QUANTIFYING LITTORAL ZONE SUBSTRATE DISTRIBUTION AND COARSE WOODY HABITAT ABUNDANCE USING LOW-COST SIDE-SCAN SONAR Christine A. Koeller College of Natural Resources University of Wisconsin-Stevens Point A Thesis Submitted in partial fulfillment of the Requirements of the degree 1
MASTER OF SCIENCE COLLEGE OF NATURAL RESOURCES UNIVERSITY OF WISCONSIN AT STEVENS POINT Stevens Point, Wisconsin August, 2014 2
APPROVED BY THE GRADUATE COMMITTEE OF: Dr. Ron Crunkilton, Committee Chairman Professor of Water Resources, College of Natural Resources UWSP Dr. Justin Sipiorski, Assistant Professor of Biology College of Natural Resources, UWSP Dr. Keith Rice, Professor of Geography College of Letters and Science, UWSP 3
CHAPTER 1: LAKE HABITAT MAPPING WITH SIDE-SCAN SONAR IN WISCONSIN LAKES... 13 ABSTRACT... 13 INTRODUCTION... 14 Substrate Distribution Defined... 16 Substrate Distribution Significance... 17 Habitat Mapping: Alternatives to Side-Scan... 18 Overview of Side-Scan Sonar... 19 Applications of Side-Scan Sonar... 19 METHODS... 22 Habitat Survey Methods... 24 Ground-truth Verification... 26 RESULTS... 28 DISCUSSION... 33 CHAPTER 2: FISH SPECIES RICHNESS AND LAKE HABITAT EVALUATION IN ELEVEN WISCONSIN LAKES... 36 INTRODUCTION... 36 Historical overview... 38 METHODS... 41 SITE DESCRIPTION... 41 Habitat Survey Methods... 44 Habitat Ground-truth Verification... 44 Fish Survey Methods... 48 4
RESULTS... 56 DISCUSSION... 72 CHAPTER 3: A COMPARISON OF HISTORICAL CHANGES IN LAKE MORPHOLOGY OF SIX INLAND WISCONSIN LAKES... 77 ABSTRACT... 77 INTRODUCTION... 77 Study Area... 79 METHODS... 79 RESULTS.86 DISCUSSION... 89 REFERENCES... 91 APPENDIX A... 96 APPENDIX B... 1118 5
TABLE OF FIGURES Figure 1: Map of study area in eastern Marathon County, Wisconsin consisting of 9 lakes with surface area ranging from 17 to 82 hectares.... 23 Figure 2: Map of study area in Eastern Marathon County, Wisconsin consisting of 11 lakes with surface area ranging from 10 to 82 ha.... 43 Figure 3: Boom-shock spatial locations for Pike, Mission, and Wadley lakes with total number of shock observations.... 55 Figure 4: Relationships between species richness (Y) and lake physical characteristics of maximum depth and volume. Reported values are coefficient of determination (R 2 ).... 65 Figure 5: Relationships between species richness (Y) and shoreline development (docks per mile). Reported values are coefficient of determination (R 2 ).... 65 Figure 6: Six study lakes positioned in eastern Marathon County, WI.... 81 Figure 7: Rectified WI DNR 1967 bathymetry map of Mayflower Lake (Marathon County, WI) over present-day aerial imagery (NAIP, 2010).... 82 Figure 8: Trimble survey equipment (UWSP-GIS Center) attached to the transom of a 6- horsepower Jon boat (UWSP CWSE) during Pike Lake bathymetry survey, Marathon County, WI (2012)... 84 Figure 9: Grid transects and near-shore perimeter XYZ data collected on Lost Lake, Marathon County WI (2012)... 84 Figure 10: TIN surface of Wadley Lake, Marathon County, WI (2012)... 85 Figure 11: Square 0.4 hectare grid with one random point per polygon area on Big Bass Lake, Marathon County, WI (2012).... 85 6
Figure 12: Relationship between lake surface area of 2012 bathymetry surveys and change in total lake volume from historic documented surveys.... 87 Figure 13: Spatial view of morphological lake-bottom between historic WI DNR maps and present-day surfaces.... 88 Figure 14: Frequency (n) at Length (cm) histograms for Lilly Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods.... 96 Figure 15: Frequency (n) at Length (cm) histograms for Mayflower Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods.... 97 Figure 16: Frequency (n) at Length (cm) histograms for Big Bass Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods.... 98 Figure 17: Frequency (n) at Length (cm) histograms for Rice Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods.... 99 Figure 18: Frequency (n) at Length (cm) histograms for Lost Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods.... 100 Figure 19: Frequency (n) at Length (cm) histograms for Pike Lake in Marathon County, WI with varying sample gear (fyke net, seine, shock). Refer to Table 3 for survey periods.... 101 Figure 20: Frequency (n) at Length (cm) histograms for Mission Lake in Marathon County, WI with varying sample gear (fyke net, seine, shock). Refer to Table 3 for survey periods... 103 Figure 21: Frequency (n) at Length (cm) histograms for Wadley Lake in Marathon County, WI with varying sample gear (fyke net, seine, shock). Refer to Table 3 for survey periods... 105 Figure 22: Frequency (n) at Length (cm) histograms for Norrie Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods.... 107 7
