Landscape-based assessment of human disturbance for michigan lakes.

Abstract

Assessment of lake impairment status and identification of threats' type and source is essential for protection of intact, enhancement of modified, and restoration of impaired lakes. For regions in which large numbers of lakes occur, such assessment has usually been done for only small fractions of lakes due to resource and time limitation. This study describes a process for assessing lake impairment status and identifying which human disturbances have the greatest impact on each lake for all lakes that are 2 ha or larger in the state of Michigan using readily available, georeferenced natural and human disturbance databases. In-lake indicators of impairment are available for only a small subset of lakes in Michigan. Using statistical relationships between the in-lake indicators and landscape natural and human-induced measures from the subset lakes, we assessed the likely human impairment condition of lakes for which in-lake indicator data were unavailable using landscape natural and human disturbance measures. Approximately 92% of lakes in Michigan were identified as being least to marginally impacted and about 8% were moderately to heavily impacted by landscape human disturbances. Among lakes that were heavily impacted, more inline lakes (92%) were impacted by human disturbances than disconnected (6%) or headwater lakes (2%). More small lakes were impacted than medium to large lakes. For inline lakes, 90% of the heavily impacted lakes were less than 40 ha, 10% were between 40 and 405 ha, and 1% was greater than 405 ha. For disconnected and headwater lakes, all of the heavily impacted lakes were less than 40 ha. Among the anthropogenic disturbances that contributed the most to lake disturbance index scores, nutrient yields and farm animal density affected the highest number of lakes, agricultural land use affected a moderate number of lakes, and point-source pollution and road measures affected least number of lakes. Our process for assessing lake condition represents a significant advantage over other routinely used methods. It permits the evaluation of lake condition across large regions and yields an overall disturbance index that is a physicochemical and biological indicator weighted sum of multiple disturbance factors. The robustness of our approach can be improved with increased availability of high-resolution disturbance datasets.

Full text

Page 1

ENVIRONMENTAL ASSESSMENT
Landscape-Based Assessment of Human Disturbance
for Michigan Lakes
Lizhu Wang•Kevin Wehrly•James E. Breck•
Lidia Szabo Kraft
Received: 12 February 2010/Accepted: 23 June 2010/Published online: 13 July 2010
? Springer Science+Business Media, LLC 2010
Abstract
identification of threats’ type and source is essential for
protection of intact, enhancement of modified, and resto-
ration of impaired lakes. For regions in which large num-
bers of lakes occur, such assessment has usually been done
for only small fractions of lakes due to resource and time
limitation. This study describes a process for assessing lake
impairment status and identifying which human distur-
