138
801
THS

 

LIBRARY
\ Michigan State
0 r U 5’ University

:,303;;

This is to certify that the
thesis entitled

DEFINING LAKE LANDSCAPE POSITION: RELATIONSHIPS TO
HYDROLOGIC CONNECTIVITY AND LANDSCAPE FEATURES

presented by
Sherry L. Martin

has been accepted towards fulfillment
of the requirements for the

MS. degree in Fisheries and Wildlife

2% £3~
Major Professor’s Signature
/Q'[¥’0¥

Date

 

MSU is an Affirmative Action/Equal Opportunity Institution

PLACE IN RETURN Box to remove this checkout from your record.
10 AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 c:/ClRC/DateDue.ind¢p.15

 

DEFINING LAKE LANDSCAPE POSITION: RELATIONSHIPS TO HYDROLOGIC
CONNECTIVITY AND LANDSCAPE FEATURES

BY

Sherry L. Martin

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Fisheries and Wildlife

2004

ABSTRACT

DEFINING LAKE LANDSCAPE POSITION: RELATIONSHIPS TO HYDROLOGIC
CONNECTIVITY AND LANDSCAPE FEATURES.

By
Sherry Martin

The concept of landscape position provides a general framework for investigating
spatial patterns among lakes in a region and can be defined as the position of a lake
relative to features of the landscape. Although several studies have found that landscape
position is related to many in-lake variables, it is unclear what underlying mechanisms
are driving the relationships. To examine this issue, I measured landscape position
emphasizing four different aspects of hydrologic connectivity, to ask two questions: (1)
Which landscape position metric is most strongly related to lake water chemistry/clarity?
And (2) what other features are related to landscape position? I found the metric that
measured stream connections was most strongly related to in-lake variables and,
therefore, the best way to measure landscape position. This result suggests that stream
inputs, as opposed to upstream lake inputs, likely drive patterns associated with lake
landscape position. I also found that landscape position was related to lake morphometry
and the amount of surrounding wetlands, suggesting additional mechanisms for why
landscape position explains variability in lake variables. These results suggest the need to
develop measures of landscape position that incorporate multiple landscape features.
Thus, landscape position is a composite variable, strongly correlated to a variety of in-

lake variables due to its relationship to these major landscape features.

ACKNOWLEDGEMENTS
My research was financially supported by the Michigan State University Agricultural
Experimental Station Disciplinary Research Grant Program. I thank Jim Breck and Lidia
Szabo Kraft for their contributions to and creation of the lake GIS databases, and Arthur
Cooper for a GIS coverage of stream order. I especially would like to thank Mike
Belligan for his tireless efforts in creating and maintaining the landscape database as well
as answering innumerable GIS questions. Additionally, Laura Barrett, Stephen Bowman,
Tyler Roasa, Remy Brim, Cassie Meier, Dave Meyers, and Sarah Wills aided with
compiling and conducting quality control on the databases. Thanks to Gary Weissmann
for help with interpreting geology data, and Scott Winterstein, Dan Hayes and Brian
Maurer for statistical guidance. I thank Seth Martin, Sarah Wills, Dave Meyers, and
Cassie Meier for help with field sampling. I thank Kendra Spence Cheruvelil, Steve
Hamilton, Aaron J ubar, Lesley Knoll, Ace Samelle, Carrie Scheele, and David Weed for
technical assistance. I thank my grad student cohort of Michelle Prehoda, Lauren Bailey,
Veronica Lopez, and Tammy Giroux for additional statistical contemplations and much
needed support. I thank my friend Kendra Spence Cheruvelil for always being there to
listen and get me back on track. I thank my husband Seth for more than I can start to say
here. Thanks to my committee members Mary Brernigan and Jan Stevenson for their
professional guidance and critical eye. Last but not least, I would like to thank my major
advisor, Pat Soranno, for enduring my machine-gun email style, helping me to extract my
thoughts from the my head, and helping me organize those thoughts while avoiding

dangling participles.

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................... v
LIST OF FIGURES ........................................................................................................... vii
INTRODUCTION ............................................................................................................... 1
METHODS .......................................................................................................................... 6
RESULTS .......................................................................................................................... 13
DISCUSSION .................................................................................................................... l7
IMPLICATIONS ............................................................................................................... 23
APPENDIX ....................................................................................................................... 46
LITERATURE CITED ...................................................................................................... 50

iv

LIST OF TABLES

Table 1. Total number of lakes sampled and number of lakes sampled per watershed for
each category of each landscape position metric. LP: landscape position, LH=
lake hydrology, LO: lake order, LNN= lake network number, LNC= lake
network complexity. ............................................................................................. 24

Table 2. List of lake morphometry and water chemistry/clarity response variables
including minimum, maximum, mean and standard deviation across 71 lakes.
SDF = shoreline development factor (unitless), DOC = dissolved organic carbon,
TN = total nitrogen, TP = total phosphorus. ......................................................... 25

Table 3. ANOVA results for landscape position metrics versus all lake response
variables. Smallest AIC values indicate the most parsimonious model fit and are
bolded. Some variables required a 2-way ANOVA model (A) due to significant
interaction term (see methods for further explanation). Only p-values for
landscape position term are provided and are bolded if significant (less than or
equal to 0.01). Positive/negative relationship with increasing landscape position
(from high to low in landscape for each metric) indicated by +/- signs. "ns"
indicates not significant. ....................................................................................... 26

Table 4. ANOVA results from individual watershed analyses. P-values less than or equal
to 0.01 in bold. ...................................................................................................... 29

Table 5. ANOVA results for landscape position metrics versus landscape/physical
features: (A) lake morphometry, (B) proportion of wetland type within the
surrounding 100 and 500 meter area, and (C) proportion of land use/cover types
within the surrounding 100 and 500 meter area. “All Human Uses” category
combines urban and agriculture land use/cover types. Agriculture land use/cover
type was not analyzed separately, see methods for further explanation. Some
variables required a 2-way ANOVA (A) due to significant interaction term (see
methods for further explanation). Only p-values for landscape position term are
provided and are bolded if significant (less than or equal to 0.01).
Positive/negative relationship with increasing landscape position (from high to
low in landscape) indicated by +/- signs. “u” indicates an unimodal relationship.
"ns" indicates not significant. "np" indicates no discemable pattern. ................. 30

Table 6. Comparison of data ranges between Riera et al. 2000, Quinlan et al. 2003, and
this study for: (A) lake water chemistry/clarity, and (B) lake morphometry
variables. Minimum, maximum, and mean values listed. it = number of

observations. Significant relationships found with LO are in bold (p-value less
than or equal to 0.05). ........................................................................................... 33

vi

LIST OF FIGURES

Figure 1. Conceptual framework for the linkages among hydrologic connections,
landscape/physical features, landscape position and in-lake variables. Solid line
and arrow indicates linkages found in published studies. Dotted line indicates
additional linkages examined in this study. .......................................................... 35

Figure 2. Description of landscape position metrics. A) Lake Hydrology (LH). B) Lake
Order (LO). C) Lake Network Number (LN N). D) Lake Network Complexity
(LNC). See text for further descriptions. Categories not included in this study are
indicated with * . .................................................................................................. 36

Figure 3. Major river drainage areas (USGS 8-digit Hydrologic Units) of Michigan with
three watersheds in study area highlighted. Heavy black lines indicate watershed
boundaries. Fine black lines indicate rivers/streams. Circles indicate location of
lakes (solid: sampled, empty: not sampled). ...................................................... 38

Figure 4. Box-plots of lake hydrology and lake network number versus chlorophyll a for
each individual watershed: (A) Muskegon, (B) Au Sable, and (C) Thunder Bay.
Solid lines show categories not significantly different using Tukey multiple means
ramparisons. Some data points were omitted for graphical purposes indicated by
. ....................................................................................................................... 39

Figure 5. Box-plots of lake hydrology, and lake order versus total nitrogen for each
individual watershed: (A) Muskegon, (B) Au Sable, and (C) Thunder Bay. Solid
lines show categories not significantly different using Tukey multiple means
comparisons, and dotted lines indicate non-contiguous categories that are not
significantly different. . ........................................................................................ 40

Figure 6. Box-plots of L0 versus all lake response variables: (A) alkalinity, (B) calcium,
(C) magnesium, (D) chloride, (E) pH, (F) conductivity, (G) silica, (H) nitrate, (I)
sulfate, (J) Secchi, (K) water color, (L) DOC, (M) chlorophyll a, (N) TN, (0) TP,
and (P) TNzTP ratio. Solid lines show categories not significantly different using
Tukey multiple means comparisons, and dotted lines indicate non-contiguous
categories that are not significantly different. Some data points were omitted for
graphical purposes indicated by ‘1 . .................................................................. 41

Figure 7. Box-plots of L0 versus morphometry variables: (A) maximum depth, (B) mean
depth, (C) lake area, (D) lake basin slope, and (E) shoreline development factor

vii

(SDF). Solid lines show categories not significantly different using Tukey
multiple means comparisons, and dotted lines indicate non-contiguous categories
that are not significantly different. Some data points were omitted for graphical
purposes indicated by ‘1 . ................................................................................ 44

Figure 8. Box-plots of lake order versus proportion of wetland types in the 100m and
500m area surrounding the study lakes: (A) forested wetlands, (B) scrub-shrub
wetlands, and (C) all wetland types. Solid lines show categories not significantly
different using Tukey multiple means comparisons, and dotted lines indicate non-
contiguous categories that are not significantly different. Note differences in
scale. . ................................................................................................................... 45

viii

Introduction

Recognition of spatial variability is an integral part of ecosystem research, and
specifically, the quantification and study of the structure of spatial mosaics is one of the
primary goals of landscape ecology (Palmer 2002, Liu and Taylor 2002, Wiens et al.
2002). Application of these principles to areas of ecological research that have not
traditionally been viewed from a landscape perspective can enhance our understanding of
ecosystem structure and function (Rabeni and Sowa 2002). For example, aquatic
ecosystems have benefited tremendously from being viewed from this perspective in the
past few decades (Hynes 1975, Kratz et al. 1997, Fisher et al. 2001).