Figure 23: Frequency (n) at Length (cm) histograms for Mud Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods.... 109 Figure 24: Frequency (n) at Length (cm) histograms for Bass Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods.... 110 TABLE OF TABLES Table 1: Size (hectare) and perimeter (kilometer) of nine study lakes in eastern Marathon County, WI ranging in size from 19-84 hectare and 1.9-4.5 km of shoreline.... 25 Table 2: Substrate classifications and descriptions for eastern Marathon County lake substrate maps, modified from the Wisconsin Department of Natural Resources critical habitat designation manual (Cunningham 2008).... 25 Table 3: Total near-shore habitat out to 30.5m available for side-scan sonar scanning in each lake and total interpreted near-shore area, substrate points, and coarse woody habitat plots that were attained in 2012 field season.... 25 Table 4: Bottom substrate percent cover by lake in 2012. Surveys of in-lake habitat cover the near-shore area extending out 30.5 meters from the shoreline.... 29 Table 5: Error matrix for substrate map classification values in nine eastern Marathon County study lakes (2012). Interpreted accuracy is total verified as true during ground-truth divided by the total random sample locations per category. User s accuracy is determined from total true per category divided by total found in category among all random sites. Overall accuracy equals total number correct divided by the total number of random sites.... 30 8
Table 6: Error matrix for coarse woody habitat map classification values in nine Marathon County study lakes (2012). Interpreted accuracy is total verified as true during groundtruth divided by the total random sample locations per category. User s accuracy is determined from total true per category divided by total found in category among all random sites. Overall accuracy equals total number correct divided by the total number of random sites.... 31 Table 7: Interpreted coarse woody habitat (CWH) logs per kilometer values in nine eastern Marathon County study lakes (2012).... 32 Table 8: Size (hectare) and perimeter (kilometer) of eleven study lakes in eastern Marathon County, WI.... 42 Table 9: Substrate classifications and descriptions for Eastern Marathon County lake substrate maps.... 47 Table 10: Total near-shore habitat out to 30.5m available for side-scan sonar scanning in each lake and total interpreted near-shore area, substrate points, and CWH plots that were attained in 2012 field season.... 47 Table 11: Fish species preferences of substrate, vegetation, woody habitat, depth (m), and water clarity of all expected fish species in the study area (Becker, 1983).... 49 Table 12: Survey periods by gear type (fyke net, seine, boom shock) of eastern Marathon County lakes in 2011 to 2012.... 51 Table 13: Forward stepwise regression results for measured habitat variables (explanatory variables) versus fish diversity (response variable) (p-value for entry = 0.25) relationship strength (r 2 ) and significance (P).... 58 9
Table 14: Forward stepwise regression results for measured habitat variables (explanatory variables) versus fish species richness (response variable) (p-value for entry = 0.25) relationship strength (r 2 ) and significance (P)... 58 Table 15: Fish species occurrence in the 2011 and 2012 study area, historical record, and WDNR stocking records where P = present in the current study, H = present historically, and S = stocked previously.... 59 Table 16: Average length per species per lake in fyke nets during the survey period. Note survey periods varied among lakes, refer to Table 3 for survey dates.... 60 Table 17: Average length per species per lake in seines during the survey period. Note survey periods varied among lakes, refer to Table 3 for survey dates.... 61 Table 18: Average length per species per lake during the boom shock survey period. Note survey periods varied among lakes, refer to Table 3 for survey dates.... 61 Table 19: Length range and regression summary per species per lake (n>25) with fyke net and seine gear (Y-intercept, Slope, Standard Error reported).... 62 Table 20: Species richness in the 2011 and 2012 study, species previously documented in WDNR records, newly documented species in 2011 and 2012, lost or undocumented species, total species richness (past/present), and percentage of species lost or undocumented.... 66 Table 21: Bottom substrate percent cover by lake in 2012 surveys of near-shore, in-lake habitat out to 30.5 meters of shoreline.... 68 Table 22: Error matrix for substrate map classification values in nine Marathon County study lakes (2012). Interpreted accuracy is total verified as true during ground-truth divided by the total random sample locations per category. User s accuracy is determined from total 10
