QUANTIFYING THE RELATIONSHIPS BETWEEN LAKE NUTRIENTS AND
AGRICULTURAL LAND USE/COVER AT DIFFERENT SPATIAL SCALES
By
Katie L. Droscha

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Fisheries and Wildlife
2012

ABSTRACT
QUANTIFYING THE RELATIONSHIPS BETWEEN LAKE NUTRIENTS AND
AGRICULTURAL LAND USE/COVER AT DIFFERENT SPATIAL SCALES
By
Katie L. Droscha

I quantified agricultural land use/cover using five spatial metrics to determine which
spatial metric was most strongly correlated with seasonal lake total phosphorus and total nitrogen
concentrations in 204 Michigan lakes. I used two distance based spatial metrics, two flow
direction spatial metrics and full lake catchments. Agricultural land use/cover was categorized
as pasture, forage, row crop and total agriculture. My research questions were 1) which
agricultural type and spatial metric is the most strongly correlated with lake nutrient
concentrations, 2) How similar are the spatial metrics that quantify agricultural land use/cover in
lake catchments, 3) Do the different spatial metrics each contribute to explaining lake nutrients,
4) Do different agricultural types have similar effects on lake nutrients. I used correlation
coefficients to quantify the relationship between agricultural types, spatial metrics and seasonal
lake TP and TN. Among agricultural types, the catchment, moderate-highly contributing area
and 100 m lake-stream buffer spatial metrics were strongly correlated with each other, meaning
they measured similar aspects of the landscape. Row crop and total agriculture had the strongest
correlations for spring and summer lake nutrients, however, linear regression analysis within the
100 m lake-stream buffer reveal that for a change in one unit of pasture or forage, there is a
larger change in lake nutrients when compared to row crop and total agriculture. Further
research is needed to more fully quantify the effects of pasture and forage lands on lake
nutrients.

ACKNOWLEDGEMENTS
I would like to thank my committee members Dr. Lois Wolfson and Dr. A. Pouyan

Nejadhashemi for their constructive comments and suggestions for the direction of my research. I
also would like to acknowledge Sarah AcMoody and Dr. Dave Lusch of the Remote Sensing & GIS
Research and Outreach Services (RS&GIS) for passing on their statistical knowledge, land use/cover
expertise and GIS resource. I would like to acknowledge the MSU limnology lab members and
friends Stacie Auvenshine, Joshua Booker, Geoff Horst, Andrea Jager-Miehls, Dianna Miller, Emily
Norton, Kim Peters, Jeff White, Heidi Ziegenmeyer, Paul Bordeau, Shikha Singe, Cara Vealey and
Lia Lilley for their support, love and suggestions. I would like to thank my family members Denise
Droscha, James Droscha, Jeremy Droscha, Jessica Droscha and Mason Droscha for their love and
support. I also wish to acknowledge my fiancé Andrew Rodgers for his patience, love and
assistance. Last but not least I would like to acknowledge my advisor, Dr. Patricia A. Soranno, who
has been a wonderful mentor preparing me both academically and professionally for a career in
limnology and agricultural analysis; I have learned so much through your patience, expertise and
mentoring style, and I look forward to a future where I can assist others in my community.

iii

TABLES OF CONTENTS

LIST OF TABLES……………………………………………………………………………………..v
LIST OF FIGURES…………………………………………………………………….......................vi

CHAPTER 1: QUANTIFYING THE RELATIONSHIPS BETWEEN LAKE NUTRIENTS
AND AGRICULTURAL LAND USE/COVER AT DIFFERENT SPATIAL SCALE………….1
Introduction…………………………………………………………………………………....1
Methods………………………………………………………………………….....................5
Results………………………………………………………………………………………..11
Discussion……………………………………………………………………………………15
Conclusions…………………………………………………………………………………..21
REFERENCES……………………………………………………………………………………….39

iv

LIST OF TABLES

Table 1. Summary statistics of the lake variables. TP is total phosphorus, TN is total nitrogen.
N= 204 lakes……………………………………………………………………………………..23
Table 2. Summary statistics of the percentage of agricultural types for each spatial metric. N =
204 lakes…………………………………………………………………………………………24
Table 3a. Correlations between spring lake total phosphorus (TP) or total nitrogen (TN)
concentrations and the different spatial metrics. Numbers in bold are either the highest
correlation value, or within 0.05 of the highest value within the respective column. All values in
the tables are significant at α ≤ 0.05. N= 204 lakes……………………………………………...25
Table 3b. Correlations between summer lake total phosphorus (TP) or total nitrogen (TN)
concentrations and the different spatial metrics. Numbers in bold are either the highest
correlation value, or within 0.05 of the highest value within the respective column. All values in
the tables are significant at α ≤ 0.05. N= 204 lakes……………………………………………...26
Table 4. Correlations between different agriculture types within each spatial metrics. All values
in the tables are significant at α ≤ 0.05, ‘n.s.’ is not significant. N= 204 lakes. Note that the value
in the columns are measured at the same spatial extent as the value in the row………………...27
Table 5. Correlations between different spatial metrics within each agricultural type. All values
in the table are significant at α=0.05. N= 204. ‘Mod.- high.’ is moderate-highly. Note that the
value in the columns are measured for the same agricultural type as the value in the row……...28

v

LIST OF FIGURES

Figure 1. Distribution of the 204 lakes used in this study. All types of agricultural land
use/cover are shown in grey……………………………………………………………………...29
Figure 2. Four different study lakes displaying the five spatial extents used to measure
agricultural land use/cover within lake catchments. Column A) equidistant buffers around either
lakes and streams, or lakes only, and column B) contributing areas estimated at highly
contributing or moderate-highly contributing areas, as well as the full catchment area. ‘C.A.’ is
contributing area………………………………………………………………………………... 30
Figure 3. Plots of agriculture measured in the 100 m lake-stream buffer (y axis) versus the
catchment for A) pasture, B) forage , C) row crop, and D) total agriculture (x axis)……….......31
Figure 4. Plots of total agriculture measured in the 100 m lake-stream buffer (y axis) versus the
residuals of total agriculture measured in the catchment (x axis) for lake nutrient
concentrations................................................................................................................................32
Figure 5. Plots of agriculture measured in the moderate-highly contributing area (y axis) versus
the 100 m lake-stream buffer for A) pasture B) forage, C) row crop, and D) total agriculture (x
axis). ‘C.A.’ is contributing area…………………………………………………………………33
Figure 6. Plots of total agriculture measured in the 100 m lake-stream buffer (y axis) versus the
residuals of total agriculture measured in the moderate-highly contributing area (x axis) for lake
nutrient concentrations…………………………………………………………………………...34
Figure 7. Plots of agriculture measured in the moderate-highly contributing area (y axis) versus
the catchment for A) pasture , B) forage, C) row crop, and D) total agriculture (x axis). ‘C.A.’ is
contributing area…………………………………………………………………………………35
Figure 8. Plots of total agriculture measured in the moderate-highly contributing area (y axis)
versus the residuals of total agriculture measured in the catchment (x axis) for lake nutrient
concentrations. ‘C.A.’ is contributing area………………………………………………………36
Figure 9. Linear regressions of spring TP (y axis) versus percent agricultural type measured in
the 100m lake-stream buffer for A) pasture, B) forage, C) row crop and D) total agriculture (x
axis)………………………………………………………………………………………………37
Figure 10. Linear regressions of spring TN (y axis) versus percent agricultural type measured in
the 100 m lake-stream buffer for A) pasture, B) forage, C) row crop and D) total agriculture (x
axis).……………………………………………………………………………………………...38

vi

CHAPTER 1: QUANTIFYING THE RELATIONSHIPS BETWEEN LAKE
NUTRIENTS AND AGRICULTURAL LAND USE/COVER AT DIFFERENT SPATIAL
SCALES

Introduction
The effects of agriculture on lakes and streams has been well recognized for some time,
and has been the focus of recent research (Arbuckle and Downing 2001, Buck et al 2004,
Egerston and Downing 2004, Egerston et al 2004, Bechmann et al 2005, Bishop et al 2005,
Lehmann et al 2005, Kroger et al 2007, Diebel et al 2008, Hoorman et al 2008, Gémesi et al
2011). Surface runoff from agricultural land is rich in nitrogen and phosphorus, and has been
identified as the leading cause of eutrophication in lakes, reservoirs and streams (Carpenter et al
1998, Arbuckle and Downing 2001, Vanni et al 2005), resulting in excessive growth of aquatic
macrophytes, algal blooms, anoxic water conditions, and undesirable fish taxa (Carpenter et al
1998, Egerston and Downing 2004, Egertson et al 2004, Vanni et al 2005). Agricultural
practices are diverse, and vary by region (Vanni et al 2001, Ekholm et al 2005), however, even
within regions of similar agricultural land use/cover, researchers have found high variation in
surface water nutrient concentrations (Kirchner 1975, Nilsson and Hakanson 1992). There are
two challenges in quantifying the effects of agriculture on surface water nutrients that likely lead
to such high variation in agricultural regions. The first is that there are different types of
agriculture that may influence surface waters in different ways. The second is that agricultural
land use/cover can be quantified at many different spatial extents. For both of the above
challenges, it is not clear which of the above factors is most related to downstream surface water
nutrients.