Stream ecologists were among the first to develop spatially explicit models for
aquatic ecosystems. For example, the River Continuum Concept (RCC) is one of the
early spatial frameworks developed to describe variability among biological communities
in rivers (Vannote et al. 1980). The RCC states that stream communities change as stream
order increases in response to changes in physical habitat and are further structured by
upstream inefficiencies in energy use. More recently, the RCC has been modified to
account for other influences in rivers, such as nutrient inputs, climate change, and
channel alterations (Brezonik 1996). Another example of a spatially explicit aquatic
model is the serial discontinuity concept (SDC), which incorporates disruptions in the
river continuum caused by reservoirs (Ward & Stanford 1983). The SDC proposes that
changing the natural flow regime of a river will obscure RCC predictions for that stream.
Since developing spatial frameworks for the study of rivers, it has become common
practice to view flowing waters along a longitudinal gradient (Frissell et al. 1986,

Hawkins et al. 1993).

In contrast, lakes have historically been viewed as independent microcosms and
have been studied largely as individual ecosystems (National Research Council 1996).
Only recently have lakes been viewed along a continuous spatial gradient, interconnected
through groundwater and/or surface water pathways (Kratz et al. 1997, Soranno et a1.
1999, Riera et al 2000, Webster et al. 2000, Lewis and Magnuson 2000, Quinlan et al
2003). By integrating concepts from landscape ecology, such as identifying and
evaluating the importance of spatial structure in an ecosystem, studies have found that
variability of some lake features follows a pattern consistent with the position of the lake
within the landscape (Kratz et al. 1997, Soranno et al. 1999, Kling et al. 2000, Lewis and
Magnuson 2000, Reed—Andersen et al. 2000, Riera et al. 2000, Webster et a]. 2000,
Quinlan et al. 2003).

The concept of lake landscape position provides a general framework to explicitly
investigate spatial patterning of lake characteristics, analogous to the RCC for streams.
Kratz et al. (1997) define the landscape position of a lake as a “combination of the
hydrologic description with information on the spatial placement of a lake within a lake
district”. For example, in northern Wisconsin, precipitation is the dominant source of
water for lakes positioned high in the landscape, whereas surface and groundwater input
are the dominant source of water of lakes lower in the landscape (Kratz et al. 1997).
Therefore, lakes higher in the landscape respond more strongly and recover more slowly
to drought than lakes lower in the landscape (Webster et al. 2000). Recognizing this
difference in landscape position is important when determining mechanisms driving a

variety of ecological processes in lakes.

Lake landscape position has been measured in three ways to date, each addressing
a distinct aspect of a lake’s hydrologic connectivity. The first measure is based on the
relative position of a lake within a groundwater flow system (Kratz et al. 1997). This
measure was developed and tested specifically in the groundwater-dominated Northern
Highland Lake District of northern Wisconsin. In this district, lakes higher in the
landscape have relatively lower groundwater inputs, and therefore, lower calcium and
magnesium concentrations, derived from groundwater sources, than lakes lower in the
landscape. Expanding upon this groundwater based system, lake chain number, measures
lake landscape position with regard to lakes directly connected along a linear chain by
surface or groundwater flow systems (Soranno et al. 1999). This measure was tested in
seven different lake districts representing a wide range of geologic and hydrologic
settings. In general, as lake chain number increased, loading of non-reactive weathering
products (such as alkalinity, calcium, and magnesium) increased. Lake chain number
was also correlated to increased concentrations of total nutrients and chlorophyll a along
surface water dominated lake chains but not along groundwater dominated lake chains.
Although lake chain number accounts for both groundwater and surface water
connections, it only considers lakes directly connected to one another in a linear fashion.
Finally, lake order has been defined by Riera et al. (2000) as measuring “the type and
strength of connections between a lake and the surface drainage network.” Lake order is
very similar to stream order and accounts only for the presence and strength of the outlet
stream connection. This measure has been tested in the Northern Highlands Lake District
of northern Wisconsin (Riera et al. 2000) as well as in Ontario, Canada (Quinlan et al.

2003). As with lake chain number, lake order also explains significant variability in

alkalinity, conductivity, calcium, and chlorophyll a (Riera et al 2000, Quinlan et a1 2003).
However, unlike lake chain number, lake order was only weakly related to concentrations
of total nutrients.

Although these studies have found that each of these individual measures has
appeared to successfully quantify some aspect of lake landscape position, each one did so
without incorporating the hydrologic connections the other measures emphasized. For
example, lake chain number accounts for the influence of upstream lakes but ignores the
influence of stream order, whereas lake order accounts for stream order but ignores
connections to other lakes. In addition, some measures of landscape position explained
significant variation in lake productivity variables whereas others did not. Thus, it is
unclear which measure of landscape position explains the most variation in lake water
chemistry/clarity variables. Furthermore, few studies have analyzed other features of the
landscape that may be related to landscape position as possible explanatory factors for
why landscape position has such a strong relationship with some lake water
chemistry/clarity variables. For example, some studies have found that heterogeneity of
surficial and bedrock geology regulates patterns of lake response variables along a
landscape position gradient (Soranno et al. 1999, Quinlan et al. 2003). However, these
features have not been specifically incorporated into studies of landscape position.

The goal of my study is to improve our understanding of lake variability and
spatial patterns in lake districts by identifying the best way to measure lake landscape
position and to examine what other features of the landscape are related to it (Figure l). I
argue that lake landscape position is characterized both by the hydrology of a lake

(defined by the presence of lake, stream or groundwater connections), and by other

landscape and physical features that may also vary along a landscape position gradient. 1
hypothesize that landscape position is strongly related to lake response variables because
of its relationship to both hydrology and to these major landscape features.

To examine these issues, I ask two questions: (1) which landscape position metric
is most strongly related to lake water chemistry/clarity variables? I define ‘metric’ simply
as a measurement system differentiating between two or more objects, based on unique
characteristics of those objects. And (2) what landscape and physical features are related
to landscape position? To answer the first question, I compare four different landscape
position metrics based on the different surface hydrologic connections of a lake (stream
only, lake only, stream and lake combined, and lake complexity). Due to the difficulty of
obtaining groundwater data, I was not able to compare landscape position based on
groundwater connections. Because previous studies have found significant relationships
between landscape position and lake water chemistry/clarity variables when measuring
landscape position based on surface connections to lakes (Soranno et al. 1999) or streams
(Riera et al. 2000), I hypothesize that a metric of landscape position that combines both
types of hydrologic connections will explain the most variability in lake water
chemistry/clarity. To answer the second question, I examine possible reasons why
landscape position is such a good predictor of many lake variables by examining whether
other important landscape features, besides hydrology, are related to landscape position.
Because of geomorphological constraints associated with landscape position, I expect
landscape position to also be correlated to lake morphometry, geology, land use/cover,

and wetlands (Figure l).

Methods
Landscape position metrics

I measured landscape position using four different metrics, each emphasizing
different surface hydrologic connections. These metrics are: ( 1) lake hydrology, (2) lake
order, (3) lake network number, and (4) lake network complexity (Figure 2). Each lake
was assigned a category for each landscape position metric using surface water data and
navigational tools provided in the National Hydrography Dataset (NHD,
http://nhd.usgs.gov/). Lakes located within the same major river watershed (USGS 8-digit
hydrologic unit, HUC-8) were considered to be a part of the same lake network.
Descriptions of each metric of landscape position follow.

Lake hydrology (LH) measures landscape position by incorporating both
connections to lakes and streams, providing the overall surface hydrologic position of a
lake (Figure 2A). Lakes are assigned to one of seven categories based on the presence or
absence of inflow and outflow stream connections and connections to other lakes in the
watershed. Seepage lakes (S) are isolated lakes unconnected to any permanent stream,
and therefore no other lakes. Inflow lakes (1) are connected to one permanent stream
(regardless of the actual direction of water flow) but not any other lakes. Inflow/outflow
lakes (10) are connected to two or more permanent streams but not to any other lakes.
Headwater lakes (H) have no inflow stream but are connected to other lakes through an
outflow stream. Inflow headwater lakes (IH) are connected to both inflow and outflow
streams as well as other downstream lakes. Flow through lakes (F) are connected to both

inflow and outflow streams as well as upstream and downstream lakes. Lastly, terminal

lakes (T) are connected to upstream lakes through inflow streams but no downstream
lakes.

Lake order (LO) measures landscape position by emphasizing connections to
streams (Figure 2B). Lakes are assigned a lake order based on the Strahler stream order
of the outflow stream (see Riera et al. 2000 for complete details). Lakes not connected to
permanent inflow streams are separated into the following four categories: (1) lakes
completely unconnected to any stream (permanent or temporary) or wetlands are
assigned lake order ‘-3’, (2) lakes unconnected to a permanent stream but do have a
connection to wetlands are assigned lake order ‘-2’, (3) lakes unconnected to a permanent
stream but do have a connection to a temporary stream (defined as a stream represented
on 1:24 000 map but not on a 1:100 000 map) are assigned lake order ‘-1’, and (4) lakes
connected to a permanent outlet stream but with no inlet stream are assigned lake order
‘0’. Lake order -1 was not included in this study due to low sample size (n=3).

Lake network number (LNN) measures landscape position by emphasizing
connections to other lakes (Figure 2C). Lakes are assigned a network number based on
the number of upstream lakes connected through the same stream, as defined by NHD
navigational tools. Lakes located on tributary streams are assigned a network number
according to the number of other lakes also located along the same tributary. However,
these tributary lakes do not influence the network number of downstream lakes on any
other streams. Lake network number is based on lake chain number as described in
Soranno et al. (1999), however I have modified it by adding a category for lakes in the
same HUG-8 major river watershed that are not directly connected to any other lakes

through stream connections (0).

Finally, lake network complexity (LNC) measures landscape position by
emphasizing connections to other lakes and accounting for the complex branching nature
of most stream networks (Figure 2D). Lakes are assigned a network complexity based on
the number and location of other lakes connected immediately upstream through any
stream, mainstem or tributary. Lakes not connected to any other lakes are assigned a
network complexity based on the presence (only streams, OS) or absence (-) of a
permanent stream connection. Lakes connected to other lakes are differentiated between:
1) those lakes connected to another lake immediately upstream through the same stream
(lake/stream, LS), and 2) those lakes connected to at least two lakes immediately
upstream through different streams (+). Lake network complexity was designed
specifically to capture the intricacy of this last category (+).

