United States
Environmental Protection
Agency
Office of Research and
Development
Washington, DC 20460
EPA/600/3-91/012
November 1990
xvEPA
National Surface Water
Survey
Temporal Variability in
Lakewater Chemistry in the
Northeastern United States:
Results of Phase II of the
Eastern Lake Survey
LAKE
-------
EPA/600/3-91/012
November 1990
TEMPORAL VARIABILITY IN LAKEWATER CHEMISTRY
IN THE NORTHEASTERN UNITED STATES:
RESULTS OF PHASE II OF THE EASTERN LAKE SURVEY
December 1990
A.T. Herlihy1, D.H. Landers2, R.F, Cusimano3, W.S. Overton4, P.J. Wigington, Jr.2
A.K. Pollack5, and I.E. Mitchell-Hall6
Utah State University, Utah Water Research Laboratory, c/o EPA Environmental Research Laboratory,
200 SW 35th St., Corvallis, OR 97333.
2 U.S. Environmental Protection Agency, EPA Environmental Research Laboratory, 200 SW 35th St.,
Corvaliis, OR 97333,
3 State of Washington, Department of Ecology, Mail Stop PV-11, Olympia, WA 98504.
4 Department of Statistics, Kidder Hall No. 8, Oregon State University, Corvallis, OR 97333.
5 Systems Applications Inc., 101 Lucas Valley Rd., San Rafeal, CA 94131.
6 Lockheed Engineering & Sciences Company, 6585 S. Paradise Rd., Las Vegas, NV 89119.
ijf) Printed on Recycled Paper
-------
ACKNOWLEDGMENTS
The ELS-II could not have been completed without the hard work and dedication of numerous
people at the EPA laboratories in Corvallis and Las Vegas, and at NSI Technology Services, Inc.,
Lockheed Engineering and Sciences Company, and Systems Applications, Inc. In addition, the advice
and assistance provided by local, state, and federal agencies has been invaluable. The help of all
involved parties is gratefully acknowledged.
The research described in this report has been funded by the U.S. Environmental Protection
Agency. This document has been prepared at the EPA Environmental Research Laboratory in Corvallis,
Oregon, through cooperative agreements CR815168 with Utah State University and CR815422 with
Oregon State University, and contract numbers 68-C8-0006 with NSI Technology Services, 68-03-3249
with Lockheed Engineering and Sciences Company, and 68-03-3439 with Kilkelly Environmental
Associates. The report has been subjected to the Agency's peer and administrative review and
approved for publication. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
-------
TABLE OF CONTENTS
Section Page
EXECUTIVE SUMMARY , xl
1, INTRODUCTION 1
1.1 Overview , 1
1.2 The National Surface Water Survey 2
1.3 Eastern Lake Survey - Phase II 3
1.4 The ELS-II Data Report 5
2. ELS-II SURVEY DESIGN 6
2.1 Overview 6
2.2 Eastern Lake Survey - Phase I 6
2.3 Eastern Lake Survey - Phase II 8
2.3.1 Defining the Population of Interest 8
2.3.2 Statistical Sampling Design 9
2.4 ELS-II Target Population 11
2.4.1 Refinement of the Target Population 11
2.4.2 Making ELS-II Population Estimates 13
2.4.3 Comparison to DDRP Population 15
3. METHODS 16
3.1 Overview 16
3.2 Field Sampling Plan 16
3.2.1 Spring Seasonal Survey 20
3.2.2 Summer Seasonal Survey 20
3.2.3 FallSurvey and Variability Study 20
3.3 Field Methods 24
3.3.1 Site Location 24
3.3.2 Sample Collection 24
3.3.3 In situ Measurements 25
3.4 Sample Handling 25
3.5 Processing Laboratory Techniques 25
3.6 Analytical Laboratory Support 27
3.7 Data Base Development , 27
3.7.1 Raw Data Set (Data Set 1) 30
3.7.2 Verified Data Set (Data Set 2) 30
3.7.3 Validated Data Set (Data Set 3) 32
3.7.4 Enhanced Data Set (Data Set 4) 33
-------
4. DATA QUALITY ASSESSMENT 34
4.1 Introduction . 34
4.2 Completeness, Comparability, and Representativeness 34
4.3 Detectability, Accuracy, Precision, and Laboratory Bias 35
4.3.1 Detectability 36
4.3.2 Accuracy 36
4.3.3 Precision 38
4.3.4 Laboratory Bias 42
4.3.5 Variable by Variable Summaries 42
4.3.5.1 ANC and pH 42
4.3.5.2 Sulfate 46
4.3.5.3 Nitrate 46
4.3.5.4 Chloride 47
4.3.5.5 Dissolved Organic Carbon 47
4.3.5.6 Base Cations 47
4.3.5.7 Aluminum 48
4.5 Conclusions 48
5. RESULTS - TEMPORAL VARIABILITY IN THE NORTHEAST 50
5.1 Overview 50
5.1.1 Components of Temporal Variability 50
5.1.2 Cumulative Distribution Functions 50
5.1.3 Characteristics of the ELS-II Target Population 52
5.1.4 Aluminum Measurements 52
5.2 Between-Year Variability 55
5.2.1 Population Estimates of Acidic and Low ANC Lakes 55
5.2.2 Chemical Changes 57
5.2.2.1 ANC/pH 57
5.2.2.2 Inorganic Acid Anions 57
5.2.2.3 Sum of Base Cations 63
5.2.2.4 DOC/A!uminum 63
5.2.3 Conclusions - Between-Year Variability 63
5.3 Within-Season Variability 68
5.3.1 Chemical Differences 68
5.3.1.1 ANC/pH 68
5.3.1.2 Inorganic Acid Anions 72
5.3.1.3 Sum of Base Cations 72
5.3.1.4 DOC/Aluminum 72
5.3.2 Conclusions: Robustness of the Fail Index 77
5.4 Among-Season Variability 77
5.4.1 Population Estimates of Acidic and low ANC Lakes 77
5.4.2 Seasonal Chemical Differences 79
5.4.2.1 ANC/pH 83
5.4.2.2 Inorganic Acid Anions 83
5.4.2.3 Sum of Base Cations 90
5.4.2.4 DOC/Aluminum 90
5.4.3 Conclusions: Among-Season Variability 96
IV
-------
6. SYNTHESIS AND DISCUSSION 97
6.1 Assessment of ELS-I! Variability 97
6.1.1 Components of Variability 97
6.1,2 Pooling Variance Estimates 97
6.1.3 Relative Importance of Variability Components 98
6.1.3.1 ANC and pH 100
6.1.3.2 Sulfate and Nitrate 100
6,1.3.3 Sum of Base Cations 100
6.1.3.4 DOC 104
6.1.4 Conclusions and Comparison to Long-Term Monitoring Data 104
6.2 Robustness of the ELS-I Fall Index Sample 105
6.2.1 Fall Index Variability 105
6.2.2 Predicting Spring Conditions From Fall Index Data 105
6.3 Spring Conditions 108
6.3.1 Seasonal Biotic Toxicity 108
6.3.2 Factors Related to Fall/Spring ANC Changes 110
6.3.3 Worst Case Spring Conditions 116
7. LITERATURE CITED 124
Appendix A - Population Estimates of Selected Physical and Chemical Variables
in the Spring, Summer, and Fall ELS-II Seasonal Surveys 129
Appendix B - Listing of Lakes Sampled in ELS-II 208
Appendix C - Chi-square Analyses of ELS-II Population Differences 213
-------
LIST OF FIGURES
Figure Page
1-1 Organization of the National Surface Water Survey, showing the two major
components, the National Lake Survey and the National Stream Survey,
and their relationship to later phases of study 4
2-1 ELS-II sample lakes in the five subregions of the northeastern United States 7
3-1 Sampling strategies for the spring seasonal survey, ELS-ll 21
3-2 Sampling strategies for the summer seasonal survey, ELS-II 22
3-3 Sampling strategies for the Fall Seasonal Survey, ELS-ll 23
3-4 Processing laboratory activities, ELS-II 26
3-5 ELS-II data base development 29
4-1 Comparison of laboratory 1 and laboratory 2 values for replicate samples
analyzed during the summer seasonal survey: (a) ANC, (b) sulfate,
(c) nitrate, (d) DOC, and (e) chloride 44
4-2 Comparison of laboratory 1 and laboratory 2 values for replicate samples
analyzed during the summer seasonal survey: (a) sodium, (b) potassium,
(c) calcium, and (d) magnesium 45
5-1 Hypothetical example of the type of cumulative distribution function plot
used in the ELS-ll report 51
5-2 Population distribution and comparison of fall 1986 ANC with fall 1984 ANC
in ELS-ll lakes 58
5-3 Population distribution and comparison of fall 1986 pH with fall 1984 pH
in ELS-II lakes 59
5-4 Population distribution and comparison of fall 1986 sulfate with fall 1984
sulfate in ELS-II lakes 61
5-5 Population distribution and comparison of fall 1986 nitrate with fall 1984
nitrate in ELS-II lakes 62
5-6 Population distribution and comparison of fall 1986 chloride with fall 1984
chloride in ELS-II lakes 64
5-7 Population distribution and comparison of fall 1986 sum of base cations with
fall 1984 sum of base cations in ELS-II lakes 65
5-8 Population distribution and comparison of fail 1986 DOC with fall 1984 DOC
in ELS-II lakes 66
5-9 Population distribution and comparison of fall 1986 MIBK-extractable aluminum
with fall 1984 MIBK-extractable aluminum in ELS-ll lakes 67
VI
-------
5-10 Population distribution of the three sample visits in the ELS-II Fall Variability
Study: (a) ANC, and (b) pH 69
5-11 Scatterplot of all possible Fall Variability Study pairs (visit 1 /visit 2,
visit 2/visit 3, and visit 1/visit 3) for (a) ANC, and (b) pH 70
5-12 Population distribution of the three sample visits in the ELS-II Fall Variability
Study: (a) sulfate, (b) nitrate, and (c) chloride 73
5-13 Scatterplot of all possible Fall Variability Study pairs (visit 1/visit 2,
visit 2/visit 3, and visit 1 /visit 3) for (a) sulfate, (b) nitrate, and
(c) chloride 74
5-14 Population distribution of the three sample visits in the ELS-II Fall Variability
Study: (a) sum of base cations, and (b) inorganic monomeric aluminum 75
5-15 Scatterplot of all possible Fall Variability Study pairs (visit 1/visit 2,
visit 2/visit 3, and visit 1/visit 3) for (a) sum of base cations, (b) DOC, and
(c) inorganic monomeric aluminum 76
5-16 ANC population distribution for the spring, summer, and fall seasonal surveys
and comparison of spring 1986 and fail 1986 ANC in ELS-II lakes 84
5-17 pH population distribution for the spring, summer, and fall seasonal surveys and
comparison of spring 1986 and fall 1986 pH in ELS-II lakes 85
5-18 Population minimum, 25th percentile, median, 75th percentile, and maximum
values in each of the three ELS-II seasonal surveys for (a) ANC, and (b) pH 86
5-19 Sulfate population distribution tor the spring, summer, and fall seasonal surveys
and comparison of spring 1986 and fall 1986 sulfate in ELS-II lakes 87
5-20 Population minimum, 25th percentile, median, 75th percentile, and maximum
values in each of the three ELS-II seasonal surveys for (a) sulfate, and
(b) nitrate 88
5-21 Nitrate population distribution for the spring, summer, and fall seasonal surveys
and comparison of spring 1986 and fall 1986 nitrate in ELS-II lakes 89
5-22 Sum of base cation population distribution for the spring, summer, and fall
seasonal surveys and comparison of spring 1986 and fall 1986 sum of base
cations in ELS-II lakes 91
5-23 DOC population distribution for the spring, summer, and fall seasonal surveys
and comparison of spring 1986 and fall 1986 DOC in ELS-II lakes 92
5-24 Population minimum, 25th percentile, median, 75th percentile, and maximum
values in each of the three ELS-II seasonal surveys for (a) sum of base cations,
(b) DOC, and (c) inorganic monomeric aluminum 93
5-25 Inorganic monomeric aluminum population distribution for the spring, summer,
and fall seasonal surveys and comparison of spring 1986 and fall 1986 inorganic
monomeric aluminum in ELS-II lakes 94
VII
-------
6-1 Population distribution of the temporal standard deviation in the 41 Fall Variability
Study lakes for (a) ANC, and (b) pH 101
6-2 Population distribution of the temporal standard deviation in the 41 Fall Variability
Study lakes for (a) sulfate, and (b) nitrate 102
6-3 Population distribution of the temporal standard deviation in the 41 Fall Variability
Study lakes for (a) sum of base cations, and (b) DOC 103
6-4 Relationship between ANC change (spring 1986 minus fall 1986) and fall 1986
index ANC in ELS-II lakes 111
6-5 Relationship between ANC change (spring 1986 minus fall 1986) and sum of base
cation concentration change (spring 1986 minus fall 1986) in ELS-II lakes with
fall 1986 index ANC < 50 jueq/L 113
6-6 Relationship between ANC change (spring 1986 minus fall 1986) and nitrate
concentration change (spring 1986 minus fall 1986) in ELS-II lakes with
fall 1986 Index ANC < 50 /xeq/L 114
6-7 Relationship between ANC change (spring 1986 minus fall 1986) and inorganic
monomeric aluminum concentration change (spring 1986 minus fall 1986)
in ELS-II lakes with fall 1986 index ANC < 50 peq/L 117
6-8 Relationship between pH change (spring 1986 minus fall 1986) and inorganic
monomeric aluminum concentration change (spring 1986 minus fall 1986)
in ELS-II lakes with fall 1986 index ANC < 50 /ueq/L 118
6-9 Hydrograph of mean daily discharge during spring 1986 (March 1 to June 1) in
Woods Lake Outlet, New York (Adirondacks subregion) 120
6-10 Hydrograph of mean daily discharge during spring 1986 (March 1 to June 1) in
Towanda Creek, Pennsylvania (Poconos/Catskills subregion) 121
6-11 Hydrograph of mean daily discharge during spring 1986 (March 1 to June 1) in
Swift River, Maine 122
via
-------
LIST OF TABLES
Table Page
2-1 Estimated Population and Sample Sizes of Northeastern Lakes (Region 1) in
ELS-I and ELS-II 12
3-1 Chemical and Physical Variables Measured in ELS-II and Methods Employed 17
3-2 Maximum Holding Times Specified for ELS-li Samples 28
3-3 Variables Validated in ELS-II 33
4-1 System Decision Limits (Mean + 1.65 Standard Deviation of Field Blanks) for
Major ELS-II Analytes 37
4-2 Mean ± One Standard Deviation of Seventh Lake (SL) and Big Moose Lake
(BML) Field Natural Audits for Major ELS-II Analytes 39
4-3 Grand Mean (XQ) and Percent Relative Standard Deviation (% RDS) of Field
Duplicate Pairs for Major ELS-II Analytes 40
4-4 Within-Batch and Among-Bateh Precision (% RDS) for Major ELS-II Analytes 41
4-5 Differences between Summer Split Samples Analyzed at the Two Analytical
Laboratories 43
5-1 Breakdown of the ELS-II Target Population 53
5-2 Physical Characteristics of the ELS-il Target Population 54
5-3 Population Estimates of the Number (N) and Percentage of Lakes with ANC and
pH below Reference Values in the Fall of 1984 (ELS-I) and 1986 (ELS-II) in the
Northeastern United States 56
5-4 Population Characteristics of Between-Year Chemical Changes (Fall 1984 - Fall
1986) in ELS-II Lakes 60
5-5 Population Distribution of the Standard Deviation of the Three Lake Samples
Collected during the Fall Index Period in the ELS-II Fall Variability Study 71
5-6 Population Estimates (N ± Standard Error) of the Number of ELS-II Lakes with
ANC (/«q/L) below Reference Values in the Three Seasonal Surveys 78
5-7 Population Estimates (N ± Standard Error) of the Number of ELS-II Lakes with
pH below Reference Values in the Three Seasonal Surveys 80
5-8 Population Characteristics of Between-Season Chemical Changes (Spring 1986 -
Fall 1986) in ELS-II Lakes 81
5-9 Regional Population Characteristics of Between-Season Chemical Changes
(Spring - Fall 1986) in Lakes with Fall 1986 ANC< 50 /ieq/L 82
5-10 Percentage of ELS-II Lakes with Inorganic Monomeric Aluminum (Aljm) Concen-
trations above Reference Values 95
IX
-------
6-1 Pooled Standard Deviation of Multiple Observations in ELS-Il for Duplicate
Samples (Analytical), between Lakes (Spatial), within Season (Fall 1986),
among Season (Spring, Winter, Fall 1986), and between Year (Fall 1984/
Fall 1986) 99
6-2 Regression Statistics for the Relationship between Spring Chemistry (Depen-
dent Variable) and Fall Chemistry (Independent Variable) in 1986 ELS-Il
Data 107
6-3 Percentage of Lakes with Chemical Conditions Exceeding Acid Stress Index
Values in the ELS-Il Target Lake Population 109
6-4 Spring Minus Fall Changes (A) in Major Anions and Cations in Low ANC ELS-Il
Lakes with Spring ANC Depressions (AANC) > 10 Meq/L 115
-------
EXECUTIVE SUMMARY
The Eastern Lake Survey - Phase II (ELS-II) was conducted in spring, summer, and tali of 1986 as
part of the U.S. Environmental Protection Agency's (EPA) National Surface Water Survey (NSWS), The
NSWS is a contribution to the National Acid Precipitation Assessment Program, which was charged by
the U.S. Congress to provide policymakers with sound technical information regarding the effects of
acidic deposition.
In Phase I of the Eastern Lake Survey (ELS-I), the acid-base status of lakes in the eastern United
States was quantitatively estimated using a statistical approach whereby a single lakewater sample was
collected in the fall (the fall index sample). Fall was selected as the index period in ELS-I because it is a
period of reasonable length (about 6 weeks) during which lakes are relatively well mixed and samples
from over 1,500 lakes could be collected. A similar well-mixed period in the spring can be very short. In
addition, chemical conditions vary less in fall than in spring, which was consistent with the objective of
ELS-I to assess chronic acidity rather than seasonal acidity.
The major objective of ELS-II was to assess the temporal variability in regional lakewater chemistry
with respect to acidic deposition effects. ELS-II had three major goals:
1. Assess the sampling error associated with the ELS-I fall index sample.
2. Estimate the number of lakes that are not acidic (ANC > 0) in the fall, but are acidic (ANC
< 0) in other seasons.
3. Establish seasonal water chemistry characteristics among lakes and relate the fall index
sample to seasonal and annual water chemistry patterns.
To accomplish these goals, water samples were collected from 145 statistically representative lakes in
the spring, summer, and fall of 1986. These three seasonal samples were used to assess among-season
variability, and to estimate between-year variability by comparing the fall 1986 ELS-II samples to the fall
1984 ELS-I samples. In 41 ELS-II lakes, two additional samples from independently selected locations in
the deepest part of the lake were collected during the fall index period to assess variability within index
periods and in site selection.
Lakes analyzed in ELS-II were chosen from those sampled in ELS-I using a variable probability
sample. Thus, ELS-II data, like ELS-I data, can be used to provide unbiased estimates of the status and
extent of acidic and low ANC lakes for the explicitly defined lake population from which the sample was
drawn. A number of restrictions were placed on ELS-I lakes included in the ELS-II target population in
order to concentrate sampling on the lakes of most interest with respect to acidic deposition effects.
Specifically, lakes with the following characteristics were jiot sampled in ELS-II:
X!
-------
1. Lakes with high acid neutralizing capacity (ANC > 400/ieq/L), because these lakes probably
will not be chronically affected at current rates of acidic deposition,
2. Lakes severely enriched with nutrients that would either (1) internally generate enough ANC
for the lake to be considered insensitive to acidic deposition, or (2) seriously distort the
natural chemical composition of the lake and confound data interpretation.
3, Shallow lakes (< 1.5 m deep).
4, Large lakes (> 20 km2) that exhibit considerable spatial variability in water chemistry and
present difficult logistic problems (e.g., Lake Champlain).
5. Lakes modified by (1) anthropogenic cultural disturbances (e.g., major wastewater treatment
plant discharge) or (2) recent in-lake management practices (e.g., liming) to such an extent
that the results would not be representative of other lakes in the population.
Lastly, only lakes in the northeastern United States (ELS-I Region 1) were sampled in ELS-II, in order to
focus resources on a region of high interest. Thus the subregions analyzed in ELS-II include the
Adirondacks, the Poconos/Catskills, northern New England, southern New England, and Maine.
The use and interpretation of any data set are restricted by the design, the quality of the data
obtained, and the sampling protocols, which are presented in detail in Sections 2, 3, and 4 of this report.
These aspects of the survey should be well understood before drawing conclusions both within and
beyond the scope of the original objectives. For example, the ELS-II spring sample was collected after
ice-out, 2 to 3 weeks after maximum lake discharge. Site intensive studies have shown that maximum
lakewater ANC depressions generally occur during peak discharge. Thus, one should not conclude that
the ELS-II estimates of acid-base status in the spring represent the maximum number of lakes that
become acidic during spring snowmelt.
RESULTS
Between-Year Variability
Changes in acid neutralizing capacity (ANC) and pH between fall 1984 and fall 1986 were small,
and estimates of the number of acidic and low pH lakes were very similar between the two years. For
example, within the ELS-II target population, there were 307 acidic lakes (ANC< 0; 8%) and 471 low pH
lakes (pH < 5.5; 12%) in fall 1984, versus 343 acidic lakes (9%) and 478 low pH lakes (12%) in fall 1986.
The median pH change in the ELS-II lake population between fall 1984 and fall 1986 was only 0.06 pH
units. Sulfate and DOC distributions were very similar in both years. Base cation and chloride concen-
trations tended to be higher in 1986 than in 1984, probably due to the drier conditions in 1986. Fall
nitrate and extractable aluminum concentrations were very low in most lakes in both years, so that
between- year differences were usually small. In lakes with elevated nitrate and extractable aluminum,
neither year had consistently higher concentrations than the other. A comparison between ELS-I and
xii
-------
ELS-II data shows that conclusions about the acid-base status of lakes in the northeastern United States
lakes would have been similar whether the assessment had been done in fall 1984 or fall 1986.
Within-Fall Season Variability: Robustness of the Fall Index
Chemical variability within the fall index period was very small. Population estimates of the
proportion of lakes with ANC or pH below reference values were not greatly affected by fall temporal
and site selection variability. Overall conclusions about the acid-base status of northeastern lakes would
have been the same if sampling had occurred during any of the three fall sample visits or at any of the
sampling locations. All observed concentrations of ANC, pH, sum of base cations, DOC, and inorganic
monomeric aluminum concentrations showed very little difference among the three fall visits. Mineral
acid anion concentrations (sulfate, nitrate, and chloride) also showed little difference among the three
lake visits, except at the highest observed concentrations, at which some within-season variability was
evident. The median standard deviation of the three lake visits was 6/^eq/L for ANC, 4 /zeq/L for
sulfate, 8/ieq/Lfor sum of base cations, and 0.05 pH units. For ANC, pH, and base cations, variability
in sample preparation and analysis was the major component of within season variability. Based on
data from the ELS-II Fall Variability Study, a single fall index sample is a robust estimator of conditions
during the fall index period.
Among-Season Variability
The ELS-II spring sample was collected during a period of spring ANC depression, generally within
two to three weeks after icemelt; however, it almost certainly does not represent the worst case
(minimum ANC) conditions. Based on intensive monitoring studies, the minimum lake outlet ANC and
pH and the maximum nitrate and inorganic monomeric aluminum concentrations usually occur around
peak lake discharge, while ice is still on the lake. Based on this observation, it would be expected that
the minimum ANC in ELS-li lakes would be lower than the observed values of the ELS-II spring samples,
which were collected 2 to 3 weeks after peak discharge. Thus, the ELS-li spring data do not indicate
worst case episodic spring conditions, but rather are indicative of post-snowmeit spring seasonal
conditions. The seasonal comparisons presented in the next three paragraphs need to be interpreted
with these facts in mind. In the following discussion, spring refers to post-snowmelt spring seasonal
conditions, not worst case spring episodic conditions.
For most chemical variables, among-season variability was greater than between-year or within-fall
variability. Chemical conditions in the spring, however, were positively correlated with fall index
conditions, although ANC, pH, DOC, and sum of base cation concentrations were typically lower in the
spring than in the summer and fall. Summer conditions for these ions were more similar to fall condi-
tions than spring conditions. Nitrate and inorganic monomeric aluminum concentrations were highest in
xiii
-------
the spring relative to summer and fall, although both were found at very low concentrations in the major-
ity of ELS-II lakes. ELS-II population estimates show that there were 24% more acidic lakes (ANC < 0) in
the spring (424 lakes) than in the fall or summer (343 lakes). This corresponds to an increase in the
percentage of acidic lakes from 9% to 11% in the ELS-II target population. Similarly, the percentage of
ELS-II lakes with ANC< 50/ieq/L increased from 25% to 34% (fall to spring) and the percentage with
ANC< 100peq/L increased from 43% to 62%. The number of low pH lakes was also higher in spring
than in fall. The percentage of ELS-II lakes with pH < 5.0 increased from 4% to 5%, the percentage of
lakes with pH < 5.5 increased from 12% to 14%, and the percentage of lakes with pH < 6.0 increased
from 19% to 27%.
The largest spring depressions in ANC (relative to fall ANC) were observed in the highest ANC
(200-400 pteq/L) ELS-II lakes. In lakes with ANC > 50 peq/L, spring ANC depressions (median ANC
depression = 60 jteq/L) were associated with decreases in base cation concentrations, probably due to
dilution by spring snowmelt runoff low in ANC and base cations. In low ANC (< 50Meq/L) lakes, spring
ANC depressions were small (median = S^eq/L, median pH decrease = 0.1 pH unit) and were associ-
ated with increases in nitrate in the Adirondacks and Poconos/Catskills. In all ELS-II regions, spring
ANC depressions in low ANC lakes were also associated with increasing spring inorganic monomeric
aluminum concentrations. Sulfate concentrations were very similar in the spring, summer, and fall
seasonal surveys.
Acid stress to fish was assessed using an acid stress index based on lakewater pH, inorganic
aluminum, and calcium concentration. The results showed that there was very little difference in the
number of lakes unsuitable for fish between spring, summer, and fall in the ELS-II lake population. Thus
it appears that the observed seasonal changes in pH, calcium, and aluminum were not great enough to
cause a significant seasonal change in the regional estimates of lakes with acid stress, as estimated by
the acid stress index. The main reason for this was that the largest ANC changes occurred in higher
ANC lakes that did not have a corresponding pH decrease into the range of stressful values. The low
ANC lakes that are most susceptible to acidic deposition effects did not have large changes in pH,
calcium, and inorganic aluminum and thus there were only small changes in estimated biological effects.
XIV
-------
SECTION 1
INTRODUCTION
1.1 OVERVIEW
The extent and magnitude of surface water acidification caused by atmospheric deposition has
been one of the most important and politically controversial environmental issues of recent times.
Research on individual lakes and streams suggests that pH and acid neutralizing capacity (ANC) have
declined over the past half century in some European and North American surface waters (Beamish et
al., 1975; Wright and Gjessing, 1976; Schofield, 1976). Atmospheric acidic deposition resulting from
fossil fuel combustion has been implicated as a cause of these declines (Schindler, 1988; Sharpe et a!.,
1984), though alternative hypotheses have been advanced and debated (Rosenqvist, 1978; Krug and
Frink, 1983). Regional paleolimnological data from the Adirondacks has shown that most lakes with
ANC < 50/^eq/L have declined in pH since industrialization and that there are three times as many
acidic lakes now as there were in preindustrial times (Sullivan et al., 1990). The timing of recent lake
acidification can best be explained by the onset of acidic deposition (Sullivan, 1990).
A prerequisite for a regional-scale understanding of the effects of acidic deposition on surface
waters is knowledge of the present chemical status of surface waters over broad regional areas. To
accomplish this goal, the U.S. Environmental Protection Agency (EPA) began the National Surface Water
Survey (NSWS) in 1983. The objective of the NSWS was to quantify the physical and chemical charac-
teristics of lake and stream populations within acid-sensitive geographic regions of the United States. In
Phase I of the NSWS, surface waters were selected in a systematic random fashion from a statistical
frame to allow unbiased estimates of the status and extent of acidic and low ANC systems. Phase I of
the NSWS determined that about 6% of the lakes > 4 ha in size in the northeastern United States are
acidic, and most of them have acid anion compositions dominated by sulfate from atmospheric depo-
sition (L Baker et al., 1990; Linthurst et al., 1985). The conclusions from Phase I of the NSWS are
based on the premise that one (for lakes) or two (for streams) samples can be used as an index to
characterize the present chemical status of surface waters. For lakes, a single fall epilimnetic sample
was used to provide an index of lake chemical status and to identify acidic and potentially sensitive lakes
(Linthurst et al., 1985). The fall index data alone, however, do not provide information on how these
descriptions of lake populations might differ if temporal or spatial variability in individual lakewater
chemistry had been addressed. In addition, a major criticism of the lake surveys was that a single water
sample is insufficient for characterizing a lake, and therefore is insufficient for characterizing lake
populations. The primary restriction of Phase I of the NSWS was that questions of the effects of
temporal variability of lake chemistry could not be addressed.
Phase II of the NSWS was designed to supplement the data and results from Phase I by providing
information on the temporal variability of surface water chemistry. This report presents the results of the
1
-------
spring, summer, and fall 1986 lakewater chemistry surveys conducted in the northeastern United States
as part of Phase II of the Eastern Lake Survey (ELS-II).
The ELS-II was designed to chemically and physically characterize a probability based subsample
of Phase I lakes during the spring, summer, and fall seasons. The primary objectives of ELS-II were to:
• Assess the sampling error associated with the ELS-I fall index sample.
» Estimate the number of low ANC (potentially susceptible) lakes not acidic in the fall that are
acidic in other seasons, emphasizing spring variability in water chemistry.
• Establish seasonal water chemistry characteristics among lakes and relate the fall index
sample to seasonal and annual water chemistry patterns.
Ideally, it would have been desirable to study lakes in all of the ELS-I regions. Logistical and resource
constraints, however, dictated that only one region could be analyzed. The Northeast Region (ELS
Region 1) was chosen because the Phase I data indicated that it was the ELS-I region most impacted by
acidic deposition.
1.2 THE NATIONAL SURFACE WATER SURVEY
In response to the need for knowledge regarding the present extent of acidic or potentially
susceptible aquatic resources and their associated biota, the U.S. EPA and cooperating scientists were
asked in 1983 to design a program that would achieve five major goals:
1. Characterize the chemistry of surface waters (both lakes and streams) in regions of the
United States presently believed to be potentially susceptible to change as a result of acid
deposition.
2. Examine associations among chemical constituents and define important factors that may
affect surface water chemistry.
3. Determine the biological resources within these systems.
4. Evaluate correlations among surface water chemistry and the corresponding biological
resources.
5. Quantify regional trends in surface water chemistry and biological resources.
The program designed to meet these goals was designated the National Surface Water Survey
(NSWS). The NSWS became an integral part of the National Acid Precipitation Assessment Program
(NAPAP), an interagency research, monitoring, and assessment effort mandated by Congress in 1980.
NAPAP provides policymakers with technical information concerning the extent and severity of the
effects of acid deposition on human, terrestrial, aquatic, and material resources.
-------
The NSWS design (Figure 1-1) incorporates two parallel components, the National Lake Survey
(NLS) and the National Stream Survey (NSS), in order to satisfy the five major research goals. In both
components, early project phases contributed to the design and interpretation of subsequent phases.
The synoptic surveys of lake and stream chemistry performed in the early phases of the NLS and NSS
contribute substantially to the design and interpretation of subsequent project phases and are essential
to the regional extrapolation of their results.
The NSWS design grew out of the recognition that although it is clearly not feasible to perform
intensive, process-oriented studies or monitoring programs on all surface waters within the United
States, it is equally inappropriate to study a few systems that later may be found to have atypical
biological and chemical characteristics. Therefore, each component of the NSWS begins with Phase I, a
synoptic survey designed to characterize and quantify the chemistry of lakes and streams throughout
areas of the United States expected to contain the majority of low ANC waters. In the NSWS, lakes and
streams were sampled on a regional basis using a statistically rigorous survey design, appropriate
standardized and documented field and analytical techniques, a relatively complete set of chemical and
physical measurements, and a quality assurance/quality control (QA/QC) program to explicitly define
uncertainty in the resulting data. The Initial survey component (Phase I) provides a snapshot of the
present condition of surface waters in regions of the United States most likely to be affected by acidic
deposition. The Phase I data also serves as a basis for classification of lakes and streams. Such
classifications allow the regional extrapolation, with known confidence, of results from past and future
intensive studies on both high-interest aquatic subpopulations and individual lakes and streams.
In the second phase of the National Lake Survey, the Eastern Lake Survey - Phase II (ELS-ll), and
in the Episodic Response Project (ERP), short-term (seasonal, weekly, or episodic) variability in water
chemistry is quantified within and among lakes and streams of most interest with respect to acidic
deposition effects. These surface waters were defined on the basis of Phase I water chemistry and
associated hydrology, aquatic organisms, regional acid deposition inputs, land use, physiography, and
other basin characteristics.
1.3 EASTERN LAKE SURVEY - PHASE II
The major component of ELS-ll was the spring, summer, and fall seasonal surveys and Fall
Variability Study of lakewater chemistry in the northeastern United States. ELS-ll lakes were sampled
once in the spring, summer, and fall at the same location on the lake where the ELS-I sample was
collected. In the fall variability study, a subset of ELS-ll lakes was sampled on two additional dates at
two independently selected locations believed to be the deepest point in the lake. ELS-ll data, in
conjunction with ELS-I data can be used to assess between-year, within-season, and among-season
chemical variability, as well as spatial variability due to site selection.
-------
NATIONAL SURFACE WATER SURVEY (NSWS)
NATIONAL LAKE SURVEY (NLS)
SYNOPTIC CHEMISTRY
Eastern Lake Survey- I (1984)
Western Lake Survey-I (1985)
EASTERN LAKE SURVEY - H
Temporal Variability in
Northeast (1986-87)
Biological Resources in
Upper Midwest (1986)
NATIONAL STREAM SURVEY (NSS)
SYNOPTIC CHEMISTRY
Phi Pilot (S. Blue Ridge) (1985)
Phi Mid-Atlantic and
Southeast (1986)
Episodic Effects (1988)
Biological Resources (1988)
Long-Term Monitoring
Figure 1-1. Organization of the National Surface Water Survey, showing the two major compo-
nents, the National Lake Survey and the National Stream Survey, and their relation-
ship to later phases of study. Dates in parentheses are years of field data collection.
-------
In addition to the northeastern chemistry surveys, biological sampling was also conducted in lakes
of the eastern Upper Midwest, Data from this study are not discussed in this report. Biological data
from the Upper Midwest sampling effort is discussed in Cusimano et al. (1990) and J. Baker et al.
(1990a).
1.4 THE ELS-II DATA REPORT
This ELS-II data report discusses only data from the seasonal surveys of lakewater chemistry in
northeastern lakes. Section 2 discusses the statistical design of ELS-II and Section 3 describes the field
and laboratory methodologies. Section 4 details the results of the quality assurance (QA) program.
Section 5 presents the major results and analyses of the three main components of temporal variability
studied in ELS-II: between-year variability, within-season variability, and among-season variability.
Between-year variability was addressed by comparing the results from ELS-l (fall 1984) with those
measured in the ELS-II fall seasonal survey (fall 1986). Among-season variability was analyzed by
comparing the spring, summer, and fall ELS-II seasonal surveys. Within-season variability was
addressed in the ELS-II fall variability study, which sampled a subset of ELS-II lakes at three different
times at independently selected locations on the lakes within the fall index period. The last section (6) of
the report is a synthesis and discussion of the relative importance of the various components of
temporal variability.
-------
SECTION 2
ELS-ll SURVEY DESIGN
2.1 OVERVIEW
Phase II of the Eastern Lake Survey (ELS-ll) was designed to assess temporal variability in
regional lakewater chemistry. The ELS-ll is based on a probability sample of 160 lakes from an explicitly
defined subpopulation of lakes sampled in Phase I of the Eastern Lake Survey (ELS-I). As in ELS-I,
ELS-ll samples are weighted proportionally to the number of lakes they represent in the target popula-
tion, so that conditions may be inferred for the target population as a whole, not just the sampled lakes.
Only lakes from the Northeast Region (Region 1) of the ELS-I were sampled in ELS-ll,
2.2 EASTERN LAKE SURVEY - PHASE I
As the lakes selected for ELS-ll are a subset of the lakes sampled in ELS-I, a short background on
the design of ELS-I is warranted. A more detailed discussion of the ELS-! design can be found in
Linthurst et al. (1986) and Landers et al. (1988). There were three stratification factors in the ELS-I
design; region (Northeast, Upper Midwest, or Southeast), subregion, and ANC map class (< 100,
100-200 or > 200 /xeq/L). A total of 33 separate strata were coded by region, subregion, and ANC map
class. For example, 1A2 designates Region 1 (Northeast), subregion A (Adirondacks), and map class 2
(expected ANC class of 100-200 /xeq/L). Only the Northeast Region (Region 1) was sampled for
chemical variability in ELS-ll. Within the Northeast Region (Figure 2-1), there were five subregions:
Adirondacks (1A), Poconos/Catskills (1B), Central New England (1C), Southern New England (1D), and
Maine (1E).
Strata boundaries were overlain on 1:250,000-scale USGS maps and a list was compiled of all
lakes on the map in each strata. This list of lakes is known as the statistical frame population and
represents the universe of lakes considered for study in the ELS-I. All population estimates computed in
ELS-I, and thus those in ELS-ll, refer to the map frame population and do not represent conditions in
lakes outside the area of coverage or in systems not depicted on the USGS maps used. The main limi-
tation of the map population is that lakes smaller than 4 ha are not part of the ELS population because
they are not generally depicted on 1:250,000-scale maps. A comparison of the number of ELS-I target
and nontarget lakes and the total number of lakes in the northeastern United States can be found in
Johnson et al. (1989).
The sampling plan for ELS-I employed a stratified random design, with equal allocation of number
of sample lakes to strata. Lakes were selected from each stratum by systematic sampling from an
ordered list following a random start. In general, 50 lakes per stratum were selected for sampling. A
target population of lakes was defined by excluding lakes with noninterest attributes (e.g., heavy
-------
LAKE SURVEY - PHASE II
Sur»«y Boundaries
Phis* II Sumy Sites
• Cluter I
a Cluster II
* CluUr III
MAINE (1C)
CENTRAL NEI EN6LAKB (1C)
SOUTHERN K£I ENSUND (ID)
! MCQHOS/eATSKIUS (18)—-
Figure 2-1. ELS-II sample lakes in the five subregions of the northeastern United States. Clusters
were delineated on the basis of 1984 fall index ANC in ELS-! (Section 2.3.2). Cluster I
had ANC< 25 peq/L, Cluster II had ANC between 25 and 100 /«q/L, and Cluster
had ANC between 100 and 400
-------
anthropogenic activity, no lake present, marsh/swamp) based on larger scale map examination and field
visitation. Each sampled lake was assigned a sample weight equivalent to the number of lakes it
represents in the target population. In ELS-I, all the lakes in a stratum have the same weight, equivalent
to N/n***, where N is the estimated total number of target lakes in the frame population in the stratum
and n*** is the number of target lakes sampled in the stratum. For example, if 1,000 target lakes are
estimated to be in a stratum and 50 target lakes were sampled, then each sample lake represents 20
lakes in the target population and has a weight of 20. In total, there are an estimated 7,157 ELS-I target
lakes in the Northeast Region out of a frame population of 10,758 lakes.
2,3 EASTERN LAKE SURVEY - PHASE II
ELS-ll was designed to sample a subset of the ELS-I lakes to estimate the number of lakes not
acidic in the fall that are acidic in other seasons and to assess the sampling error associated with the
ELS-I fall index sample. In addition, ELS-ll was designed to establish the seasonal characteristics
among lakes thought to be in an acid-sensitive class of lakes, in order to aid in detecting trends and
evaluating episodic acidification.