bances have the greatest impact on each lake for all lakes
that are 2 ha or larger in the state of Michigan using readily
available, georeferenced natural and human disturbance
databases. In-lake indicators of impairment are available
for only a small subset of lakes in Michigan. Using sta-
tistical relationships between the in-lake indicators and
landscape natural and human-induced measures from the
subset lakes, we assessed the likely human impairment
condition of lakes for which in-lake indicator data were
unavailable using landscape natural and human disturbance
measures. Approximately 92% of lakes in Michigan were
identified as being least to marginally impacted and about
8% were moderately to heavily impacted by landscape
human disturbances. Among lakes that were heavily
impacted, more inline lakes (92%) were impacted by
human disturbances than disconnected (6%) or headwa-
ter lakes (2%). More small lakes were impacted than
medium to large lakes. For inline lakes, 90% of the heavily
impacted lakes were less than 40 ha, 10% were between
40 and 405 ha, and 1% was greater than 405 ha. For
Assessment of lake impairment status and
disconnected and headwater lakes, all of the heavily
impacted lakes were less than 40 ha. Among the anthro-
pogenic disturbances that contributed the most to lake
disturbance index scores, nutrient yields and farm animal
density affected the highest number of lakes, agricultural
land use affected a moderate number of lakes, and point-
source pollution and road measures affected least number
of lakes. Our process for assessing lake condition repre-
sents a significant advantage over other routinely used
methods. It permits the evaluation of lake condition across
large regions and yields an overall disturbance index that is
a physicochemical and biological indicator weighted sum
of multiple disturbance factors. The robustness of our
approach can be improved with increased availability of
high-resolution disturbance datasets.
Keywords
Catchment ? Disturbance index
Bioassessment ? Nutrient ? Lake condition ?
Introduction
Natural and man-made lakes in the United States and many
other parts of the world are substantially impacted by
human activities that occur within the lakes, along lake
shorelines, and in lake catchments, as well as by human-
released materials in the atmosphere. For large geographic
areas, such as a state or a multistate region, usually insuf-
ficient field data collected from individual lakes are avail-
able for identifying how many and which lakes are affected,
the geographic distribution of those lakes, and the serious-
ness in impairment of those lakes. Such information is
essential for developing effective lake management strate-
gies to maintain and improve water quality, aquatic habi-
tats, fisheries, and other recreational uses. Such information
L. Wang (&) ? K. Wehrly ? J. E. Breck ? L. S. Kraft
Institute for Fisheries Research, Michigan Department of Natural
Resources and Environment and University of Michigan,
212 Museum Annex, 1109 N. University, Ann Arbor,
MI 48109, USA
e-mail: lizuwang@umich.edu
123
Environmental Management (2010) 46:471–483
DOI 10.1007/s00267-010-9525-z

