 0.29 5.2 + 0.45 (mg/l) Mean 7.7 + 1.19 7.2 + 0.50 6.8 + 0.51 6.6 + 0.81 Total-P Load Median 1.1 + 0.10 1.2 + 0.05 1.0 + 0.06 0.9 + 0.10 (kg/cap-y) Inorganic-N Conc. Median 6.4 + 1.00 7.1 + 0.38 6.5 + 0.45 1.3 + 0.29 (mg/l) Mean 8.3 + 1. 40 8.2 + 0.60 8.4 + 0.69 5.5 + 1.95 Total-N Conc. Median 22.4 + 1.30 16.4 + 0.54 13.6 + 0.62 11.5 + 0.84 (mg/l) Mean 23.8 + 3.48 17.9 + 1.23 15.8 + 1.16 17.1 + 3.59 Total-N Load Median 4.2 + 0.40 2.9 + 0.17 2.2 + 0.15 2.0 + 0.26 TN:TP Ratio Median 3.4 2.4 2.4 2.2 Per Capita Flow Median 473 + 72 439 + 19 394 + 26 378 + 38 (l/cap.d) Value + 1 standard error. (1) From Gakstatter et al. (1978)

PAGE 23

loadings from wastewater treatment plants, they should not be regarded as substitutes for actual measurements when expensive management decisions are made. Precipitation Inputs. For many lakes, particularly seepage lakes with long detention times, bulk precipitation (wetfall + dryfall) may be a major source of nutrients. A recent study on the chemical composition of bulk precipitation in Florida (Brezonik et al. 1981) includes data on nitrogen and phosphorus loadings for 24 sites throughout Florida, providing a data base that can be used to estimate precipitation loadings to lake surfaces (Table 111-2) The mean deposition rate for nitrogen was 0.76 g N/m2-yr, but the deposition rates at individual stations varied considerably. The lowest nitrogen deposition rate, 0.32 g N/m2-yr, occurred at Bahia Honda Key, while the highest rate, 1.13 g N/m2-yr, occurred at Belle Glade in the intensively cultivated Everglades Agricultural Area. When sampling stations were grouped according to location and local land use, deposition rates were found to be lowest at the coastal and non-agricultural sites and highest at the agricultural sites (Table 111-2). On a statewide basis, 69% of the total nitrogen deposited was in the form of inorganic species, and speciation generally followed the sequence NHt>NO;> organic N. Phosphorus deposition was also site dependent and ranged from 17 mg P/m2-yr at Bahia Honda to 111 mg P/m2;....yr at Jasper, with a mean of 51 mg P/m2-yr. As with nitrogen deposition, phosphorus deposition was generally lowest at the coastal and non-agricultural rural sites and highest at agricultural sites. Soluble reactive phosphorus was the dominant species, accounting for 68% of the total phosphorus in bulk precipitation. The significance of nutrient inputs from precipitation with respect to lake eutrophication can be evaluated by comparing the magnitude of these loadings with critical loading values. For example, using Vollenweider's (1968) original critical loading criteria, a lake a mean depth of 3 m has a critical loading of 97 mg P/m2-yr and 1.45 g N/m2-yr. According to these criteria, the mean precipitation loading for Florida corresponds to 53% of the critical loading for phosphorus and 52% of the critical loading for nitrogen. Thus, precipitation inputs of nutrients may be a significant component of the total nutrient budget for a lake. Non-point soux>ce loadingai!'ortJ.FZoridawatei'$heds. Export coefficients for nitrogen and phosphorus determined for Florida watersheds are presented by land-use category (urban residential, agricultural, forest) in Tables 111-3 to 111-6 For comparison, the results of a literature review of export coefficients for studies conducted throughout the u.S. and Canada (Reckhow et al. 1980) are presented. Before discussing the magnitude of these export coefficients, several comments are in order concerning the nature of these studies. First, the methods used to determine nutrient loading are highly variable among investigators, and some are based on rather limited data. For example, Lamonds (1974) based his estimates of Nand P export from a residential area in Eustis, Florida, on data collected during only seven storms, while Wanielista 15

PAGE 24

Table 1II-2. Atmosplieric (via bulk precipitation) of total nitrogen and total phosphorus at stationsgrouped according to dominant land use in the area.(l) TN TP kg/ha-yr kg/ha-yr Coastal 5.8 0.31 Urban 7.6 0.50 Rural (non-agricultural) 6.2 0.27 Rural (agricultural) 8.8 0.66 State average 7.5 0.51 (1) From Brezonik et al. (1981) 16

PAGE 25

I-' -...J Table III-3. Nutrient export from urban areas. Source Burton and Turner (1977) Wanie1ista (1977) Miller et ale (1979) Location Nr. L. Jackson Fla. Orlando, Fla. Ft. Lauderdale, Fla. Total p 7.49 3.5 0.26 Loading (Rg/ha-yr) Ortho Total N Org. P N 0.19 0.37 (25%) (inorg.) 2.0 10 57% (TKN + NO;) 0.11 2.88 2.12 42% (74%) NO-.,..N 3 0.18 (0.02 NO;) 6 0.55 (.042 N02) (19.1%) + NH-N 4 0.17 0.10 (3.5%) Comments 80% residential & commercial. 423 samples analyzed Commercial area. Only two storms sam-pled -results ex-trapolated. 97.9% impervious area 31 storms sam-pled. Loadings calc. from raw data.

PAGE 26

I--' 00 Table 111-4. Nutrient export from residential areas. Source Burton and Turner (977) Hanielista et a1. (1977) Bedient et al. (1978) Mattraw & Sherwood (1977) Lamonds (1974) Location Vicinity of L. Jackson, Fla. Near Orlando, Fla. Houston, Texas Broward Co., Fla. Eustis, Fla. Total P 4.74 2.24 0.745 0.21 0.82 Loading (kg/ha-yr) Ortho-P 0.09 '(1. 9%) 0.80 (35.7%) Total N 3.98 (TKN + N03 ) 1.48 7.36 Organic N N03-N 0.58 (.03 N02 ) 2.17 0.29 + NH4-N 0.16 Comments Area is lightly developed but includes 10% under highway eenst. Stream receives package plant eff. from school. Loading est. extrapolated from data on two storms. Residential development included new construction. Clayey soils. Uniform single family dwellings. Only 510% of rainwater collected as runoff. Flow estimated only 7 storms sampled.

PAGE 27

Table III-5. Nutrient export from agricultural areas. Loading (kg/ha-yr) + Source Location Total P Ortho-P Total N Org. N NO--N Comments 3 Campbell (1978) Alachua Co. 1975-76 1.34 1.21 6.36 5.30 -0.37 0.68 Land in intensive crop pro-1976-77 0.86 0.63 2.;1.0 1.92 0.09 0.09 duction w/some pasture near stream. 3-8% slope, sandy soil w/claypan at 1-2 m.
PAGE 28

Table 111-5. Nutrient export from agricultural areas (cont'd) Total P 0.65 2.37 0.55 0.77 0.58 0.41 0.75 Loading (kg/ha-yr) Ortho-P Total Org. N N NO--N 3 NH+-N 4 Comments 0.141 0.24 Cropland 36.8%. Forest, swamp, etc. 0.145 0.14 27.18 38.86 12.36 50.1 27.0 30.3 32.3 All areas backpumped & irrigated. Organic soils mean of 3 sites. Mean of 3 sites. Mean of 2 sites. Note: drainage areas poorly defined in EAA 88% agricultural (71.2% truck crops, 8.9% pasture) 93% agricultural (36% sugarcane, 15% truck farming, 45% pasture. 78% agricultural (69% sugarcane, 9% pasture). 22% forest & wetland. 30.5% agricultural (26% sugarcane). 69.3% forest and wetlands.

PAGE 29

Table III-6. Nutrient from forested areas. Loading (kg/ha-yr) Source Location Total Ortho Total Org. P .-P N N CSl'lpbell (l278} Alachua Co. 1975-76 0.33 0.30 1.43 1.21 1976-77 0.68 0.52 1.65 1.49 N Bedient et a1. I-' Houston, 0.21 (1978) Texas Reikerk, et Bradford 0.4 .0.2 6.1 5.5 a1. (1978) Co., Fla. Duffy et Northern 0.30 0.029 al. (1978) Mississippi N03 -N NU+-N 4 0.12 0.11 0:09 0.07 0.29 0.1 0.5 (inc. N02 ) Comments Native forest w/small amount of cropland. Sandy soils w/claypan, 0-3% slope. Heavily forested, clayey soils, 0.1% slope. 25 storms sampled. Data collected from 3 coastal plain flatwoods areas for 1 water year. Mean values shown here. Data collected for storm events (no base flow between storms). watersheds sampled. Pine (mix) loess soils. only Five forest;

PAGE 30

et al. (1977) computed export coefficients for several watersheds in the Orlandd area using data collected during only two storms. In the most careful studies (including Riekerk et al. 1978; Campbell 1979; Ritter et al. 1979; Burton et al. 1977) data on baseflow concentrations were collected on a regular basis (usually weekly), and stormflow events were sampled at frequent intervals using an automated sampler. Concentration data were then combined with continuous flow measurement data to produce relatively accurate estimates of loading. In applying export coefficients reported in the literature, the reader should consider the quality of the methods used to compute nutrient loadings. A second weakness of many of these studies is that water budgets were not constructed. Because of this, it cannot be certain that the streamflow at the sampling station represents the sole outlet of water (and nutiients) from the watershed, although this is generally assumed. Conversely, unless a water budget is completed, it cannot be assumed that all of the water coming from a watershed results from precipitation falling on that watershed rather than from seeps and springs whose flow may originate from outside the boundaries of the watershed in question. In order to develop reliable nutrient export coefficients as a function of land-use, it is important that all inflows and outflows to the watershed are determined; this is particularly true in Florida where the groundwater table is fairly shallow and springs are common. Finally, only a few of the studies reported here involved more than one annual cycle, and therefore they did not evaluate annual variations in loading. The data of Campbell (1978), Stewart et al. (1978) and Asmussen et al. (1979) illustrate the extent of variation that may occur from year to year (Table 111-5). Camp'bellobserved decreases of 36% and 67% for total P and total N loads, respectively, between the 1975-76 sampling period and the 1976-77 sampling period. Most of the difference in loading between the two annual periods occurred as the result of a decrease in streamflow rather than changes in the concentrations of constituents. For the Upper Taylor Creek watershed, the highest annual export rate of orthophosphate was 5.5 times the lowest export rate, and the highest export rate of NO)-N was 6.1 times the lowest export rate over a three year period (Stewart et al. 1978). Changes in management practices, as well as variations in precipitation, were considered to be responsible for the observed fluctuations in loading. Asmussen et al. (1979) reported relatively constant orthophosphate loads but a 2X variation in the nitrate load over a two year period for the Little River watershed in southern Georgia. As can be seen from these examples, the magnitude of the variations in loadings can be substantial, at least for agricultural watersheds. Studies cited by Reckhow et al. (1980) indicate that there are also substantial annual variations in the loadings from forested watersheds. The extant of variations in nutrient loadings observed in multiple-year studies suggests that for critical applications, calculations of nutrient loadings should be based on data acquired over the period of several years. Furthermore, it can be concluded that the variation in ex-port coefficients reported for a given land use is attributed in part to temporal variability. -22

PAGE 31

The range of export coefficients for Nand P reported in literature for various land uses is shown in Figures 111-1 and 111-2, together with export coefficients determined from the statistical analysis (to be discussed in the following section). The range of phosphorus export coefficients reported for Florida watersheds falls within the range of coefficients found by Reckhow. et""'al. (1980), for. all three major land uses., although the range of phosphorus export coefficients for urban areas in Florida is near the lower end of the range reported by Reckhow and coworkers. The range of nitrogen export coefficients found for Florida watersheds also falls within the range of values reported by Reckhow et a1. (1980) for all three major land uses. Limited .data on speciation of nitrogen in runoff from agricultural and forested watersheds (Fig. 111-3) indicates that while the majority of nitrogen in runoff from forested watersheds is in the organic form, both nitrate and organic nitrogen are important fractions in the runoff from agricultural wathersheds. It is tempting to conclude that the narrower range of each land-use export coefficient for Florida watersheds compared to the range reported by Reckhow et a1. (1980) reflects the greater similarities in climate, soils, and topography among the Florida watersheds. However, it is also possible that the narrower ranges of export coefficients for Florida watersheds simply reflects the fact that fewer studies have been conducted for Florida watersheds. Unfortunately, a statistical analysis of the variation in export coefficients reported for Florida watersheds cannot be conducted because only a few values have been reported for each land use. The values of export coefficients reported for each land use in Florida watersheds may vary by more than an order of magnitude for most land uses. Thus, phosphorus export coefficients range from 0.26 to 7.49 kg/ha-yr for urban areas, 0.21 to 4.74 kg/ha-yr for residential areas, 0.41 to 2.37 kg/hayr for agricultural areas, and 0.21 to 0.68 kg/ha-yr for forested watersheds. Nitrogen export coefficients exhibit similar variability: 0.37 to 2.88 kg/ha-yr for urban areas, 1.48 to 7.36 kg/ha-yr for residential area, 2.1 to 50.1 kg/ha-yr for agricultural areas .and 1. 43 to 6.1 kg/ha-yr for forested areas. These broad ranges of values limit the accuracy of loading estimates that can be made using a literature-base approach. A further disadvantage of this approach is that the accuracy of predictions cannot be evaluated using statistical analysis. STATISTICAL ANALYSIS OF NUTRIENT EXPORT FROM THE NES WATERSHEDS. An alternative approach in predicting NPS nutrient export from watershed land-use characteristics is to use multiple regression techniques whereby land-use characteristics (the independent variables) are used to predict nutrient export (the dependent variable). In this study, data on nutrient export, flow-weighted nutrient concentrations, flow, and land-use characteristics were compiled for 41 NES watersheds, as described in Chapter II. These data were used to develop statistically significant regression relationships between land use and NPS nutrient loading, land use and f10w weighted nutrient concentration, and land use and flow. -23

PAGE 32

N -I>-(3) Urban Urban' URBAN Residential' I (41) I ALLURB Mixed agricultural Row' crops Nonrow crops I (41). I CPAST AGRICUL TURAL (5) FOREST 0.1 0.5 1.0 5 Reckhow t et 01. (1980) fJ:fil!t;WW\;(\:!diiHf Florida studies fMean a 90% CI for NES watersheds Number in parenthesis indicates number of watersheds 10 50 Phosphorus Export. Kg/ha-yr Figure III-I. Phosphorus export from urban, agricultural. and forested watersheds. 100

PAGE 33

U R B A N A G R I C U L I T U R E F 0 R E S T 0.1 0.1 50 Urban (2) Urban Residential M bled agriculture 'Row crops Non -row crops' Grazing and Pasture Mixed agriculture CPAST I ", 'I (41)', I Reckhow,et 01. (980) I (3) Florid a stud iea I I Mean a 90% CI for NES watersheds Forest Forest 0.5 1.0 5 10 Nitrogen Export Kg/ho yr t 50 Figure 111-2. Nitrogen export from urban, agrtcultural, and forested watersheds. 100 100

PAGE 34

N 0\ F o R I NOI + NO. E S T NH. (3) A ORGANIC N G R I C NOI + NO. U L T U R E, o 10 ORGANIC N (4) 20 30 40 50 80 70 80 90 100 Per Cent of Total N Figure III-3. Range of nitrogen fractions in runoff from forested and agricultural watersheds.

PAGE 35

In order to determine which land uses were statistically significant as independent variables in these regression equations, the backward elimi nation procedure (STEPWISE/B) of SAS was used. In this procedure, all potentially significant independent variables (in this case, area in each land use) are entered in the regression equation. The least significant variables are deleted in sequential steps until all remaining independent variables are found to be statistically significant in contributing to the variability of the dependent variable. For these analyses, the 0.05 level. of significance was used as the criterion for selecting independent vari. abIes for rtrhe,final regression equations. The 95% CLI for the final regression equations were determined using the GLM procedure of SAS. Prior to performing the regression analyses, a correlation matrix of the land use categories was computed, and it indicated that the categories "other urban" (OURB) and "residential" (RES) were significantly correlated (r2 = 0.25). In order to avoid the problem of multicollinearity among independent variables (Neter and Wasserman 1974), these two categories were combined into an "all urban" (ALLURB) category. Thus, eight land use categories were initially included as independent variables in the regression equations: ALLURB, "crops and pastureland" (CPAST), "other agriculture" (OAG) "forest" (FOR), "rangeland" (RA), "non-forested wetland" (FOR), "open water" (WA) and "strip mine" (SMINE) (see Figure 11-2). The. land-use characteristics of the 41 NES watersheds used in these analyses are shown in Table 111-7. The utility of using land-use areas as independent variables in these equations was evaluated by also using total drainage area (DA) as the sole independent variable in equations to predict NPS nutrient loading, nutrient concentration, and flow. Regression results are summarized in Table III-S. Equations to Predict Loading (TPL). The best equation to predict TPL, eq. 3-3, has an r of 0.71 (P >0.0001) and includes three land use areas as statistically significant independent variables (ALLURB, CPAST, and RA). The relationship between DA and TPL (eq. 3-2) was much weaker (r2 0.21), indicating that the use of individual land use areas results in much better predictions of TPL than does the use of DA alone. (Table The S value (phosphorus export coefficient) of ALLURB in eq. 3-3, 6.0 + 1.4 kg/ha-yr, is high compared with phosphorus export coefficients reported for urban watersheds in other studies (Figure III-I). One likely explanation for this is that ALLURB is a well-defined and fairly restrictive land-use category (See Figure II-I), while the "urban" watersheds in other studies often included areas of forest and other land uses that tend to decrease the overall phosphorus export. The S value for CPAST in eq. 3-3, 1.2 + 1.0 kg/ha-yr, falls within the range of phosphorus export coefficients reported for other agricultural watersheds in Florida and throughout North America (Figure III-I). The S value for RA in eq. 3-3, (-1.4 kg/ha-yr) is a statistical anomoly that reflects an inherent weakness of the regression approach in evaluating nutrient export. However, since RA comprised little of the average watershed area (S%), the magnitude of its coefficient has relatively little effect on the predictions produced using eq. 3-3. In contrast, CPAST comprised 27

PAGE 36

Table 111-7. Land use characteristics of study watersheds. Land Use (%)(1) Other urban Residential Total urban (2) Crop and pastura1and Other agriculture Forest Range Non-forested wetland Open water Strip mine Drainage Area (km2 ) Computer Code OURB RES ALLURB CPAST OAG FOR RA NFWET WA SMINE DA Hean 8.2 18.5 26.7 19.6 11.2 19.2 8.3 7.6 5.7 0.9 135 (1) Land uses defined in Chapter II. (2) ALLURB = OURB + RES. 28 Minimum Maximum 0.0 32.5 0.0 62.8 0.0 78.5 0.0 70.0 0.0 60.5 0.0 93.5 0.0 50.6 0.0 46.1 0.0 33.3 0.0 25.4 0.4 978

PAGE 37

TElble III-8. Regression equations to predict nutrient lOElding, nutrient concentration and flow for 41 NES watersheds. Equation Dependent Intercept Independent S 2 P > F n r variable Estimate Std. Error variables Estimate Std. Error 3-2 TPL 7123 3663 DA 0.48 14.6 41 0.21 0.002 (kg/yr) (ha) 3-3 TPL 1705 2373 ALLURB 6.0 0.7 41 0.71 0.0001 (kg/yr) CPAST 1.2 0.5 RA -1.4 0.7 (ha) (kg/hEl-yr) 3-4 TNL 8235 8897 DA 5.1 0.4 41 0.84 0.001 (kg/yr) (ha) (kg/ha-yr) 3-5 TNL 15145 8179 CPAST 4.2 1.2 41 0.87 0.001 N (kg/yr) 16.1 1.8 1.0 NFWET WA 16.0 3.4 (ha) (kg/ha-yr) 3-6 0.36 0.06 -5 -5 41 0.24 0.005 TOTAL P ALLURB 5.6 x 10_ 5 1.8 x 10_ 5 (mg/L) RA -2.6 x 10 1.2 x 10 (hEl) 3-7 FLOW 1,314,01#0 3,794,000 DA 0.31 0.02 41 0.91 0.0001 (m3/yr) (m2 ) (m/yr) 3-8 FLOW 2,765,000 2,978,000 ALLURB 0.25 0.09 41 0.96 0.0001 (m3/yr) FOR 0.55 0.04 RA 0.49 0.08 CPAST 0.15 0.06 OAG 0.75 0.09 (m2 ) (m/yr)

PAGE 38

an average of 19.6% of the total watershed area, while ALLURB comprised an average of 26.7% of the total watershed area. The best predictive equation found to relate the concentration of total phosphorus (TOTAL P) to land use is eq. 3-6, which includes only ALLURB and RA as independent variables. The relatively low r2 for this equation (0.24, P> .005) suggests that TOTAL P is not greatly affected by land use in these watersheds. The strong correlation found between TPL and land use thus suggests that variations in flow rather than concentration are responsible for variations in TPL among watersheds. The mean non-point source loading of phosphorus for all 41 watersheds is 1.0 kg/ha-yr. Since precipitation contributes only 0.3 to 0.7 kg P/ha-yr to Florida watersheds (Table 111-2), it can reasonably be concluded that there is a net addition of phosphorus to the runoff water from within the watersheds. Predictions of Nitrogen Loading (TNL). Unlike the situation for TPL, there is a strong correlation between TNL and DA (r2 = 0.84, P> 0.001), indicating that drainage area alone is a reasonably good predictor of nonpoint source nitrogen loading Ceq. 3-4). When the areas in individual land uses were used as independent variables, three terms CCPAST, NFWET and WA) were found to be significant at the 0.05 level, producing an equation Ceq. 3-5) with an r2 of 0.87 (P> 0.001). Thus, the use of individual land use areas as independent variables rather than total watershed area alone contributes little toward improving predictions of TNL. Two of the significant terms in eq. 3-5 are NFWET and WA, both of which have nitrogen coefficients near 16 kg/ha-yr. The magnitude of these export coefficients is high compared to the mean input of nitrogen from precipitation (7.5 kg/ha-yr). The high loadings for open water and non-forested wetland may be caused by inputs from surrounding land uses, such as loadings from lawn fertilization and septic tanks associated with shoreline development. Alternatively, the high reported non-point source loadings for these land uses may reflect inaccuracies in the computation of non-point source loadings. The only other significant term in eq. 3-5 is CPAST, which has a nitrogen export coefficient of 4.2 + 2.4 kg/ha-yr. It can be seen (Figure 111-2) that this 95% C.l. for nitrogen loading from agricultural areas is within the lower end of the range of values reported for agricultural watersheds in Florida (Table 111-5) and throughout North America. The mean non-point source loading for nitrogen in all 41 study watersheds is 5.7 kg/ha-yr., which is somewhat lower than the mean precipitation loading of 7.5 kg/ha-yr reported by Hendry et al. (1981) and Brezonik et al. (1981). Thus it appears that there is a net accumulation of nitrogen in these watersheds. Equations to Predict FZow (FLOW). As shown by eq. 3-7, FLOW is highly correlated with DA Cr2 = 0.91, P> 0.0001). The S value for DA in eq. 3-7, 0.31 m/yr, is about 24% of the mean annual precipitation for the NES watersheds. The use of individual land use areas rather than DA improves the 30