2.3.1 Defining the Population of Interest
To better characterize the population of interest, a group of lakes of low interest to the goals of
ELS-ll was defined. These low-interest lakes were not subsampled for inclusion in the ELS-ll population
and are not represented by any of the data discussed in this report. Lakes of low interest were defined
as having any of the following characteristics (Thornton et al., 1986):
• ANC levels such that a lake probably would not become acidic at current rates of deposition
(termed capacity protected systems). High ANC lakes were defined as those having ELS-I
index chemistry > 400 p. eq/L.
• Highly enriched by nutrients that would either (1) internally generate enough ANC for a lake
to be considered capacity protected, or (2) seriously distort the chemical composition of the
lake and confound data interpretation. Nutrient enriched lakes were defined as those having
either total phosphorous > 90/ig/L, NO3" > 50/aeq/L, NH4+ > 30Meq/L, turbidity > 7 NTU,
or Secchi disk depth < 0.5 m, based on ELS-I index chemistry.
• Shallow lakes (defined as those with ELS-I site depth < 1.5 m). At the time of ELS-ll site
selection, it was believed that shallow lakes did not contain a significant fishery resource. In
hindsight, this does not appear to have been a useful exclusion. Shallow lakes do support a
fishery resource.
• Lakes so large that they exhibit considerable spatial variability in water chemistry and present
difficult logistic problems (e.g., Lake Champlain or the Finger Lakes of central New York).
Large lakes were defined as those with surface areas > 2,000 ha (20 km ).
-------
Modified by anthropogenic cultural disturbances to such an extent that the results would not
be representative of other lakes in the population (e.g., major wastewater treatment plant
discharge into the lake).
Modified by recent in-lake management practices resulting in a serious disturbance of either
the biota or the lake chemistry (e.g., rotenone treatment, removal of a dam, liming).
The first four factors were used to exclude ELS-I lakes from inclusion in the ELS-II sample. Factors 5
and 6 were evaluated after site selection on an individual lake basis to further refine the ELS-II population
of interest.
2.3.2 Statistical Sampling Design
The ELS-II is a double, or two-phase, sampling design. The first phase was the ELS-I sampling
and the second phase was the ELS-II subsampling of lakes selected from the ELS-I sampled lakes.
Logistical considerations for ELS-II limited the number of lakes that could be adequately analyzed to a
total of about 150. Based on ELS-I experience, statistical precision requirements for making population
estimates (proportion of lakes with values below reference values) indicated that a sample size of about
50 lakes was desirable for a stratum (Linthurst et al., 1986). Thus the ELS-II design was based on a
sampling of 50 lakes from each of three clusters of ELS-I lakes. Clusters were chosen by analyzing data
from all 15 ELS-I strata in the Northeast (Region 1) using cluster analysis to identify meaningful
subgroups, on the basis of lake chemistry or other attributes. Lakes of low interest (as defined in
Section 2.3.1) were assigned to a reserved fourth cluster and not used in site selection. The results
indicated that a univariate criterion based on ANC coincided closely with the results of cluster analysis
using a variety of variables, including pH, ANC, DOC, color, sulfate, methyl-isobutyl-ketone (MIBK)
extractable aluminum, and base cations (Thornton et al., 1986). Thus, the three clusters used as
stratification factors in the ELS-II design were:
Cluster I. ANC < 25 /i eq/L
Cluster II. 25 < ANC< 10Q/ueq/L
Cluster III. 100 < ANC< 400/ieq/L
The ELS-II subsampling process employed a fixed-size systematic variable probability sample
(Overton, 1987). Selection was with probability proportional to the ELS-I sample weight (inversely pro-
portional to the ELS-I inclusion probability), with the result that the total ELS-II sample weights within the
three clusters are nearly equal. The final ELS-li sample weight for lake i, W2j, is equal to (equation 1):
.WI, (1)
-------
where Wcond f is the ELS-II conditional inclusion weight and W1S is the ELS-I sample weight, Wcond. Is
calculated for each cluster in each strata from equation 2:
wcond J = Tw / (W1, • n) (2)
where TW is the estimated number of mapped target lakes (the sum of the ELS-I sample weights) in the
cluster and n is the number of lakes to be sampled in the cluster.
In selection, clusters were treated as stratification factors. Prior to selection, the ELS-I sample
lakes in each cluster were sorted by subregion (major) and site depth (minor). Sorting by these two
factors increased the probability of obtaining representative spatial coverage of lakes throughout the
region and varied lake types. Measured ANC was the primary basis of the clusters so that all three of
these factors were thus controlled in lake selection. After a random start, ELS-II sample lakes were then
chosen from the sorted list of ELS-I lakes at equal intervals (i.e., every 10th lake, or every 23rd lake,
depending on the desired sample size).
Three minor difficulties arose. First, five lakes in cluster I and two lakes in cluster II had ELS-I
weights (W1j) > Tw/n and had to be sampled with Wcond. equal to 1, otherwise the Wcor)d. calculated
from equation 2 would be < 1 (implying an inclusion probability > 1). Thus, these seven ELS-I lakes
were included with certainty in the ELS-II sample. Those samples selected with a Wcond. of 1 had an
ELS-II sample weight the same as the ELS-I sample weight (see equation 1). Sampling these lakes with
certainty affected the Wcond. in clusters I and II, because n and Tw change. Tw is decreased by the sum
of the ELS-I weights sampled with certainty in each cluster and n is decreased by five lakes in cluster I
and two lakes in cluster II.
A second difficulty is that it is awkward to construct an expanding sample using this scheme. That
is, if the sample size were set to 50 and it was necessary to add two lakes to bring the number to 52,
and keep the same statistical properties of the resultant sample, the required procedure would be com-
plex. It is easier to draw a larger sample, and reduce it while retaining the desired statistical properties.
The lakes eliminated by the reduction are thus available for expansion, if necessary. In ELS-II, it was
decided to select 60 lakes in clusters II and III to provide for 10 alternate lakes in each of these two
clusters. The alternates were needed in cluster II because of post-selection exclusions (see Section
2.4.1), but were deemed unnecessary in cluster III. Therefore, 160 sample lakes represent the initial
ELS-II target population: 50 from cluster I, 60 from cluster II, and 50 from cluster III.
A final complication was that after selection of the ELS-II sample, small revisions were made to the
ELS-I weights in the 1B2, 1D2, 1D3, 1E1, and 1E2 strata because some of the ELS-I lakes chosen for
sampling were not actually sampled. Wcond. is fixed at the time of site selection and does not change,
but the change in the ELS-I weight (Wlf) does change the final ELS-II sample weight (see equation 1).
The resulting changes were fairly small, but the final ELS-II sample weights are not quite uniform in each
cluster as they would have been if the ELS-I weights had not been revised.
10
-------
2.4 ELS-H TARGET POPULATION
Of the estimated 7,157 target ELS-I lakes, an estimated 4,426 are represented by the 160 ELS-II
lake samples (Table 2-1). The remaining 2,731 lakes were of low interest to ELS-II for the reasons
described in Section 2.3,1 and enumerated in Table 2-1. The major reasons for a low interest designa-
tion were high ANC and shallow depth. The sample lakes in the initial ELS-II target population were then
examined in more detail to check for low interest attributes.
2.4.1 Refinement of the Target Population
As discussed in Section 2.3.1, lakes that were managed or modified by cultural disturbances were
considered of low interest for ELS-II because their inclusion would interfere with an assessment of
natural chemical variability. After selection, the 160 study lakes were evaluated on an individual basis for
evidence of perturbance by consulting external data sources (state files, liming records, etc.). As a
result of this analysis, 147 lakes were targeted for field visitation and 13 lakes were eliminated from the
population of interest. Seven of the 13 excluded lakes were removed because they had been limed and
were being managed, 5 were removed due to point sources of pollution, and 1 was removed because it
was a bog.
After field visitation, an additional lake (1B3-025) was removed from the population of interest.
This lake was managed by a private club that denied permission for access to the lake. This lake was
classified as of low interest due to management. In addition, one lake (1A2-058) in cluster III that could
not be sampled in the fall due to inclement weather was sampled in the spring and summer. Rather
than have unequal sample sizes and thus different weights in the different seasonal data sets, it was
decided that this lake would be considered a random miss in all of the seasons. Data from this lake for
spring and summer were not used to make population estimates. W2j and Wcond j were adjusted for the
random miss by decreasing the sample size, n, in cluster III by 1 and recalculating the weights using
equations 1 and 2.
In sum, 145 of the 160 lakes in the initial ELS-II population were used to make population esti-
mates. Of the 15 eliminated lakes, 14 were classified as low interest and one was a random miss.
Extrapolating to the population, about 10% (an estimated 443 lakes) of the 4,426 lakes in the initial ELS-II
population were excluded for low interest attributes. A complete breakdown of the ELS-I and ELS-II
target populations is presented in Table 2-1. The refined ELS-II target population with which population
estimates are made is based on a sample of 145 lakes representing a population of 3,993 mapped lakes.
The location and cluster of the 145 ELS-II sample lakes is shown in Figure 2-1. The distribution of ELS-II
lakes in the Northeast is patchy due to either a low density of lakes greater than 4 ha in some regions
(e.g. Vermont), the prevalence of high ANC lakes in some areas (e.g. the St. Lawrence Lowlands), or
random chance.
11
-------
Table 2-1. Estimated Population and Sample Sizes of Northeastern Lakes (Region 1) in ELS-1
and ELS-li
Sample Size
Population Size
ELS - Phase I
Map Frame Population
ELS-I Target Population
ELS-II Selection Exclusions
High ANC Lakes (> 400 jieq/L)
Culturally Enriched Lakes
Shallow Lakes (< 1.5 m)
Big Lakes (> 2000 ha)
Exclusions Subtotal
Potential ELS-II Lakes
768
114
24
119
5
262
506
10,758
7,157
1,409
264
997
61
2,731
4,426
ELS - Phase II
Initial Target Population
Post Selection Exclusions
Managed Lakes
Polluted Lakes
Bog Lakes
Random Miss
160
8
5
1
1
4,426
245
139
49
Qa
Exclusions Subtotal
15
433
Refined Target Population
145
3,993
Sample weights were adjusted to account for the random miss so that there was no change to the population size.
All ELS-II population estimates presented in this report are based on the refined target population.
12
-------
2.4.2 Making ELS-ll Population Estimates
Population estimates in ELS-II were made in much the same manner as those in other surveys of
the NSWS (Linthurst et al., 1986; Landers et al., 1987; Kaufmann et al, 1988). Details of estimation,
along with the statistical foundation of the methods, are provided elsewhere (Overton, 1987; Blick et ai.,
1987), The general form of estimator:
T = XW2jy (3)
S
where:
T is the estimate of the total of any attribute, y, over the population.
y is any lake attribute of interest (e.g., number or area), known over the sample, S.
I indicates summation over the appropriate sample of target reaches, whether the full sample, a
S subsample, or mixed cluster sample.
W2j is the ELS-II sample weight assigned to the lake in making population estimates (see Appendix B),
This estimator is similar to the Horvitz-Thompson (1952) estimator for variable-probability samples, but it
is not exactly the same due to the two-phase nature of the sampling (Sarndal and Swensson, 1987). By
assigning different definitions to y, and by summing over different sets of sample units, S, the various
attributes of the target population of lakes can be estimated from this one equation. Specifically, in the
ELS-II, two attributes are identified as parameters of the resource of interest:
1. Total number of target lakes (N), y = 1: N = ZW2j
2. Total area of target lakes (A), y = A: A = IW2jA
where A is the surface area of the lake.
Specific subpopulations are assessed by combining the samples from those subpopulations, and
in general it is necessary only for the sample subset to be defined in exactly the same way as the
estimated population. For example, the number of ELS-II target lakes in the Adirondack Mountains with
pH < 6.0 is estimated by summing the sample weights of all Adirondack Mountain lakes with pH < 6.0
in the ELS-II sample. Similarly, the area of ELS-II target lakes in Maine with elevation > 500 m is
calculated by summing the product of the weight and lake area for all ELS-II sample lakes in Maine with
elevation > 500 m.
13
-------
Estimates of the variance of estimated target population totals were calculated by using appro-
priate variable-probability variance estimators adapted for the two-phase nature of the sampling
(Overton, 1987). The variance estimator again closely resembles the Horvitz-Thompson variance esti-
mator, but it is not exactly the same because of the two-phase nature of the sampling. Additional
complications due to the systematic nature of the sampling are addressed by Stehman and Overton
(1987). The formula for estimated variance is:
V(fy) = I y2W2j(W2j - 1) + I I y,y,(W2m - W ) (4)
S ie S ]c S
where W8 = A-EL is the inverse of the pairwise inclusion probability (see Appendix A for details). The
second term in equation 4 represents a pairwise comparison of every possible i,j pair in the population
of interest. The estimated standard error (SE) of the estimated population total is then calculated as the
square root of the variance estimate, V(T ).
The primary outputs of these population estimates are descriptions of the various distributions of
the chemical variables. Distributions of chemistry within any subset of the target population were
analyzed in the same way. That is, an estimate was made of the number of lakes in the subpopulation
having a value of the variable less than or equal to a particular value.
N(x) = I W2j (5)
X
-------
Other population statistics of interest can be generated from the distributions. Each distribution
has identified quantiles, for example the median and the two quartiles, Qt, and Q3. The median of any
variable is the value of x such that F(x) = 0.5. The first quartile of any variable, Qv is the value of x
such that F(x) = 0.25. These statistics can be defined for all distributions. Additionally, from the
frequency distribution, F(x), the mean and standard deviation of the variable x on the population Is
estimated.
Mean(x) = IV^x/ZW^ (7)
SD(x) = (EW^x2 /ZW2, - [IW^x /IW2j]2)a5 (8)
2.4.3 Comparison to DDRP Population
Northeastern lakes studied in the EPA's Direct/Delayed Response Project (DDRP) were selected
concurrently and with the same statistical design as ELS-II (Church et al., 1989). The DDRP, however,
had additional site restrictions (e.g., watershed areas had to be < 3,000 ha) so that there is not a
complete overlap between the DDRP and ELS-II sites. In addition, some of the lakes that were dropped
from ELS-II after site selection, during the target population refinement (Section 2.4.1), were analyzed in
the DDRP. Overall, both surveys sampled 145 target lakes but the DDRP had an estimated target popu-
lation of 3,668 lakes, whereas ELS-II had an estimated target population of 3,993 lakes. There were 118
sample lakes representing an estimated 3,078 lakes in the target population common to both surveys
(77% of the DDRP population was in the refined ELS-II population).
15
-------
SECTION 3
METHODS
3.1 OVERVIEW
This section discusses the methods employed in Phase II of the Eastern Lake Survey (ELS-II) for
sample collection, handling, and analysis. It also describes the quality assurance and quality control
(QA/QC) protocols that were implemented, including the procedures for data management (verification,
validation, and construction of the final data bases used for analysis).
Much of the ELS-II methodology was shared with other components of the National Surface Water
Survey (NSWS): Phase I of the Eastern (Linthurst et al., 1986) and Western (Landers et al., 1987) Lake
Surveys and the National Stream Survey (Kaufmann et al., 1988). The QA/QC sample collection design,
data management protocol, and processing laboratory activity for ELS-II were similar to the procedures
implemented in the other surface water surveys.
The chemical and physical characteristics of lake water that were analyzed for all ELS-II samples
are listed in Table 3-1. A description of each parameter is given in the ELS-II analytical methods manual
(Kerfoot et al., 1988). Once collected, water samples were transported via express courier to a central
processing laboratory where they were preserved and split into aliquots within 36 hours after sample
collection. In addition to preparing the samples for shipment to the contract analytical laboratory, the
processing laboratory measured pH, dissolved inorganic carbon (DIG), color, turbidity, specific conduc-
tance, and certain aluminum species. At the analytical laboratory, 24 major chemical variables were
measured (Table 3-1). Data from the analytical laboratory were entered into a data base, which then
underwent a series of QA checks.
3.2 FIELD SAMPLING PLAN
A total of 147 statistically selected lakes were targeted for field visitation during ELS-II (as
described in Section 2; however, only 145 lakes are in the refined target population) during the spring,
summer, and fall of 1986. In addition to these seasonal surveys, a variability survey was conducted
during the Fall Seasonal Survey. The Fall Variability Survey was designed to sample a subset of 50 of
the ELS-II lakes at three different times at independently selected locations believed to be the deepest
points in the lakes during the fall index period.
In situ measurements of pH, specific conductance, temperature, and dissolved oxygen (Table 3-1)
were made at 1.5 m below the surface and 1.5 m above the bottom in all lakes. If the lake was stratified
(temperature difference > f C), depth profiles of the in situ measurements were made. Water samples
for laboratory analyses were collected from the epilimnion at 1.5 m below the surface in lakes more than
16
-------
Table 3-1, Chemical and Physical Variables Measured in ELS-II and Methods Employed
Parameters
Instrument or
analytical methods
Reference
laboratory
methods3
FIELD SITE
pH, in situ
Specific conductance
and temperature,
in situ
Dissolved oxygen,
in situ
Secchi Disk Transparency
Potentiometer Hydrolab
(Surveyor II)
Conductivity cell and Thermistor
Hydrolab (Surveyor II)
Oxygen probe Hydrolab
(Surveyor II)
Secchi disk
Merritt and Sheppe (1988)
Merritt and Sheppe (1988)
Merritt and Sheppe (1988)
Merritt and Sheppe (1988)
PROCESSING LABORATORY
Aluminum
Total monomeric
Nonexchangeable
monomeric
Specific conductance
pH, closed system
Dissolved inorganic carbon,
closed system
True color
Turbidity
Colorimetry (pyrocatechol violet,
automated flow injection
analyzer)
Colorimetry as with total mono-
meric (after passing through
strong cation-exchange column)
YSI conductivity meter (Model 32);
YSIcell (YSI 3417)
pH meter (Orion Model 611);
glass combination electrode
(Orion Model 8104)
Infrared spectrophotometry
(Dohrmann DC-80 carbon
analyzer)
Comparator (Hach Model CO-1)
Nephelometer (Monitek Model 21)
Arentetal. (1988)
Arent etal. (1988)
EPA 120.1
EPA 150.1
EPA 415.2 (modified)
EPA 110.2 (modified)
EPA 180.1
EPA methods are taken from U.S. EPA (1983); USGS methods are from Skougstad et al. (1979).
(Continued)
17
-------
Table 3-1. Chemical and Physical Variables Measured in ELS-II and Methods Employed
(Continued)
Parameters
Instrument or
analytical methods
Reference
laboratory
methods3
CONTRACT ANALYTICAL LABORATORY
Acid neutralizing
capacity (ANC)
Aluminum
Extractable
Total
Ammonium
Base neutralizing
capacity (BNC)
Calcium
Chloride
Dissolved inorganic
carbon (DIG)
Initial
Air equilibrated
Dissolved organic
carbon (DOC)
Fluoride, total dissolved
Iron
Acidimetric titration,
modified Gran analysis
Atomic absorption spectroscopy
(furnace) on methyl-isobutyl-
ketone extract
Atomic absorption spectroscopy
(furnace)
Colorimetry (phenate, automated)
Alkalimetric titration, modified
Gran analysis
Atomic absorption spectroscopy
(flame)
Ion chromatography
Infrared spectrophotometry
Infrared spectrophotometry,
after bubbling with 300
ppm C02 air for 20 minutes
Infrared spectrophotometry,
after acidification and
sparging to remove DIG
Ion-specific electrode
Atomic absorption spectroscopy
(flame)
Hillman et al. (1987);
Kramer (1984)
Hillman et al. (1987);
EPA 202.2
EPA 202.2
EPA 350.1
Hillman et al. (1987);
Kramer (1984)
EPA 215.1
ASTM (1984); O'Dell
etal. (1984)
EPA 415.2 (modified)
EPA 415.2 (modified)
EPA 415.2
EPA 340.2 (modified)
EPA 236.1
EPA methods are taken from U.S. EPA (1983); USGS methods are from Skougstad et al. (1979).
(Continued)
18
-------
Table 3-1. Chemical and Physical Variables Measured in ELS-II and Methods Employed
(Continued)
Parameters
Instrument or
analytical methods
Reference
laboratory
methods3
Magnesium
Manganese
Nitrate
pH
Air equilibrated
Initial ANC
Initial BNC
Phosphorus, total
Potassium
Silica (SiO2)
Sodium
Specific conductance
Sulfate
Atomic absorption spectroscopy
(flame)
Atomic absorption spectroscopy
(flame)
Ion chromatography
pH electrode and meter, after
bubbling with 300 ppm CO2
air for 20 minutes
pH electrode and meter, at
start of ANC titration
pH electrode and meter, at
start of BNC titration
Colorimetry (phosphomolybdate
automated)
Atomic absorption spectroscopy
(flame)
Colorimetry (silicomolybdate,
automated)
Atomic absorption spectroscopy
Conductivity cell and meter
Ion chromatography
EPA 242.1
EPA 243.1
ASTM (1984); O'Dell
etal. (1984)
EPA 150.1
EPA 150.1
EPA 150.1
USGS I-4600-78
(modified)
EPA 258.1
USGS 1-2700-78
EPA 273.1
EPA 120.1
ASTM (1984);
O'Dell et al. (1984)
EPA methods are taken from U.S. EPA (1983); USGS methods are from Skougstad et al. (1979).
19
-------
3 m deep. In lakes < 3 m deep, epilimnetic samples were collected at 0.5 m below the surface. ELS-II
field operations are discussed in detail by Merritt and Sheppe (1988).
3.2.1 Spring Seasonal Survey
ELS-II lakes in the Northeast were sampled once in the spring of 1986 in the epilimnion at the
same location in the lake as the fall index sample in ELS-I (Figure 3-1). Lakes were sampled as soon
after iceout as was practical. Water samples were collected from 146 lakes between March 25 and May
3, 1986. The relationship between the ELS-II sampling window and lake outlet discharge in ELS-II
subregions is discussed in Section 6.3.3.
3.2.2 Summer Seasonal Survey
ELS-II Lakes were sampled once in the epilimnion in the summer of 1986 at the same location in
the lake as the fall index sample in ELS-I (Figure 3-2). Water samples were collected from 147 lakes. In
addition to the epilimnetic sample, a hypolimnetic sample was collected in 123 of these lakes. Hypo-
limnetic samples were drawn from the middle of the hypolimnion in stratified takes, and at 1.5 m from
the bottom in nonstratified lakes and in lakes < 5 m deep. Hypolimnetic samples were not collected
from lakes < 3 m deep. Summer seasonal survey samples were collected between July 23 and August
11, 1986. A number of special studies were also conducted during the summer seasonal survey (e.g.,
zooplankton analyses, chlorophyll-a; see Figure 3-2). Data from these studies, as well as the hypo-
limnetic data, are not discussed in this report. Details of the zooplankton study can be found in Tessier
and Horwitz (1988, 1990).
3.2.3 Fall Survey and Variability Study
A variability study was conducted in fall 1986 along with the regular seasonal survey to assess the
within-season and within-lake spatial variability in index chemistry (Figure 3-3). In addition to the fall
seasonal survey visits at the ELS-I sample locations, a subset of 50 ELS-II lakes, in two sets of 25, was
selected for sampling at two additional times during the fall index period at two independently selected
locations in each lake. The Fall Variability Study sampling locations were chosen at the time of sample
visits (by independent field crews) by locating a spot that appeared to be the deepest part of a lake,
according to lake shape, surrounding topography, and depth measurements, in the same manner as the
original ELS-l sites were chosen. A set of 25 fall variability lakes was randomly selected from all the
ELS-II lakes in each of two geographic regions of the northeastern United States: the Adirondacks, and
southeastern New England (Connecticut, Rhode Island, Massachusetts, and southern New Hampshire).
20
-------
SPRING SEASONAL
FALL INDEX SITE
"» SPECIAL STUDY SAMPLE
j=ALIQUOTS DRAWN FROM
! ! LAKE SAMPLE
C\ _ POINT OF SAMPLE
CJ "COLLECTION
LAKi WATER MEASUREMENTS
• IN SITU ANALYSES»T£MPERATUR£. CONDUCTANCE. pH.
DO, SECCHI DISK TRANSPARENCY
* PROCESSING LAB ANA LYSES«D 1C. pH, TRUE COLOR,
TURBIDITY. PCV ALUMINUM (DISSOLVED: ORGANIC)
• ANALYTICAL LAB ANALYSES "ALL 24 ELS-I VARIABLES
• SPECIAL STUOIES*TRACE METALS; Pb. Cd, Hi. Mn. Cu
Figure 3-1. Sampling strategies for the Spring Seasonal Survey, ELS-II (from Merrill and Sheppe,
1988).
21
-------
SUMMER SEASONAL
FALL INDEX SITE
1.5m
I TRACE I
« TOTAL N.I'
I TOTAL P I
METALIMNION^X ~
I TOTAL N J
ITOTALfJ
L -\ CH LOR OPH Y U."l ~{
\ _—__——_——
M ANALYTICAL LAB |
I BIAS SPLIT I
HYPOLIMNION<
MtO HYPOLYMNION
-------
FALL SEASONAL
FALL INDEX SITE
I*.
SPECIAL STUDY SAMPLE
= ALIQUOTS DRAWN FROM
1— ! LAKE SAMPLE
X)= POINT Of SAMPLE COLLECTION
SAMPLING SITES OF A SUBSET OF
X>* LAKES: SITE MAY VARY IN RELATION
TO FALL INDEX SITE
LAKE WATER MEASUREMENTS
• IN SITU ANALYSES'TEMPERATURE. CONDUCTANCE, pH.
DO. SECCHI DISK TRANSPARENCY
» PROCESSING LAB ANALYSES*DIC. pH. TRUE COLOR.
TUHBIOITY. PCV ALUMINUM (DISSOLVED: ORGANIC!
* ANALYTICAL LAB ANALYSES'ALL 24 ELS-4 VARIABLES
• SPECIAL STUDIES
- TRACE METALS; Pb. Cd. Ni. Mn, Cu
- VARIABILITY OF SAMPLING SITIJSPATIALI
Figure 3-3. Sampling strategies for the Fall Seasonal Survey, ELS-II (from Merritt and Sheppe,
1988).
23
-------
Due to inclement weather and logistic constraints, only 41 of the 50 selected lakes (17 in the
Adirondaeks and 24 in central and southeastern New England) were sampled three times in the fall.
The ELS-II lakes not included in the Fall Variability Study were sampled only once in the fall at the
same location as the ELS-I sample. In total, water samples were collected from 152 lakes, 7 of which
were special interest sites (lakes that are not part of the statistical sample but are of interest in
comparing ELS-II results with those of other researchers). Only epilimnetic water samples were
collected. All fall seasonal and variability study samples were collected between October 8 and
November 14, 1986. In the fall variability study, there were typically (median) 10 days between the first
and third sample visits (interquartile range = 7-15 days).
3.3 FIELD METHODS
3.3,1 Site Location
Samplers identified the study lakes by comparing their features with those depicted on 1:24,000-
scale USGS maps. Helicopter crews verified the location of each lake by comparing lake latitude and
longitude obtained via a LORAN-C guidance system with those established during ELS-I, also using
LORAN-C. In the spring, summer, and fall Seasonal Surveys, samples were collected from the same
general location on a lake as those in ELS-I by using ELS-I data sheets showing the lake outline, sample
location, and sample depth (Merritt and Sheppe, 1988). Lakes were accessed by either helicopter or
boat. A comparability study of the two access methods conducted during the Western Lake Survey
demonstrated that there were no significant differences that would affect data interpretation (Landers et
al., 1987).
3.3.2 Sample Collection
Water samples were collected from the lake using a 6.2-L Van Dorn bottle fitted with a nylon Leur-
lok valve. Before any container was filled, it was rinsed three times with sample water from the Van
Dorn. Four gas-tight 60-mL polypropylene syringe samples were filled from the Van Dorn without
exposing the samples to the atmosphere, in order to minimize changes (e.g., degassing) in the water
sample prior to analysis. These syringes were analyzed in the processing laboratory for pH, dissolved
inorganic carbon (DIG), and total monomeric and nonexchangeable aluminum (Table 3-1). Next, a 4-L
cubitainer was filled with water from the Van Dorn bottle. Thus, a routine sample consisted of four
syringes and one cubitainer.
Two types of QC samples were collected in the field, blanks and replicates. Blanks were collected
by rinsing and filling the Van Dorn bottle with deionized water and then filling the syringes and
cubitainer. For replicates, another set of four syringes and one cubitainer were filled with lake water
24
-------
from the Van Dorn bottle. At least one field blank was collected per day for each laboratory batch. A
replicate water sample was collected from at least one lake per day. In the summer, triplicate samples
were also collected to assess analytical laboratory bias.
3.3.3 In Situ Measurements
After confirming the sampling location on the lake, in situ chemical measurements were made
(Table 3-1). In situ chemistry measurements included pH, specific conductance, dissolved oxygen,
Secchi depth, and temperature. All field pH, specific conductance, temperature, and dissolved oxygen
determinations were made using Hydrolabs (Surveyor II). The Hydrolab pH probes were calibrated each
morning using commercially available high ionic strength buffer solutions (pH 4.0 and 7.0). A low ionic
strength quality control check sample (QCCS) was used to check the calibration of the meter before
leaving for the field and again before and after in situ measurements. Measurement and calibration
techniques for all in situ measurements are described in detail in Merritt and Sheppe (1988).
3.4 SAMPLE HANDLING
Water samples were transported from the sample site in coolers containing frozen chemical
refrigerant packs that maintained a temperature of approximately 4° C until the samples arrived at the
processing laboratory. Samples were shipped by overnight courier to ensure their arrival at the
processing laboratory in Las Vegas, Nevada, on the morning after collection. Upon arrival at the
processing laboratory, samples were organized into a batch for processing. A sample batch consisted
of a group of routine lake samples and related QC samples. In almost all cases, processing laboratory
analyses were completed and samples were preserved and split into aliquots within 36 hours after
sampling.
Within each batch, each sample was assigned a unique identification number to distinguish it from
all other samples in the survey. After the batch of samples was preserved and split into aliquots, it was
shipped by overnight courier to a contracted analytical laboratory for chemical analysis (Table 3-1).
3.5 PROCESSING LABORATORY TECHNIQUES
The processing laboratory provided a controlled environment for processing and preserving water
samples and performing certain chemical measurements that needed to be completed as soon as
possible after sample collection. Figure 3-4 illustrates the processing laboratory activities. Chemical
parameters that tend to become unstable over time (i.e., pH, DIG, and the aluminum species) were
measured in the processing laboratory.
25
-------
ISJ
o>
Samples Organized
into Batch for Processing
t
Syringes
DIG
pH
Total Monomeric
Aluminum
Non-exchangeable
Aluminum
t
4 - L Containers
- Turbidity
- True Color
- Conductivity
Aliquots 1-7
f
FILTERED;
1 I NITRIC ACID_
PRESERVED
FILTERED;
BULFURIC ACID
PRESERVED
5( UNFILTERED_
SULFURIC ACID t
PRESERVED
UNFILTEBED;
NITRIC ACID
PRESERVED
.' FILTERED; METHYL-
2 I ISOBUTYL KETONE |
, EXTRACTION
FILTERED
Ca, Mg, K. Na.
Fo, Mn
if Al-exl
cr, F , SOA'
NO3"% Si02
DOC,
DIC-tnH, DlC-oq,
pH-ANC, pH-BNC.
Cond-lab, pH-eq,
ANC, BNC
Al-lola!
ANALYTICAL
Figure 3-4. Processing laboratory activities, ELS-II.
-------
Each 4-L cubitainer sample was split into seven aliquots and prepared for shipping to analytical
laboratories for additional analyses. Subsamples were also taken from each Cubitainer for measuring
turbidity, specific conductance, and true color (Table 3-1). Field crews capped the syringe samples with
air-tight teflon valves to prevent air equilibration from occurring before analysis. The syringe samples
were used to measure pH, DIG, and total and nonexchangeable monomeric aluminum species and to
prepare the total extractable aluminum aliquot (Table 3-1). Processing laboratory measurements of
these variables were essential for providing quality data within holding time requirements. Processing
laboratory analytical methods are described in Arent et a!. (1988) and Hillman et a!. (1987).
Figure 3-4 depicts the seven aliquots prepared from each Cubitainer. The aliquots were stabilized
by filtration (0.45-/im filter), acid preservation, or refrigeration, or some combination of these procedures.
All aliquots were stored and shipped at 4°C to reduce biological activity and, for total extractable
aluminum aliquots, to reduce volatilization of solvent. Arent et al. (1988) give a detailed discussion of
processing laboratory activities for ELS-II.
3.6 ANALYTICAL LABORATORY SUPPORT
Standard EPA contract laboratory procurement procedures were used to secure the services of
two analytical laboratories (Mitchell-Hall et al., 1989). Laboratory 1 analyzed the spring samples and
laboratory 2 analyzed the fall samples. Both laboratories were involved in the analysis of summer
samples. Table 3-1 lists the analytical instruments and methods used by the analytical laboratories. A
detailed description of each analytical method is given in the ELS-II analytical methods manual (Kerfoot
et al., 1988). The maximum allowable holding time for each analyte before analysis is given in Table 3-2.
3.7 DATA BASE DEVELOPMENT
ELS-II data management and analysis were patterned after procedures developed for the ELS-I
(Kanciruk et al., 1986). The ELS-II data base used in this report has been subjected to four levels of QA
evaluation to ensure that the data collected during ELS-II is representative of the physical and chemical
characteristics of the lakes at the time of sampling. The completion of each level of QA produced a new
working data set of increased refinement. These working data bases are defined as: raw (Data Set 1),
verified (Data Set 2), validated (Data Set 3), and enhanced (Data Set 4). The final product of this refine-
ment process, the enhanced data set (Data Set 4), incorporates data substitution and replacement of
missing values. This is the data set that is used for calculating ELS-II population estimates. Figure 3-5
summarizes the development of these working data bases. A detailed discussion of the data base
creation and contents is included in the ELS-II Data Base Dictionary (Jimenez et al., in press).
After the completion of the ELS-II survey, all lakes successfully sampled were targeted for morpho-
metric analysis. The bathymetrie mapping was successfully completed for 129 lakes in the summer of
27
-------
Table 3-2. Maximum Holding Times Specified for ELS-ll Samples3
Variable Holding Time
N03"b; MIBK-extractable aluminum 7 days
ANC; BNC; specific conductance; DIG; DOC; pHc 14 days
P; NH4+; CP S042"' F; Si02 28 days
Ca; Fe; K; Mg; Mn; Na; total aluminum 28 daysd
Sample preservation methodology is summarized in Figure 3-4.
Although the EPA (U.S. EPA, 1983) recommends that nitrate in unpreserved samples (unacidified) be determined within 48
hours of collection, evidence exists (Peden, 1981; APHA, 1985) that nitrate is stable for 2 to 4 weeks, if the sample is stored in
the dark at 4° C.
c Although the EPA (U.S. EPA, 1983) recommends that pH be measured immediately after sample collection, evidence exists
(MeQuaker et at., 1983) that it is stable for as long as 15 days, if the sample is stored at 4*0 and sealed from the
atmosphere. The pH was also measured in a sealed sample at the processing laboratory within 24-36 hours of sample
collection.
Although the EPA (U.S. EPA, 1983) recommends a 6-month holding time for these metals, this study required that all the
metals be determined within 28 days, which ensured that significant changes would not occur and that data would be
obtained in a timely manner.
28
-------
rield Sampling^
[and Processing
d.aborator\
Analytical
Laboratories
Error and
Range Check
RAW DATA SET
(Data Set I)
Batch Reports
Verification
Data Editing
and Flagging
SPECIAL
DATA
ASSESSMENT
VERIFIED DATA SET
(Data Set 2)
Site
Reports
Maps
Substitution
and
Replacement
Validation
Data Editing
and Flagging
VALIDATED DATA SET
(Data Set 3)
ENHANCED DATA SET
(Data Set 4)
Figure 3-5. ELS-II data base development.
29
-------
1987 using small craft equipped with a recording depth sounding instrument. Data from each lake were
then analyzed using a computer program developed by the Adirondack Lake Survey Corporation that
calculated the lake volume and residence time, and created a bathymetric map (J. Baker et al., 199Gb).
These data are contained in the validated and enhanced ELS-II data bases.
3.7.1 Raw Data Set (Data Set 1)
The collective data from all components of the sampling and analysis made up the raw data set.
Field, processing laboratory, and analytical laboratory personnel sent the original data forms to the QA
staff at the Environmental Monitoring Systems Laboratory (EMSL) in Las Vegas for review, in order to
ensure that data were correct and consistent. Completed forms were then forwarded to Systems
Applications, Inc. (SAI), where the data were entered into the data base. To ensure accurate data
transfer from the data forms to computer files, the information was double-entered into computer files
and subjected to automated checking procedures. The raw data set was used to screen the data for
problems, perform exploratory analysis, and evaluate the need for any adjustments in the data analysis
plan.
3.7.2 Verified Data Set (Data Set 2)
The objectives of the data verification process were to identify, correct, and flag raw data of
questionable or unacceptable quality and to identify data that might need to be corrected during or after
data validation. These objectives were met by reviewing the QC data measured and recorded at the
sampling site, at the processing laboratory, and at the analytical laboratories. The verification process
was automated as much as possible through applicable computer programs.
Verification began with receipt of the data forms from the field and processing laboratory. An
auditor reviewed the forms for completeness, agreement between field and laboratory forms, and proper
assignment of sample identification codes and data qualifier tags. Data anomalies were reported to the
field base site and processing laboratory coordinators for corrective action. Data reporting errors were
usually corrected on the data forms before the data were entered into the raw data set. During verifi-
cation, each sample was evaluated individually and by analytical batch. Individual values that were
identified as exceptions (as a result of cation/anion balances, specific conductance balance, or protolyte
analysis) or that did not meet the acceptance criteria, were flagged in the data base. Suspect values
were also identified by examining QC data (blank, duplicate, and audit samples) measured and recorded
at the processing and analytical laboratories. In addition, data qualifier flags were added when QC
samples did not meet acceptance criteria, or when sample analysis holding time requirements or instru-
ment detection limits were not met. The output from these checks, along with original data and field
notebooks, was used to evaluate the quality of the analytical results. Based on the evaluation of the
30
-------
analytical results reported for QC samples, analytical laboratories were directed to confirm reported
values or to reanalyze selected samples. If a value was identified as an exception to expected results, a
flag was placed in the data base.
During the ELS-II data verification and validation activities, concern arose regarding several data
quality issues. This concern, focusing primarily on spring data from the chloride, nitrate, sulfate, and
ANC analyses, prompted a special data assessment, which took place after the completion of the official
verified data base. This special assessment included an extensive examination of the raw data from
both analytical laboratories for many parameters and resulted in the creation of a modified verified data
set. The validated and enhanced data sets were then constructed from this modified verified data set. A
complete description and listing of all changes made between the original verified data set and the
modified verified data set are given in the ELS-II QA Report (Mitchell-Hall et a!., 1989).
In the special data assessment, analytical data were reevaluated for fall sodium, spring anions (Cf,
SO42", NO3~), and two batches of spring ANC values. In addition to these changes, the data base was
also updated to reflect sample switches and transcription errors, and missing values that had been
inadvertently reported as zero were set back to missing values.
At laboratory 2, fall sodium was measured by both atomic absorption spectroscopy (AAS) and
inductively coupled plasma emission spectroscopy (ICPES). The original verified data were based
exclusively on the AAS data. Examination of the precision of the two methods for fall sodium audit
samples showed that ICPES was much more precise than AAS in all but three batches of samples in the
fall data. Thus, ICPES sodium data were substituted for AAS sodium data in the fall modified verified
data base for all samples not in one of the three batches that showed better AAS audit precision (see
Mitchell-Hall et al., 1989, p. 39 for more detail).
Problems with the spring chloride, nitrate, and sulfate data were recognized early in the data vali-
dation process. Several types of errors contributed to the problems with these anion data, including (1)
dilution errors (sample values outside the range of calibration standards), (2) QC check solution values
exceeding contract-required detection limits, and (3) reporting errors. As a result, spring anion aliquots
were analyzed twice and in some cases three times for sulfate, nitrate, and chloride. Analysis 1 was the
original analysis in March and April of 1986. Analysis 2, a reanalysis, occurred in May and June of 1986.