Description oLstudy area

My study was conducted in three different lake networks that were defined as
major river watersheds (HUC-8) of Michigan’s lower peninsula: Muskegon, Au Sable,
and Thunder Bay (Figure 3). These watersheds are all hydrologically variable (Sellbach
et al. 1997) and were chosen to minimize differences in climate, land use/cover, and
geology and to maximize the number of lakes per landscape position category. Forested
land use/cover makes up 65% of the study area (Muskegon 53%, Au Sable 79%, Thunder
Bay 67%). The bedrock geology of the study area is 94% clastic sedimentary rock
(Muskegon 96%, Au Sable 100%, Thunder Bay 78%). Surficial geology is 51% outwash
(Muskegon 46%, Au Sable 72%, Thunder Bay 27%), 21% glacial till (Muskegon 18%,
Au Sable 5%, Thunder Bay 50%), and 18% moraine (Muskegon 26%, Au Sable 8%,

Thunder Bay 13%).

Sampling design

Lakes larger than 20 hectares were included in the study. For each of the three
watersheds, I randomly selected 3-5 lakes from each the following lake hydrology (LH)
categories: (1) seepage, (2) headwater, (3) inflow headwater, (4) flow through lakes with
three or fewer upstream lakes, and (5) flow through lakes with more than three upstream
lakes. Inflow, inflow/outflow, and terminal lakes were not included due to low sample
size. I sampled a total of 71 lakes, although, only 68 lakes were analyzed for lake order
because the three -1 lakes were dropped from the analysis (Table 1).

Lake saleing and chemical analysis

 

I sampled each lake one time in 2003 during the summer stratification period
(mid-July through August, although four lakes were visited in mid-September while the
lakes were still strongly stratified) for a variety of physical, chemical, and biological
variables. I conducted depth profiles using a YSI 6920 multi-probe (Yellow Springs Inc,
Yellow Springs, Ohio) for dissolved oxygen, temperature, conductivity and pH. All water
chemistry and clarity samples were taken using an integrated tube sampler from the
epilimnion at the deepest point in the lake. Alkalinity samples were processed within 8-
12 hours of sample collection using Gran titration (Wetzel and Likens 2000). Calcium
and magnesium concentrations were determined by flame atomic absorption
spectrophotometry (Wetzel and Likens 2000). Chloride, nitrate, and sulfate
concentrations were determined using membrane-suppression ion chromatography
(Wetzel and Likens 2000). Silica concentrations were determined using the molybdate
colorimetric method (Wetzel and Likens 2000). Dissolved organic carbon (DOC)

concentrations were determined using hi gh-temperature platinum-catalyzed combustion

followed by infra-red gas analysis of C02 (Wetzel and Likens 2000). Water color was
determined using a Hach model CO-l color test kit (Loveland, Colorado). Chlorophyll a
samples were filtered within 8—12 hours of sample collection through glass fiber filters,
stored in a dark container, and immediately frozen. Filters were soaked in 95% ethanol
overnight and chlorophyll a concentrations determined fluorometrically using
phaeopigment correction (Nusch 1980, Sartory and Grobbelaar 1984). Total nitrogen
concentrations were determined using the 2nd derivative of the absorbance curve at 224
nm following persulfate digestion (Crumpton et al. 1992, Bachmann and Canfield 1996).
Total phosphorus concentrations were determined spectrophotometrically following
persulfate digestion (Murphy and Riley 1962, Menzel and Corwin 1965).

Landscape and physical features

Lake morphometry data were quantified from bathymetric maps. Maximum depth
was obtained using a handheld depth finder. Mean depth was calculated by taking the
average depth of approximately 100 points evenly spaced across each lake bathymetry
map (Omemik and Kinney 1983). Lake basin slope was calculated as (surface
area)”2/mean depth (Numberg 1995). Shoreline development factor (SDF) was calculated
as the ratio of shoreline perimeter divided by the circumference of a perfect circle of the
same area (Wetzel and Likens 2000).

A GIS-based landscape feature database was created for all lakes in the study
area. Wetland data were obtained from the National Wetlands Inventory (NW1,
http://wetlands.fws.gov/) where wetland location, type, and extent were determined using
aerial photography in conjunction with USGS 1:24,000 topographic maps following

Cowardin et al. (1979). For my study, wetland types were grouped by dominant

10

vegetation (forest or scrub-shrub). All wetland types were also combined to produce a
category for overall wetland coverage. The proportion of each wetland type present in the
100 m and 500 m buffer areas around each lake was calculated. Land use/cover data were
obtained from the Michigan Resource Information Service (MIRIS 2000), where the
location and extent of urban, agriculture, upland field, and forest land use/cover type was
determined using the Anderson Classification scheme (Anderson et al. 1976) from aerial
photographs taken between 1978 and 1985 at a resolution of 2.5 ha. Urban and
agricultural land use/cover types were combined to form a new land use/cover category
for analysis of all land dominated by human uses. The proportion of each land use/cover
type present in the 100 m and 500 m buffer areas around each lake was calculated.
Bedrock geology data were obtained from the Geologic Survey Division of the Michigan
Department of Environmental Quality. Bedrock geology types were grouped into the
following categories: (1) carbonate, (2) elastic, (3) hard rock, (4) salt, and, (5) iron.
Surficial geology data were provided by the Michigan Natural Features Inventory and
Michigan Department of Natural Resources. For my study, surficial geology types were
grouped into the following types: (1) dune sand, (2) glacial till = fine, medium and
coarse-textured glacial till, (3) lacustrine, (4) moraine = fine, medium, and coarse-
textured end moraine till, and, (5) outwash = glacial outwash sand and gravel and
postglacial alluvium, ice-contact outwash sand and gravel. The proportion of each
bedrock and surficial geology type present was determined for the 500 m buffer areas
around each lake.

Statistical analyses

11

The relationship between landscape position and lake response variables was
tested for each landscape position metric using two-way analysis of variance (2-way
ANOVA). These models include landscape position and watershed as categorical
predictor variables, as well as an interaction term. This design evaluates the role of each
watershed and landscape position metric and determines whether patterns with landscape
position vary among watersheds. An example of a 2-way ANOVA model is as follows:
(1) Yijk=ll+xj+Yk+(Xj*)’k)+€ijk
Where:

Yij'k: response variable for lake i in watershed j with landscape position k

,u = overall mean of response variable

x,- = watershed j

yk = landscape position k

()9 * y.) = watershed j by landscape position k interaction

eUk = error term for lake i in watershed j with landscape position k
All response variables were first analyzed using a 2-way AN OVA. If the interaction term
from the 2—way ANOVA was not significant (p-value less than or equal to 0.01), a one
way analysis of variance ( 1-way ANOVA) with landscape position as the only predictor
variable was used to analyze the relationship. An example of a 1-way AN OVA follows:
(2) Yijk = l1 “I” Yk + eijk
Where:

Yijk: response variable for lake i in watershed j with landscape position k

p = overall mean of response variable

yk = landscape position k

12

ejjk = error term for lake i in watershed j with landscape position k
If the interaction term was significant, then 1-way ANOVA was used to analyze the
relationship for each watershed separately.

A conservative significance level was chosen (p-value less than or equal to 0.01)
to minimize the probability of finding spurious significance due to the large number of
comparisons (minimizing type 11 error). All response variables were transformed to meet
normality assumptions. Tukey multiple means comparisons were used to determine
which landscape position categories differed significantly (p-value 0.05). Because each of
the metrics of landscape position had different numbers of categories, Akaike
Information Criterion (AIC) values were calculated to determine which of the four
metrics provided the best fit to the data. AIC allows for the unbiased comparison between
models of different size whereas comparing R2 does not. Metrics with AIC values more
than 7 units lower compared to other metrics were considered to be substantially better
(Bumham and Anderson 2002). SAS version 8 software was used to compute all statistics

(SAS Institute Inc.).

Results

The study lakes varied widely in water chemistry/clarity and morphometry
characteristics (Table 2). Lake area ranged from fairly small lakes (20 ha, the lower limit
included in the dataset) to the largest inland lake in Michigan (Houghton Lake, 8124 ha).
On average, the study lakes were slightly basic, moderately to highly buffered, and
moderately clear. However, chlorophyll a had a narrow range and most lakes were

oligotrophic to mesotrophic.

13

Measuring landscape position using different hydrologic connections

To determine how best to measure landscape position, I evaluated the ability of
the four different metrics of landscape position to explain variation in lake water
chemistry/clarity variables. All landscape position metrics explained significant variation
in some water chemistry variables (Table 3). In particular, a majority of dissolved
conservative ions and dissolved reactive ions were significantly related to each of the
landscape position metrics (Table 3). In contrast, only one productivity variable (TNzTP
ratio) was significantly related to the landscape position metrics, although, TN is

moderately significant for LH and LO (Table 3). Also, only two of the landscape

 

position metrics (LH and LO) explained significant variation in any measures of water L
clarity (DOC). Therefore, variation in many dissolved ions, as well as TNzTP ratio, is

explained by landscape position metrics that measure either type of surface hydrologic

connection (lake, stream, or a combination of the two), whereas DOC is explained only

by landscape position metrics that incorporate connections to streams (LH and LO). All

significant models show a positive relationship with landscape position metrics,

suggesting that dissolved materials accumulate along the landscape position gradient,

from high to low in the landscape. The ratio TNzTP, on the other hand, shows a negative

relationship with landscape position.

Although a majority of lake response variables were analyzed using a 1-way
ANOVA model, that included landscape position only, some variables required a 2-way
ANOVA model, that included landscape position and terms for watershed and the
interaction between landscape position and watershed (Table 3). Two measures of lake

productivity (TN and chlorophyll a) resulted in significant interaction terms, meaning

14

that the watersheds show different patterns between landscape position and the lake
response variable, so each watershed required its own model. In the Au Sable watershed,
chlorophyll a was negatively related to landscape position as measured by LH and LNN,
both explaining similar amounts of variation (64% and 62%, respectively) (Table 4,
Figure 4). Landscape position was not significantly related to chlorophyll a in either the
Muskegon or Thunder Bay watersheds, although a negative pattern in the Thunder Bay
watershed is apparent, similar to that found in the Au Sable watershed (Table 4, Figure
4). There was no significant relationship between TN and landscape position in any of
the three watersheds (Table 4, Figure 5).