PAGE 39

prediction of flow (eq. 3-8) only slightly (r2 = 0.96, P> 0.0001). The 8 values in eq. 3-8 are runoff coefficients (m/yr), which are signifioant at the 0.05 level for five land uses. APPLICATION The accuracy of predictions frtrirnutrient loading and flow can be evaluated using the 90% and 95% eLI's shown in Figures III-A to 1119. The eLI is the confidence limit for individual predictions, and represents the degree of confidence with which a regression equation can be used to produce new predictions. Since the width of the confidence interval remains approximately constant, predictions of high values are relatively more accurate than are predictions of low values. For example, the 90% eLI for predictions of TPL using equation 3-3 is approximately + 21,000 kg P/yr. Thus, a watershed having a predicted phosphorus export of 13,600 kg P/yr (the mean value) has a 90% eLI of + 21,000 kg P/yr, or + 154% of the predicted value. However, a watershed having a predicted phosphorus export of 40,000 kg P/yr also has a 90% eLI of around 21,000 kg P/yr., so the relative accuracy is improved to + 53% of the predicted value. Predictions of TNL and FLOW are relatively more accurate than are predictions of TPL. For predictions of TNL the 90% eLI is approximately + 80,000 kg N/yr, or + 104% of the predicted value for a watershed having than mean value of nitrogen export (77,000 kg/yr), using either equation 3-5 (Figure 111-6) or equation 3-6 (Fig. III-n. The 90% eLI for predictions of FLOW using equation 3-8, is + 32.5 x 106 m 3/yr, or + 75%.:of thepnedicted value for a watershed having the mean flow of 43.1 x 106.m3/yr (Figure 111-9). However, as seen in Figure 111-8, the confidence intervals for equation 3-7, in which DA is the only independent variable, are comparable. Since the use of individual land use areas rather than drainage area alone does little to improve predictions of TNL and FLOW, it is suggested that equations 3-5 and 3-7, in which DA is the sole independent variable, be used for predictions of these parameters. For predictions of TPL, the use of equation 3-3, in which land use areas are used as independent variables is recommended. In applying these equations, confidence intervals should be used to assesses the degree ,of reliability associated with each prediction. In using these equations to predict nutrient loadings, nutrient concentration or runoff in other Florida watersheds, several points should be emphasized: 1) These predictive equations were developed using a relatively small group of watersheds that are not necessarily representative of all Florida watersheds. Predictions made using these equations are valid for only those watersheds that are from the same population as the test watersheds. Thus, in evaluating the applicability of these equations for a new situation, the user is urged to compare the land use characteristics of the new watershed(s) with those of the watersheds used in these analyses (See Table 111-7). For example, it would be inappropriate to apply the predictive tions developed for urbanized watersheds to a new watershed that is composed 31

PAGE 40

12 TPL 7.123 x 103 + 0.48 DA r2 = 0.21 10 90% and '95% eLI sho'liffi .. b1l P :8 6 ell a H CIJ H .8 4 p.. CIJ a ..c: P-t Cf.l 2 o o 1 2 3 4 5 6 7 Drainage Area, 10 4 ha 8 9 10 Figure 111-4. Non-point source phosphorus loading vs. drainage area for 41 NES watersheds. 32

PAGE 41

12 1-1 :>, 10 ---bfl ..oe: -:::t o r-l n 8 gf "ri 'lj (1j o H 6 4 2 o o 2 TPL = 1,705 +.6'.0 ALLURB + 2 ,1.2 CPAST -1.4 RA r = 0.71 90% and 95% CLI shown 6 8 10 12 Measured NPS Phosphorus Loading, 104 kg/iyr. Figure 111-5. Predicted vs. measured NPS phosphorus loading using Equation 3-3. -33

PAGE 42

H :>-. ........ co -::t 0 .-l co P .r-! "d cd 0 H P OJ co 0 H .\J 'r-! z Cf.l P-l z 60 55 50 45 40 35 30 25 20 15 10 TNL = 8.24 x 10 3 + 5.1 DA 5 r2 = 0.84 90% and 95% eLI shown 0 o 1 2 3 4 5 6 7 8 9 Drainage Area, 10 4 ha Figure 111-6. Non-point source nitrogen loading vs. drainage area for 41 NES watersheds. -34 10

PAGE 43

-!-I :>-, ----OJ) '7 0 r-l OJ) (:1 'r! '1:). C1j 0 H (:1 (j) OJ) 0 !-I .w 'r! Z C/) '1:) (j) .w c.J 'r! '1:) (j) !-I p., 60 55 50 Lf5 LfO 35 30 25 20 15 10 5 TNL 2 r = 15,145 + 4.2 CPAST + 16.1 NFWET + 16.0 WA 0.87 90% and 95% CLl shown o 5 10 15 20 25 30 35 40 45 50 55 60 Measured NPS Nitrogen Loading, 105 kg/yr Figure 111-7. Predicted vs. measured NPS nitrogen loading using Equation 3-5. 35

PAGE 44

H ?> --.... \"'") S r---0 H 0 H f:J:.< Ct:J 35 30 25 20 15 10 5 FLOW = 1.31 x 106 + 0.31 DA r2 = 0.91 90% and 95% eLI shown o 2 3 4 5 6 7 S 9 10 Drainage Area, 104 ha. Figure III-S. Non-point flow vs. drainage area for 41 NES watersheds. 36

PAGE 45

:: I-l :>-, "-C"') S r-.... 0 r-l .. 0 r-l fI.! tf.l '1j aJ .j..J CJ ',-j '1j aJ I-l p.; !..1. ;:.:" 35 30 25 I 20 15 10 5 o o 5 10 FLOW = 2.77 x 106 + 0.25 ALLURB + 0.55 FOR + 0.49 RA + 0.15 CPAST + 0.75 OAG r2 = 0.96 90% and 95% CLI shown 15 20 25 30 7 3 Measured NPS Flow, 10 m /yr 35 Figure 111-9. Predicted vs. measured NPS flow using Eq. 3-8. -37

PAGE 46

of 90% ALLURB, since the largest fraction of any of the watersheds used in developing these equations covered by ALLURB is 78.5% (Table 111-7). 2) It would be inappropriate to use these predictive equations to estimate the change in loading that would occur if a portion of a watershed is converted from one land use to another (i.e., forest to residential). The reason for this is that regression analysis does not necessarily imply a causal relationship between the independent variables (i.e., nutTient loading) and the dependent variables (land uses). Underlying factors, such as soil type, drainage and slope may be related to both land use and nutrient loading. Thus, the difference in loading between forest and urban areas reflects, in part, differences in edaphic features of the landscape that render certain areas suitable for urban development and others less so. However, as development pressure will be a tendency to urbanize areas that are not currently considered suitable for urban development (Le. under the conditions in which these predictiveequatilions weredevel oped). Because of this, the loading of nutrients from urban areas developed in the future may be different from the loading from urban areas currentlyin existence. In a similar vein, it should be realized that changes in management practices of existing land uses may alter the rate of nutrient export from some areas. For example, in the past 10 years, the practice of. nitrogen fertilization in agricultural areas has changed considerably. These trends include 1) decreased application rates, 2}the improvement of tillage methods, 3) the use of fertilizers that are better retained in the soil column (i.e., urea and ammonium instead of nitrate), and 4) the application of nitrification inhibitors (Calvert 1975; and Allen 1970). These trends should act to conserve fertilizers and reduce the export of nitrogen from agricultural areas. 3) The predictive equations developed here are based on mean annual loadings. As mentioned earlier, there may be considerable year-to-year variation in the annual nutrient loading from non-point sources. For some applications, such as the assessment of lake restoration techniques, it is important to be able to determine actual annual loadings of nutrients into a lake for a period of several years. For this type of study the predictive equations developed here would not be suitable. 4) In the development of these predictive equations it was assumed that all of the flow in a tributary other than that emanating from-known pOint sources (i.e.; sewage treatment plants), was derived from precipitation falling on the watershed. Several watersheds were excluded from the analyses because it was believed that the flows in the tributaries were too high to be the result of natural precipitation. ,In applying these predictive equations, the usr should also be reasonably certain that the flow in a partic ular tributary is derived from natural precipitation and runoff. For some watersheds, the use of these equations would not be suitable, either because of the limitations cited above or because their predictive capability is not adequate for a particular purpose. In some cases, the use of export coefficients obtained from the literature (Tables 111-3 to 111-6) may be more appropriate in generating preliminary estimates of NPS nutrient loading. This is particularly true for small watersheds in which one land use is predominant. When the literature-based approach is used, -38

PAGE 47

we suggest that original studies be consulted and that reported export coefficients be used only when the watersheds in which they were determined are similar, with respect to topography, soils and other characteristics, to the watershed under investigation. When more accurate results are required or when the effects of proposed management strategies on nutrient loadings are being evaluated, the use of simulation models should be considered. Examples of these include a number of agricultural models (reviewed by Haith 1980) and the Storm Water Management Model, designed for use in urban areas (Heaney eb a1. 1976). -39

PAGE 48

CHAPTER IV, LIMNOLOGICAL CHARACTERISTICS OF FLORIDA LAKES The data compiled in this study enable a broad characterization of the limnological conditions of Florida's lakes. Data were collected for most major limnological parameters in all three studies (Table 11-2), with the exception of color and turbidity, which were not collected by the National Eutrophication Survey. It should be noted that the sampling intensity for most of the lakes in the three surveys was low; most lakes were sampled only 3-4 times during an annual period. Considering the large variations (often order of magnitude) that may occur in algal standing crop and in the concentrations of major nutrient species during an annual period, computed means for any single lake must be accepted with caution. However, considering the large number of lakes in the data base, use of the data to make general inferences on limnological relationships in the lakes is justified. MORPHOLOGICAL CHARACTERISTICS Although Florida's lakes vary considerably in size and morphometry, most are quite shallow. Of the 101 lakes included in this only ten have maximum depths greater than 10 m, and only three (Annie, Kingsley and Mize) are over 20 m deep. The study lakes have a considerable range in surface area (Table IV-I). Many of the lakes in and Putnam counties have surface areas of only a few hectares, but several of Florida's largest lakes are included in this study. By far the largest is Lake Okeechobee (1890 km2), which after the Laurentian Great Lakes is the largest freshwater lake (in surface area) entirely in the United States. The morphometry of most Florida lakes has been affected by limestone solution processes and many have been formed in sinkhole depressions. Some of these lakes, like Lake Santa Rosa (Putnam County) are nearly circular, while others are complex dolines that have been formed in adjoining solution basins (e.g. Cowpen Lake in Putnam County). The smaller lakes are often hydraulically connected to perched water tables that are separated from the main (Floridan) limestone aquifer by an clayey Many lakes are in seepage basins and lack distinct inflows or outflows. The water level in these lakes may fluctuate by as much as several meters between wet and dry periods. Water levels in the larger lakes are usually structurally controlled to minimize natural variations. Because of their shallowness and the mild climate, most Florida lakes do not exhibit stable thermal stratification. Of the 55 lakes studied by Brezonik and Shannon (197l), only eight developed stable thermal stratification during the warm season. Approximately eight additional lakes showed evidence of temporary thermal stratification lasting for periods of a few weeks to a few months. The shallowness of Florida's lakes also encourages resuspension of sediments to the overlying water, particularly in large lakes and lakes with loose, flocculent sediments. Studies on Lake Apopka 40

PAGE 49

Table IV-L General characteristics of study lakes. Morphological I Mean Minimum Maximum I Surface area, km2 23.0 0.01 1890.7 Volume, 3 x 106 6L4 0.03 2494.0 m Mean depth (z), m 2.9 0.7 8.3 Maximum depth ), m 5.4 0.9 25.3 max Chemical & Biological (annual means) Color, units 116 2 539 pH 7.1 4.7 10.4 Alkalinity, mg/L as CaC03 32 0 163 Chlorophyll a (chl a), ]Jg/L 29.1 0.9 276.6 Total nitrogen (N), mg/L L5l 0.19 5.56 Total phosphorus (P) ]Jg/L 231 7 2,750 41

PAGE 50

(Pollman and Brezonik 1981) indicate that phosphorus is released to the water column during wind events as the result of desorption from suspended sediments. This phenomenon undoubtedly contributes to eutrophication prob lems in such lakes. CHEMICAL AND PHYSICAL CHARACTERISTICS. The majority of Florida's lakes are poorly buffered (mean alkalinity 32 mg/L as CaC03)' even though they ar.e underlain by limestone. This apparently contradictory situation occurs because many lakes are not hydraulically connected to the underlying limestone formations, but receive the bulk of their water either directly from precipitation or by surface and subsurface runoff from the sandy, low-calcareous soils. Many are also highly colored; the mean color for the study lakes is 116 CPU (chloroplatinate units). The frequency and intensity of color reflects the abundance of watersheds composed of pine forests and wetlands (swamps). Relationship Between pH and Alkalinity. The pH of Florida lakewaters is strongly correlated with alkalinity (Figure IV-I). The solid line in the figure shows the relationship between pH and alkalinity in a water system where the carbonate buffering system controls pH at atmospheric pressure (PC02 = 10-3 5 atm). The data suggest that except for the most acidic lakes the pH is near the equilibrium value defined by the CO2-bi carbonate system. However, for many of the poorly buffered lakes in the Trail Ridge region, the observed pH is considerably lower than that expected for pure water in equilibrium with atmospheric CO2 (pH 5.7). Brezonik et al. (1981) found that the pH of some of these lakes has decreased by. as much as 0.5 pH units over the past 20 years, apparently as the result of inputs of acid.precipitation. This trend is expected to continue as Florida increases its production of electricity by coal-fired electric plants. Factors Affecting Transparency. Seechi disk transparency is one of the most commonly measured parameters in the study of lake eutrophication. In addition to being of scientific interest, Secchi disk transparency is readily comprehended by the public as a measure of water clarity. Several investigators (Bachman and Jones 1974; Carlson 1977; Brezonik 1978) have found a hyperbolic relationship between the concentration of chlorophyll a and Seechi disk transparency. Brezonik (1978) reviewed the theoretical relationship between light attenuation and Secchi disk transparency (SD) and concluded that for many lakes the equation: l/SD = a [color] + b (4-1) can be used to describe the variation in Secchi disk transparency. Since turbidity is often closely related to chlorophyll a, it was hypothesized that l/SD = a [color] + b [chI a] (4-2) The non-linearity between inverse Secchi disk and chlorophyll a observed by Carlson (1977) and Brezonik (1978) may be explained by 1) an increase in the amount of chlorophyll per cell in more eutrophic situations, 2) a -42

PAGE 51

12 10 8 p. 6 4 2 o 50 100 150 200 Mean alkalinity, mg/L as CaC0 3 Figure IV-I. Mean pH VS, mean alkalinity for 101 study lakes. Solid line shows equilibrium relationship between pH and alkalinity at pC0 2 = 10-3 5 atm; 43

PAGE 52

change in the size distribution of the seston and in the morphology of algae with increasing eutrophication, or 3) light attentuation due to color (Brezonik 1978). A statistical analysis of the relationship between inverse Secchi disk and color, turbidity, and chlorophyll a was conducted by Brezonik (1978) using data from the 55 lake study. A similar analysis was conducted for this report using the additional data from the two other lake surveys (NES 1977 and Brezonik et al. 1981). Results of the statistical analyses (Table (IV-2) are similar to those of Brezonik (1978), although the coefficients differ slightly, and the coefficients of determination (r2 values) are slightly lower. The lower r2 values may result from the more diverse group of lakes used in the present analysis. Differences in sampling and analytical procedures among the studies also may account for the slightly lower r2 values. When chlorophyll a is the only independent variable, a log-log model yields better predictions of SD transparency than does an inverse relationship (cf. eq. 4-2 and eq. 4-8). This result is consistent with the findings of Carlson (1977), Bachman and Jones (1974), and Brezonik (1978). The coefficient of determination for the log SD vs. log (chI a) relationship is lower for Florida lakes (r2 = 0.70) than reported by Carlson (1977) and Bachman_and Jones (1974) for temperate zone lakes (r2 0.86 and 0.90, respectively) because color affects SD transparency more in Florida lakes than in temperate zone lakes. As seen by equations 4-4 and 4-9, color alone is reasonably good predictor of SD transparency. The best equation for predicting l/SD (eq. 4-6) uses turbidity and color as independent variables (equation 4-6). Apparently, the use of turbidity overcomes some of the problems mentioned above that are encountered when (chI a) is used to represent light attenuation. Despite the apparent good fit (Figure IV-2), this equation does not predict SD transparency accurately in some of the clear, oligotrophic lakes in the Trail Ridge group. The intercept value of eq. 4-6 (0. 17rrll) corresponds to a SD of only 5.9 m at zero color and turbidity in the water column. This transparency is far lower than that expected in a water column devoid of algae and color. Hutchinson (1957) reported that the maximum SD transparency ever measured in a lake is about 40 m, which would give an intercept term in eq. 4-6 of 0.025. The failure of this equation to predict high SD values accurately reflects the fact that high values of SD have very low inverse values that do not affect the fit of the regression equation as much as do lower SD values. Furthermore, the data base does not include any lakes with very high transparency values; the maximum mean SD in the data set is 7.9 m (l/SD = 0.013 m-l ) for Lake Sheeler. However, eq. 4-6 does produce reasonable estimates of SD transparency for predicted SD values less than 3 m. This situation accounts for most of the lakes in the data set, and probably most lakes in Florida. For lakes with a predicted SD >3m., eq. 4-6 underestimates the actual SD transparency. BIOLOGICAL CHARACTERISTICS. PhytopZankton Communities. The and standing crop of phytoplankton communities in the study lakes is highly variable. The most pris44

PAGE 53

Table IV-2. Regression equations to predict Secchi disk transparency. (1) Inverse relationships l/SD = l/SD = l/SD = l/SD l/SD = 0.80 + 0.01 (chla) 2 r = 0.33 n = 100 0.49 + 0.12 (T) r2 = 0.62 n = 63 0.70 + 0.002 (e) 2 r = 0.70 n = 63 0.37 + 0.03 (chI a) + 0.001 (e) r2 = .62 n = 63 0.17 + 0.11 (T)+ 0.002 (e) r2 = 0.82 n = 63 Log-log relationships log (SD)= log (SD)= log (SD)= 0.49 = 0.76 log (T) 2 r = 0.55 n = 63 0.55 0.47 log (chI a) 2 r = 0.70 n = 100 0.78 0.39 log (e) 2 r = 0.53 n = 63 (4-2) (4-3) (4-4) (4-5) (4-6) (4-7) (4-8) (4-9) (1) SD = mean Secchi disk transparency, m; C = mean color, Pt uni ts; T = mean turbidity, FTU, !clU a) = mean chlorophyll a concentration, -45

PAGE 54

3 s ft p... CJ 2 p. Q) H cO p. (J) p cO H +J (J) OM .j::.. OM :;(. 0\ CJ CJ Q) Cf.l '"d 1 Q) +J CJ .-/ OM '"d Q) H ./ P-I o 0' 1 2 3 Measured Secchi disk transparency, m Figure IV-2. Predicted vs. measured Secchi disk transparency using Eq. 4-6.

PAGE 55

tine are those in the Trail Ridge region. Most of these lakes (except Altho, Geneva and Kingsley) have mean pH values of <5.6 and phytoplankton communities typical of acidic, oligotrophic lakes (Schulze 1980). These communities are dominated by Staurastrum sp., Scenedesmus sp., Ankistrodesmus fa1catus, Peridinium inconspicuum and several small green coccoids. Blue-green algae occur in low abundance in these lakes and are represented mainly by Osci11atoria 1imnetica and Anacystis incerta. At the other extreme, many of the study lakes are highly eutrophic (e.g. Biven's Arm, Newnan's, Apopka and Kanapaha). These lakes may exhibit dense algal blooms dominated by blue green and green algae; in some of the most fertile these blooms are virtually continuous. Fish Populati,ons. A recent survey of 22 Florida lakes describes the changes that occur in fish communities with increasing eutrophication (Kautz 1981). The oligotrophic lakes, characterized by well-developed communities of littoral vegetation, limited planktonic production, and sandy bottoms covered with a thin layer of detritus, have fish communities dominated by populations of sport fishes (largemouth bass, bluegill and other sunfish, striped bass, and pickere1!) and forage fishes (Fig. IV-3). Populations of rough fishes (gar, gizzard shad, bowfin, and ti1apia) and commercial fishes (primarily catfishes) are limited and the total biomass and species diversity of the fish communities is low. The meso trophic-eutrophic lakes have the best developed populations of sport and forage fishes (Fig. IV-3). These lakes are characterized by we11-developed connnunities of littoral vegetation, 1arge-
PAGE 56

UJ UJ o .r! ,.Cl ..r::: UJ .r! 4-4 r-l cd .w 90 80 70 60 50 40 .3 30 4-4 o 20 10 Oligotrophic lakes Sport fishes Forage fishes Rough fishes I: Commercial fishes Mesotrophiceutrophic lakes Hypereutrophic lakes Figure IV-3. Structure of fish communities in Florida lakes. Data from Kautz (1981). -48

PAGE 57

Of the methods used to evaluate nutrient limitation, the most common are (1) enrichment bioassays and (2) the computation of nutrient ratios, usually the ratio of total soluble inorganic N to soluble reactive phosphate (SIN:SRP). In enrichment bioassays, various nutrients are added to aliquots of the test water. Either a test species of algal (usually Selenastrum capricornutum) or an ihdigenous mixed culture, are added to each treatment, and the limiting nutrient is determined by comparing the growth of algae in the nutrient-spiked aliquots with a control. The procedure has become highly standardized in the Algal Assay Procedure: Bottle Test (Miller et al. 1978) and is widely used. The second method involves the measurement of nutrient concentrations and computation of the ratio SIN:SRP. ratio has been compared with the results of nutrient enrichment bioassays (cf. Chiaudini and Vighi 1974; Miller et a1. 1975). The "critical ratio" usually falls between 10:1 and 20:1. Porcella and Bishop (1975) stated that a SIN:SRP ratio less than 10:1 clearly indicates N-limitation, a ratio greater than 20:1 indicates P-limitation, and an intermediate ratio indicates mixed nutrient limitation. Algal bioassays (AAP:BT) conducted for 31 of the NES lakes (NES 1977) indicated that 23 (74%) were nitrogen limited, seven (23%) were phosphorus limited and one had mixed nutrient limitation. When the results of nutrient bioassays are compared with the criteria of Porcella and Bishop for nutrient limitation (Figure IV-4), few misclassifications occur. For the 27 lakes having an SIN:SRP ratio of less than 10:1, 23 were found to be nitrogen limited in the AAP:BT and four were found to be phosphorus limited or have mixed nutrient limitation. All four lakes with SIN:SRP ratios >10:1 in which biassays were conducted were phosphorus limited. Thus, the criteria of Porcella and Bishop seem to be reasonably valid, although more data are needed for lakes having high SIN:SRP ratios to make conclusive remarks concerning the application of these criteria in Florida lakes. A frequency distribution of mean SIN:SRP ratio for all 101 study lakes (Figure IV-5) shows that 46% have SIN:SRP ratios <10:1, 21% have ratios between 10:1 and 20:1, and 33% have ratios >20:1. Thus, if the proposed criteria of nutrient limitation are valid, nearly half of the study lakes are nitrogen-limited while only a third are phosphorus-limited. A plot of the mean SIN:SRP ratio versus the mean chlorophyll a for each lake (Figure IV-6) shows an inverse relationship. Lakes with high algal standing crops (mean chI a >50 pg/L) have values of SIN: SRP
PAGE 58

Figure 'iEV-4.. Total inorganic nitroge,n vs. orthophosphate concentrations (mean values) in the Florida NES lakes. Nand P next to data points indicate nutrient found limiting in lake by Algal Assay Procedure.