Finally, several samples (19% of the chloride data, and 11% of the sulfate and nitrate data) were selected
for a third analysis, a year later, in May 1987.
Much of the effort of the special data assessment centered on deciding which analysis results to
use in the modified data base for each batch of samples. Accuracy estimates (from audit samples) and
precision estimates (from duplicate samples) were calculated for each of the spring sample batches for
sulfate, nitrate, and chloride (see Section 4 for details on accuracy and precision estimates). In each of
the sample batches, data from the analysis with the best accuracy and precision were used in the modi-
fied verified data base. For sulfate, among the 28 batches analyzed in the spring ELS-II survey, data
from 11 of the batches were based on analysis 1, data from 16 batches were based on analysis 2, and
31
-------
data from 1 batch was based on analysis 3. For nitrate, data from 12 batches were based on analysis 1
and data from 16 batches were based on analysis 2. For chloride, data from 8 batches were based on
analysis 1, data from 17 batches were based on analysis 2, and data from 3 batches were based on
analysis 3. Details of the three analyses, the analysis selection process, and within-batch accuracy and
precision estimates are presented in Mitchell-Hall et al. (1989). The resulting overall estimates for spring
precision and accuracy are discussed in Section 4.
In the original verified spring data base, 2 batches of samples (out of 28} had ANC values much
higher than the carbonate alkalinity values calculated from both processing and analytical laboratory pH
and DIG measurements. A detailed investigation indicated that the actual concentration of the acid
titrant reported by the laboratory and used in calculating ANC in these two batches was less than one-
half the value that the laboratory reported. The ANC values for these batches were recalculated for the
modified verified data base using a calculated concentration of the acid titrant and the data from the
autotitrator (see Mitchell-Hall et al., 1989, pp. 35-38).
Other than the changes just described for fall sodium, spring anions, and the two batches of
spring ANC samples, the number of changes made to the data base during the special data assessment
was relatively small. For the variables analyzed in Sections 5 and 6, changes were made to routine/
duplicate lakewater data for two sodium values and one calcium value in the spring. In the summer,
three DOC, seven nitrate, and five sulfate values were changed. Other than fixing one sample switch,
none of the fall data presented in Sections 5 and 6 were changed by the special data assessment.
3.7.3 Validated Data Set (Data Set 3}
The purpose of the verification procedures was to evaluate data at the sample and batch level.
Validation, on the other hand, was intended to compare samples across the population of lakes and
across variables. The two main components of the validation process were (1) identification of outliers
from regional distributions of chemistry and (2) evaluation of possible systematic errors in the measure-
ment process. During validation, missing values and values with known errors, based on relationships
with other variables, were identified and assigned validation flags. Values with validation flags were
deleted in the creation of the enhanced data set. Only the major variables (shown in Table 3-3), whose
cumulative distribution functions (CDF) are presented in Appendix A, were validated in ELS-II.
The creation of the seasonal validated data sets from the modified verified data sets involved (1)
changing units from mg/L to/^eq/L where appropriate, (2) creating two new variables, mean Secchi
depth, and labile (inorganic) monomeric aluminum calculated as the difference between total monomeric
aluminum and organic monomeric aluminum, (3) adding Adirondack Lake Survey Corporation physical
data, (4) adding cluster and phase II weights, (5) rounding, and (6) concatenating the validation flags to
the verification flags.
32
-------
Table 3-3. Variables Validated in ELS-I
Base neutralizing capacity
MIBK-extractable aluminum
Chloride
Closed headspace DIG
Total fluoride
Manganese
Nitrate
Total phosphorous
Monomeric (PCV) aluminum Organic monomeric (PCV) aluminum
Acid neutralizing capacity
Color
Dissolved organic carbon
Potassium
Sodium
Air-equilibrated pH
Sulfate
Calcium
Specific conductance
Iron
Magnesium
Ammonium
Closed headspace pH
Silica
3.7.4 Enhanced Data Set (Data Set 4)
The enhanced (final) data set for each seasonal chemistry survey was constructed from the vali-
dated data set by (1) deleting all values with a validation flag, (2) averaging all duplicate values (from
field duplicate samples), (3) changing negative values (except for ANC and BNC) to zero, and (4) replac-
ing all missing values with values calculated from regression models, so that population estimates could
be made.
Calculation of population estimates and their confidence bounds is difficult if there are missing
values in the data, thus missing values and deleted validation outliers were replaced in the enhanced
data set if they were necessary for making population estimates (epilimnion samples from target lakes).
When necessary, substitutions were determined, using a linear regression model, by calculating a pre-
dicted value based on observed relationships with other chemical variables. Of the more than 10,000
major chemical values in the ELS-II data base, 12 values were replaced in the spring data, 14 values in
the summer data, and 4 in the fall data.
33
-------
SECTION 4
DATA QUALITY ASSESSMENT
4.1 INTRODUCTION
Objectives of Phase II of the Eastern Lake Survey (ELS-II) were that the data be of high quality,
have low and quantifiable analytical error, have known precision, and be representative of the state-of-
the-art analysis attainable in high-volume contract analytical laboratories. The quality assurance and
quality control (QA/QC) program of ELS-II was designed to maximize the utility of the collected data and
to minimize the likelihood of erroneous chemical data. Data quality objectives (DQOs) were established
and used as guidelines in maintaining data integrity during the sampling, analysis, and data collection
and recording (Mitchell-Hall etal., 1989).
This section describes the analytical approach, results, and conclusions for each of the five
aspects of data quality (completeness, comparability, representativeness, accuracy, and precision), plus
detectability, applicable to ELS-II data. Detectability was analyzed because the low ionic strength of
many of the ELS-II samples necessitated an evaluation of the background levels of analytes.
Representativeness, completeness, and comparability were considered important data quality
goals in the development of the statistical sampling design of the ELS-II and the QA plan. As explained
in Section 2, they were affected by uncontrollable events that influenced the number of samples actually
collected and analyzed by the proper protocols during the course of the survey. Detectability, accuracy,
laboratory bias, and precision were quantitatively assessed by using the analytical results from QC
samples.
Analysis of the QC samples (blanks, natural audit samples, and field duplicates) provided two
kinds of information for the assessment of data quality. Sampling and laboratory performance could be
gauged against the DQOs established for precision, accuracy, and detectability. In addition, unforeseen
effects of the collection and measurement process on analytical results could be quantified and their
impact on data interpretation discerned. For example, the addition of background levels of an analyte
during sample collection and subsequent handling can hinder the comparison and interpretation of data
from lakes having naturally low levels of that analyte. interlaboratory bias also can confound statistical
comparisons of data, because true differences may not be distinguishable from differences resulting
from systematic measurement errors at the different laboratories.
4.2 COMPLETENESS, COMPARABILITY, AND REPRESENTATIVENESS
Of the 147 lakes initially selected for field sampling, 145 (99%) were actually visited in the field in
all three seasons. One lake was not sampled in the fall due to inclement weather and the other lake was
not sampled in the spring and fall because of access permission difficulties (see Section 2.4.1). Sample
34
-------
weights were adjusted for these missing samples as described in Section 2 so that the ELS-II project
objectives were not affected by the missing lake data. Of the 742 lakewater samples collected during
ELS-II, 732 were analyzed for all planned chemical variables. Only 10 missing values were replaced
during the creation of the enhanced data set.
All the ELS-II field crews and laboratory personnel used standardized protocols (Arent et al., 1988;
Merritt and Sheppe, 1988; Kerfoot et al., 1988), which maximized internal comparability. Similarly, the
use of these standard methods, combined with the quantitative results of QC sample analysis (Mitchell-
Hall et al., 1989), allowed comparison with data from other studies, such as the ELS-I (Linthurst et al.,
1986), WLS (Landers et al., 1987), and NSS (Kaufmann et al., 1988), and facilitated detailed process
oriented studies on ELS-II special interest sites.
Representativeness can be viewed in a hierarchical manner from specific analyses to the general
representativeness of the group of sample lakes. At the lowest level, representativeness refers to how
well our chemical and physical analyses reflect the physical and chemical conditions at the sampling
location in the lake. This aspect of data quality is discussed in Section 4.3 (detectability, accuracy, and
precision).
At a higher level, the concept of representativeness refers to how well a water sample character-
izes a specific lake. Representativeness in this context is influenced by the location of the sampling site,
the specific location (or microhabitat) from which the sample was collected, and the local conditions at
the time of collection. One goal of ELS-II is to assess the representativeness of the ELS-I fall index
sample by taking multiple lake samples during the Fall Variability Study. At the population level,
representativeness refers to whether the ELS-II sample lakes were representative of the target population
as a whole. The probability sampling design employed by ELS-II makes it extremely unlikely that sub-
stantial bias might result from seriously undersampling any particular geographic area or class of lakes
defined as part of the target population. Recall that in the ELS-II, specific populations of lakes were
deliberately not sampled (high ANC or shallow lakes; see Section 2.3.1).
4.3 DETECTABILITY, ACCURACY, PRECISION, AND LABORATORY BIAS
Indices of detectability, accuracy, and precision can be calculated at both the method level and
the system level. Indices of analytical error are calculated from laboratory blank, audit, and duplicate
samples and refer to the quality of the analytical methodology. On the other hand, indices of sampling
error are determined from blank, audit and duplicate samples processed in the field, and they apply to
detectability, accuracy, and precision of the whole sampling process (sample collection, handling,
processing, and shipping, as well as analytical methodology). This section discusses only the sampling
error indices, because they are more inclusive of the total variability involved in collecting the data. Also,
the discussion is restricted to the 12 major variables on which this report concentrates (ANC, pH, SO42",
NO3~, Cf, base cations, DOC, and aluminum). A more complete description of the ELS-II QA process,
35
-------
QA data, and information on all ELS-II variables can be found in Mitchell-Hall et al. (1989). This section
first explains how the indices were determined and concludes with a variable-by-variable assessment of
detectablilty, accuracy, precision, and laboratory bias.
4.3.1 Detectability
Background levels of analyte added during the analysis, collection, or handling of samples were
estimated by computing a system decision limit (SDL). The SDL represents the lowest measured sample
value that can be distinguished from a blank sample or from background noise. In ELS-II, the SDL was
based on the mean and standard deviation of field blanks. Field blanks were reagent-grade deionized
water put into Cubitainers and syringes in the field and treated as samples throughout shipping,
processing, and analysis. SDLs were calculated as:
SDL = Meanb|anks + 1.65(standard deviationb|anks) (9)
Thus, the SDL corresponds to the 95th percentile of the blank measurements (assumes a normal distri-
bution) and there is a 0.05 probability of concluding that an analyte is present when in fact it is not.
SDLs for the major ELS-II chemical variables are listed in Table 4-1.
System decision limits are important in the interpretation of ELS-II chemical distributions. Popu-
lation estimates of lake resources with concentrations at or below the SDL should be interpreted with
caution. There can be little confidence that values reported at these very low analyte levels are signifi-
cantly different from zero. It is likely that variations in the distributions of observed values less than the
SDL are artifacts of sample collection, handling, and analysis. For the same reason, groups of lakes
characterized by analyte levels below the SDL should not be compared.
4.3.2 Accuracy
Accuracy is defined as the closeness of a measurement to a true or known value. Typically, the
EPA has evaluated accuracy by calculating a percent difference from a known value. Although this
approach to data quality has proven effective for studies of a single analyte over a limited range of
concentrations, it is not as effective when evaluating multianalyte survey data with a wide range of con-
centration. In ELS-II, the approach for assessing accuracy was different than in previous NSWS surveys
(Mitchell-Hall et al., 1989). Accuracy was assessed by comparing the distribution of field natural audit
samples to a target value. The target value was the median of all of the available natural audit sample
data (including ELS-II, National Stream Survey, and Spring Variability Pilot Study data). Two natural
audit samples were analyzed repeatedly in the ELS-II: lake water from Big Moose Lake (an acidic lake;
36
-------
Table 4-1. System Decision Limits (Mean + 1.65 Standard Deviation of Field Blanks) for Major
ELS-II Analytes
Variable
A'MIBK" (W/D
Mmc e«/g
Ca2+ (/leq/L)
CT 0*q/L)
DOC (mg/L)
K* 0«q/L)
2-1-
Mg (/zeq/L)
Na+ (peq/L)
NO3" (/^eq/L)
SO/' Oieq/L)
Spring
1
16
1.2
8.0
0.6
0.5
0.6
3.1
0.5
1.4
Seasonal Survey3
Summer
12
16
1.4
2.2
0.4
0.2
1.1
0.4
0.9
1.0
Fall
8
17
0.7
0,8
0.3
0.2
0.2
2.8
0.5
0.8
29 blank samples were analyzed in spring, 17 in summer, and 26 in fall.
AyBK = MIBK extractable aluminum.
AL = PCV reactive (total monomeric) aluminum.
37
-------
ANC = -3/xeq/L) and lake water from Seventh Lake (a circumneutral lake; ANC =155 ^eq/L), Field
natural audit samples were prepared at a support laboratory and processed in the processing laboratory
as routine samples. Natural audit samples were taken from the same large containers of lake water
throughout all the ELS-II seasonal surveys. Target values and the mean ± standard deviation of natural
audit sample data in each season are given in Table 4-2 for the major ELS-II chemical variables.
4.3.3 Precision
Analysis of field duplicate pairs provided an estimate of overall sample precision within a batch,
including the effects of sample collection, processing, and analysis. Repeated measurements of natural
audit samples provided estimates of among-batch precision within a laboratory, and when pooled across
the two laboratories, as they were in summer, this included the effects of interlaboratory bias. A batch
refers to a batch of samples grouped together at the processing laboratory and kept together through-
out the analytical process with the same batch identifier.
For each variable, within-batch precision was estimated as a pooled standard deviation based on
the means and variances of the duplicate sample pairs. The individual variances of each sample pair
were summed to calculate a pooled variance, and a pooled standard deviation was calculated as the
square root of the pooled variance after dividing by the sample size. The percent relative standard
deviation (%RSD) was then calculated as:
pooled standard deviation
%RSD = * 100 (10)
grand mean of duplicate samples
where the grand mean equals the average of all of the sample pair means. Grand means and %RSD for
within-batch precision are given in Table 4-3 for the major ELS-II chemical variables. Because of the
dilute nature of the waters sampled during the ELS-II, the single %RSD estimate of precision for the
entire range of analyte values should be interpreted with caution. If many of the sample pairs have very
low concentrations, the %RSD can be high, even though the actual variability may be small. For
example, the %RSD for nitrate in the summer seasonal survey was 28% (Table 4-3). The grand mean,
however, was 1 \i eq/L, so the pooled standard deviation was only 0.28 n eq/L. Therefore, it is important
to consider the grand mean as well as the %RSD when interpreting these data.
The %RSD for among-batch precision in Table 4-4 was calculated from the data in Table 4-2 by
dividing the standard deviation of the repeated analyses of the natural audit samples by the mean value.
%RSD for within-batch (from Table 4-3) and among-batch data precision are compared in Table 4-4.
Summer data for the among-batch %RSDs (natural audit samples) were pooled from data from both
analytical laboratories and thus also include the effects of interlaboratory bias.
38
-------
Table 4-2. Mean ± One Standard Deviation of Seventh Lake (SL) and Big Moose Lake (BML)
Field Natural Audits for Major ELS-II Analytes3
Variable
M
Alm
ANC
Ca2*
cr
DOC
K+
Mg2+
Na+
NCV
pH
so42-
Audit
SL
BML
SL
BML
SL
BML
SL
BML
SL
BML
SL
BML
SL
BML
SL
BML
SL
BML
SL
BML
SL
BML
SL
BML
Target13
Value
16.3
166
21.2
193
155
-3
252
95.9
82.9
11.6
3.6
3.6
12.5
10.5
67.8
26.8
94.7
26.4
20.5
19.7
6.9
5.1
143
132
Spring
20.517.4
142130
22.515.0
200113
150+18
-4 + 4
24515.8
96.2+2.0
89.6+31.1
11.512.2
3.510.3
3.510.2
12.6+0.4
10.4+0.4
67.811.3
26.810.8
93.0+3.6
26.4+1.7
20.3+1.7
17.5+2.2
6.910.03
5.2+0.04
14216.7
127+12
Seasonal Survey
Summer
8.9+9.2
147154
20.915.8
178+20
160 + 21
-314
24915.7
95.4+1.8
75.5+8.3
11.111.0
3.7+0.1
3.5+0.2
12.710.5
10.610.4
66.511.9
26.210.7
96.7+10.4
26.111.0
20.810.6
19.6+0.5
6.7+0.09
5.110.06
14319.7
135110
Fall
13.2+4.3
137+17
24.6+3.4
19618.1
15919
-2 ± 2
258+4.7
98.212.0
82.2+5.1
11.610.2
3.510.1
3.410.2
11.9+1.5
9.7+1.1
67.811.0
27.311.1
95.812.9
27.111.8
20.0+1.2
19.410.5
6.910.04
5.110.03
141119
132+4.7
All units are neq/L except for pH, DOC (mg/L), and aluminum (/jg/L). The sample size for the Seventh Lake audit was 11 in
the spring, 7 in the summer, and 11 in fall; for the Big Moose Lake audit it was 10 in the spring, 12 in summer and 11 in fall.
Target values represent the median of all audit samples measured in 1986 in the ELS-II, National Stream Survey, and Spring
Variability Study.
39
-------
Table 4-3. Grand Mean (Kg) and Percent Relative Standard Deviation (% RSD) of Field Duplicate
Pairs for Major ELS-II Analytesa
Variable
AIMIBK
Alm
ANC
Ca2+
cr
DOC
K+
Mg2+
Na+
N03-
PHb
S042'
Spring
XG
43
65
52
121
159
3.4
11.4
57.8
149
7.2
5.910.03
123
n=26
%RSD
56
4.5
8.8
2.0
20
11
5.4
1.7
2.8
6.7
3.0
Summer i
XG
13
41
132
173
137
4.2
11.0
75.1
138
1.0
6.310.1
117
n = 17
%RSD
71
6.6
4.7
1.8
25
9.6
2.8
3.5
4.5
28
2.0
xc
34
63
94
149
115
4
10,
52
134
2
6
108
Fall n = 29
. %RSD
11
4.5
5.4
1.1
4.4
.6 6.2
.7 12
.1 1.1
1.2
.5 11.6
.010.02
2.4
X units are in fieq/L except for pH, DOC (mg/L), and aluminum (/xg/L). % RSD = (SC|Jool / )^) * 100. SC^00, is the pooled
standard deviation = (X pair variance / n)as.
Data for pH expressed as X~ ± pooled standard deviation.
40
-------
Table 4-4. Within-Batch and Among-Batch Precision3 (% RSD) for Major ELS-II Analytes
Spring
Variable
UMBK
Alm
ANC
Ca2+
Cl"
DOC
K+
Mg2+
Na+
N03-
pHb
so/
Within
Batch
56
4.5
8.8
2.0
20
11
5.4
1.7
2.8
6.7
0.03
3.0
Among
SL
36
22
Batch
BML
21
6.5
12 100
2.4
35
8.6
3.2
1.9
3.9
8.4
0.03
4.7
2.1
19
5.7
3.8
3.0
6.4
13
0.04
9.4
Summer
Within
Batch
71
6.6
4.7
1.8
25
9.6
2.8
3.5
4.5
28
0.10
2.0
Among
SL
103
28
Batch
BML
12
11
13 133
2.3
11
2.7
3.9
2,9
11
2.9
0.09
6.8
1.8
9
5.7
3.8
2.7
3.8
2.6
0.06
7.4
Fall
Within
Batch
11
4.5
5.4
1.1
4.4
6.2
12
1.1
1.2
12
0.02
2.4
Among
SL
33
14
Batch
BML
12
4.1
5.6 100
1.8
6.2
2.9
13
1.4
3.0
6.1
0.04
13
2.0
1.7
5.9
11
4.0
6.6
2.6
0.03
3.6
Precision estimates are expressed as %RSD (see text). Within-batch precision is based on duplicate sample pairs. Among-
batch precision is based on repeated analyses of natural audit material (SL = Seventh Lake, BML = Big Moose Lake; see
Table 4-2).
pH precision is expressed in pH units as pooled standard deviation.
41
-------
4.3.4 Laboratory Bias
In the ELS-11, two analytical laboratories were used to analyze the data. All the spring data were
analyzed by laboratory 1 and all the fall data were analyzed by laboratory 2. In the summer, both
laboratories were involved in sample analysis. Thus, if significant analytical bias exists between the
laboratories, apparent seasonal differences between samples could simply be the result of interiabora-
tory bias rather than actual differences. Laboratory bias can be evaluated three different ways. One way
is to compare split samples that were analyzed by both laboratories during the summer seasonal survey.
Split differences are summarized in Table 4-5, and plots of laboratory 1 values versus laboratory 2 values
are shown in Figures 4-1 and 4-2 for select variables. Secondly, laboratory bias may also be examined
by comparing mean natural audit sample values in the spring (all laboratory 1) with those in the fall (all
laboratory 2) (Table 4-2). A third way to assess laboratory bias is to compare the among-batch pre-
cision estimates made in the summer with those made in the fall and spring. Spring and fall %RSD
values reflect only the precision of the one laboratory that did the analyses in that season. The summer
data are pooled across laboratories and laboratory bias is incorporated into the %RSD. If the two
laboratories are measuring substantially different values in the summer data for the audit sample
material, then this would be reflected in an elevated %RSD relative to spring and fall %RSDs.
4.3.5 Variable by Variable Summaries
4.3.5.1 ANC and pH
For obvious reasons, detectabillty was not assessed for ANC (it can be negative) and pH. In this
section, pH refers to the elosed-headspace (syringe sample) pH measured at the processing laboratory.
Closed-headspaee pH Is also the pH variable used in this report to make population estimates of lake-
water pH conditions. In terms of accuracy, mean ANC values of the natural audit samples were within 5
j«eq/L of the target value in each of the seasonal surveys for the high ANC (Seventh Lake) audit sample.
In the acidic audit sample (Big Moose Lake), mean values were within 1 fj/eq/L of the target value (Table
4-2). Mean audit sample values for pH were within 0.1 pH unit of target values, except for the summer
Seventh Lake audit sample, which was within 0.2 pH unit.
Duplicate sample precision (%RSD) for ANC ranged from 4% to 7% in the three seasonal surveys
(Table 4-3). Precision for the higher ANC natural audit sample ranged from 5% to 13%. Precision for
the acidic audit sample was about 100%, because the mean values were only -2 to -4 fjeq/L. Pooled
standard deviations in the acidic audit samples, however, were very low (2-4 jueq/L; Table 4-2). Pooled
standard deviations for both audit samples and duplicate samples were < 0.1 pH unit for closed-system
(processing laboratory) pH. Precision for pH was not expressed as %RSD because of its logarithmic
nature.
42
-------
Table 4-5. Differences between Summer Split Samples Analyzed at the Two Analytical
Laboratories
Variable
ANC
S042'
N03
CBa
DOC
Variable
ANC
SO42"
N03-
CBa
DOC
Minimum
-77
-33
-0.9
-45
-0.9
Minimum
-11
-33
-0.9
-11
-0.9
0,
-18
-12
0.1
-11
-0.2
ANC
Q,
-3
-28
0.1
-6
-0.2
All data, n = 25
Lab 1 - Lab 2
Median
-2
-4
0.4
-2
0.1
< 50 peq/L, n = 1 1
Lab 1 - Lab 2
Median
0
-13
0.5
-2
0.1
Q-3
1
1
0.7
6
0.4
3
1
0.8
6
0.3
Maximum
10
17
9.4
18
1.2
Maximum
10
8
1.5
18
0.4
Q, =
Kf.
43
-------
u
I
400-
200-
0-
0 200
LABORATORY 2
400
400"
3
s
CO
200-
200 400
LABORATORY 2 SO4 (jieq/L)
6.4-
I UH
g
so
0-
0 32 6A
LABORATORY 2 NO3 (peq/L)
O
O
D
8 -
4 -
4 8
LABORATORY 2 DOC(mg/L)
U
o!
O
a
560-
280-
280 560
LABORATORY 2 Cl(neq/L)
Figure 4-1. Comparison of laboratory 1 and laboratory 2 values for replicate samples analyzed
during the summer seasonal survey: (a) ANC, (b) sulfate, (c) nitrate, (d) DOC, and
(e) chloride. Taken from Mitchell-Hall et al. (1989), Appendix E.
44
-------
2
1
g
CQ
430-
215-
21S 430
LABORATORY 2 Na(jie<]/L)
~ 50.0-
I
O
0
25.0-
0.0
0.0 23.0 50.0
LABORATORY 2 K(flcq/L)
U
1
400-
200-
200 400
LABORATORY 2 Ca(ncq/L)
I
s
0
240-
160-
80 160
LABORATORY 2 Mg (|teq/L)
Figure 4-2. Comparison of laboratory 1 and laboratory 2 values for replicate samples analyzed
during the summer seasonal survey: (a) sodium, (b) potassium, (c) calcium, and (d)
magnesium. Taken from Mitchell-Hall et al. (1989), Appendix F.
45
-------
There was no evidence of laboratory bias for ANC in the acidic audit sample or in low ANC (< 50
peq/L) summer replicate (split) samples (Figure 4-1; Table 4-5). At higher ANC levels, there was some
indication of laboratory bias in the summer split samples. Overall, the interquartile range for the
difference between laboratory split samples was 19/*eq/L (laboratory 2 ANC values were usually higher
than laboratory 1 values). Also, there was a 9 M eq/L difference between the Seventh Lake audit sample
mean concentrations for fall and spring. The pH values used to make population estimates in this report
are those measured at the processing laboratory. As there was only one processing laboratory, labora-
tory bias is not an issue for the pH measurements presented in this report.
4.3.5.2 Sulfate
The system decision limit for sulfate was < 1.5 ji©q/Lfor ail the seasonal surveys, and thus
background contamination or detectability should cause no problems for sulfate data interpretation in
ELS-H (Table 4-1). Seasonal mean sulfate audit sample values were always within 4% of the target value
(Table 4-2). Sample duplicate (within-batch) precision was < 3% (%RSD) in all the seasonal surveys.
Among-batch precision was < 10% in all the audit sample data except for the fall Seventh Lake data,
which had a %RSD of 13% (Table 4-4). The relatively high variability in this seasonal audit sample
appears to be due to two outliers rather than to systematic imprecision. Laboratory bias does not
appear to be a problem for sulfate, based on laboratory split samples (Figure 4-1) and a comparison of
seasonal audit sample means.
4.3.5.3 Nitrate
The system decision limit for nitrate ranged from 0.5 to 0.9 p, eq/L in the three seasonal surveys
(Table 4-1). Because many of the ELS-II samples have low nitrate, these numbers should be kept in
mind when interpreting the data. Seasonal mean nitrate audit sample values were usually within 3% of
the target value, except for one seasonal audit sample that was within 10% of the target value (Table
4-2). Nitrate concentrations were relatively high (20 n eq/L) in the audit samples and among-batch
precision ranged from 2% to 13%. Sample duplicate precision had fairly high %RSD (7% to 28%), but
this was mainly due to the low nitrate concentrations in most of the sample pairs (grand means ranged
from 1 to 7 \ieq/L). Pooled standard deviations in the sample pairs ranged from 0.3 to 0.5 /ieq/L (Table
4-3). There was poor correlation between the summer laboratory split samples (Figure 4-1), probably
due to the low nitrate concentrations in the split samples and not to laboratory bias. Most of the split
samples had concentrations below the SDL for nitrate (0.9 fj, eq/L) and thus were not significantly
different from deionized water blank samples. A seasonal comparison of the audit samples shows only
small (< 2^ eq/L) differences between means and %RSDs (Table 4-4). There are, however, two data
points on the laboratory split plots (Figure 4-1 b) that show relatively high nitrate in laboratory 1
measurements (9/*eq/L), but almost no nitrate (< 1 neq/L) in laboratory 2 measurements.
46
-------
4.3.5.4 Chloride
There were data quality problems with chloride in analyses performed at laboratory 1.
Precision and detectability at laboratory 1 greatly exceeded the DQOs. We recommend not using
any of the chloride values from laboratory 1 and they have not been used in this report. Thus
ELS-II population estimates were not made for summer or spring chloride distributions. These data
quality problems can be seen for spring chloride detectability (SDL = 8/ieq/L), and precision (within-
batch %RSD = 20%; among-batch %RSD = 19-35%). Data quality does not appear as poor in the
summer because only about half the data were analyzed at laboratory 1, When the summer data are
separated by laboratory, laboratory 1 data are still poor, whereas laboratory 2 summer chloride data are
comparable to the fall results (Mitchell-Hall et al., 1989), There do not appear to be any chloride data
quality problems in the laboratory 2 fall and summer data (SDL < 1 ^eq/L; %RSD = 2-6%).
4.3.5.5 Dissolved Organic Carbon
The SDL for DOC ranged from 0.3 to 0.6 mg/L over the three seasonal surveys. Seasonal mean
audit sample DOC concentrations were within 0.2 mg/L of the target values (Table 4-2). Duplicate
sample precision ranges from 6 to 11 %RSD and audit sample precision ranged from 3 to 9 %RSD. In
terms of laboratory bias, the summer split samples showed close agreement for DOC values < 5 mg/L,
but there was some scatter (but no bias) at higher DOC (Figure 4-1). Summer split sample differences
between the two laboratories were typically (interquartile range) < 0.4 mg/L Very similar seasonal audit
sample means (within 0.2 mg/L) also indicate a lack of laboratory bias for DOC < 5 mg/L.
4.3.5.6 Base Cations
The base cations (Na+, K+, Mg2+, and Ca2+) all had SDLs < 1.5/*eq/L, except for sodium in the
spring (3/*eq/L), Base cation seasonal audit sample mean concentrations were usually within 3% and
always within 8% of the target values (Table 4-2). Base cation sample duplicate (within-batch) precision
and natural audit sample (among-batch) precision almost always had %RSDs < 5% Only K+ in the fall
(11-13%) and the Seventh Lake Na+ audit sample in the summer (11%) had a %RSD > 10% (Table 4-4).
The imprecision in the summer Na+ data, however, is due to a single outlying value rather than to a
general scatter in the data. Summer split samples show very good agreement between the two labora-
tories for all the base cations (Figure 4-2). Seasonal mean audit sample concentrations were within
3/ieq/L of each other for all base cations except the Seventh Lake calcium audit sample, which had a
fall/spring (laboratory #1/laboratory #2) difference of 13/ieq/L (target Ca2+ = 252^eq/L). Sum of
base cation differences in the summer split samples between the two laboratories were typically < 11
[6/^eq/L in low ANC (< 50 ^eq/L) systems].
47
-------
4.3.5.7 Aluminum
MIBK-extractable aluminum (AIM|BK) was measured in ELS-I, therefore it was also measured in
ELS-II for comparison purposes. PCV-reactive aluminum (Alm), another measure of total monomeric
aluminum, was also measured in ELS-II and is used to calculate inorganic monomeric aluminum by
subtracting of organic monomeric aluminum (measured by PCV colorimetry after passing the sample
through a strong cation exchange column). The SDL of both aluminum species was usually between 10
and 20 HQ/L (Table 4-1), which should be kept in mind when interpreting the aluminum data because
many ELS-II samples have very low aluminum concentrations. It is uncertain whether samples with
aluminum below the SDL really contain any aluminum at all.
In interpreting accuracy, precision, and laboratory bias for the aluminum variables, it is important
to remember that many duplicate sample pairs and the Seventh Lake natural audit sample were very
close to or below the SDL and would be expected to be somewhat imprecise because the magnitude of
the signal is not much greater than the noise. Thus the Seventh Lake audit samples (Alm = 21 M9/U
AIMIBK = 16 jug/L) had %RSDs ranging from 14% to 103% for the aluminum variables. In the Big Moose
Lake audit sample (Alm = 193jug/L; AIM[BK = 166/ug/L), Alm was more precise (%RSD = 4-11%)than
AIM|BK (12-21%). A similar pattern of precision was seen in the field duplicate sample pairs (Table 4-3).
For Alm, Big Moose Lake mean audit samples were within 8% of the target value in all seasons, whereas
AIM|BK means were from 11% to 17% lower than the target value. Overall, Al data seem to be of better
quality than AIM|BK data. In this report, only the AIM|BK data are used to make between-year com-
parisons (fall 1984/1986), because Alm was not measured in the ELS-I fall 1984 survey.
4.5 CONCLUSIONS
In a final assessment, data quality can be considered good if it meets the initially established data
quality objectives (DQO). A detailed discussion of the ELS-II QA data in comparison to the ELS-II DQOs
is given in the ELS-II QA report (Mitchell-Hall et al., 1989). In brief, Mitchell-Hall et al. found that data
quality for a majority of analytes was within or very close to the DQOs. In terms of the 12 analytes
focused on in this report, only chloride and AIMJBK were problem variables. Rather than concentrating
on DQOs, we have tried in this section to present the QA data in relation to the degree of precision,
bias, and background contamination, in order to aid in data interpretation. In short, how do the
observed data precision, detectability, and laboratory bias affect the conclusions that can be drawn from
the data? Our overall conclusions are:
1. For ANC, pH, PCV-aluminum, base cations, and DOC, the data are of sufficient quality to make
reliable population estimates of seasonal lakewater chemistry.
2. Based on the sample QA data, sulfate and nitrate data appear to be of sufficient quality to make
reliable population estimates. The quality of sulfate and nitrate data from laboratory 1 (spring and
48
-------
half of summer), however, has an element of uncertainty due to the two reanalyses and best batch
selection processes that occurred during the special data assessment.
3. Chloride data quality from laboratory 1 were suspect and we believe that the data cannot be used
to make population estimates of temporal variability. Thus, spring and summer population esti-
mates were not made for chloride chemistry. No problems were found in chloride data from
laboratory 2 (all fall data) that would confound data interpretation. Thus, chloride population
estimates were made for the fall seasonal data and variability survey.
4. The poor quality of the laboratory 1 chloride data renders the typical QA procedures for between-
variable assessments (charge balance, calculated versus measured conductivity) meaningless.
Chloride is present in significant concentrations relative to the total anion charge in most lakes, so
that there were poor charge balances and conductivity checks in much of the laboratory 1 data.
Charge balance and conductivity checks for laboratory 2 data, however, were good (see Mitchell-
Hall et al., 1989).
5. Due to the low lakewater concentrations of nitrate and aluminum, the system decision limit (SDL)
must be considered when interpreting these data. The SDL for aluminum variables was typically
between 10 and 2Q/ig/L; for nitrate, the SDL was between 0.5 and 1 /zeq/L Lakewater concen-
trations below these SDLs are not significantly different from field blank concentrations at a 95%
confidence level.
6. MIBK-extractable aluminum is less precise and accurate than the other major ELS-ll variables
analyzed in this section (%RSD 10-20%).
7. Any assessment of temporal variability must take into account the variability inherent in sampling
and measuring the system. Temporal changes must be interpreted in light of the sampling varia-
bilities reported in this section.
49
-------
SECTION 5
RESULTS - TEMPORAL VARIABILITY IN THE NORTHEAST
5.1 OVERVIEW
This section presents the effects of temporal variability on lakewater chemistry in the northeastern
United States. Temporal variability was examined in terms of population estimates of the distribution of
chemical variables across the target lake population and the magnitude of observed chemical changes.
This section focuses on those variables of primary interest in acidic deposition research: ANC, pH,
sulfate, nitrate, base cations, DOC, and aluminum. Information on other variables is presented in
Appendix A.
5.1.1 Components of Temporal Variability
Three components of temporal variability were examined in Phase II of the Eastern Lake Survey
(ELS-II): between-year, within-season, and among-season temporal variability. Between-year variability
was examined by comparing the fall seasonal survey data collected in 1986 during the ELS-II with fall
data from the same lakes collected in 1984 during ELS-I. Within-season variability was assessed using
data from the Fall Variability Study. In the Fall Variability Study, lakes were sampled three times during
the fall index period at three independently selected locations on the lake (selected using a consistent
protocol). Among-season variability was determined by comparing the results of the ELS-II spring,
summer, and fall seasonal surveys undertaken in 1986.
5.1.2 Cumulative Distribution Functions
One of the most useful ways to present results from ELS-II, as was the case in other components
of the National Surface Water Survey (NSWS), is the cumulative distribution function (CDF). CDFs show
the distribution of the study variable in the target population and are used extensively in this report.
Appendix A presents CDFs for many of the ELS-II variables in each of the seasonal surveys. Over-
lapping CDFs are used in this section of the report to show the effects of the different temporal
components on the variables in question.
A CDF (illustrated in Figure 5-1) is a frequency distribution, interpreted as the proportion of ELS-II
target lakes having values of the attribute X less than or equal to the value x. The cumulative proportion,
F(x), is calculated as described in Section 2.4.2 (equation 6). To read the example CDF (Figure 5-1),
pick a value, x, for the variable of interest along the horizontal axis (ANC in this example) and read the
y-axis value of the curve, F(x), at this value. F(x) is the estimated cumulative proportion of lakes in the
50
-------
1.0
0.8
g
Q.0.6
I
1
3 0.4
I '
O
0.2
0.0
10
20
30
40 50
ANC
60
70
90
100
/L)
Figure 5-1. Hypothetical example of the type of cumulative distribution function plot used in the
ELS-II report.
51
-------
population with an ANC equal to or less than the selected value x. By overlaying F(x) lines from differ-
ent populations on the same CDF, it is easy to make population comparisons across the distribution. In
this section we use overlying CDFs to compare variable distributions in different seasons, years, and
sample visits within season.
In addition to CDFs, we also present scatter plots of the 1984/1986 and spring/fall data. Scatter
plots are useful in that they show how each individual lake varied between times, rather than how the
population changed, as is the case in the CDFs. The scatter plots, however, do not reflect the sample
weights and may give a misleading indication of how the population as a whole was changing. A statis-
tical analysis of 1984/1986 and spring/fall population differences using chi-square analyses is presented
in Appendix C.
5.1.3 Characteristics of the ELS-II Target Population
In the ELS-ll, 145 sample lakes represented 3,993 lakes in the target population with a total lake
?
area of 3,549 km . Lakes were sampled from five subregions, divided for the purposes of this report into
three regions: Adirondacks subregion, Poconos/Catskills subregion, and New England subregions
(Table 5-1). The majority of ELS-II lakes were in New York and Maine. ELS-II lakes are not uniformly
distributed across the Northeast (see Figure 2-1) because of the selection factors (Section 2.2) used in
defining the interest attributes for ELS-II. The median ELS-II lake was 6.5 m deep at an elevation of 275
m and had a surface area of 24 ha (Table 5-2). Most ELS-II lakes were drainage lakes (72%) or reser-
voirs (15%). The remaining 13% were seepage or closed lakes. In the fall, most lakes (85%) were well
mixed (top-bottom temperature difference < 1°C), whereas in summer, most lakes (66%) were strongly
stratified (> 4°C change). Spring stratification conditions were intermediate between fall and summer
(Table 5-2).
5.1.4 Aluminum Measurements
In the ELS-II, inorganic monomeric aluminum (Aljm) was determined as the difference between
total pyrocatechol violet (PCV) reactive aluminum and nonexchangeable (organic) PCV reactive
aluminum. Aljm is thought to be the form of aluminum most toxic to aquatic organisms (Baker and
Schofieid, 1982; J. Baker et al., 1990a; Driscoll et al., 1980) and is thus of most interest in terms of acidic
deposition effects. Alim, however, was not measured in ELS-I, preventing fall 1984/faII 1986 compari-
sons. Methyl-isobutyl-ketone (MIBK) extractable aluminum (AIM(BK), an alternate estimate of total
monomeric aluminum, was measured in both ELS-I and ELS-ll. Thus, we are able to make year-to-year
comparisons for aluminum with AIM|BK. For all other temporal comparisons, we use Aljm because it is of
more direct biological interest.