1

In most studies, different types of agricultural land use/cover are lumped into a single
category (Norvell et al 1979, Omernik et al 1981, Detenbeck et al 1993, Hunsaker and Levine
1995, Jones et al 2004, Gergel 2005, Baker et al 2006, Fraterrigo and Downing 2008, Zampella
and Procopio 2009). However, different agricultural types have very different effects on lake
and stream nutrient concentrations.
Row crops are a high-cost agricultural practice that inverts the soil and buries any
vegetative residue to decay below the soil surface (Stephens 1998). Row crops are a highly
researched cause of non-point source pollution due to the amounts of nutrients and fertilizers
washed into streams and agricultural ditches (Bechman et al 2005, Jones and Knowlton 2005,
Zaimes et al 2008). The disturbance of topsoil, repeated fertilizer applications and seasonal
precipitation can result in elevated levels of nutrients and suspended solids within freshwater
ecosystems, and catchments with high percentages of row crops have been noted to have high
N:P ratios within streams and lakes (Arbuckle and Downing 2001, Zedler 2003, Bechman et al
2005, Jones and Knowlton 2005, Vanni et al 2005, Renwick et al 2008).
Animals grazed on pasture do not involve the increased levels of soil disturbance and
fertilizer applications associated with row crops. However, grazing can cause adverse effects to
freshwater ecosystems, as livestock are attracted to riparian areas and when unregulated, can
result in stream bank erosion and sediment loss due to wallowing and foraging (Zaimes et al
2008). The waste produced by animals results in low N:P concentrations within streams, and is a
source of concern for groundwater contamination and the threat of fecal bacteria entering
drinking water (Arbuckle and Downing 2001, Monaghan et al 2008).
In comparison to row crops and pastureland, forage crops are vegetation such as hay,
grain or legumes grown as a fodder provided to domestic animals (Lewis 2002). Fields that

2

grow forage such as alfalfa or grass have soil remain undisturbed for several years, as the crops
are harvested from year to year with no need to replant seed between harvests (Lipton 1995).
Harvested forage crops may be fed to animals fresh, dried or cured after ensiling (Lipton 1995,
Stephens 1998). Forage crops do not contribute the same amount of soil disturbance as row
crops, but require nitrogen fertilizers for increased production (Malhi et al 2004). Different
agricultural types can result in different disturbance to surface water. Defining specific
agricultural types may help to explain nutrient concentration variations between water-bodies
with similar land use/cover (Nerbonne and Vondraek 2001, Vanni et al 2001, Ekholm et al
2005).
Agricultural land use/cover within close proximity to lakes and streams have been the
focus of management studies due to concerns over erosion, nutrient loading and water quality
(Dieble et al 2009, Zampella and Procopio 2009). However, many studies have found that total
catchment was more relevant to quantify agricultural land use/cover and predict nutrient
concentrations than buffers surrounding riparian zones or land use/cover measured near stream
channels (Hunsaker and Levine 1995, Jones et al 2004, Meynendonchx et al 2006, Fraterrigo and
Downing 2008, Taranu and Gregory-Eaves 2008). Hunsaker and Levine (1995) found that
catchments were better at predicting nutrient concentrations than land use/cover measured in 200
m and 400 m buffers on either side of the study streams. Jones et al (2004) found the catchment
scale to be the most appropriate when considering the influence of hydrology and topography on
reservoir nutrient concentrations, while buffers surrounding riparian areas were unable to capture
the range of variables and conditions measured in the study. Similarly, Meynendonckx et al
(2006) found that stream nutrient concentrations were largely regulated by drainage
characteristics and hydrological pathways of the catchment, and that land use/cover measured

3

close to the stream channel was not a better predictor of nitrate or phosphate concentrations than
land use/cover measured at farther distances away from the stream channels.
Other studies have shown the importance of land use/cover measured near surface water,
as nutrients respond differently to land use/cover measured at different scales due to variations in
nutrient mobility; areas closer to water tend to contribute more to surface transport of nutrients
than areas farther away (Soranno et al. 1996, Johnson et al 1997, Buck et al 2004, Diebel et al
2009). In comparison to catchment studies, narrow riparian buffers have been found to be the
strongest predictor of sediments than more extensive scales (Diebel et al 2009). Johnson et al
(1997) determined that land use/cover measured within the100 m buffer strip explained
variability in several chemical parameters better than whole catchment data, though noted that
ecotonal areas may simply reflect dominant catchment land uses. Soranno et al (1996) found
that agricultural land within 100 m of open water contributes directly to loading, and depending
upon topography and runoff conditions, the greatest contribution of loading to streams came
from a heterogeneous riparian corridor measuring between 0.1-6 km.
Some researchers have argued that there is no optimal scale to measure the effects of land
use/cover on lakes and streams (Strayer et al 2003). Some studies showed that land use/cover
values quantified at the catchment scale were strongly correlated to land use/cover values
quantified at riparian scales (Arscott et al 2006, Baker et al 2006) and demonstrated that it would
be difficult to separate the relative influence of near and far land uses on water quality (Zampella
and Procopio 2009). Other studies have shown that configuration. i.e. the spatial aggregation of
agricultural land use/cover is more important than catchment composition or measures of
proximity (Gémesi et al 2011).

4

Uncertainty about the accuracy of distance-based spatial metrics has led to the use of
empirical modeling and the development of more complex spatial metrics. Researchers combine
agricultural land use/cover proximity with catchment dynamics to simulate contributing areas
(Jones et al 2001, Gergel 2005, Fraterrigo and Downing 2008). Contributing area spatial metrics
are developed by combining topography, hydrology and catchment variables to simulate
hydrogeomorphology, and are then used to identify areas within a catchment that have the
highest likelihood of contributing sediments and nutrients to surface water (Erskine et al 2006,
Van Sickle and Johnson 2008).
My research builds on the results of previous studies that have used spatial metrics to
quantify agricultural land use/cover, and that have defined different agricultural types and
practices when predicting surface water nutrients (Jones et al 2001, Strayer et al 2003, Jones et al
2004, Buck et al 2004, Fraterrigo and Downing 2008). In this study, I compared different ways
to measure spatial extent (e.g., distance-based spatial metrics, contributing-area spatial metrics
and whole catchment measurements), and different agricultural types (e.g., row crop, forage,
pasture, and total sum of all agricultural types) for predicting lake nutrients. My research
questions are: 1) Which agricultural type and spatial metric is the most strongly correlated with
lake nutrient concentrations, 2) How similar are the spatial metrics that quantify agricultural land
use/cover in lake catchments, 3) Do the different spatial metrics each contribute to explaining
lake nutrients, 4) Do different agricultural types have similar effects on lake nutrients
Methods
Study Region
The study area is the state of Michigan, USA (Figure 1). Michigan has three major rock
types created by glacial deposits and magma: igneous rock, sediment rock, and metamorphic
5

rock (Brandt 2008). Bedrock geology differs by region and is covered by unconsolidated
sediments such as gravel, sand, silt and silt. The bedrock of the Lower Peninsula and the eastern
part of the Upper Peninsula is made of sediment rock, while the bedrock of the western part of
the Upper Peninsula is made of igneous and metamorphic rock (Brandt 2009).
Michigan’s state soil is Kalkaska sand (Schaetzl 2009). However, due to glacial deposits
and various sediments, there is a wide variety of parent materials. The soils reflect their parent
material, resulting in complex soils profiles within close proximity throughout Michigan
(Schaetzl 2009). There are typically five major soil types: clay rich soils, wet sands, organic
soils, loamy wet soils and sandy spodsols (Schaetzl 2009).
Michigan has 62 major watersheds with numerous inland and coastal wetlands (Wolfson
2009). There are 87,389 km of perennial streams and rivers that flow continuously due to baseflow. Eighty percent of the annual stream flow in the Lower Peninsula comes from base-flow.
The base-flow provides uniform flow and relatively stable temperatures throughout the year
(Wolfson 2009). Rivers maintain annual flow due to base-flow in the Upper Peninsula and the
north western part of the Lower Peninsula as well. In comparison, rivers in southern part of the
Lower Peninsula and thumb area of the state have lower annual flow and variable discharges due
to overland flow and storm events (Wolfson 2009).
Michigan has more than 11,000 lakes > 2 ha in size, with 1000 lakes > 40 ha (Wolfson
2009, Fuller and Minnerick 2008). Inland lakes were formed by glacial processes, resulting in
four basic lakes types: scour channel lakes formed by pressurized subglacial water creating small
crescent filled basins; deep kettle lakes with irregularly shaped outlines and bathymetry; riverine
lakes with wide floodplains; and horseshoe or oxbow lakes formed by the erosion of tributary