To compare the different landscape position metrics, I used AIC and R2 values. 1
found that L0 was consistently the landscape position metric with the best AIC value
(lower by 7 units) and the highest R2 value for all significant models, ranging from 22%
(DOC) to 53% (conductivity and calcium) of variation explained (Table 3). Lake
hydrology was the second best metric for all lake water chemistry/clarity variables, with
the exception of TNzTP ratio, where the AIC values for LO, LH, and LNN were very
close, with an overall difference of only 6 units among the three metrics, thus being
statistically indistinguishable. Because LO was found to be the best overall metric of
landscape position, only box-plots of L0 versus response variables are shown (Figure 6).

I used Tukey multiple means comparisons to examine differences among LO
categories (Figure 6). Generally, I found the formation of two groups: lakes not
connected or minimally connected to streams (-3, -2, and 0), and, lakes more highly
connected to streams (l, 2, and >=3) (Figure 6), although there is much variation in this

general pattern depending on the lake variable examined. For example, although L0

15

categories have significantly different DOC concentrations, Tukey comparisons do not
show significant differences between any individual pairs.
Relationships between landscape position and landscape features

I examined whether several landscape features were related to landscape position
by analyzing landscape position as a predictor variable and various landscape features as
the response variable. Landscape position was significantly related to two general
features of the landscape - lake morphometry and the proportion of wetland types in
buffer areas surrounding the study lakes (Table 5). All landscape position metrics were
positively related to lake area, with LO category >=3 having the largest lakes (Table 5
and Figure 7). Three landscape position metrics (LH, LO, LNN) were significantly
related to SDF, although patterns differed slightly among the three metrics (Table 5).
Only two metrics (LNN and LNC) were significantly related to lake basin slope, both
showing a positive relationship (Table 5). All four landscape position metrics were
significantly related to the proportion of all wetlands in both the 100 m and 500 m buffer
areas, with the proportion of wetlands generally increasing or showing a unimodal
relationship along the landscape position gradient (Table 5). Various landscape position
metrics were also significantly related to wetlands when grouped by dominant vegetation
type (Table 5). Tukey multiple means comparisons show similarities between many LO
categories, with few categories significantly different from one another (Figure 8). The
percent variance explained for significant relationships was generally low, ranging from
16% to 33% (Table 5).

I found no significant relationship between any landscape position metrics and

land use/cover (Table 5). I was not able to statistically analyze total agricultural land

16

use/cover due to zero values for many study lakes (58 zeros in the 100m area, 40 zeros in
the 500m area). I was also not able to statistically analyze bedrock or surficial geology
due to zero values for most geology types. Thus, geology and land use/cover change

little in these watersheds along the landscape position gradient.

Discussion

There are two important conclusions about lake landscape position stemming
from this research. The first is that the landscape position metric that measures the
presence and magnitude of stream connections (rather than other surface hydrologic
connections) was most strongly related to lake water chemistry/clarity variables. This
result suggests that the magnitude of stream inputs is a major factor driving patterns
associated with lake landscape position. The second conclusion is that landscape and
physical features, such as lake morphometry and the presence and magnitude of wetland
connections, are significantly related to lake landscape position. These relationships
should help to uncover additional possible mechanisms driving the relationships between
landscape position and lake response variables. In addition, these patterns may allow us
to define landscape position better in regions where surface hydrologic connections are
difficult to measure or absent.
Landscape position and hydrologic connections

Previous studies have shown many common patterns associated with the
landscape position of a lake. My results confirm many of these patterns. For example,
many dissolved conservative ions and some dissolved reactive ions increase with

increasing landscape position as measured by groundwater connections (Kratz et al.

17

1997, Webster et al. 2000), lake connections (Soranno et al. 1999, Kling et al. 2000), and
stream connections (Lewis and Magnuson 2000, Riera et al. 2000, Quinlan et al. 2003).
Even though these previous studies have shown the utility of using landscape position in
modeling the spatial variation of lake water chemistry/clarity variables, none has
compared the different measures of landscape position.

Because LO was most strongly correlated to lake response variables, it appears
that the presence and magnitude of stream connections is a stronger driver of the
relationship between landscape position and lake water chemistry/clarity variables than
the presence and magnitude of lake connections. However, I also found strong
relationships with landscape position metrics that measure hydrologic connections to
other lakes (LH, LNN, LNC). Lake hydrology (LH) explained only slightly less variation
in lake response variables than L0, and all landscape position metrics (except LNC)
explained TNzTP ratio equally well. Only one lake response variable (DOC) was
explained solely by metrics of landscape position that included stream connections (L0
and LH). This last pattern is not surprising given that a large portion of DOC originates
from allochthonous material in the watershed and is transported via surface water flow
(Likens and Borrnann 1979,Kaplan et a1. 1980, Molot and Dillon 1997, Schiff et al.
1997). In addition, previous studies have found that watershed characteristics explained
more variation in DOC for drainage lakes (lakes connected to streams) than for seepage
lakes, suggesting that streams provide a constant source of DOC to lakes (Kortelainen
1993, Gergel et al.1999). Based on the above evidence, it is clear that both connections to
lakes and to streams influence lake water chemistry/clarity variables through landscape

position. In some instances choosing the best metric of landscape position may depend

18

not only on what one is trying to predict, but also on data availability. All of these metrics
are based on relatively coarse map data (1:100 000); however, for L0, some categories
are more difficult to measure because they also require wetland and finer scale stream
maps (1:24 000), making LO the most difficult to quantify.

To examine the usefulness of L0 across a wide range of different geomorphic
settings, I examined whether there were common patterns in studies from three different
lake districts by comparing my results with two previously published studies (Table 6): a
groundwater dominated lake district in Wisconsin (Riera et al. 2000), and a surface water
dominated lake district in Ontario (Quinlan et al. 2003). Although there are many
common patterns, there are also some interesting differences across regions. Lake order
explains a significant amount of variation in many dissolved conservative ions and
dissolved reactive ions, all of which are related to weathering. However, Riera et al.
(2000) did not find a relationship between L0 and sulfate, which may be due to the small
range and low mean concentration of sulfate found in their study lakes compared to the
other lake districts (Table 6A). In contrast to weathering variables, productivity variables
seem to be the most difficult to predict from L0. For example, none of the three studies
found a significant relationship between L0 and TP and there are mixed results for TN
and chlorophyll a. Riera et al. (2000) found that L0 explained a significant amount of
variation in chlorophyll a concentrations whereas I did not. This result, again, may be due
to regional differences in the ranges of values. The range of chlorophyll a values in
Wisconsin lakes was larger than in Michigan lakes. Differences among Michigan
watersheds may also explain why there was no relationship with chlorophyll a. Lake

order was significant with chlorophyll a in only one of the three watersheds, and

19

surprisingly exhibited a decreasing pattern with LO (Table 4, Figure 4). This result may
be due to the contrasting patterns of nutrients and DOC, which may increase the
importance of DOC induced shading in controlling chlorophyll a (Carpenter et al. 1998,
Beisner et al. 2003). Analyzing data at different levels of aggregation (ie. watersheds
combined vs. watersheds separated) as well as regional differences in the data ranges can
help clarify many of the discrepancies in findings across these three studies.
Landscape position relationships to landscapefitysical features

In order to properly infer mechanisms driving the relationships between landscape
position and lake water chemistry/clarity, it is important to also understand the
relationship that landscape position has with other features that may influence lake water
chemistry/clarity. Many features of the landscape have been shown to be directly related
to lake water chemistry/clarity and may explain as much variation as landscape position.
For example, the presence and amount of wetlands surrounding lakes have been found to
explain a significant amount of variation in concentrations of lake DOC (Gergel et al.
1999, Prepas et al, 2001, Xenopoulos et al. 2003), TP (Detenbeck et al. 1993, Devito et
al. 2000, Prepas et al. 2001), and TN (Detenbeck et al. 1993, Prepas et a1, 2001). My
results show that landscape position was significantly related to some measures of lake
morphometry and the amount of wetlands surrounding lakes. Lake area and SDF were
related to increasing landscape position (Table 5A), and specifically to increasing LO
across three regions (Table 6B). Therefore, analyzing relationships between landscape
position and measures of morphometry may be important in understanding shifts in the

food web from lake to lake. For example, lake area is an important factor driving fish and

20

zooplankton community structure (Fryer 1985, Dodson 1992), and may be important in
understanding subsequent relationships with landscape position (Kratz, 1997).

Other measures of lake morphometry show slightly different relationships with
landscape position when compared among the three studies measuring LO. Maximum
depth appears to increase with increasing landscape position, but only when a wide range
of lake depth was included in the study. Although Wisconsin and Michigan lakes had
similar maximum values for lake depth, the Wisconsin study also included very shallow
lakes. On the other hand, the Ontario and Michigan lakes shared similar minimum values
for lake depth, but the Ontario lakes were far deeper than the Michigan lakes. Although
Riera et al. (2000) reports a significant relationship between L0 and mean depth, they
temper their results in light of statistical artifacts stemming from their highly unbalanced
sample design. This comparative analysis suggests that regional differences in data
ranges may influence the relationships between landscape position and
landscape/physical features.

For my study site, I also analyzed relationships to other landscape features such as
land use/cover, geology and presence and magnitude of wetlands surrounding lakes.
Given that the design of this study was to minimize variation in land use/cover and
geology, it is not surprising that none of these features were related to landscape position.
However, the presence and magnitude of wetland areas increased significantly along the
landscape position gradient. Therefore, the relationship between landscape position and
lake water chemistry/clarity variables may be responding to some combination of
increasing wetland areas and increasing hydrologic connectivity. Specifically, the

relationship I found with DOC may be a product of changes in landscape position or

21

increasing amounts of wetland areas. In contrast, although total phosphorus has been
found to be significantly related to increasing wetlands; and landscape position was
significantly related to increasing wetlands in this study; nevertheless, total phosphorus
was not related to any landscape position metrics in any of the three regions (Wisconsin,
Ontario, Michigan). For Michigan lakes, this finding may be due the low overall
proportion of wetlands (>40%) included in the study. Prepas et al. (2001) found a
significant relationship between total phosphorus and wetlands only when wetland cover
exceeded 40%, approximately. Studies have also found an interaction between total
phosphorus and wetland type (wet meadow, marsh, bog, poor fen, rich fen, etc.), with
different types acting as a source or sink of total phosphorus (Detenbeck et al. 1993,
Prepas et al. 2001). In addition, Devito et al. (2000) found that total phosphorus
decreased in lakes as groundwater input (measured by calcium and magnesium
concentrations) increased. The contradictory nature of these findings helps to explain the
lack of pattern in total phosphorus in my study. The addition of groundwater information
and more detailed wetland information may improve the analysis of the relationship
between landscape position and water clarity and productivity variables. Although LO
includes information on the presence of wetlands, which may explain why this metric had
the strongest relationship to water chemistry/clarity variables, it may be more beneficial
to incorporate wetland information as a separate factor from hydrologic connectivity,
allowing for the interpretation of each factor. In conclusion, incorporating other
landscape features, such as the presence and magnitude of wetlands around lakes, into
future definitions of landscape position may explain additional variation in lake water

chemistry/clarity and help to identify underlying mechanisms.