PAGE 59

50 40 r-.. '-' Q} ()I t=l Q) I-l I-l 30 ;:l () () 0 4-4 0 :>. () U1 t=l Q) 20 t-' ;:l 0-Q) I-i r: 10 o ff-"-0-10.00 10.01-20.00 )(=22.1 20.01-30.00 30.01-40.00 40.01-50.00 50.01-60.00 r 60.01-70.00 Soluble soluble reactive phosphorus >70 Figure IV-S. Frequency distribution of mean SIN:SRP ratios inthe 101 study lakes.

PAGE 60

a .,.-l +J cO !-l Ul .. f:l Ul Q) U1 N 130r 120 110 100 90 60 r 50 40 :.. 30r ..... 10 t .. 1 ........ ot, I I 60 o 20 40 80 100 120 140 160 180 200 220 240 260 280 Hean chlorophyll a, j.1g/L Figure IV-6. Relationship between soluble nitrogen: soluble reactive phosphorus (SIN:SRP) and mean chlorophyll a, j.1g/L.

PAGE 61

Florida streams is related to their proximity to phosphate deposits, and an analysis of the SIN:SRP ratios in the NES tributaries indicates that all the streams in the vicinity of the phosphate deposits had ratios less than 5 (n = 6). However, about 30% of the streams not in the vicinity of these deposits also had such low ratio, indicating that factors other than imity to phosphate deposits are responsible for the low SIN:SRP ratios in many of Florida's streams. In-lake mechanisms, particularly denitrification, also may result in a decrease in the ratio of nitrogen to phosphorus with increasing eutrophication. In many lakes, loss of nitrogen via denitrification can represent a large fraction of the input of nitrogen. For example, Messer et ale (1979) concluded that up to 26% of the nitrogen input into Lake Okeechobee is removed by denitrification; losses of over 10% have been reported in the literature for a number of other lakes. Although no direct correlation has been demonstrated between trophic state and denitrification rates, conditions existing in eutrophic lakes encourage denitrification. These conditions include a supply of organic matter to supply energy, anoxic conditions near the bottom, and high levels of nitrate. The Relationship Between Nutrients and Chlorophyll a Standing Crops. Several investigators (Dillon and Rigler 1974b Sakamato 1966; Jones and Bachman 1976) have demonstrated a strong log-log relationship between the mean concentration of phosphorus in the water column during turnover and the mean epilimnetic chlorophyll a concentration during the summer (Table IV-3). This relationship has the basis for constructing input/output (I/O) models that can be used to predict chlorophyll concentrations in lakes using only data on phosphorus loading and hydraulic characteristics (Chapter 5). However, these relationships were developed in temperate zone lakes that are usually phosphorus limited (Miller et ale 1974). The lakes in this .study differ from these temperate zone lakes in that 1) a dimictic pattern of stratification is usually not observed, and 2) most are N-limited rather than P-limited. Thus we examined the relationship between chlorophyll a and total Nand P for Florida lakes. Since there are no distinct periods of turnover and stratification in these lakes, annual means usually were used, although the relationship between mean spring phosphorus concentration (spring being defined as March through May) and mean summer chlorophyll a (summer being defined as June through September) also was evaluated. Log-log plots of (P)l and (N)l versus (chl a) (annual means) are shown in Figures IV-7 and IV-8, together with the lines of the regression equations that describe the best fit (equations 4-13 and 4-16). The regression line describing the relationship between (P)l and (chI a) (eq. 4-13) for the study lakes has a much lower slope that do the regression lines of Dillon and (1974b)or Jones and Bachman (1976) (see Figure IV-9). A second regression line, determined using only spring total P and summer chlorophyll a values, has a similar but slightly lower slope (eq. 4-15) than that for the annual means. These two lines demonstrate that the amount of chlorophyll a associated with a given level of total P is lower in Florida lakes than for most temperate zone lakes. This is the situation that one would expect for a group of lakes that are largely N-limited. In order to test the hypothesis -53

PAGE 62

Table IV-3. Regression equations of total phosphorus & total nitrogen vs. chlorophyll a.(l) Phosphorus Jones & Bachman (1976): log (chI a)s = -1.09 + 1.46 log (P) r2 = 0.98 n = 143 Dillon and Rigler (1974b): Using data from Sakamoto (1966): log (chI a)s = -1.134 + 1.583 log (P)sp r2 = 0.95 n = 30 Using data from North American lakes: log (chI a)s = -1.136 + 1.449 log (P)sp r2 = 0.90 n = 46 This study: Entire data set, annual mean (P) vs. annual mean (chI a) log (chI a) = + 0.79 log (P) r2 0.72 n = 100 P-limited lakes only: log (chI a) = -0.71 + 1.03 log (P) r2 = 0.53 n = 33 Entire data set, spring P vs. summer chI a: log (chI a)s -0.16 + 0.71 log (P)sp r2 = 0.57 n = 63 Nitrogen Entire data set: log (chI a) = 1.03 + 1.46 log (N) r2 0.77 n = 100 N-limited lakes only: log (chI a) = 1.12 + 1.53 log (N) r2 = 0.77 n = 44 (4-10) (4-11) (4-12) (4-14) (4-15) (4-16) (4-17) (1) chI a = mean chlorophyll a, Vg/L., P = mean total phosphorus, Vg/L. N = mean total nitrogen, mg/L. Subscripts: s = summer mean, sp = spring mean, none = annual mean. S4

PAGE 63

H -bIl ;::l. .. tl ..-! ..-! i ,..c:: p.. 0 1-1 0 ..-! ,..c:: (j (l) ;:;:; 1000 100 ..... I-10 .i.!- 1 ChI a 0.392 [P] 1 0.79 2 0.72 r 95% CLM shown 0.1 1 10 100 1000 10000 Mean total phosphorus, )lg/L Figure IV-7. Relationship between mean total phosphorus and mean chlorophyll a for 101 study lakes. 55

PAGE 64

1000 OJ) ;:l. ., tl r-1 r-1 E p.. o H o r-1 '5 10 \:::! (1j QJ ::>:: 1 0.1 ... .. -. I .. r 1.0 ChI a 2 r 95% 10. 7 [N] 11. 46 0.77 CLM shown 10.0 Mean total nitrogen, mg/L Figure IV-S. Relationship between mean total nitrogen and mean chlorophyll a in 101 study lakes. 56

PAGE 65

H -... 1/ I, II 1/ I, 1/ 1/ I, 1/ / 1./ 100 1/ I' 1/ 1/ I. 1/ Ii 1/ I' /'/ 1/ /, 1/ 1/ /, /1 1/ I, 1,/ II /.' I? /1 / I'/;' / /" /" /" ,/ ,/' /' ----JONES AND BACHM-AN, 1976 (Eq.4-10) -'-'-' DILLON AND RIGLER, 1974 (Eq.4-11) ----THiS STUDY, P LIMITED LAK ES ONLY (Eq.4-14) --"--"THIS STUDY, SPRING P vs, SUMMER CHL A(Eq-4-15) IL-__________ ____ __________ ____ __________ __ 10 100 1000 Total phosphorus, Figure IV-9. Regression lines of total phosphorus vs. chlorophyll a determined by several investigators. -57

PAGE 66

that nitrogen limitation is actually the factor that results in the low slope for the Florida lakes, a regression of (P)l vs. was determined for a group of P-1imited lakes (eq. 4-14). In this analysis, phosphorus limitation was defined by a ratio of SIN:SRP of >20:1. The slope of this line is steeper than that of the regression line determined for the entire group of Florida lakes (Figure IV-9), but it still is lower than the slopes of the lines determined for temperate zone lakes. There are several factors that could account for the lower slope for phosphorus-limited Florida lakes than for the temperate zone lakes. First, it is possible that a ratio of 20:1 for SIN:SRP is not an accurate criterion of phosphorus limitation; that is, some higher ratio would be more realistic. This is unlikely, since a critical ratio of 20:1 is quite conservative. Second, factors other than phosphorus concentration may limit algal standing crop even in lakes that are considered phosphorus-limited on the basis of nutrient ratios. For example, nutrients other than nitrogen or phosphorus may limit algal productivity in some lakes, as may toxic substances. Miller et a1. (1974) found that constituents other than N or P were limiting to algal productivity in 6 of the 49 temperate zone lakes included in their survey. Additional bioassay data are needed to determine to what extent other nutrients or toxic constituents may be limiting for algae in Florida lakes. Third, biological interactions may limit the standing crop of algae in Florida's lakes below the level in temperate zone lakes containing the same level of phosphorus. Grazing by herbivorous grazing fish (e.q., shad) or other structural differences in the food chain may control the algal standing crop more effectively in Florida's lakes than in temperate lakes, as may interactions with macrophytes. A detailed discussion of ecological inter aetions that affect the algal standing crops in lakes is presented by Shapiro (1979). Finally, the much longer growing season (essentially year-round in Florida, compared to the compressed growing season in temperate lakes) may also affect the chlorophyll a-total phosphorus relationship. The long period of ice cover and dormancy in temperate lakes leads to a build-up of inorganic nutrients, culminating in the "spring maximum". This in turn leads to the spring and early summer pulses of algal blooms. In contrast, inorganic nutrient levels exhibit less seasonal variations in warm temperate lakes, and algal growth is more evenly distributed throughout the year (see following section). Coupled with the biological interactions (e.g. grazing) mentioned above, this may result in a lower standing crop of algae (hence chlorophyll a) for a given total phosphorus level than is found in most pulsed systems. The relationship between (N)l and (ch1 q) in the study lakes is shown by Eq. 4-16. The r2 for this relationship (0.77) is slightly higher than the relationship between (P)l and (ch1 a) (r2 = 0.72), as one would expect for a group of lakes that is largely nitrogen-limited. When only nitrogenlimited lakes were considered (according to the criterion of SIN:SRP<10:1), the regression equation was only slightly altered (eq. 4-17). 58

PAGE 67

Analysis of Seasonal Trends. In temperate zone lakes, pronounced seasonal variations occur in the standing crop of algae and in the concentrations of major nutrients. Major nutrients tend to reach peak concentrations during spring and fall turnover, and the algal standing crop tends to reach maximum levels either following the turnover periods or during the summer stratification period. In Florida, where most lakes do not undergo stable thermal stratification and seasonal variations in temperature are less pronounced than in the temperate zone, it is reasonable to hypothesize that seasonal variations in algal standing crop and in the concentrations of major nutrients will be minimal. To test this hypothesis for the study lakes, normalized parameter values were computed for each lakes: N .. = X .. 1J ---.!:l. where 1'-l 1J X .. 1J X .. 1J X .. 1J the normalized value for variable i and lake j; the variable value during a sampling period; the annual mean value. An overall normalized mean can be calculated: 1>.1 n N .. = L: j=l n where n = the number of lakes in the study group. (4-18) (4-19) Thus, if a study lake had a value for a variable during a particular sampling period equal to the annual mean, Nii would be 1.0. Values greater than 1.0 indicate a positive seasonal trend while values below 1.0 indicate a negative seasonal trend. Seasonal trends of chlorophyll a, total phosphorus and total nitrogen were analyzed for the two geographical subgroups of lakes included in acid lake study: the 13 Trail Ridge lakes in northern Florida and the seven Highlands Ridge lakes in southern Florida. These data are particularly we11-suited for an analysis of seasonal trends for several reasons. First, these two groups of lakes are located at opposite ends of the state and therefore represent the extremes in climatic conditions in Florida. For the northern the difference in mean daily air temperature between January and July is for the southern group the 'diffenenceis gn1y 16 QC... Second, most of the lakes in both groups are relatively undist1!l1!bac:tbyhuman activity and pulses in algal productivity are not likely to reflect man's activities (e.g., nutrient-rich runoff during agricultural fertilization). Finally, since all 20 lakes were studied by one group of investigators using standarized methods, differences between the two groups are not likely to be the result of variations in methodology. A plot of the mean normalized values for chlorophyll a throughout the year (Figure IV-10), shown with 95% confidence intervals, indicates that there are no statistically significant seasonal trends in chlorophyll a values for 59

PAGE 68

tl r-1 ..c C) '"Cl (]) N .,-l r-1 I-i 0 I=l I=l cU C]\ (]) a ;:E: 3.0 2.5 2.0 1.5 1.0 ...-,../ ....-0.5 D J F M A M Month ...-'" J SOUTHERN GROUP 4.,---+ -. NORTHERN GROUP T I I J A S T I I t 1\' I: o N Figure IV-10. Seasonal trends in chlorophyll a concentration for 10 northern Florida lakes and 10 southern Florida lakes. See text for calculations of mean normalized parameter values.

PAGE 69

either group of lakes. Similar results were obtained when total phosphorus and total nitrogen data were analyzed. Data compiled for Lake Apopka and the other Oklawaha lakes (Brezonik et al. 1978; Tuschall et al. 1979; Pollman et al. 1980) also indicate that while seasonal trends in chlorophyll a do occur, the trends are not consistent among lakes within a given year or in anyone lake throughout the threeyear study period. A plot of chlorophyll a data for Lake Apopka (Figure IV-II) shows that while there are major oscillations in the chlorophyll a standing crop, the timing of periods of bloom and scenescence vary from year to year. For example, in 1977 peak chlorophyll a concentrations occurred in the fall, while in 1978 peaks occurred in May and August and in the first half of 1979 a peak occurred in April. Periods of scenescence show a similar lack of temporal regularity: minimum chlorophyll a values occurred in March and May of 1977 and in February and June of 1978. This analysis suggests that there are no (or only small) regular seasonal trends in nutrient and chlorophyll a levels in Florida rakes. This conclusion serves as a basis for using annual means for nutrient and chlorophyll a concentrations, rather than seasonal means (e.g., mean spring total phosphorus; mean summer chlorophyll a) in the refinement of nutrient loading models for Florida lakes (Chapter V). DEVELOPMENT OF A TROPHIC STATE INDEX A trophic state index (TSI) that allows the ranking of lakes along a linear gradient is useful for several reasons: 1) a linear ranking system facilitates comparisons of trophic state within a group of lakes, 2) use of a TSI obviates the need to place a lake into a discrete trophic class (oligotrophic, meso trophic eutrophic), 3) a TSI can be used to quantify historical changes in trophic state and thereby assess the impacts of cultural perturbations, and 4) a numerical index can be comprehended by the public. Trophic state indices have been developed based on both single measures of trophic state (univariate indices) and on a composite of several trophic state indicators (multivariate indices). Trophic state variables that have been used in indices include dissolved oxygen, total and inorganic phosphorus and nitrogen, Secchi disk transparency, chlorophyll a, primary production, and the relative abundance of major ions). Concepts and applications of trophic state indices and the relative merits of univariate versus multivariate indices have been reviewed and discussed by Shapiro (1976), Brezonik (1976) and Carlson (1977). Trophic state indices that might be applied to Florida lakes include those of Shannon and Brezonik (1972), the National Eutrophication Survey (1975), and Carlson (1977). The National Eutrophication Survey developed a "water quality index" (WQI) based on six parameters related to trophic state. The WQI has two major shortcomings in ranking Florida's lakes: 1) its use of a dissolved oxygen parameter and 2) its use of two phosphorus parameters (median total phosphorus and median dissolved phosphorus). The dissolved oxygen parameter (a constant value minus the dissolved oxygen concentration in the bottom waters) is inappropriate because most of Florida's -61

PAGE 70

0\ N H ....... bD ;:::l. 100 tl ,..., ,..., :>, ...c p.. 0 H 0 ,..., ...c u 7St 50 !\ /\ II 4II "...25 ... J __ ...IL_,.J, .... ..J.I--III-.l--I-...&.......1..-.4 J F 11 AM J j AS 0 N D J F M A MJ J AS 0 N D J F M A M J Figure IV-II. Chlorophyll a concentrations in Lake Apopka, January, 1977 to June, 1979.

PAGE 71

lakes are not stratified. Rankings made using the D.O. parameter for the Florida NES lakes were widely scattered and not well-correlated with other trophic state indicators. The use of two phosphorus parameters in the WQI overemphasizes the importance of phosphorus as a trophic state indicator, particularly for Florida's many nitrogen-limited lakes. The index developed by Brezonik and Shannon (1971) using data from 55 Florida lakes could reasonably be applied to other lakes in the present data set except that data on primary production and cation composition required for the computation of this TSI were not collected in the Florida NES and the acid lake survey. Furthermore, values of chlorophyll a, total organic nitrogen and specific conductivity for several lakes lie outside the range of values in the original data base used to construct the TSI. Carlson (1977) developed three separate univariate indices of trophic state, based respectively on Secchi disk transparency, total phosphorus and chlorophyll a. Carlson's index was initially based on Secchi disk transparency, with values scaled so that the zero point corresponded to a Secchi disk value greater than any value yet reported (64 m): TSI(SD) = 10(6 1n(SD)/ln 2), (4-20) where SD = Secchi disk transparency (m). Trophic state indices for phosphorus and chlorophyll a were developed using the relationships between these parameters and Secchi disk transparency. Thus, for a group of 147 lakes the relationship between Secchi disk transparency and chlorophyll a concentration was best expressed (Carlson 1977) using the equation: 2 1n SD = 2.04 0.68 1n (ch1 a); r 0.86. (4-21) A TSI based on chlorophyll a was computed by combining equations 4-20 and 4-21: TSI(CHA) = 10(6 -(2.04 -0.68 1n(ch1 a)/ln 2. (4-22) A TSI based on phosphorus was based on an observed inverse relationship between total phosphorus and Secchi disk transparency: SD = 64.9/(P)1. (4-23) Combining eqs. 4-21 and 4-23 results in a TSI based on phosphorus: TSI(TP) = 10(6 -1n(48/(P)1)/ln 2). (4-24) The indices developed by Carlson have several advantages, including small data requirements, objectivity, and reliance upon commonly measuI!ed and understood indicators of trophic state. However, linear regressions of Carlson's TSI(TP) against both TSI(SD) and TSI(CHA) showed that the TSI(TP) values were significantly different (at the 95% confidence level) from the other two TSI values for the Florida NES lakes. This is not surprising, 63

PAGE 72

considering the predominance of nitrogen-limited lakes in the NES data set. Consequently, we developed a nitrogen-based TSI using a computation method analogous to that used by Carlson in developing the phosphorus-based TSI (Kratzer and Brezonik 1981). For the NES lakes, the relationship between chlorophyll a and total nitrogen is expressed by the equation: 2 In(chl a) = 2.44 + 1.6 In(N)l; r = 0.89 (n = 39). Combining equations 4-21 and 4-25 yields: SD '" 1. 46 (N)l A TSI(TN) can be calculated by substituting eq. 4-26 into eq. 4-20: TSI(TN) = 10(6 -In(1.46/(N)1)/ln 2). (4-25) (4-26) (4-27) Ideally, when both TSI(TP) and TSI(TN) are computed, the smaller of the two TSI's should represent the limiting nutrient for any given lake. This hypothesis is generally supported by the Florida NES data. The five lakes (Yale, Kissimmee, Marion, Reedy and Apopka) with TSI(TP) values considerably lower than TSI(TN) all were phosphorus-limited according to the NES algal bioassay results. For this study the lesser of TSI(TP) and TSI(TN) was averaged with the corresponding TSI(SD) and TSI(CHA) to compute a TSI(AVE). The values of the four univariate TSI's plus TSI(AVE) are listed for the 101 study lakes in Table IV-4. The TSI(AVE) values correspond well with assessments made by the NES limnologists (Lakes 1 to 40) and by Brezonik and Shannon (1971) (lakes 41 to 92). In addition, the TSI(AVE) values for the NES lakes agree reasonably well with the values of the EPA's Water Quality Index (r2 = 0.64). The use of a TSI(AVE) has the advantage over the use individual TSI's in that the use of a single index of trophic status can be more easily used for comparative and management purposes. Furthermore, the compositing of physical, biological and chemical components of trophic state into one index reflects the multidimensional nature of the eutrophication phenomenon, since it is generally agreed that no single trophic indicator adequately measures the underlying concept. Combining the major physical, chemical and biological indicators of trophic state into a single index smooths out the variability associated with individual indicators and provides a reasonable composite measure of trophic conditions in a lake. 64

PAGE 73

Table IV-4. Trophic Status Index for study lakes. Lake TSI(CHL) TSI(TP) TSI(TN) TSI(SD) TSI(AVE) Assessment 101 30 31 31 30 30 84 35 44 33 34 34 u 89 34 40 30 39 34 U 93 35 34 35 35 35 78 34 41 36 42 38 U 77 36 42 41 38 38 U 91 38 41 37 40 38 U 79 37 41 38 40 38 U 82 35 43 37 43 38 U 86 39 45 40 37 39 U 92 36 46 42 43 40 U 81 34 45 46 43 41 U 80 41 41 53 41 41 U 87 43 46 43 42 43 U 90 40 49 49 42 43 M 96 40 42 35 55 43 53 39 41 49 54 44 U 1 42 47 49 47 46 O-M 100 50 41 42 47 46 47 40 50 59 48 46 0 99 42 63 38 58 46 65 41 56 51 51 48 M 41 47 56 48 49 48 0 98 52 43 46 / 50 49 5 49 56 51 47 49 M 62 48 102 46 54 49 H 76 49 50 55 50 50 0 61 45 57 56: 51 50 M 95 54 48 43 55 51 43 49 54 51 52 51 (j 94 55 50 45 52 51 42 45 56 53 55 51 0 83 45 52 53 56 51 0 44 48 58 50 56 0 2 47 57 54 52 M 3 52 61 52 53 52 M 52 51 71 46 60 52 0 88 44 56 53 61 53 0 51 51 59 54 54 53 0 66 60 74 52 50 54 M 4 51 59 52 59 54 M 50 54 66 55 55 55 M 85 47 56 53 64 55 U 21 53 70 51 60 55 E 70 50 61 56 60 55 U 49 50 74 56 61 56 M 73 52 87 56 60 56 M 6 49 59 57 62 56 M-E 67 63 86 51 54 56 M 46 49 62 53 66 56 M 65

PAGE 74

Table IV-4. continued ... Lake TSICHL): TSI(TP) TSI(TN) TSI(SD) TSI(AVE) Assessment 97 52 65 51 66 56 7 62 52 59 54 56 E 10 55 60 55 60 57 E 56 54 57 57 59 57 M 69 56 61 55 60 57 M 9 63 61 54 57 58 E 74 57 57 61 60 58 M 55 56 64 58 61 59 M 16 49 68 63 64 59 E 15 53 66 56 66 60 E 24 63 74 58 58 60 E 11 63 63 56 60 60 E 8 62 56 62 62 60 E 54 59 63 60 62 60 M 45 60 78 60 61 60 M 19 61 79 59 64 61E 12 61 66 61 62 61 E 18 66 55 64 64 62 E 17 63 68 61 62 62 E 13 57 65 64 65 62 E 14 64 60 66 63 62 E 29 57 80 64 67 63 E 25 66 99 6il 62 63 H 64 65 90 60 63 63 E 59 60 79 64 65 63 E 58 66 74 56 67 63 M 72 65 78 64 61 63 E 27 65 76 62 64 64 E 32 70 107 62 62 64 E 28 64 88 64 65 64 E 48 69 68 63 64 65 E 75 61 79 72 64 66 H 57 66 71 63 68 66 E 63 68 94 64 66 66 H 31 72 93 67 62 67 E 68 64 92 70 69 68 :H 71 70 95 67 67 68 E 23 67 81 71 67 68 E 36 73 105 70 64 69 H 35 74 100 69 66 70 H 26 76 83 68 66 70 E 60 74 93 66 72 71 H 20 69 90 76 74 73 H 34 73 94 69 77 73 H 30 76 98 77 73 75 E 22 68 89 78 81 76 H 33 83 97 76 71 77 H 38 79 113 79 74 78. H 40 86 109 79 72 79 E 39 85 109 78 77 80 H 37 74 118 79 89 81 E U = u1trao1igotrophic, 0 = oligotrophic, M = mesotrophic, E = eutrophic, H = hyperutrophic. Assessment by NES (lakes 1 to 40) or Shannon and Brezonik, 1972 (lakes 41-92). 66