6

river channels. A rare additional type are karst lakes that are found in the northern part of the
state, and are made from basins dissolved from highly soluble limestone rock (Wolfson 2009).
Michigan surface water quality varies depending on watershed characteristics; basins,
geology, soils, land use/cover and human population centers (Fuller and Minnerick 2008,
Woflson 2009). Inland lakes within the Upper Peninsula and northern Lower Peninsula have
good to excellent quality. In comparison, the water quality in the central and southern portions
of the Lower Peninsula range from good to poor water quality, depending upon the percentage of
agriculture and urban land use/cover measured in the surrounding catchment (Wolfson 2009).
Land use/cover varies throughout Michigan. The majority of agricultural land use/cover
is located in the Lower Peninsula, in the middle and southwestern part of the state (Figure 1).
Lakes located within the Lower Peninsula were more likely to be in close proximity to
agricultural land use/cover than lakes located within the Upper Peninsula (Figure 1). The state of
Michigan is an ideal location for examining research questions about agriculturally-based lake
nutrient variation due to the diversity of agricultural types, the range of agricultural cover from
low to high, and the high density of lakes.
Selecting study lakes
Study lakes were selected from a dataset of 364 inland lakes >14 ha, that were monitored
under the Lake Water Quality Assessment Program (LWQA) (Fuller and Minnerick 2008).
These lakes were sampled twice a year, spring and summer (epilimnion), from discrete water
samples collected at different depths and within multiple basins from 2001-2008 (further
sampling details are provided in Fuller and Minnerick 2008). Of these 364 lakes, I selected those
that had at least 1 % of their catchment in total agricultural cover, and that had no more than 25%
of the 500 m buffer in urban land use/cover. The resulting dataset contained 204 lakes located in

7

both the Lower and Upper Peninsulas (Table 1, Figure 2). Lakes used for analysis had surface
area between 14 ha – 4,200 ha, and mean depth between 0.8 m – 20.5 m (Table 1, Figure 2).
These selection criteria were used to first, ensure that at least some agriculture occurred in the
catchment, to avoid excessive data points with zero agricultural land use/cover; and second, to
remove the potentially large effects of urbanization that may mask patterns with agricultural land
cover. We used the 500 m buffer to quantify urban land use/cover due to the amount of urban
and suburban land use/cover concentrated in the areas around lakes (Walsh et al. 2000), thus
providing more information about urban land use/cover than measured in a catchment-wide
extent.
Agricultural land use/cover
I used the 2007 USDA Agricultural Statistics (NASS) Crop Data Layer (CDL) to
quantify land use/cover. The Remote Sensing & GIS Research and Outreach Services
(RS&GIS) assisted converting the CDL raster data (56 m X 56 m) with 72 land use/cover
categories into a vector format with 13 main categories based off of defined uses (urban,
agricultural, forest, etc.). Agricultural land use/cover types within the CDL were categorized
using general use, planting and harvest practices into four classes; pasture, forage, row crop and
total agriculture (all agricultural types combined) (Stephens 1998, Lewis 2002, House 2006,
Zaimes et al 2008).
Distance based spatial metrics
Buffers are commonly used for water quality spatial analysis in relation to land use/cover,
but it is not always clear what the optimal width should be (Stephenson and Morin 2009). I
chose two different buffer distances to test the relationship between agricultural land use/cover
and lake nutrient concentrations (Table 2, Figure 1). First, I created a 500 m equidistant buffer

8

around lake shorelines. Second, I created a 100 m equidistant buffer around streams that flow
into the lake and the lake shoreline itself. I used the National Hydrology Dataset (NHD) at
1:24,000 resolution for the stream network and a lake polygon layer from the Michigan
Department of Natural Resources (MDNR) for the lake shorelines.
Flow direction spatial metrics
Two different spatial metrics were created using flow path estimates using the flow direction D8
algorithm (Erskine et al 2006) and the USGS National Elevation Dataset (NED) at 30 meter
resolution. Flow direction was calculated using the ESRI Spatial Analyst Hydrology tools within
ArcGIS to determine the flow of water to the lowest lying areas within a lake catchment. I
selected parameters to create two different spatial metrics, which I will refer to as contributing
areas, or those areas that can potentially produce runoff to the location (i.e., a grid cell) (Erskine
et al 2006) to simulate hydrogeomorphology within the catchment. One of the metrics
represented highly contributing areas, which included those cells that have a flow accumulation
greater or equal to 100,000 pixels within the grid; those cells that were connected to upstream
cells or had the highest number of cells that drained into a singular cell, i.e. the lowest –lying
areas within the grid and so have the potential to be focal areas of nutrient transport. The second
metric represented moderate-highly contributing areas, which included the same cells in the
highly contributing area that have a flow accumulation greater or equal to 100,000 pixels within
the grid, and also the additional cells that had flow accumulation greater or equal 75,000 pixels
within the grid, or moderate amounts of upstream cells flowing into them. The parameters were
chosen by visual inspection of cell contributing area maps to arrive at contrasting sizes of
contributing areas. The highly contributing areas are the lowest lying areas within lake
catchments, and often closely follow the stream flow-lines (Table 2, Figure 1). In comparison,

9

the moderate-highly contributing area captures additional land use/cover that extends beyond
streams and the highly contributing area, and includes more areas that may contain ephemeral
stream beds and drainage lines on the land (Table 2, Figure 1).
Catchment delineations and metric
Catchments were delineated using existing watershed features created for stream
catchments from the Michigan Department of Natural Resources (MDNR) and hand-digitizing
the coverage for modifications needed for lake catchments using the USGS Digital Raster
Graphic Mosaic (DRG) coverage. The NHD stream coverage was used to determine connected
streams and catchments flowing into lakes. Lake catchments were defined as all land and
streams draining into the lake with upstream lakes < 10 ha in area. If a lake > 10 ha was
upstream and connected to the downstream lake, then the downstream lake catchment did not
include the catchment of the upstream 10 ha lake. This criterion was established to attempt to
create ‘local’ catchments that included the area of land and streams that drained into a lake and
was not unduly influenced by upstream lakes.
Statistical analysis
I calculated Pearson correlation coefficients (significance α < 0.05), between the four
agricultural types and spring and summer TP and TN to compare the relationships between
nutrients, different spatial extents and agricultural types. I then examined whether the
agricultural types and metrics themselves were correlated with each other by calculating Pearson
correlation coefficients among the spatial metrics and the agricultural types themselves.
For those spatial metrics that were strongly correlated with each other and that were also
strongly correlated to lake nutrients, I conducted a regression residual analysis to determine if the
two different spatial extents provided new predictive capability or were equally providing the

10

same information about agricultural cover. For this analysis, I ran regressions among each of the
nutrients and the spatial scales that were strongly correlated and calculated the residuals from the
regression. I regressed the residuals against the agricultural variable that was also strongly
correlated to a lake nutrient to see if additional variation in lake nutrients as explained by the
other agricultural variable. Finally, to examine whether the slopes between seasonal nutrients
and agricultural types were similar, I conducted a linear regression analysis between total
agriculture, row crop, forage and pasture measured within the 100 m lake-stream buffer with
spring TP and TN.

Results
Study lakes
The study lakes are widely distributed throughout Michigan (Figure 2) and they had a
wide range in water chemistry, surface area, catchment area and lake depth (Table 1). Spring TP
concentrations ranged from 1 to 176 µg L

-1

and summer TP concentrations ranged from 1 to 290

-1

-1

µg L (Table 1). Spring TN concentrations ranged from 146-7,800 µg L , and summer TN
-1

concentrations ranged from 90-3,180 µg L (Table 1). However, the median values for spring
and summer nutrient concentrations were at the lower end of the concentration range.
Agricultural land use/cover around the study lakes
The percentage of agricultural cover was highly variable as measured by the spatial metrics.
The 100 m lake-stream buffer had the lowest maximum percentages measured for the
agricultural types except for forage, which had a similar range for all of the spatial metrics. In
comparison, the highly contributing area had the highest maximum percentage measured for the