22

Implications

This study should help to refine the concept of landscape position and suggest
possible underlying mechanisms driving variability among lakes in seemingly similar
settings. By taking a comparative approach, I have identified the type of hydrologic
connection most related to lake landscape position in a region of variable hydrology. 1
have also broadened the view of landscape position beyond solely considering hydrologic
connections to specifically incorporate relationships to other landscape/physical features.
This more comprehensive definition of landscape position should help characterize lakes
in regions where landscape features may play a larger role than hydrologic connectivity
in explaining lake variability, such as in extremely wet or arid regions. The definition of
landscape position will continue to expand as the concept is tested in diverse regions,
allowing more accurate extrapolation to unsampled lakes and a clearer understanding of

lake variability at the landscape scale.

23

 

Table 1. Total number of lakes sampled and number of lakes sampled per
watershed for each category of each landscape position metric. LP: landscape
position, LH= lake hydrology, LO: lake order, LNN= lake network number, LNC=
lake network complexity.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

24

Watershed
LP Metric LP category Muskegon Au Sable Thunder Bay TOTAL

LH S 8 8 6 22
H 5 4 3 1 2
1H 8 5 5 1 8
F 9 6 4 1 9

subtotal 30 23 1 8 71

L0 -3 3 3 1 7
-2 5 3 4 1 2
5 4 3 12

1 3 1 3 7
9 6 2 1 7
>=3 5 4 4 1 3
subtotal 30 21 17 68
LNN 0 8 8 6 22
1 13 9 8 30
2 4 3 3 1 O

>=3 5 3 1 9

subtotal 3O 23 1 8 71

LNC - 8 a 6 22 J
LS 1 5 12 9 l 36
+ 7 3 3 13 7
subtotal 3O 23 1 8 71

 

 

 

Table 2. List of lake morphometry and water chemistry/clarity response variables
including minimum, maximum, mean and standard deviation across 71 lakes.
SDF = shoreline development factor (unitless), DOC = dissolved organic carbon,
TN = total nitrogen, TP = total phosphorus.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Morphometry Min Max Mean SD
Maximum depth (m) 3 36 13.8 7.7
Mean depth (m) 1.5 15.2 5.2 3.1
Lake area (ha) 20 8124 309 1052
Lake basin slope 143 5654 609 786
SDF 1.0 4.9 2.1 0.8
Dissolved Conservative Ions Min Max Mean SD
Alkalinity (ueq/L) 1 70 3722 2232 792
Conductivity (uS/cm) 24.0 401.0 253.8 79.7
Calcium (mg/L) 2.1 57.7 30.7 10.6
Magnesium (mg/L) 0.9 18.0 10.6 4.2
Chloride (mg/L) 0.4 26.5 8.2 5.9
pH 6.5 9.3 8.1 0.5
Dissolved Reactive Ions Min Max Mean SD
Silica (mg/L) 0.04 5.53 2.20 1.63
Nitrate (ug/L) 0 913 39 1 1 1
Sulfate (mg/L) 2.5 27.0 8.2 4.5
Water Clarity Min Max Mean SD
Secchi (m) 1.3 9.2 3.9 1.4
Water color (PCU) 0 30 10 9
DOC (mfl) 0.0 27.6 10.5 6.5
Productivity Min Max Mean SD
Chlorophyll a (ug/L) 0.3 15.4 3.0 2.5
Total nitrogen (ug/L) 102 1509 540 289
Total phosphorus (ug/L) 2.6 34.0 1 1.5 6.3
TNzTP ratio 15 130 52 25

 

25

 

 

 

Table 3. ANOVA results for landscape position metrics versus all lake response
variables. Smallest AIC values indicate the most parsimonious model fit and are
bolded. Some variables required a 2-way ANOVA model (A) due to significant
interaction term (see methods for further explanation). Only p-values for
landscape position term are provided and are bolded if significant (less than or
equal to 0.01). Positive/negative relationship with increasing landscape position
(from high to low in landscape for each metric) indicated by +/- signs. "ns'
indicates not significant.

26

Amie w.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. . 5.6 I 90.0 PI. 5.8 0304 F0.
0.000200 003002321030 10-8.50 N5 9.2“.» 0%3 0-<ml.cm >.O 3» peak
2.3.3.... A983 .80 9.5 + A983 08 98 +
0030:0232 A983 um» 98 + A983 28 99... +
00.063 A983 80 98 + A983 53 9mm +
2.080903 A983 8m 98 + A983 93 98 +
03.0100 9...» wok. 90m :0 98.. .8 9.m :m
0... 98.x .mm 90» :0 98m .3. 98 :0
0.000200 300220 .030
mace A983 ..m 98 + A983 .8 98 +
2.2.0.0 930 .ou 98 :m 93m .8 93 am
9:08 A983 mm 9m. + A983 8 93m +
203.. 0.012
money. 9qu um 90» :m 908 um 93 :m
508400.04 9.8 B.mm 93 :m 930 can 9.0 :m
000 98. to 9m. + 908 3w 9mm +
bananas?
020303.. m 908 .mm 98 3m. 9.3m .mm 9.0 :m
...o.m_ 3.2000: 93. mo 98 :m 934 > mm 9m. :0
408.030.30.325 90mm .5 98 3m 98» ..© 98 :m
92”....0 820 A983 8 9mm - A983 .3 0b.. -

 

 

27

400.0 0 003..

 

5.000200 003002030

0.0 20320} 2:300“ F22.

r0x0 20.5.0} 0030.082 2.20.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.030 0-<0l_m0 >.0 .0» 09903 0-<0|=W 20 3» 00303
2.0.3.2 A983 .00. 900 + A908. .000 90m +
003000.25. A908. «00 0.00 + A9003 «3. 0.00 +
00.0.03 A9003 k.00 90m + A980. 000 00.. +
7.0030053 A983 00.0 9mm + A983 000 9m0 +
03.0300 9000 wow 9.0 30 0..m0 N00 0.00 30
B. 90.0 ..0 900 30 90.0 .m. 90m 30
0.000200 3000.20 .030
9:00 98» .mm 0.m0 + 0.00. .mm 0.. 0 +
250.0 0.000 .00. 900 30 998 Sn 0.. . 30
05.0.0 903 .00 9m. + 0.03 .00 0.. 0 +
50.0.. 0.04.2
00003., 9000 E 900 30 9.00 No 900 30
20.0.. 00.0_. 0.. .0 00m 900 30 0.0mm .000 0.00 30
000 0.00. 000 9.0 30 90.0 000 0..m 30
03000332
03.0603... 0 903 .N0 90. 30 9000 .00 0.0. 30
...0.0_ 3.20003 9000 . .0 0.00 30 903 .8 90m 30
.38. 030003030 9.00 ..0 0.00 30 0.00. . .0 904 30
42.....0 4030 A980. 40 0.0. - 900.. 0. 9.0 -

 

 

28

Table 4. ANOVA results from individual watershed analyses. P-values less than
or equal to 0.01 in bold.

 

Lake Watershed

 

response Muskegon Au Sable Thunder Bay
variable LP metric p-value R’- p-value R2 p-value R2

 

LH 0.801 0.04 0.000 0.64 0.239 0.25

 

Ch'°'°phy"a LNN 0.734 0.05 0.000 0.62 0.632 0.11

 

LH 0.233 0.15 0.012 0.43 0.101 0.35

 

TN LO 0.458 0.17 0.072 0.46 0.048 0.60

 

 

 

 

 

 

29

 

Table 5. ANOVA results for landscape position metrics versus
landscape/physical features: (A) lake morphometry, (B) proportion of wetland
type within the surrounding 100 and 500 meter area, and (C) proportion of land
use/cover types within the surrounding 100 and 500 meter area . “All Human
Uses” category combines urban and agriculture land use/cover types.
Agriculture land use/cover type was not analyzed separately, see methods for
further explanation. Some variables required a 2-way ANOVA (A) due to
significant interaction term (see methods for further explanation). Only p-values
for landscape position term are provided and are bolded if significant (less than
or equal to 0.01). Positive/negative relationship with increasing landscape
position (from high to low in landscape) indicated by +/- signs. “u” indicates an
unimodal relationship. ”ns" indicates not significant. "np" indicates no
discemable pattern.

30

400.0 0.

>.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

r020 r020 1403.815... r020 030.. F9
2.2030303. 0-<0...0 20 0» 0020.3 0.5.00 20 0» 0020.3
2.0.3303 000.3 0.000 .00 900 30 0.000 .00 0.00 30
2.003 000.3 9004 .m0 900 30 90.0 ..0 900 30
r0x0 0.00 980 -040 9.0 + 980 -00.. 0.0.» +
.-0.0 000.3 0.000 9.4» .08 0.00 30 9.00 -000 0.... 30
mo.” V983 -0» 0.8 + 930 -00 0.». 30
. 502.030 4.400 5.8 1.3.0.09. .5... 5.6 030.. F0.
.83 00204 0.350 >5 0» 0020.3 0.3.30 20 0» 0020.3
00.00.00 .2000 98. 4.. 98 +4: 980 40 0.00 +
0030-033... 2000 0.0mm :0 0.. 4 30 980 .04 904 +4:
>._ 20:03.. .4000 V983 -00 0.8 +4: 903 .8 0.8 +
0003 0:204
00.00.00 £000 908 .m0 0.3 30 — 0.000 .mm 9.0 30
028-03% 280 9.00 .0. 0.04 :0 _ 0.80 :0 0.0.. +3
>__ 20.02. .0000 v0.08. .8» 0.0.. +3 _ 0.80 -04 on. +
. .1030 000.0054 5.6 1840.00.42.10 p.020 0301.5.
.83 0:20.. 0.5.00 20 0» 0020.3 W350 >.0 0» 0020.3
:30: 0.000 .0 0.00 30 0.040 -40 0.00 30
C0033 0.0.3 9.0» 00 9.» 30 0.040 40 9mm 30
00400. 900» m0 900 30 0.000 -. m 0.00 30
>._ 1.303 C000 9.0. m. 0.00 0.000 -4 0.00 30
083 0:20..
:30: 94?. -00 0.0» 30 0.0mm mm 0.0» 30
C0030 0.0.3 0.0.0 > 00 0.00 30 0.000 04 0.04 30
00400. 0.004 -. 0 0.00 30 9000 m4 0.00 30
>__ 1.303 C000 90.. -.0 0.0. 30 0.0.0 m0 900 30

 

 

31

400.0 0 003.