Page 2

is also necessary to inform policy makers and to meet the
requirements of the US Clean Water Act. Section 305b
requires states and tribes to prepare a comprehensive
inventory of the condition of their waters and Section 303d
requires states and tribes to list all waters not attaining their
designateduses(http://www.epa.gov/owow/monitoring/
305bguide/v1ch1.pdf).
Many state and federal agencies have established com-
prehensive monitoring and assessment programs to meet
these information needs. A common sampling design used
by water resource agencies is to target lakes for which local
and catchment human influences are high or where infor-
mation is needed for assessing compliance or for making
regulatory or management decisions. Targeted sampling
provides information for specific lakes and can be used for
determining whether the sampled lakes meet their desig-
nated uses. This sampling design, however, cannot provide
information on the overall quality of lakes in a state or a
region because targeted lakes are rarely representative of
the larger population of lakes. To assess regional condi-
tions, state and federal agencies typically employ a strati-
fied random selection of lakes from a particular basin or
ecoregion. In this design, a portion of lakes are sampled
within each basin or ecoregion on a rotational basis with a
target of completing an entire state or all ecoregions within
5–10 years. Stratified random sampling provides reason-
able estimates of the proportion of lakes that are impaired
for a particular basin or ecoregion. This design, however,
does not provide information on unsampled lakes, and the
status of many individual lakes in the basin or ecoregion
remains unknown. Assessing the status of all lakes,
including unsampled waters, would be extremely helpful in
directing regulatory and management efforts. Conse-
quently, an assessment approach that can be used for not
only assessing what proportion of lakes is impacted by
human activities but also for assessing impairment condi-
tions of unsampled lakes is highly desirable.
Another challenge to assessing lake condition involves
choosing appropriate stressors and indicators so that vari-
ous physical, chemical, and biological conditions can be
realistically assessed with a limited amount of resources
and time. The US Environmental Protection Agency (US
EPA) recommended numerous core and supplemental
indicators to state and tribal water monitoring and assess-
ment programs (USEPA 2003). The US EPA also recom-
mended that agencies use a different set of indicators to
monitor waters with different designated uses (USEPA
2003). Assessing the recommended suite of indicators,
which includes conductivity, temperature, water clarity,
dissolved oxygen (DO), pH, nutrients, eutrophic condition,
chlorophyll-a, Secchi depth, pathogens, trace metals,
landscape conditions, sediment contaminants, catchment
land uses, and at least two biological assemblages, makes
the assessmentprogram veryexpensive andtime-consuming.
Assessing the entire suite of indicators becomes cost-prohib-
itive when many lakes are to be sampled, and in states or
regions where a large number of lakes occur, only a small
proportionoflakescanbesampled.Hence,ascreeningtoolis
neededfor identifying lakesthatneedintensive sampling and
lakes that can be assessed using less intensive field sampling
or alternative approaches (Wang and others 2008). Such a
screeningtoolwouldprovideacost-effectivemeanstoobtain
information on the condition of all lakes in a region.
It is even more challenging to integrate all stressors and
indicators so that overall conditions within and among lakes
can be assessed and compared. The most commonly used
index for lake condition assessment is Carlson’s Trophic
State Index (TSI; Carlson 1977). Carlson’s TSI is an index
of potential lake productivity and uses chlorophyll con-
centration, Secchi depth, and total phosphorus to indepen-
dently estimate algal biomass. Carlson’s TSI has limited
utility in assessing overall lake conditions because it only
measures trophic state. In addition, the TSI does not dis-
tinguish whether high algal production results from natural
or human-derived sources and can only be used in lakes that
have few rooted aquatic plants and little nonalgal turbidity.
More recently, considerable effort has been devoted to
developing biological indexes because they are believed to
better represent overall lake conditions. Efforts have been
made to develop indexes of biotic integrity for lakes using
fish (e.g., Drake and Pereira 2002; Lyons and others 2000;
Minns and others 1994; Schulz and others 1999), macro-
invertebrates (e.g., Lewis and others. 2001; Ma and others.
2008), macrophytes (Melzer 1999; Nichols and others
2000), and diatoms (e.g., Dixit and Smol 1994; Stenger-
Kova ´cs and others 2007). The success of these biotic
indexes in accurately depicting overall lake condition has
been mixed when compared to alternative measures of lake
condition (e.g., Drake and Valley 2005; Schulz and others
1999). Because each index assesses different aspects of lake
impairment and reference conditions need to be identified
for each lake type, the interpretation of assessment results
from different indicators for comparisons among lakes is
often challenging. Hence, an index that integrates all of the
measured indicators and that reflects the overall physical,
chemical, and biological conditions of lakes, without bias
toward one or the other, is needed.
Identifying dominant individual disturbance factors that
determine lake condition is essential for directing man-
agement efforts toward impairment remediation. Although
biological indexes are believed to integrate the effects of
all disturbance sources, types, and pathways for river sys-
tems (Fausch and others 1990; Karr and Chu 1999), con-
nections between biological indexes and specific human
disturbances are complex and poorly understood. This lack
of understanding makes it difficult to pinpoint sources of
472Environmental Management (2010) 46:471–483
123