11

agricultural types, except for forage, which tied with forage within the moderate-highly
contributing area (Table 2). The median percentage value for row crop measured in the
catchment was <6% for the study lakes. However, some of the study lakes had very high
percentages of row crop within the catchment (Table 2). The pattern of high maximum values
for agriculture percentages with low median values were repeated for all of the spatial metrics
(Table 2).
Relationships between metrics of agriculture land use/cover and lake nutrients
Correlation strengths for TP and TN concentrations varied for the spatial metrics,
depending on season and agricultural type. The 100 m lake-stream buffer, moderate-highly
contributing area, and catchment were strongly correlated to Spring TP and TN for all three
agricultural types and total agriculture (Table 3a). Additionally, for spring TN, strong
correlation strengths were found for row crop and total agriculture (Table 3a).
Relationships between agricultural land use/cover types
The strongest correlations among total agriculture and the three agricultural types were
between total agriculture and row crop (r ≥ 0.90) (Table 4). Total agriculture had moderate to
strong correlations with forage and pasture for all of the spatial metrics, except the moderatehighly contributing area and catchment metrics. The weakest correlations were between pasture,
forage and row crop for the highly contributing area and moderate-highly contributing area (r <
0.30) (Table 4).
Relationships among spatial metrics within individual agricultural types
The strongest correlations were between the catchment, moderate-highly contributing
area and highly contributing area (Table 5). The weakest correlations were between the 100 m
lake-stream buffer and 500 m lake buffer (Table 5). The range in spatial metric correlation

12

strengths and the results from Table 3a and 3b indicate that differences may result from the wide
range of values across the datasets for local agricultural land use/cover (Table 1, Table 3a, Table
3b). Next, I examined whether the catchment metrics contribute the same information to the
relationship with lake nutrients, or if some of the spatial metrics are providing additional
information related to land use/cover that helps to further explain lake nutrient concentrations. I
explored such relationships between two sets of spatial metrics below.
100 m lake-stream buffer vs. catchment: The 100 m lake-stream buffer and catchment
were strongly correlated to both spring and summer lake nutrients (Tables 3a and 3b). There was
a strong positive correlation between the 100 m lake-stream buffer and catchment spatial metrics
for total agriculture and row crop (Table 5, Figure 3). Agricultural percent cover was higher in
the catchment compared to the 100 m lake-stream buffer (Figure 3), suggesting that although the
two metrics were correlated, the distribution of agricultural land use/cover in lake catchments is
not uniform, and agricultural land use/cover within the 100 m lake-stream buffer is less common
than in the rest of the catchment. To attempt to separate the possible effects of the two spatial
metrics, I first ran regressions of lake nutrients against the catchment spatial metric; then I
regressed the residuals from this model on lake nutrients. The fact that this latter correlation was
significant for spring and summer TP, and spring TN shows that the 100 m lake-stream buffer
helps to explain additional variation in lake nutrients (Figure 4).
100 m lake-stream buffer vs. moderate-highly contributing area: The 100 m lake-stream
buffer and moderate-highly contributing area were weakly correlated with spring and summer
TP, and moderately correlated with spring and summer TN (Tables 3a and 3b). There was a
strong positive correlation between both spatial metrics for total agriculture and row crop (Table
5, Figure 5). Agricultural land use/cover was higher overall in the moderate-highly contributing

13

area compared to the 100 m lake-stream buffer (Figure 5). I found that the 100 m lake-stream
buffer gives information in addition to that which was explained by the moderate-highly
contributing area. The correlation between the moderate-highly contributing area residuals and
the 100 m lake-stream buffer was significant for both TP and TN for both seasons (range in r
from 0.15 to 0.30) (Figure 6).
Catchment vs. moderate-highly contributing area: Both spatial metrics have similar
relationships with the 100 m lake-stream buffer, and were moderately correlated with seasonal
lake nutrient concentrations (Tables 3a and 3b). There was a strong positive correlation between
the catchment and moderate-highly contributing area for total agriculture and row crop (Table 5,
Figure 7). Percentages of total agriculture and the three agricultural types were higher in the
catchment compared to the moderate-highly contributing area (Figure 7). The very strong
correlation between the two spatial metrics suggests that they capture the same amount and
spatial pattern of land use/cover, or that within both spatial metrics the measured spatial patterns
are very similar. The lack of a significant correlation between the catchment residuals with the
moderate-highly contributing area suggests that this metric is not providing additional
information beyond the catchment (Figure 8).
Linear regressions between spring nutrients and agricultural types: I compared the slopes
of the relationships between agriculture cover and land nutrients by examining spring nutrients
and the 100 m lake-stream buffer only. I found that row crop and total agriculture had similar
slopes for spring TP and TN, whereas pasture and forage had larger slopes than row crop and
total agriculture. Thus, there was a larger increase in TP and TN for forage or pasture per unit
increase in cover compared to row crop and total agriculture (Figures 9, 10).

14

Comparisons across seasons: Correlation strengths between agricultural types and
nutrient concentrations varied for the spatial metrics depending upon season (Tables 3a, 3b).
The 100 m lake-stream buffer, highly contributing area, moderate-highly contributing area, and
catchment had relatively strong correlations ( r > 0.50) with spring TN for row crop and total
agriculture, but were moderate to weakly correlated with pasture or forage (Table 3a). The 100
m lake-stream buffer, moderate-highly contributing area and catchment had low correlations ( r >
0.30) with spring TP for forage, row crop and total agriculture, but had overall moderate to weak
correlations (r < 0.50) for spring TN (Table 3a). Correlation strengths were moderate to weak (r
< 0.50) between all five spatial metrics and agricultural types for summer TP and TN (Table 3b).
The strongest correlations were between the 100 m lake-stream buffer, highly contributing area,
moderate-highly contributing area, and catchment with summer TN for row crop and total
agriculture.

Discussion
I took a unique approach in my study by comparing spatial metrics that have been
traditionally used for stream studies to quantify different agricultural types to predict lake
nutrients. I found that differences in the relationships between agricultural spatial metrics and
lake nutrients depended upon season, nutrient, and dominant agricultural type. However, despite
differences, some common patterns emerged. Row crop and total agricultural were strongly
correlated with one another, as 50-70% of total agriculture consists of row crop across all study
lake catchments. This suggests that row crop is the dominant agricultural type measured around
selected study lakes, and may have the most impact on lake nutrient concentrations. The 100 m
lake-stream buffer and catchment were moderate to strongly correlated among the three

15

agricultural types and total agriculture. In comparison, the catchment and moderate-highly
contributing area spatial metrics were also strongly correlated, with further analysis indicating
that the spatial metrics were likely measuring similar aspects of catchment agricultural land
use/cover. More importantly, my analysis shows that several of the spatial scales were strongly
correlated with one another, and even though there were some differences among the different
spatial metrics, no single scale stood out as the best at predicting lake nutrients.
Correlations among agricultural types and total agriculture
Few studies have separated out the influence of individual agricultural types on surface
water nutrients (Johnson et al 1997, Strayer et al 2003, Buck et al 2004, Meynendonckx et al
2006, Diebel et al 2009). Studies will often define different agricultural types, but often lump
them together as total agriculture or cropland when analyzing the impact of land use/cover on
surface water (Osborne and Wiley 1988, Soranno et al 1996, Ahearn et al 2005, Arscott et al
2006, Taranu and Gregory-Eaves 2008, Gémesi et al 2011). Other studies measure all
agriculture within the landscape as one category of land use/cover, without distinguishing the
agricultural types or practices make up the category (Norvell et al 1979, Omernik et al 1981,
Detenbeck et al 1993, Hunsaker and Levine 1995, Jones et al 2004, Gergel 2005, Baker et al
2006, Fraterrigo and Downing 2008, Zampella and Procopio 2009). Separating out the possible
effects of different agricultural types on lake nutrients was an important goal for this study.
Different agricultural practices have been noted to have very different effects on water body
nutrient concentrations (Nerbonne and Vondraek 2001), and defining specific agricultural
practices may help to explain nutrient concentration variations between water-bodies with
similar land use/cover (Nerbonne and Vondraek 2001,Vanni et al 2001, Ekholm et al 2005). My
results show that row crop and total agriculture had the strongest correlation with spring and

16

summer TN, and weaker correlations with spring and summer TP depending on the spatial
metric (Tables 3a, 3b).
Relationships between agricultural spatial metrics and lake nutrients
I compared commonly used spatial metrics in an attempt to determine which spatial
metric, if any, was the best at predicting lake nutrient concentrations (Tables 3a, 3b). The 100 m
lake-stream buffer, moderate-highly contributing area, and catchment had the strongest
correlations with TP and TN for all three agricultural types and total agriculture (Table 3a).
Previous studies have observed strong correlations between surface water nutrient concentrations
and land use/cover quantified within close proximity of surface waters (Soranno et al 1996,
Johnson et al 1997, Diebel 2009), or with land use/cover measured at the catchment scale
(Hunsaker and Levine 1995, Jones et al 2004, Meynendonchx et al 2006, Fraterrigo and
Downing 2008, Taranu and Gregory-Eaves 2008). Other studies have emphasized the
importance of landscape arrangement, topography, and the amount of detail provided by
contributing areas and simulated models to predict nutrients (Hunsaker and Levine 1995, Gergel
2005, Baker et al. 2006). Many studies have quantified agricultural land use/cover at various
scales and measured their effects on stream nutrient concentrations (Omernik et al 1981, Osborne
and Wiley 1988, Hunsaker and Levine 1995, Strayer et al 2003, Baker et al 2006,
Meynendonckx et al 2006, Diebel et al 2009, Zampella and Procopio 2009). In comparison,
relatively few studies have examined the relationships between fixed-distance spatial metrics and
contributing areas to predict nutrients within reservoirs (Jones et al 2004, Gémesi et al 2011) or
lakes (Norvell et al 1979, Taranu and Gregory-Eaves 2008). My results suggest that many of the
spatial metrics are correlated and that in similar landscapes to Michigan, any of several extents
would work equally well at predicting lake nutrients. However, other studies have shown that