52
-------
Table 5-1. Breakdown of the ELS-ll Target Population3
Area n N A (km2)
SUBREGION
Adirondacks (1A) 36 754 1,004
Poconos/Catskiils (1B) 18 608 134
New England Subregions 91 2,631 2,411
Central (1C) 32 931 802
Southern (1D) 25 659 297
Maine (1E) 34 1,040 1,312
TOTAL ELS-H 145 3,993 3,549
STATE
Connecticut
Massachusetts
Maine
New Hampshire
New York
Pennsylvania
Rhode Island
4
21
45
16
41
14
4
144
459
1,475
406
901
477
131
45
223
1,454
640
1,024
101
64
n is the sample size; N is the estimated number of lakes in the target population; A is the estimated lake area (km?) in the
target population.
53
-------
Table 5-2. Physical Characteristics of the ELS-II Target Population
Variable
Lake Depth (m)
Elevation (m)
Lake Area (ha)
Watershed Area (ha)
Watershed: Lake Area
Residence Time (yr)
Temperature Stratification
(Top-Bottom)
< rc
1-2°C
2-4°C
> 4°C
% Drainage Lakes0
% Seepage Lakes0
% Closed Lakes0
% Reservoirs
Minimum Q.,3 Median Q3a
1,2 4,0 6.5 10.5
2 116 275 459
4 10 24 92
13 145 425 1,417
2 7 15 32
0.001 0.13 0.30 0.84
Percent of Lakes
Spring Summer
47 18
7 9
23 7
23 66
72
9
4
15
Maximum
30.5
791
1,619
81,420
2932
10.0
Fail
85
4
7
4
Q, = 25th percentile, Q^ = 75th percentile.
Residence time (RT) calculated as: RT = /ijake Z / [R(\s - ^ake) + P(^ake)]' wnere Aake = lake area' Am = watershed
area, Z = site depth, R = runoff, and P = precipitation.
Drainage lakes have mapped outlets, seepage lakes have no mapped inlets or outlets, and closed lakes have inlets but no
mapped outlets.
54
-------
5.2 BETWEEN-YEAR VARIABILITY (FALL 1986 VS. FALL 1984)
As only two different years were sampled in both phases of the ELS, it is not our intent to make a
rigorous statement about interannual variability. However, the comparison of ELS-I (fall 1984 sample)
and ELS-II (fall 1986 sample) data does provide useful information about annual difference in our ability
to estimate attributes of regional lake populations. In this subsection, only ELS-II data collected from the
ELS-I fall index site as defined on the original data sheet are presented. Data collected during the two
subsequent lake visits of the fall variability study were not used to assess between-year variability. Thus,
the fall 1984 and 1986 chemistry data analyzed in this subsection were collected from the same general
lake location.
In terms of hydrology, 1984 was a "wet water year" in the northeastern United States. Runoff for
the year was above average to excessive throughout the Northeast (upper 25th percentile compared to
long-term records; Blackey et al., 1985; Firda et al., 1985; Gadoury et al., 1985; Haskell et al., 1985). In
general, these excesses were due to large precipitation inputs in the summer of 1984, which caused
some flooding and filled reservoirs to capacity. Runoff tended to return to more normal levels during the
fall of 1984. In contrast, runoff throughout the 1986 water year was average to above average (Firda et
al., 1987). Water levels In lakes in the Adirondacks and Poconos/Catskills were at or slightly above
long-term averages and fall 1986 runoff in eastern New York was about 100-110% of 30-year averages
(Firda et al., 1988).
5.2.1 Population Estimates of Acidic and Low ANC Lakes
The percentage of acidic (ANC< O/xeq/L) ELS-i lakes in fall 1984 (6.2%) was almost the same as
that calculated for the fall of 1984 using only the ELS-II (7.7%) subset of sample lakes (Table 5-3). ELS-II
and ELS-I percentages are similar, despite the exclusion of high ANC (> 400peq/L) lakes in ELS-II,
because a number of acidic and low ANC lakes were excluded from ELS-II because they were shallow
or historically had been limed. As can be seen in Table 5-3, an estimated 134 of the estimated 441
acidic ELS-I lakes were excluded from the ELS-II target population. All these acidic lakes were excluded
because they were too shallow (site depths < 1.5 m).
There were roughly 10% more acidic ELS-II lakes in the fall of 1986 than in fall 1984 (Table 5-3).
On the other hand, there were about 10% fewer lakes with ANC < 50, 100, and 200 /ueq/L in fall 1986.
Overall, these between-year differences correspond to small changes in the percentage of the target
population with ANC below reference values. The reference values for pH and ANC used in Table 5-3
are those that are commonly used in acidic deposition research. Lakes with ANC < 0 have lost the
capacity to neutralize incoming acids and lakes with ANC< 50/ueq/L have been defined as extremely
sensitive to acidic deposition (Schindler, 1988).
55
-------
Table 5-3. Population Estimates of the Number (N) and Percentage of Lakes with ANC and pH
below Reference Values in the Fall of 1984 (ELS-I) and 1986 (ELS-II) in the Northeastern
United States
Reference
Values3
ANC<0/ieq/L
ANC < 50
ANC < 100
ANC < 200
pH<5.0
pH<5.5
pH<6.0
pH<6.5
Sample size
Population size
ELS-
Fall
N
441
1,556
2,632
4,324
240
613
916
1,868
768
7,157
I Lakes
1984b
%
6.2
21.7
36.8
60.4
3.4
8.6
12.8
26.1
ELS-II Lakes
N
307
1,089
1,873
3,005
155
471
660
1,346
145
3,993
Fall 1984°
%
7.7
27.3
46.9
75.3
3.9
11.8
16.5
33.7
Fall
N
343
1,000
1,711
2,760
167
478
111
1,324
145
3,993
1986d
%
8.6
25.1
42.8
69.1
4.2
12.0
19.5
33.2
Reference values for ANC and pH are those that are commonly used for classification purposes in acidic deposition
research.
Estimates based on ELS-I chemistry measured in (all 1984 for the ELS-I target population.
Estimates based on ELS-I chemistry measured in fall 1984 at the 145 ELS-II lakes (population estimates made using ELS-II
weights).
Estimates based on ELS-II fall seasonal survey.
56
-------
ELS-II population estimates of the number of lakes with pH below reference values were also very
similar in fall 1984 and fall 1986 (Table 5-3). Percentage estimates were within 1% of each other for
lakes with pH < 5.0, 5.5, and 6.5. Based on ELS-II data, conclusions about the acid-base status of lakes
in the northeastern United States would have been similar whether the surveys had been implemented in
fall 1984 or fall 1986.
5.2.2 Chemical Changes
5.2.2.1 ANC/pH
Overall, CDFs for ANC (Figure 5-2) and pH (Figure 5-3) show similar distributions for fall 1984 and
fall 1986. ANC values were very similar between the two years at low ANC (< 50 /ieq/L). At higher
ANC values, 1986 ANC tended to be slightly higher than 1984 ANC (Figure 5-2). The median change in
ANC between the two years was a gain of 15 /zeq/L from 1984 to 1986 (Table 5-4). One-fourth of the
lakes increased by more than 33Meq/Lfrom fall 1984 to fall 1986. All of these lakes, however, had ANC
over 50 /ueq/L In low ANC systems (< 50 peq/L), the median 1984/1986 ANC change was only 2
/ueq/L, and over half the low ANC lakes changed by less than 6/zeq/L Between-year differences in pH
were low, with 1984 pH levels tending to be slightly higher than 1986 pH levels (Figure 5-3). The median
pH change in ELS-II lakes was 0.06 pH units, with an interquartile range of 0.18 pH units (Table 5-4),
Note that the pooled standard deviation of duplicate lake samples was 5.1 /teq/L for ANC and 0.02 pH
units in fall 1986 (Table 4-3). Differences smaller than these should be interpreted with caution.
5.2.2.2 Inorganic Acid Anions
Distribution of sulfate concentration in ELS-II lakes were nearly identical in 1984 and 1986 (Figure
5-4). Fall 1984/Fall 1986 scatter plots, however, show some large deviations from the 1:1 line at sulfate
concentrations > 200/ieq/L. At these relatively high concentrations, these lakes are dominated by
internal watershed sources of sulfate rather than deposition sulfate sources (L. Baker et al., 1990).
Compared to deposition, which is a relatively stable sulfate source, fluxes of watershed-derived sulfate
are more variable because they are more dependent on flowpaths, weathering rates, and other blogeo-
chemical processes. Thus, lakes with watershed sources of sulfate are likely to experience greater
variability over time than lakes whose sulfate is derived primarily from deposition. Overall, in most lakes
the year-to-year changes in sulfate concentration were small. The median change was a decrease of 3
A*eq/L from fall 1984 to fall 1986, and over half of both the low and high ANC lakes changed by less than
10/zeq/L Sulfate duplicate sample precision (pooled SD) in fall 1986 was 2.6 ^eq/L.
More lakes had nitrate concentrations > 1 /^eq/L in fall 1986 than fall 1984 (Figure 5-5). In both
years, however, most lakes had low nitrate concentrations and thus between-year nitrate changes were
small (majority < 2 /*eq/L; Table 5-4). The largest observed nitrate changes were a 13 neq/L increase
and a 7/^eq/L decrease from 1984 to 1986. The main difference between the two years was an
57
-------
0.0 i
-50 0
50 100
150 200
ANC tu,eq /L)
250
300 350 400
cr
O
400
350
300-
0-
-50-t
-50
50 100 150 200 250 300 350
Fail 1986 ANC Oueq /L)
400
Figure 5-2. Population distribution and comparison of fall 1986 ANC with fall 1984 ANC in
ELS-II lakes.
58
-------
4.00 4,25 4.50 4.75 5.00 5.25 5.50 5.75 6.00 6.25 6.50 6.75 7.00 7.25 7.50
pH
4.00-I*
4.00 4.25 4.50 4.75 5.00 5.25 5.50 5.75 6.00 6.25 6.50 6.75 7.00
Fall 1986 pH
7.25 7.50
Figure 5-3. Population distribution and comparison of fall 1986 pH with fall 1984 pH in ELS-II
lakes.
59
-------
Table 5-4. Population Characteristics of Between-Year Chemical Changes (Fall 1984 - Fall 1986)
in ELS-II Lakes3
Fall 1984b - Fall 1986
Variable
All ELS-II (N =
ANC
pH
S042'
NO3"
cr
Base cations
DOC
A'MBK
Fall 1986 ANC
ANC fceq/L)
PH
S042'
NO3"
cr
Base Cations
DOC
Al
Minimum
3,993)
-112
-0.37
-101
-13
-201
-338
-4.6
-166
<50/*eq/L(N = 1,000)
-16
-0.37
-101
-10
-156
-321
-4.2
-166
Q7
-33
-0.03
-4
-2
-18
-74
-0.6
-6
-6
-0.10
-8
-1
-5
-38
-0.4
-20
Median
-15
0.06
3
-0.4
-3
-38
-0.1
-2
-2
0.04
2
0
0
-18
0
-7
Q3
-2
0.15
10
-0.1
0.2
-16
0.3
3
4
0.14
9
0
1
-9
0.4
2
Maximum
39
1.2
44
7
139
69
2.7
79
39
1.2
44
7
139
34
2.7
79
All units are in ^eq/L except for pH DOC (mg/L), and aluminum (/ig/L).
b Fall 1984 data is from ELS-I.
Qj = 25th percentile, C^ = 75th percentile.
60
-------
1.0 f
0,8
g
t
a
2
a.
••8
"5 0,4
E
d
0.2
o.o-I
50 100 150 200
Sulfate {/^eq /L)
250
300
350
cr
-------
1.0 f
Nitrate Cueq /L)
cr
0)
1
I
10-
5-
5 10 15
Fall 1986 Nitrate (Meq /L)
20
Figure 5-5. Population distribution and comparison of fall 1986 nitrate with fall 1984 nitrate in
ELS-II lakes.
62
-------
increase in the number of lakes with nitrate between 1 and S^eq/Lfrom fail 1984 to fall 1986. Accord-
ing to chi-square analysis, nitrate concentrations in the fall 1986 population were different than those in
the fall 1984 population in all three ELS-II sample clusters (see Appendix C, Table C-1).
Similar chloride distributions were observed for lakes with low chloride values (< IQQ/ueq/L) in fall
1984 and 1986 (Figure 5-6). In lakes with higher chloride concentrations, 1986 values were slightly
higher than 1984 values. As with sulfate, the biggest year-to-year changes were observed in systems
with the highest concentrations. These high chloride lakes receive either extensive inputs of marine salts
or internal watershed inputs of chloride (e.g., road salt). The majority of low ANC lakes showed only
small changes (< 5^eq/L) in chloride (Table 5-4).
5.2.2.3 Sum of Base Cations
Sum of base cation concentrations in the fall of 1986 were consistently higher than those observed
in the fall of 1984 (Figure 5-7). The median base cation increase was 38/zeq/L (18/ueq/L in low ANC
lakes; Table 5-4). Three-fourths of the ELS-II lakes had base cation increases of more than 16/^eq/L
between fall index periods in 1984 and 1986. These differences are most likely attributable to the
different hydrological conditions in the two years. 1984 was a "wetter" year, resulting in more dilute
conditions and thus lower base cations. According to chi-square analysis, the fall 1986 population had
significantly different (p < 0.01) base cation concentrations than the fall 1984 population (see Appendix
C, Table C-1).
5.2.2.4 DOC/Aluminum
DOC distributions for fall 1984 and fall 1986 were nearly identical (Figure 5-8). Fall 1984/Fall 1986
scatter plots show no tendency for higher DOC in either year, and the scatter in the relationship was
fairly consistent across all DOC levels. Overall, the median change in DOC (0.1 mg/L) was very small
and was within the precision of the sampling process. The majority of lakes changed by less than 0.6
mg/L (Table 5-4).
MIBK-extractable aluminum (AIM|BK) is an estimate of total monomeric aluminum and was the only
aluminum species measured in ELS-I. AIM)BK population distributions were about the same in fall 1984
as in fall 1986 (Figure 5-9). As with nitrate, most of the lakes had low AIMIBK concentrations and thus
had small year to year changes (median = 2 /zg/L; Table 5-4). The majority of low ANC lakes also had
small changes in AIMIBK (< 20/ug/L).
5.2.3 Conclusions: Between-Year Variability
Fall 1984/Fall 1986 changes in ANC and pH were small and estimates of the number of lakes with
ANC and pH below reference values commonly used in acidic deposition research were very similar
between the two years. Based on ELS-II data, major conclusions about the acid-base status of lakes in
the northeastern United States would have been very similar, whether the surveys had occurred in fall
63
-------
1.01
cr
T3
_o
6
600
500-
400
300
200-
100-
o-l
— Fall 1984
Fall 1986
,1
100
200 300 400
Chloride (^eq /L)
500
600
1:1
100
200 300 400
Fall 1986 Chloride (^eq /L)
500
600
Figure 5-6. Population distribution and comparison of fall 1986 chloride with fall 1984 chloride
in ELS-ll lakes.
64
-------
1.0
0.8-
g
OL
"5 0.4-
I
o
0,2-
0.0-1
Fall 1984
Fall 1986
100 200 300 400 500 600 700
Base Cations (^.eq /L)
800
900 1000
O"
CD
to
c
O
CD
8
CQ
3
CD
1000
900
800
700
600
500
400
300
200
100
1:1
0 100 200 300 400 500 600 700 800 900 1000
Fall 1986 Base Cations (/^eq /L)
Figure 5-7. Population distribution and comparison of fall 1986 sum of base cations with fall
1984 sum of base cations in ELS-II lakes.
65
-------
1.0 i
4567
DOC (mg /L)
8 9 10
O)
H
O
o
Q
3
if
34567
Fall 1986 DOC (mg /L)
8 9 10
Figure 5-8. Population distribution and comparison of fall 1986 DOC with fall 1984 DOC in
ELS-II lakes.
66
-------
0.04.
300
250
O)
^
<
GO
CO
O3
50
100 150 200
MIBK Al Gag /L)
100 150 200
Fall 1986 MIBK Al (^g /L)
250
300
250
300
Figure 5-9. Population distribution and comparison of fall 1986 MIBK-extractable aluminum with
fall 1984 MIBK-extractable aluminum in ELS-II lakes.
67
-------
1984 or fall 1986, Sulfate and DOC distributions were very similar in both years. Base cation and
chloride concentrations tended to be higher In 1986 than In 1984, probably because of the dryer condi-
tions in 1986, Nitrate and AIM|BK concentrations were very low in most lakes in both years, so that
between-year differences were usually small. In lakes with elevated nitrate and AIMIBK. neither year
appeared to have higher concentrations than the other.
5.3 WITHIN-SEASON VARIABILITY (FALL 1986)
A subset of 41 lakes was sampled three times at three independently selected locations on each
lake during the fall index period of 1986. One visit was made at the mapped ELS-I index site, whereas
the other two visits were made, following the established protocol, to the "deepest spot on the lake" in
the judgement of independent field crews. The average time between the first and third sample visits
was 12 days (SO = 7). Thus the fall variability study has a spatial component as well as the within-
season temporal component. These 41 fall variability lakes represent 1,012 lakes with a surface area of
551 km2 in the ELS-II target population. As in the entire ELS-ll population, the vast majority of the fall
variability lakes were drainage lakes (59%) or reservoirs (21%). Of these 41 lakes, 17 were in the
Adirondack Mountains, and 24 were in the coastal areas of southern New England (within 61 km of the
ocean). In examining within-season variability, no spatial (regional) patterns were observed in the data,
so we will not report separate results for the two groups of fall variability lakes. In this section, Fall
Variability Study data are presented in two ways: (1) by CDFs showing the population distributions
during visits 1, 2, and 3, and (2) by scatterplots showing the relationships between all possible study
pairs (visit 1/visit 2, visit 2/visit 3, visit 1/visit 3).
5.3.1 Chemical Differences
5.3.1.1 ANC/pH
Distribution of ANC and pH were very similar among samples from the three lake visits (Figures
5-10 and 5-11). Visit 2 was the visit to the mapped 1984 ELS-I fall index site. Median ANC and pH
values were within 2 peq/L and 0.05 pH units of each other among the three lake visits. The median
standard deviation of the 41 sets of lake triplicate (three visits) samples was 6/*eq/Lfor ANC and 0.05
units for pH (Table 5-5). Over three-fourths of the fall variability lake population had triplicate standard
deviations < 12/zeq/L and 0.07 pH units. From the CDFs of ANC and pH, and from the scatterplots of
all possible visit pairs, it is apparent that population estimates of the proportion of lakes with ANC or pH
below reference values were not greatly affected by fall temporal and site selection variability. Overall
conclusions about the acid-base status of northeastern lakes would have been the same if sampling had
occurred only during visit 1, 2, or 3. Therefore, the fall index sample appears to be quite stable for ANC
and pH.
68
-------
1.01
0.0
ELS-I! Fall Variability Study
-100 -50
50 100 150 200
ANC (^eq A.)
250
300 350
1.0 -f
I Visit 3
4.00 4.25 4.50 4.75 5.00 5.25 5.50 5.75 6.00 6.25 6.50 6.75 7.00 7.25 7.50
Figure 5-10. Population distribution of the three sample visits in the ELS-II Fall Variability Study:
(a) ANC, and (b) pH.
69
-------
Fall Variability Study Pairs
cr
0
Q
400
300
200
100
o-
-100
Ca)
-100 0 100 200 300 400
ANC
81
0)3
6
pH
Figure 5-11. Scatterplot of all possible Fall Variability Study pairs (visit 1/visit 2, visit 2/visit 3,
and visit 1/visit 3) for (a) ANC, and (b) pH.
70
-------
Table 5-5. Population Distribution of the Standard Deviation of the Three Lake Samples Collected
during the Fail Index Period in the ELS-II Fall Variability Study3
Variable
ANC
PH
so42-
NO3
cr
Base Cations
DOC
A'im
Grandb
Mean
61
5.9
124
4
161
374
3.8
42
Triplicate Standard Deviation0
Minimum
0.3
0.01
0.5
0.02
0.1
1.5
0.01
0.50
QI
2.1
0.03
2.8
0.1
0.8
4.4
0.09
1.7
Median
6.1
0.05
4.1
0.3
4.9
7.7
0.17
3.3
%
12
0.07
8.6
0.6
18
18
0.24
7.8
Maximum
19
0.11
57
4.8
96
35
0.46
13
All units are in/ieq/L except for pH, DOC (mg/L), and aluminum
Grand mean is the mean of all the triplicate means in the Fall Variability Study (n = 41).
Population distribution of the standard deviation of the three lake samples collected in the Fall Variability Study.
71
-------
5.3.1.2 Inorganic Acid Anions
Sulfate changes were minimal among the three lake visits of the fall variability study (Figures 5-12
and 5-13). Median and quartile values were within 10/ieq/L of each other among the three lake visits.
Maximum suifate values showed the greatest among-visit difference (328-438/xeq/L), probably due to
varying inputs of watershed suifate. Over three-fourths of the fall variability lake population are estimated
to have fall variability (standard deviation of the three lake samples) less than 9peq/L
Nitrate distributions were fairly similar among the three lake visits with most (> 60%) lakes having
very low nitrate (< 3/ieq/L) concentrations (Figures 5-12 and 5-13). The 75th percentile for nitrate
varied from 8.8 to 10.0 /xeq/L and maximum nitrate varied from 19 to 27 fieq/L among the three visits.
Most (> 75%) of the fall variability lakes had triplicate nitrate standard deviations < 1 /ieq/L and all of
them had standard deviations < 5/^eq/L (Table 5-5).
Median chloride concentrations were almost identical (138-140/^eq/L) among the three lake visits
(Figure 5-13). As with the other acid anions, the maximum variability among the three lake visits was
observed at the high end of the distribution (Figure 5-13). Overall, the median chloride triplicate
standard deviation in the fall variability lakes was 5/ieq/Land over 75% of the population had standard
deviations below 18/ieq/L (Table 5-5).
5.3.1.3 Sum of Base Cations
CDFs of the three lake visits during the Fall Variability Study almost completely overlap for the sum
of base cations (Figure 5-14). Population minimums, quartiles, medians, and maximums are nearly iden-
tical among the three visits. The median triplicate standard deviation for the sum of base cations was
7.7 jueq/L (maximum = 35 /^eq/L) for the fall variability lake population (Table 5-5) scatterplot. The fall
index appears to be extremely stable for base cations.
5.3.1.4 DOC/Aluminum
DOC values are very similar among all the possible pairs among the three lake visits in the Fall
Variability Study (Figure 5-15). CDFs for DOC showed almost no differences among the three lake visits
(data not shown). The largest triplicate DOC standard deviation in any of the lakes was 0.46 mg/L
(Table 5-5). The median standard deviation was only 0.17 mg/L. The fall index appears to be extremely
stable for DOC.
Inorganic monomeric aluminum (Alfm) concentrations were low in most fall variability lakes (Figure
5-14). The SDL for Aljm was 17 ^Q/L and values below this value are not significantly different than field
blanks. All the median Aljm values for the three lake visits are below the SDL. At all concentrations,
variability among the three lake visits was small (Figure 5-15). Triplicate visit standard deviation for A!jm
ranged from 0.5 to 13 M9/L in the fall variability lakes. The fall index appears to be extremely stable for
72
-------
ELS-II Fall Variability Study
1.0-r
0.8
.0
10.6
2
CL
I
"3 0.4
o
0.2
o.o-l
50 100 150 200 250
Sulfate Cu,eq /L)
300
350
400
450
1.0
0.8
O
t
8.0.6
O
i
'
"5 0.4
E
O
0.2
0.0 4-
10 15
Nitrate {/^eq /L)
20
25
30
Figure 5-12. Population distribution of the three sample visits in the ELS-II Fall Variability Study:
(a) sulfate, and (b) nitrate.
73
-------
Fall Variability Study Pairs
cr
-------
ELS-II Fall Variability Study
1.0 f
0.0 i
100 200 300 400 500 600 700 800 900 1000 1100 1200 1300
Base Cations (/ieq /L)
1.0
0.8
c
o
"t±
§.0.6
O
-------
Fall Variability Study Pairs
§
en
o
'IS
O
CD
a
CQ
C/3
O)
o
•c
a>
o
o
.9
CO
1500
1200
900
600
300
(a)
0 300 600 900 1200 1500
Sum Base Cations (ueq/L)
16
12
01
O
O
Q
250
200
Cb)
0 50 100 150 200 250
Inorganic Monomeric AI
4 8 12
DOC (mg/L)
16
Figure 5-15. Scatterplot of all possible Fall Variability Study pairs (visit 1/visit 2, visit 2/visit 3,
and visit 1/visit 3) for (a) sum of base cations, (b) DOC, and (c) inorganic mono-
meric aluminum.
76
-------
5.3.2 Conclusions: Robustness of the Fall Index
Based on data from the ELS-II Fall Variability Study, a single fall index sample is a robust estimator
of conditions during the fall index period (after fall turnover). Overall, variability among the three lake
visits was very small. Population estimates of the proportion of lakes with ANC or phi below reference
values were not greatly affected by fall temporal and site selection variability. Overall conclusions about
the acid-base status of northeastern lakes would have been the same if sampling had occurred only
during visit 1, 2, or 3. ANC, pH, sum of base cations, DOC, and Aljm showed almost identical distribu-
tions among the three lake visits over all observed concentrations. Acid anions (sulfate, nitrate, and
chloride) showed very little differences among the three lake visits except at the highest observed
concentrations where some larger within-season differences were evident. Sample processing and
analytical variability are components of within-season variability. Section 6.1 provides a comparison of
the magnitude of analytical and temporal variability.
5.4 AMONG SEASON VARIABILITY (SPRING, SUMMER, AND FALL 1986)
Among-season variability is assessed by comparing the results of the 1986 ELS-II spring, summer,
and fall seasonal surveys. All of these seasonal samples were collected from the same general location
in the lakes as the 1984 ELS-I sample. Fall samples were collected after turnover (October to mid-
November) when conditions were isothermal (Table 5-2) and chemical variability low (Section 5.3).
Summer samples were collected from mid-July to mid-August, when most lakes were stratified (Table
5-2). Only epilimnetic sample data are discussed in this section. Spring data were collected in April as
soon after ice-out as possible. However, since peak lake discharge usually precedes ice-out, the ELS-II
spring sample does not represent peak flow conditions when episodic acidification effects would be
most severe. A detailed discussion of the ELS-II spring sample in relation to lake discharge is given in
Section 6.3.3. In summary, the ELS-II fall sample represents epilimnetic conditions in the profundal zone
after fall turnover. The ELS-II spring sample represents post-snowmelt (2-3 weeks), non-episodic spring
conditions.
5.4.1 Population Estimates of Acidic and Low ANC Lakes
One of the major objectives of ELS-II was to estimate the number of systems that would be acidic
or have low ANC in seasons other than the fall index period sampled in ELS-I. ELS-II population esti-
mates show that there were 24% more acidic lakes (ANC < 0) in the spring (an estimated 424 lakes) than
in the fall (an estimated 343 lakes; Table 5-6). This corresponds to an increase in the percentage of
acidic ELS-II lakes from 9% to 11%. It should be noted that the number of acidic lakes in the spring is
77
-------
Table 5-6. Population Estimates (N ± Standard Error) of the Number of ELS-II Lakes with ANC
below Reference Values in the Three Seasonal Surveys3
Reference
Value
ANC<0
ANC< 50
ANC< 100
Region
Adirondacks
Poconos/Catskills
New England
All ELS-II
Adirondacks
Poconos/Catskills
New England
All ELS-II
Adirondacks
Poconos/Catskills
New England
AH ELS-II
Spring
1986
184±42
67139
173145
424169
440182
185161
713+98
1,338+128
555199
31 1 1 94
1,5981169
2,4651192
Summer
1986
160139
67±39
117+37
343+64
346+ 62
185161
558+88
1,0891113
456± 74
212+66
1,1151117
1,7831130
Fall
1986
135+37
67139
142140
343+64
346+62
141154
514+85
1,0001109
456+ 74
185161
1,0701117
1,711 + 129
Total ELS-II population size = 3,993 lakes, 754 in the Adirondacks, 608 in the Poconos/Catskills, and 2,631 in New England.
78
-------
less than the upper 95% confidence limit of the number of acidic ELS-II lakes in the fall (343 + 1.65 * 64
= 449 lakes; Table 5-6). There were also increases in the numbers of low ANC lakes in the spring
relative to fall; the percentage of ELS-II lakes with ANC< 5Q/*eq/L increased from 25% to 34% (fall to
spring) and the percentage with ANC< 100/ieq/L increased from 43% to 62%. Summer conditions
were usually more similar to fall conditions than to spring conditions. Broken down by region, the
Adirondacks showed the greatest relative fall-to-spring increase in the number of acidic lakes, but the
smallest relative increase in the number of lakes with ANC < 50 or 100/ieq/L {Table 5-6).
There was a 13% increase in the number of lakes with pH < 5.0 in the spring relative to fall and
there were slightly more lakes with pH < 5.0 in the summer than in the spring (Table 5-7). The largest
numbers of lakes with pH < 5.5 and pH < 6.0 were observed in the spring in all regions. The lowest
number of lakes with pH < 5.5 was observed in the summer, whereas the lowest number of lakes with
pH< 6.0 was observed in the fall. In a fall/spring comparison, the percentage of ELS-II lakes with: (1)
pH< 5.0 increased from 4% to 5%, (2) pH< 5.5 increased from 12% to 14%, and (3) pH< 6.0 increased
from 19% to 27% (Table 5-7). Similarly small increases in the percentages of acidic and low pH lakes in
the spring relative to fall were also observed in the 1,715 lakes sampled in Massachusetts during Phase
II of the Acid Rain Monitoring Project (ARM). For ARM lakes > 4 ha, 9% were acidic and 27% had pH
< 6.0 in the spring (April) versus 6% and 23%, respectively, in the fall (Ruby et al., 1988). For smaller
lakes, spring/fall percentage differences were even smaller.
5.4.2 Seasonal Chemical Differences
Seasonal differences are presented by overlaying distributions of the study variable in different
seasons on one plot. Also, seasonal minimums, quartiles, medians, and maximums are presented for
each season using floating bar diagrams. In more detailed analyses, we have concentrated on pre-
senting spring/fall differences because spring samples had the lowest ANC in the ELS-II survey, and we
wished to compare those values to the index period measured in ELS-I. In examining seasonal chemical
differences, there were different patterns of seasonal change in low and high ANC lakes. Thus, we have
presented spring/fall comparisons for lakes greater than and less than 50^eq/L as well as for all lakes
(Table 5-8). An ANC cutoff of 50 ^eq/L was chosen to define those lakes of most interest with respect
to acidic deposition effects. Paleolimnological data in the Adirondacks have shown that most ELS lakes
with current ANC < 50 neq/L have acidified since pre-industrial times, whereas those with ANC > 50
fjteq/L have increased in ANC (Sullivan et al., 1990). Also, an ANC level of 50/ieq/L has been
commonly used to define "extremely sensitive" systems (Schindler, 1988). There were also regional
differences in spring/fall changes in low ANC lakes, so regional estimates are also presented (Table 5-9).
All among-season differences should be interpreted in light of the laboratory bias (Table 4-5) and
precision estimates presented in Section 4.
79
-------
Table 5-7. Population Estimates (N± Standard Error) of the Number of ELS-II Lakes with pH
below Reference Values in the Three Seasonal Surveys8
Reference
Value
pH < 5.0
pH < 5.5
pH < 6.0
Region
Adirondacks
Poconos/Catskills
New England
All ELS-II
Adirondacks
Poconos/Catskills
New England
All ELS-II
Adirondacks
Poconos/Catskills
New England
All ELS-II
Spring
1986
111133
12112
69130
191145
233145
79141
2641 60
576181
368164
185161
51 1 1 85
1,0641112
Summer
1986
123135
12112
69130
204146
197±43
67139
173145
4361 70
302155
151167
3791 72
8321107
Fall
1986
98132
12112
56±28
167+43
197143
94147
188147
4781 75
302155
106148
3691 70
777195
a Total ELS-II population size = 3,993 lakes, 754 in the Adirondacks, 608 in the Poconos/Catskills, and 2,631 in New England.
80
-------
Table 5-8. Population Characteristics of Between-Season Chemical Changes (Spring 1986 - Fall
1986) in ELS-II Lakes3
Spring 1986 -Fall 1986
Variable
AH ELS-II
ANC
pH
S042'
N03~
Base cations
DOC
^im
Fall ANC<50jieq/Lc
ANC
pH
S042"
N03'
Base cations
DOC
*im
Fall ANC > 50/*eq/Lc
ANC
pH
S042'
N03-
Base cations
DOC
^im
Minimum
-177
-0.97
-70
-6
-335
-7.0
-101
-31
-0.76
-70
-2
-335
-7.0
-101
-177
-0.97
-65
-6
-279
-5.8
-24
QID
-84
-0.34
-6
-€.1
-85
-1.5
-10
-10
-0.28
-9
1
-22
-0.8
-9
-97
-0.39
-5
0
-112
-1.7
-10
Median
-44
-0.18
3
2
-49
-0.5
-7
-3
-0.11
-2
2
-6
-0.4
-1
-60
-0.19
3
2
-69
-0.6
-8
Q?
-10
-0.02
16
6
-16
0.1
-2
4
0.07
6
9
2
0.2
42
-35
-0.06
20
5
-35
0.1
-4
Maximum
21
0.7
97
33
32
5.0
217
21
0.70
42
33
32
5.0
217
11
0.31
97
14
8
3.3
29
All units are in peq/L except for pH, DOC (mg/L). and aluminum (pg/L).
QI = 25th percentile, Qj = 75th percentile.
c There were an estimated 1,000 ELS-II lakes with fall ANC < 50/ieq/L and 2,993 lakes with fall ANC > 50^eq/L.
81
-------
Table 5-9. Regional Population Characteristics of Between-Season Chemical Changes (Spring
1986 - Fall 1986) in Lakes with Fall 1986 ANC< 50/ieq/La
Fall ANC<50/ieq/L
Spring 1986 - Fall 1986
Variable
Adirondacks0
ANC
PH
so42-
NO3~
Base cations
DOC
A'inn
Poconos/Catskillsc
ANC
pH
so42-
N03"
Base cations
DOC
*.m
Minimum
-28
-0.50
-25
1
-16
-2.6
-101
-31
-0.58
-71
-1
-83
-1.2
-29
QID
-10
-0.24
-14
6
-5
-1.4
-5
-5
-0.33
-27
0
-7
-1.0
-6
Median
-7
-0.08
-6
13
1
-0.7
36
-4
-0.26
-6
1
-6
-0.4
15
Q3
0
0.04
1
19
7
-0.3
71
-1
-0.12
9
6
5
0.1
74
Maximum
15
0.27
9
33
25
5.0
217
19
0.7
39
24
32
0.2
202
New England Subregions0
ANC
pH
so42-
N03~
Base cations
DOC
A'im
-23
-0.76
-42
-2
-335
-7.0
-24
-12
-0.30
-5
0
-34
-0.6
-12
-1
-0.05
3
1
-19
-0.1
-8
10
0.08
12
2
-6
0.4
19
21
0.33
42
10
19
1.3
48
All units are in peq/L except for pH, DOC (mg/L), and aluminum
Q, = 25th percentile, O^ = 75th percentile.
There were an estimated 346 lakes in the Adirondacks, 141 lakes in the Poconos/Catskills, and 514 lakes in New England
with fall ANC < 5Qjieq/L in the ELS-H.
82
-------
5.4.2.1 ANC/pH
Seasonal distributions of ANC show that ANC was lowest in spring, highest in fall, and
intermediate in summer (Figure 5-16), ANC differences were most pronounced at higher ANC (> 50-100
/xeq/L). The typical ELS-II lake (median) had ANC 44^eq/L lower in spring than fall (Table 5-8). In
lakes with ANC< 50peq/L, the median spring minus fall ANC change was only -3/ieq/L, and over half
the lake population changed by < 10/ieq/L On the other hand, high ANC lakes decreased by a
median 60 ^teq/L between spring and fall. Three-fourths of the high ANC lakes had a spring ANC
depression > 35/*eq/L Chi-square analysis indicates that the spring population had a significantly
different ANC distribution than the fall population in all ELS-II sample clusters (see Appendix C, Table
C-2).
As with ANC, pH was generally lowest in the spring, highest in the fall, and intermediate in the
summer (Figure 5-17). The largest seasonal differences in pH were observed in higher pH lakes. Note
that the maximum observed pH (9.08) occurred in the summer, probably due to high rates of photosyn-
thesis. The 75th percentile in summer, however, was lower than that measured in the fall (Figure 5-18).
In the ELS-II target population, the median pH decrease from fall to spring was 0.18 (Table 5-8).
Fall/spring pH changes were similar in both high and low ANC lakes. No lake changed by more than 1
pH unit, and most lakes changed by < 0.5 pH units. Fall/spring changes in ANC and pH in low ANC
systems were fairly similar between regions (Table 5-9).
5.4.2.2 Inorganic Acid Anions
Seasonal distributions of sulfate concentration were similar in spring, summer, and fall (Figures
5-19 and 5-20). Fall/spring changes in sulfate concentration were small in both high and low ANC lakes
(median = 3/xeq/L; interquartile range = -6 to 16/ieq/L; Table 5-8).
Seasonal distributions of nitrate concentration show the lowest concentrations in the summer and
the highest concentrations in the spring (Figure 5-21). Fall values are intermediate between spring and
summer values. Over 75% of the ELS-II lakes have nitrate concentrations < 1 Meq/L in the summer. In
contrast, the 75th percentile for nitrate in the spring is 7.7/^eq/L (Figure 5-20). Thus biological activity in
the summer is very effective at removing nitrate in almost all the ELS-II lakes. Reduced biological activ-
ity as temperatures decrease in the fall is probably responsible for the observed nitrate increase between
summer and fall. It should be noted, however, that the 25th percentile for nitrate is below the SDL for
nftrate (0.5-0.9/ieq/L) in all seasons. Due to the low spring nitrate concentrations in most ELS-II lakes,
fall/spring nitrate changes were typically small (interquartile range = 0-6/ieq/L). The largest observed
spring increase in nitrate concentrations was 32/ieq/L
Kelly et al. (1990) have proposed that a summer surface nitrate concentration of 1 neq/L be used
as an indicator of lakes that have exceeded their algal nitrogen requirements. Lakes with summer nitrate
concentrations > 1 ^eq/L would thus be susceptible to lake acidification if nitrate loadings were to
increase. Based on ELS-II summer surface nitrate data (Figure 5-21), 25% of the ELS-II target lake
population had nitrate concentrations > 1 neq/L. Just over 8% of the ELS-II lake population had
summer nitrate concentrations > S/ieq/L These higher nitrate lakes lakes were located primarily in the
83
-------
1.0 i
0.8
c
o
"•e
2
Q.
I
'•Jo
"3 0.4
E
d
0.2-
o.o v__
-100
-50
50 100 150 200
ANC (ueq /!_)
250 300 350 400
o
00
CD
T—
o>
o.
CO
400 f
350
3QO
250
200
150
100
50
°
-50:
-1004;:
-100 -50
1:1
50 100 150 200 250 300 350 400
Fall 1986 ANC Gueq /L)
Figure 5-16. ANC population distribution for the spring, summer, and fall seasonal surveys and
comparison of spring 1986 and fall 1986 ANC in ELS-II lakes.
84
-------
0.04
4.25 4.50 4.75 5.00 5.25 5.50 5.75 6.00 6.25 6.50 6.75 7.00 7.25 7.50
PH
4.00
4.00 4.25 4.50 4.75 5.00 5.25 5.50 5.75 6.00 6.25 6.50 6.75 7.00 7.25 7.50
Fall 1986 pH
Figure 5-17. pH population distribution for the spring, summer, and fall seasonal surveys and
comparison of spring 1986 and fall 1986 pH in ELS-II lakes.
85
-------
SPRING 1986
SUMMER 1986
FALL 1986
-100 0 100 200 300 400 500
ANC
SPRING 1986
SUMMER 1986
FALL 1986
4.0 4.5 5.0 5.5 6.0 6.5 7.0 7,5 8.0 8.5 9.0 9.5
pH
Figure 5-18. Population minimum, 25th percentile, median, 75th percentile, and maximum values
in each of the three ELS-II seasonal surveys for (a) ANC, and (b) pH.