>.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

00x0 00x0 20020.... 20302 2.22. 00.3 20220:. 0030.902 2.20.
2.2000303. 0.<0.00 >5 0» 0020.3 b.<0.00 >5 0» 0020.3
2.0.0303 0002. 9400 .00 900 30 0.000 .00 0.04 :0
2.00: 000.: 9000 .m. 900 30 9044 ..0 900 30
00x0 000 v0.000. -000 900 + 0.00. -000 9.0 +
00x0 000.: 0.000 0.00. .00.. 900 + 0.00. -000 90. +
000 V9000. -00 0.0. + 9000 -00 0.00 :0
. 5.02.030 ...<00 00.8 20020:. 20300.. 0.22. 00x0 20220.... 0030.902 2.20.
.003 0020« 0.5.00 >5 0» 00203 0.<0.00 >5 .0» 00203
00.00.00 2000 9000 0. 0.. 0 30 9000 0. 0.. 0 30
00.00.0300 2000 0.00» .00 900 30 0.004 . . . 0.. 4 0
>.. <<0..0:0 €000 V9000. -00 0.00 0 v0.000. -00 900 0
0003 0020..
028.0... 000,-. 0.4.... .00 0.8 :0 _ 0.000 .00 Pom :0
02:33:... 2.80 0.80 .00 9.4 c _ 0.000 .00 0.0.. :0
>_. <<0..030 QW00 0.000 -00 0.. 4 +3 _ 0.00» -. 0. 0.. 0 t0
. 0030 0004020.. 00x0 20020.... 20300.. 2.22. 00.6 20020:. 0030.00.21.25.
.003 0020.. 0.<0.00 >5 0» 0020:. 0.<0.00 >5 0» 00203
C603 0.. 00 -00 0.04 :0 0.000 .0 900 30
C0.0:0 0.0.0 0.... 00 9.0 30 9040 00 9.0 30
00.00. 0.040 -00 0.00 30 9000 m0 900 :0
>.. 1030.. C000 9.40 0.04 30 0.000 .0 0.00 30
0003 0020a
C60... 90. m .0 0.00 :0 0.. 04 -00 0.00 30
C0.030 0.0.0 9400 00 0.00 :0 0.400 > 04 0.00 30
00.00. 900. 00 0.00 :0 9000 -00 900 :0
>._ 10303 C000 9000 mm 0.0. :0 0.004 -.0 900 30

 

 

 

32

Table 6. Comparison of data ranges between Riera et al. 2000, Quinlan et al.
2003, and this study for: (A) lake water chemistry/clarity, and (B) lake
morphometry variables. Minimum, maximum, and mean values listed. n =

number of observations. Significant relationships found with LO are in bold (p-
value less than or equal to 0.05).

33

 

 

400.0 0.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

>. 00.00 20.04 0.0.0 0. 0.. ~08 00:03 02 0.. ~80 40.0 900..

02.03.03.331... 3.3 30x 3003 3 3.3 30x 3003 3 3.3 30x 3003 3
2.0.3.... .0 ~0~0 000.0 000 0 000 .000 00 .40 04- -0~ 4.
0030092... 0 ~00 0~.. 000 .0.0 ..0 00.0 00 ~0 00. ~00 4.
00.0.03 9.0 ~. .00 0.00 .00 ..0~ .0.0 0.00 00 ~.. 04.4 094 4.
2.0030053 0.00 0.- ~... .40 0.04 ~.0 ..00 00 90 .0 .90 4.
00.0300 9.0 0.0 0.00 04 3.0 0.0 ~00 0.~ 4.

E. 0.0 0.0 0.0 000 0.0. 4.00 0.00 00 0.0 0.0 0.. 4.
0:80 0 .00 ~.. .0. 0... ~..0 900 00 0.0 0.0 ~.~ 40

00:08 0.0. 0.00 0.00 .0~ 0.4 0.0 4.04 00 ~.0 ~4.0 0.~ 4.
00003. 0.0 .0.. 0.0 ~00 3\0 . .0 9~ 0.0 40
<<0.0. 00.04 0 .0~.0 ~4.4 ~.0 30 0 00 .0 40

GOO .0 .~.0 0.~ 40 .00 0.0 0.00 00 0 ~40 .90 4.

03.0603... 0 0.4 00.0 0.00 40 3.0 0.0 .00 0.0 4.

42 . 00 000 000 . 00 . 00 00. .0 ~0~ 00 .0~ .000 000 4.

.4.”v ~ 0m .0 0m ~.4 00.4 0.0 00 ~.0 00 ...0 4.

42.40 3.0 4.0 0.4 0.0 00 .0 .00 0~ 4.

0. 00:0 3.040 0. 0.. ~80 00:03 0. 0.. ~80 40.0 ”0%

2.2000303. 3.3 30x 3003 3 3.3 30x 3003 3 3.3 30x 3003 3
2.0x 000.0 0.0 00.. 0.0 0. 0 0.0 00 00.0 00 0 00 .0 4.
2.003 000.3 ..0 .0 0.0 00 3.0 . .0 .0.~ 0.~ 00

00x0 0.00 0.0 .000 00.4 000 0.0 .040 ~00. 00 ~90 0. ~0 009~ 4.

000 ..0. 0.~. ..00 000 3.0 . .0 0.0 ~.. 4.

 

 

 

 

 

 

    

 
  
  

   
  

Landscape/Physical
Features

Land Lake
use/cover morphometry
Geology Wetlands

O
O

Hydrologic
Connectivity

  
 

            

 

 
   
   
 

Lake water
’ chemistry 8. clarity

 

Figure 1. Conceptual framework for the linkages among hydrologic connections,
landscape/physical features, landscape position and in-lake variables. Solid line
and arrow indicates linkages found in published studies. Dotted line indicates
additional linkages examined in this study.

35

Figure 2. Description of landscape position metrics. A) Lake Hydrology (LH). B)
Lake Order (LO). C) Lake Network Number (LNN). D) Lake Network Complexity
(LNC). See text for further descriptions. Categories not included in this study are
indicated with * .

36

 

 

 

 

.I

0.00---o-'--Ot:

ANC'OCDONQQ)

 

00.0 020.000 . +

0.0000430... H0..
00.00.00 200 “00
00.0 020000 N u

0;”

 

 

 

.3

0000003

 

00.000... H...
0030.0. 26.". up.
0.03.000... 26.00. .1.

0003000.... U...

30.030.30.00. H0.
>.600. N_
0000000 .0

 

 

 

 

37

..00.0E00 .00 "0.00.0

00.00.00 "0:00. 09.0. .0 02.000.
0.00.00. 00.8.0 000000.202.
0.00.00. 000.. V.00.... 00.”.

000000300 00020.02. 0.00.00. 000..
0.0.0.0 .900... 00.00.30... 00.0 >030
0. 000000.02. 00.0. 0..; 000.00....
.0 .903 202231.03 0003.
000.0 0000.0... .02. 8.0.). .0 0.30. H.

              

0000 3.0.

7‘ v‘@
. . x
.0..

38

A. Muskegon ‘

I F I I 10 I

 

 

.3
O

 

 

 

 

Chlorophyll a (ug/L)
*3 9F 0: 9
0+ 0? 91L
.{D o

l 021:]— «I»
l L 1 l

 

 

O
p

O
P

 

 

 

 

 

 

 

 

 

 

 

B Au Sable
10 I I I I 10 l V I I
2 8r- - 8r -
O
a
(B 6” '4 6' “l
E
:1 4r ~ 4r ~
2 Q
2 ~ 0
.c .. ..
O 2‘ a 2
o . . . 5E 0 . . ? 'r
C. Thunder Bay
16 T V l F 10 1 f I I
8'- ‘ 8- -
6 ~ 6- J

Chlorophyll a (ug/L)
.h

I

N

l

 

 

 

 

l

 

O
O

 

 

_ - ¢
-B*$B..B¢a

s H lH F
Lake Hydrology Lake Network Number

>=3

Figure 4. Box-plots of lake hydrology and lake network number versus
chlorophyll a for each individual watershed: (A) Muskegon, (B) Au Sable, and (C)
Thunder Bay. Solid lines show categories not significantly different using Tukey
multiple means comparisons. Some data points were omitted for graphical
purposes indicated by e.

39

A. Muskegon

 

 

 

 

 

 

20005 I I I I 2000 I I I I I I—
1500- 0 1500- o -
E10004— ~ 1000- 0 -<
E ,._., a
m'éjé Ei~5w0o$¢trm
L L 4 I 1 l

O
b
O

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

B. Au Sable
2000 l l l I 2000 l I I T f l
1500- ~ 1500- ~
3
Smoo- - 1000- _ ~
2
'—
0-0053 -0.-w $~
G j 1 "l... (I) G l I l l l I
C. Thunder Bay
2 1 I I I 2000 I I I I I I
1500* 0 1500' 1
31000- ~ 1000- -
Z
I- _
500i [II] é - 500— a] B ~
5!
0 L l I l l l l l l I
s H IH F -3 -2 0 1 2 >=3
Lake Hydrology Lake Order

Figure 5. Box-plots of lake hydrology, and lake order versus total nitrogen for
each individual watershed: (A) Muskegon, (B) Au Sable, and (C) Thunder Bay.
Solid lines show categories not significantly different using Tukey multiple means
comparisons, and dotted lines indicate non-contiguous categories that are not
significantly different.