17

agricultural types predict nutrients differently depending on the scale used to measure land
use/cover (Buck et al 2004) and the season in which they are measured (Vanni et al 2001,
Bechman et al 2005, Diebel et al 2009).
Other studies have found that the fluctuations in nitrogen and phosphorus concentrations
can be caused by different types of agricultural land use/cover and seasonal precipitation (Vanni
et al 2001, Bechman et al 2005). Possibilities for TN variability between spring and summer
have been well established due to complexities of the nitrogen cycle, local agricultural types,
hydrology and seasonal precipitations. Kroger et al (2007) noted fluctuations in nitrogen
concentrations due to temperature dependency for microbial immobilization and denitrification,
plant assimilation, seasonal precipitation and storm events. Likewise, Bechmann et al (2005)
found that internal lake processes, such as light, temperature and algal uptake of phosphorus
significantly influenced the concentrations and availability of phosphorus within the water
column. The uptake of phosphorus and nitrogen during summer months may explain the weak
correlations found between spatial metrics and summer TN and TP (Table 3b).
Relationships among agricultural spatial metrics
My results showed that more than one spatial metric was strongly correlated with lake
nutrients (Tables 3a, 3b), which is supported by studies that have conflicting results in regards to
which is the best scale to quantify land use/cover (Hunsaker and Levine 1995, Soranno et al
1996, Johnson et al 1997, Jones et al 2004, Gergel 2005, Baker et al 2006, Meynendonchx et al
2006, Fraterrigo and Downing 2008, Taranu and Gregory-Eaves 2008, Dibel 2009). Since there
is more than one spatial metric that is strongly correlated with lake nutrients for agricultural
types, it is therefore necessary to determine how strongly correlated those spatial metrics are
with one another (Table 5).

18

My results showed that the 100 m lake-stream buffer and catchment have similar
correlation strengths with nutrients for all three agricultural types and total agriculture (Table 3a,
3b) and are strongly correlated with each other for each agricultural type (Figure 3). However,
the results from the residual analysis show that the 100 m lake-stream buffer helps to explain
additional variation in lake nutrients beyond what the catchment explains (Figure 4). I founds
similar results for the moderate-highly contributing area and the 100 m lake-stream buffer
(Figure 5). However, it appears that the moderate-highly contributing area and the catchment
measure nearly identical amounts of agricultural cover and so do not predict unique amounts of
lake nutrient variation (Figures 7, 8).
Other studies have found that different spatial metrics are strongly correlated with one
another, and that there are often difficulties in separating out the relative influence of near and
far land use/cover on water quality (Zampella and Procopio 2009). Arscott et al (2006) used
numerous spatial metrics to predict nutrients in reservoirs and streams, and found that land
use/cover quantified at the catchment scale were strongly related to land use/cover quantified
within buffers around riparian areas. Likewise, Baker et al (2006) found that agricultural land
use/cover measured within fixed-distance buffers around riparian areas were strongly correlated
with whole-catchment land use/cover, stating that land use/cover measured within the buffer
cannot reveal much about nutrient flux that is not already captured by the catchment.
Strayer et al (2003) showed that there was not a single optimal scale for quantifying the
effects of land use/cover on stream nutrients for empirical models. In addition, Gergel (2005)
showed that landscape arrangement, i.e. the general pattern of land use/cover throughout the
catchment coupled with topography may be the most important variables to consider when
measuring land use/cover in relation to non-point source pollution. Recent studies, such as

19

Gémesi et al (2011) show that catchment configuration (i.e., spatial arrangement of agricultural
land use/cover) was more important than composition (i.e, percent composition of land use/cover
types) at either catchment or riparian scales. Evidence from other studies, coupled with results
that spatial metrics are strongly correlated with one another among different agricultural types
and with lake nutrients, indicates that there may not be a single spatial metric that best predicts
lake nutrients.
Effect size of different agricultural types
Because total agriculture was comprised of 50-70% row crop within each spatial metric
(Table 2), row crops are likely contributing the most nutrients to lakes. However, it is important
to compare the effect sizes of each of the agricultural types (i.e., the slopes of the relationships)
to determine if different agricultural types have the same effect per unit increase in area. I
quantified univariate regressions of each of the nutrients and seasons against the different
agricultural types in the 100 m lake-stream buffer. I chose the 100 m lake-stream buffer because
this scale was consistently correlated with both nutrients and seasons. I found that the slopes for
forage and pasture are much steeper compared to row crops and total agriculture (Figures 9, 10).
These results suggest that lake nutrients react more quickly to pasture and forage than to row
crop, i.e. there is a larger increase in TP and TN for every increased unit of pasture and forage.
The percentage of row crop measured within the 100 m lake-stream buffer was higher than either
pasture or forage (Table 2). However, the linear regressions results indicate that forage and
pasture land use /cover could potentially have a stronger influence on lake nutrients if the
percentage of either land use/cover increased. These results are somewhat preliminary because
in my catchments, the percentages of forage and pasture were relatively low, so the regressions

20

for these types are not as well defined as they are for row crop. This result warrants further
investigation in catchments with higher proportions of forage and pasture.
However, there is evidence from the literature supporting the overall pattern that I found.
As stated previously, the influence of row crops on surface water is well known and has become
the focus of management and conservation practices to prevent excess nutrients from entering
waterways (Arbuckle and Downing 2001). In comparison, forage crops and pasture cover are
very efficient at nutrient utilization and are not associated with the same amount of soil erosion
and agrichemical runoff as row crops. However, sediment loss and stream bank erosion due to
unregulated access of cattle is a known source of excess nutrients in surface water (Zaimes et al
2008). Excessive fertilizers on forage and pasture grass are also known to leach to surface runoff
and are a particular problem on steep slopes in late growing seasons (Doo Hong Min 2002).
Forage crops and pastures require the application of fertilizers based off of soil test results , but
without proper attention to crop health and management considerations, the crops can decrease in
fertility, resulting in cover loss and increased erosion and runoff (Daniels et al 2011). A
combination of steep topography, proximity, excessive fertilizer applications and unregulated
management practices may help to explain why lake nutrients react more quickly for forage and
pasture than with row crop and total agriculture when measured in the 100 m lake-stream buffer.
The influence of pasture and forage may be strong on lake nutrients, but the magnitude of their
effects are difficult to quantify because these lands are less common across the landscape.

Conclusions
Recommendations for quantifying land use/cover around lakes

21

I found little statistical differences among the spatial metrics. Quantifying land use/cover
in relation to lake nutrients can be a challenge for scientists and managers due to the
complexities of the systems they are analyzing, as well as constraints from the time and cost of
collecting data. As stated previously, many fresh water studies analyze either land within close
proximity to water bodies, or land throughout the catchment. Due to the time and cost
constraints of digitizing lake catchments, my results support measuring agricultural land
use/cover at the 100 m lake-stream buffer. Even though the 100 m lake-stream buffer helps to
explain additional variation in lake nutrients, both the catchment and the 100 m lake-stream
buffer are measuring similar features of the agricultural land use/cover.
Total agriculture and row crop had strong to moderate correlations with spring TN, which
is supported by many other studies. Pasture and forage had weak correlations with lake nutrients
for all five of the spatial metrics when compared to row crop and total agriculture. However,
scientists and managers should not disregard the impacts of pasture and forage types when
analyzing the cumulative effects of agricultural land use/cover on lake nutrients concentrations;
even though pasture and forage make up a small percentage of measured land use/cover in the
lake catchments, their impacts may be overshadowed by the prevalence of row crop in an
analysis such as this. Further research is needed to more fully quantify the effects of pasture and
forage lands on lake nutrients.