86
-------
1.0
0.0 i
100
150 200
Suifate (/^eq /L)
250
300
350
I
I
C/3
CD
CD
.
a
350f
300-
50 100 150 200 250
Fall 1986 Suifate (ufiq /L)
300
350
Figure 5-19. Suifate population distribution lor the spring, summer, and fall seasonal surveys and
comparison of spring 1986 and fall 1986 sulfate in ELS-II lakes.
87
-------
SPRING 1986
SUMMER 1986
FALL 1986
0 50 100 150 200 250 300 350
SULFATE (^eq/L)
SPRING 1986
SUMMER 1986
FALL 1986
0
10 20 30
NITRATE (^eq/L)
40
50
Figure 5-20. Population minimum, 25th percentile, median, 75th percentile, and maximum values
in each of the three ELS-II seasonal surveys for (a) sulfate, and (b) nitrate.
88
-------
1.0 f
0.8
2
'•e
a 0.6
o
I
"5 0,4
E
O
0,2
0.0-I
50
«=!
cr
-------
Adirondacks and are of high interest with respect to possible future increases in nitrate due to nitrate
breakthrough.
5.4.2.3 Sum of Base Cations
Sum of base cations was generally lowest in the spring and highest in the fall (Figure 5-22). At
low base cation concentrations (< 200/xeq/L), spring and fall concentrations were similar, whereas at
higher concentrations, spring values were almost all lower than fall values (Figure 5-22). As base cation
concentrations are closely related to ANC, these changes are clearly evident in the comparison of low
and high ANC lakes (Table 5-8). Median spring/fall change in base cations decreased by only 6^eq/L
in low ANC (< 50 /*eq/L) lakes compared to a 69 /xeq/L decrease in high ANC lakes.
5.4.2.4 DOC/A!uminum
Fall and summer distributions of DOC are very similar but spring distributions are noticeably lower
(Figures 5-23 and 5-24). Lower spring concentrations are probably attributable to dilution during spring
runoff and decreased decomposition during the colder winter months. The median DOC increase from
spring to fall was 0.5 mg/L (interquartile range = -0.1 to 1.5 mg/L) and was about the same in both high
and low ANC lakes (Table 5-8).
Before discussing inorganic monomeric aluminum (Aljm), it should be emphasized that the A!jm
SDL is about 20/ug/L and values below this may not actually contain any Al. . About 80% of the lakes
in all seasons had Aljm < 20/ug/L (Figure 5-25). Due to the large number of lakes with Aljm below the
SDL, medians and quartiles of spring/fall Aljm changes are negligible (Table 5-8). There was, however, a
consistent pattern of higher spring than fall Al!m concentrations in lakes with Aljm concentrations > 50
/zg/L (Figure 5-25). Due to the strong control of pH on aluminum solubility, all the lakes with elevated
Alim (> 50/ig/L) had pH < 6.0. Thus lakes with ANC > 50^eq/L (all had pH > 6.0) had very low Aljm
concentrations and showed no significant seasonal changes (Table 5-8). In lakes with ANC< 50/aeq/L,
the upper quartile showed significantly higher Aljm in the spring than in the fall (42-217/ug/L). The
majority of low ANC lakes, however, had little change in Aljm (Table 5-8). In low ANC lakes, spring/fall
changes in Aljm concentrations were most noticeable in the Adirondacks and Poconos/Catskills (Table
5-9). Fewer low ANC lakes in New England showed significant fall to spring Aljm increases (75th percen-
tile = 19/ig/L; maximum = 48/ig/L). In contrast, the 75th percentiles for spring Aljm increases in the
Adirondacks and Poconos/Catskills low ANC lakes were 74 and 71 jug/L, respectively (maximum
increases were 217 and 202/ig/L).
Alj concentrations of 50-200/ug/L have often been used as rough indicators of stressful levels of
aluminum on some aquatic organisms (L Baker et al., 1990), although the toxic effects of aluminum vary
greatly among species and are dependent on the concentrations of other ions (especially H+ and Ca2+;
J. Baker et al., 1990a). Percentages of ELS-II lakes with Aljm concentrations above the indicator values
were always highest or the same in the spring compared to fall and summer (Table 5-10). The
90
-------
1.0
0.8
c
o
i±
a 0.6
8
CL
w
c
.2
"3
O
cu
m
CO
CO
CD
DJ
_C
'C
CL
1000-
900-
800-
j
700^
j
•i
600-
500-
7.
X.'
400-
300-
200
100-
o-
X= ."* *
/• * * *
,X •'.__•
/'"*'' • •'
• X- .' ." . • '
x^":""
X
/*
X
0 100 200 300 400 50
1:1
600 700 800 900 1000
Fail 1986 Base Cations (/j.eq /L)
Figure 5-22. Sum of base cation population distribution for the spring, summer, and fall
seasonal surveys and comparison of spring 1986 and fall 1986 sum of base cations
in ELS-II lakes.
91
-------
1.0
O)
O
O
Q
03
CD
O)
_c
Q.
151
14
13
12
11
10
9
7
6:
1
3i
2!
1:1
o 1
2 3 4 5 6 7 8 9 10 11 12 13 14 15
Fall 1986 DOC (mg /L)
Figure 5-23. DOC population distribution for the spring, summer, and fall seasonal surveys and
comparison of spring 1986 and fail 1986 DOC in ELS-II lakes.
92
-------
SPRING 1986
SUMMER 1986
FALL 1986
SPRING 1986
SUMMER 1986
FALL 1986
SPRING 1986
SUMMER 1986
FALL 1986
0
0 200 400 600 800 1000 1200 1400
SUM CATIONS (^eq/L)
6 8 10
DOC (mg/L)
12
14 16
0 100 200 300 400
INORGANIC MONOMERIC AL
500
Figure 5-24. Population minimum, 25th percentile, median, 75th percentile, and maximum values
in each of the three ELS-II seasonal surveys for (a) sum of base cations, (b) DOC,
and (c) inorganic monomeric aluminum.
93
-------
1.0
0.8
g
"
0.6
8
o.
I
"5 0.4
E
O
0.2-
0.0
Spring
Summer
50 100 150 200
Inorganic Monomeric Al fag /L)
250
300
CD
-------
Table 5-10. Percentage of ELS-II Lakes with Inorganic Monomeric Aluminum (Aljm) Concentra-
tions above Reference Values
Reference Value
% of ELS-II Lakes
Region
Spring
Summer
Fall
Alim > 50/ig/L
Adirondacks
Poconos/Catskills
New England
AH ELS-II
32
13
5
11
29
7
4
9
24
7
4
AL> 100/ig/L
Adirondacks
Poconos/Catskills
New England
All ELS-II
24
7
2
7
16
2
1
4
15
2
2
4
Al,
Adirondacks
Poconos/Catskills
New England
All ELS-II
13
2
1
3
8
2
0
2
10
2
0
2
95
-------
Adirondacks have the highest percentage of lakes with elevated AIjm concentrations (32% > 50/xg/L
and 13% > 200/ug/L in the spring). Fewer lakes In New England had elevated Aljm (5% > 50^g/L) and
there was almost no change in the percentage of high Alim systems among the spring, summer, and fal!
seasonal surveys (Table 5-10), These similar percentages reflect the small Aljm changes between spring
and fall in New England (Table 5-9). Seasonal changes in the percentage of high Aljrn lakes were much
more pronounced In the Adirondacks and the Poconos/Catskills.
5.4.3 Conclusions: Among Season Variability
ANC, pH, DOC, and sum of base cations were lower in the spring than in the summer or fall.
Summer concentrations for these ions were more similar to concentrations in the fall than in spring.
Nitrate and inorganic monomeric aluminum concentrations were highest in the spring relative to summer
and fall, although they were found in very low concentrations in all seasons in the majority of ELS-II
lakes. ELS-II population estimates show that there were 24% more acidic lakes (ANC < 0) in the spring
than in the fall. This corresponds to an increase in the percentage of acidic ELS-II lakes from 9% to
11%. Similarly, the percentage of ELS-II lakes with ANC < 50 /xeq/L increased from 25% to 34% (fall to
spring) and the percentage with ANC< 100/ieq/L increased from 43% to 62%. The number of low pH
lakes was also higher in the spring than the fall. The percentage of ELS-II lakes with pH < 5.0 increased
from 4% to 5%, those with pH < 5.5 increased from 12% to 14%, and lakes with pH < 6.0 increased from
19% to 27%. The greatest spring depressions in ANC (relative to fall) were observed in ELS-II lakes with
the highest ANC (200-400 ^eq/L). In lakes with ANC > SO^eq/L, the median ANC depression was 60
A*eq/L and was associated with decreases in base cation concentrations, probably due to dilution by low
ANC, low base cation spring snowmelt runoff. In low ANC (< 50 jueq/L) lakes, spring ANC depressions
were small (median = 3/ieq/L; median pH decrease = 0.1 pH unit) and were associated with increases
in nitrate and inorganic monomeric aluminum. Sulfate concentrations were very similar in the spring,
summer, and fall seasonal surveys.
96
-------
SECTION 6
SYNTHESIS AND DISCUSSION
6.1 ASSESSMENT OF ELS-II VARIABILITY
6.1.1 Components of Variability
Data from ELS-Ii provides an excellent opportunity to assess the relative importance of different
factors on the observed variability of lakewater chemistry on a regional scale. Lakewater chemistry
varies along a number of different annual, seasonal, daily, and hourly time scales. Spatial variability is
also an important source of lakewater chemical variability. Spatial variability includes both differences in
lakewater chemistry within an individual basin and differences among lakes in a region. Variability Is also
introduced by the methods used to collect, transport, and store water samples, and by the analytical
techniques used to measure the variables of interest. The variability in ELS-II data was examined to
determine the Influence of these temporal, spatial, and analytical components.
6.1.2 Pooling Variance Estimates
Five components of variability were assessed for the ELS-II data: analytical, within-season
temporal, among-season temporal, between-year temporal, and among-lake spatial. Withln-lake spatial
variability is marginally accounted for in the within-season temporal component In the ELS-II data,
because the multiple within-season observations in the Fall Variability Study were collected from
independently selected locations on the study lakes. As the objective was to collect samples from
epilimnetic locations in the deepest part of the lake, this assessment does not address the questions of
littoral versus profundal variability and depth variability.
Temporal variability components were assessed by calculating a standard deviation and sum of
squares error (SSE) for each study lake with multiple observations for the temporal component of
interest. For example, the among-season component was analyzed by calculating the mean, standard
deviation, and SSE of the three seasonal measurements (spring, summer, and fall seasonal surveys) for
each of the 145 ELS-II study lakes:
SSE= Z^-X^J2 (11)
: -i
where Xj is the observation, Xmean is the mean of the multiple observations, and n equals the number of
observations in each study lake for the temporal component of interest. For the among-season and
within-fall-season components, there were three observations at each lake: three visits in the Fall
97
-------
Variability Study and three separate seasonal visits (n = 3). There were only two observations (n = 2)
for between-year variability (fall 1984/fall 1986), Note that within-fall variability could only be assessed
for the 41 lakes in the Fall Variability Study that had multiple within-season observations. A pooled
standard deviation (SD 0|) was calculated for each temporal component by dividing the total SSE in the
population by the total degrees of freedom:
SD = [ISSE/I(n-1)]a5 (12)
S S
where Z over S indicates the summation of all the lakes in the population of interest.
In the rest of this section, we refer to among-lake variability as spatial variability. Spatial variability
was simply calculated as the standard deviation of the lake chemistry at the ELS-I index site in fall 1986
in all lakes in the study population. The fall 1986 data were chosen, as opposed to spring or summer
data, because fall data are common to all the components of temporal variability. Analytical variability
was assessed in the same manner as temporal variability using sample duplicates. An SD , was calcu-
lated for each variable using equations 11 and 12 and the 84 duplicate sample pairs (n = 2) collected
throughout ELS-II.
6.1.3 Relative Importance of Variability Components
The SD | is a single value that can be used to compare the magnitude of variability of the
different components. It is also possible to plot CDFs of the individual lake standard deviations for the
different temporal components to study the distribution of lake temporal variability across the ELS-II
target population. We have constructed two CDFs for each study variable: one for all 145 lakes in the
ELS-II target population in which among-season and between-year variability is compared, and one for
the 41 Fall Variability Study lakes in which within-season, among-season, and between-year variability
can be compared. SD ( values are also presented for all data and the Fall Variability Study lake
subset. Only one estimate of the SD o| for analytical variability, based on all available ELS-II sample
duplicates, was calculated and it is presented for both groups. Sample duplicates were not collected
from all lakes. The analytical SD (is based on 84 duplicate pairs from 60 different lakes.
As measured in ELS-II, the different components of variability are not independent. Analytical
variability is nested within all the other variability components. Similarly, within-season variability is a
part of among-season and between-year variability because ELS-II samples were collected throughout
the seasonal index period. The SD ( values for the different components of variability in ELS-II are
presented in Table 6-1. To assess pooled variance, in order to make estimates of the percent of vari-
ance accounted for by the different components, the SD , values in Table 6-1 can be squared.
98
-------
Table 6-1. Pooled Standard Deviation of Multiple Observations in ELS-II for Duplicate Samples
(Analytical), between Lakes (Spatial), within Season (Fall 1986), among Season (Spring,
Winter, Fall 1986), and between Year (Fall 1984/Fall 1986)a
Variable
ALL DATA (n = 145)
ANC
PH
so/-
N03"
Base cations
DOC
Analytical13
8.3
0.10
2.9
0.36
12.4
0.10
Pooled Standard
Within Among
28.0
0.25
13.4
5.1
38.6
1.1
Deviation
Between
18.8
0.14
12.8
1.9
57.6
0.75
Spatial0
96
0.77
47
4.4
225
2.6
FALL VARIABILITY LAKES (n = 41)
ANC
pH
so42-
N03
Base cations
DOC
8.3
0.10
2.9
0.36
12.4
0.10
7.5 22.8
0.06 0.19
11.2 13.8
0.95 5.3
12.2 41.2
0.20 1.1
16.0
0.13
15.1
2.3
71.7
0.58
87.3
0.76
52.6
5.3
267.6
2.5
Units are in p, eq/L except for pH and DOC (mg/L),
Analytical pooled standard deviation is calculated from all available ELS-II data and is based on 84 sample duplicate pairs
collected from 60 different lakes. As such, it is not based on all 145 ELS-II lakes or all 41 fall variability lakes as are the other
pooled standard deviations.
Spatial pooled standard deviation is the standard deviation of all the Fall 1986 index samples.
99
-------
6.1.3.1 ANCand pH
For ANC and pH, within-season SD ( values and analytical SD , values were very similar, indi-
cating that the major component of within-season variability is due to analytical variability (Table 6-1).
ANC and pH spatial variability among lakes is much greater than any of the temporal or analytical
components (SD ( 3-10 times higher). Between-year SD ( values for ANC and pH were about twice
those of within- season values, whereas among-season SD ( values were about three times within-
season values (Table 6-1). For both ANC and pH, among-season variability was consistently higher than
between-year variability, which was almost always higher than within-season variability (Figures 6-1 a,b).
6.1.3.2 Sulfate and Nitrate
Sulfate spatial variability among lakes was much greater than temporal or analytical variability
(Table 6-1), indicating that regional sulfate deposition differences or watershed sulfate source differences
exert a much greater control on lakewater sulfate concentrations than temporal changes within a lake.
Within-season, among-season, and between-year sulfate changes were fairly similar in both the Fall
Variability Study lakes and all ELS-II lakes (Figure 6-2a), although among-season variability was usually
the largest. Over 80% of the ELS-II lake population had temporal standard deviations in sulfate
concentration < 20/zeq/L As within-season changes are nested within among-season and between-
year changes, these data indicate that short-term (day-to-day or week-to-week) changes are an
important component of longer term sulfate temporal variability. The within-season SD (for sulfate
was about four times the analytical SD ,, so analytical variability accounts for only a minor portion of
the sulfate temporal variability.
Among-season SD ( values for nitrate were the same or greater than fall spatial variability (Table
6-1). Thus, the seasonal changes in nitrate concentration within a lake are about the same as the
differences in fall nitrate concentration among the ELS-II lakes in the northeastern United States. Spring
spatial variability (SD | = 10/aeq/L), however, was twice as high as fall spatial variability. Among-
season nitrate SD 0| values are also more than twice as high as between-year SD , values, 5 times
higher than within-season SD ( values, and 10 times higher than analytical SD ( values. Therefore,
population estimates of the regional distribution of nitrate are very dependent on the season in which
they are measured. Nitrate temporal standard deviations, however, were very low (75% < 5/ieq/L) due
to the low nitrate concentrations in most lakes (Figure 6-2b).
6.1.3.3 Sum of Base Cations
The spatial SD , for sum of base cations is four times that of the largest temporal component
(Table 6-1), indicating that differences in watershed base cation mobilization among lakes accounts for
far more regional variability than temporal changes within lakes. In terms of temporal variability,
between-year SD ( values are the same or slightly higher than among-season SD , values (Figure
6-3a). Thus, year-to-year changes in hydrologic conditions appear to be equally as important in
100
-------
Standard Deviation - Fall Variability Lakes
1.0
c
g
1 0.6
o
(t
I
"5 0.4
E
3
O
0.2
0.0 4
10
1.0 f
0.0
- Between Season
- Between Year
- Within Fail
20
30
ANC Cueq /L)
40
50
60
0.00 0.05 0.10
0.15
0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60
pH
Figure 6-1. Population distribution of the temporal standard deviation in the 41 Fall Variability
Study lakes for (a) ANC, and (b) pH.
101
-------
Standard Deviation - Fall Variability Lakes
1.0
0.8
c
o
••e
a 0.6
s
a.
••a
"5 0.4
I
0.2
0.04
Between Season
Between Year
Within Fall
10 20 30 40 50
Sulfate (ueq /L)
60
70
80
1.01
0.8-
g
"•e
a 0.6
8
o_
|
3
O
0.4-
0.2-
o.o-I
Between Season
Between Year
Within Fan
10 15
Nitrate (/^eq /L)
20
25
Figure 6-2. Population distribution of the temporal standard deviation in the 41 Fall Variability
Study lakes for (a) sulfate, and (b) nitrate.
102
-------
Standard Deviation - Fall Variability Lakes
1.0
0.8
o
•-E
8.0.6
"5 0.4
I
0.2
0.0-I
Between Season
Between Year
Within Fall
25
50
75
100 125
Base Cations
150
175
200
225
250
/L)
1.0
0.8
O
t
8.0.6
O
'•s
"5 0.4
|
O
0.2
o.o-i
Between Season
Between Year
Within Fall
0.0 0.5 1.0 1.5 2.0 2.5
DOC (mg /L)
3.0
3.5
4.0
Figure 6-3. Population distribution of the temporal standard deviation in the 41 Fall Variability
Study lakes for (a) sum of base cations, and (b) DOC.
103
-------
controlling base cation concentrations as seasonal changes, at least between the 1984 and 1986 years.
Whether between-year regional lakewater base cation changes would be as high for two years that were
more hydrologically similar than 1984 and 1986 is unknown, Within-season SD { values were about
equal to analytical SD ( values and were much lower than between-year and among-season SD (.
For base cations, analytical variability is thus the major component of within-season variability. Base
cations were very stable during the fall index period compared to annual and seasonal variability.
6.1.3.4 DOC
Within-season DOC SD , values were twice as high as analytical SD , values (Table 6-1).
Among-season variability was consistently higher than between-year variability, which was higher or the
same as within-season variability (Figure 6-3b). Overall, among-lake spatial variability was higher than
any of the temporal variability components.
6.1.4 Conclusions and Comparison to Long Term Monitoring Data
With the exception of nitrate, among-lake spatial variability is much larger than temporal or
analytical variability. Thus the regional factors explaining differences among lakes are more important
than within-lake temporal variability in explaining the distribution of ANC, pH, sulfate, DOC, and base
cations in the ELS-II population. In terms of temporal variability, for ANC, pH, DOC and nitrate:
among-season variability > between-year variability > > within-season variability.
For base cations:
between-year variability > among-season variability > > within-season variability.
For sulfate, the three components of temporal variability were about equal. Analytical variability was the
major component of within-season variability for ANC, pH and base cations.
L Baker et al. (1990) analyzed data from EPA's Long Term Monitoring (LTM) Project (Newell et al.,
1987) to assess among-year and among-season variability in lakewater chemistry. Based on 15 lakes in
the Adirondacks, and 5 lakes in Maine, they found that year-to-year changes in fall index values between
1982 and 1988 fell within a consistent range. No one year stood out as being atypical. Comparisons of
spring and fall chemistry in the LTM data showed strong correlations. Spring ANC and pH were typically
lower than or the same as fall values and spring values could often be predicted with moderate success
from fall values. These predictive relationships, however, were not consistent among different years in
the 1982-1988 study period (Newell, 1987). Overall, in all the LTM data from lakes in Maine, Vermont,
the Adirondacks, and the Upper Midwest, among-lake variability exceeded annual and seasonal temporal
104
-------
variability (Newell et al., 1987). As in ELS-il, spatial variability accounts for more variability in regional
lakewater chemistry than temporal variability. Thus, Newell et al. (1987) have suggested that sampling
more lakes in a region would allow for a more accurate characterization of regional trends than would
sampling a similar number of lakes more intensively: an important consideration for future regional
monitoring efforts that is supported by the analyses reported here.
6.2 ROBUSTNESS OF THE ELS-I FALL INDEX SAMPLE
6.2.1 Fall Index Variability
One of the major goals of ELS-II was to assess the sampling error associated with the ELS-I fall
index sample. The success of ELS-I was based on the premise that a single fall epilimnion sample
adequately characterizes the acid-base status of a study lake and that these data could then be used to
estimate regional chronic lake acidity. All the within-season data collected in ELS-II support the
adequacy of the fall index concept. Overall, variability among the three lake visits at independently
selected sites in the Fall Variability Study was very small and was about equal to analytical variability for
ANC, pH, and base cations. Population estimates of the proportion of lakes with ANC or pH below
reference values were not greatly affected by fall temporal or within-lake site selection variability.
Conclusions about the acid-base status of northeastern lakes would have been the same if sampling had
occurred at any of the three sample visits or sampling locations. Thus, in terms of variables of interest
for acidic deposition effects, mid-lake epilimnetic conditions during the fall index period are well
represented by a single sample. Collection of additional samples during the fall index period would have
yielded little additional information on the acid-base status of lakes.
6.2.2 Predicting Spring Conditions from Fall Index Data
A second aspect of the robustness of the fall index sample Is how well it relates to conditions in
the other seasons of the year. In particular, we are interested in how well the fall index period relates to
spring conditions when ANC and pH are lowest. The large-scale regional survey effort in ELS-I was not
conducted in the spring because of the difficulty in sampling a large number of lakes in four regions of
the country during a narrow spring window that was difficult to predict. Fall was selected as the index
period in ELS-I because it was a well-mixed period of reasonable duration (about 6 weeks) during which
samples from more than 1,600 lakes could be collected. In addition, chemical conditions are less
varying in fall than in spring, which fit the objective of ELS-I, which was to assess chronic acidity rather
than episodic acidity. By utilizing the spring/fall relationship in ELS-II lakes, however, we can estimate
what spring conditions would have been like in ELS-I.
105
-------
Spring chemistry in ELS-II lakes was strongly correlated with fall chemistry (Table 6-2; also see
Figures 5-16, 5-17, 5-19, 5-21, 5-22, 5-23, and 5-25). Due to the fact that spring/fall chemical differences
varied in the different ANC classes or clusters, spring/fall regressions were performed separately within
the three ANC clusters used in the ELS-II site selection (see Section 2.3.2). In the lowest ANC sample
cluster (< 25 jueq/L), fall chemistry accounted for more than 60% of the variance in spring chemistry (r2
> 0.6) for all of the studied variables (Table 6-2). The root mean square error (RMSE) or average
residual for predicting spring ANC and pH from fall values was 11 ^eq/L and 0.2 pH units, respectively.
Similar r2 values were seen in the two higher ANC clusters, with the exception of Aljm (Table 6-2). Aljm
concentrations were almost always below the system decision limit (not different from blank values) in
the higher ANC clusters (due to high pH) and thus there was poor fall/spring predictive ability. For
other variables (SO42", NO3", base cations, DOC), there was a strong relationship between the measured
fall and spring chemistry. A strong relationship (r = 0.82) between spring 1986 minimum lake outlet
ANC and ELS-I fall index ANC has also been noted by Eshleman (1988), using the data of Driscoli
(1986), in nine Adirondack lakes.
We can make a rough estimate of the acid-base status of the northeastern United States ELS-I
target population in the spring by using the regression equations in Table 6-2 for the appropriate ANC
clusters in the ELS-I data. For this exercise, we assume that lakes with ANC > 400 (above the ELS-II
cutoff) have ANC > 50 and pH > 6 in the spring. Recognize that this estimate makes the assumption
that the spring/fall relationship in ELS-II (1986) is the same as the spring/fall relationship in ELS-I (1984).
As noted earlier, Newell (1987) found that the spring/fall relationship in the LTM data was not always the
same in different years or in different regions. The conclusions from LTM data, however, are based on
individual lakes. Changes in regional lake populations are likely to be different and possibly more stable
among years.
In fall 1984, 6.2% (441 lakes) of the ELS-I lakes in the Northeast were acidic, and 21.7% had ANC
< 50/ieq/L (see Table 5-3). Using the fall/spring regressions, we would estimate that 8.1% (577 lakes)
of the ELS-I lakes would have been acidic in the spring and 28.0% would have had ANC< 50 jueq/L
Similarly, in terms of pH, 3.4% of the ELS-I lakes in the Northeast had pH < 5.0 and 8.6% had pH < 5.5
in fall 1984 (see Table 5-3). Regression estimates for spring conditions in the northeastern ELS-l lakes
show only small increases in the percentage of low pH lakes (3.6% with pH < 5.0, and 8.8% with pH
< 5.5 in spring). Estimates of the percentage of ELS-I lakes with pH < 6.0 in the spring (17.1%),
however, are substantially larger than the measured fall percentage (12.8%). Although conditions are
more acidic in the spring, spring conditions are strongly related to fall index conditions. Overall, ELS-II
results support the conclusion that the ELS fall index sample is a robust estimator of the acid-base
status of lakes, at least in the northeastern United States.
106
-------
Table 6-2. Regression Statistics for the Relationship between Spring Chemistry (Dependent
Variable) and Fall Chemistry (Independent Variable) in 1986 ELS-II Data3
Variable
Cluster
ANC
pH
so42-
N03"
CB
DOC
A"im
Cluster
ANC
pH
so42-
NO3~
CB
DOC
*lim
Cluster
ANC
pH
S042~
NO3"
CB
DOC
Alim
r2
1 (ANC<25/ieq/L)
0.65
0.77
0.72
0.74
0.95
0.63
0.82
II (25 < ANC< 100Aieq/L)
0.43
0.53
0.83
0.42
0.95
0.70
0.26
III (100 < ANC< 400/ieq/L)
0.88
0.43
0.81
0.60
0.94
0.36
0.12
slope
0.974
0.834
0.744
2.14
0.810
0.522
1.20
0.512
0.795
0.898
0.814
0.883
0.500
0.993
0.766
0.739
0.893
1.29
0.863
0.548
0.647
y-intercept
-3.74
0.815
28.3
3.19
32.2
1.07
15.6
12.7
1.02
15.6
2.13
-3.72
1.22
-3.70
-21.9
1.63
19.2
2.46
-14.2
1.33
-3.08
RMSEb
11.1
0.21
16.6
6.7
36.1
1.0
56.3
18.1
0.25
20.8
3.6
33.7
1.01
12.0
29.0
0.21
22.2
4.2
50.0
1.2
7.6
Units are in peq/L except for pH, DOC (mg/L), and Ajm (ng/L).
RMSE = root mean square error.
107
-------
6.3 SPRING CONDITIONS
Although the fall index period was a very stable period in which to assess the acid-base status of
lakes, it is not necessarily the season of lowest ANC and pH. In regions that accumulate snowpacks,
lakewater ANC and pH typically decline during spring snowmelt (Galloway et al., 1980; Jeffries et al.,
1979). Based on the ELS-II seasonal surveys (Section 5.4), ANC and pH were lower in the spring than
in the summer or fall in lakes in the northeastern United States. Thus, spring is of high interest from a
perspective of acidic deposition effects. Of specific interest is the question of whether conditions in the
spring were more toxic to biota than fall conditions. Also, the factors related to the fall/spring changes
in ANC and pH are important in understanding the processes responsible for the observed spring ANC
depressions. Lastly, it is also important to place the ELS-II spring sample in context with spring
snowmelt hydrology. As is shown in Section 6.3.3, the ELS-II samples were generally collected 2 to 3
wkks after peak snowmelt discharges. Thus, the ELLS-II spring data do not indicate worst case episodic
spring conditions, but rather post-snowmelt spring seasonal (baseflow) conditions. The seasonal
comparisons made in this report need to be interpreted with these facts in mind. In this section,
therefore, spring refers to post-snowmelt spring baseflow conditions, not worst case spring episodic
conditions.
6.3.1 Seasonal Biotic Toxicity
The effects of acidity on aquatic organisms are determined by a number of different water quality
variables, the most important of which are pH, calcium, and inorganic monomeric aluminum (Al. ) (J,
Baker et al., 1990a). Declining pH caused by acidic deposition can mobilize Alim to surface waters
resulting in both low pH and high Aljm concentrations that are toxic to biota. The toxic effects of pH and
Aljm, however, can be ameliorated to some extent by high Ca2+ concentrations. Toxicity is very species
dependent and chemical conditions that are extremely harmful to one species may have little or no
effect on another species. J. Baker et al. (1990a) have used an acid-stress index (ASI) to estimate acid
stress to fish. The ASIs, which take into account the combined effects of pH, Aljm, and calcium, were
developed from laboratory bioassay data of fish mortality. The ASI is based on a logistic regression
equation and represents expected percent mortality. ASIs were developed for three fish species that
vary in their sensitivity to acidic conditions: brook trout (acid-tolerant), smallmouth bass (intermediate),
and rainbow trout (acid-sensitive). Full details of model development and equations are given in J.
Baker et al. (1990a),
Sensitive, intermediate, and tolerant ASIs were calculated for the ELS-II lakes in each season from
*y 4.
observed pH, Ca , and Aljm concentrations. The results showed that there was very little difference in
the number of lakes unsuitable for fish between spring, summer, and fall in the ELS-II lake population
(Table 6-3). Thus, it appears as if the observed spring/fall changes in pH, calcium and aluminum were
108
-------
Table 6-3. Percentage of Lakes with Chemical Conditions Exceeding Acid Stress Index Values in
the ELS-II Target Lake Population
Acid Stress Index (AS1) Value8
Acid-Sensitive Species
ASI > 10
ASi > 30
ASI > 50
ASI > 80
Intermediate Species
ASI > 10
ASI > 30
ASI > 50
ASI > 80
Acid-Tolerant Species
ASI > 10
ASI > 30
ASI > 50
ASI > 80
Percentage of ELS-II Lakes
Spring
72
36
23
17
18
14
13
11
3.7
1,2
0.3
0
Summer
80
37
22
15
18
13
11
8.4
4.5
0.9
0
0
Fall
73
32
21
15
16
13
8,8
7.2
3.4
0.9
0,3
0
The acid stress index (ASI) is a model estimate of percent fish mortality based on laboratory bioassays of the combined
effects of calcium, pH, and inorganic monomeric aluminum (J. Baker et at., 1990a). The ASI in ELS-II lakes was calculated
from the observed seasonal concentrations of pH, Ca?1", and A|m.
The fall index ASI values were related to observed fish presence in lakes in order to make the following generalizations: lakes
with a tolerant ASI > 30 were unsuitable for ail fish species, lakes with a tolerant ASI > 10 were unsuitable for brook trout,
lakes with an intermediate ASI > 80 were unsuitable for other sport fish, such as srnallmouth bass and lake trout, and lakes
with a sensitive ASI > 80 were unsuitable for acid-sensitive species such as minnows.
109
-------
not great enough to cause a significant regional change in biological effects as measured by the ASI.
The main reason for the lack of a seasonal change was that the largest ANC changes occurred in higher
ANC lakes that did not have a corresponding pH decrease or Alim increase into the range of stressful
values. The low ANC lakes that are most susceptible to acidic deposition effects did not have large
changes in pH, Ca2+, and Alim; thus, there were only small changes in estimated biological effects. The
ASI values listed in Table 6-3 are due to the observed chemical conditions in the lake and are not
necessarily the result of acidic deposition. Other causes of the acidic conditions in ELS-II lakes (e.g.,
organic acidity) would be expected to cause stressful conditions for some fish species. Regardless, fall
index chemistry appears to be a stable estimator of biological effects. The relationship between ELS-ll
spring conditions and worst-case conditions is discussed in Section 6.3.3.
6,3.2 Factors Related to Fall/Spring ANC Changes
Of particular interest in terms of acidic deposition effects are the factors related to the observed
spring depressions in ANC. An attempt was made to correlate physical variables with the observed
fall/spring change in ANC. However, in the entire ELS-II data set, there was no significant relationship
(r < 0.05; p > 0.01) between ANC change and elevation, lake area, watershed area, residence time,
lake depth, or lake volume. When examined within specific regions or within high and low ANC lake
groups, there were some significant correlations with elevation and residence time. In higher ANC (> 50
jieq/L) ELS-II lakes, residence time was positively related (r2 = 0.13; p = .002) to the general increase in
ANC from spring to fall in 1986. In other words, lakes with shorter residence times experienced larger
spring ANC depressions. This makes intuitive sense because lakes with faster flow-through times would
be the ones most susceptible to the dilution effects of low ANC runoff or inputs of acid anions. Among
low ANC lakes (fall ANC< 50/ieq/L), elevation was negatively related (r2 = 0.16; p = 0.001) to the
spring minus fall difference in ANC in ELS-II lakes. This correlation was significant only In the
Adirondacks (r2 = 0.55; p = .004) and Poconos/Catskills (r2 = 0.74; p = 0.05) regions and not in New
England (r2 = 0.19; p = 0.3). Thus, low ANC lakes in New York and Pennsylvania tended to have larger
spring ANC depressions at higher elevations. Measured physical variables, however, only explain a
minor percentage (low r2) of the variance in fall/spring changes in ANC. Other factors are likely to be
more important.
Fall/spring changes in ANC were similar in different hydrologie lake types (e.g., drainage, seepage
lakes). For example, in lakes with fall ANC > 50 j«q/L, the median spring minus fall ANC change in
drainage lakes (including reservoirs) was-60/ieq/L (interquartile range = -41 to-98/^eq/L). The
corresponding median change in seepage lakes (including closed lakes) was-69/zeq/L (interquartile
range = -9to-88/ieq/L). A similar pattern was also seen in lakes with fall ANC < 50 /^eq/L
Correlations between fall/spring ANC change and changes in other chemical variables were also
examined. As can be seen in Figure 6-4, there was a strong relationship between fall/spring ANC
110
-------
ELS-II
o
O)
c
OJ
Q.
CO
25-
;
-25
-50-
-75^
100-
125-
150-
175
?nn
o ^
c
Q*M
±5
«^
X
S^-® o
•LK -/oL-V; /™CJ
|^W^@f"AO
•^ _fT~S
^ Cy o A
0Q-) ^"^ O O O
A
c£ CB ^ 9)
^ o o
* o o Qo
Q
0)0 ® **°
o
* £ £ °
o
*
A Adirondacks
.A, O
w Poconos/Catekills
° New England
-100
0
100 200 300
Fall 1986 ANC (^eq/L)
400
500
Figure 6-4. Relationship between ANC change (spring 1986 minus fall 1986) and fall 1986 index
ANC in ELS-II lakes.
111
-------
change and fail index ANC. Maximum fall/spring ANC changes were observed in the ELS-il lakes with
the highest fall ANC valuess (200-450//eq/L). The largest spring ANC depression (-177/xeq/L) occurred
in a lake with a fall ANC of 326/i eq/L. This is in contrast to the results reported by Schaefer et al.
(1990) who found the biggest ANC depressions in Adirondack lakes with intermediate baseline ANC
values (50-100 p eq/L), However, they had only one lake with baseline ANC over 150 ^teq/L with which
to judge ANC changes in higher ANC lakes. In the rest of this section, lakes with ANC above and below
50/^ eq/L are analyzed separately because they exhibited different geochemical changes associated with
spring ANC depressions.
In higher ANC (> 50/^eq/L) ELS-II lakes, the only significant chemical change associated with
fail/spring ANC changes were changes in base cation concentrations (Figure 6-5; r2 = 0.36). Spring
depressions in ANC fall around the 1:1 line coincident with spring depressions in base cations.
Spring/fall changes in sulfate, nitrate, DOC, and aluminum were not related to ANC change (r2 < 0.05; p
> 0.05). The spring/fall changes are also evident in Table 5-8, which shows that in lakes with fall ANC
> 50 p eq/L, median and quartile estimates of spring decreases in ANC are quite similar lo decreases in
base cations (median ANC change = -eo^eq/L; median base cation change = -69/jeq/L). This phen-
omenon is most likely the result of the dilution of lake water and/or runoff by low ANC, low base cation
snowmelt in the spring. In higher ANC systems, episodic ANC losses during storm events have often
been attributed primarily to base cation dilution (Wigington et al., 1990). Similarly, Molot et al., (1989)
reported that the major contributor to spring snowrnelt alkalinity depressions in Ontario streams was
base cation dilution. A base cation dilution mechanism also helps explain the small fall/spring ANC
changes observed in low ANC lakes. Fall "baseline" conditions in low ANC lakes more closely approxi-
mate spring snowmelt runoff and thus would not experience nearly as much dilution as higher ANC
lakes. Thus, low ANC lakes would require inputs of H+ acid anions to cause an ANC depression.
In low ANC (< 50jieq/L) ELS-II lakes, spring increases in both nitrate (r2 = 0.19) and Alim (r2 =
0.22) were positively correlated with spring ANC depressions. Spring depressions in ANC in low ANC
lakes were not significantly correlated with either sulfate or base cation changes (r < 0.1; p > 0.05).
Relationships between seasonal ANC change and changes in other ions were different among the differ-
ent regions. The spring decrease in ANC was significantly related to nitrate Increases in the Adirondacks
(r2 = 0.22) but was not related to nitrate increases in New England (Figure 6-6). Most of these low ANC
Adirondack lakes had spring nitrate concentration increases greater than spring ANC depressions
(Figure 6-6; Table 6-4). The same pattern of nitrate increase associated with ANC loss during snowmelt
has been reported by Schaefer et al. (1990) for the RILWAS lakes in the Adirondacks. In the Poconos/
Catskills there were only seven low ANC ELS-II sample lakes, so correlation relationships are not very
useful. In one of the low ANC Pocono Mountain Lakes, however, a 31 /xeq/L decrease in ANC was
nearly balanced by a 24 n eq/L increase in nitrate (Table 6-4). Overall, increasing nitrate concentrations
are associated with spring ANC depressions in low ANC Adirondacks and Poconos lakes but not New
England lakes.
112
-------
sr
3>
o
c
O)
o.
CO
25
0
-25
-50
-75
-100
-125
-150
-175
-200
Fall ANC > 50 ueq/L
O
* Adirondacks
^ Poeonos/Catskills
° New England
o
-3QQ
O
O
250
-200 -150 -100 -50
Spring minus Fall Base Cations (^eq/L)
0
50
Figure 6-5. Relationship between ANC change (spring 1986 minus fall 1986) and sum of base
cation concentration change (spring 1986 minus fall 1986) in ELS-II lakes with fall
1986 index ANC > 50 jieq/L. Solid line is the -1:1 line.
113
-------
Fall ANC s 50 ueq/L
S"
3>
o
3
C
1
O>
Q.
C/3
30
20
10
0
-10
-20
-30
-40V
Q
0°
o
o
-5
o
°
o o
o
* Adirondacks
^ Poeonos/Catskills
° New England
o
o
5 10 15 20 25
Spring minus Fall Nitrate (/^eq/L)
30
35
Figure 6-6. Relationship between ANC change (spring 1986 minus fall 1986) and nitrate con-
centration change (spring 1986 minus fall 1986) in ELS-II lakes with fall 1986 index
ANC< 50peq/L Solid line is the -1:1 lines.