40

Figure 6. Box-plots of L0 versus all lake response variables: (A) alkalinity, (B)
calcium, (C) magnesium, (D) chloride, (E) pH, (F) conductivity, (G) silica, (H)
nitrate, (I) sulfate, (J) Secchi, (K) water color, (L) DOC, (M) chlorophyll a, (N) TN,
(0) TP, and (P) TNzTP ratio. Solid lines show categories not significantly different
using Tukey multiple means comparisons, and dotted lines indicate non-
contiguous wtegories that are not significantly different. Some data points were
omitted for graphical purposes indicated by a- .

41

 

 

 

4.0

quL

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I I I. .I I I 500 l I l T I I 60 I I I I I f
A. AlkaIInrty B. ConductiVIty # _ C. Ca 0
30- - 400' E] 5° . é,
' [if] 40* 4)
5300' " _l
O
. r ‘ ‘ 323m '
20 ‘3200 é éfi] . ‘ 520* I
1.0' " 100- . - 10’ '
I I I I I I 0 I I l ___..l l l o I I I I I 0
20 I I I I I I 30r I I I. I I I 10 I I I I I I
0. M9 E. Chloride F. pH
_ . _ 9 ..
‘5 I? 420) . - [j
_.I
— - a) I -
2’10 E “8 Ea] $
10r $ [3 ~ _
5 b ‘ - 7 I
O .l_..._l 1 l ..l..__l. O l ‘ 1 6 I I I I I I
- I .l. I I I l 20- ' I I I ' ' 3“ I I I I I I
_ 6' Silica _ H. Nitrate "' l. Sulfate o
it - ~ - .
4- _ _ 150 420’ .
'6', _ _ BI _ a) o .
E o 3100 g l E ‘ $
)- —- ' _.
2 . 50-.Dé $8 10$ééé¢
¢ 9 l l l l L l
0 _‘_......_‘_ ° ° _ ........... ——:
10l_ stgccai ' ' ' _ 35_K'.vvatéchIo} ' q 3” L: DOC '
P . - :25— L - £200 ' -
_ _ o - _ g) 0
E5 - El ‘ O. 15 - 10_ -
: é : 5.- $ ‘ 0 $I Q I I
1 I I I I I 7 o l I l
-3-2012>=3 -3-2012>=3 -3-2012>=3

Lake Order Lake Order Lake Order

42

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

20 I I I I I I 2 “ I I I I I f 40 I I I I I I
M. Chlorophyll a N. TN l 0. TP
15' ° ‘ 1500’ ° ‘ ‘ 30f ‘
_, a ' ~ . ‘
010* . v 91000" “azof .
:3 ‘ 3
“P000 ‘ m’iaéfiifl‘ W W '
o I I I I G I l l l 1 ‘ 0 I I I I I I
“3’2012>=3 -3-2o12>=3
‘ I 1 I I I I Lake Order
150 P. TNzTP ratio Lake Order
100‘ o ‘
sor HT; é] —
0 _..i...‘ 4 ' 1
-3 -2 0 1 2 >=3
Lake Order
Figure6cont.

43

..

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I I I I I I 20 ' r j I ' ' I I I I I f
40 A. Max Depfl'l 8. Mean depth 150° 0. Lake area
. ' - 15' ‘ ‘ .
30 . 1000~ 'l
220: : E10” 2 * g
500r
10’ ‘ 5’ Q Q ‘
o I I I I I I o I I I I I I 0 if.¢ [:1 é
‘_.__._ -3-2012>=3
I I I I I I 5 I I I r I 7
1500' ° ° : 4' t
1000’ I 3' ‘
500’ a ‘ 2' E] g ‘
a 9 .0 Q B
o I I I I I I 1 4 I 4 I I
-3-2012>=3 -3-2012>=3
Lake Order Lake Order

Figure 7. Box-plots of L0 versus morphometry variables: (A) maximum depth,
(B) mean depth, (C) lake area, (D) lake basin slope, and (E) shoreline
development factor (SDF). Solid lines show categories not significantly different
using Tukey multiple means comparisons, and dotted lines indicate non-
contiguous categories that are not significantly different. Some data points were
omitted for graphical purposes indicated by 'C'.

A. Forested

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.5 I I I I Y I 0.5 I I T fir I j
0.4 r- , ‘ 0.4 ' 1
'6 33
”5 0.3 r ~ :5 0.3» a
.D .0
CE, 0.2 ~ 4 CE, 020 -
o o
‘— . l0 10 J
00 iii 0.0 [‘1'
B. Scrub-shrub
0.5 I I I ‘I I I 0.5 I I I I I I
0.4 - o - 0.4 . a
'5 5
"5' 0.3 - ° - E 0.3) a
.0 J '0 I
g 0.2 F ° 5 0.2 - . ~
53 8
0.1— Fl19 ~ 0.1— éé¢ 3 ~
0.0 '1‘ ‘ __ Eb 0.0 £3 é E.
C. All wetlands
0.7 I I I fi I I 0.7 I I I I r I
0.6" . 0.6- o -
'5 5
:5 0.4 - ’5- 0.4- . -
.o F .0
CE) 03* a g 03- -
o o
‘- U)
0.1 ” a 0.1 - g Q g ..
0.00—EL _ 0-0 ‘ '_-.._
-3 .2 o 1 2 >=3 -3 -2 o 1 2 >=3
Lake Order Lake Order

Figure 8. Box-plots of lake order versus proportion of wetland types in the 100m
and 500m area surrounding the study lakes: (A) forested wetlands, (B) scrub-
shrub wetlands, and (C) all wetland types. Solid lines show categories not
significantly different using Tukey multiple means comparisons, and dotted lines
indicate non-contiguous categories that are not significantly different. Note
differences in scale.

45

APPENDIX

46

Table 7. ANOVA results for selected landscape/physical features versus all lake
response variables. SDF and 100m scrub-shrub did not have any significant
results. Some variables required a 2-way ANOVA model (A) due to significant
interaction term (see methods for further explanation) . P-values and R2 are
provided and are bolded if significant (less than or equal to 0.01).
Positive/negative relationship with increasing landscape position (from high to
low in landscape) indicated by +/- signs. ”ns“ indicates not significant.

47

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

00 00.0 000.0 00 00.0 50.0 00 00.0 000.0 00 00.0 000.0 - who 000.0 0.00.902...
00 5.0 000.0 00 5.0 000.0 00 5.0 000.0 00 00.0 nmmd 00 00.0 50.0 020000003008.
00 «0.0 2.0.0 00 00.0 000.0 00 5.0 50.0 - 3.0 50.0 00 5.0 50.0 0000.00.38
00 00.0 09.0 + Nfio 50.0 00 5.0 50.0 00 00.0 000.0 00 < 00.0 50.0 0.3009200
3.383020
+ Nfio 50.0 00 5.0 05.0 + 9.0 000.0 + 5.0 000.0 00 00.0 09.0 000.
+ mfio 50.0 00 00.0 000.0 + 9.0 50.0 00 :0 5.0.0 00 No.0 00m.0 .o.00.0.0>>
00 00.0 000.0 00 < N00 000.0 00 00.0 000.0 00 5.0 03.0 00 00.0 50.0 .00000
3:05 .0003
00 00.0 05.0 00 00.0 05.0 00 00.0 000.0 + 3.0 000.0 + 3.0 50.0 0.0.30
00 00.0 000.0 00 0.0.0 000.0 00 5.0 000.0 00 00.0 000.0 00 00.0 50.0 0.00.2
+ 000 50.0 00 00.0 000.0 + < 00.0 000.0 + 00.0 50.0 00 5.0 50.0 00.:0
002 0>.~000m
0020005
00 00.0 N000 00 00.0 50.0 00 00.0 05.0 00 00.0 000.0 00 5.0 03.0 10
00 5.0 05.0 00 00.0 50.0 00 5.0 000.0 00 5.0 000.0 + 3.0 50.0 0000.00
00 00.0 09.0 00 00.0 50.0 00 00.0 05.0 00 9.0 2.0.0 00 5.0 500 0.2000005.
00 00.0 05.0 00 00.0 05.0 00 00.0 05.0 + 00.0 000.0 00 00.0 05.0 0.0.0.00
00 00.0 05.0 00 00.0 03.0 00 00.0 05.0 + 00.0 000.0 + 00.0 «00.0 0.2600000
00 5.0 000.0 00 00.0 09.0 00 00.0 05.0 + 5.0 50.0 + 9.0 0.00.0 3.5.00.2.
0m 0 mm. 0 mm. 0 0m 0 mm. 0 002
0003.? 0.000 0200-00.00 0.000 0003.? 0.09 00.00.00 0.09 00.0 000.. 0200300000
00.: 00030.5 B~0EO0QBE 0020005

 

 

0 203

48

Because some measures of lake morphometry and the presence and magnitude of
wetlands were significantly related to lake order, I examined whether these landscape
features alone could explain variation in lake response variables. I ran regressions
(ignoring watershed) and ANCOVA’s (with watershed as a covariate), using
landscape/physical features as the predictors in these new analyses, and lake water
chemistry/clarity variables were used as the response variables. I found that lake area and
the presence and magnitude of wetlands explained a significant amount of variation in
some dissolved conservative ions and dissolved reactive ions, but the explanatory power
of these direct relationships was far lower than of landscape position (Tables 3 and 7).
For example, the presence and magnitude of forested wetlands in the 100m buffer area
explained 33% of variation in calcium concentrations, whereas lake order explained 53%.
These results suggest the use of landscape position instead of directly modeling the
relationship between landscape features and these lake response variables. However, for
some productivity and water clarity variables, which were not related to landscape
position, I did find direct relationships to landscape features. In particular, DOC,

chlorophyll a and TN were all related to some measure of wetland connection (Table 7).

49

LITERATURE CITED

Anderson J .R., E.H. Harvey, J.T. Roach, and RE. Whitman. 1976. A land use sensor and
land cover classification system for use with remote sensor data geological
survey. Professional Paper 964. US Government Printing Office, Washington,
DC.

Bachmann, R.W. and DE. Canfield, Jr. 1996. Use of an alternative method for
monitoring total nitrogen concentrations in Florida lakes. Hydrobiologia 323: 1-8.

Beisner, B.E., C.L. Dent, and SR. Carpenter. 2003. Variability of lakes on the landscape:
roles of phosphorus, food webs, and dissolved organic carbon. Ecology 84: 1563-
1575.

Brezonik, P. L. 1996. Organizing paradigms for the study of inland aquatic ecosystems.
In: Freshwater Ecosystems: revitalizing educational programs in limnology. Ed.
Washington, DC: National Academy Press. p 181-202.