22

Table 1. Summary statistics of the lake variables. TP is total phosphorus, TN is total nitrogen.
N= 204 lakes.
Variable
Surface area (ha)

Minimum
14

Maximum
4,183

Median
75

Catchment area (ha)

52

241,640

769

Mean depth (m)

0.8

21

5

Maximum depth (m)

3.0

39

13

1

176

16

3

290

12

146

7,800

655

90

3,180

575

-1

TP, spring (ug L )
-1

TP, summer (ug L )
-1

TN, spring (ug L )
-1

TN, summer (ug L )

23

Table 2. Summary statistics of the percentage of agricultural types for each spatial metric. N =
204 lakes.
Spatial metric
Lake-stream buffer, 100 m
Pasture
Forage
Row crop
Total agriculture

Minimum

Maximum

Median

0
0
0
0

16
23
46
53

3
2
1
4

Buffer, 500 m
Pasture
Forage
Row crop
Total agriculture

0
0
0
0

19
18
47
60

3
1
3
7

Contributing area, highly
Pasture
Forage
Row crop
Total agriculture

0
0
0
0

52
24
65
66

3
2
4
9

Contributing area, moderatehighly
Pasture
Forage
Row crop
Total agriculture

0
0
0
0

22
24
57
65

4
20
6
12

Catchment
Pasture
Forage
Row crop
Total agriculture

0
0
0
1

21
27
61
70

5
2
6
17

24

Table 3a. Correlations between spring lake total phosphorus (TP) or total nitrogen (TN) concentrations and the different spatial
metrics. Numbers in bold are either the highest correlation value, or within 0.05 of the highest value within the respective column.
All values in the tables are significant at α ≤ 0.05. N= 204 lakes.
Spatial metric

Lake-stream buffer, 100 m
Lake buffer, 500 m
Contributing area, highly
Contributing area,
moderate-highly
Catchment

TP
TP
Pasture Forage

TP
Row crop

0.33
0.26
0.17

0.42
0.25
0.33

0.32
0.30
0.27

TP
Total
agriculture
0.41
0.34
0.32

0.20

0.39

0.32

0.36

0.24

0.30

0.64

0.65

0.23

0.40

0.33

0.40

0.20

0.30

0.60

0.66

25

TN
Pasture

TN
Forage

TN
Row crop

0.44
0.23
0.22

0.42
0.22
0.23

0.65
0.43
0.57

TN
Total
agriculture
0.69
0.45
0.60

Table 3b. Correlations between summer lake total phosphorus (TP) or total nitrogen (TN) concentrations and the different spatial
metrics. Numbers in bold are either the highest correlation value, or within 0.05 of the highest value within the respective column.
All values in the tables are significant at α ≤ 0.05. N= 204 lakes.
Spatial metric

Lake-stream buffer, 100 m
Lake buffer, 500 m
Contributing area, highly
Contributing area,
moderate-highly
Catchment

TP
Pasture

TP
Forage

TP
Row crop

TN
Pasture

TN
Forage

TN
Row crop

0.28
0.11
0.31

TP
Total
agriculture
0.33
0.17
0.34

0.32
0.16
0.19

0.36
0.26
0.30

0.39
0.32
0.44

TN
Total
agriculture
0.44
0.34
0.44

0.21
0.18
n.s.

0.25
n.s.
0.39

0.15

0.20

0.24

0.26

0.21

0.25

0.44

0.47

0.17

0.21

0.22

0.27

0.19

0.26

0.45

0.49

26

Table 4. Correlations between different agriculture types within each spatial metrics. All values
in the tables are significant at α ≤ 0.05, ‘n.s.’ is not significant. N= 204 lakes. Note that the value
in the columns are measured at the same spatial extent as the value in the row.
Spatial metric
Pasture
Lake-stream buffer, 100 m
Buffer, 500 m
Contributing area, highly
Contributing area,
moderate-highly
Catchment

Pasture

Forage

Row crop

Total
agriculture

-

0.39
0.36
0.19
0.33

0.34
0.42
0.29
0.37

0.61
0.67
0.54
0.60

-

0.36

n.s.

0.61

-

0.40
0.30
0.27
0.17

0.64
0.63
0.53
0.46

-

n.s.

0.45

-

0.93
0.91
0.93
0.94

-

0.94

Forage
Lake-stream buffer, 100 m
Buffer, 500 m
Contributing area, highly
Contributing area,
moderate-highly
Catchment
Row crop
Lake-stream buffer, 100 m
Buffer, 500 m
Contributing area, highly
Contributing area,
moderate-highly
Catchment
Total agriculture
Lake-stream buffer, 100 m
Buffer, 500 m
Contributing area, highly
Contributing area,
moderate-highly
Catchment

-

27

Table 5. Correlations between different spatial metrics within each agricultural type. All values in the table are significant at α=0.05.
N= 204. ‘Mod.- high.’ is moderate-highly. Note that the value in the columns are measured for the same agricultural type as the value
in the row.
Spatial metric

Lake-stream buffer,
100 m

Buffer,
500 m

Contributing area,
highly

Contributing area,
mod.-high.

Pasture
Lake-stream buffer, 100 m
Buffer, 500 m
Contributing area, highly
Contributing area, moderate-highly
Catchment

0.38
0.38
0.47
0.45

0.53
0.62
0.78

0.74
0.68

0.94

Forage
Lake-stream buffer, 100 m
Buffer, 500 m
Contributing area, highly
Contributing area, moderate-highly
Catchment

0.22
0.49
0.57
0.56

0.57
0.64
0.66

0.73
0.70

0.97

Row crop
Lake-stream buffer, 100 m
Buffer, 500 m
Contributing area, highly
Contributing area, moderate-highly
Catchment

0.44
0.80
0.76
0.75

0.50
0.64
0.70

0.90
0.86

0.98

Total agriculture
Lake-stream buffer, 100 m
Buffer, 500 m
Contributing area, highly
Contributing area, moderate-highly
Catchment

0.46
0.70
0.73
0.71

0.63
0.73
0.77

0.88
0.83

0.98

28

Study lakes
Agricultural land use/cover
Other land use/cover

0

150 km

Figure 1. Distribution of the 204 lakes used in this study. All types of agricultural land
use/cover are shown in grey.

29

A)

B)

N

catchment boundary
0

6 km

100 m lake-stream buffer

Highly C.A.

500 m lake buffer

Moderate-highly C.A.

Figure 2. Four different study lakes displaying the five spatial extents used to measure agricultural
land use/cover within lake catchments. Column A) equidistant buffers around either lakes and
streams, or lakes only, and column B) contributing areas estimated at highly contributing or
moderate-highly contributing areas, as well as the full catchment area. ‘C.A.’ is contributing area.
30

0

10

20

% Forage in the 100 m lake-stream buffer
40 50 60 70
0
10 20 30
30

40

50

60

70

70
60
10

20

30

40

50

C) Row crop
r = 0.75
p < 0.0001

0

% Row crop in the 100 m lake-stream buffer

% Pasture in catchment

0

10

20
30
40
50
% Row crop in catchment

60

70

B) Forage
r = 0.56
p < 0.0001

0
% Total agriculture in the 100 m lake-stream buffer
0
10 20 30 40 50 60 70

70
60
50
40
30
20
10
0

% Pasture in the 100 m lake-stream buffer

A) Pasture
r = 0.46
p < 0.0001

10

20

30

40

50

60

70

10
20 30
40
50 60
% Total agriculture in catchment

70

% Forage in catchment
D) Total agriculture
r = 0.71
p < 0.0001

0

Figure 3. Plots of agriculture measured in the 100 m lake-stream buffer (y axis) versus the catchment for A) pasture B) forage, C) row
crop, and D) total agriculture (x axis).

31

4
3
1
-1

0

Residuals

1
0
-1
-2

Residuals

B) Spring TN

2

2

3

A) Spring TP

2

2

R = 0.09
p < 0.0001

-2

-3

R = 0.02
p = 0.04
0
10
20
30
40
50
% Total agriculture in the 100 m lake-stream buffer

0
10
20
30
40
50
% Total agriculture in the 100 m lake-stream buffer
D) Summer TN

-1
-3

2

R = 0.02
p = 0.04

4

-2

2

Residuals

0

Residuals

2

0

1

4

2

C) Summer TP

0
10
20
30
40
50
% Total agriculture in the 100 m lake-stream buffer

2

R = 0.02
p = 0.06
0
10
20
30
40
50
% Total agriculture in the 100 m lake-stream buffer

Figure 4. Plots of total agriculture measured in the 100 m lake-stream buffer (y axis) versus the residuals of
total agriculture measured in the catchment (x axis) for lake nutrient concentrations.
32

70

70
0

10

20

B) Forage
r = 0.97
p < 0.0001

% Forage in Moderate-highly C.A.
0 10 20 30 40 50 60

% Pasture in Moderate-highly C.A.
0 10 20 30 40 50 60

A) Pasture
r = 0.94
p < 0.0001

30

40

50

60

70

0

20

30

40

50

60

70

0
10
20 30
40
50
60 70
% Row crop in the 100 m lake-stream buffer

60
10

20

30

40

50

D) Pasture
r = 0.46
p < 0.0001

0

C) Row crop
r = 0.76
p < 0.0001

70

% Forage in the 100 m lake-stream buffer
% Total agriculture in Moderate-highly C.A.