114
-------
Table 6-4. Spring Minus Fall Changes (A) in Major Anions and Cations in Low ANC ELS-II Lakes
with Spring ANC Depressions (&ANC) > 10peq/La
Lake ID AANC
Adirondacks
1A2-004 -27.6
1A1-061 -23.1
1A1-003 -15.7
1A1-057 -12.8
1A1-017 -11.6
1A1-012 -10.8
Poconos/Catskills
1B1-010 -30.6
1B2-028 -11.3
New England
1C2-057 -23.4
1D3-002 -21.2
1E1-106 -18.6
1C1-086 -16.4
1E1-054 -13.4
1D3-029 -13.1
1C2-037 -12.3
ANO3~
19.7
13.6
33,0
13.4
26.5
15.2
23.9
1.3
0.9
10.0
-1.1
1.9
1.1
1.5
0.8
AAIn +
6.8
-10.3
16.7
19.4
8.6
5.8
18.7
1.2
2.1
2.1
3.1
-0.1
-0.2
3.4
-0.4
ACb
-4.4
3.0
9.2
9.4
-5.2
4.9
-4.0
-20.9
-17.7
-335
-12.1
-37.4
-12.8
-23.2
-32.3
AS042'
-18.3
-24.8
-6.1
2.7
-4.9
-8.1
-13.1
-27.5
36.7
6.8
-3.9
3.1
-7.6
12.5
13.9
AOrg A"
2.8
-16.2
0.9
-15.4
-7.3
0.1
0.05
-2.3
-6.7
1.4
6.1
-5.3
5.5
4.4
4.1
AH +
-3.3
-2.6
2.2
5.3
2.3
2.7
4.9
1.3
2.8
2.8
3.9
0.7
0.2
0.5
1.4
All units are in jieq/L. Low ANC lakes were defined as those lakes with fall 1986 ANC < SOj/eq/L. AAf+ is the spring-fall
change in inorganic monomeric aluminum concentration in^eq/L, Inorganic aluminum was converted from^M tojieq/L
based on an aluminum charge calculated from an empirical equation derived from equilibrium model speciation and National
Stream Survey data (A(m charge = 7.06 - pH; also see Figure 1 in Sullivan et al., 1989). A Org A* is the spring-fall change in
organic anion concentration in j*eq/L. Organic anion concentration was estimated from DOC and pH using the equation of
Oliver et al., (1983).
115
-------
Spring increases in Aljm are also associated with spring ANC depressions (Figure 6-7) in the
Poconos/Catskills, New England (r2 = 0,33), and the Adirondacks (r2 = 0.32 after removing one outlier),
Spring aluminum increases in New England, however, are quite small (< 3/^eq/L) relative to the ANC
decreases (< 25^eq/L; Figure 6-7), In the Adirondacks and the Poconos, some of the Aljm increases
are about the same as the ANC decreases (5-20/ieq/L). Spring H+ increases were quite small (< 5
A*eq/L) in all low ANC ELS-ll lakes. Thus, large spring ANC depressions (> 10/zeq/L) related to acid
cation increases (AIn+, H+) are associated mainly with aiuminum increases and not H+ increases (Table
6-4). Aluminum is an effective pH buffer in acidic waters (Driscoll and Bisogni, 1984). The increase in
Alim in ELS-II lakes, however, does not appear to be associated with decreases in pH (Figure 6-8). For
example, the three largest individual spring Aljm increases in ELS-ll lakes all had spring pH depressions
smaller than 0.2 pH units. Also, three lakes in which spring pH was higher than fall pH had spring Al|m
increases over 50 fj. g/L. It should be noted, however, that small changes in pH at low pH levels can
mobilize significant amounts of aluminum.
In low ANC lakes in New England, spring/fail ANC changes were also correlated with base cation
(r2 = 0,13) and DOC (r2 = 0.26) changes. The significant relationship with DOC was due to four lakes
that had spring ANC increases (10-20 jueq/L) associated with large DOC (4-7 mg/L) decreases. It is
possible that strong organic acids were diluted in these lakes during snowmelt, resulting in increased
ANC. All four of these lakes are high DOC systems (fall index DOC = 7-15 mg/L) dominated by organic
anions. Although base cation concentration changes were only weakly related to ANC changes in low
ANC New England lakes, the lakes in New England with spring ANC depressions greater than 10/zeq/L
all had base cation depressions about the same or greater than the ANC depression (Table 6-4). Thus
the main factor responsible for the larger ANC depressions in low ANC New England ELS-ll lakes
appears to be base cation dilution. It should be reiterated, however, that the observed spring
depressions in ANC were fairly small in all ELS-ll low ANC lakes. Only 25% of the low ANC ELS-ll lakes
decreased by more than 10 jueq/L and the maximum depression was 31 ^teq/L
6.3.3 Worst Case Spring Conditions
Conditions during the fall index period (after turnover) have been shown to be extremely stable
(Section 5.3), Spring conditions, however, are much more variable. A number of studies have shown a
significant amount of episodic acidification (ANC/pH decreases, Aiim/NO3" increases) associated with
snowmelt episodes in the Adirondacks (Galloway et al,, 1987; Sehaefer et al., 1990; Schafran and
Driscoll, 1987). A major factor behind this phenomenon is that during snowmelt, runoff occurs more
rapidly through shallow flowpaths where acid neutralization is incomplete (Driscoll et al., 1987; Peters
and Driscoll, 1987). The peak runoff typically occurs while ice cover is still present on the lakes,
whereas the ELS-ll spring index sample was scheduled (for logistics and safety reasons) to be taken
within two weeks of ice-out. During baseflow conditions, acidic inputs flow through deeper flowpaths
where they have longer contact times and contact with more material rich in base cations, Sehaefer et
116
-------
Fall ANC s 50 ueq/L
IT
£
o
•2'
CO
c
"E
O)
Q.
CO
301
20
10
-10
-20
A Adirondacks
Poconos/Catskiils
° New England
-10 -50 5 10 15
Spring minus Fall Inorganic Aluminum (/^.eq/L)
20
Figure 6-7. Relationship between ANC change (spring 1986 minus fall 1986) and inorganic
monomeric aluminum concentration change (spring 1986 minus fall 1986) in ELS-I!
lakes with fall 1986 index ANC < 50/ieq/L Inorganic aluminum was converted from
p,M to/jeq/L based on an aluminum charge calculated from an empirical equation
derived from equilibrium model speciation and National Stream Survey data (AIjm
charge = 7.06 - pH; also see Figure 1 in Sullivan et al., 1989). Solid line is the -1:1
line.
117
-------
Fall ANC g 50 ueq/L
0.8
0.6
0.4
i :
Q.
= 0.2
if
CO
— n r\ -
c u.u
1 :
DJ
•i -0.2
D.
CO
-0.4:
-0.6
-0.8-
•k
O
, A ,-
-------
al. (1990) reported that lake outlet ANC depressions tended to persist for 4-8 weeks after peak outlet
discharge during spring snowmelt in 1986 and 1987 in the Adirondacks. Acidic conditions, however,
were most severe about the time of peak snowmelt discharge. ANC then increases as the system
returns to baseflow.
Example hydrographs for conditions in the spring of 1986, along with the associated ELS-II sample
windows, are shown for the Adirondacks (Figure 6-9), Pennsylvania (Figure 6-10), and Maine (Figure
6-11). As can be seen in these figures, the ELS-II spring sample was conducted 2-3 weeks after peak
discharge in each of these three areas. For example, peak discharge occurred on April 1, 1986, in
Woods Lake Outlet in the Adirondacks. ELS-II spring samples in the Adirondacks were collected
between April 16 and 25, 1986, just before the time when the lake outlet hydrograph returned to
baseflow levels at the end of April (Figure 6-9).
ELS-II spring data definitely indicate a spring ANC depression relative to a fall baseline, yet it is
doubtful, and certainly not expected, that the spring ELS-H data correspond to the minimum lakewater
ANC. The data of Schaefer et al. (1990) on Adirondack lake outlet chemistry during snowmelt indicate
that the minimum ANC occurred at or right after peak discharge. In addition, minimum spring lake outlet
ANC predicted from fall 1984 ELS-l ANC index values using the regression equation of Eshleman (1988)
was typically (median) 32 /zeq/L lower than the observed spring lakewater ANC in ELS-II Adirondack
lakes. The ANC in Long-Term Monitoring Program lake outlets in the Adirondacks were on average 22
/^eq/L lower during spring peak flow than in late April when the ELS-II was conducted (Driscoll, pers.
comm.). Also, maximum nitrate increases during spring snowmelt episodes in the Adirondacks range
from 5 to 80 /«q/L (Wigington, 1990) as opposed to the 2-33 fjeq/L increases observed in the ELS-II
Adirondack lakes (Figure 6-6). Based on these observations, it seems safe to conclude that the
minimum ANC in ELS-II lakes would be lower than the observed ELS-II spring index values that were
collected during a period of decreasing lake outflow 2-3 weeks after peak snowmelt runoff.
In addition to temporal changes in chemistry during spring snowmelt, spatial variability is also
important. When colder and lighter (0-4°C) acidic snowmelt enters a lake, it usually flows across the
lake surface under the ice directly to the lake outlet without mixing with the denser 4°C water already in
the lake. Thus, the bottom portion of the lake may be relatively unchanged by the acidic snowmelt
inputs, whereas the surface water undergoes a severe ANC depression. This phenomenon has been
observed in Cone Pond, New Hampshire (Baird et al,, 1985), and in both pelagic and near-shore regions
of Woods Lake in the Adirondacks (Gubala et al., in review). Gubala et al. also found that small
amounts of groundwater in-seepage rich in base cations helped maintain baseline ANC conditions in the
bottom water of the shallow near-shore area. However, the maintenance of a circumneutral bottom
water refuge does not necessarily mean that there are no adverse biological effects in the lake from
snowmelt. Fish fry (e.g., brook trout) that must migrate from the benthos to the surface water to fill their
air bladders after hatching in the spring may be greatly impacted by the acidic, high Aljm concentrations
119
-------
Woods Lake Outlet, New York
Spring 1986
to
o
-------
Towanda Creek, Pennsylvania
Spring 1986
10
ll I ^^
O 10
*~"O
c
0) id
fr (0
M 3
id o
10
•H
Q
10
8
6
4
ELS-I
0
03/01 03/15 03/29 04/12 04/26 05/10 05/24
03/08 03/22 04/05 04/19 05/03 05/17 05/31
Date
Figure 6-10. Hydrograph of mean daily discharge during spring 1986 (March 1 to June 1) in
Towanda Creek, Pennsylvania (Poconos/Catskills subregion). The spring ELS-II
seasonal sampling window in Pennsylvania is shown in brackets at top. Data from
USGS Hydrodata.
121
-------
Swift River, Maine
Spring 1986
1500
w
o
%»x
0)
fd
O
l/l
•H
Q
1000
0
03/01 03/15 03/29 04/12 04/26 05/10 05/24
03/08 03/22 04/05 04/19 05/03 05/17 05/31
Date
Figure 6-11. Hydrograph of mean daily discharge during spring 1986 (March 1 to June 1) in
Swift River, Maine. The spring ELS-II seasonal sampling window in Maine is shown
in brackets at top. Data from USGS Hydrodata.
122
-------
they experience in the surface water. Thus, assessing worst case acidic conditions during spring
snowmelt is not a simple task and is complicated by both spatial and temporal variability in the chemical
environment. The ELS-II spring index sample is indicative of post-snowmelt spring conditions in the
surface water. It should not be construed to represent worst case spring episodic conditions.
It would have been almost impossible to implement a regional synoptic survey of lakes at the time
of peak runoff. For one, the large number of lakes to be sampled necessitates a long sampling window.
Another problem is that identifying the peak discharge time is easy in hindsight, but identifying it in the
field over a large heterogeneous region is virtually an impossible task. Finally, lake sampling during
snowmelt Is difficult and potentially dangerous. Sampling from the middle of the lake is unsafe until after
ice-out, which is generally after the period of maximum discharge (Driscoll et al., in press). Thus the
ELS-II spring sample was probably the most reasonable index of lakewater chemistry that could be
made for a synoptic survey in the spring. Again, it should be emphasized that it does not represent
worst case episodic spring conditions. Minimum ANC values in the spring are almost assuredly lower
than the ELS-II spring index values and occur earlier.
123
-------
SECTION 7
LITERATURE CITED
Arent, L.J., M.O, Morison, and C.S. Soong. 1988. Eastern Lake Survey - Phase II, National Stream
Survey - Phase I, Processing Laboratory Operations Report. EPA 600/4-88/025, U.S. Environ-
mental Protection Agency, Office of Research and Development, Washington D.C.
American Public Health Association (APHA), American Water Works Association, and Water Pollution
Control Federation. 1985. Standard Methods for the Examination of Water and Wastewater. 16th
ed. APHA, Washington, D.C.
ASTM. 1984. Annual Book of ASTM Standards, Vol. 11.01, Standard Test Methods for Anions in Water
by Ion Chromatography, American Society for Testing and Materials, Philadelphia, PA.
Baird, F., D.C. Buso, and J.W. Hornbeck. 1985. Access pipes for multiple sampling under ice. Limnol.
Oceanogr. 30:1129-1130.
Baker, J.P., and C.L Schofieid. 1982. Aluminum toxicity to fish in acidic waters. Water Air Soil Pollut.
18:289-309.
Baker, J.P., D.P. Bernard, S.W. Christensen, M,J. Sale, and Others. 19903. Biological Effects of
Changes in Surface Water Acid-Base Chemistry. NAPAP Report 13, Acidic Deposition: State of
Science and Technology, National Acid Precipitation Assessment Program, Washington, D.C.
Baker, J.P., S.A. GherinI, S.W. Christensen, C.T. Driscoll, J. Gallagher, R.K. Munson, R.M. Newton, K.H.
Reckhow, and C.L Schofieid. 1990b. Adirondack Lake Survey: an interpretive analysis of fish
communities and water chemistry, 1984-87. Adirondack Lake Survey Corporation, Ray Brook, NY.
Baker, L.A., P.R. Kaufmann, A.T. Herlihy, and J.M. Eilers. 1990. Current Status of Surface Water Acid-
Base Chemistry. NAPAP Report 9, Acidic Deposition: State of Science and Technology, National
Acid Precipitation Assessment Program, Washington, D.C.
Beamish, R.J., W.L Lockhart, J.C. Van Loon, and H.H. Harvey. 1975. Long-term acidification of a lake
and resulting effects on fishes. Ambio 4:98-102.
Blakey, F.E., J.E. Cotton, and K.W. Toppin. 1985. Water Resources Data, New Hampshire and Vermont,
Water Year 1984. U.S. Geological Survey Water-Data Report NH-VT-84-1. U.S. Geological Survey,
Denver, CO.
Blick, D.J., J.J. Messer, D.H. Landers, and W.S. Overton. 1987, Statistical basis for the design and
interpretation of the National Surface Water Survey, Phase I: lakes and streams. Lake Reserv.
Manage. 3:470-475.
Church, M.R., and Others. 1989. Direct/Delayed Response Project: Future Effects of Long-Term Sulfur
Deposition on Surface Water Chemistry in the Northeast and Southern Blue Ridge Province.
EPA/600/3-89/061 a-d. U.S. Environmental Protection Agency, Washington D.C. 4 volumes.
Cusimano, R.F., J.P. Baker, W.J. Warren-Hicks, V. Lesser, W.W. Taylor, M.C. Fabrizio, D.B. Hayes, and
B.P. Baldigo. 1990. Fish Communities in Lakes in Subregion 2B (Upper Peninsula of Michigan) in
Relation to Lake Acidity. EPA/600/3-89/021. U.S. Environmental Protection Agency, Washington
D.C. 240 pp.
124
-------
Driscoll, C.T., and J.J. Bisogni. 1984. Weak acid/base systems in dilute acidified lakes and streams of
the Adirondack region of New York State. Pages 53-72 in J.L Schnoor, ed. Modeling of Total
Acidification Impacts. Butterworth Publishers, New York.
Driscoll, C.T., J.P. Baker, J.J. Bisogni, and C.L. Schofield. 1980. Effect of aluminum speciation on fish in
dilute acidified waters. Nature 284:161-164.
Driscoll, C.T. 1986. A Program to Evaluate Long-Term Changes in the Chemistry of Adirondack Lakes.
Annual Report to the Environmental Research Laboratory, U.S. Environmental Protection Agency,
Corvallis, OR.
Driscoll, C.T., B.J. Wyskowski, C.C. Cosentini, and M.E. Smith. 1987. Processes regulating temporal
and longitudinal chemistry of a low-order woodland stream in the Adirondack region of New York.
Biogeochemistry 3:225-242.
Driscoll, C.T, R.M. Newton, C.P. Gubala, J.P. Baker, and S.W. Christensen. In press. Adirondack
Mountains. Chapter 6 in D.F. Charles, ed. Acidic Deposition and Aquatic Ecosystems: Regional
Case Studies. Springer-Verlag, New York.
Eshieman, K.N. 1988. Predicting regional episodic acidification of surface waters using empirical
techniques. Water Resour. Res. 24:1118-1126.
EPA. 1983. Methods for Chemical Analysis of Water and Wastes. EPA/600/4-79/020. U.S. Environ-
mental Protection Agency, Washington, D.C.
Firda, G.D., R.R. Cumia, and P.M. Burke. 1985. Water Resources Data, New York, Water Year 1984.
Volume 1, Eastern New York, Excluding Long Island. U.S. Geological Survey Water-Data Report
NY-84-1. U.S. Geological Survey, Denver, CO.
Firda, G.D., R.R. Cumia, and P.M. Burke. 1987. Water Resources Data, New York, Water Year 1986.
Volume 1, Eastern New York, Excluding Long Island. U.S. Geological Survey Water-Data Report
NY-86-1. U.S. Geological Survey, Denver, CO.
Firda, G.D., R.R. Cumia, and P.M. Burke. 1988. Water Resources Data, New York, Water Year 1987.
Volume 1, Eastern New York, Excluding Long Island. U.S. Geological Survey Water-Data Report
NY-87-1. U.S. Geological Survey, Denver, CO.
Gadoury, R.A., G.G. Girouard, K.G. Ries, and H.L White. 1985. Water Resources Data, Massachusetts
and Rhode Island, Water Year 1984. U.S. Geological Survey Water-Data Report MA-RI-84-1. U.S.
Geological Survey, Denver, CO.
Galloway, J.N., C.L Schofield, G.R. Hendrey, N.E. Peters, and A.H. Johannes. 1980. Sources of acidity
in three lakes acidified during snowmelt. Pages 264-265 in D. Drablos and A. Tollan, eds. Eco-
logical Impact of Acid Precipitation. SNSF Project, Oslo, Norway.
Gubala, C.P., C.T. Driscoll, R.M. Newton, and C.L Schofield. In review. The chemistry of a near-shore
lake region during spring snowmelt. Environ. Sci. Techno!.
Haskell, C.R., W.P. Bartlett, Jr., W.B. Higgins, and W.J. Nichols, Jr. 1985. Water Resources Data, Maine,
Water Year 1984. U.S. Geological Survey Water-Data Report ME-84-1. U.S. Geological Survey,
Denver, CO.
Hillman, D.C., S.L Pia, and S.J. Simon. 1987. National Surface Water Survey, National Stream Survey,
Analytical Methods Manual. EPA/600/8-87/005. U.S. Environmental Protection Agency, Las
Vegas, NV.
125
-------
Horvitz, D.G., and D.J. Thompson. 1952. A generalization of sampling without replacement form a finite
universe. J. Amer. Stat. Assoc. 47:663-685.
Jeffries, D.S., C.M, Cox, and P.J. Dillon. 1979, Depression of pH In lakes and streams in central Ontario
during snowmelt. J. Fish. Res. Board Can. 36:640-646.
Jimenez, M.M., R.G. Johnson, T.C. Myers, B.D. Miller, M.M. Lang, A.K. Pollack, and S.D. Reynolds, in
Press. National Surface Water Survey: Eastern Lake Survey - Phase II, Data Base Dictionary.
U.S. Environmental Protection Agency, Corvallis, OR.
Johnson, C.B., T.J. Sullivan, and D.J. Blick. 1989. Defining regional populations of lakes for the
assessment of surface water quality. Water Resour. Bull. 25:565-572.
Kanciruk, P., J.M. Eilers, R.A. McCord, D.H. Landers, D.F. Brakke, and R.A. Linthurst. 1986.
Characteristics of Lakes in the Eastern United States. Volume III. Data Compendium of Site
Characteristics and Chemical Variables. EPA/600/4-86/007c. U.S. Environmental Protection
Agency, Washington D.C. 439 pp.
Kaufmann, P.R., A.T. Heriihy, J.W. Eiwood, M.E. Mitch, W.S. Overton, M.J. Sale, J.J. Messer, K.A.
Cougan, D.V. Peck, K.H. Reckhow, A.J. Kinney, S.J. Christie, D.D. Brown, C.A. Hagley, and H.I.
Jager. 1988. Chemical Characteristics of Streams in the Mid-Atlantic and Southeastern United
States. Volume I. Population Descriptions and Physieo-Chemieal Relationships. EPA/600/
3-88/021 a, U.S. Environmental Protection Agency, Washington, D.C. 397 pp.
Kelly, C.A., J.W.M. Rudd, and D.W. Schindler. 1990. Acidification by nitric acid - future considerations.
Water Air Soil Pollut. 50:49-61.
Kerfoot, H.B., I.E. Lewis, D.C. Hillman, M.L Faber, and I.E. Mitchell-Hall. 1988. Eastern Lakes Survey
(Phase II) Analytical Methods Manual. EPA/600/4-88/031. U.S. Environmental Protection
Agency, Environmental Monitoring Systems Laboratory, Las Vegas, NV.
Kramer, J.R. 1984. Modified Gran analysis for acid and base titrations. Environmental Geochemistry
Report No. 1984-2. McMaster University, Hamilton, Ontario, Canada.
Krug, E.G., and C.R. Frink. 1983. Acid rain on acid soil: a new perspective. Science 221:520-525.
Landers, D.H., J.M. Eilers, D.F. Brakke, W.S. Overton, P.E. Kellar, M.E. Silverstein, R.D. Schnobrod, R.E.
Crowe, R.A. Linthurst, J.M. Omernik, S.A. league, and E.P. Meier. 1987. Characteristics of Lakes
in the Western United States. Volume I. Population Descriptions and Physico-Chemical Relation-
ships. EPA/600/3-86/054a. U.S. Environmental Protection Agency, Washington, D.C. 176 pp.
Landers, D.H., W.S. Overton, R.A. Linthurst, and D.F. Brakke. 1988. Eastern Lake Survey: regional
estimates of lake chemistry. Environ. Sci. Technol. 22:128-135.
Linthurst, R.A., D.H. Landers, J.M. Eilers, D.F. Brakke, W.S. Overton, E.P. Meier, and R.E. Crowe. 1986.
Characteristics of Lakes in the Eastern United States. Volume I, Population Descriptions and
Physico-Chemical Relationships EPA/600/4-86/007a. U.S. Environmental Protection Agency,
Washington, D.C. 136 pp.
McQuaker, N.R., P.O. Kluckner, and O.K. Sandberg. 1983. Chemical analysis of acid precipiation: pH
and acidity determinations. Environ. Sci. Technol. 17:431-435.
Merritt, G.D., and V.A. Sheppe. 1988. Eastern Lake Survey - Phase II, Field Operations Report.
EPA/600/4-88/024. U.S. Environmental Protection Agency, Las Vegas, NV.
126
-------
Mitchell-Hall, I.E., A.C. Neale, S.G. Paulsen, and J.E. Pollard. 1989. Eastern Lake Survey - Phase II:
Quality Assurance Report. EPA/600/4-89/029. U.S. Environmental Protection Agency, Las Vegas,
NV.
Molot, LA., P.J. Dillon, and B.D. LaZerte. 1989. Factors affecting alkalinity concentrations of
streamwater during snowmelt in Central Ontario. Can. J. Fish. Aquat. Sci. 46:1658-1666.
Newell, A.D., C.F. Powers, and S.J. Christie. 1987. Analysis of Data from Long-Term Monitoring of
Lakes. EPA/600/4-87/014. U.S. Environmental Protection Agency, Washington, D.C. 150 pp.
O'Dell, J.W., J.D. Pluff, M.E. Gales, and G.D. McKee. 1984. Technical Additions to: Methods for
Chemical Analysis of Water and Wastes. Method 300.0. The Determination of Inorganic Anions in
Water by Ion Chromatography. EPA/600/4-85/017. U.S. Environmental Protection Agency, Ohio.
Omernik, J.M., and C.F. Powers. 1983. Total alkalinity of surface waters-a national map. Ann. Assoc.
Am. Geog. 73:133-136.
Overton, W.S., P. Kanciruk, LA. Hook, J.M. Eilers, D.H. Landers, D.F. Brakke, D.J. Blick, R.A. Linthurst,
M.D. DeHaan, and J.M. Omernik. 1986. Characteristics of lakes in the Eastern United States.
Volume II. Lakes Sampled and Descriptive Statistics for Physical and Chemical Variables.
EPA/600/4-86/007b. U.S. Environmental Protection Agency, Washington D.C. 374 pp.
Overton, W.S. 1987. Phase II Analysis Plan, National Lake Survey - Working Draft, April 15, 1987.
Technical Report 115. Department of Statistics, Oregon State University, Corvallis.
Overton, W.S. 1989. Calibration Methodology for the Double Sample Structure of the National Lake
Survey - Phase II Sample. Technical Report 130. Department of Statistics, Oregon State
University, Corvallis.
Peden, M.E. 1981. Sampling Analytical and Quality Assurance Protocols for the National Atmospheric
Deposition Program. ASTM D-22 Symposium and Workshop on Sampling and Analysis of Rain.
ASTM, Philadelphia, PA.
Peter, N.E., and C.T. Driscoll. 1987. Hydrogeological controls on surface water chemistry in the
Adirondack region of New York State. Biogeochemistry 3:163-180.
Rosenqvist, I. 1978. Acid precipitation and other possible sources for acidification of rivers and lakes,
Sci. Total. Environ. 10:271-272.
Ruby, A., P.J. Godfrey, and O.T. Zajicek. 1988. The Massachusetts Acid Rain Monitoring Project, Phase
II. Water Resources Center Publication No. 159. University of Massachusetts, Amherst.
Sarndal, C.E., and B. Swensson. 1987. A general view of estimation for two phases of selection with
applications to two-phase sampling and nonresponse. International Statistical Review 55:279-294.
Schaefer, D.A., C.T. Driscoll, R. Van Dreason, and C.P. Yatsko. 1990. The episodic acidification of
Adirondack Lakes During Snowmelt. Water Resour. Res. 26:1639-1647.
Schindler, D.W. 1988. Effects of Acid Rain on Freshwater Ecosystems. Science 239:149-157.
Schofield, C.L. 1976. Acid precipitation: effects on fish. Ambio 5:338-340.
127
-------
Skougstad, M.W., J. Fishman, LC. Friedman, D.E. Erdmary, and S.S. Duncan, eds. 1979. Methods for
Determination of Inorganic Substances in Water and Fluvial Sediments: Techniques of Water
Resources Investigations of the United States Geological Survey, Book 5, Chapter A1. U.S.
Government Printing Office, Washington, D.C.
Snedecor, G.W., and W.G. Cochran. 1980. Statistical Methods. The Iowa State University Press, Ames,
Iowa.
Stehman, S.V., and W.S. Overton. 1987. Estimating the variance of the Horvitz-Thompson estimator for
variable probability, systematic samples. Proc. 1987, ASA Survey Research Section. San Fran-
cisco, CA.
Sullivan, T.J. 1990. Historical Changes in Surface Water Acid-Base Chemistry in Response to Acidic
Deposition. NAPAP Report 11, Acidic Deposition: State of Science and Technology, National
Acid Precipitation Assessment Program, Washington, D.C.
Sullivan, T.J., D.F. Charles, J.P. Smol, B.F. Gumming, A.R. Selle, D.R. Thomas, J.A. Bernert, and S.S.
Dixit. 1990. Quantification of changes in lakewater chemistry in response to acidic deposition.
Nature 345:54-58.
Sullivan, T.J., C.T. Driscoll, S.A. Gherini, R.K. Munson, R.B. Cook, D.F. Charles, and C.P. Yatsko. 1989.
Influence of aqueous aluminum and organic acids on measurement of acid neutralizing capacity in
surface waters. Nature 338:408-410.
Tessier, A.J., and R.J. Horwitz. 1988. Analysis and interpretation of zooplankton samples collected
during Phase II of the National Lake Survey. Acad. Nat. Sci. Phila. Rep. 88-18.
Tessier, A.J., and R.J. Horwitz. 1990. Influence of water chemistry on size structure of zooplankton
assemblages. Can. J. Fish. Aquat. Sci. 47:1937-1943.
Thornton, K., J.P. Baker, K.H. Reckhow, D.H. Landers, P.J. Wigington, Jr. 1986. National Surface Water
Survey, Eastern Lake Survey-Phase II, Research Plan. U.S. Environmental Protection Agency,
Washington, D.C.
Wigington, P.J., T.D. Davies, M. Tranter, and K.N. Eshleman. 1990. Epidsodic Acidification of Surface
Waters Due to Acidic Deposition. NAPAP Report 12, Acidic Deposition: State of Science and
Technology, National Acid Precipitation Assessment Program, Washington, D.C.
Wright, R.F., and E.T. Gjessing. 1976. Acid precipitation: changes in the chemical composition of
lakes. Ambio 5:219-223.
128
-------
APPENDIX A
POPULATION ESTIMATES OF SELECTED PHYSICAL AND CHEMICAL VARIABLES
IN THE SPRING, SUMMER, AND FALL ELS-II SEASONAL SURVEYS
This appendix presents the cumulative distribution functions (CDF) for the major ELS-II variables
measured in the spring, summer, and fall seasonal surveys (Table A-1). CDFs are described in Section
5.1.2 and procedures for making these types of population estimates are presented in Section 2.4.2. To
read these figures, pick a value, x, of the attribute X, along the horizontal axis, and read the y-axis value
of the two curves, F(x) (solid line) and FU(X) (dotted line) at this value. F(x) is the estimated proportion
of the number of lakes in the population with a value of the attribute equal to or less than x. FU(X) is the
upper confidence bound on this proportion, to be read as: one is 95% confident that the true proportion
is less than this bound. Some distributions plot the declining function, 1 - F(x). For these, read the
distribution as the estimated proportion of reaches having a value of the attribute equal to or greater
than X. Confidence bounds are as before: an upper confidence bound on the true proportion.
A lower one-sided confidence bound on F(x) can be generated, if needed, by measuring the dis-
tance between the two curves, and projecting the identical distance below F(x). This curve is not
presented because it could be confused with two-sided confidence bounds, which are of different width.
The bound provided is the one usually considered appropriate for expressing the status of the resource.
By generating distribution functions for other characteristics, such as lake surface area, other
types of distributions can be obtained. In this appendix, G(x) represents the proportion of lake surface
area (in hectares) associated with lakes having values of the attribute X< some value x. In all cases,
these distributions and the represented confidence bounds are dimensionless, and a shift from hectares
to square miles has no effect on G(x) or Gu(x). Any other distribution, such as lake watershed area or
shoreline length can be analyzed in exactly the same way, requiring only an appropriate lake attribute
that sums to the population attribute.
Other population statistics of interest can be generated from the distributions. For each distribu-
tion, the quantiles, median, and four quintiles, C^, Q2...,Q4 (20th percentile, ... 80th percentile) are
identified. The median of the population is the value of x such that F(x) = 1/2. The Q| (quintile) of the
population is the value of x such that F(x) = i/5. These statistics are defined for all distributions.
Additionally, the mean and standard deviation of the variable x for the population is estimated as in
Section 2.4, equations 7 and 8.
Making ELS-II Variance Estimates
^
The formula for estimated variance of the estimated population totals (T ) is:
V(f) = I y2W2|(W2, - 1) + I Z y.y.^WZ - W,,) (A-1)
S ieS jeS
]/'
129
-------
Table A-1. Listing of Variable Names and Descriptions for CDFs Presented in Appendix A.
Variable Name
Description
ACC011
ALD02
ALO_02
ALDI98
ALEX11
ALKA11
CA98
CL98a
COLOR02
COND11
DIC02
DOC11
FE11
FTL98
K98
MG98
MN11
NA98
NH498
NO398
PH02
PTL11
SECME98
SIO211
SO498
SOBC98
Base Neutralizing Capacity
Total PCV reactive (monomeric) aluminum
Nonexchangeable (organic) PCV reactive aluminum
Inorganic monomeric aluminum (ALD02 - ALO_02)
MIBK-extractable aluminum
Acid neutralizing capacity
Calcium
Chloride
Color
Specific conductance - analytical laboratory
Closed headspace dissolved inorganic carbon - processing laboratory
Dissolved organic carbon
Iron
Total Fluoride
Potassium
Magnesium
Manganese
Sodium
Ammonium
Nitrate
Closed headspace pH - processing laboratory
Total phosphorous
Secchi depth
Silica
Sulfate
Sum of base cations (CA98 + MG98 + K98 + NA98)
Chloride CDFs are not presented for spring and summer seasonal surveys because of data quality concerns (see Section
4).
130
-------
where W- = A^B- is calculated from the following equations.
If lakes i and j are from the same ELS-I strata then: Ay = W1m S(N* - 1) / (N*/W1mJ - 1).
* If lakes I and ] are from different ELS-I strata then: Ay = W1m ,W1m .
• If the conditional ELS-II weight (Wcond) of lake i equals 1 then: B.. = WooncJ.,
• If the conditional ELS-II weight (Wcond) of lake j equals 1 then: B» = Wcorid _..
• If lakes i and j are from different ELS-II clusters then, B^ = Woond jWcond j.
Otherwise, if lakes i and j are from the same ELS-II cluster then:
6^ = (N'(N' -a)) / (n'(n- -1)W10.W1OJ); where N' = nW^W^j and a = (W1Q>j + W1oj)/2.
Variables are defined as:
* W10 = original ELS-I sample weight at the time of ELS-II site selection.
* ^1m = m°dified ELS-I sample weight reflecting minor adjustments made after ELS-li site
selection.
* ^cond = ELS-II conditional weight (see Section 2, equation 2).
* n' = the cluster sample size of the initial ELS-II target population after subtracting lakes
entered with certainty (Wcond = 1). Cluster I n' equals 45, cluster II n' equals 58, and cluster
III n' equals 49,
The standard error (SE) of the estimated population total is then calculated as the square root of the
variance estimate, V(T ). One-sided upper 95% confidence bounds (Tu) are then calculated as Tu = T
+ 1.645(SE(fy)).
For example, there were 20 sample lakes in the ELS-II with fall DOC < 2 mg/L. These represent
a population of 298 lakes (calculated by summing the sample weights) or 7.5% of the target population
(F(x) = 298/3,993 = 0.075). By using equation A-1, the estimated variance of the subpopulation was
4,173 with a standard error of 64.6, The 95% upper confidence level Is 404 (298 + 1.645 * 64.6). In
other words, we would say that there are 298 (± 64.6) ELS-II target lakes in the northeastern United
States that had fail 1986 DOC < 2 mg/L; we are 95% sure that there are no more than 404 lakes with
DOC < 2 mg/L.
131
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: ACCO1 1 (/ueq/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
1.0
0.6-
0.4-
0.2-
0.0
DIRECT
25
50.
75.
100.
125.
150.
Min: 5.40 Q : 41,95
Q2: 53.70
61.30
Median:
Mean:
Std. Dev.:
58.80
58.84
20.92
72.70 Max: 174.7
z
100.
125.
150.
Min: 5.40 Q : 37.10
= 48.10 Q3: 58.80 Q4: 62.60 Max: 174.7
Median:
Mean:
Std. Dev.:
48.30
53.12
15.84
o
UJ
f/1
* Confidence bounds ore for number and area of lakes, but have been scaled.
132
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: ACCO1 1 (/xeq/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A); 354924 SE(A): 68995
1.0r
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
0.6
0.4-
0.2
0.0+
0
Min: 15.00
Median:
Mean:
Std. Dev.:
50.
75.
100.
125.
150.
26.60 Q2: 33.50 Qg: 41.60 Q4: 59.80 Max: 204.4
37.50
46.68
29.50
G(X)'
0.8-
0.6
0.2-
0.04-
0
25
DIRECT
50,
75.
100.
Min: 15.00 Q,: 25.60 Q0: 28.50 Q,: 35.05 Q
1 u *>
125. 150.
38.70 Max: 204.4
Median:
Mean:
Std. Dev.:
30.85
33.80
13.33
_J
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: ACC011 (/ueq/l)
09 MAY 1990
Populat
Lake Ar
1.0-
0.8-
0.6-
F(X)'
0.4-
0.2-
0.0-
ion Size (N); 3993 SE(tM): 191.6 Sample Size; 145
ea (A); 354924 SE(A); 68995
/
J
). 25.
Min: 3.00 Q.: 29
Med
Mea
Std
1.0-
0.8-
0.6-
G(X)'
0.4-
0.2-
o.oj
ian: 38
n: 43
. Dev.: 22
I/
/ [
J
25.
Min: 3.00 Qji 27.
Median: 34
Mean: 36
Std. Dev.: 11
__~— — ' ™ — : — "
f/
/"'/
' ^
/
50, 75. 100.
REGION 1
CLUSTERS
1,2, & 3
DIRECT
125. 150.
.40 Q : 34.60 Q_: 41.30 Q . : 52.80 Max: 222.2
t) o 4
.90
.81
.48
y~~
50, 75. . 100.
DIRECT
125. 150.
85 Q0: 32.80 «„: 37.50 Q.: 42.50 Max: 222.2
i*5^
60
59
70
00
"Z.
Confidence bounds are for number ond area of lakes, but hove been scaled.
Q
UJ
00
<
I
a,
134
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE; ALD02 (jug/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
Sample Size: 145
REGION 1
CLUSTERS
1,2, & 3
1.0
I-F(X)'
0.8-
0.6-
0.4-
0.2-
0.0-
0. 50.
Min: 6.10
Median:
Mean:
Std. Dev.:
DIRECT
100.
150.
200.
250.
300.
350.
400.
16.10
28.70
52.50
76.03
Q2: 24.30
Qg: 33.90
Q4: 62.50
450. 500.
Max: 517.2
1.0-
DIRECT
TOO
150.
Min; 6.10 Q : 15.90
200. 250. 300. 350. 400. 450. 500.
25.10 Q,: 35.40 Q • 61.10 Max: 517.2
Median:
Mean:
Std. Dev.
28.30
42.40
41.49
* Confidence bounds are for number and area of lakes, but have been scaled.
I/)
<
'<
0
LJ
00
I
CL
135
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: ALD02 Og/0 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68997
Sample Size: 145
0.6-
0.4-
0.2-
0.0-
DIRECT
0.
50.
100.
150.
200.
250.
300.
350.
Min: 3.10
Median:
Mean:
Std. Dev.:
11.30
18.80
31.68
43.75
: 16.40
23.80 Q4: 32.40 Max: 328.5
l-G(X)'
1 0
0.8-
0.6-
0.4-
0.2'
o.o
DIRECT
50.
100.
150.
200.,
250.
300.
350.
DO
Min: 3.10
Median:
Mean:
Std. Dev.:
12.60 Q2: 17.80 Q3: 20.00 Q4: 24.80 Max: 328.5 Q
18.80
22.71
21.43
Ld
Confidence bounds ore for number and area of lakes, but have been scaled.
CL
136
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE; ALD02 G/g/l) 09 MAY 1990
Population Size (N): 3993 SE(N): 191,6
Lake Area (A); 354924 SE(A): 68996
1.0
Sample Size. 1 45
REGION 1
CLUSTERS
1,2, & 3
0.6-
0.4-
0.2-
0.0
50
DIRECT
100 150. 200. 250. 300. 350.
400.
450. 500.
Min: 8.60 Qn: 16.50 Q0: 20.10 Q«: 26.00 Q.: 41.10 Max: 626.5
i & «3 4
Median: 22.70
Mean: 40.12
Std. Dev.: 60.48
0.6-
0.4-
0.2-
o.o-l
0.