PAGE 75

CHAPTER V. APPLICATION OF NUTRIENT LOADING MODELS TO THE FLORIDA NES LAKES INTRODUCTION In the past decade considerable efforts have been made to quantify the relationships between nutrient loading and lake trophic status using simple input-output (I/O) models. Nutrient loading models have been widely used to predict the effects of changes in nutrient loading on lake trophic status (Vollenweider 1969, 1975, 1976; Patalas and SaIki 1973; Dillon and Rigler 1975; and many others) and to predict the trophic status of new reservoirs (e.g. Bradford and Maiero 1978; Baker et al. 1978; Huber and Brezonik 1980). The utility of these models is greatly enhanced by their modest data requirements and computational simplicity. Thschapter reviews the developments made in mass balance nutrient models for lakes over the past decade and the resulting advances in our ability to predict trophic conditions in lakes from information on phosphorus and hydraulic loadings and basic lake morphometry. Nearly all these predictive models have been developed using data on temperate lakes. This chapter describes the application of these models to warm-temperate and subtropical lakes in Florida and evaluates their usefulness in predicting trophic conditions in Florida lakes. HISTORICAL DEVELOPMENT Phosphorus Input/Output Models The relationship between nutrient concentrations and algal productivity in lakes has long been recognized (see review by Vollenweider 1968). Since phosphorus has been identified as the most common limiting nutrient in temperate lakes, the development of nutrient loading models has focused entirelyon phosphorus, although Vollenweider (1969) noted that the principles involved could be applied to other nutrients. In the development of his phosphorus loading model, Vollenweider (1969) expressed the change in the mass of phosphorus in a simplified form: dP/dt = J -L -out S (5-1) where dP/dt = rate of change of the mass of P in a lake. J = flux of P into the lake, L = out flux of P from the lake via its and -67

PAGE 76

s the rate of loss of P via sedimentation and other mechanisms other than loss through the outlet. In formulating his model, Vollenweider made several simplifying assumptions: (1) A lake behaves like a continuously stirred tank reactor. That is, any substance entering a lake becomes completely mixed as soon as it enters. (2) The rate of sedimentation is proportional to the amount of phosphorus in the lake, i.e., S = Gp P, where P = the mass of phosphorus in the lake and cr is a first order seaimentation coefficient with units of yr-1 p (3) The concentration of phosphorus in the outflow is equal to the concentration of phosphorus in the lake. Thus, Lout = Pw(P)l' where Pw is the hydraulic flushing coefficient (Q/V) in yr-1 ana (P)l = mean lake phosphorus concentration. (4) There is no seasonal fluctuation in loading. Although none of these assumptions is entirely valid, they allow the development of a simple expression to compute (P)l: dP/dt = J cr P -P P, (5-2) P w If it is further assumed that the system is at steady state, i.e., dP/dt = 0: J = p P + cr P; or P w p J Dividing through by lake volume (V) yields P J/V -= V P + cr w p and since P/V = [PI1 and J/V L [P] 1 = __ v,--_ crp + Pw or Lv = crp[P]l + = L (the volumetric loading rate), v (5-3) (5-4) (5-5a) (5-5b) Phgsphorus loading usually is expressed on an areal basis (in g/m2_yr), as Ln = L /z, where z = mean depth: K" V [P] 1 z (p + cr ) w p = L It 68 -(5-6)

PAGE 77

where q = hydraulic loading (m/yr) = z p. Since L z and p can readily be measITred, the major difficulty in this mode! is the of the sedimentation' coefficient, cr. Vollenweider (1975) found that for a group of 25 temperate zone cr = 10/2 Substituting this expression into eq";' (5-6) yields p 10 + q s (5-7) This relationship is tantamount to stating that the apparent deposition or settling v'e1ocity for total phosphorus in lakes is constant. If cr = 10/2, then cr. z = constant; 0"'.2 has units of m/yr, which dimensionally iE a velocity. TRis term can be iRterpreted as the settling velocity for total P and is given the It should be noted that both vp and cr are not subject to strict physical interpretation and measurement, since p there is more than a single mechanism (and a single form of phosphorus) involved in deposition to sediments. Moreover, sediment deposition is not the sole internal sink for phosphorus, although on a long term basis it is by far the most important. Phosphorus also can be lost from the water column via uptake by macrophytes or incorporation in fish biomass; neither of these reservoirs is measured in typical phosphorus budgets. Thus the basic mass balance model for phosphorus is a simplification of reality, and the sedimentation term is a composite of all internal sink processes, including deposition (settling) of detritus, adsorption of orthophosphate by sediments, and uptake by macrophytes followed by direct incorporation of dead macrophyte tissue into the sediments. Consequently, it is not possible to measure crp or vp directly, although sedimentation traps may provide good approximations under certain limited conditions. Jones and Bachman (1976) found that for a group of 16 Iowa lakes, the best fit for eq. (5-6) was obtained using a constant value of 0.65 for cr Thus for their data sedimentation rate was independent of depth, and eq.p (5-6) becomes [P] = 0.84Lp ==-=-=-q + O.65z s More recently, Vollenweider (1976) proposed that cr = 1/;.r-, is the hydraulic retention time (TW = Pw-1 ; hence TwP= dlen becomes: q (1 + s (5-8) where Eq. 5-6 (5-9) It is to be noted that in developing eq. 5-9, Vollenweider has gone beyond the dimensiona11y-and theoretically-correct (albeit perhaps simplistic) mass balance model and has interjected an element of empiricism into the phosphorus predictive equation. An alternative approach to that requlrlng a determination of cr was proposed by Dillon and Rigler (1974). These workers proposed that Rn easily measured retention coefficient, R can be used to replace cr in Vollenweider's model: p p 69

PAGE 78

R P 1 Qout [P] out Qin [P]in (5-10) which is simply the fraction of the input phosphorus that is retained within a lake. If the lake is at steady state, it follows that the phosphorus retained must be added to the sediment. The sedimentation coefficient can be expressed in terms of R. Since L = cr [P]l + p [P]l (eq. 5-5b), p v p w L eJ + P V P w (5-11) [P]l From eq. 5-5b and the nature of steady state systems, it also is clear that RpLv = eJp[P] 1. Substituting this relationship into eq. 5-11, we obtain: eJ P and R P Rp' P w 1 -R P cr p cr + P P w Substituting eq. 5-12 into eq. 5-6 and noting that q s L (1 -R ) P P [P] = L (1 -R ) 1 P P Z P w (5-12) (5-13) z P ,'( we obtain w (5-14) Dillon and Rigler (1974a) developed loading criteria plots based on eq. 5-14. Plots of Lp(l -R )/p vs. z have a slope of equivalent to the steady state concentration of phosphorus ([P]l) in a lake and can be used to predict trophic conditions (see following sect10n). Larsen and Mercier (1975) presented an alternative approach that is derived directly from Vollenweider's model but emphasizes the importance of fne in:fIuentconcent:ra-fion 6rpliospnorus;[P] .; rather than thetotaLToaaing: Since L:Q=JiiA, where Ji is the total input phosphorus (g/yr), and since q = Q.IA, it is clear that Lp/q = J.IQ. = [PJ .. Substitution of this rela into the Dillon-Rigler (e&. 5-14) results in: [P] 1 = [P]. (1 -R ). 1 P (5-15) Finally, Chapra (1975) introduced the concept 0 f an "apparent settling velocity", vp where vp = cr Z, as discussed earlier. Substituting this term into eq. 5-13 results in p R P v p ., v + q p s Combining equations 5-14 and 5-15 yields v + q P s 70 -(5-16) (5-17)

PAGE 79

The apparent settling velocity can be determined by rearranging eq. 5-16 and solving. If vp is assumed to be constant, R can then be calculated directly knowing only qs. This relationship is sound in that when q = 0, R must be 0, regardless of the magnitude of v. Chapra (1975) foundsa of 16 m/yr for v in a group of Ontario whereas for a somewhat larger group of lakes PKirchner and Dillon (1975) found that vp was 13.3 m/yr (using eq. 5-16 and measured values of Rp and qs). As noted earlier, Vollenweider's (1969, 1975) formulation, crp = 10/z, corresponds to a vp of 10 m/yr. In addition to the fundamental I/O approach to modeling lake phosphorus dynamics, several empirical models have been developed. Two such models recently were proposed as statistical improvements on the previous predictive models for total phosphorus concentration (Reckhow 1977, Walker 1977). The Reckhow (1977) model was developed by using a nonlinear regression to account for those variables that produce a nonlinear response in the total phosphorus concentration. The resulting equation, based on 33 north temperate lakes with qs < 50 m/yr, was: L (P) = P 1 10 + z + 1.05 qs exp (0.012 qs) (5-18) Walker (1977) developed a model and came up with the predictive L from a data base of 105 north temperate lakes equation: (P) = _____ ....... 1 (1 + 0.824 T 0.454) qs w f5-19) Imboden (1974) and Snodgrass and 0'Me1ia (1975) have developed more p1icated models of phosphorus that divide lakes vertically into two compartments (the epi1imnion or upper, mixed layer, and the hypolimnion, or lower stagnant layer). Their models require the solution of four coupled differential equations for the summer (stratification) period (ortho-P and particu1ateto be solved for the winter (circulation) period (ortho-P and particu1ate-P for the entire lake). Data requirements for these models are more compli-cated and include: (1) phosphorus exchange coefficients between the epi1imn-ion and the hypolimnion and between the sediment and overlying water; (2) rate coefficients for photosynthesis and mineralization; (3) settling coefficients for particulate phosphorus; (4) loading rates for both forms of phosphorus; (5) water flow rates; and (6) depths of the epi1imnion, thermocline, and hypolimnion. These models produced one surprising result: deep lakes were predicted to have higher phosphorus concentrations in their euphotic zones than shallow lakes, a result that is contradictory to the observations of several investigators. Snodgrass and 0'Me1ia (1975) proposed to resolve this discrepancy by assigning a depth dependence to the exchange doefficient across the thermocline and to the effective settling velocity of particulate phosphorus. Since most Florida lakes are thermally unstratified at all times, the complications introduced by these two-compartment models are unnecessary. -71

PAGE 80

Prediction of Rp An advantage of the Dillon-Rigler ,and Larsen-Mercier models is that can be determined experimentally (eq. 5"":LO) from phosphorus budget measure ments. However, in some important applications, is not known. For example, in predictive studies on proposed reservoirs, Rn obviously cannot be measured directly. Also, it is often desirable to preaict for an existing lake for which a complete phosphorus budget is not available or for which proposed management activities may change hydraulic and or nutrient loading characteristics. These needs have led to several empirical attempts to predict from other easily measured limnologica1 variables, particularly hydraulic parameters such as qs and pw. Kirchner and Dillon (1975) derived a double exponential equation to predict from qs using a data base of 15 southern Ontario lakes: R = 0.426 exp (-0.271 q ) + 0.574 exp (-0.00949 q ). p s s (5-20) The correlation coefficient for this relationship based on 15 lakes in southern Ontario was 0.94; furthermore, the relationship is reasonable in that it gives an R of 1.0 when qs = 0 (i.e. when there is no flow from the lake). p Larsen and Mercier (1975) evaluated a variety of empirical formulations to estimate using a data base of 20 temperate lakes, for example: Rp = 0.854 0.142 qs' (5-21) and R = 0.482 0.112 1np p w (5-22}1 Equation 5-21 yielded a correlation coefficient of 0.92 using all 20 lakes in the data set. When the two shallowest lakes were excluded, eq. 5-22 gave nearly as good a fit. Larsen and Mercier noted that these equations do not provide theoretically correct predictions for lakes with extreme values of qs or Pw, in which cases unreasonable predictions of Rp (>1 or <0) could occur. A relationship that was more theoretically sound was obtained by Larsen and Mercier from the finding that (Jp (as estimated by mass balance models) was related to the flushing coefficient (p ) for the 20 lakes: w 1n (J = 0.472 1n p 0.273, (r = 0.84). 'p w (5-23) Substituting this term into eq. 5-13 and simplifying resulted in the approximate relationship: R P 1 1 + p 2 W Prediction of Ch1orophy11rq Concentration (5-24) A major advance in the development of nutrient loading models came with the recognition of a relationship between the concentration of phosphorus 72

PAGE 81

during spring turnover (P)sp and the mean summer concentration of chlorophyll a (Dillon and Rigler 1974a; Sakamoto 1966; Jones and Bachman 1976; see Chapter 4). These correlations have led to the development of models that can predict the chlorophyll a concentration directly from data on nutrient loading, morphological features and hydraulic parameters (Vollenweider, 1976; Jones and Bachman 1976; Chapra and Tarapchak 1976). For Vollenweider (1976) regressed the right side term of eq. 5-9 against mean summer ch1orophy.11 a (ch1 a)s to obtain the relationship: [ch1oJ = 0.367 [L/qs ]0.91. (5-25) s 1 + lIT w In a similar manner, Chapra and Tarapchak (1976) combined Dillon and Rigler's equation (5-14) and Chapra's equation (5-16), setting vp = 12.4, and regressing the resulting term against [chI gJ s to obtaih [ch1 a] + 12.4 L 1.449 p ] (5-26) Finally, Hand (1975) used an empirical approach to predict [ch1 a] for Florida lakes using a shape factor (5) and the outflow phosphorus concentration (P / out Qout): [ch1 a] = 114 (N t/3 + P )/Q ou out ou (5-27) Nutrient Loading Criteria: Graphical Approaches Para11eI1ing the development of models to predict [P] 1 ,and later (ch1 fO s is the development of critical loading plots that depict critical loadings (i.e., permissible loadings that maintain oligotrophic conditions and excessive rates above which eutrophic conditions occur). These criteria. are plotted as functions of lake morphometry and/or hydraulic conditions. The first such plot was developed by Vollenweider (1968) and was empirical in nature. graph (Figure V-1J--Uepicts trophic status as a function of areal phosphorus loading (Lo) and mean depth (z). Brezonik and Shannon (1971) developed a similar graph showing the relationships between nitrogen and phosphorus loading and a trophic state index for Florida lakes. The permissible and excessive loading rates they developed were higher than Vollenweider's corresponding loading criteria, implying that Florida lakes are capable of assimilating more nutrients than are the temperate lakes used by Vollenweider. Although Vollenweider's 1968 loading plots were widely used following their introduction, Vollenweider and others realized that the failure of the plots to account for hydraulic characteristics limited their usefulness. Thus Vollenweider (1975) introduced a second critical loading plot based on his phosphorus loading model (eq. 5-6). Critical phosphorus levels were considered to be 10 and 20 pg/L, respectively, as lower limits for mesotrophic and eutrophic conditions. The resulting plot of Lp vs. qs (Figure V-2) has three distinct segments: 1) q < 4 m/yr. L -constanu. (dependent only on (J ). s cr1t p -73

PAGE 82

10 100 Q) 5", .I-J cd p::: 50 bO I=l -r-! 0 1 H C/J H Eutrophic Zone 10 ::J :>.. 0.5 H ON 5 ..c: is p.. ....... C/JP-! 0 ..c:bO P-!' Hi 0.1 cd i .I-J 1 0 0.105 E-i H 0.5 cd g 0.01 0.1 0.5 1 5 10 50 100 Mean Depth (z) m Figure V-I. Vollenweider's phosphorus and nitrogen loading criteria (1968), L versus z. Q) .I-J bO I=l -r-! "t:l cd 0 H 10 5 Eutrophic 1 Excessive Permissible Oligotrophic Zone ______ ______ __ ______ 0.1 0.5 1 5 10 50 100 Hydraulic Loading Rate (q ), m/yr s Figure V-2. Vollenweider's phosphorus loading criteria (1975), L versus q p s -74

PAGE 83

2) 40 m/yr > q > 4 m/yr. L .t proportional to q (and a ). s cr1 s p 3) qs > 40 m/yr. Lcrit proportional to q and a ,I ... being damiuant. s p s This model is an improvement over Vollenweider's early depth-loading plots in that the critical loading is recognized to be a function of the areal water. Dillon (1975) introduced a critical loading plot based on the model developed by Dillon and Rigler (eq. 5-14), again using 10 and 20 for cri tical levels of [P] 1. In this case, L (1 Rp) TW is plot ted vs. z; the slopes of the critical lines are 10 and 28 respectively (Figure V-3). Finally, Larsen and Mercier (1975) used eq. 5-15 to construct a plot relating (P). to Rp (Figure V-4). This plot delineates zones of oligotro-Phic, mesotr6phic, and eutrophic lakes on the b f Rd aS1S 0 p an average 1n-f1uence concentrations of total phosphorus. APPLICATION OF NUTRIENT LOADING MODELS TO FLORIDA LAKES The nutrient loading models described above were developed using data from temperate lakes, and their validity has not been evaluated for subtropical lakes. As demonstrated in Chapter IV, Florida lakes differ considerab1y from temperate zone lakes in their 1imno1ogica1 characteristics. Unlike temperate zone 1akes,:F1orida lakes do not undergo a dimictic pattern of stratification, as do most temperate zone lakes. Many are quite shallow and harbor large beds of macrophytes. Overall, they are considerably more colored than most temperate zone lakes. Finally, nitrogen tends to be the limiting nutrient for many Florida lakes, raising the question of whether loading models based on phosphorus are applicable. Thus, the objective of this phase of the study was to evaluate the use of I/O models using the Florida NES data base .. Models to predict (P)l, (N)l' (chI: a) and .... retained coefficients(Rp and RIl) were analyzed using regression analyses. The GLM procedure of SAS was used to select the best models and to determine the 95% CL] for these models. Since seasonal variability is less pronounced in Florida lakes compared to temperate zone lakes (Chapter IV), annual mean values of (P)l, (N)l and (chla) were used in these analyses. Available phosphorus loading plots (Vollenweider 1968, 1975; Dillon 1975; Larsen and Mercier 1975) were examined and modified to fit Florida lakes. Analogous loading plots for nitrogen also were developed since many Florida lakes are nitrogen-limited. TotaZ Phosphorus Concentration. Most of the predictive equations for total phosphorus ([P]1))ana1yzed here are based on equations developed by previous investigators. Coefficients for the equations were modified using regression analysis to improve their predictive capability for the Florida NES lakes. Log-log transformations of both (P)l and the predictive terms in the equations were used because of the wide ranges of encountered values. -75

PAGE 84

N S P-i co o 5., 1 0.5 Eutrophic Zone Oligotrophic Zone 1 5 10 50 100 500 1000 Mean Depth (z), m Figure V-3. Dillon's phosphorus loading criteria (1975), with lines of constant phosphorus concentration distinguishing trophic states. 1000 .t:: 500 P-i co ;:L !=l 100 o 'M +J t1! t 50 C) !=l o u C/l ;:l H o 10 ,..c: Po C/l o ,..c: P-i 5 +J Eutrophic Zone Oligotrophic Zone ;:l r-I 44 1 o 0.1 O. 2 O. 3 o. 4 o. 5 o. 6 O. 7 O. 8 o. 9 R Figure V-4. The Larsen and Mercier phosphorus loading criteria (1975), 76

PAGE 85

Equation TP5 is based on the concept of a "mean apparent settling velocity" for phosphorus (Chapra 1975); vp was calculated for each of the Florida NES lakes by rearranging equation 5-16: qs Rp vp= __ (5-28) 1 -Rp The mean value of vp for the Florida NES lakes, 8.53 m/yr, was then substituted into equation 5-17 to produce TP5. Eight equations were evaluated (Table V-I), and the r2 values ranged from 0.77 to 0.91. The best predictive equation, TP2, is based on Dillon and Rigler's (1975) model. Prediction of total phosphorus using TP2 requires data on areal phosphorus loading, hydraulic loading, and the phosphorus retention coefficient (R p)' To obtain the best estimates of lake phosphorus concentration the measured phosphorus retention coefficient [Rp = (EPin EPout)/EPinJ should be used in TP2. However, since the outflow phosphorus loading may not be known or economically determined, it may be necessary to predict Rp from morphometric and hydrological parameters. The ability to predict Rp in Florida lakes by equations involving such parameters is limited (see p. 80), and estimates of (P)l made using predicted values of Rp in TP2 will not be very accurate. Several other equations that do not require data on outflow phosphorus loading (TP3 to TP6) have r2 values 0.82 and 0.84, but these equations all have C.V. values higher than The 95% confidence limits for individual predictions shown in Figure Vj:-5 can be used to evaluate the ,accuracy of predictions made using TP2. For example, a lake whose predicted phosphorus concentration is 0.100 1Ilg/L has a 95% CLI of 0.057 to 0.116mg/L. TotaZ Nitrogen Concentrations. The nitrogen balance in a lake can be expressed by the following equation: dN/dt = N. -N + N N N out fix -sed -den (5-29) where dN/dt change in the mass of nitrogen in a lake, N. flux of nitrogen ,into lake, N out Nf Nden = flux of rate of rate of rate of nitrogen throughout outflow, nitrogen fixation, denitrification, and nitrogen loss by sedimentation. Since N f and Nden were not determined for the NES lakes, these fluxes are grouped with Nsed as parts of a composite loss term, aN N. The nitrogen mass balance equation thus simplifies to: dN/dt = N. -p N a N (5-30) w N which is analogous to the pho.sphorus mass balance (eq. 5-2). At steady state: dN dt = 0 = Nin (5-31) or N (5-32) 77

PAGE 86

-....:J 00 Table V-I Predictive 1 equations for total phosphorus concentration Predictive Equation* TP1=0. 682 [L /q (l+h)] 0.934 p s w TP2=0.748[L (l-R )/q ]0.862 p p s TP3=0.984[L (l-R )/q] 0.851 p (K+D) s TP4=O. 706 [0. 84L / (0. 65z+q )] 0.9:64 p s TP5=0.952[L /(8.53+q )].860 p s L TP6=0.885[ P jO.968 18z -+1.05q exp(0.0121 ) 10+z s s L TP7=0.643 [ p l 01. 932 Qs(1+0.824TwO.454) TP8=0.416[L /q ] 0.873 p s Original Equation Investigator Eqn. no. in text Vollenweider 5-9 (1976) Dillon and 5-14 Rigler (1975) Dillon and 5-14 (1975) Jones and 5-8 Bachman (1976) Chapra (1975) 5-17 Reckhow 5-18 (1977) Walker (1977) 5-19 This study Modified Equation 2 r n** 29 25 28 0.84 29 29 0.88 29 0.91 29 26 2 c.v. r 0.79 31.3 0.91 14.8 0.84 23.6 0.82 29.3 0.83 28.1 0.82 29.1 0.78 32.:6 0.77 30.0 For each predictive equationl (except TP10) TP. = a [Original Equation]b where a and b are constants determined by regression. 1 ** n is the number of lakes inc[uded in the regression.