% Row crop in Moderate-highly C.A.
0 10 20 30 40 50 60 70

% Pasture in the 100 m lake-stream buffer

10

0
10
20
30
40
50
60 70
% Total agriculture in the 100 m lake-stream buffer

Figure 5. Plots of agriculture measured in the moderate-highly contributing area (y axis) versus the 100 m lake-stream
buffer for A) pasture, B) forage, C) row crop, and D) total agriculture (x axis). ‘C.A.’ is contributing area.
33

4

3

B) Spring TN

10

20

30

40

1
-1

2

R = 0.02
p = 0.04

-2

-4

R = 0.02
p = 0.03
0

0

Residuals

-1
-2

2

-3

Residuals

0

2

1

3

2

A) Spring TP

50

0

20

30

40

50

% Total agriculture in the 100 m lake-stream buffer
3

% Total agriculture in the 100 m lake-stream buffer

10

C) Summer TP

0
-2

2

-3

R = 0.02
p = 0.04

-4

-2

-1

Residuals

2
0

Residuals

1

4

2

D) Summer TN

0
10
20
30
40
50
% Total agriculture in the 100 m lake-stream buffer

2

R = 0.02
p = 0.05
0
10
20
30
40
50
% Total agriculture in the 100 m lake-stream buffer

Figure 6. Plots of total agriculture measured in the 100 m lake-stream buffer (y axis) versus the residuals of total agriculture
measured in the moderate-highly contributing area (x axis) for lake nutrient concentrations.
34

70

0

10

20

30

40

50

10

20

30

40

50

% Row crop in catchment

60

70

60
50
40

10

20

30

40

50

60

70

60

70

70

% Forage in catchment

60

C) Row crop
r = 0.97
p < 0.0001

0

0
% Total agriculture crop in Moderate-highly C.A.

70

30
0

70

D) Total agriculture
r = 0.97
p < 0.0001

50

60

40

50

30

40

% Pasture in catchment

60

20

% Forage in Moderate-highly C.A.
30

20

20

10

10

0

0

% Row crop in Moderate-highly C.A.

B) Forage
r = 0.97
p < 0.0001

10

70
60
50
40
30
20
10
0

% Pasture in Moderate-highly C.A.

A) Pasture
r = 0.94
p < 0.0001

0

10

20

30

40

50

% Total agriculture in catchment

Figure 7. Plots of agriculture measured in the moderate-highly contributing area (y axis) versus the catchment for A) pasture, B)
forage, C) row crop and D) total agriculture (x axis). ‘C.A.’ is contributing area.
35

4

2

R = 0.004
p = 0.79

B) Spring TN

3

3

A) Spring TP

R = 0.004
p = 0.79

2

1
-3

-3

-2

-2

-1

0

Residuals

0
-1

Residuals

1

2

2

5

0
10
20
30
40
50
60
% Total agriculture in Moderate-highly C.A.

0

C) Summer TP

D) Summer TN

2

2

R = 0.004
p = 0.84

0

Residuals

1
-2

-2

-1

-1

0

Residuals

2

1

3

2

R = 0.0001
p = 0.93

4

10
20
30
40
50
60
% Total agriculture in Moderate-highly C.A.

0

10

20

30

40

50

60

% Total agriculture in Moderate-highly C.A.

0

10

20

30

40

50

60

% Total agriculture in Moderate-highly C.A.

Figure 8. Plots of total agriculture measured in the moderate-highly contributing area (y axis) versus the residuals of total
agriculture measured in the catchment (x axis) for lake nutrient concentrations. ‘C.A.’ is contributing area.
36

10

10

2

4

6

Spring TP

6
2

4

Spring TP

8

B)

8

A)

y = 0.085x + 2.577

y = 0.104x + 2.726

2

0

10

60

20

30

40

50

0
60

2

2

4

6

6

Spring TP

8

8

D)

4

Spring TP

10
20
30
40
50
% Forage in the 100 m lake-stream buffer

10

10

% Pasture in the 100 m lake-stream buffer
C)

y = 0.024x + 2.734

y = 0.024x + 2.619

0

R = 0.102
10
20
30
40
% Row crop in the 100 m lake-stream
buffer
regressions of spring TP (y axis) versus

2

0

2

0

60

2

R = 0.192

0

0

R = 0.146

50
60

R = 0.188
0
10
20
30
40
50
% Total agriculture in the 100 m lake-stream buffer

Figure 9. Linear
percent agricultural type measured in the 100m lake-stream buffer for A)
pasture, B) forage, C) row crop and D) total agriculture (x axis).
37

10

10

8
6
2

4

Spring TN

8
6
4
2

Spring TN

B)

A)

y = 0.089x + 6.358

y = 0.088x + 6.576

2

0
10

10
20
30
40
50
% Pasture in the 100 m lake-stream buffer

10
20
30
40
50
% Forage in the 100 m lake-stream buffer

8

D)

2

4

6

Spring TN
6
4

8

C)

2

Spring TN

R = 0.175

0

0
0
10

2

R = 0.204

y = 0.043x + 6.453

y = 0.035x + 6.316

2

2

0

0

0

R = 0.423
10
20
30
40
50
% Row crop in the 100 m lake-stream buffer

R = 0.484
0
10
20
30
40
50
% Total agriculture in the 100 m lake-stream buffer

Figure 10. Linear regressions of spring TN (y axis) versus percent agricultural type measured in the 100 m lake-stream buffer for
A) pasture, B) forage, C) row crop and D) total agriculture (x axis).
38

REFERENCES

39

REFERENCES

Arbuckle K.E. and J.A. Downing. 2001. Use on lake N:P in a predominantly agricultural
landscape. Limnology and Oceanography 46(4): 970-975.
Arhearn, D.S., R.W. Sheibley, R.A. Dahlgren, M. Anderson, J. Johnsoc and K.W. Tate. 2005.
Land use and land cover influence on water quality in the last free-flowing river draining the
western Sierra Nevada, California. Journal of Hydrology 313: 234–247.
Arscott, D.B., C.L. Dow and B. W. Sweeney. 2006. Landscape template of New York City’s
drinking-water- supply watersheds. Journal of the North American Benthological Society,
25(4):867–886.
Baker M.E., D.E. Weller, and T.E. Jordan. 2006. Improved methods for quantifying potential
nutrient interception by riparian buffers. Landscape Ecol 21:1327–1345.
Bechmann, M.E., D. Bergeb, H.O. Eggestada and S.M. Vandsemb. 2005. Phosphorus
transfer from agricultural areas and its impact on the eutrophication of lakes: two longterm
integrated studies from Norway. Journal of Hydrology 304: 238-250.
Bishop, P.L., W.D. Hively, J.R. Stedinger, M.R. Rafferty, J.L. Lojpersberger and J.A.
Bloomfield. 2005. Multivariate analysis of paired watershed data to evaluate agricultural
best management practice effects on stream water phosphorus. Journal of Environmental
Quality 34:1087-1101.
Brandt, D. 2009. Surface water resources, p. 315-328. In R. Shaetzl, J. Darden, and D Brandt
[eds.], Michigan Geopgraphy and Geology. Pearson, New York, New York.
Buck, O, D.K. Niyogi and C.R. Townsend. 2004. Scale-dependence of land use effects on water
quality of streams in agricultural catchments. Environmental Pollution 130: 287-299.
Carpenter SR, Caraco NF, Correll DL, Howarth RW, Sharpley AN, and Smith VH. 1998.
Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecological
Applications. 8:559–568.
Carpenter, S.R., D. Ludwig and W.A. Brock. 1999. Management of eutrophication for
lakes subject to potentially irreversible change. Ecological Applications 9(3): 751-771.
Daniels, M. T. Daniel, D. Carman, J. Langston, and K. VanDevenden. Soil phosphorus levels:
concerns and recommendations. 2011. Agriculture and Natural Resources. University of
Arkansas, Division of Agriculture, Cooperative Senstitive. FSA1029-4M-6-98N.
Detenbeck, N.E., C.A. Johnston and G.J. Niemi. 1993. Wetland effects on lake water quality in
the Minneapolis/St. Paul metropolitan area. Landscape Ecology 8 (1): 39-61.
40