Burnham, K.P. and DR. Anderson. 2002. Model selection and multimodel inference (2nd
ed). Springer-Verlag, New York, NY.

Carpenter, S.R., JJ. Cole, J.F. Kitchell, and ML. Pace. 1998. Impact of dissolved organic
carbon, phosphorus, and grazing on phytoplankton biomass and production in
experimental lakes. Limnol. Oceanogr. 43: 73-80.

Cowardin, L.M., V. Caner, F.C. Golet, and ET. LaRoe. 1979. Classification of Wetlands
and Deepwater Habitats of the United States. US. Department of the Interior,
Fish and Wildlife Service, Washington, DC. FWS/OBS-79/3 I.

Crumpton, W.G., T.M. Isenhart, and PD. Mitchell. 1992. Nitrate and organic N analyses
with second-derivative spectroscopy. Limnol. Oceanogr. 37(4): 907-913

Detenbeck, N., C. Johnston, and G. Niemi. 1993. Wetland effects on lake water quality in
the Minneapolis/St. Paul metropolitan area. Landscape Ecology 8: 39-60.

Devito, K.J., I.F. Creed, R.L. Rothwell, and BE. Prepas. 2000. Landscape controls on
phosphorus loading to boreal lakes: implications for the potential impacts of
forest harvesting. Can. J. Fish. Aquat. Sci. 57: 1977-1984.

Dillon, PJ. and LA. Molot. 1997. Dissolved organic and inorganic carbon mass balances
in central Ontario lakes. Biogeochemistry. 36: 29-42.

Dodson, S. 1992. Predicting crustacean zooplankton species richness. . Limnol.
Oceanogr. 37(4): 848-856.

50

Fisher, S. G., J. Welter, J. Schade, and J. Henry. 2001. Landscape challenges to
ecosystem thinking: creative flood and drought in the American Southwest.
Scientia Marina 65(2): 181-192.

Frissell, C., W. Liss, C. Warren, and M. Hurley. 1986. A hierarchical framework for
stream habitat classification: viewing streams in a watershed context.
Environmental Management 10:199-214.

Fryer, G. 1985. Crustacean diversity in relation to the size of water bodies: Some
facts and problems. Freshwater Biology 15(3): 347-361.

Gergel S.E., M.G. Turner, and T.I(. Kratz. 1999. Dissolved organic carbon as an

indicator of the scale of watershed influence on lakes and rivers. Ecological
Applications 9(4): 1377-1390.

Hawkins, C., J. Kershner, P. Bisson, M. Bryant, L. Decker, S. Gregory, D. McCullough,
C. K. Overton, G. Reeves, R. Steedman, and M. Young. 1993. A hierarchical
approach to classifying stream habitat features. Fisheries 18:3-12.

Hynes, H.B.N. I975. The stream and its valley. Verh. Int. Ver. Theor. Angew. Limnol.
10: 1-15.

Kaplan, L.A., R.A. Larson, and TL. Bott. 1980. Patterns of dissolved organic carbon in
transport. Limnol. Oceanogr. 25: 1034-1043.

Kling, G. W., G.W. Kipphut, M.M. Miller, and WJ. O'Brien. 2000. Integration of lakes
and streams in a landscape perspective: the importance of material processin on
the spatial patterns and temporal coherence. Freshwater Biology 43: 477-497.

Kortelainen, P. 1993. Content of total organic carbon in Finnish lakes and its relationship
to catchment characteristics. Can. J. Fish. Aquat. Sci. 50: 1477-1483.

Kratz, T. K., K.E. Webster, CJ. Bowser, J.J. Magnuson, and BJ. Benson. 1997. The
influence of landscape position on lakes in northern Wisconsin. Freshwater
Biology 37: 209-217.

Lewis, D. B., and J .J . Magnuson. 2000. Landscape spatial patterns in freshwater snail
assemblages across Nothem Highland catchments. Freshwater Biology 43: 409-
420.

Likens, GE, and RH. Borrnann. 1979. The role of watershed and airshed in lake
metabolism. Arch. Hydrobiol. Beih. Ergebn. Limnol. 13: 195-211.

Liu, J ., and W.W. Taylor. 2002. Coupling landscape ecology with natural resource
management: paradigm shifts and new approaches. In: Integrating Landscape

51

Ecology into Natural Resource Management. J. Liu, and W.W. Taylor Ed. New
York, NY: Cambridge University Press. p 3-19.

Menzel, D. W., and Corwin, N. 1965. The measurement of total phosphorus in seawater
based on the liberation of organically bound fractions by persulfate oxidation.
Limnol. Oceanogr. 10: 280-282.

MIRIS (Michigan Resource Information System). 2000. [Online] Michigan Department
of Natural Resources. Available at:
http://www.ciesin.org/datasets/mirislcover/miriscov.html, 1 May, 2000.

Molot, L.A., and PI. Dillon. 1997. Colour-mass balances and colour-dissolved organic
carbon relationships in lakes and streams in central Ontario. Can. J. Fish. Aquat.
Sci. 54: 2789-2795.

Murphy, J ., and Riley, L. P. 1962. A modified single solution method for the
determination of phosphate in natural waters. Anal. Chim. Acta. 27: 31-36.

National Research Council. 1996. Freshwater Ecosystems: revitalizing educational
programs in limnology. Washington, DC, National Academy Press.

Nusch, EA. 1980. Comparison of different methods for chlorophyll and phaeopigment
determination. Ergeb. Limnol. 14: 14-36.

Numberg, GK. 1995. Quantifying anoxia in lakes. Limnol. Oceanogr. 40: 1100-1111.

Omemik, J .M. and A.J. Kinney. 1983 An improved technique for estimating mean depth
of lakes. Water Research. 17(11): 1603-1607.

Palmer, D. W. 2002. Scale detection using semivariograms and autocorrelograms. In:
Learning landscape ecology. A practical guide to concepts and techniques. S. F.
Gergel, M.G. Turner Ed. New York, NY: Springer-Verlag. p

Prepas, E.E., D. Planas, J .J . Gibson, D.H. Vitt, T.D. Prowse, W.P. Dinsmore, L.A.
Halsey, P.M. McEachem, S. Paquet, GJ. Scrimgeour, W.M. Tonn, C.A.
Paszkowski, and K. Wolfstein. 2001. Landscape variables influencing nutrients
and phytoplankton communities in Boreal Plain lakes of northern Alberta: a
comparison of wetland- and upland-dominated catchments. Can. J. Fish. Aquat.
Sci. 58: 1286-1299.

Quinlan, R., A.M. Paterson, R]. Hall, P.J. Dillon, A.N. Wilkinson, B.F. Cumming, M.V.
Douglas, and J .P. Smol. 2003. A landscape approach to examining spatial
patterns of limnological variables and long-term environmental change in a
southern Canadian lake district. Freshwater Biology 48: 1676-1697.

52

Rabeni, C. F., and SP. Sowa. 2002. A landscape approach to managing the biota of
streams. In: Integrating Landscape Ecology into Natural Resource Management.
J. Liu, and W.W. Taylor Ed. New York, NY: Cambridge University Press. p 114-
142.

Reed-Andersen, T., E.M. Bennett, B.S. Jorgensen, G. Lauster, D.B. Lewis, D. Nowacek,
J.L. Reira, B.L. Sanderson, and R. Stedman. 2000. Distribution of recreational

boating across lakes: do landscape variables affect recreational use? Freshwater
Biology 43: 439-448.

Riera, J. L., J .J . Magnuson, T.K. Kratz, and KB. Webster. 2000. A geomorphic template
for the analysis of lake districts applied to the Northern Highland Lake District,
Wisconsin, USA. Freshwater Biology 43: 301-318.

Sartory, DP. and J .U. Grobbelaar. 1984. Extraction of chlorophyll a from freshwater
phytoplankton for spectrophotometric analysis. Hydrobiologia 114: 177-187.

Schiff, S.L., R. Aravena, S.E. Trumbore, M.J. Hinton, R. Elgood, and P.J. Dillon. 1997.
Export of DOC from forested catchments on the Precambrian Shield of Central
Ontario: Clues from 13C and 14C. Biogeochemistry 36: 43-65.

Seelbach, P.W., M.J. Wiley, J .C. Kotanchik, and ME. Baker. 1997. A landscape-based
ecological classification system for river valley segments in Lower Michigan (MI-
VSEC version 1.0). Michigan Department of Natural Resources Fisheries
Division Fisheries Research Report 2036. 51 pg.

Soranno, P. A., K.E. Webster, J .L. Reira, T.K. Kratz, J .S. Baron, P.A. Bukaveckas, G.W.
Kling, D.S. White, N. Caine, R.C. Lathrop, and PR. Leavitt. 1999. Spatial
variation among lakes within landscapes: ecological organization along lake
chains. Ecosystems 2: 395-410.

Vannote, R. L., G.W. Minshall, K.W. Cummins, J.R. Sedell, and CE. Cushing. 1980.
The river continuum concept. Canadian Journal of Fisheries and Aquatic Sciences
37: 130-137.

Ward, J. V., and J .A. Stanford. 1983. The serial discontinuity concept of lotic
ecosystems. 29-42.

Webster, K. E., P.A. Soranno, S.B. Baines, T.K. Kratz, C.J. Bowser, P.J. Dillon, P.
Campbell, E.J. Fee, and RE. Hecky. 2000. Structuring features of lake districts:
landscape controls on lake chemical responses to drought. Freshwater Biology 43:
499-515.

Wetzel, R.G., and GE. Likens. 2000. Limnological Analyses, third edition. Springer-
Verlag New York Inc.

53

Wiens, J. A., B.Van Horne, and ER. Noon. 2002. Integrating landscape structure and
scale into natural resource management. In: Integrating Landscape Ecology into
Natural Resource Management. J. Liu, and W.W. Taylor Ed. New York, NY:
Cambridge University Press. p 23-67.

Xenopoulos, M.A., D.M. Lodge, J. Frentress, T.A. Kreps, S.D. Bridgham, E. Grossman,
and C]. Jackson. 2003. Regional comparisons of watershed determinant of
dissolved organic carbon on temperate lakes from the Upper Great Lakes region
and selected regions globally. Limnol. Oceanogr. 48: 2321-2334.

54

 

11111111311111th111111311