Page 3

ecosystem change and to prescribe preventive or restor-
ative management actions (Norris and Hawkins 2000; Suter
and others 2002; USEPA 2000). Therefore, it would be
highly desirable to develop an assessment approach that
could identify sources of degradation for each lake.
Most lake condition assessments have focused on in-lake
physicochemical and biological conditions because of the
unavailabilityoflarge-scalehuman-disturbancedataandthe
resources required to delineate lake tributary and local
catchment boundaries. As the availability of regional dat-
abases and the development of geographic information
technologieshaveincreased,usinglandscapedisturbancesto
directlyassessaquaticsystemconditionhasbecomefeasible
and cost-effective (Danz and others 2007; Mattson and
Angermeier 2007; Wang and others 2006a, 2008). Land-
scape-based assessments are attractive because they rely on
thenaturallydefinedhydrologiccatchmentastheintegrating
management unit. Consequently, landscape-based assess-
ments can be used to assess lake resources as well as the
streams and rivers that flow into lakes and the lands that
contribute water to the lakes. Thus, landscape-based
assessments provide an integrated, holistic strategy for pro-
tecting and managing aquatic resources and achieving
broaderenvironmentalprotectionobjectives(USEPA2003).
In this study, we adapted for lakes in Michigan an
approach that was developed for streams (Wang and others
2008) for quantifying human-disturbance gradients and
identifying key disturbance factors. This approach incor-
porates natural environmental variability of landscapes at
multiple spatial scales and uses readily available anthro-
pogenic disturbances and lake-specific physicochemical
and biological measures to link levels of human distur-
bance with those physicochemical and biological changes
to assess lake conditions. Our specific objectives were to
(1) compare the relative importance of landscape distur-
bances at tributary and local catchment scales in influ-
encing in-lake indicators, (2) assess human-disturbance
gradients for all lakes that are 2 ha (5 acres) or larger in the
state of Michigan based on observed relationships between
landscape human disturbances and in-lake physicochemical
and biological measures, and (3) determine major sources
of degradation for lakes that are moderately to heavily
impacted and identify potential threats for lakes that are
least impacted.
Methods
Lake Polygons and Their Associated Catchment
Boundaries
We first identified all natural and man-made lake polygons
that are 2 ha or larger from the 1:24,000 National
Hydrography Dataset for the entire state of Michigan using a
geographic information system (GIS; ESRI 2002). We then
delineated tributary and local catchment boundaries for all
lakes.Wedefinedatributarycatchmentasthelandareawhere
surface water drains directly into rivers and then into a lake.
We defined a local catchment as the land area where surface
runoffdrainsdirectly into a lake. Catchmentboundarieswere
delineated using GIS algorithms to identify runoff directions
basedon30-mresolutionDigitalElevationModel(DEM)and
to restrict the outmost catchment boundaries using a 12-digit
Hydrological Unit (HU) or aggregated HUs that were devel-
oped by Michigan Department of Natural Resources and
Environment (MDNRE).
We calculated several measures of lake landscape
position and morphometry that are known to strongly
influence lake physicochemical and biological character-
istics (Martin and Soranno 2006). These measures included
lake order calculated as the stream order of the largest
tributary flowing into each lake, total number of lakes in
the tributary catchment of each lake, total number of lakes
downstream between each lake and the Great Lakes (all
rivers in Michigan flow into the Great Lakes), tributary and
local catchment areas, shoreline development index
[D = L/(2(pA)?], where L is the perimeter and A is lake
area), and lake fetch (length of the longest unobstructed
distance across a lake). We identified lakes based on
hydrologic connectivity, including both perennial and
intermittent streams. Inline lakes were defined as having
both inflows and outflows in a river network; headwater
lakes were defined as having only outflows; and discon-
nected lakes were defined as having no inflows or outflows.
Natural and Human-Disturbance Landscape Variables
We calculated natural and human disturbance measures in
tributary and local catchments for lakes from databases of
various sources. For natural landscape measures, non-
human-disturbance land-cover types were measured from
2001 Michigan Land Use/Cover Data (http://www.mcgi.
state.mi.us/mgdl). Surficial geology texture and formation
types were calculated from the Michigan Quarternary
geology geographic theme under geology (http://www.
mcgi.state.mi.us/mgdl/?rel=ext&action=sext). Soil perme-
ability was calculated based on the US Geological Survey’s
soils data for the conterminous United States derived from
the Natural Resource Conservation Service state soil geo-
graphic (STATSGO) database (http://water.usgs.gov/GIS/
dsdl/ussoils_04.e00.gz). Annual and July mean air tem-
peratures were obtained from the Oregon State University/
Spatial Climate Analysis Service for the conterminous
United States(www.climatesource.com/us/fact_sheets/
fact_tmean_us.html). Annual growing degree days and
annual precipitation were also obtained from the Oregon
Environmental Management (2010) 46:471–483473
123