Diebel, M.W., J.T. Maxted, D.M. Robertson, S. Han and M.K. Vander Zanden. 2009.
Landscape planning for agricultural nonpoint source pollution reduction iii: assessing
phosphorus and sediment reduction potential. Environmental Management 43:69–83.
Doo-Hong Min. 2002. Forage management: Re-evaluating manure application to alfalfa.
Michigan Dairy Review 15: January 2002.
Egertson, C.J. and J.A. Downing. 2004. Relationship of fish catch and composition to
water quality in a suite of agriculturally eutrophic lakes. Canadian Journal of Fisheries
and Aquatic Science 61:1784-1796.
Egertson, C.J., J.A. Kopaska and J.A. Downing. 2004. A century of change in macrophyte
abundance and composition in response to agricultural eutrophication. Hydrobiologia 524:
145–156.
Ekholm, P., E. Turtolab, J. Gronroosb, P. Seurib and K. Ylivainio. 2005. Phosphorus loss
from different farming systems estimated from soil surface phosphorus balance.
Agriculture, Ecosystems and Environment 110(3-4): 266-278.
Erskine, R.H., T.R. Green, J.A. Ramirez and L.H. MacDonald. 2006. Comparison of grid-based
algorithms for computing upslope contributing area. Water Resources Research 42: W09416,
doi:10.1029/2005WR004648, 2006.
Fraterrigo, Jennifer and John A. Downing. 2008. The influence of land use on lake nutrients
varies with catchment transport capacity. Ecosystems 11: 1021–1034.
Fuller, L.M. and R.J. Minnerick. 2008. State and regional water quality characteristics
and trophic conditions of Michigan’s inland lakes, 2001-2005. United States
Geological Survey: 1-66.
Gémesi, Zsolt and John A. Downing, Richard M. Cruse, and Paul F. Anderson. 2011. Effects of
catchment configuration and composition on downstream lake water quality. Journal of
Environmental Quality 40: 517–527.
Gergel, S.E. 2005. Spatial and non-spatial factors: When do they affect landscapes indicators of
catchment loading? Landscape Ecology 20:177–189.
Hoorman, J., T. Hones, T. Sudman Jr., T. Dirksen, J. Iles and K.R. Islam. 2008. Agricultural
impacts on lake and stream water quality in Grand Lake St. Marys, Western Ohio. Water Air
Soil Pollution 193:309–322. DOI 10.1007/s11270-008-9692-1.
House, F.J. 2006. Agricultural Programs, Terms and Laws. Novika Books, Nova Science
Publishers: 1-304.
Hunsaker, C.T. and D.A. Levine. 1995. Hierarchical Approaches to the Study of Water Quality
in Rivers. BioScience 45 (3) Ecology of Large Rivers: 193-203.
41

Johnson, L.B., C. Richards, G.E. Host and J.W. Arthur. 1997. Landscape influences on water
chemistry in Midwestern stream ecosystems. Freshwater Biology 37: 193-208.
Jones, K.B., A. C. Neale, M. S. Nash, R. D. Van Remortel, J.D. Wickham, K. H. Riitters and
R.V. O’Neill. 2001. Predicting nutrient and sediment loadings to streams from landscape
metrics: A multiple watershed study from the United States Mid-Atlantic Region. Landscape
Ecology 16: 301-312.
Jones, J.R. and M.F. Knowlton. 2005. Suspended solids in Missouri reservoirs in relation
to catchment features and internal processes. Water Reaserch 39(15): 3629-3635.
Jones, J.R. and M.F. Knowlton, D.V. Obrecht, and E.A. Cook. 2004. Importance of landscape
variables and morphology on nutrients in Missouri reservoirs. Canadian Journal of Fish and
Aquatic Science 61: 1503–1512.
Kirchner W.B. 1975. An examination of the relationship between drainage basin
morphology and the export of phosphorus. American Society of Limnology and
Oceanography 20: 267-230.
Kroger, R., M. M. Holland, M. T. Moore and C. M. Cooper. 2007. Hydrological
variability and agricultural drainage ditch inorganic nitrogen reduction capacity. Journal
of Environmental Quality 36:1646 -1652.
Lehmann, J., Z. Lan, C. Hyland, S. Sato, D. Soloman and Q. M. Ketterings. 2005. Long-term
dynamics of phosphorus forms and retention in manure-amended soils. Environmental Science
and Technology 39: 6672-6680.
Lewis, R.A. 2002. CRC Dictionary of Agricultural Sciences. CRC Press, LLC:1-284.
Lipton, L.K. 1995. Dictionary of Agriculture : From Abaca to Zoonosis. Boulder Lynne
Rienner Publishers.
Malhi, S.S., K.S. Gill, D.H. McCartney and R. Malmgren. 2004. Fertilizer management of forage
crops in the Canadian Great Plains. Recent Research and Developement of of Crop
Sciences.1: 237-271 ISBN: 81-271-0046-3.
Meynendonckx, J., G. Heuvelmans, B. Muys and J. Feyen. 2006. Effects of watershed and
riparian zone characteristics on nutrient concentrations in the River Scheldt Basin. Hydrology
and Earth System Sciences Discussions 3: 653–679.
Monaghana, R.M., C.A.M. de Kleina and R.W. Muirhead. 2008. Prioritisation of farm
scale remediation efforts for reducing losses of nutrients and faecal indicator organisms
to waterways: A case study of New Zealand dairy farming. Journal of Environmental
Management 87(4): 609-622.

42

Nerbonne, B.A. and B. Vondracek. 2001. Effects of local land use on physical habitat,
benthic macroinvertebrates and fish in the Whitewater River, Minnesota, USA.
Environmental Management 28(1): 87-99.
Nilsson, A. and L. Hakanson. 1992. Relationship between drainage area characteristics
and lake water quality. Environmental Geology and Water Science 19(2): 75-81.
Norvell, W.A., C.R. Frink and D.E. Hill. 1979. Phosphorus in Connecticut lakes predicted by
land use. Proceedings of the National Academy of Sciences of the United States of America
76(11): 5426-5429.
Omernik, J.M., A.R. Abernathy and L.M. Male. 1981. Stream nutreint levels and proximity of
agricultural and forest land to streams: some relationships. Journal of Soil and Water
Conservation July/August 36 (4):227-231.
Osborne, J.L. and M.J. Wiley. 1988. Empirical relationships between land use/cover and stream
water quality in an agricultural watershed. Journal of Environmental Management 26: 9-27.
Renwick, W.H., M.J. Vanni, Q. Zhang and J. Patton. 2008. Water quality trends and
changing agricultural practices in a midwest U.S. watershed, 1994–2006. Journal of
Environmental Quality 37:1862-1874.
Schaetzl, R.J. 2009. Surface water resources, p. 315-328. In R. Shaetzl, J. Darden, and D Brandt
[eds.], Michigan Geopgraphy and Geology. Pearson, New York, New York.
Soranno, P.A., S.L. Hubler, S.R. Carpenter and R.C. Lathrop. 1996. Phosphorus loads to
surface waters: a simple model to account for spatial pattern of land use. Ecological
Applications 6(3): 865-878.
Spectrum Analytic Incorporated. Fertilizing alfalfa. Spectrum Analytic Inc. n.d.,Web.,
9/15/2011.
Stephens, A. 1998. Dictionary of Agriculture. P.H. Collin, Fitzroy Dearborn Publishers.
Stephenson J. M. and. A. Morin. 2009. Covariation of stream community structure and biomass
of algae, invertebrates and fish with forest cover at multiple spatial scales. Freshwater
Biology. 54 (10): 2139-2154. DOI: 10.1111/j.1365-2427.2008.02142.x.
Strayer, D.L., R.E. Beighley, L.C. Thompson, S. Brooks, C. Nilsson, G.Pinay and R.J. Naiman.
2003. Effects of land cover on stream ecosystsems: roles of empirical models and scaling
issues. Ecosystems 6: 407–423.
Taranu, Z.E. and I. Gregory-Eaves. 2008. Quantifying relationships among phosphorus,
agriculture, and lake depth at an inter-regional scale. Ecosystems. DOI: 10.1007/s10021-0089153-0.

43

Vanni, M.J., W.H. Renwick, J.L. Headworth, J.D. Auch and M. H. Schaus. 2001.
Dissolved and particulate nutrient flux from three adjacent agricultural watersheds: A
five-year study. Biogeochemistry 54: 85–114.
Vanni, M.J., K.K. Arend, M.T. Bremigan, D.B. Bunnell, J.E. Garvey, M.J. González,
W.H. Renwick, P.A. Soranno, and R.A. Stein. 2005. Linking landscape and food webs:
effects of omnivorous fish and watersheds on reservoir ecosystems. Bioscience
55(2):155-167.
Van Sickle, J. and C.B. Johnson. 2008. Parametric distance weighting of landscape influence on
streams. Landscape Ecology 23:427–438. DOI 10.1007/s10980-008-9200-4.
Walsh, S. 2000. Lakes, wetlands, and streams as dynamic drivers of land use/cover pattern and
change: a unique view of aquatic ecosystems in the landscape. MS Thesis. MSU. 67 pp.
Wolfson, L. 2009. Surface water resources, p. 206-222. In R. Shaetzl, J. Darden, and D Brandt
[eds.], Michigan Geopgraphy and Geology. Pearson, New York, New York.
Zaimes, G. N., R.C. Schultz, and T.M. Isenhart, 2008. Streambank soil and phosphorus
losses under different riparian land-uses in Iowa. Journal of the American Water
Resources Association (JAWRA) 44(4):935-947.
Zampella, R. A. and N. A. Procopio. 2009. Landscape patterns and water-quality relationships in
New Jersey Pinelands streams. Pinelands Commission, New Lisbon, New Jersey, USA.
Zedler, J.B. 2003. Wetlands at your service: reducing impacts of agriculture at the
watershed scale. Frontiers in Ecology and the Environment 1(2): 65-72.

44