PAGE 87

H p... b.O S "' !=l 0 .,-j .(..J til H .(..J !=l Q) CJ !=l 0 CJ C/l ;:I H 0 ,.c: p. C/l 0 ,.c: P. '"Cl Q) .(..J CJ .,-j '"Cl Q) H p... 10.0 1.0 0.10 = 0.748[L (l-R)/ ]0.862 p r2 = 0.91 95% eLI shown 0.01 0.10 1.0 Measured total phosphorus, mg P/L Figure V-5. Predicted vs. measured total phosphorus concentration using equation TP2. 79 10.0

PAGE 88

Dividing by volume (V), we obtain a predictive equation for the steady-state concentration of total nitrogen: (5-33) Since there are no existing equations to predict total nitrogen concentration, the equations evaluated here were based on modifications of phosphorus loading equations. For example, TNI was derived by setting aN = 0.65z and substituting Ln for Lp; TN2 -TN3 were derived by substituting Ln for Lp and R Equation TN5 was derived by computing a mean apparent settling tor nitrogen (vn) in Florida NES lakes in a manner analogous to that used above .. The predictive equations (TNl-TN4) have r2 values ranging from 0.52 to 0.77 (Table V-2). The'best predictive equation, TN2, is based on Dillon and Rigler's (1974a) model, with Ln and Ru substituted for Lp and Rp. TN2 requires data on areal nitrogen loading, hydraulic loading and the nitrogen retention coefficient. The use of measured retention coefficients will produce the most accurate estimates of mean lake nitrogen concentration. However, in some cases measured values of Rn may not be available; it may therefore be necessary to use predicted values of Ru. Equations to predict Ru, evaluated below, are unfortunately not highly accurate. Thus, when it is necessary to use predicted values of Ru in TN2, the resulting predictions of (N)l will be of limited usefulness. Although the other three predictive equations do not require data on the outflow nitrogen loading, their predictive capacity is considerably lower than that of TN2. The small 95% confidence limits for individual predictions shown in Figure V-6, indicates that good accuracy can be achieved in predicting [N]l using TN2. Phosphorus and Nitrogen Retention Coefficients. Since the best equations to predict total phosphorus and total nitrogen require the use of retention coefficients, it is desirable to be able to predict Rp and using data on the hydrologic and morphologic characteristics of a lake. Several investigators (Kirchner and Dillon 1975; Larsen and Mercier 1975) have shown that this approach can be successful in The seven predichere (Table V 3) were-based on vious equations using the parameters Z. TW, and qs. Unfortunately, none of the equations is successful in predicting The r2 values for the equations ranged from 0.41 to 0.53. A plot of the measured phosphorus retention Rp versus the yalues by the best predictive equation (RP6) shown in Figure V-7 illustrates the substantial scatter in the relationship, and the 95% eLI for new predictions shows that at mean value of Rp (0.48) the predicted values is approximately + 0.4 of the actual value. The best predictive equation (RP6) was developed here by combining mean depth (z) and water residence time (Tw) by multiplication, rather than by division as is done to define the areal hydraulic loading rate. The physical meaning of this factor (Z.TW ) is uncertain, and the difference in predictive capability between RP6 and the other equations is too small to warrant conclusions about the relative importance of mean depth and water residence time as predictive parameters of phosphorus retention. 80

PAGE 89

00 J-L Table V-2. Predictive equatiops for total nitrogen concentration Predictive Equations* TNI = 1.08 [L /(0.65i + q )]0.859 n s TN2 = 0.899 [L /(1 -R )/q ] 0.976 n n s TN3 = 0.841 [Lh/qs]0.877 TN4 = 1.29 [To /q (1 + h)] 0.8158 11 s TW TN5 = 1. 69 [Ln/ (5.49 + 1 Original Basis for Equation Jones and Bachmann (1976) Dillon and Rigler (1975) This study Vollenweider (1976) Chapra (1975) n 27 24 27 27 29 2 r 0.53 0.77 0.52 0.55 0.19 For. each predictive TN3) the original equation was, a total phosphate predictive equatioi with Lp and Rp replaced by Ln and Rn' Thus, TNt = a[Transformed Original Equation]b where a and b are constants determined by the regression C.V. 69.7 47.6 70.2 67.9 91.5

PAGE 90

H Z '" l=l OJ bO 0 1-1 -I-l .,-l l=l r-l Clj -I-l 0 -I-l '1j OJ -I-l CJ .,-l '1j OJ 1-1 P-! 1.0 0.1 0.1 0.899[L (l-R)/ ] 0.976 nn qs r2 = 0.77 95% eLI shown 1.0 Measured total nitrogen, mg NIL 10.0 Figure V-6. Predicted vs. measured total nitrogen concentration using equation TN2. -82

PAGE 91

00 tN Table V-3. Predictive equations I for phosphorus retention coefficient Original Equation Modil:rea Equatlon Predictive Equation* RPl = 0.767 0.367 log (q ) s RP2 = 0.639 + 0.355 log (TW) RP3 = -0.056 + 1.40/(l+;P-) w RP4 = -0.009 + 8.17/(lO+q ) s RP5 = -0.249 + 0.487 exp(-0.271 rs) +0.656 exp(-0.00949 qs) RP6 = 0.500 + 0.353 log (z TW) Investigator Eqn. no. in text Larsen and Mercier (1975) Larsen and Mercier (1975) Larsen and Mercier (1975) Larsen and Mercier (1975) Kirchner and Dillon (1975) This study 5-21 5-22 5-24 5-20 RP7 = -0.131 + 1.07 R(K+D) Hand (1975) -0 172(N /Q ) (R out out (K+D) shape/z o RP8 = 0.734[8.53/(8.53 + q )] Chapra (1975) s 5-16 r2 n 0.88 27 0.86 27 0.88 27 0.86 27 0.88 i 27 26 0.50 26 26 2 r C.V. 0.46 42.2 0.46 41.9 0.47 41.6 0.46 41.9 0.46 41.9 0.53 40.3 0.41 44.4 0.34 67.2 For each predictive equation (lrxcept. RP6) the variables used were the :same as those in the original equation, but he equation is altered by new constants a and b, such that RPi = a + b [Original Variable].

PAGE 92

r::t.P-< 0 'rl +J g +J CIJ !-I (J) ::I !-I 0 ..c: P-< (J) 0 ..c: P-< 'U CIJ +J tJ 'rl 'U CIJ !-I P-! 1.0 0.8 0.4 0.2 0.0 -0.2 -0.2 0.0 0.2 0.4 = 0.500 + 0.353 log (zoT) r2 = 0.53 95% eLI shown 0.6 0.8 Measured phosphorus retention coefficient, R p 1.0 Figure V-7. Predicted vs. measured phosphorus retention coefficients using Equation RP6. 84

PAGE 93

The predictive equations examined here to estimate Ru similarly were based on hydrologic and morphologic characteristics (RNI to RN5). These equations (Table V-4) like those examined to predict R p also were not very successful in. predicting nitrogen retention coefficients for Florida's NES lakes. The r2 values of the equations ranged from 0.12 to 0.51, and C.V. values ranged from 67.0 to 89.6. The width of the 95% confidence intervals shown for individual predictions in a plot of measured nitrogen retention vs. values predicted by the best equation (RN4) illustrates its poor predictive ability (Figure Equation RN4 predicts Rn as a function of the mean inflow nitrogen concentration (Ln/qs). The inclusion of the areal nitrogen loading Ln.. in the predictive equation resulted in avast improvement over-the other--equationsthafwere based eniirely--on-hydrolOgIc-and morphologic data. The poor correlations between observed and predicted values of Ru and R using predictive equations RNI-RN5 and could be due to several First, it is likely there are substantial errors in many of the NES water and nutrient budgets, particularly for many lakes where portions of the nutrient and water budgets were estimated rather than measured directly. In particular, nutrient inputs were estimated for ungauged tributaries and for some municipal wastewater treatment plants. It is also possible that morphologic and hydrologic factors are simply not good predictors of nutrient retention in Florida's lakes. Unlike temperate zone lakes, most of the Florida NES lakes are shallow and do not undergo thermal stratification. Nutrients lost via sedimentation thus may re-enter the water column more readily in Florida in temperate zone lakes. Pollman (unpublished data) has shown that resuspension of sediments in Lake Apopka during storms causes a significant, temporary increase in the concentration of soluble reactive phosphorus from sediment particles. Although the effect of this phenomenon on the long-term retention of not known, the data suggest that long-term retention may be affected. Biological processes also may be more important in affecting nutrient retention in Florida lakes than in temperate zone lakes because of the warm climate, long growing season and generally nutrient-enriched conditions. For example, release of sedimentderived phosphorus to the column by rooted aquatic macrophytes has been shown to be an important mechanism affecting the concentration of phosphorus in several lakes (Smith 1978, Lie unpublished ms). It seems reasonable that macrophytes may have important effects on nutrient retention in many Florida lakes, although this has not been studied. of chlorophylla. The predictive models for chlorophyll evaluated here (Table V-5) generally were based on loading expressions developed by previous investigators, although regressions also were determined for relationships between chlorophyll a levels and lake nutrient concentrations. CHA12 and CHA13 are regression equations that describe the relationship between chlorophyll a and the concentrations of phosphorus and nitrogen, respectively, in the Florida NES lakes. As would be expected for a group of primarily nitrogen-limited lakes, nitrogen is better correlated with ohi a (r2 = 0.79) than is phosphorus (r2 = 0.63). The coefficients for CHA13 are, nearly identical to those of eq. 4-11, which describes the relatiojJ.ship between (chI a) and (N) 1 for the 44 nitrogen-limited lakesi.n the entire set of UiOl study lakes. -85

PAGE 94

" Table V-4. predictive equationsi for nitrogen retention coefficient

PAGE 95

1.0 o.s 0.6 -0.2 0.0 0.2 0.4 ... 0.01 + 0.597 log (Ln/qs) r2 = 0.51 95% eLI shown 0.6 o.s Measured nitrogen retention coefficient, R n La Figure V-So Predicted vs. measured nitrogen retention coefficients using equation RN4. 87

PAGE 96

Table V-5. Predictive equations chlorophyll a concentration. Original Equation Modified Equation Eqn. no. 2 2 Predictive E uations* Investigator in text r n r C.V. CHAl=81. 3 [L /q (1 + .r.r)J 0.652 Vollenweider 24 0.75 29 0.59 21.1 P s w (1976) L CHA2=91.6 [ p JO.549 Vollenweider 5-30d 29 0.52 22.8 8.53+qs (1975) L CHA3=21.9[ n ]0.786 5-34a 29 0.32 27.1 5.49+qs CHA4=9.82[L (l-R )/q ] 1.58 .Dillon arid 26 0.60 19.0 n n s Rigler (J.974a) CHA5=89.3[L (l-R )/q ] 0.604 Dillon and 27 0.52 22.7 p P s Rigler (1974a) CHA6=8.30[Ln!(0.65Z+qs)] 1.71 Jones and 26 0.69 19.2 CHA7=73.4[L /(0.65z+q )] 0.667 Bachmann (1976) 00 Jones and 29 0.60 20.9 00 \ P L S J .676 Bachmann (1976) CHA8=97.5 P Reckhow (1977) 29 0.61 .20.5 18z [----+1.05q exp(0.012q 10+z s s L CHA9=78.7[ E ]0.6571 Walker (1977 29 0.59 21.1 q (1+0.824T 0.454) s w CHAIO=59.2[Lp/qs] 0.663 This study 29 0.57 21. 7 CHAll=5.62[LnlqS] 1.61 This study 26 0.58 22.5 CHA12=101(P}1 0.640 This study 40 0.63 20.0 CHA13=11.S(N), 60 This study 39 0.79 14.7 CHA14=97.8 [Nout/3+P out)'; shape/z/Qou ] 1.04 Hand (1975) 25 0.94 26 0.59 20.1 For CHAI CHA9, and CHA14, CHA. a [Original where a and b are determined by the regression. However, in1t e case of CHAl, CHA2, and CHA14, a and bare t e result of multi-plication by the original regress'on constants also.

PAGE 97

For predictive models based on loading terms, CHA6 gave the best results (r2 = 0.69). This equation is based on the equation of Jones and Bachman (1976) with Lsubstituted for Lp. A plot of predicted versus measured values of chrorophy11 a, using CHA6 (Figure V-9) shows the 95% CLI for this equation. Thus, for a lake having a predicted chlorophyll a concentration of 20 the 95% CLI is 5 to 74 Although this level of predictibi1ity is useful when considering the range of chlorophyll a values in the entire data set (3 to 208 further refinement of models to predict chlorophyll a concentrations from nutrient loading data is needed. NUTRIENT LOADING CRITERIA FOR FLORIDA LAKES One ,of the major objectives of the project this report summarizes was to develop nutrient loading criteria for Florida lakes. In this section, existing nutrient loading criteria (Vollenweider 1968; Shannon and Brezonik 1972; Vollenweider 1975; Dillon 1975) are analyzed for their ability to predict trophic status in the Florida NES lakes. The phosphorus loading criteria based on mass balance models are then modified, using the predictive equations developed earlier in the previous section, to improve their predictive ability for the Florida NES lakes. Since many of Florida's lakes are nitrogen-limited, nitrogen loading criteria have been developed using I/O models analogous to those used for the development of phosphorus loading criteria. The loading criteria are evaluated according to their ability to predict the trophic status of the Florida NES lakes. In addition, a "trophic ratio", defined as the ratio of a lake's nutrient loading to the minimum eutrophic loading for that lake, is used to evaluate the degree of eutrophication predicted by each model. In earlier developments of nutrient loading criteria, the' terms "ex cessive" and "permissible" have been used to describe the minimum loading levels that result in eutrophic and mesotrophic conditions, respectively, (Vollenwelder Dillon 1975; Larsen and Mercier 1975). These terms invoke a value judgement that many aquatic scientists now regard as unnecessary and unjustified. Thus, in this report we have used "minimum eutrophic loading" (MEL) to designate the minimum loading required to cause eutrophic conditions and "minimum mesotrophic loading" (MML) to designate the minimum loading required to cause mesotrophic conditions. These terms correspond respectively to the "excessive" and "permissible" loadings of earlier investigators. Phosphorus Loading Models. Vollenweider (1968) Loading Model. The Vollenweider (1968) loading model delineates trophic state as a function of mean lake depth. The minimum eutrophic loading and the minimum mesotrophic loadings are: = 0.05z0.6 0.025z0.6 -89 ,(5-34-8.) (5-34b)

PAGE 98

H co ;:::t 1000 100 -10 1 a a 1 [chI a] 8.30 [Lnl 65'Z + qs)] 1. 71 r2 ;" 0.69 95% CLI shown 10 100 Measured chlorophyll a, Figure V-9. Predicted vs. measured chlorophyll a using equation CHA6. 90

PAGE 99

According to the Vollenweider (1968) criteria (Figure V-lO) all of the Florida NES lakes are eutrophic. Thus, these criteria are conservative with respect to the classification of the mesotrophic lakes, since all of the mesotrophic lakes lie above the minimum eutrophic loading line (MEPp)' The failure of the Vollenweider (1968) model is not surprising since th1s model does not include any hydrologic variables. The Florida NES lakes are extremely diverse with respect to hydrologic conditions (0.03 yr < < 2.90 yr), and hydrologic conditions are an important factor affecting trophic state in these lakes. Shannon and Brezonik (1972) Model. Phosphorus loading criteria were aeve10pea By Shannon ana Brezon1k (1972) for F1or1aa lakes uS1ng a aata base of 55 lakes in the northern part of the state. Their criteria were based on volumetric loading rates (0.22 and 0.12 g P/m 3-yr, respectively, for excessive and permissible loading rates), and like the original Vollenweider criteria, they ignore hydrologic conditions. Although these criteria are more successful in predicting the trophic status of mesotrophic NES lakes than are Vollenweider's criteria, the extent of eutrophicati'on expressed is excessive for many lakes (Table V-6) particularly those with high flushing rates. For the lakes in which 'w < 0.10 years (Monroe, LC 29; Howell LC 32; Banana, LC 33; Trout, LC 36; Lawne, LC 37 and Munson, LC 38) the degree of eutrophication expressed by the Shannon and Brezonik criteria is greater than that expressed by the models that incorporate hydrologic variables. Vollenweider (1975) Model. The criteria proposed by Vollenweider (1975) are based on the equation: Lp (5-35a) Vollenweider found O'pz '" 10 m/yr for his study lakes. He also considered "permissible" (Le., minimum mesotrophic) and "excessive" (Le., minimum eutrophic) levels of total phosphorus to be 0.01 mg/L and 0.02 mg/L, respectively. Substituting these values into eq. 5-30a produces MEL P MML P = 0.20 + 0.02 qs (5-35b) 0.10 + 0.01 qs (5-35e) These criteria (Figure V-II) are an improvement over Vollenweider's 1968 criteria, although most of the meso trophic lakes still appear in the eutrophic zone. In order to improve this model for application to Florida lakes, two modifications were made. First, the concentration criteria for phosphorus were revised to account for the higher level of phosphorus associated with a given concentration of chlorophyll a in Florida lakes than found in most temperate lakes. For the phosphorus-limited lakes in the study set (n = 33), the relationship between phosphorus and chlorophyll a is given by equation 4-14: (chI a) = 0.195 [P]l -91

PAGE 100

H 32E 50 EUTROPHIC ZONE 33H >0-f!<} 10 E ..... a. C> z 0 9 en ::::> 0:: 0 ::CI a. en 0 ::c a. 0.5 -.J f2 -.J ::::> z z 0.1 0.05 0.010.1 35H 37E -29E _21E -34H -28E -30E E -27E .20H E -23E .15E 2M IIE1 9E._17E -22H---8E5M 10M -14E M OLIGOTROPHIC ZONE 5 MEAN DEPTH (z),m 50 Figure V-lO. Trophic state delineation of the Florida NES lakes by the Vollenweider (1968) phosphorus model. 92 100

PAGE 101

\0 C,N Table V-6. Comparison of phosphoru s models by trophic ratios and trophic state c1assificat LC EPA-NES TSI(AVG) Vol 1enweider Shannon and Vollenweider (1975) Assessment Brezonik Original Modified Orig (1972a) 1 O-M 46 (O-M) 98 (E) 0.45 (0) 0.76 (M) 0.51 (M) 1.31 2 M 52 (M) 7.93 (E) 1. 21 (E) 2.57 (E) 1. 75 (E) 1.28 3 M 52 (M) L52 (E) 0.95 (E) 1.48 (E) 0.97 (M) 2.15 4 M 54 (M-E) 1.81 (E) 0.34 (0) 0.49 (0) 0.34 (0) 1.69 5 M 49 (M) c.19 (E) 0.132 (M) 1.52 (E) 1.02 (E) 1.82 6 M-E 56 (E) (E) 0.76 (M) 1.05 (E) 0.71 (M) 2.67 7 E 56 (E) ---------8 E 60 (E) L89 (E) 1. 24 (E) 1.40 (E) 0.93 (M) 1. 78 9 E 58 (E) 4L9 (E) .7.39 (E) -----10 E 57 (E) ---------11 E 60 (E) .85 (E) 1.24 (E) 3.l3 (E) 2.0 (E) 3.16 12 E 61 (E) --------l3 E 62 (E) .02 (E) 0.45 (0) 1.25 (E) 0.88 (M) 1.29 14 E 62 (E) .56 (E) 0.52 (M) 0.63 (M) 0.43 (0) 2.12 15 E 60 (E) 1 .5 (E) 2.16 (E) 3.22 (E) 2.17 (E) 5.00 16 E 59 (E) -------17 E 62 (E) .97 (E) 1.17 (E) 3.28 (E) 2.28 (E) 5.02 18 E 62 (E) {.35 (E) 0.90 (M) 2.66 (E) 1. 84 (E) 1.62 19 E 61 (E) 1 .5 (E) 1. 32 (E) 3.88 (E) 2.73 (E) 9.25 20 H 73 (H) 1 .9 (E) 2.77 (E) 6.12 (E) 4.l3 (E) 5.58 21 E 55 (E) 6 .8 (E) 9.76 (E) 6.82 (E) 4.28 (E) 4.55 22 H 76 (H) .16 (E) 1. 68 (E) 2.94 (E) 2.08 (E) 21.2 23 E 68 (H) 1 .9 (E) 2.21 (E) 3.80 (E) 2.51 (E) 4.79 24 E 60 (E) ------25 H 63 (E) 20 (E) 32.7 (E) 27.0 (E) 17.11 (E) 36.8 26 E 71 (H) 1 .7 (E) 3.01 (E) 5.06 (E) 3.35 (E) 9.87 27 E 64 (E) 3 .8 (E) 5.09 (E) 5.81 (E) 3.72 (E) 8.07 28 E 64 (E) 5 .8 (E) 9.26 (E) 16.2 (E) 10.91 (E) 8.15 29 E 63 (E) 20 (E) 37.3 (E) 9.77 (E) 6.04 (E) 12.8 30 E 75 (E) 6 .8 (E) 15.6 (E) 12.5 (E) 8.48 (E) -31 E 67 (E) ------32 E 64 (E) 67 (E) 106 (E) 72.4 (E) 45.58 (E) 66.4 33 H 77 (H) 98 (E) 224 (E) 95.4 (E) 61.39 (E) 86.0 34 H 73 (H) 8 (E) 17.5 (E) 14.2 (E) 9.60 (E) 22.5 35 H 70 (E-H) 24 (E) 47.4 (E) -----36 H 69 (E-H) --------37 E 81 (H) 28 (E) 62.1 (E) 35.5 (E) 23.18 (E) 20.5 38 H 78 (H) 119 (E) 231 (E) 63.8 (E) 39.64 (E) -39 H 80 (H) ------40 E 79 (H) r .... -.----I ons. ilIon (1975) na1 Modified (E) 0.61 (M) (E) 0.59 (M) (E) 0.98 (M) (E) 0.80 (M) (E) 0.85 (M)-(E) 1.25 (E) --(E) 0.82 (M) ---(E) 1.53 (E) --(E) 0.61 (M) (E) 0.99 (E) (E) 2.32 (E) --(E) 2.35 (E) (E) 0.75 (M) (E) 4.33 (E) (E) 2.61 (E) (E) 2.18 (E) (E) 9.73 (E) (E) 2.23 (E) --(E) 16.68 (E) (E) 4.62 (E) (E) 3.72 (E) (E) 3.80 (E) (E) 5.42 (E) ---(E) 31. 72 (E) (E) 41.60 (E) (E) 10.66 (E) ----(E) 9.37 (E) ----

PAGE 102

>. I E ...... 0.. .. <.!) 50 10 9 CJ) 5 :::> 0:: o :r: 0.. CJ) o :r: 0.. -.oJ I--.oJ 0.5 z z <{ H .33H EUTROPHIC ZONE H -29E E .27E -20H -26E -23E -15E -19E -17E _liE -18E -2M -22H -13E OLIGOTROPHIC ZONE __ l____ 1____________ JL __ 0.1 OJ 0.5 I 5 10 50 HYDRAULIC LOADING (qs)' m/y Figure V-II. Trophic state delineation of the Florida NES lakes by the Vollenweider (i975) phosphorus model. 94 100