Page 4

State University/Spatial Climate Analysis Service for the
conterminous United States (www.climatesource.com/us/
fact_sheets/fact_gdd50f_us.html; www.climatesource.com/
us/fact_sheets/fact_precip_us.html). Lake elevation and
catchment slopes were calculated based on the DEM using
GIS.
For human-disturbance measures (Table 1), urban and
agricultural land uses were determined from 2001 Michigan
Land Use/Cover Data (http://www.mcgi.state.mi.us/mgdl).
Percent of catchment area in urban and agriculture and ratio
of catchment area in urban and agriculture to lake area were
computed for all tributary and local catchments. Length of
roads and number of road crossings in the catchment per
catchment area and per lake area were calculated using the
Michigan Geographic Data Library roads (http://www.
mcgi.state.mi.us/mgdl/?rel=ext&action=sext, under Trans-
portation). Numbers of residents in the catchment per
catchment area and per lake area were calculated using
2000 census data (http://www.mcgi.state.mi.us/mgdl/?rel=
ext&action=sext, under the Census). Yields of total nitro-
gen and total phosphorus were calculated as kilograms per
catchment area and per lake area per year from data that was
predicted by the US Geological Survey (USGS) using the
Spatially Referenced Regressions on Watershed Attributes
model (Smith and others 1997; http://water.usgs.gov/
nawqa/sparrow/wrr97/results.html). The total number of
permitted discharge facilities in each catchment and the
number of permitted discharge facilities directly connected
to streams in each catchment were obtained from the
MDNRE unpublished database and were calculated as the
number of facilities per catchment area and as the number
of facilities per lake area. The total number of toxic release
sites in the catchment and the number of toxic release sites
directly connected with streams were obtained from the
US EPA’s toxic release inventory (http://www.epa.gov/
tri/tridata/index.htm) and were calculated as the number
of facilities per catchment area and per lake area. The
proportion of catchment area and the ratio of catchment
area to lake area that were treated with fertilizers, herbi-
cides, insecticides, and manure were calculated using the
US Department of Agriculture (USDA)’s 2002 Census
of Agriculture database (http://www.agcensus.usda.gov/)
and the USDA’s 2002–2005 Performance Results System
(http://ias.sc.egov.usda.gov/prshome/default.html). Numbers
of active mining sites per catchment area and per lake area
were calculated from USGS’s mineral resource database
(http://tin.er.usgs.gov/mineplant/).
In-Lake Variables
Fish assemblage composition data were from the MDNRE
Fish Collection System (MDNRE unpublished data). We
selected lakes with fish data collected using gill nets, trap
nets, beach seines, and electrofishing following the stan-
dard MDNRE status and trends protocol (Wehrly and
others 2010). There were 287 lakes with fish data that met
our data selection criteria. Using these fish data, we cal-
culated the index of biotic integrity (IBI) following the
method developed for Minnesota lakes (Drake and Pereira
2002). The IBI consisted of 16 metrics, including 2 metrics
calculated from gill nets, 3 from trap nets, 3 from near-
shore gear (seining and eletrofishing), and 8 species-rich-
ness metrics from a combination of all gear types. Because
the Minnesota lake IBI was developed for lakes less than
202 ha, its ability to determine fish biotic condition is
unknown for lakes larger than 202 ha in our dataset.
Despite this uncertainty, we used this IBI because it is the
only validated IBI version available for Midwestern lakes
and because the majority of Michigan lakes used in our
analysis are smaller than 202 ha.
Lake indicator and stressor data were obtained from
various sources, including the USGS’s Water Data for the
Nation, data maintained by MDNRE, MDNRE Fish Col-
lection Systems, and the Institute for Fisheries Research
Water Atlas Database (Breck 2004). From these databases,
we selected 750 lakes having total nitrogen, total phos-
phorus, chlorophyll-a, Secchi depth data, and DO profiles
that were collected during the lake stratification period
(July, August, and September). The amount of hypoxia and
anoxia in each lake were estimated as the depth and pro-
portion of the water column having oxygen concentrations
below 4.0 mg/L and 0.5 mg/L during the stratification
period, respectively.
In-lake physical habitat-disturbance data included per-
cent of shoreline armoring (sea walls, riprap, and other
man-made structures), number of houses per kilometer of
shoreline, number of docks per kilometer of shoreline, and
number of submerged trees that were 7.6 cm or larger in
diameter per kilometer of shoreline. These measures were
visually estimated using standard lake survey protocol
(Wehrly and others 2010) from 2002 to 2009.
Lake mean depth and maximum depth were obtained by
digitizing bathymetry maps. Lake ice-free days and mean
and maximum ice-free surface water temperatures were
modeled based on the relationship among mean annual air
temperature, fetch, lake area, and measured surface water
temperature for a subset of lakes (Breck unpublished data;
Shuter and others 1983).
Data Analyses
We organized our data into two datasets. The first dataset
included all lakes in Michigan that were 2 ha or larger and
the second was a subset of these lakes that had in-lake
variables. The subset-lake dataset consisted of 152 various
size lakes distributed across the state (Fig. 1). The criteria
474 Environmental Management (2010) 46:471–483
123