DIRECT
50. 100. 150. 200 250. 300. 350. 400. 450. 500.
Q: 24.50 Q: 35.40 Max: 626.5
Min: 8.60 Q^ 16.50 Q2: 21.60
Median:
Mean:
Std. Dev.:
22.30
27.44
26.05
* Confidence bounds are for number and area of lakes, but have been scaled.
if)
<
Q
LJ
CO
IE
Q..
137
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: ALO_02 Gug/l) 10 MAY 1990
Population Size (N): 3993 SE(N); 191.6
Lake Area (A): 354924 SE(A); 68996
Sample Size: 1 45
1.0-
0.6-
0.4-
0.2-
0.0
DIRECT
Min: 6.20
Median:
Mean:
Std. Dev.:
25.
18.10
26.60
31.06
19.64
50.
75.
23.10
Qg: 29.80
100. 125. 150.
L: 39.80 Max: 155.6
1.0-
0.8-
0.6-
0.1
0.2-
0.0
Median:
Mean:
Std. Dev.:
DIRECT
25.
Min: 6.20 Q 21.40
50.
100.
125.
150.
: 26.80 Qg: 30.60 Q4: 41.10 Max: 155.6
30.60
32.93
15.1?
* Confidence bounds ore for number and area of lakes, but have been scaled.
CO
Q
LjJ
00
<
31
n..
138
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: AL0^02 (/ug/l) 10 MAY 1990
Population Size (N); 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
0.0
Min: 0.00 Q,: 2.90 Q0: 5.50
Median:
Mean:
Std. Dev.:
7.50
8.30
7.3?
Q3:
8.80
Q : 12.60
Max: 42.70
i-G(xr
0.8~
0.6
0.4-
0.0-
DIRECT
10
15. 20. 25. 30. 35. 40. 45.
50
Min: 0.00 Q1: 2.90 Q : 5.50 Q •
JL & o
Median:
Mean:
Std. Dev.:
8.00
7.83
5.47
9.80 Q4: 12.60 Max: 42,70
Confidence bounds ore for number and area of lakes, but hove been scaled.
139
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: ALO_02 Og/l) 09 MAY 1990
Population Size (N): 3993 SE(N); 191.6
Lake Area (A). 354924 SE(A): 68996
Sample Size; 1 45
I-F(X)'
100.
125.
150.
Min: 0.00 Qji 7.40 Q£: 10.30 Q3: 14.10 Q4: 20.70 Max: 154.7
Median: 12.50
Mean: 17.11
Std. Dev.: 17.26
1.0-
I-G(X)'
0.8-
0.6-
0.4-
0.2-
DIRECT
25.
Min: 0.00 Q.: 7.50
50. 75. . 100.
= 12.50 Q: 14.50
125.
150.
18.60 Max: 154.7
Median:
Mean:
Std. Dev.:
14.50
13.70
8.88
CO
GO
<
I—
Q
Ld
Confidence bounds ore for number and area of lakes, but have been scaled.
IE
CX.
140
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: ALD198 (fig/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
Sample Size; 145
1.C
0.8-
0.6-
0.4-
0.2-
o.a
-50. 0.
Min: 0.00
Median:
Mean:
Std. Dev.:
DIRECT
50. 100. 150. 200. 250. 300. 350. 400. 450. 500.
0.00
2.20
23.60
66.75
Q2: 0.20 Q3: 4.30 Q4: 16.50 Max: 481.6
0.8-
0.6-
0.4-
0.2-
.0*
niprp
\
%
v
1
-50 0
Min: 0.00
Median:
Mean:
Std. Dev.:
50. 100
0.00
0.00
12.19
34.38
j
150. 200. 250. ,300. 350. 400. 450. 500
0.00 QQ: 3.20 Q.: 16.80 Max: 481.6
«J T
o
* Confidence bounds are for number and area of lakes, but have been scaled.
141
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: ALD198 0*9/0 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
0.6-
0.4-
0.2-
0.0
Sample Size; 1 45
DIRECT
-50.
0.
50.
Min: 0.00 Q.: 5.40 Q :
Median:
Mean:
Std. Dev.;
11.70
23.52
43.06
100.
150.
200.
250.
300.
9.40 Q3: 13.80 Q4: 25.00 Max: 319.9
1.0
0-t
0.6-
0.4-
0.2-
0.0
-50.
0.
50.
100.
Min: 0.00 Q^. 5.40 Q2= 10.60
Median:
Mean:
Std. Dev.:
11.30
14.92
20.93
DIRECT
150.
200.
250.
300.
-. 12.20 Q4: 16.80 Max: 319.9
GO
o
Ld
CO
Confidence bounds ore for number and area of lakes, but have been scaled.
Ou
142
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: ALDI98 (ywg/l)
09 MAY 1990
Population Size (N): 3993 SE(N): 191,6
Lake Area (A); 354924 SE(A): 68996
Sample Size: 145
l-F(X)'
Median:
Mean:
Std. Dev.:
Q4: 20.20
450. 500.
Max: 471.8
10.90
23.01
47.82
1.0-
1~G(XV
0.8
06
0.4-
0.2
0.0
DIRECT
50. 100, 150. 200. 250. 300. 350. 400. 450. 500.
Min; 1.50
Median:
Mean;
Std. Dev.:
6.60
8.90
13.73
21.03
8.30 Q : 10.70 Q : 17.80 Max: 471,
o 4
o
LaJ
fonfidence bounds
for number and area of lakes, but have been scaled.
IE
O.
143
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: ALEX11 (fig/l) 09 MAY 1990
Population Size (N): 3993 SE(N); 191,6
Lake Area (A): 354924 SE(A): 68996
Sample Size: 145
REGION 1
CLUSTERS
1,2, & 3
0.8
0.6-
0.4
0.0+-
0.
DIRECT
50.
100.
150.
200.
250.
Min: 0.00
Median:
Mean:
Std. Dev.:
0.00
12.10
32.47
61.46
6.80 Q3: 20.70 Q4: 42,10 Max: 435.2
Min: 0.00
Median:
Mean:
Std. Dev.:
DIRECT
100.
150.
200.
0.00
10.30
23.69
36.17
Qg: 4.60 Q3: 13.90 Q4: 42.10
250. 300.
Max: 435.2
I —
Q
Ld
* Confidence bounds ore for number and area of lakes, but. have been scaled.
I
0,
144
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: ALEX1 1 Gug/0 09 MAY 1990
Population Size (N): 3993 SE(N): 191,6
Lake Area (A): 354924 SE(A); 68996
Sample Size; 145
i-O
0.6-
0.4-
0.2-
o.a
o.
DIRECT
50.
100.
150.
200.
250.
Min: 0.00 Q,: 0.00
Median:
Mean:
Std. Dev.:
5.10
19.38
40.48
J2: 3.10 Qg: 9.00 Q4". 25.30 Max: 298.8
1.0-
0.8-
0.6-
0.4-
o ?-
0.0
50.
DIRECT
TOO.
150,
200.
250.
GO
Min: 0.00 CL: 0.30 Q
1*
Median:
Mean:
Std. Dev.:
4.40
8.82
20.81
1.60 Qg: 7.30 Q4: 10.40 Max: 298,8
Confidence bounds ore for number and area of lakes, but have been scaled.
Q
00
<
X
Qu
145
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: ALEX1 1 (/ug/l)
09 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
I.O-i
0.6-
0.4-
O.O4-
0.
DIRECT
50
150.
200.
250.
300.
Min: 0.00
Median:
Mean:
Std. Dev.:
4.40
7.40
22.64
48.17
6.30 Qg: 9.80 Q4: 24.10 Max: 467,2
1.0
0.8-
0.6-
0.4-
0.2-
DIRECT
100.
150.
200.
250.
300.
10
z
Min: 0.00
Median:
Mean:
Std. Dev.:
2.90 Q2: 4.50
6.30
12.04
21.63
3
8.30 Q4: 18.30 Max: 467.2
Confidence bounds ore for number and area of lakes, but hove been scaled
Q
LJ
00
<
IE
146
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: ALKA1 1 (/ueq/l) 09 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
1.0
0.8
0.6-
150. 200.
250.
300. 350.
Min: -63,1 Q : 20.10 Q,: 60.80 Q_: 99.20 Q . : 168.1
I / o 4
Max: 338.3
Median:
Mean:
Std. Dev.:
79.10
100.78
92.38
0.6
0.4-
0.2-
0 04—
-50.
DIRECT
50.
100.
150.
200.
250.
300.
350.
Min: -63.1 Q.: 30.60 Q»: 68.90 Q,,: 109.2 Q.: 235.8 Max: 338.3
Median:
Mean:
Std. Dev.:
88.90
116.96
93.95
Q
Ld
Confidence bounds are for number and area of lakes, but hove been scaled.
147
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: ALKA1 1 (/ueq/l) 09 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A); 354924 SE(A); 68996
1.0
Sample Size: 145
F(X)'
0.8-
0.6-
50. 100
150.
200,
250.
300.
350.
Min: -48.6 Q1: 28.80 Q : 84.70 Q • 140.2
•I- w «J
Median: 107.80
Mean: 132.89
Std. Dev.: 115.09
Q4: 230.4 Max: 454.8
G(X)'
10
0.6-
0.4-
0.0
-50.
0.
DIRECT
50.
100.
150.
200
250.
300.
350,
"Z.
Min: -48.6 Q : 52.55 Q2'. 94.70
: 146.1 Q4: 286.7 Max: 454.8
Median:
Mean:
Std. Dev.:
109.80
151.49
108.54
* Confidence bounds ore for number and area of lakes, but have been scaled.
o
UJ
00
<
X
CL
148
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE; ALKA1 1 (/ieq/l)
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
1 n
! , UT""
09 MAY 1990
Sample Size: 1 45
REGION 1
CLUSTERS
F(X)'
: 26,60 Qg: 92.20 Qg: 179.5
Median: 119.20
Mean: 149.39
Std. Dev.: 126.16
Q.t 261.3 Max: 461.2
4
G(X)'
1.0
0 6-
0.4-
0.2-
0 0
*f
50 0 50
Min: -54.3 Q : 56.80
Median: 117.20
Mean: 166.11
Std. Dev.: 118.16
100.
150.
200.
DIRECT
3(30 350.
: 101.1 Qg: 181.4 Q4: 298.3 Max: 461.2
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: CA98 (p^eq/l) 09 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68995
•t
I -
Sample Size: 145
F(X)<
0.6-
0.4-
0.2-
O.O
DIRECT
100.
200.
300.
400.
500.
Min: 27.44 Q: 88.27
Median;
Mean:
Std. Dev.:
139.17
162,45
89.61
,: 125.2 Q3: 150.4 Q4: 246.8 Max: 441.9
G(X)«
i.O-
0.6-
0.4^
100.
200.
300.
DIRECT
400.
500.
00
Min: 27.44
Median:
Mean:
Std. Dev.:
: 96.16
140.62
179.88
90.52
134.4 Qg: 168.5 Q4: 275.2 Max: 441.9
Confidence bounds are for number and area of takes, but have been scaled.
o
oo
<
I
a.
150
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE. CA98 (/-ieq/l) 09 MAY 1990
F(X)'
G(X)'
Populal
Lake Ar
1,0
0.8
0.6-
0.4-
0.2-
0.0-
REGION 1
ion Size (N): 3993 SE(N): 191.6 Sample Size: 145 CLUSTERS
ea (A); 354924 5E(A): 68994 1 ,2, & 3
^.,^=^^
^S
f J
/ * *
j ft
tJ J
//
x7
,Jf DIRECT
) 100- 200- 300- 400, 500.
Min: 23.40 Q^ 88.12 Qg: 128.2 Q : 169.2 Q.: 260.2 Max: 431.6
Mec
Mea
Std
1,0
0.&
0.6-
0 4
0.2-
0.0-
0
tian: 153.89
in: 175.76
. Dev.: 101.95
/ 'r~^^
r—"
^\
A^
Jf DIRECT
— ^
100, 200. 300- 400 500-
Min: 23.40 Q, : 109.2 Q0: 144.6 Q_: 182.5 Q.: 309.7 Max: 431.6
1 I 3 4
Median: 162.92
Mean: 197.51
Std. Dev.: 102.12
CO
* Confidence bounds are for number and area of lakes, but have been scaled.
Q
LiJ
00
<
jC
0_
151
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: CA98 (/ueq/l) 09 MAY 1990
Population Size (N); 3993 5E(N): 191.6
Lake Area (A): 354924 SE(A): 68995
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
F(X)'
0.6-
0.4-
0.2-
0.0+
0
DIRECT
100.
Median:
Mean:
Std. Dev,:
163.67
190.69
109.84
200.
Min: 29.89 Q1: 94.31 Q : 137.1
J. £j
300.
400.
500.
190.3 Q4: 293.4 Max: 480.2
G(X)'
1.0-
0.8
0.6-
0.4
0.2-
O.o-l~-
0.
DIRECT
100.
200.
300.
400.
500.
CO
Min: 29.89
Median:
Mean:
Std. Dev.:
QJ-. 118.4
182.88
212.58
109.85
Q2: 151.4
Q3: 208.6
337.0 Max: 480.2
Confidence bounds ore for number and area of lakes, but have been scaled.
Q
Ld
CO
<
X
CL
152
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: CL98 (/ueq/l)
09 MAY 1990
Population Size (N): 3993 SE(N): 191,5
Lake Area (A); 354924 SE(A); 68996
1.0'
0.8-
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
I-F(X)'
0.6-
0 4-
0.2-
0.0*
400
500.
600.
700.
Min: 5.61 Q1: 14.11 Q0: 40,45 Q«: 87.90 Q • 224.6 Max: 667.9
J- £* . 134.4 Max: 667.9
1 & J 4
Median:
Mean:
Std. Dev.;
59.13
106.91
129,61
<
Q
LJ
Confidence bounds ore for number ond area of lakes, but have been scaled.
I
a_
153
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: COLOR02 (PCU) 10 MAY 1990
PC
La
FIX)
G(X)'
>pulat
ke Ar
1.0-
0.8-
0.6-
0.4-
0.2-
0.0-
ion Size (N): 3993 SE(N): 191.6 Sample Size: 145
ea (A): 354924 SE(A): 68996
^
sv
). 10 20
Min: 0.00 Q : 10
Medi
Mean
Std.
I.Oi
0,8-
0.6-
0.4-
0.2-
0.0-
an: 20
: 23
Dev.: 13
30. 40. 50. 60. 70. 80
rREGION 1
CLUSTERS
1,2, & 3
DIRECT
90. 100.
.00 Q2: 20.00 Qg: 25.00 Q4: 30.00 Max: 95.00
.00
.64
.98
^~"~A '\ DIRECT
10 20
Min: 0.00 Q..: 15
Median: 25
Mean: 22
Std. Dev.: 11
30 40 50. 60. 70. 80
(/)
t/1
X
_j
Z
90. 100.
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEi, SUMMLR I
-------
09 MAY 1990
Population Size (N): 3993 SE(N): 191.5
Lake Area (A): 354924 SE(A): 68996
Sample Size: 1 45
O.f
0.4
0.2-I
0.04
REGION 1
CLUSTERS
1,2, & 3
DIRECT
10. 20. 30. 40. 50. 60. 70. 80.
90. 100.
Min: 0.00 Qj: 15.00 Q£: 25,00 Qg: 30.00 Q4: 40.00 Max: 200.0
Median: 25.00
Mean: 30.35
Std. Dev.: 21.89
o-H
Min: 0.00
Median:
Mean:
Std. Dev.:
40
DIRECT
50. 60. 70. 80. 90. 100.
15.00 Q0: 20.00 «„: 30.00
u O
25.00
27.13
14.77
; 40.00 Max: 200.0
00
CO
h-
<
Q
UJ
CO
Confidence bounds ore for numoer and area of lakes, but hove been scaled.
Q_
156
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: COND11 (/tS/cm) 14 MA^i 199C
REGION 1
Population Size (N): 3993 SE(N): 191.6 Sample Size: 145 CLUSTERS
Lake Area (A): 354924 SE(A); 68995 _LfL& 3
1.0-
0.8-
l
0.4
DIRECT
0 20. 40
Min: 12.40 Q^ 25.30
Median: 34.20
Mean: 42.40
Std. Dev.: 22.86
60.
80.
100
120.
: 31.40 Qg: 40.00 Q4= 59.30 Max: 126.2
G(X)'
1.0
0.6
0.4-
0.0
20
I/)
Min: 12.40 Q : 28.50 Q0: 30.40 Q0: 40.00 Q.: 59.30 Max: 126.2
1234
Median: 35.20
Mean: 40.92
Std. Dev.: 18.82
Confidence bounds are for number and area of lakes, but have been scaled.
Q
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: COND11 OS/cm) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68994
1.0-
0.8
0.6-
Sample Size: 1 45
120.
140.
Median:
Mean:
Std. Dev.:
24.40 Q2: 32.70 Q3: 41.90 Q4: 63.00 Max: 125.6
36.30
43.75
23.64
1.0
0.
0.6-
0.4-
0.2-
o.o
DIRECT
0.
20.
40.
60.
80.
100.
120.
140.
CO
00
Min: 11.20
Median:
Mean:
Std. Dev.:
29.40
39.30
42.94
19.78
Q2: 32.60
43.30
: 63.60 Max: 125.6
LJ
Confidence bounds ore for number and area of lakes, but have been scaled.
Q.
158
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: CONDI! OuS/cm) 09 MAY 1990
Population Size (N): 3993 SE(N): 191,6
Lake Area (A); 354924 SE(A): 68995
1 .On
F(X)-
0.8
0.6-
0.4-
0.2-
0 04
0
Min: 11.40
Median:
Mean:
Std. Dev.:
40.
: 26.00
39.00
45.59
24.67
60.
34.70
Sample Size: 145
80.
100
120
140.
: 43.30 Q.: 63.00 Max: 138,
O *
0 8
0.6-
0.2-
20
40.
60.
DIRECT
80.
100.
120.
140
Min: 11.40 Q1 : 30.40 Q
J.
Median:
Mean:
Std. Dev.:
41.30
44.25
20.88
33.50 Q«: 44.10 Q : 65.70 Max: 138.8
O T
a
LiJ
Confidence bounds ore for number and area of lakes, but hcve been scaled.
X
n.
159
-------
POPULATION DESCRIPTION PHASE I! LAKE SURVEY, SPRING 1986
VARIABLE: DIC02 (mg/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A); 354924 SE(A); 68995
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
1,0
0.8
0 6-
0.4-
0.2-
O.Ot
0,0 0.5 1 0
Min: 0.02 Q : 0.97
Median: 1.76
Mean: 1.94
Std. Dev.: 1.09
1.5
20
1.54
3,0
Q3: 1.98
DIRECT 1
3.5 4.0 4.5 50
.: 3.03 Max: 4.60
G(X)'
1,0
0.8-
0 6-
0.2-
'•$
0 0.5
Min: 0.02
Median:
Mean:
Std. Dev.:
1.0 15
1.07 Q.
t
1.92
2.16
1.16
20 25 3,0 3,5
4,0
1.66 Q«: 2.15 Q.: 3.84 Max: 4.60
«J T;
Confidence bounds ore for numDer and area of lakes, but have been scaled.
00
DO
O
UJ
GO
<
X
0,.,
160
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMED 1'J86
VARIABLE: DIC02 (mg/l) 10 MAY I9'j
Population Size (N); 3993 SE(N). 191.6
Lake Area (A); 354924 SE(A): 68996
Sample Size: 1 45
REGION 1
F(X)'
0.4-
0.0 0 5
Min: 0.04 Q
Median:
Mean:
Std. Dev.:
1.0
0.66
1.56
1.84
1.35
1.5
2.0
Q_: 1.21
&
2.5
Qo
3.0
3.5
4.0
1.93 Q4: 2.75 Max; 7.70
1 .0-
G(X)'
0.0-1
'00
0.5
Min: 0.04 Q..: 0.76 Q : 1.33 Q,:
-I M tJ
Median:
Mean:
Std. Dev.:
1.39
1.82
1.12
1.85 Q4: 3.09 Max: 7.70
Ci
Confidence bounds ore for number and area of lakes, but have been scaled.
0..
161
-------
PUPUl. AIION DESCRIPTION PHASE II LAKE SURVEY, FALL ! 986
VARIABLE: DIC02 (mg/l)
09 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
1.0-r
F(X)-
0.8-
0.6-
0.5 1.0 1
Min: 0.23 Qji 0.83 Q2: 1.42
Median:
Mean:
Std. Dev.:
1.85
2.14
1.44
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
2 0 2.5 3.0 3.5
4.0
DIRECT
4.5 5.0
2.19 Q4: 3.22 Max: 6.18
1 .0-r
O.fr
0.4-
2.5 3.0
3.5
DIRECT
4.0 4.5
5.0
<
Min: 0.23 Q,: 1.04 Q
1'
1.49 Qg: 2.07 Q4: 3.42 Max: 6.18
Median:
Mean:
Std. Dev.:
1.78
2.14
1.28
U.I
Confidence bounds ore for number and area of lakes, but have been scaled.
T.
162
-------
I'Uf ULAHON DESCRIPTION PHASE 11 LAKE SURVEY, SPRING 1986
VARIABLE: DOC1 1 (mg/l) 09 MAY 19!)«,
Somple Size 145 CLUSTERS
0.6-j
0 4-
0.0
DIREC1
0.0 10 20 3,0 4.0 5.0 6.0 7.0 80 9.0 10.0 11.0 1?
Min: 0.20 Qj: 2.40 Q^. 3.07 Qg: 3.75 Q4: 4.72 Max: 10.92
Median:
Mean:
Std. Dev.:
3.27
3.62
1.69
0.6
0 A i
DIRECT
30 40 50 6.0 7.0 8.0 90 100 110 12 u
Min: 0,20 Q : 2.62 Q • 3.13 Q0: 4.02 Q,: 5.93 Max: 10.92
Median:
Mean:
Std, Dev.:
3.45
3.84
1.51
163
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: OOC1 1 (mg/l) 09 MAY 1990
Populot
Lake Ac
1 0-
0.8
0.6
F(X)'
0 4-
0.2-
O.Q~
0
ion Stze (N): 3993 SE(N): 191,6 Sample Size: 145
ea (A), 354924 SE(A): 68996
''"'' /
./
;•' /
/'
.,
--"^f^''
0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0
REGION 1
CLUSTERS
1,2, & 3
DIRECT
10.0 11.0 12.0
Min: 0.30 t^: 2.86 Qg: 3.64 Qg: 4.66 Q : 6.72 Max: 11.92
Med
Mea
Std
1 .0-i
0 8-
0.6-
G(X)'
0.4-
0.2-
°-°o
ian: 4.14
n: 4.65
. Dev.: 2.35
j J
-- -" ^^~r
1
f"~ J
?J /
1 J
JJ
__Jr
0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9,0
DIRECT
10.0 11.0 12.0
Min: 0.30 Q, : 2.76 Q0: 3.49 QQ: 4.66 Q . : 7.03 Max: 11.92
1 t* A 4
Median: 3.74
Mean: 4.37
Std. Dev.: 1.97
00
z:
Confidence bounds ore for number and area of lakes, but have been scaled.
a
CO
<
m
o..
164
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE. DOC1 1 (mg/l) 09 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
I.Oi
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
0.2-
'O.o' "
DIRECT
20
4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0
Min: 0.29 Q : 2.72 Q :
X Ci
Median:
Mean:
Std, Dev.:
3.59 Q3: 4.48 Q4: 5.96 Max: 15.30
3.89
4.47
2.36
0.8-
06
0 4
0.0-
DIRECT
0,0 1.0 2.0 3.0 40 5.0 6.0 7.0 8.0 9.0 100 110 120
00
(7)
Q3: 4.68
6.54 Max: 15.30
Min: 0.29 Q,: 2.72 Q : 3.36
JL &
Median: 3.68
Mean: 4.44
Std. Dev.: 2.16
-oiinds ore for number and urea of iokes, but have been scaled.
<
165
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: FE11 (/Ltg/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.5
Lake Area (A); 354924 SE(A): 68996
Sample Size: 145
REGION 1
CLUSTERS
1,2, & 3
1.0-
0.8-
0.6-
0.2-
0.0
0 50
Min: 0.00 Q
Median:
Mean:
Std. Dev.:
DIRECT
100. 150 200 250.
22.00
44.00
62.20
59.6?
: 35.00 Qg: 56.00
300 350. 400. 450.
Q,: 91.00 Max: 433.0
i-c(xy
0.6-
0.4
0.2-
DIRECT
150 200 250. 300 350 400 450
Min: 0.00 Q,: 22.00 Q : 33.00 Q : 44.00 Q : 55.00 Max: 433.0
3. « v Tt
Median:
Mean:
Std. Dev.:
40.00
43.30
33.88
Confidence bounds are for number and area of lakes, but have been scaled.
oo
h-
Q
CO
a.
166
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: FE11 (^9/0 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68995
Sample Size: 145
REGION 1
CLUSTERS
1,2, & 3
DIRECT
100
150. 200. 250. 300. 350. 400.
450.
Min:
0.00 Q : 11.00
X
Q : 21.00
: 46,00
: 100.0 Max: 895.0
Median:
Mean:
Std. Dev.:
34.00
62.21
101.25
1.0
0.8
0.6-
0.4-
0.2-
DIRECT
50
100. 150. 200. 250. 300. 350. 400.
450.
00
GO
Min: 0.00 Q,: 9.00 Q0: 18.00 Q : 33.00 Q.: 45.00 Max: 895.0
1 i 6 4
Median:
Mean:
Std. Dev.:
25.00
34.69
51.42
Confidence hounds ore for number and area of lakes, but have been scaled.
LiJ
CO
<
X
0...
167
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: FE11 (/-tg/l)
O9 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
1.0-
OS-
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
0.6-
-F(XV
Median:
Mean:
Std. Dev.:
400. 450.
14.00 Q£: 39.00 Qg: 62.00 Q4: 116.0 Max: 556.0
54.00
76.54
92.87
0.6-
0.2-
0.04
DIRECT
50 100 150 200 250. 300. 350. 400. 450.
00
"Z.
Min: 0.00 Q..: 14.00 Q,: 43.00 QQ: 58.00
.i. Ct O
: 84.00 Max: 556.0
Median:
Mean:
Std. Dev.:
43.00
62.23
69.42
Q
Ld
CO
* Confidence bounds ore for number ond area of lakes, but have been scaled.
X
a,
168
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1'-J8f-.
VARIABLE: FTL98 (/^eq/l) 10 MAY 1990
REGION !
Population Size (N): 3993 SE(N): 191.6 Sample Size: 145 CLUSTERS
Lake Area (A): 354924 SE(A): 68996 [ 1.2,_& 3_
1.0-r
1-EfxV
0.8H
0.6-
0.4-
0.2-
0.0-1—
0.0
Min: 0.32
Median:
Mean:
Std. Dev.:
4 0
1,33
2.05
2.67
2.09
DIRECT
6.0
8.0
,: 1.85 Qg: 2.35 Q4: 3.14 Max: 14,74
1.0-
0.6-
0.4-
0.04
2.0
Min: 0.32 Q
Median:
Mean:
Std. Dev.:
r
4 0
1.26
2.35
3.15
2.53
6.0
Q2: 2.10
8.0
10.0
! 4 0
2.79 Q4: 3.83 Max: 14.74
O
Confidence bounds ore- for nurnbet and arpc.i of lokes,
169
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: FTL.98 (jueq/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
1.0-r
Sample Size; 1 45
REGION 1
CLUSTERS
1,2, & 3
0.8-
0.6-
0.4-
0.2-
0.0+—
00
2.0
Min: 0.54
Median:
Mean:
Std. Dev.;
4.0
1.62
2.55
3.16
2.45
DIRECT
6.0
10.0
: 2.26 Q3: 2.90
12.0 14.0 16.0
„: 4.10 Max: 15.97
0.8-
0.6-
0.4-
4.0
6.0
3.0
Min: 0.54 Q :
Median:
Mean:
Std. Dev.:
1.45 Q2: 2.28
2.73
3.75
3.35
10.0
DIRECT
12.0
14,0
160
3.25 Q4: 4.45 Max: 15.97
02
z
t—
o
L.L.I
oro for number and area of lakes, but have been scaled.
j.:
n
170
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE; FTL98 (/./.eq/l)
Population Size (N): 3993 SE(N): 191-6
Lake Area (A): 354924 SE(A): 68997
1.0-r
Sample Size; 1 45
09 MAY I 9'-30
REGION 1 |
CLUSTERS !
1,2, & 5 j
l-F(X)'
0 6-
0 4
0.2-
2.0 -1.0
Min: 0.00 Q : 1.68
Median: 2.55
Mean: 3.43
Std. Dev.: 3.24
6.0
8.0
DIRECT
1 4 0
2.26 Q3: 3.08 Q4: 4.12 Max: 19.93
0.8
0.2-
2 0
4.0
DiRECl
60
8.0
10.0
120 140 16
Min: 0.00 Q,: 2.00 Q_: 3.18
ll
Median:
Mean:
Std. Dev.:
3.28
5.20
5.26
3.86
4.46 Max: 19.93
171
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE; K98 (/xeq/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.5 Sample Size: 145
Lake Ar
1.0-
08-
0.6-
xj-
0.4-
0.2-
0 0-
0
ea (A); 354924 SE(A): 68995
'-X^N
\
\ \
\\
\ '• •
\v
\-
\ — ;•
^ — ^~™^^ ""• _
0 5.0 100 150 200 25.0 30,0 35.0
Min: 2.56 Q, : 6.63 Q0: 8.44 Q0: 10.87 Q.: 15.37
i ^ J 4
Median: 9.21
Mean: 11.77
Std.
1.0-j
0.8-
0 6-
v V
x)
0.4-
0.2-
°-w
Dev.: 7.59
~\Y
V
h
V_^';
\_,z:i:,.:._
" -— -— ~~_ _^,
0 5.0 10.0 15.0 20.0 25.0 30.0 35.0
REGION 1
CLUSTERS
1,2. & 3
DIRECT
40.0 45.0
Max: 42.22
DIRECT
01
i
21
40.0 45.0
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: K98 (jueq/l) 10 MAY 1990
Populat
Lake Ar
1 0-
0.8-
0.6
X\*
;
0.4-
0 ?.•
0.0-
0
on Size (N); 3993 SE(N): 191.5 Sample Size: 145
ea (A): 354924 SE(A): 68995
^\;..
V-,,
I \
\\,
NVN
^^^
REGION 1
CLUSTERS
1,2, & 3
DIRECT
^N* •— — — —~~.~—^____ - - - •
0 5.0 10.0 15.0 20.0 25.0 30.0 35.0
Min: 1.38 Q, : 5.78 Q0: 7.80 Q_: 10.74 CL: 16.38
A £ o 4
Median: 9.15
Mean: 12.06
Std. Dev.: 9.75
40.0 45.0
Max: 53.57
0.6-
0.4
00
Min: 1.38
Median:
Mean:
Std. Dev.:
5.0
10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0
7.08 Q2: 8.13
8.36
10.48
6.48
Q3: 9.15
Q4: 12.50
Max: 53.57
I—
O
LJ
".onfidence bounds ore for number ond area of lakes, but have been scaled.
n.
173
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: K98
Population Size (N): 3993 SE(N): 191.5 Sample Size: 145
Lake Area (A): 354924 SE(A); 68995
I.Oi—
09 MAY 1990
REGION 1
CLUSTERS
1,2, & 3
0.6-
0.4-
0.2-
0.0
DIRECT
0.0
5.0
10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0
Min: 0.90 Q. : 6.39 Q0: 8.26 Q_: 12.22 Q.: 17.59 Max: 57.87
1 i d 4
Median:
Mean:
Std. Dev.:
9.90
12,81
9.27
0.6-
0.4-
5.0 100 15.0 20,0 25.0 30.0
Min: 0.90 Q.: 7.70 Q2: 8.90 Q3: 9.90
DIRECT
35,0 40.0 45.0
,: 13.83 Max: 57.87
Median:
Mean:
Std. Dev.:
8.90
10.95
6.63
Confidence bounds ore for number and area of lakes, but have been scaled.
Q
00
<
X
Q_
174
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE. MG98 (jueq/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
1.0-,—•
0.8-
0,6
Sample Size; 1 45
REGION 1
CLUSTERS
1,2, & 3
125. 150. 175, 200. 225.
Min: 11.93 Q,: 34.39 Q0: 42.16 Q_: 55.28 Q,: 85.14 Max: 225.0
1 2 J 4
Median:
Mean:
Std. Dev.:
49.93
59.49
35.96
0 2-
0.0
125. 150. 175. 200. 225.
Min: 11.93 Q : 32.08 Q9: 44.83
X £t
Median:
Mean:
Std. Dev.:
49.93
59.98
29.96
Q3: 61.86 Q4: 94.68 Max: 225.0
Q
LJ
Confidence bounds ore for number and area of lakes, but have been scaled.
0_
175
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: MG98 (/ueq/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191,6
Lake Areo (A): 354924 SE(A); 68995
Sample Size: 145
REGION 1
CLUSTERS
1,2, & 3
JT
0.6-
0.4-
0.2-
0.0
DIRECT
0. 25. 50. 75. 100 125. 150. 175. 200. 225
Min: 11.76 Q1: 33.48 Q0: 45.57 Q : 56.92 Q : 87.61 Max: 251.3
X £i \J 4
Median:
Mean:
Std. Dev.:
50.01
63,73
40.96
G(x)-
1 O-i
0 6-
0.4-
0.04-
0.
50. 75 100. 125. 150. 175.
DIRECT
"200.
25.
CO
00
Min: 11.76 Q 33.23
Median:
Mean:
Std. Dev.:
55.03
63.27
32.38
,: 46.72 Q3: 63.92
: 94.85 Max: 251.3
LJ
CO
Confidence bounds ore for number and area of lakes, but hove been scaled.
a
176
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE. MG98 (yiteq/l)
09 MAY 1990
Population Size (N): 3993 SE(N). 191.6
Lake Area (A): 354924 SE(A): 68996
Sample Size: 145
REGION 1
CLUSTERS
1,2, & 3
50
75.
100
125.
150.
Min: 11,68 Q,: 38.00 Q0: 50.84 QQ: 63.01 Q.
J, Lt O
175. 200. 225
96.41 Max: 277.9
Median:
Mean:
Std. Dev,:
57.75
69.43
44.18
1 O-i
G(X)-
0.6-
0.4-
0.2-
0.04
25.
125. 150. 175. 200
225.
Min: 11.68 Q : 34.96 Q0: 49.77
i z
Median:
Mean:
Std. Dev.:
57.75
68.71
36.43
: 63.01
: 104.1 Max: 277.9
LJ
LJ
< r\
* Confidence bounds ore for number and area of lakes, but have been scaled.
I
CL
177
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: MN11 (/.iq/l) 10 MAY 1990
I-F(X)'
Sample Size: 1 45
REGION 1
CLUSTERS
DIRECT
Median:
Mean:
Std. Dev.:
19.00
35.09
49.43
150. 200.
250.
300.
350.
400.
= 14.00 Q3: 26.00 Q4«. 50.00 Max: 369.0
0.8'
0 6-
0.4-
o.o
50
Min: 0.00 Q
Median:
Mean:
Std. Dev.:
DIRECT
100
5.00
13.00
21.53
30.13
150
200.
Q0: 8.00 Q
w *
250. 300. 350. 400
23.00 Q,: 28.00 Max: 369.0
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: MN11 (/Ltg/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A). 354924 SE(A): 68996
1 ,
Sample Size: 1 45
REGION 1
CLUSTERS
O.f
0.6-
0.4-
O.O
Min: 0.00
Median:
Mean:
Std. Dev.:
50.
100.
0.00 Q2: 0.00
1.00
17.98
44.91
DIRECT
150.
200.
4.00 Q4: 22.00 Max: 337.0
1 .On
0 8-
0.6
0 0+
Min: 0.00
Median:
Mean:
Std. Dev.:
50
0.00
0.00
7.43
28.60
100
,: 0.00 Q,
150,
DIRECT
200.
0.00 Q : 3.00 Max: 337,0
Confidence bounds ore for number end Q.reo of lakes, but have been scaled
179
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: MM 1 1 (/ug/l)
09 MAY 1990
Population Size (N): 3993 SE(N): 191.5
Lake Area (A): 354924 SE(A); 68996
I , vJ~~¥"- : : : : ———i — ' " •••'--•"-"• •••'•«•
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
0.6-
0.4-
0.2-
0-0
50
DIRECT
100
150
200.
250.
300.
350
400.
Min: 0.00 Q,: 3.00 Q •
Median:
Mean:
Std. Dev.:
8.00 Q3: 18.00 Q4: 43.00 Max: 406.0
12.00
30.99
53.32
1.0-
0.6-
0.4-
0.2-
100.
150.
200.
250.
Min: 0.00 Q : 2.00 Qg: 4.00 Qg: 12.00
Median:
Mean:
Std. Dev.:
8.00
17.96
32.59
DIRECT
300. 350. 400.
,: 23.00 Max: 406.0
* Confidence bounds are for number and area of lakes, but have been scaled.
CO
to
Q
Ld
00
<
X
0.
180
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: NA98 (;ueq/0 14 MAY IS00
"~ REGION 1 i
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
Sample Size: 1 45
CLUSTERS
1,2, & 3
0.6
0.4-
0.04
0
Median:
Mean:
Std. Dev.:
100.
Min: 3.35 Q 33.10
69.56
122.43
123.72
DIRECT
200.
300.
400.
59.33 Qg: 93.39 Q4: 217.5 Max: 590,3
0.8-
0.6-
0,4
Min: 3.35
Median:
Mean:
Std. Dev.:
100
: 32.36 Q«: 56.77
i i
79.78
106.43
102.54
DIRECT
300.
400.
: 93.39 Q4: 113.8 Max: 590.3
Confidence bounds ore for number and area of lakes, but have been scaled.
181
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: NA96 (/ueq/0 10 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
Sample Size: 1 45
1.O
i-F(xy
0.8-
0.6-
0.4-
0.2-
0.0
Median:
Mean:
Std. Dev.:
100.
76.52
133.47
130.25
DIRECT
200.
Min: 3.91 Qi 34.97 Q: 62.20
300.
400,
500.
t 111.6 Q : 230.9 Max: 587.2
1.0-
0.6-
0.4-
0.2-
100.
Min: 3.91 QJ: 34.45
Median:
Mean:
Std. Dev.:
85.91
123.12
110.37
DIRECT
200
67.36
300.
400.
500.
: 112.9 Q4: 150.0 Max: 587.2
* Confidence bounds are for number and area of lakes, but hove been scaled.
in
Q
LjJ
in
<
T.
a,
182
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: NA98 (yueq/l)
09 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68996
Sample Size: 1 45
1.0
0.6-
0.4-
0.2-
0.04-.
100.
200.
Min: 6.83 QJ: 39.15 Q2: 74.04
Median:
Mean:
Std. Dev.:
86.83
144.48
141.37
DIRECT
300.
400.
500
(: 108.5 Q4: 251.7 Max: 780.6
I.O
0.6-
0 4-
0,0-
100
200
DIRECT
300.
400.
500
00
Min: 6.83 Q,: 39.15 Q0: 72.30 Q,: 111.5 Q,: 164.0 Max: 780.6
Median:
Mean:
Std. Dev.:
87.26
128.83
117.71
Q
UJ
Confidence bounds ore for number and area of lakes, but have been scaled.
n:
n
183
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: NH498 (/ueq/0 10 MAY 1990
Population Size (N): 3993 SE(N): 191.5
Lake Area (A): 354924 SE(A): 68997
Sample Size: 145
0.0
00 2.0 4.0
Min: 0.00 Q.: 0.00
Median: 0.22
Mean: 1.19
Std. Dev.: 2.51
10.0
12.0
14,0
0.00 Qg: 0.50 Q4: 1.44 Max: 13.92
4,0
Min: 0.00
Median:
Mean:
Std. Dev.:
0.00
0.00
0.78
2.03
6,0
DIRECT
.O
10.0
12.0
14.0
: 0.00 ^: 0.05 Q4: 1.28 Max: 13.92
tn
<
o
* Confidence bounds are for number and area of lakes, but have been scaled.
31
a.