PAGE 103

The lower limit of chlorophyll a concentrations for meso trophic conditions is commonly considered to be 5 and 10 is usually considered the lower limit for eutrophic conditions. The corresponding phosphorus concentrations using equation 4-14 are 23 and 46 Thus, in revising the phosphorus loading criteria, the minimum meso trophic and eutrophic phosphorus concentrations were considered to be 25 and 50 respectively. The model was further modified by using the statistically fitted form of equation 5-35a (TP5) determined in the previous section: [P]] = 0.952 [L /(8.53 + q )] 0.860, p s (5-35d) where 8.53 is the mean value of the apparent settling velocity for phosphorus (vp ) as determined earlier (p.77) using eq. 5-28. When values of 0.050 and 0.025 mg/L phosphorus are used with this equation, the following loading criteria result: MEL = 0.033q + 0.28 p s (5-35e) MML = 0.015q + 0.12 p s (5-35f) The revised criteria, shown in Figure V-12, are an improvement over the original criteria, particularly with respect to the classification of mesotrophic lakes. While the original criteria place three meso trophic lakes in the eutrophic zone, one of these lakes, plus one classified as mesotrophiceutrophic are placed in the mesotrophic zone, although two moderately eutrophic lakes (Okeechobee, LC 13 and Kissimmee, LC 8) are now placed in the meso trophic zone. Dillon (1975) Model. Dillon (1975) proposed loading criteria for phosphorus based on the predictive equation for [m 1 of Dillon and Rigler (1974a) based on phosphorus loading and hydrologic data (eq. 5-14). setting IP] 1 in eq. 5-14 to 0.010 and 0.020 mg P/L to present minimum mesotrophic and eutrophic levels of phosphorus and substituting qs = Z/TW resulted in the criteria: (MEL )'(l-R)T = 0.020z p p w (MML )'(l-R)T = 0.010z p p w (5-3h) (5-36b) Like the phosphorus loading criteria developed by Vollenweider (1968 and 1975), these criteria appear conservative when applied to the Florida NES lakes (Figure V-13). To improve these criteria for Florida lakes, the minimum meso trophic and eutrophic levels of phosphorus were adjusted upward to 0.025 and 0.050 mg/L, as was done with the Vollenweider (1975) model. Furthermore, equation TP2, a modification of the original Dillon and Rigler (1974a) predictive equation, was used to express the relationship between phosphorus loading and phosphorus concentration in the revised criteria. Thus, the minimum eutrophic loading of phosphorus becomes: 95

PAGE 104

50 I C'I Z C C 10 ...J en ::::> 0:: 5 o J: a.. en o J: a.. ...J o .... ...J ::::> Z 0.5 Z MEL MMLp EUTROPHIC ZONE .4M -33H .37E .20H .26E -23E E -2M -25H .21E .Z7E OLIGOTROPHIC ZONE 0.1 0.'5 I 5 10 50 100 HYDRAULIC LOADING,(qs)' m/yr Figure V-12. Trophic state delineation of the Florida NES lakes by the modified Vollenweider (1975) model using loading criteria developed in this study. 96

PAGE 105

C\f E c:: 0'1 -a. 0:: I J-EUTROPHIC ZONE 5 0.5 0.1 E -25H -22H .26E -29E -27E -28E i9E .20H -17E E _15E-2IE -liE M 6ME -18E -5M 8E 2M,.13 0.05 ___ __ OLIGOTROPHIC ZONE 0.0 1'=0":"".1 MEAN DEPTH (z ) ,m Figure V-13. Trophic state delineation of the Florida NES lakes by the Dillon (1975) phosphorus model. -97

PAGE 106

TP2 = 0.748[(MEL )'(l-R )/q ]0.862 p p s 0.050 = 0.748 [(MEL)' (l-R)/ ].0.862 p p qs' (MEL )'(l-R)T = 0.043z p p w and the minimum mesotrophic loading is: 0.02S = 0.748[(MML )'(l-R )/q 1 0.862 p P ff (MML )'(l-R)T = 0.019z p p w (S-36c) (S-36d) (S-36e) (5-36f) (S-36g) These criteria, shown in Figure V-14, appear more reasonable than the original criteria, particularly with respect to the classification of mesotrophic lakes. Thus, while the original criteria placed all of the mesotrophic lakes in the eutrophic zone, the revised criteria place all of the meso trophic lakes in the mesotrophic zone, with only three eutrophic lakes classified incorrectly (all three in the mesotrophic zone). Although the modified loading criteria improve the predictive ability of the Dillon model, the resulting plot can be further modified in order to spread the points out along the abscissa and improve the clarity of the plot. To accomplish this, each side of equations S-31e and g was multiplied by T : w L (1 p L (1 p Rp) = 0.043 qs R) 0.019 q p s (S-36h) (S-36i) The resulting plot, shown in Figure V-1S, places each lake in the same zone as does Figure V-14 but has the graphical advantages noted above. Nitrogen Loading Models Vollenweider (1968) Model. Vollenweider (1968) developed loading criteria based on his phosphorus loading criteria and an ffl-ratio of 15. The minimum eutrophic and mesotrophic areal nitrogen loading rates were: = 0.7S0:l,6 0.375z0 6 (S-37a) As with the Vollenweider (1968) phosphorus criteria, the nitrogen criteria classify all of the Florida NES lakes as eutrophic (Figure V-16); these criteria are therefore unacceptable for Florida lakes. Shannon and Brezonik (1972) Model. Based on their study of Florida lakes, Shannon and Brezonik (1972) proposed criteria of 1.Sl g/ma_yr and 0.86 g m 2/yr as minimum eutrophic and minimum meso trophic volumetric nitrogen loading rates. These criteria were developed using lakes having relatively long hydraulic residence times and are not successful in classifying the Florida NES lakes since all of these lakes except Lake Okeechobee (LC 13) 98

PAGE 107

C\I E a:: 01 .. IQ. a:: I Q. ..J ________ __ ________ __ ________ __ --, 5 EUTROPHIC ZONE .25H. .34H 0.5 e37E 0.1 0.05 OLIGOTROPHIC ZONE 0.01 MEAN DEPTH ('i), m Figure V-14. Trophic state delineation of the Florida NES lakes by the Dillon (1975) phosphorus model using the loading criteria developed in this study. 99

PAGE 108

>I C\J E 50 10 5 a: -a.. 0:: I -0..05 ..J 0.1 0.05 EUTROPHIC ZONE -32E -33H -37E -27E -21E -34H -26E -28E -23E -22H -19E -17E OLIGOTROPHIC ZONE 0.1 0.5 I 5 10 50 100 HYDRAULIC LOADING RATE (qs) ,m/yr Figure V-IS. Trophic state delineation of the Florida NES lakes of the Dillon (1975) phosphorus model using'loading criteria developed in this study. 100

PAGE 109

EUTROPHIC ZONE 500 100 f" C\I E "-z50 0'1 .. (!) z is
PAGE 110

are classified as eutrophic. Furthermore, the degree of eutrophication expressed by these criteria is excessive for lakes with short hydraulic residence times (Table V-7). This fact again illustrates the need to include hydrologic parameters in models to predict lake trophic status. Vollenweider (19?5)-Type Nitrogen Model. A nitrogen-based analog of Vollenweider's (1975) phosphorus loading model was described earlier (equation TN5, p. 80) that predicts [N]l as a function of nitrogen loading and qs' This model was used to develop nitrogen loading criteria for Florida's lakes. The first step in developing these criteria was to determine minimum concentrations of [N]i for mesotrophic and eutrophic conditions. Since most of the Flor1da NES la es were nitrogen limited, equat10n eRA 13 was used to establish the relationship between nitrogen and chlorophyll a concentrations: (chI a) = 11.5(N)11.60 (5-38) For chlorophyll a concentrations of 5 and 10 the corresponding nitrogen concentrations are 0.59 and 0.92 mg/L. Thus, reasonable criteria for nitrogen concentrations are 1.0 mg/L-N for minimum eutrophic conditions and 0.5 mg/L-N for minimum meso trophic conditions. Equation TN5 was then used to compute the minimum eutrophic nitrogen loading: (N)l 1.0 MELN The minimum mesotrophic 0.50 = 1. 69 [L /(5.49 + qJ 0.341 n = 1. 69 [L /(5.49 + qg1 0.341 n = 1.18 + 0.21 q s loading was determined in the 1.69 [L /(5.49 + q )] 0.341 n s = 0.154 + 0.03 q s same way: (5-39a) (5-39b) (5-39c) (5-39d) (5-3ge) The value 5.49 represents the mean for the NES lakes of the apparent settling velocity (vn ) for nitrogen (in m/yr) as determined from an equation analogous to the equation used for v (eq. 5-28). The criteria given as eqs. 5-39-c and e are plotted as dashea lines in Figure V-17, and they result in place ment of all the NES lakes in the eutrophic zone. Thus on this basis they are not satisfactory. It should be noted that the r2 value for the equation on which these criteria were based (TN5) was very low (0.18), and its predictive capability is limited. Moreover, the low fractional exponent in TN5 (and hence in eqs. 5-39 b and d) renders these equation unsatisfactory as semi-theoretical (mass balance) models of the behavior of nitrogen in aquatic systems.' Thus the regression (TN5) yielded poor predictions at the expense of increased empiricism and decreased realism (Le. the fractional exponent). Consequently, we developed a set of loading criteria for nitrogen based on the unaltered equation (i.e. without the statistically determined coefficients). This approach at least has the merit of maintaining the model's theoretical appeal and realism. The derived equations for and MELN criteria are: 102

PAGE 111

Table V-7. Comparison of nitrogen models by trophic ratios and trophic state classifications. EPA-NES Vollenweider Shannon & Vollenweider Dillon LC AssessTSI (AVG) (1968) Brezonik (1975) type (1975) ment (1972) type 1 O-M 46 (M) 5.07 (E) 1. 69 .(E) 0.52 (M) 1. 93 (E) 2 M 52 (M) 5.44 (E) 1. 82 (E) 0.78 (M) 1. 05 (E) 3 M 52 (M) 7.10 (.EY 2.27 (EI 0.61 (M) -(M) 4 M 54 (M-E) 3.82 (E) 0.70 (M) 0.49 (0) 1. 32 (E) 5 M 49 (M) 1. 30 (E) 2.86 (E) 1. 04 (E) 0.70 (M) 6 M-E 56 (E) 4.97 (E) 1. 95 (E) 0.54 (M) 1. 07 (E) 7 E 56 (E) 8 E 60 (E) l3 .1 (E) 5.15 (E) 1. 08 (E) 1.26 (E) 9 E 58 eE) 8.27 (E) 3.11 (E) 10 E 57 (E) 11 E 60 (E) 15.7 (E) 4.l3 (E) 2.46 (E) 1.47 (E) 12 E 61 (E) l3 E 62 (E) 2.66 (E) 0.88 (M) 0.55 (M) 0.78 (M) 14 E 62 (E) 5.70 (E) 2.55 (E) 0.67 (M) 1. 91 (E) 15 E 60 (E) 8.27 (E) 3.11 0.92 (M) 1.47 (E) 16 E 59 (E) 17 E 62 (E) 4.35 (E) 1.39 (E) 0.87 (M) 2.74 (E) 18 E 62 (E) 49.1 (E) 15.2 (E) 9.83 (E) 1.33 (E) 19 E 61 (E) 3.69 (E) 1. 39 (E) 0.63 (M) 1. 59 (E) 20 H 73 (H) l3.0 (E) 4.17 (E) 1. 81 (E) 2.53 (E) 21 E 55 (E) 36.3 (E) 11. 9 (E) 1.28 (E) 1. 05 (E) 22 H 76 (H) 3.20 (E) 1. 29 (E) 0.53 (M) 3.68 (E) 23 E 68 (H) 17.9 (E) 5.82 (E) 1. 81 (E) 2.71 (E) 24 E 60 (E) 25 H 63 (E) 33.1 (E) 8.72 (E) 1.14 (E) 1. 35 (E) 26 E 71 (H) 8.67 (E) 2.89 (E) 0.89 (M) 2.64 (E) 27 E 64 (E) 47.0 (E) 15.0 (E) 2.79 (E) 1. 76 (E) 28 E 64 (E) 16.9 (E) 5.91 (E) 2.02 (E) 2.94 (E) 29 E 63 (E) 151 (E) 59.3 (E) 2.26 (E) 2.22 (E) 30 E 75 (H) 12.4 (E) 2.05 (E) 1. 00 (E) ---3r K---tiT---CE) ------(-E-)------------------------------------------32 E 64 (E) 115 (E) 30.4 (E) 3.20 (E) 1. 71 (E) 33 H 77 (H) l32 (E) 65.7 (E) 4.64 (E) 3.74 (E) 34 H 73 (H) 16.7 (E) 8.01 (E) 1.28 (E) 2.51 (E) 35 H 70 (E-H) 63.6 (E) 26.8 (E) 36 H 69 (E-H) -' 37 E 81 (H) 70.4 (E) 33.7 (E) 3.36 (E) 2.08 (E) 38 H 78 (H) 217 (E) 91.6 (E) 3.75 (E) 39 H 80 (H) 40 E 79 (H) 103

PAGE 112

500 >I "'e ...... z QlIOO .. (!) z 50 o Z lLJ (!) o a:: IZ ....J 10 ....J 5 ::: z z or.t MELn MMLn EUTROPHIC ZONE -18E -liE _.'-./ .32E -33H -27E -23E / OLIGOTROPHIC .ZONE // ,// 0.1 0.5 I 5 1 HYDRAULIC LOADING (qs) m/yr Figure V-17. Trophic state delineation of the Florida NES lakes by a modified Vollenweider (1975) model for nitrogen using loading criteria developed in this study. Dotted lines represent MELn and according to equations 5-39c and e (See text). 104

PAGE 113

(N)l = L n/(S.49 + qs) 1.0 =MEL N/(S.49 + qs) (S-40b) MELN = S.49 + qs (S-40c) O.S = MMLN/(S.49 + qs) (S-3Sd) 2.7S + O.S q s (S-40e) These criteria, also plotted in Figure V-17, place nearly all of the meso-troph1c lakes 1n the mesotrophic zone. However, they also result in place ment of many eutrophic lakes in the mesotrophic zone. Particularly affected are lakes with low areal water loading rates (i.e., q < S m/yr). Since most of these eutrophic lakes are indeed (except Apopka; LC 22), the incorrect classification of these lakes seems to reflect a general weakness in the model. Specifically, the use of a uniform value for the apparent settling velocity (vn ) as a basis for estimating nutrient losses appears to have limited validity for nitrogen, as reflected by the large standard deviation associated with vn for the NES lakes (S.S + 11.2 m/yr). By contrast, the estimate of vp (although still large) was more precise (8.S + 8.8), and the fitted equation to predict (P)l using vp had an r2 value of 0.83. The weakness of this approach for nitrogen may reflect the more complex biogeochemical cycle for this element than for phosphorus. DiZZon (1975)-Type Nitrogen ModeZ. The relationship between nitrogen loading rate and nitrogen concentration can be expressed in a form analogous to that used by Dillon (197S) to express the relationship between phosphorus loading rate and phosphorus concentration: This equation was modified statistically in an earlier section (p. 80) to fit the Florida NES data set. The resulting equation (TN2) can be used as a basis for constructing a plot to show the relationship between nitrogen loading rate and trophic status: (N) = 0.899 [L (1 -R)/ 1 0.976 1 n n (S-41b) To construct the loading plot, values of 1.0 and O.S mg/L were used as minimum eutrophic and mesotrophic concentrations of total nitrogen, and the term z/'w was substituted for qs' The minimum eutrophic and meso trophic loadings can be expressed as follows: (MELN)'(l-R), = 1.12z n w (S-:41c) (S-.41d) The resulting plot (Figure V-18) shows the positions of the Florida NES lakes with respect to the lines of minimum eutrophic and minimum mesotrophic nitrogen concentration. This model works well, since nearly all of 105

PAGE 114

50 (\IE ....... Z CJI "10 .-! c:: 0:: I ..J 5 33H I I' H EUTROPHIC ZONE 2"'2E .17E v 20H E 27E 5 10 MEAN DEPTH (z), m OLIGOTROPHIC ZONE 50 100 Figure V-IS. Trophic state delination of the Florida NES lakes by a modified Dillon (1975) model for nitrogen using loading criteria developed in this study. 106

PAGE 115

the mesotrophic lakes, but only two eutrophic lakes (both moderately eutrophic) are located in the mesotrophic zone. This model is an improvement over the Vollenweider (1975)-type nitrogen model (Figure V-17) in its placement of eutrophic lakes with low areal hydraulic loading rates (qs < 5 m/yr). Several lakes this type that were located in the mesotrophic zone using the Vollenweider-type model including Marion (LC 14), Doctors (LC 17), Gibson (LC 19), and Apopka (LC 22) are located in the eutrophic zone using the Dillon-type model. The clarity of the Dillon-type graph for nitrogen loading can be improved by dividing both sides of equations 5-36c and d by T : w (MELN)o(l-R n ) = = 1.12 q s 0.55 q s (5-4le) (5-41) The resulting graph (Figure V-19) spreads the lakes out along the abscissa but does not change the relative position of the lakes with respect to trophic state zone (cf. Figures V-IS and V-19). APPLICATION Predictive equations were developed in a preceding section for total nitrogen, total phosphorus and chlorophyll a that enable reasonably reliable estimates of these parameters from nutrient loading and hydrologic data. The best equations to predict total nitrogen (TN2) total phosphorus (TN2) are both based on Dillon and Rigler's (1974a) model, and they require data on both the inputs and the outputs of these nutrients to the lake to compute the retention coefficients. The best equation to predict chlorophyll a is CHA6, which is a nitrogen-based analog of the Jones and Bachman (1976) model. Equations to predict Rand R were not very successful, and even the best equations proauce accurate preaictions of Rp ana R. The failure of these equations to predict Rp and Rn may reflect errors iN the nutrient or water budgets or may indicate an inherently poor correlation between hydrologic and morphologic parameters and nutrient retention in Florida lakes. The best equations for (P)l' (N)l' (chI a), Rn and Rp were plotted in Figures V-5 to V-9 together with the 95% confidence limits for individual predictions. New predictions made using equations TP2, TN2, CHA6, RP6 and RN4 should be accompanied by the confidence intervals shown in these figures to express the degree of confidence associated with those predictions. The predictive equations described above have been developed for a group of lakes that are primarily nitrogen-limited, and their application to phosphorus-limited lakes should be considered with caution. Unfortunately there are few data on nutrient loadings for phosphorus-limited lakes in Florida, and predictive equations that are directly applicable to these lakes have not been developed. 107

PAGE 116

500 .29E 100 EUTROPHIC ZONE 50 >-N' E ..... Z 10 en .. -c 5 I -.22H c ...J 0.5 OLIGOTROPHIC ZONE 0.1 0.1 0.5 5 10 50 100 HYDRAULIC LOADING (qs), m/yr Figure V-19. Trophic state delineation of the Florida NES lakes using a further transformation of Dillon-type nitrogen modeL 108

PAGE 117

The nutrient loading criteria developed for Florida lakes can be used to evaluate the effects of management strategies on lake trophic status. Either the modified Vollenweider (1975) criteria (Figure V-12) or the modified Dillon (1975) criteria (Figures V-13 and V-14) can be used to evaluate phosphorus loadings, but the Vollenweider (1975) model does not require data on Rp and thus can be used when outflow phosphorus loadings are not known. For nitrogen loading, the Dillon-type loading criteria (Figures V-IS and V-19) are more valid than the Vollenweider-type criteria and should be used. In using these loading plots, the user must decide whether to use nitrogen or phosphorus loading criteria. This should be done on the basis of the observed nutrient linutation of the lake; phosphorus loading criteria should be used for phosphorus-limited lakes and nitrogen loading criteria should be used for nitrogen-limited lakes. For lakes with mixed nutrient limitation, both nitrogen and phosphorus loading criteria should be used, since the addition of either nutrient could result in enhanced productivity. The need to use appropriate loading criteria can be demonstrated by applying both nitrogen and phosphorus loading criteria to lakes with large nutrient imbalances. as shown in Table V-So For the one lake with a high SIN:SRP ratio (52.0, LC 18), the trophic ratios based on: nitrogen criteria are higher than the trophic ratios based on phosphorus criteria using both the modified Vollenweider (1975) model and the modified Dillon (1975) model. Conversely, for the seven lakes with very low SIN:SRP ratios, the trophic ratios based on phosphorus criteria are always higher than the trophic ratios based on nitrogen. When large increases or reductions in nutrient loadings occur, the limiting nutrient may change, particularly when the altered inflow has a large nutrient imbalance. For example, treated municipal sewage is usually nitrogen-limited (See and the addition, of treatecimunicipal sewage to an initially phosphorus-limited lake may tend to shift the nutrient limitation towards nitrogen. In this case, phosphorus loading criteria are applicable before the addition of sewage, and nitrogen loading criteria may be applicable after the addition of sewage (see Figure V-20). In such a case, the use of phosphorus loading criteria to evaluate the effect of sewage addition would result in a prediction of greatly enhanced productivity (Figure V-20c). However, if the lake had shifted to nitrogen-limitation following this perturbation, its actual trophic status would be predicted more accurately by nitrogen loading criteria (Figure V-20d). Thus, when evaluating the effects of altered loading, both nitrogen and phosphorus loading criteria should be used, and the correct prediction should be based on the criteria that produces the lower trophic ratio. 109

PAGE 118

..... ..... o ,. Table V-8. Trophi c ratios of lakes with large nutrient irr balances Trophic Ratios Average An nua1 Modified Vollenweider Modi fied Dillon LC Inorg-N/ SBP (1975) type models type models .. p N P N 18 52.0 1.84 9.83 0.75 1.33 25 0.2 17.11 1.14 16.68 1.35 28 0.9 10.91 2.02 3.80 2.94 30 2.2 8.48 1.00 --32 0.2 45.58 3.20 31.72 1. 71 33 1.8 61.39 4.64 41.60 3.74 34 1.0 9.60 1.28 10.6 6 2.51 38 1.3 39.64 3.75 --(1975

PAGE 119

50 CORRECT 500 INCORRECT 10 EUTROPHIC EUTROPHIC I N 5 N E E 50 '" '" a.. a.. en Ol .. 10 c 0: 0.5 0: 5 -0. C ...J 0.1 ...J 005 0.5 0.5 I 5 10" 50 05 I 5 10 50 qs t m/yr qs t m /yr A. Prior to addition of sewage (Lp = 0.2 g P/m2-yr.; Ln = 5.0 g N/m2-yr). Correct trophic state is predicted in phosphorus loading plot. 50 INCORRECT 5 CORRECT !-.'-10 EUTROPHIC EUTROPHIC 5 I 50 1: '" a.. Ol 10 C' .. -5 .. c a:: 0. 0: I -0.1 c -...J 0. ...J a05 0.5 0.0010.1 0.5 I. 5 10 50 05 I 5 10 50 qs t m/yr qs t m / yr B. Following addition of 5 x 104 kg P and 15 x 104 kg N from sewage (N:P =3:1), lake is still mesotrophic due to shift towards N limitation. Figure V-20. Application of loading criteria developed in this study to a hypothetical lake before and after addition of municipal sewage. Tw = 1 yr, = 5m, qs = 5 m/yr, Rp = 0.5, Ru = 0.4. 111