Page 5

Table 1 Mean, median, range, and standard deviation of human-disturbance factors for lakes that are greater or equal to 2 ha in Michigan, from
which lake human disturbance was assessed
Disturbance variablesMeanMedianRange Standard
deviation
Agricultural disturbance variables
Cattle density (No./km2catchment)
Cattle density (No./km2lake area)
400–566
4280 0–110,041 2,888
Manure, pesticide, herbicide application (% of catchment)
Manure, pesticide, herbicide application (km2catchment/km2lake area)
171 0–204 31
2173 0–17,636808
Other agriculture (% of catchment)
Other agriculture (km2catchment/km2lake area)
101 0–100 15
10434 0–125,0245,370
Rowcrop (% of catchment)
Rowcrop (km2catchment/km2lake area)
Total nitrogen yield (kg/km2/year)
Total nitrogen yield (1000 kg/km2lake area/year)
Total phosphorus yield (kg/km2/year)
Total phosphorus yield (1000 kg/km2lake area/year)
60 0–10013
8830 0–278,1816,871
6524280–2,900 461
656 0–16,106374
48230–692 69
50 0–1,80837
Urban variables
Commercial–industrial (% of catchment)
Commercial–industrial (km2catchment/km2lake area)
Human population (No./km2catchment)
Human population (No./km2lake area)
100–1005
5800–18,397 495
5816 0–3,979 157
5,322 1670–1,637,44342,639
Imperviousness (% of catchment)
Imperviousness (km2catchment/km2lake area)
20 0–606
18900–59,713 1,597
Other urban (% of catchment)
Other urban (km2catchment/km2lake area)
00 0–512
260 0–15,169265
Resident (% of catchment)
Resident (km2catchment/km2lake area)
100–604
11840–47,403 916
Road–park (% of catchment)
Road–park (km2catchment/km2lake area)
310–905
202 15 0–67,0961,229
Shoreline house density (No./km lake shoreline)1010 0–518
Point-source variables
Mines (No./km2catchment)
Mines (No./km2lake area)
US EPA’s toxic release inventory sites do not drain to surface water (No./km2catchment)
US EPA’s toxic release inventory sites do not drain to surface (No./km2lake area)
US EPA’s toxic release inventory sites draining to surface (No./km2catchment)
US EPA’s toxic release inventory sites draining to surface (number/km2lake area)
MDEQ’s permitted point-source facilities do not drain to surface water (No./km2catchment)
MDEQ’s permitted point-source facilities do not drain to surface water (No./km2lake area)
MDEQ’s permitted point-source facilities draining to surface (No./km2catchment)
MDEQ’s permitted point-source facilities draining to surface (No./km2lake area)
00 0–481
000–691
000–3274
100–1926
000–1231
000–431
100–1,11117
10 0–1,66623
400–2,15442
200–97024
Other disturbance variables
Catchment runoff (cm/year)
Catchment runoff (dm3/year/km2lake area)
333318–538
20.8 2.40–19,373249.3
Chlorine–sulfur atmospheric deposition (1000 kg/ha)
Road density (km/km2catchment)
Road density (km/km2lake area)
Road crossings (No./km2catchment)
Road crossings (No./km2lake area)
129150–39,654 731
1400–11,256 244
1475 0–42,940 838
000–1322
3200–13,419 246
Environmental Management (2010) 46:471–483 475
123