184
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: NH498 (/ueq/l) 10 MAY 1990
Population Size (N): 3993 SE(N). 191.6
Lake Area (A): 354924 SE(A): 68996
1.&T
0.8-
0.6-
0 4-
0.2-
o a
0.0
Min: 0.00
Median:
Mean:
Std. Dev.:
\_
1 o
0.00
0.00
0.47
1.16
2 0
0.00
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
DIRECT
3.0
4.0
5,0
(: 0.22 Q4: 0.89 Max: 18.02
0,8
0.6
i-c(xy
0.2-
0.0
'o o
Min: 0.00
Median:
Mean:
Std. Dev.:
1 o
0.00
0.00
0.36
0.61
2 0
0.00
DIRECT
3.0 4.0 5.0
0.03 Q,: 1.11 Max: 18.02
Confidence bounds are for number and area of lakes, but have been scaled.
in
Q
LJ
CO
-------
POPULATION DF SCRIP! ION PHASE II LAKE SURVEY, FALL 1986
VARIABLE. NH498 (jUPq/O
Population Size (Nj 3993 SE(N). 191.5
Lake Area (A): 354924 SE(A): 68996
i
i ,\
Sample Size: 1 45
09 MAY 1990
REGION 1
CLUSTERS
1,2, & 3
0 6-
l-F(X)'
Y
0 2-
.
0-0
2.0
4.0
DIREC1
6.0
10.0
12.0
Min: 0.00
Median:
Mean:
Std. Dev.:
0.00
0.72
2.02
3.96
0.44 Q3: 1.05 Q4-' 2.50 Max: 25.70
1 ,0'T
0.8-
06-
i-G(xy
O.OH—
0.0
2.0
4,0
Min: 0.00 Qn: 0.00 Q
DIRECT
1'
6.0
0.28 Q3: 0.55
10.0
12.0
14.0
Q : 1.05 Max: 25.70
Median:
Mean:
Std. Dev.:
0.28
0.85
2.01
* Confidence bounds ore for number ond area of lakes, but have been scaled.
Z.
O
LLJ
00
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: N0398 (/ueq/l) 09 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68997
Sample Size: 1 45
l.O
0.6-
0.4-
0.2-
0.04-
DIRECT
'0.0 5.0 10.0 15.0
Min: 0.00 Q^ 0.01 Q :
Median: 3.15
Mean: 6.14
Std. Dev.: 7.85
20.0 25.0 30.0 35.0 40.0 45.0 50.0
2.30 Q3: 5.22 Q4: 8.78 Max: 47.16
l-G(X)'
0.
0.6-
0.4-
DIRECT
0.04-
0.0 5.0 10.0 15.0
Min: 0.00 Q,: 1.71 Q9:
20.0 25.0 30.0 35.0 40.0 45.0 50.0
4.43 Q : 6.78 Q : 14.61 Max: 47.16
Median:
Mean:
Std. Dev.:
5.81
7.44
7.25
Confidence bounds are for number and area of lakes, but have been scaled.
(/)
CO
a
Ld
(f)
IT
0-
187
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE. NO398 (jweq/l) 09 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68997
Sample Size: 145
Median:
Mean:
Std. Dev.:
30.0
1.10 Max: 22.70
0.46
1.59
3.40
1-G(XV
0.6-
0.4-
0 2-
O.C
3.0
DIRECT
5.0
10.0
15.0
20.0
25.0
30.0
Min: 0.03 Q,: 0.19 Q • 0.33
0.86 Q4: 3.45 Max: 22.70
Median:
Mean:
Std. Dev.:
0.48
2.61
4.61
CO
00
Q
Ld
to
Confidence hounds ore for number cmd area of lakes, but hove been scaled.
a,
188
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: N0398 (/ueq/l) 09 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Loke Area (A). 354924 SE(A): 68996
1.0!
Sample Size: 145
l-F(X)'
DIRECT
0.04
0.0
4.0
8.0
12.0
16.0
20.0
Min: 0,03 Q : 0.24 Q2: 0.59 Qg:
Median:
Mean:
Std. Dev.:
0.73
2.66
4.09
1.49 Q4: 4.15 Max: 18.84
I.O-
0.8-
1-G(X)'
0.4-
0.0-
0.0
Min: 0.03
Median:
Mean:
Std. Dev.:
4.0
0.32
1.74
3.55
4.60
8.0
: 0.70
12.0
,: 1.77
16.0
20.0
8.40 Max: 18.84
(/)
I—
UJ
Confidence bounds are for number and area of lakes, but have been scaled.
189
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: PH02 09 MAY 199O
Population Size (N); 3993 SE(N): 191,6
Lake Area (A): 354924 SE(A); 68997
i.O
Sample Size: 1 45
F(Xj"
.34 Q0: 6.63 Q.: 6.85 Max: 7.39
0.6-
0.4-
0.2-
0.04-
4.0
Min: 4.44 Q :
Median:
Mean:
Std. Dev.:
6.4?
6.32
0.65
G(X)'
1 o
0.4-
0.2-
0.0
.
4.0
4.5
5.0
5.5
DIRECT
6.0
6.5
7.0
7.5
CO
CO
Min: 4.44 Q.: 6.09 Q_: 6.45
V
Median:
Mean:
Std. Dev.:
6.60
6.45
0.4?
6.68
6,77 Max: 7.39
Q
UJ
Confidence bounds are for number and area of lakes, but have been scaled.
190
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: PH02 09 MAY 1990
Population Size (N): 3993 SE(N): 191,6
Lake Area (A): 354924 SE(A): 68995
!,OT
Sample Size; 145
REGION 1
CLUSTERS
1,2, & 3
F(X)-
0.8-
0.6-
0.4-
0.2-
'4.5 5.0 5.5
Min: 4.50 Qt : 5.92 Q_
Median:
Mean:
Std. Dev.:
6,53
6.47
0.74
e.O
6.5
DIRECT
7.5
6.46 QQ: 6.73 Q.: 6.98 Max: 9.08
3 4
G(X)'
1.0
0.8-
0.6
0.
Min: 4.50
Median:
Mean:
Std. Dev.:
6.46
6.73
6.73
0.53
: 6.61 Qg: 6.88 Q4: 7.09 Max: 9.08
CO
_J
21
<
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: PH02
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68995
0.6-
0.4]
0.2-
0,0
09 MAY 1990
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
Min: 4.41 Q.: 6.09
1'
Median:
Mean:
Std. Dev.:
6.69
6.51
0.67
DIRECT
5.5 6.0 6.5 7.0 7.5
6.58 Q«: 6.81 Q.: 7.01 Max: 7.33
G(X)'
1.0
0.81
0.61
0.4-
0.2-
5.0
Min: 4.41 Q : 6,60 Q
Jl
5.5
6.0
6.5
DIRECT
7 0
6.68 Q3: 6.93 Q4: 7.20 Max: 7.33
Median:
Mean:
Std. Dev.:
6.84
6.78
0.46
* Confidence bounds ore for number and area of lakes, but have been scaled.
00
Q
LJ
00
<
IE
a.
192
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING i )8G
VARIABLE: PTL11 O9/0 10 MAY 1990
Populotion Sire (N): 3993 SE(N): 191.5
Lake Area (A): 354924 SE(A): 68995
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
I-F(X)'
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50,0 55.0
Min: 0.00 Q,: 4.00 Q9: 8.00 Q : 10.00 Q.: 13.00 Max: 52.00
L £» -o *t
Median:
Mean:
Std. Dev.:
9.00
10.87
9.71
1.0
0.8
0.6-
0,4-
0.2-
DIRECT
GO
GO
3.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 55.0
Min: 0.00 Q.: 1.00 Q
1'
7.00 Q3: 9.00 Q4: 12.00 Max: 52.00
Median:
Mean:
Std. Dev.:
8.00
9.24
8.36
Q
* Confidence bounds ore for number and area of lakes, but nave oeen
CL
193
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE; PTL11 Og/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.5
Lake Area (A): 354924 SE(A): 68994
Sample Size: 145
00+
0.0 5.0 10.0 15.0
Min: 0.40 Q : 4.90
Median: 8.60
Mean: 12,50
Std. Dev.: 16.59
20.0 250 30.0 35.0 40.0 45.0 50.0 55.0
i 7.20 Q 10.40 Q: 16.90 Max: 137.0
1 0
0.8-
0.6-
0.4-
0.2-
OQ-J-
DIRECT
5.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 55.0
Min: 0.40 Q,: 2.70 Q0: 4.20 Q,: 6.10 Q : 10.00 Max: 137.0
1 Z, o t
Median:
Mean:
Std. Dev.:
6.00
7.29
9.48
* Confidence bounds are for number and area of lakes, but have been scaled.
CO
a
LJ
00
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: PTL11 (/ig/l) 09 MAY 1990
Population Size (N): 3993 SE(N): 191,5
Lake Area (A): 354924 SE(A): 68995
Sample Size; 1 45
l.O
0.8-
0.6-
04
0.2-
0.0
DIREC1
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 500 55.0
Min: 1.00 Q^ 4.90 Q2: 6.90 Qg: 10.30 Q4: 15.50 Max: 54.30
Median: 8.00
Mean: 10.99
Std. Dev.: 8.73
i.o
0.6-
0 4-
0.2-
DIREC.T
50 100 150 20.0 25.0 30.0 35.0 40.0 45.0 50.0 55,0
Min: 1.00 QJ: 4.20 ^ 5.30
: 6.20 Q: 8.50 Max: 54.30
Median:
Mean:
Std. Dev.:
5.40
6.9?
5.09
t/)
h-
O
LtJ
Confidence bounds ore for number and area of lakes, but have been scaled
n
195
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: SECME98 (m) 10 MAY 199O
Population Size (N): 3735 SE(N); 197.3
Lake Area (A): 344194 SE(A): 69029
1.O
0.8-
Sample Size: 1 32
REGION 1
CLUSTERS
1,2, & 3
F(X)'
Q : 3.70 Q.: 4.70 Max: 8.85
Median:
Mean:
Std. Dev.:
10.0
3.30
3.59
1.57
NOTE: Secchi Depth data is missing for
13 lakes in the spring data set. Thus,
these two CDFs may not completely
represent the population.
GCXV
CO
6.0 7.0 8.0 9.0
10.0
Min: 1.10 Q,: 3.00 Q£: 3,90 Qg:
Median:
Mean:
Std. Dev.:
4.00
4.29
1.61
4.45 Q.: 4.60 Max: 8.85
4
Q
LJ
Confidence bounds ore for number and area of lakes, but hove been scaled.
I
Q.
196
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: SECME98 (m) 10 MAY 1990
Population Size (N): 3993 SE(N): 191.5 Sample Size: 145
Lake Ar
1.0
0.8-
0 6
0.4-
0.2-
0.0-
0
ea (A): 354924 SE(A): 68995
r;:^:^^
Jf
/f
^
•Y
REGION 1 1
CLUSTERS
1,2, & 3
DIRECT
.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 100
Min: 0.55 Q : 1.65 Q : 2,55 Q : 3,60 Q , ; 5.30 Max: 12.60
-*•«** 4
Median: 3.15
Mean: 3.54
Std.
10
0.8-
0.6-
0 4-
0 2-
Q.OJ
Dev.: 2.08
- J J
I ^
; ;"W
i ^
r'
,
^j--:-
0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
DIRECT
in
^_
ij~\
i
<
9.0 10.0 ff
Min: 0.55 Q, : 2.75 Q»: 3.80 Q_: 4.60 Q.: 5.90 Max: 12.60 <
1234 O
Median: 4.60
Mean: 4.46
Std. Dev.: 1.88
„
LLJ
Cr:nfidence bounds orr- for number and area of lakes, but have been scaled.
n.
197
-------
POPULATION DESCRIPTION PHASE !! LAKE SURVEY, FALL 1986
VARIABLE: SECME98 (m)
09 MAY 1990
Population Size (N): 3993 SE(N): 191.5
Lake Area (A): 354924 SE(A): 68995
1.0
Sample Size: 145
7.0 8.0 9.0 10.0
Min: 0.55 Q1: 1.80 Qr
Median;
Mean:
Std. Dev.:
2.55 Q3: 3.20 Q4: 4.65 Max: 10.90
3.00
3.25
1.72
G(X)-
V
10.0
3.35 Q4: 5.00 Max: 10.90
Median:
Mean:
Std. Dev.:
3.00
3.56
1.73
if}
z
a
LJ
Confidence bounds are for number and oreo of lakes, but have been scaled.
X
0.
198
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: SI0211 (rng/l) 10 MAY 1990
Populot
Lake Ar
1.O
0.8
0.6
l-F(X)'
0.4
0.2-
°-°o
ion Size (N): 3993 SE(N): 191.6 Sample Size: 145
eo (A): 354924 SE(A): 68997
^ ^ ^
^^\?X
\^ ^
^O\
^^--
*~x v
.0 1.0 2.0 3.0 4.0 5.0 6.0
REGION 1
CLUSTERS
1,2, & 3
DIRECT
7.0 8
]
0
Min: 0.00 Q, : 1.84 Q,: 2.84 Q0: 3.73 Q.: 4.78 Max: 7.13
1 6 O 4
Med]
Meai
Std
'•°1
0 8-
0.6-
l-G(X)'
0.4-
0.2-
o.ol
tan: 3.18
i: 3.22
Dev.: 1.69
*^_
\ "" --..,
^T^^^
~o
^-L
0 1.0 20 3.0 4.0 5.0 6.0
Min: 0.00 Q, : 2.34 Q_: 3.05 Q0: 4.36 Q,: 5.66
1234
Median: 3.46
Mean: 3.81
Std. Dev.: 1.58
DIRECT
7.0 8
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE. SIO211 (mg/l) 10 MAY 1990
Populotion Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68997
I.OT
0.8-
0.6-
0.4-
0.2-
Sample Size: 145
REGION 1
CLUSTERS
1,2, & 3
0.0
Min: 0.03
Median:
Mean:
Std. Dev.:
DIRECT
1.0 2.0
,: 0.66 Q : 1.40
X Z
1.77
1.98
1.38
3.0
.o
s.o
e.o
2.12 Q4: 3.06 Max: 5.81
1.0
0.8-
0.4-
0.2-
o
2.0
DIRECT
3.0
40
5.0
6.0
m
Z.
Min: 0.03 Q : 1.20 Q
Jl
Median:
Mean:
Std. Dev.:
2.01
2.54
1.62
1.59 Q • 2.35 Q . : 4.23 Max: 5.81
o ~f
bJ
« Confidence bounds ore for number and area of lakes, but hove been scaled.
0...
200
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: S10211 (mg/l) 09 MAY 1990
Population Size (N): 3993 SE(N): 191.6 Sample Size: 145
Lake Area (A): 354924 SE(A): 68997
1.0
REGION 1
CLUSTERS
1,2, & 3
0.6-
0.4-
0.2-
0.0
0.0 1.0
Min: 0.01 Q
Median:
Mean:
Std. Dev.:
DIRECT
Qn: 1.72 Qn: 2.79 Q.: 4.40
2.31
2.59
1.85
1.O
0.8-
0.6-
0.4-
0.2-
O.Q-f-
'.0 1.0
Min: 0.01 Qj
Median:
Mean:
Std. Dev.:
3.0
4.0
5.0
DIRECT
6.0
7.0
1.36 Q_: 2.50 QQ: 3.40 Q.: 6.04 Max:
fj O *
2.52
3.28
2.03
8.0
7.25
00
00
•x
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE; S0498 (,ueq/l) 09 MAY 1990
Population Size (N): 3993 SE(N):
191 .5 Somple Size: 1 45
Lake Area (A): 354924 SE(A): 68997
F(X)-
i n
1 .U
0 8
0.6-
0.4-
0.2-
0 OT
c
Min
~^""~-vA
H. "-x..
\ \
^^X'-
j \
x\
\ •
\
). 50. 100.
: 25.86 Q^ 77.74 Q :
Median: 110.55
Mean: 115.61
Std. Dev.: 48.34
,,
0.6-
0.4-
0.2-
o.oj
~\ \
"V.
\
50. 100.
\
^
^^^
REGION 1
CLUSTERS
1,2, & 3
DIRECT
iiii
150. 200. 250 300.
97,85 Q3: 119,4 Q4: 141.6 Max: 289.1
"S
\
~ —^ -IZ^;
DIRECT
150. 200. 250. 300.
Min: 25.86 Q : 82.43 Q : 92.25 QQ: 110.4 Q,: 126.7 Max: 289.1
1 Z ,1 4
Median: 99.44
Mean: 107.02
Std. Dev.: 36.20
CO
o
Confidonre bounds ore- for number and area of lakes, but have been scaled.
202
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: S0498 Cueq/l) 09 MAY 1990
Population Size (N): 3993 SE(N): 191.5
Lake Area (A); 354924 SE(A): 68996
1 .Or—==—' ; ~
0.6-
0.4-
0.2-
00
Sample Size: 1 45
30 60. 90.
Min: 30.81 Q • 74.78
Median:
Mean:
Std. Dev.:
105.35
107.32
43.77
REGION 1
CLUSTERS
1,2, & 3
DIRECT
120 150 180.
,: 93.06 Q • 111.4
210 240. 270
,: 130.2 Max: 331.7
60.
Median:
Mean:
Std. Dev.:
90.
120
150.
180.
210.
240.
270.
Min: 30.81 Q^. 74.78 Q£: 93.06 Q3: 110.8 Q4: 127.5 Max: 331.7
96.81
105.64
38.18
i i
Confidence bounds ore for number and area of lakes, but have been sca'ed.
203
-------
POPULATION DESCRIPTION PHASE 11 LAKE SURVEY, FALL 1986
VARIABLE; SO498 (/ueq/l)
09 MAY 1990
Population Size (N): 3993 SE(N): 191.5
Lake Area (A); 354924 SE(A): 68996
Sample Size: 145
0,0
Min: 28.94
Median:
Mean:
Std. DeY.:
50.
100
150.
200,
250.
300.
L: 73.39 Q2: 90.25
104.02
110.52
49.49
112.0 Q4: 138.0 Max: 327.9
1.0-
l-G(X)-
0.6-
0.4-
°-°o.
Min: 28.94
Median:
Mean:
Std. Dev.:
50.
DIRECT
; 73.12
97.44
105.02
41.88
100.
Q2: 88.38
150.
200.
250.
300.
Qg: 107.8
130.8 Max: 327.9
00
<
bJ
* Confidence bounds ore for number and area of lakes, but have been scaled.
1C
CL
204
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SPRING 1986
VARIABLE: SOBC98 Oeq/l) 14 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A); 354924 SE(A): 68995
•t ry __ : m vv;
Sample Size: 1 45
F(X)'
0.8
0.6-
0.4
0.2-
0.0
DIRECT
200.
Min; 52,23 : 193.1
400.
Q2: 267,7
600.
800.
Median:
Mean:
Std. Dev,:
300,97
356.15
199.80
1000.
,: 350.8 Q.: 472.0 Max: 1101
G(xy
0.8
0.6
0.4-
0.2-
0.0
0.
200.
Min: 52.23 QJ: 234.3 Qg: 256.4
DIRECT
600. 80O WOO. 1200.
409.1 Q4: 477.7 Max: 1101
Median:
Mean:
Std. Dev.:
304.24
356,49
165.56
O
Ui
* Confidence bounds ore for number ond area of lakes, but hove been scaled.
205
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, SUMMER 1986
VARIABLE: SOBC98 (/xeq/l) 10 MAY 1990
Population Size (N): 3993 SE(N): 191,6
Lake Area (A): ,354924 SE(A): 68995
Sample Size; 145
0.6-
0.4-
0.2-
O.O
0.
200.
REGION 1
CLUSTERS
1,2, & 3
DIRECT
400.
600.
800.
1000,
1200.
Min: 54.80 Q : 207.3 Q : 302.6 Q
Jl £*
374.3 Q.: 547.8 Max: 1136
Median:
Mean;
Std. Dev.:
339.77
385.01
218.06
1 0-,—
G(X)'
0.6-
0.4-
0.2-
0.0
o.
Min: 54.80
Median:
Mean:
Std. Dev.:
DIRECT
200.
400.
600.
800.
1000.
: 264.5 Q : 299.7
Ci
345.62
394.39
183.52
: 434.0 Q : 555.3 Max:
1200.
1136
o
Ld
CO
Confidence bounds ore for number and area of lakes, but have been scaled.
I
11
206
-------
POPULATION DESCRIPTION PHASE II LAKE SURVEY, FALL 1986
VARIABLE: SOBC98 (/ueq/l)
09 MAY 1990
Population Size (N): 3993 SE(N): 191.6
Lake Area (A): 354924 SE(A): 68995
10
Sample Size: 1 45
REGION 1
CLUSTERS
1,2, & 3
F(X)*
0.2-
0.0
Min: 61.72
Median:
Mean:
Std. Dev.:
200.
3^ 221.8
370.31
417.41
234.30
400.
600.
800.
1000.
: 340.8 Qg: 411.3 Q4: 573.0 Max:
1200.
1230
1.0
0.6-
0.4-
0.0-
0.
Min: 61.72
Median:
Mean:
Std. Dev.:
DIRECT
200.
400,
600.
800.
1000.
L: 280.2 Q2: 323.6 Qg: 443.1 Q4: 573.0 Max:
383.10
421.07
198.89
1200.
1230
Confidence bounds ore for number and area of lakes, but hove been scaled.
(J
h-
<
C
LJ
GO
-------
APPENDIX B
LISTING OF LAKES SAMPLED IN ELS-II
Table B-1 presents a list of the 145 target lakes sampled during Phase II of the Eastern Lake
Survey (ELS-II). Lakes are ordered by state location and sorted by Lake ID within each state. A more
complete description of the sample lakes (latitude, longitude, physical characteristics, etc.) can be found
in volumes II and III of the Eastern Lake Survey - Phase I data report (Overton et al., 1986; Kanciruk et
al., 1986).
208
-------
Table B-1. Target lakes sampled in Phase-ll of the Eastern Lake Survey.
State
CT
CT
CT
CT
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
Lake ID
1B3-056
1D3-025
1 D3-029
1D3-033
1C1-068
1C1-070
1C2-050
1C2-054
1D1-014
1D1-Q31
1D1-034
1D1-037
1D1-046
1D1-054
1D1-056
1D1-068
1D2-025
1 D2-027
1D2-036
1D2-074
1 D2-084
1 D2-093
1 D2-094
1D3-002
1D3-003
1C1-017
1C1-018
1C1-021
1C1-031
1 C2-002
1C2-012
1C2-016
1C2-056
1C2-064
1C2-068
1C3-032
1E1-009
1E1-010
1E1-011
1E1-025
1E1-040
1E1-050
Lake Name
RIGA LAKE
LONG POND
KILLINGLY POND
(NO NAME)
LINCOLN POND
PACKARD POND
MOORES POND
LAKE WAMPONOAG
HAMILTON RESERVOIR
KINGS POND
ROCKY POND
EZEKIEL POND
ROBBINS POND
UPPER MILLPOND
LITTLE WEST POND
LITTLE SANDY POND
LITTLE QUITTACAS POND
SANDY POND
MICAH POND
STETSON POND
GOOSE POND
ASHLAND RESERVOIR
SNOWS POND
DYKES POND
SANDY POND
WELHERN POND
DECKER PONDS (EASTERN)
CLEAR POND
HUNT POND
IRON POND
BLACK POND
TRAFTON POND
DRURY POND
HANCOCK POND
QUIMBY POND
BEAR POND
PEEP LAKE
SIX PONDS
FOURTH DAVIS POND
BEAN PONDS (MIDDLE)
LT. GREENWOOD POND (WEST)
LOWER OXBROOK LAKE
ELS-II
Cluster
1
3
1
3
1
2
2
1
3
1
1
1
1
2
1
1
2
1
1
2
3
3
1
1
3
3
3
3
2
2
2
3
3
2
3
3
1
3
2
3
2
2
Modif.
ELS-I
Weight
27.209
19.426
19.426
19.426
7.822
7.822
10.743
10.743
6.572
6.572
6.572
6.572
6.572
6.572
6.572
6.572
6.905
6.905
6.905
6.905
6.905
6.905
6.905
19.426
19.426
7.822
7.822
7.822
7.822
10.743
10.743
10.743
10.743
10.743
10.743
8.953
8.070
8.070
8.070
8.070
8.070
8.070
ELS-I!
Cond.
Weight
1 .0000
2.4954
1.0000
2.4954
1.5706
2.8181
2.0518
1.1436
7.5402
1 .8694
1 .8694
1.8694
1.8694
3.3541
1.8694
1 .8694
3.1342
1 .7468
1.7468
3.1342
7.0459
7.0459
1 .7468
1 .0000
2.4954
6.3352
6.3352
6.3352
2.8181
2.0518
2.0518
4.6127
4.6127
2.0518
4.6127
5.5349
1 .5386
6.2059
2.7605
6.2059
2.7605
2.7605
ELS-I!
Sample
Weight
27.209
48.477
19.426
48.477
12.285
22.043
22.043
12.285
49.554
12.285
12.285
12.285
12.285
22.043
12.285
12.285
21.642
12.062
12.062
21.642
48.652
48.652
12.062
19.426
48.477
49.554
49.554
49.554
22.043
22.043
22.043
49.554
49.554
22.043
49.554
49,554
12.416
50.082
22.278
50.082
22.278
22.278
(Continued)
209
-------
Table B-1. Target lakes sampled in Phase-ll of the Eastern Lake Survey (Continued).
State
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
ME
.ME
ME
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
Lake ID
1E1-054
1E1-061
1E1-062
1E1-073
1E1-074
1E1-077
1E1-082
1E1-092
1E1-096
1E1-106
1E1-111
1E1-120
1E1-123
1E2-007
1E2-016
1E2-018
1 E2-030
1 E2-038
1 E2-G49
1 E2-054
1 E2-056
1 E2-063
1 E3-022
1 E3-041
1 E3-042
1 E3-045
1E3-055
1 E3-060
1C1-009
1 C1 -039
1C1-050
1C1-066
1C1-084
1 C1 -086
1 C2-024
1C2-028
1C2-035
1C2-037
1C2-041
1C2-057
1C2-062
1C2-066
1C3-055
1C3-063
Lake Name
DUCK LAKE
LITTLE SEAVEY LAKE
LONG POND
GEORGES POND
CRAIG POND
PARKER POND
STEVENS POND
GREAT POND
MIDDLE CHAIN LAKE
GREENWOOD POND
LONG POND
(NO NAME)
FIRST POND
FAIRBANKS POND
ROUND POND
WEBSTER LAKE
ROUND LAKE
NELSON POND
GROSS POND
BRETTUNS POND
PEABODY POND
KALERS POND
NUMBER NINE LAKE
ROUND POND
SAND POND
MCCLURE POND
TOGUE POND
MILLINOCKET LAKE
UPPER BAKER POND
OSSIPEE LAKE
BILLINGS POND
HAUNTED LAKE
UPPER BEECH POND
STAR LAKE
LAKE WAUKEWAN
SUNSET LAKE
SMITH POND
MENDUMS POND
JUGGERNAUT POND
BABBIDGE RESERVOIR
PEMIGEWASSET LAKE
TURTLE POND
DARRAH POND
MARTiN MEADOW POND
ELS-II
Cluster
2
2
2
2
2
2
2
2
2
2
1
2
2
2
3
3
3
1
1
3
2
2
3
3
3
3
3
3
3
2
2
2
2
2
3
2
2
1
1
2
2
2
1
3
Modif.
ELS-I
Weight
8.070
8.070
8.070
8.070
8.070
8.070
8.070
8.070
8.070
8.070
8.070
8.070
8.070
8.344
8.344
8.344
8.344
8.344
8.344
8.344
8.344
8.344
10.333
10.333
10.333
10.333
10.333
10.333
7.822
7.822
7.822
7.822
7.822
7.822
10.743
10.743
10.743
10.743
10.743
10.743
10.743
10.743
8.953
8.953
ELS-II
Cond.
Weight
2.7605
2.7605
2.7605
2.7605
2.7605
2.7605
2.7605
2.7605
2.7605
2.7605
1.5386
2.7605
2.7605
2.5921
5.8271
5.8271
5.8271
1 .4447
1 .4447
5.8271
2.5921
2.5921
4.7957
4.7957
4.7957
4,7957
4.7957
4.7957
6.3352
2.8181
2.8181
2.8181
2.8181
2.8181
4.6127
2.0518
2.0518
1.1436
1.1436
2.0518
2.0518
2.0518
1 .3722
5.5349
ELS-II
Sample
Weight
22.278
22.278
22.278
22.278
22.278
22.278
22.278
22.278
22.278
22.278
12.416
22.278
22.278
21.628
48.621
48.621
48.621
12.054
12.054
48.621
21.628
21.628
49.554
49.554
49.554
49.554
49.554
49.554
49.554
22.043
22.043
22.043
22.043
22.043
49.554
22.043
22,043
12.285
12.285
22.043
22.043
22.043
12.285
49.554
(Continued)
210
-------
Table B-1. Target lakes sampled in Phase-ll of the Eastern Lake Survey (Continued).
State
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
NY
Lake ID
1A1-003
1A1-008
1A1-012
1A1-015
1A1-017
1A1-028
1A1-029
1A1-033
1A1-039
1A1-044
1A1-049
1A1-057
1A1-060
1A1-061
1A1-064
1A1-066
1A1-070
1A1-073
1A2-002
1A2-004
1A2-006
1A2-041
1A2-042
1A2-045
1A2-048
1A2-052
1A2-054
1A3-001
1A3-028
1A3-040
1A3-042
1A3-043
1A3-046
1A3-048
1A3-063
1A3-065
1 B3-032
1 B3-052
1 B3-059
1 C2-048
1 D3-044
Lake Name
HAWK POND
CEDAR RIVER FLOW
WHITNEY LAKE
HENDERSON LAKE
CONSTABLE POND
DRY CHANNEL POND
MIDDLE POND
KIWASSA LAKE
JOHN POND
LONG LAKE
MIDDLE SOUTH POND
HITCHCOCK LAKE
SEVENTH LAKE (FULTON CHAIN)
WOLF LAKE
MT. ARAB LAKE
WOODHULL LAKE
PARADOX LAKE
GULL LAKES (SOUTH)
ST. JOHN LAKE
DUCK LAKE
LAKE FRANCES
MUD LAKE
NORTH BRANCH LAKE
WOODS LAKE
(NO NAME)
CHUB LAKE
TROUT LAKE
NATE POND
CURTIS LAKE
ZACK POND
CHENEY POND
UNKNOWN POND
LONG POND
GRASS POND
(NO NAME)
SOUTH LAKE (EAST BRANCH)
WIXON POND
(NO NAME)
ISLAND POND
CRANBERRY POND
MIDDLE FARMS POND
ELS-II
Cluster
1
2
1
1
1
1
3
3
1
2
1
1
3
1
2
1
3
1
1
1
2
2
1
1
1
1
1
2
1
2
2
3
2
1
3
1
3
1
1
2
2
Modif.
ELS-I
Weight
9.633
9.633
9.633
9.633
9.633
9.633
9.633
9.633
9.633
9.633
9.633
9.633
9.633
9.633
9.633
9.633
9.633
9.633
8.338
8.338
8.338
8.338
8.338
8.338
8.338
8.338
8.338
6.719
6.719
6.719
6.719
6.719
6.719
6.719
6.719
6.719
27.209
27.209
27.209
10.743
19.426
ELS-II
Cond.
Weight
1 .2754
2.2883
1.2754
1.2754
1 .2754
1 .2754
5.1442
5.1442
1.2754
2.2883
1.2754
1.2754
5.1442
1.2754
2.2883
1.2754
5.1442
1.2754
1 .4734
1.4734
2.6437
2.6437
1.4734
1.4734
1.4734
1 .4734
1.4734
3.2807
1.8285
3.2807
3.2807
7.3752
3.2807
1 .8285
7.3752
1 .8285
1.8212
1.0000
1.0000
2.0518
1.1100
ELS-II
Sample
Weight
12.285
22.043
12.285
12.285
12.285
12.285
49.554
49.554
12.285
22.043
12.285
12.285
49.554
12.285
22.043
12.285
49.554
12.285
12.285
12.285
22.043
22.043
12.285
12.285
12.285
12.285
12.285
22.043
12.285
22.0-13
22.043
49.554
22.043
12.285
49.554
12.285
49.554
27.209
27.209
22.043
21.563
(Continued)
211
-------
Table B-1. Target lakes sampled in Phase-ll of the Eastern Lake Survey (Continued).
State
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
RI
RI
RI
RI
Lake ID
1B1-010
1B1-023
1B1-029
1B1-043
1B1-055
1B1-064
1B2-028
1B3-012
1B3-019
1B3-041
1B3-043
1B3-053
1 B3-060
1 B3-062
1D1-027
1D1-067
1D2-049
1 D3-026
ELS-ll
Lake Name Cluster
GANOGA LAKE
TWIN LAKES (BRINK POND)
NO NAME(WILSON CREEK DAM)
PENN LAKE
ROCK HILL POND
MILLPOND NO. 1
MILL CREEK RESERVOIR
LITTLE BUTLER LAKE
HARTLEY POND
EAST STROUDSBURG RESERVOIR
TROUT LAKE
NO NAME (SNOWFLAKE LAKE)
SLY LAKE
BASSETT POND
SCHOOL HOUSE POND
ROUND POND
SPRING GROVE POND
ARNOLD MILLS RESERVOIR
1
2
3
1
2
2
1
3
3
2
3
3
3
3
2
1
3
3
Mod if.
ELS-I
Weight
3.192
3.192
3.192
3.192
3.192
3.192
1.477
27.209
27.209
27.209
27.209
27.209
27.209
27.209
6.572
6.572
6.905
19.426
ELS-ll
Cond.
Weight
3.8488
6.9057
15.5245
3.8488
6.9057
6.9057
8.4786
1.8212
1.8212
1.0000
1.8212
1.8212
1.8212
1.8212
3.3541
1.8694
7.0459
2.4954
ELS-ll
Sample
Weight
12.285
22.043
49.554
12.285
22.043
22.043
12.523
49.554
49.554
27.209
49.554
49.554
49.554
49.554
22.043
12.285
48.652
48.477
212
-------
APPENDIX C
CHI-SQUARE ANALYSES OF ELS-II POPULATION DIFFERENCES
One way to test whether two populations are different (e.g., spring 1986 ANC versus fall 1986
ANC) is to do a chi-square test as shown in the top box in Figure C-1. In this one degree of freedom
sign test, the number of samples above the 1:1 line is compared to the number of samples below the
1:1 line to see if the observed differences are significantly different than random chance. If the ehi-
square value for this sign test is > 3.84, we are 95% confident that these two populations are different
Chi-square values for the sign test for the fall 1984/fall 1986 comparison are summarized in Table C-1;
values for the spring 1986/fall 1986 comparisons are summarized in Table C-2. Another question
evolving from these comparisons is whether or not the population differences are independent across
the range of concentrations in the cluster. To address this question, the population is divided in half
(using the cluster median of the pair sums) or in quarters (using cluster quartiles of the pair sums) along
the 1:1 line to construct a 2x2 and a 2x4 contigency table as depicted in the lower two boxes in Figure
C-1. If the sign test indicates that there is a significant difference between the two populations (p -
0.05), then a line is drawn parallel to the 1:1 line offset by the median of the pair differences to divide the
population In half (see Figure C-1). If the populations are not significantly different, then no offset line is
used. If the chi-square test on these contigency tables indicates a significant difference from random
chance, we can conclude that the population differences are not independent of concentration. The
independence tests associated with each sign test are summarized in Tables C-1 and C-2.
Points that fall on the 1:1 line, the median value, or the quartile values were excluded from the
analysis. These analyses treat each lake as a unit, Ignoring the ELS-II weighting factor. Thus, the chi-
square tests were done separately for each cluster, because ELS-II weights are nearly equal within each
cluster. A more complete discussion of the Chi-square analyses for ELS-II data is given in Overt on
(1987, 1989).
213
-------
0 20 40 60 80 100 120 140 160
ANC CLUSTER II
FALL 1986 VS. FALL 1984
THE TOTAL NUMBER OF POINTS PLOTTED: S3
NUMBER OF POINTS ON DIAGONAL LINE: 0
CHI-SQUARE- 15,86792
1 OF CHI-SQUARE CRITICAL VALUE - 3.84
NUMBER OF POINTS ON SHIFTED LINE: 1
ADOTTIONAL POINTS WfTH SUM EQUAL TO MEDIAN: 1
CHI-SQUARE FOR INDEPENDENCE - 5.684118
1 Of CHI-SQUARE CRITICAL VALUE - 3.84
0 20 40 60 80 100 120 140 180
ADDITIONAL POINTS WITH SUM EQUAL TO A QUAHTILE: 0
CHI-SQUARE FOR ^DEPENDENCE . 12.52367
3 OF CHI-SQUARE CRITICAL VALUE . 7B1473
0 20 40 60 80 100 120 140 160
Dept. of Statistics, Oregon State University
January 30,1989
Figure C-1. Example of the chi-square analysis for fall 1986/fall 1984 population differences for
ANC in cluster 2. (Taken from Overton, 1989.)
214
-------
Table C-1. Chi-Square Analysis of Between-Year (Fall 1984 vs. Fall 1986) Differences in ELS-II
Population Distributions
Variable Cluster
ANC I
II
III
pH I
II
III
so42- i
II
III
N03- I
II
111
Base Cations I
II
III
DOC I
II
III
AIM!BK '
Sign8
Test
0.75
15.9**
26.3**
0.19
8.32**
3.93
1.33
9.98**
3.27
8.33**
15.1**
23.3**
27.0**
34.9**
29.5**
0.08
0.47
0.09
2.08
Chi-Square Values
Independence
2x2
0.09
5.68*
3.27
0.52
0.08
1.46
3.09
0.96
3.27
5.33*
6.24*
13.1**
8.33**
12.3**
1.46
0.75
0.31
0.09
0.09
Testb
2x4
4.32
12.5**
6.91
4.77
8.68*
10.5*
4.81
3.03
3.14
8.00*
11.3**
17.8**
10.0*
16.3**
3.27
1.59
4.72
6.47
0.26
Sign test is a 1-df chi-square test indicating if the populations are different.
Independence test checks if the population distribution differences are independent of concentration (2x2 test has 1 df; 2x4
test has 3 df),
k*Significant at p < 0.01 (X2 > 6.53 when df = 1; X? > 11.3 when df = 3).
* Significant at p < 0.05 (X2 > 3.84 when df = 1; X2 > 7.81 when df = 3).
215
-------
Table C-2. Chi-Square Analysis of Spring 1986/FalI 1986 Differences in ELS-II Population
Distributions
Variable Cluster
ANC I
II
HI
pH I
II
III
S042' I
II
III
N03- I
II
III
Base Cations I
II
III
DOC I
II
111
ALim I
Sign3
Test
8.33**
15.9**
44.0**
3.00
28.70**
19.56**
1.33
0.08
4.45*
27.0**
2.28
9.09**
3.00
45.3**
40.1**
3.60
13.8**
9.09**
4.08*
Chi-Square Values
Independence
2x2
0.33
1.60
3.27
0.36
2.00
0.38
0.34
0.17
1.53
33.3**
3.91*
17.8**
5.69*
8.67**
1.46
2.65
7.07**
1.46
16.3**
Test5
2x4
2.00
4.91
6.91
111
5.35
1.31
0.69
5.57
2.51
34.0**
22.6**
21.5**
6.04
11.1**
4.73
3.42
9.75*
3.27
19.3**
Sign test is a 1-df ohi-square test indicating if the populations are different.
Independence test checks if the population distribution differences are independent of concentration (2x2 test has 1 df; 2x4
test has 3 df).
**Significant at p < 0.01 (X2 > 6.53 when df = 1; X2 > 11.3 when df = 3).
* Significant at p < 0.05 (X2 > 3.84 when df = 1; X2 > 7.81 when df = 3).
*U.S, GOVERNMENT PRINTING OFFICE: 1992- 6 "• B - 00 'if, 0671
216
-------
-------
-------
United States Center for Environmental Research
Environmental Protection Information BULK RATE
Agency Cincinnati OH 45268-1072 POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use, S300
EPA/600/3-91/012
------- |