PAGE 120

CHAPlER VI: SUfv1\v\RY AND CONCLUS IONS Nonpoint source loadings of nitrogen and phosphorus from Florida watersheds can be estimated from land use characteristics using export coeffiencts obtained from the literature or using a multiple regression approach. The literature-based approach produced a wide range of export coefficients for each land use: 0.2-0.7 kg P/ha-yr and 1.5-6.1 kg N/ha-yr for forests; 0.4-2.4 kg P/ha yr and 2 SOkg N/ha y1 for cropland, 0.2-4.7 kg P/ha yr and 1.5-7.4 kg N/ha-yr for residential areas; 0.3-7.5 kg P/ha-yr and 3-10 kg N/ha-yr for urban areas. NPS nutrient loadings (dependent variables) and land use characteristics (indpendent variables) for 41 NES watersheds were analyzed using multiple regression analysis (stepwise deletion procedure) to improve the predictive capability of the land use-nutrient loading approach. For NPS phosphorus loading, a model that includes three land use terms (cropland, rangeland and forest) has better predictive ability than does a model that includes total drainage area as the sole independent variable (r2 = 0.71 vs 0.21). Both NPS nitrogen loadings and flow were highly correlated with total drainage area (r2 = 0.84 and 0.91, respectively) and the inclusion of specific land use terms as independent variables results in a very modest improvement in predictive ability. Although the use of literature-based export coefficients may be appropriate for small watersheds with a single dominant land use, the regression approach produces reasonably precise estimates of NPS loadings for larger watersheds and has the advantage that the reliability of predictions can be statistically evaluated using confidence bands. An evaluation of the limnological characteristics of 101 Florida lakes indicates that Florida lakes are different in several important respects from temperate lakes. Most of the study lakes are shallow and well-mixed; few exhibit stable seasonal stratification or have anoxic hypolimnia. Furthermore, there is no evidence of seasonal variation in chlorophyll a standing crops or nutrient concentrations. Unlike temperate lakes which are usually phosphorus limited, many of the study lakes are nitrogelFttmited (46% had SRP: SIN ratios < 10:1). Finally, for a given concentration of phosphorus, Florida's lakes have less chlorophyll a than do temperate lakes; this is true even for phosphoruslimited Florida lakes. Carlson's trophic status index (TSI) was modified for Florida lakes by inclusion of a nitrogen index to reflect the importance of nitrogen as a limiting nutrient. A composite TSI was developed by averaging the TSI's based on Secchi disk transparency, chlorophyll a and nutrient concentration (the smaller of the nitrogen or phosphorus index). Various nutrient loading models were evaluated statistically for their ability to predict chlorophyll a, nitrogen and phosphorus in the Florida NES lakes. Models based on a Dillon and Rigler-type loading term gave the best predictions of total nitrogen and total phosphorus (r2 = 0.77 and 0.91, tively). The best predictions of chlorophyll a were obtained using a Jones and Bachman-type loading term with areal nitrogen loading substituted for areal phosphorus loading (r2 = 0.69). The confidence band associated with these -112

PAGE 121

models indicate that they produce reasonably reliable predictions of phosphorus, nitrogen and chlorophyll a. None of the models evaluated to predict nitrogen and phosphorus retention coefficients produce reliable estimates of these parameters. Existing phosphorus loading criteria (Vollenweider 1968, 1975; Dillon, 1975; Shannon and Brezonik, 1972) tend to overestimate the trophic status of the Florida NES lakes. The 1975 Vollenweider criteria and the 1975 Dillon criteria were modified to improve their predictive capabilities; in both cases the revisions result in higher critical loading rates than the original criteria. The two sets of revised criteria are equally successful in delineating eutrophic lakes from mesotroplric lakes. Nitrogen loading criteria were developed using concepts analogous to those used in the development of phosphorus loading criteria. The most successful nitrogen criteria were based on a Dillon-type model; these criteria allow adequate discrimination between meso trophic and eutrophic lakes. In evaluating the impact of a proposed management strategy (alteration of nutrient inputs) on lake trophic status both nitrogen and phosphorus loading criteria should be used. The correct response of the lake will be obtained using the criteria that predicts the lower trophic status. 113

PAGE 122

REFERENCES Asmussen, L. E., J. M. Sheriden and C. V. Boorman, Jr. 1979: Nutrient move ment from agricultural watersheds in the Georgia coastal plain. Trans. ASAE, p. 809-821. Bachmann, R. W. and J. R. Jones. 1974. in lakes. Iowa State J. Res. 49: Phosphorus inputs and algal blooms 155-160. Baker, L. A., V. D. Adams, J. S. Fifield, L. G. Terry, and D. Sorenson. 1981. Predlcted l1mnology of the Ridges Basin reserVOlr. Utah water Research Lab. Pub., Logan UT (in press). Bedient, P. B. D. A. Harned and W. G. Characl.dis. 1978. Storm-water analysis and prediction in Houston. J. Env. Eng. ASCE 104: 1087-1100. Bradford, W. L. and D. J. Maiero. 1978. Lake process models applied to reser-voir management. J. Env. Div. ASCE 104, No. EE5: 981-986. Brezonik, P. L. 1978. Effect of organic color and turbidity on Secchi disk transparency. J. Fish. Res. Bd. Canada. 35: 1410-1416. Brezonik, P. L., C. D. Hendry, Jr., E. S. Edgerton, R. L. Schulze, and T. L. Crisman. 1981. Acidity, nutrients, and minerals in atmospheric precipitation over Florida: deposition patterns, mechanisms, and ecological effects. Completion rept. on grant no. 805560, u.S. EPA, Corvallis, Oregon (in press). Brezonik, P. L., W. H. Morgan, E. E. Shannon,and H. D. Putnam. 1969. Eutrophication factors in north central Florida lakes. Publ. No.5, Univ. Florida Water Resources Center, Gainesville, Fl., rOl p. Brezonik, P. L., C. D. Pollman, T. L. Crisman, J. N. Allison,and J. L. Fox. 1978. Limnological studies on Lake Apopka and the Oklawaha chain of lakes. 1: water qual1ty ln 1977. Rpt. No. ENV 07 78 01, Dept. of Env. Eng. Sci., Univ. of Florida, Gainesville, 282 p. Brezonik, P. L. and E. E. Shannon. 1971. Trophic state of lakes in north central Florida. Publ. No. 13, Univ. of Florida Water Resources Res. Cent., Gainesville, Fl., 101 p. Burton, T. M., R. Turner, and R. C. Harriss. 1977. Nutrient export from three north Florida watersheds in contrasting land use, pp. 323-342. In: P. L. (ed.). Watershed Research in Eastern North America, Vol. 1. Tidemark Printing Co., Edgewater, Md. Calvert, D. V. 1975. Nitrate, phosphate, and potassium movement into drainage lines under three soil management systems. J. Environ. Qual. 4: 183-186. Campbell, K. L. 1978. Pollution in runoff.from,..rIonpoint sources. Publ. 42, Florida Water Resources Res. Cent., Univ. Florida, Gainesville, 49 p. -114

PAGE 123

Carlson, R. E. 1977. A trophic state index for lakes. Limnol. Oceanogr. 22: 361-369. Chapra, S. C. 1975. Comment on 'An empirical method of estimating the retention of phosphorus in lakes' byW. B. Kirchner and P. J. Dillon. Water Resources Res. 11: 1033-1036. Chapra, S. C. and S. T. Tarapchak. 1976. A chlorophyll model and its relationship to phosphorus loading plots for lakes. Water Resources Res. 12: 1260-1264. ChiaudanL._G. and M. Vighi. 1974. The N:P ratio and tests with Selenastrum to predict eutrophication in lakes. Water Research 8: 1063-1069. C&2M Hill; 1978. Water quality studies in the Everglades Agriculture Area of Florida. Report submitted to the Florida Sugar Cane League, Clewiston, Fl. Dillon, P. J. and F. H. Rigler. 1974a. A test of a simple nutrient budget model predicting the phosphorus content of lake waters. J. Fish. Res. Bd. Canada 31: 1771-1778. Dillon, P. J. and F. H. Rigler. 1974b. ship in lakes. Limnol. Oceanogr. The phosphorus-chlorophyll relation-19: 767-773. Dillon, P. J. 1975. The P04 budget of Cameron Lake, Ontario: tance of flushing rate to the degree of eutrophy of lakes. Oceanogr. 20: 28-39. The imp orLimnol. Duffy, P. D., J. D. Schreiber, D. C. McClurkin and L. L. McDowell. 1978. Aqueous and phosphorus yields from five southern pine watersheds. J. of Environ. Qual. 7: 45-50. Gakstatter, J. H., M. o. Allum, S. E. Dominiquez, and M. R. Crouse. 1978. A survey of phosphorus and nitrogen levels in treated municipal waste-water. J.W.P.C.F. 50: 718-722. Goldman, C. R. 1972. The role of minor nutrients in limiting the productivity of aquatic ecosystems, p. 21-33. In: Likens, G. E. (ed.), Nutrients and eutrophication: the limiting nutrient controversy, American Society of Limnology and Oceanography special symposium, Allen Press, Lawrence, Kansas. Grubbs, F. E. 1969. Procedures for detecting outlying observations in samples. Technometrics 11: 1-21. Haith, D. A. 1980. Models for the analysis of agricultural nonpoint source pollution. Collaborative Paper CP-80-27, Internat. Inst. Appl. Systems Analysis, Laxenburg, Austria, 39 p. Hand, J. G. 1975. Water quality modeling of the Middle St. Johns River Sys tem. M.S. thesis, Univ. of Florida, Gainesville, 125 p. -115

PAGE 124

Heaney, J. P., W. C. S. J. Nix. 1976. level I: preliminary screening procedure. U.S. EPA, Washington, D.C., 77 p. Storm water management model Tech. Rept. EPA-600/2-76-275. Hendry, C. D., P. L. Brezonik, and E. S. Edgerton, 1981. Atmospheric deposition of nitrogen and phosphorus in Florida. In Atmospheric pollutants in natural waters, S. J. Eisenreich (ed.) Science Publ., Inc., Ann Arbor, Mich. (in press). Huber, W. C. and P. L. Brezonik. 1980. Assessment of projected water quality in man-made lakes and waterways of unit 24 and unit 30 developments near Marco Island, Rept. to Deltona Corp., Fl. 91 p. Hutchinson, G. E. 1957. Treatise on limnology, vol 1. J. Wiley, New York, 1013 p. Imboden, D. M. ogre 19: 1974. Phosphorus model of lake eutrophication. Limnol. Ocean297-304. Jones, J. R. and R. Bachman. 1976. Prediction of phosphorus and chlorophyll levels in lakes. J.W.P.C.F. 48: 2176-2182. Kautz, R. S. 1981. Effects of eutrophication on fish communities of Florida lakes. Unpublished MS, Florida Game and Fresh Water Fish Commission, Tallahassee, 24 p. King, D. L. 1972. Carbon limitation in sewage lagoons, p. 98-105. In: Likens, G. E. (ed.), Nutrients and eutrophication: the limiting nutrient controversy, American Society of Limnology and Oceanography Special Symposium, Allen Press, Lawrence, Kansas. Kirchner, W. B. and P. J. Dillon. 1975. the retention of phosphate in lakes. An empirical method of estimating Water Resources Res. 11: 182-84. Kratzer, C. R. 1979. Application of input-output models to Florida lakes. MS Dept. Env. Eng. of 169 p. Kratzer, C. R. and P. L. Brezonik. for nitrogen in Florida lakes. 1981. A Carlson-type trophic state index Water Res. Bull. 17(4): (in press). Lamonds, A. G. 1974. Chemical and biological quality of Lake Dice at Eustis, Florida, with emphasis on the effects of storm runoff. Rept. Water Resources Invest. 36-74. U.S.G.S., Washington, D.C. Larson, D. P. and H. T. Mercier. 1975. Lake phosphorus loading graphs: an alternative. National Eutrophication Survey Working Paper # 174. U.S. EPA, Corvallis, Oregon. Lee, G. F. (chairman). 1966. Report of the Nutrient Sources of Lake Mendota. Report by the Nutrient Sources Subcommittee, Technical Committee of the Lake Mendota Problems Committee. Univ. Wisconsin, Madison, 41 p. Lie, G. B. Phosphorus cycling by freshwater macrophytes--the case of Shagawa lake. Unpublished manuscript, Limnol. Res. Center, Univ. Minnesota, Minneapolis. 56 p. -116

PAGE 125

Loehr, R. C. 1974. Characteristics and comparative magnitude of non-point sources. J.W.P.C.F. 46: 1849-1872. Lutz, J. R. 1977. Water quality and nutrient loadings of the major inflows from the Everglades Agricultural Area to the conservation areas, southeast Florida. Tech. Publ. 77-6, South Florida Water Manag. Dist., West Palm Beach, Fl., 40 p. + appendices. MacGill, R. A., S. E. Gatewood, C. Hutchinson, and D. D. Walker. 1976. Final report on the special project to prevent eutrophication of Lake Okeechobee. Fla. Dept. of State Planning DSP-BCP-36-76, Tallahassee, Fl. Mattraw, H. C. and C. B. Sherwood. 1977. Quality of stormwater runoff from a residential area, Broward County, Florida. J. Res. U.S.G.S. 5: 823. Messer, J. J., P. L. Brezonik and B. R. Snyder. 1979. Denitrification in Lake Okeechobee. Rept. No. 07-79-03 to Fla. Wat. Manag. Dist.; Dept. Env. Eng. Sci., Univ. of Florida, Gainesville, 135 p. Miller, R. A., H. C. Mattraw, Jr., and J. Hardee. 1979. data for a commercial area, Broward County, Florida. File Rept. 79-982, 127 p. Stormwater-runoff U.S.G.S. Open Miller, W. E., T. E. Maloney, and J. C. Greene. 1974. Algal productivity in 49 lakes as determined by algal assays. Water Research 8: 667-679. Miller, W. E., J. C. Greene, and T. Shiromaya. 1975. Application of algal bioassays to define the effects of wastewater effluents upon algal growth in multiple use river systems, pp. 71-92. In: Middlebrooks, E. J., D. H. Falkenborg and T. E. Maloney (eds.) Biostimulation and Nutrient Assessment. Water Research Lab PRWG 168-1. Utah State Univ., Logan, UT. Miller, W. E., J. C. Greene, and T. Shiromaya. 1978. The Selenastrum capri-cornutum Printz Algal Assay Bottle Test: experimental design, applica tion and data interpretation protocol. EPA-600/9-78-018. U.S. EPA, Corvallis, Oregon, 126 p. National Eutrophication Survey. 1975. National Eutrophication Survey methods, 1973-1976. Working Paper No. 175. U.S. EPA, Corvallis, Oregon. National Eutrophication Survey. 1977. Papers No. 224, 225 and 243-280. National Eutrophication Survey Working U.S. EPA, Corvallis, Oregon. Neter, J. and W. Wasserman. 1974. Applied linear statistical methods. Richard D. Irwin, Inc., Homewood, Ill. 842 p. Odum, H. T. 1953. Dissolved phosphorus in Florida waters. Rept. Invest. 9 Pt. I. Fla. Geol. Surv. Tallahassee, 39 p. Omernik, J. M. 1976. The influence of land use on stream nutrient levels. Ecol. Res. Ser. EPA-600/3-76-014. U.S. EPA, Corvallis, Oregon, 106 p. -117

PAGE 126

Omernik, J. M. 1977. Non-point source-stream nutrient level relationships: a nationwide study. Ecol. Res. Sere EPA-600/3-77-105. U.S. EPA, Corvallis, Oregon, 151 p. Patalas, K. and A. SaIki. 1973. Crustacean plankton and eutrophication of lakes in the Okanagan Valley, British Colombia. J. Fish. Res. Bd. Canada 30: 519-542. Pollman, C. D. and P. L. Brezonik. 1981. A model for nutrient recycling from sediments in shallow lakes. Paper presented at 44th ann. meeting Amer. Soc. Limnol. Oceanogr., Milwaukee, Wis. Pollman, C. D., T. L. Crisman, P. L. P. Sacco. 1980 Limnological studies on Lake Apopka and the Oklawaha chain of lakes 3: Water quality in 1979. Rept. No. ENV-07-80-02, Dept. Env. Eng. Sci. ,Univ. Florida, Gainesville, 109 p. Porcella, D. B. and A. B. Bishop. 1975. Comprehensive management of phosphorus water pollution. Ann Arbor Science, Ann Arbor, Mich., 303 p. Reckhow, K. H. 1977. Phosphorus models for lake management. Ph.D. dissertation, Harvard U., Cambridge, Mass. Reckhow, K. H., N. Beaulac, and J. T. Simpson. 1980. Modeling phosphorus loading and lake response under uncertainty: a manual and compilation of export coefficients. (Draft). Dept. Resource Development, Michigan State U., East Lansing, Mich., 214 p. Riekerk, H., S. A. Jones, L. A. Morris, and D. A. Pratt. 1978. Hydrology and water quality of three small lower coastal plain forested watersheds. Proc. Soil Crop Sci. Soc. Fla. 38: 105-111. Ritter, W. F., R. P. Eastburn and J. P. Jones. tion from coastal plain soils in Deleware. 1053. 1979. Nonpoint source polluTrans. A.S.A.E. 22: 1044Sakamoto, M. 1966. Primary production by phytoplankton community in some Japanese lakes and its dependence on depth. Arch. Hydrobiol. 62: 1-28. SAS Institute. 1979. Statistical Analysis System (SAS) users manual. Schulze, R. L. 1980. The biotic response to acid precipitation in Florida lakes. M.S. thesis Dept. Env. Eng. Sci., Univ. Florida, Gainesville. 146 p. Shannon, E. E. and P. L. Brezonik. 1972a. Limnological characterstics of north and central Florida lakes. Limnol. Oceanogr. 17: 97-110. Shannon, E. E. and P. L. Brezonik. 1972b. Eutrophication analysis: a multivariate approach. J. San. Eng. Div., A.S.C.E. 98: 37-57. Shannon, E. E. and P. L. Brezonik. 1972c. Relationships between lake trophic state and Nand P loading rates. Environ. Sci. Technol. 6: 719-725. Shapiro, J. 1975. The current status of lake trophic indices-a review. Interim rept. No. 15, Limnol. Res. Center, Univ. Minnesota, Minneapolis. -118

PAGE 127

Shapiro, J. 1979. The importance of trophic level interactions to the abundance and species composition of algae in lakes. Contribution No. 218, Limno10gica1 Research Center, U. of Minnesota, Minneapolis, 25 p. Smith, C. S. 1978. Phosphorus Uptake by Roots and Shoots of Myriophyllum spicatum L. Ph.D. Dissertation, Dept. of Botany, U. of Wisconsin, Madison, 113 p. Snedecor, G. W. and W. G. Cochran. 1967. Iowa State U. Press, Iowa, 593 p. Snodgrass, w. J. and C. R. O'Mella. 1975. in lakes. Environ. Sci. Techno1. 9: Statistical methods (6th ed.), Predictive model for phosphorus 937-944. Stewart, E. H., L. H .. Allen, Jr., and D. V. Calvert. 1978. Water quality of streams on the Upper Tayler Creek watershed, Okeechobee County, Fla. Proc. Soil and Crop Sci. Soc. Fla. 37: 117-120. -Terman, G.L. and L. H. Allen. 1979. fertiiizers from Lakeland sand Sci. Soc. Fla. 30: 130-140. Leaching of soluble and slow-release Nand K under grass and fallow. Proc. Soil Crop Tuscha11, J. R., T. L. Crisman, P. L. Brezonik,and J. N. Allison. 1979. Limno10gica1 studies on Lake Apopka and the Ok1awaha chain of lakes 2: water quality in 1978. Rept. No. Env. Eng. Sci., Univ. Florida, Gainesville, Fl., 114 p. U.S. Environmental Protecti:on Agency. 1978. Draft Environmental Impact State ment: Lake Apopka Restoration ProJect, Lake and Orange Counties, Florida. Prepared by the USEPA Region IV office, Atlanta. 191 p. Uttormark, P. D., J. D. Chapin, and K. M. Green. 1974. Estimating nutrient loadings to lakes from non-point sources. Eco1. Res. Sere EPA-660/3-74-020. U.S. EPA, Corvallis, Oregon, 112 p. Uttormark, P. D. and M. L. Rutchlns. 197B. Input/output models as decisl0n criteria for lake restoration. Tech. Comp1. Rept. C-7232, Land and Water Resources Institute, Univ. Maine, Orono, 61 p. Vollenweider, R. A. 1968. Scientific fundamentals of the eutrophication of lakes and flowing waters with particular reference to nitrogen and phosphorus as factors in eutrophication. Tecm. Rept. DAS/CSI/68.27. OECD, Paris, 159 p. (Revised 1971). Vollenweider, R. A. 1969. Possibilities and limits of elementary models concerning the budget of substances in lakes. Arch. Hydrobio1. 66: 1-36 Vollenweider, R. A. 1975. Input-output models with special reference to the phosphorus loading concept in limnology. Schweiz. Z. Hydro1. 37: 53-84. Vollenweider, R. A. 1976. Advances in defining critical loading levels for phosphorus in lake eutrophication. Mem. !st. Ita1. Idrobio1. 33: 53-83. 119

PAGE 128

Walker, W. W., Jr. 1977. Some analytical methods applied to lake water quality problems. Ph.D. dissertation, Harvard D., Cambridge, Mass. Wanielista, M. P., Y. A. Yousef,and W. M. McLellon. 1977. Nonpoint source effects on water quality. J.W.P.C.F. 49: 441-451. 120

PAGE 129

Table A-1. Table A-2. Table A-3. Table A-4. APPENDICES Limnological data for study lakes (mean values) Morphometric data for study lakes Hydrologic and nutrient loading data for the Florida NBS lakes Data used in the development of land-.use/nutrient loading relationships for the Florida NBS water-. sheds 121

PAGE 130

Table A-I. Limnological for study lakes (mean values). [chi al S.D. Turbidity pH AIJ. D.O. Condo Color Total SRP TKN NH: NO) + Lakl8 ]Jg/L m N.T.D. mg/Ld'as mg/L ]Jmho/cm C.P.D. P mg N/L mg N/L N0i Code CaC :3 mg P/L mg P/L mg N/L 1 3.3 2.4 8.00 7.17 10.0 85.3 0.020 0.011 0.687 0.046 0.029 2 5.2 1. 5 7. 6.83 12.0 119.3 0.040 0.015 1.007 0.049 0.040 3 8.7 1.6 8.27 B.07 41.0 182.7 0.051 0.031 0 .. 813 0.043 0 041 4 7.8 1.1 7.67 7.00 10.0 73.7 0.045 0.011 0.848 0.063 0:050 5 6.9 2.5 7. 60 10.0 1597.0 0.035 0.022 0.794 0.047 0.032 6 6.6 0.9 7.50 7.,3 10.0 114.0 0.044 0.014 1.207 0.068 0.069 7 25.4 1. 5 8.7"7 8.63 110.0 301. 7 0.028 0.016 1.369 0.089 0.098 8 24. 1 0.9 7 .'>0 7.50 19.3 127.7 0.037 0.008 1.652 0.085 0.081 9 26.3 1. 2 7.70 B.20 38.3 189.3 0.053 0.015 0.955 0.053 0.049 10 12. 1 1. 0 6. '!3 7.50 10.0 96.7 0.049 0.027 1.068 0.073 0.052 11 26.6 1.0 6.40 7.47 13.7 207.7 0.061 0.014 1.122 0.067 0.061 12 23.2 0.9 6.00 B.53 78.3 1085.0 0.072 0.035 1.557 0.073 0.057 13 14.4 0.7 7.83 B.50 129.0 621.0 0.066 0.018 1.948 0.072 0.146 14 29.9 0.8 B.77 B.!O 34.7 127.7 0.048 0.021 2.231 0.126 0.142 15 10. 2 O. 7 6. 7. 0 lB. 7 372. 3 O. 073 O. 037 1. 411 O. 077 O. 077 16 6.5 0.8 5. 7 .. 3 40.0 483.7 0.083 0.060 1.858 0.088 O. 108 17 27.1 0.9 6.73 7.97 4B.3 2022.3 0.085 0.037 1.593 0.064 0.097 18 36.7 0.8 7.87 B.23 51.3 215.3 0.035 0.011 1 .. 937 0.258 0.176 f-' 19 21. 2 0.8 7.60 7. iO 19.0 149.3 0.175 0.070 1..378 0.072 0.076 N 20 48.7 0.4 25.8 9.94 9. 7 120.5 315.5 87 0.386 0.131 4.403 0.167 0.063 N 21 9.5 1. 0 4.83 6 .. 0 13.7 64.3 0.094 0.033 0 767 0.159 0.143 22 47.0 0.2 26.1 10.0'! 9.20 130.6 307.2 63 0.358 0.051 4.939 0.222 0.099 23 39.8 0.6 18.0 9.12 B.90 102.2 255.8 27 0.200 0.022 3.124 0.154 0.057 24 27.7 1.2 4.67 7'10 16.0 116.3 0.128 0.064 1.297 0.120 0.073 25 37.7 0.9 6.70 B. 0 77.0 266.0 0.711 0.530 1.512 0.060 0.053 26 102.0 0.7 0.17 B.O 117.3 500.7 0.232 0.027 2:498 0.087 0.098 27 34.6 0.7 6.20 7.53 52.3 1003.0 0.142 0.075 1.662 0.091 0.079 28 30.6 0.7 6.60 7.90 24.0 168.0 0.342 0.245 L 917 0.094 O. 106 29 14.8 0.6 5. 60 39.7 953.7 0.199 0.143 1.952 0.187 0.125 30 97.9 0.4 10. 15 7. 86.0 373.0 0.692 0.251 4 .. 682 O. 143 0.198 31 70.2 0.9 8.23 9.23 81.3349.0 0.484 0.318 2,403 0.089 0.089 32 54.1 0.9 ?60 B.87 66.0 252.3 1.258 1.139 1,664 0.082 0.215 33 208.6 0.5 14. 10'10 80.0 334.7 0.631 0.256 4 ... 457 0.093 0.390 34 76.5 0.3 7.70 8. 7 '73.3 792.7 0.505 0.322 2.720 0.133 0.153 35 87.7 0.7 7.33 8. '7 51. 3 170.3 0.753 0.554 2.718 0.168 0.151 36 77.0 0.7 5.67 7.1;>7 30.3 198.3 1.072 0.933 2.833 0.230 0.317 37 84.9 O. 1 5. 7.93 42.7 231. 0 2.750 0.254 5 .. 463 0.200 1. 063 38 140.3 0.4 12.03 ?po 64.3 226.7 1.907 1.157 5,433 1.257 0.483 39 261.4 0.3 1.33 151.0 340.3 1.447 0.816 4.997 0.260 0.197 40 276.6 0.4 5.60? l3 87.0 348.3 1.483 0.943 5;562 0.238 0.845 41 5.4 2.2 2. 'l 8.65 6. 1.5 52.3 65 0.036 0.004 0,646 0.107 0.042 42 4.5 1. 4 8. ,14 5. 0.8 52.4 208 0.036 0.006 0,.880 O. 156 0.095 43 6.7 1.8 7. /0 6. 6 1.6 43.4 90 0.033 0.010 0.808 0.078 0.040 44 5.8 1.4 2.7 8.34 6. 6 2.0 50.2 162 0.041 0.010 0.726 0.104 0.021 45 20.9 0.9 5. J. 7. 7.24 5.2 60.6 133 0.171 0.024 1. 463 0.216 0.033 46 6.5 0.7 4.4 7.14 5.92 2.0 44.0 216 0.054 0.006 0,921 0.110 0.096 47 2.7 2.2 1.7 7. 24 0.2 38.8 19 0.023 0.007 1.407 0.070 0.017 48 47.9 0.8 4. ;:, 8.05 B .. 6 68. 5 165.8 62 0.081 0.007 1, 839 0.076 0.011 49 7.3 0.9 3.1 7. 6. 0 3.0 49.6 289 0.125 0.060 1.120 0.166 0.090 50 10.7 1.4 ?O 7.72 6.11 3.0 48.5 162 0.071 0.030 1,053 0.116 0.080 Z.8 1. 5 B. S. !. B 4?1 163 0.044 O. ()J.3 ().948 Q. ()::;6 0.116

PAGE 131

Table A-I (continued). Total SRP' NO; + Lake [chI a] S.D. Turbidity_pH mgt L as D.O. Condo Color P TKN NH+ NOCode llg/L m N.T.D. C C03 mg/L llmho/cm C.P.D. mg P/L mg/L m N/L mg N7L mg !:Ii:! /.7 1.0 :,. J U. Otl .24 0.6 ::;.q..o 241 0.104 0.051 O. 551 O. 090 0.019 53 2. 3 1.6 2.3 8. 52 .40 2.0 37.8 39 0.013 0.005 0.666 0.048 0.011 54 17.8 0.9 5. 11 9. 18 .90 17.8 84. 8 130 0.059 0.002 1. 511 O. 138 0.013 55 14.0 0.9 'I 8. 61 .55 15. 1 74.4 120 0.062 O. 007 1.270 O. 130 0.065 ., 56 11. 5 1.1 R.l 5. 78 .76 0.0 25.2 125 O. 039 0.003 1. 180 O. 074 0.000 57 38. 2 0.6 'I.B 8. ::-12 .50 7.0 59.3 235 0.104 0.011 1. 743 0.305 0.060 58 38.3 0.6 2. 1 2. 05 .10 1.6 52.0 390 0.123 0.045 1. 112 0.296 0.116 59 20. 3 O. 7 1.5 3. 97 .57 1.0 43.0 539 0.178 0.098 1. 918 0.410 0.060 60 87. 7 0.4 18.0 7.42 .25 119.6 305. 1 94 0.461 o. 150 2.245 O. 369 0.012 61 4. 5 1-.9 3. ::-1 8. ::i2 .74 28.4 99. 2 45 0.039 0.009 1.070 O. 182 0.055 62 5. 8 1.5 2. 1 6. Hl .86 163.0 541. 3 46 0.857 O. 439 O. 561 0.057 0.021 63 44. 8 O. 6 11. 0 9. .84 107. 8 256. 1 48 O. 521 0.225 1.987 0.149 0.044 64 34. 0 O. 8 5.9 7. riO .04 37.0 131. 4 137 0.374 O. 123 1.481 0.174 0.018 65 2. 8 1.8 1.8 8. 86 .10 42.4 99.8 42 O. 038 O. 006 O. 795 0.042 0.018 66 20.7 2.0 1.0 3. 58 .50 1.8 41. 2 304 0.124 0.059 0.820 O. 090 0.070 67 27.0 1.5 3. 1 2.92 .06 5.8 50. 0 135 0.297 0.116 O. 793 0.082 0.131 68 29. 6 O. 5 7. 5 5. 96 .72 58.2 133.8 139 0.431 0.099 2.915 0.246 0.030 69 13.6 1.0 5. 9 8.46 .72 0.0 36. 6 156 0.051 0.012 1. 014 O. 086 0.069 70 7. 6 1.0 2. 4 5. 15 .05 0.0 21. 0 254 0.052 O. 003 1. 080 O. 075 0.010 71 57. 1 O. 6 5. 5. 413 .86 11. 6 61. 2 370 O. 535 O. 186 2.324 0.360 0.181 72 32.9 O. 9 4.b 8.41 .79 15.0 64.9 63 O. 167 O. 034 1.899 O. 189 0.025 H 73 8. 5 1.0 5 4.'10 .60 11.0 49. 5 435 0.315 O. 127 1. 110 0.035 0.030 N 74 14. 5 1.0 8.B 9.20 .48 73.4 203.6 15 0.039 O. 009 1. 535 O. 150 0.042 75 23. 0 O. 8 16. 4 9. 14 .90 92. 6 252.6 36 O. 184 O. 018 3. 296 O. 144 0.026 76 6. 3 2. 0 2.2 8. T1 .14 10.6 134.4 0.024 0.007. 1.054 0.171 0.031 77 1.7 4. 5 1.'1 s.p:;? .90 5. 8 52. 9 0.014 0.002 O. 399 0.101 O. 058 78 1.5 3. 4 a.3 9. .72 1.0 28. 0 7 0.013 0.003 O. 271 0.026 O. 081 79 1.9 3. 9 1 B.20 .50 O. 6 25. 2 13 O. 013 0.003 O. 318 0.042 O. 086 80 3. 0 3. 7 2. 5 9.06 .76 O. 6 30.2 6 0.013 O. 004 0.915 O. 096 0.060. 81 1.5 3. 3 'I 9. J.5 .42 1.0 48. 1 8 0.017 0.003 O. 554 0.039 0.179 82 1.6 3. 2 1.9 8. '74 .92 1.5 44. 4 3 0.014 O. 005 0.295 O. 060 0.015 83 4. 3 1'.3 4 6.76 .36 O. 6 40. 4 178 0.028 O. 008 0.916 O. 122 0.046 84 1.6 6.2 l..7 8. "/8 .26 O. 2 35.0 8 0.016 O. 006 0.224 0.070 O. 063 85 5. 1 O. 8 1.7 6. :-i2 .68 0.0 37.4 538 0.035 0.011 O. 916 0.116 0.058 86 2. 4 4.8 1. 1 7. "/4 .06 0.0 35.9 2 0.017 0.005 O. 373 0.051 O. 023 87 3. 6 3. 5 1.2 8. .81 0.0 38.0 2 O. 018 0.003 O. 449 O. 150 0.093 88 4. 0 0.9 2. 2 7. 11 .89 O. 1 44. 1 397 0.036 0.007 0.872 O. 100 0.051 89 1.4 4. 3 2. 2 9.02 .00 0.0 50. 0 9 0.012 0.005 0.185 0.028 O. 040 90 2. 6 3. 6 1.8 7. riO .40 2.2 39. 6 21 0.023 0.006 0.663. 0.086 O. 068 91 2.1 3. 9 ::'>.:::1 9. .44 O. 4 46. 4 5 0.013 O. 007 0.295 0.046 0.005 92 1.8 3.2 ::1..3 8.78 .76 1.0 38. 6 21 0.018 0.005 0.411 0.034 0.011 93 1.5 5. 5 1..11 8. J" 75 O. 0 34. 5 7 0.008 O. 004 0.260 0.014 O. 012 94 12.7 1.8 1.8 7. r,2 .71 12.4 146. 5 19 0.024 O. 006 O. 532 0.038 0.017 95 10. 7 1.4 2. 4 8. .79 11.9 125.0 32 0.021 0.008 0.458 0.030 O. 020 96 2. 5 1.4 2.4 8. .24 O. 0 23.2 31 0.014 0.007 0.262 0.045 O. 007 97 8. 5 O. 7 4. ? 9. .14 4. 4 93.3 154 0.068 0.027 O. 781 0.070 0.026 98 9. 3 2.0 2. 0 7.95 .67 11. 5 129. 0 21 O. 014 0.007 O. 539 0.020 0.026 99 3.2 1. 1 ;;!.9 B. 55 .50 5. 4 142.8 28 0.058 0.015 0.326 0.037 O. 009 100 7.2 2. 4 l.. "' B.47 .53 5. 4 69. 8 15 0.013 0.005 0.431 0.090 O. 016 101 0.9 7.9 O. 6 8. 71 .14 O. 0 18. 6 5 O. 007 O. 004 O. 189 O. 043 0.006

PAGE 132

Table A-2. Morphometriedata for study lakes. Lake 1_ ,",uue J r, I! I (j '/ JO J 1 1 18 JI! Jb 17 Hi 6"1 70 7J Fe' 7:{ 71! 7' 7t ; 73 77 80 81 83 8'1 85 86 87 88 87 90 91 92-73 91! r"/!) r/6 97 98 '17 JOO lUi xlO m 20.63 133. 11 1. 17 48.80 25.63 201. 73 58.29 254. 52 1. 54 0.46 7;25 4.01 5.293..00 16.67 141. 22 26.40 38.88 46 56 3.84 67. 11 122.36 161. 36 95.47 3.24 8.35 4.99 558.48 182.88 63.90 18.29 14.07 4.00 1. 36 35.20 2.05 0.61 0.69 1. 54 0.41 1. 83 22.42 0.92 7.96 1. 89 0.40 0.07 O. 56 9.27 0.93 .1. 87 1. 92 0.05 73. 75 80.96 0.03 44.83 0.04 0.13 0.08 0.03 0.26 1. 14 0.06 0.06 0.06 O. 19 o 57 3. 19 0.06 O. 48 3. 84 3.22 234.36 123.61 144.96 48.62 24. 384 6.640 14. 478 28.372 10.896 3.591 3.402 1. 435 O. 120 0.100 1. 175 3. 536 4.420 8. 880 1. 190 2. 905 5.880 6. 264 9: 880 84 180 106: 320 O. 399 Area, l{m2 7. 64 49.30 O. 39 30. 50 10.25 112.07 15.95 141. 40 O. 77 0.38 2.90 4.46 1890. 71 12. 82 70. 61 17.60 12.96 J4 11 1. 92 22. 37 43. 70 124. 12 39. 78 O. 72 3.34 1. 85 186. 16 76.20 35. 50 18.29 4. 69 1. 60 1. 36 32.00 1. 37 0.41 0.63 1. 03 O. 41 1. 22 4. 67 0.27 2.21 0.86 0.27 0.05 0.20 3.31 0.29 O. 52 0.64 O. 05 25. 43 44. 98 O. 04 29.89 O. 01 O. 11 O. 04 O. 02 0.29 O. 76 0.04 O. 06 0.02. 0.05 o 82 2. 13 O. 05 O. 22 1. 01 2. 48 55. 80 30. 15 23. 01 6. 66 5. 08 O. 83 2. 54 6. 92 2. 27 1. 71 O. 42 O. 41 O. 06 O. 05 O. 47 1. 04 O. 85 2.40 O. 34 0.35 1. 47 2. 16 1. 40 4. 94 14. 03 1. 91 13.29 o 07 124 Zmax, m 2. 7 2.7 3.0 1.6 2. 5 1.8 3.7 1.8 2.0 1.2 2. 5 0.9 2.8 1.3 2.0 1.5 3.0 3 3 2.0 3.0 2.8 1.3 2.4 4. 5 2. 5 2.7 3.0 2.4 1.8 1.0 3.0 2. 5 1.0 1. 1 1. 5 1.5 1. 1 1.5 1.0 1.5 4.8 3.4 3.6 2.2 1.5 1.5 2.8 2.8 3.2 3.6 3.0 1. 1 2.9 1.8 0.8 1.5 4.0 1.2 1.9 1.6 O. 9 1.5 1.6 1.0 3. 1 3.8 0.7 1.5 1.2 2.2 3.8 1.3 4.2 4. 1 6.3 7.3 4. 8 8. 0 5. 7 4. 1 4.8 2. 1 8. 1 3. 5 2.0 2.0 2. 5 3. 4 5. 2 3.7 3. 5 8.3 4. 0 2.9 2:0 6. 0 8. 0 5. 7 Z, m 4. 3 4. 5 3. 9 3. 0 4. 6 2. 7 4. 6 2. 4 3. 3 2. 4 4.3 1.0 4. 7 3.0 3. 4 2. 5 4. 0 4 3 2. 7 4. 9 15. 0 3. 4 4. 3 8. 4 4. 3 4: 5 4. 0 5. 8 1.5 5. 2 4. 0 1. 5 2. 7 2. 5 2. 7 1.8 2. 0 1.5 2.7. 7. 0 5. 2 5. 8 3. 4 2. 4 3. 0 4. 3 4. 3 4. 6 5. 8 4. 9 1.5 3. 7 3. 0 1.2 4. 0 25. 3 2. 1 3. 4-2. 4 1.5 1.9 2. 7 1.8 6. 7. 0 o 9 3. 4 1.8 2. 7 5. 2 1.8 7. 6 5. 2 9. 8 22. 9 9. 1 14.0 10. 4 8. 8 9. 4 2. 7 13. 4 5. 2 3. 7 4. 6 3. 7 6. 4 8. 8 8.8 5. 8 20. 0 8. 0 B. 0 4. 0 3. 0 12. 0 3. 0 17.0 17 0

PAGE 133

Table A-3. Hydrologic $.nd nutrient loading data for the Florida NES lakes, Lak.e 'w: qs' in-P Oytj R N. N R p ?ut n Code yr m/yr k /yr kg yr kg yr 1 0.351 7. 70 1540 O. 25 52345 125420 -1. 40 2 0.677 3.99 5025 O. 86 363650 228820 0.37 3 0.266 11.29 185 O. 24 4020 3855 0.04 4 O. 721 2.22 2260 O. 37 116980 100050 O. 14 5 O. 518 4.83 1830 O. 61 110820 38520 O. 65 6 0.425 4.24 25570 O. 24 595380 567575 0.05 7. 1. 501 2. 43 36865 0: 46 1983505 1478835 0:25 8 0.241 7.47 9 7265 10 2:901 0:86 165 0:92 45155 4080 0:91 11 12 2:329 t: 20 58010 0: 89 6970160 1953590 0:72 13 14 0.644 2. 02 1065 O. 43 63540 54375 0.14 I-' 15 0.419 4. 77 33775 0.50 661810 549230 0.17 16 N 17 1: 699 t: 77 2295 0: 77 82085 70110 0: 15 U1 18 1.499 2. 20 1010 O. 89 1066300 45680 O. 96 19 1.699 1. 18 420 O. 75 8040 3970 O. 51 20 0.605 4.95 12450 O. 70 423890 312365 O. 26 21 O. 082 34.07 139420 O. 47 2205015 1735985 0.21 22 2. 477 0.69 35925 0.54 417730 354670 0.15 23 0.340 8.54 29225 O. 42 908380 913825 -0.01 24 0: 107 6 110 56055 110035 -0:07 25 23.40 O. 07 117245 26 O. 351 7. 70 310 2830 O. 15 21810 41635 -0. 91 27 O. 159 18. 88 870 561005 O. 10 12675785 6865460 O. 46 28 0.471 5. 09 265 63410 O. 83 1631585 1269745 O. 22 29 O. 027 65. 70 075 543740 -0. 04 5723790 5760010 -0.01 30 0.263 3. 80 945 170365 31 0:082 610 0:29 183600 92470 0: 50 32 30. 42 66375 33 O. 063 15.87 100 38605 O. 42 134930 89725 0.34 34 O. 225 4.90 590 71750 O. 47 427090 438290 -0.03 35 0:036 425 17910 83235 59790 36 42. 12 460 0: 70 35210 0: 54 37 O. 099 11. 15 2830 16220 38 O. 030 49. 77 7 500 213655 39 40

PAGE 134

"-I-1:--) Qj: -" It Ta.ble A-4. Data used in the develop NES watersheds. Land u Tr:ibutary ALLURB CPAST OAG FOR Code 02Al 0.356 0.225 O. 048 0.006 03Bl 0.622 0.000 O. 087 0.084 04Al 0.057 0.065 O. 465 0.050 05Al 0.090 0.046 O. 493 0.221 06Bl 0.057 0.257 0.166 O. 104 0801 O. 022 0.264 O. 001 0.028 09Al 0.318 0.069 0.215 0.039 13Cl 0.037 O. 700 O. 020 0.008 1301 O. 101 0.694 O. 000 0.023 13Fl 0.000 O. 663 0.017 0.027 13Gl 0.021 0.386 O. 064 0.047 1401 0.004 0.343 0.192 0.295 15Bl 0.024 0.082 0.015 O. 563 15Cl 0.002 0.286 0.000' O. 247 19B1 0.048 0.296 O. 605 0.024 20Al 0.055 0.223 O. 295 O. 041 20Bl O. 748 0.000 O. 018 O. 162 21Bl 0.000 0.000 o. 000 0.935 2101 0.003 0.075 O. 000 0.896 23A1 O. 159 0.302 0.198 O. 103 25A1 0.243 0.388 O. 216 0.065 26Bl O. 516 0.072 0.196 0.062 26C1 0.626 O. 102 O. 046 0.059 28Al 0.330 0.098 O. 095 0.025 28Bl 0.125 0.621 O. 055 0.000 29Bl 0.020 O. 146 O. 001 0.446 30A2 O. 122 0.217 O. 076 O. 166 30Bl 0.314 0.064 O. 189 0.100 30Cl O. 573 0.037 O. 063 0.031 32Al O. 785 0.000 O. 037 0.000 34Al 0.695 O. 009 O. 036 0.119 34Bl 0.294 0.057 0.311 0.148 34Cl 0.214 O. 189 0.100 0.264 3401 0.385 O. 150 O. 171 O. 117 34El 0.016 0.132 O. 003 O. 721 35A1 0.339 0.242 0.109 0.230 35Bl O. 765 0.000 O. 000 0.179 35Cl 0.415 0.016 O. 000 0.420 37Bl 0.445 O. 049 O. 000 0.366 37Cl O. 578 O. 421 O. 000 0.000 38Bl 0.458 0.051 O. 000 0.441 = urban + residential, CPAS forest, RA = rangeland, NFWET = no alnd other barren land. (See Chapt a:['ea. DA = drainage area, TPL and ent of land-use;nutrient loading relationships lor the Florida e characteristics RA NFWET WA SHINE DAi TPL, T L, FLm'l, m 3/sec Km kg/yr kg O. 196 O. 103 O. 063 0.003 223. 5 12695 O. 117 0.000 O. 076 0.000 2. 6 115 O. 012 0.284 O. 040 0.027 143. 4 1185 O. 083 0.000 O. 050 0.000 8. 6 210 O. 191 O. 147 O. 078 0.000 978. 1 17510 O. 506 0.088 O. 091 0.000 254. 3 2900 O. 049 0.000 O. 295 0.000 7. 1 1525 0.130 O. 104 O. 000 0.000 289. 0 38560 O. 081 O. 101 O. 000 0.000 115.7 5640 O. 203 0.089 O. 000 O. 000 386. 3 7190 O. 395 0.086 O. 003 0.000 496. 5 26315 0.143 0.000 O. 007 0.000 8. 5 130 O. 000 0.299 O. 016 0.000 726. 5 30680 O. 000 0.461 O. 004 0.000 100. 1 13340 O. 000 O. 000 O. 000 0.000 1.6 205 O. 000 0.052 O. 333 0.001 381. 7 35920 O. 000 0.000 O. 010 0.000 1. 1 20 O. 000 0.065 O. 000 0.000 74.1 615 O. 000 0.027 O. 000 O. 000 26. 6 1435 O. 000 O. 198 O. 040 O. 000 12. 9 215 O. 000 0.054 O. 018 0.017 139. 5 49940 0.103 O. 000 O. 010 0.000 1.0 115 0.138 0.000 O. 014 0.000 18.5 895 O. 068 0.325 O. 058 0.000 472. 9 113285 0.125 0.043 O. 031 0.000 26.1 6405 O. 333 0.000 0.075 0.000 15.8 375 O. 062 0.075 O. 222 0.060 56. 0 12705 O. 000 0.027 O. 054 0.253 41. 6 11800 O. 002 0.028 O. 261 0.005 53. 9 5230 O. 000 0.025 O. 153 0.000 57. 2 55110 O. 024 0.054 O. 061 0.000 33. 8 10990 O. 000 0.098 O. 085 0.008 55. 0 1470 O. 000 0.179 O. 055 0.000 20. 5 5725 O. 000 0.077 0.100 0.001 124. 0 43435 0.127 0.000 O. 001 0.000 10. 6 3750 O. 055 0.000 O. 012 0.000 1.7 205 O. 007 0.000 O. 000 0.000 1.5 470 O. 131 0.000 O. 000 0.000 3. 8 755 O. 105 0.000 O. 004 0.000 5. 4 6100 O. 000 0.000 O. 000 0.000 O. 4 150 O. 000 0.029 O. 008 0.012 156. 5 30440' = cropland and pasture, OAG = other agricultur -forested wetlands, WA = open water, SHINE = st r II.) All land use data as fraction of total TNL = non-point source phosphorus and nitrogen yr 1. 60 O. 06 1. 39 O. 08 10. 39 2.30 0.07 2. 87 1. 01 1. 52 5. 96 0.08 7.90 0.80 O. 02 3. 50 0.02 1. 40 O. 50 0.06 2. 40 0.01 0.24 2.67 0.31 O. 20 0.60 O. 70 0.30 1. 30 0.35 0.22 0.25 2.67 0.20 O. 02 0.02 O. 10 0.12 O. 01 1. 75 FOR = ip mines atershed oading.