true per category divided by total found in category among all random sites. Overall accuracy equals total number correct divided by the total number of random sites.... 69 Table 23: Error matrix for coarse woody habitat map classification values in nine Marathon County study lakes. Interpreted accuracy is total verified as true during ground-truth divided by the total random sample locations per category. User s accuracy is determined from total true per category divided by total found in category among all random sites. Overall accuracy equals total number correct divided by the total number of random sites.... 70 Table 24: Interpreted Logs per kilometer values in nine Marathon County study lakes.... 71 Table 25: Lake area in hectares of study lakes located in Marathon County, WI.... 82 Table 26: T-test results (t0.05(two-tail),df>50) of paired data on each lake. Results show significant differences exist between historic and present-day surfaces for each lake.... 87 Table 27: Historic versus present lake volume (acre-feet) as calculated with surface volume 3Danalyst tool in ArcMap 10.0 (ESRI 2012) mapping software.... 87 Table 28: Total catch and lengths (min/max/average) of species in Lilly Lake during the 2012 fyke net and seining surveys.... 96 Table 29: Total catch and lengths (min/max/average) of species in Mayflower Lake during 2012 fyke netting and seining efforts.... 97 Table 30: Total catch and lengths (min/max/average) of species in Big Bass Lake during the 2012 fyke netting and seining surveys.... 98 Table 31: Total catch and lengths (min/max/average) of species in Rice Lake during the 2011 fyke net survey.... 99 11
Table 32: Total catch and lengths (min/max/average) of species in Lost Lake during the 2011 fyke netting and seining surveys.... 100 Table 33: Total catch and lengths (min/max/average) of species in Pike Lake during 2012 fyke netting and seining efforts.... 102 Table 34: Total catch and lengths (min/max/average) of species in Pike Lake during the 2012 boom shocking survey.... 102 Table 35: Total catch and lengths (min/max/average) of species in Mission Lake during the 2012 fyke net and seining surveys.... 104 Table 36: Total catch and lengths (min/max/average) of species in Mission Lake during the 2012 boom shocking survey.... 104 Table 37: Total catch and lengths (min/max/average) of species in Wadley Lake during the 2012 fyke net and seining surveys.... 106 Table 38: Total catch and lengths (min/max/average) of species in Wadley Lake during the 2012 boom shocking survey.... 106 Table 39: Total catch and lengths (min/max/average) of species in Norrie Lake during 2012 fyke netting and seining efforts.... 108 Table 40: Table 2: Total catch and lengths (min/max/average) of species in Mud Lake during the 2012 fyke net survey.... 109 Table 41: Total catch and lengths (min/max/average) of species in Bass Lake during the 2012 survey... 110 12
CHAPTER 1: LAKE HABITAT MAPPING WITH SIDE-SCAN SONAR IN WISCONSIN LAKES ABSTRACT Lake-habitat features such as substrate, coarse woody habitat (CWH), depth, and vegetation are important components of the ecosystem, and are used by fish and other aquatic organisms for foraging, refuge, and spawning. Traditionally, habitat has been measured manually using quadrats and transects. Depth maps for the study area were constructed without GPS/Sonar technology. Side-scan sonar technology has more recently been used to map underwater habitat features such as substrate and coarse woody habitat; however, this technology has not been employed for habitat mapping of inland Wisconsin lakes. Our main objectives were to determine the accuracy of sonar interpretation for in-lake habitat mapping for nine Eastern Marathon County kettle lakes with glacial origin, to determine the accuracy of new depth maps created from sonar depth readings, and to determine if different habitat combinations can be used to predict fish diversity. The Lowrance HDS5 side-scan (or structure scan) was used to collect data from nine lakes in the study area, around the entire lake perimeter starting at the shoreline, in a swath 30.5 meters wide. This data was compared to ground truth data collected from random sampling points within each habitat type using an error matrix. Results of substrate interpretation accuracy was 53 percent; whereas, CWH abundance/absence interpretation accuracy was 74 percent. This method could provide a quick, economic alternative over manual estimation for fish biologists to use when measuring in-lake habitat for future projects. 13
INTRODUCTION The purpose of this study was to apply a new method for mapping lake habitat attributes of small inland lakes (<84 hectares in size) in Marathon County, Wisconsin. Physical habitat map attributes recorded in previous studies include, but are not limited to coarse woody habitat (CWH) (Kaeser and Litts 2008) or coarse woody structure (CWS) (Newbrey et al. 2005, Sass et al. 2006a), and sediment distribution (Kaeser and Litts 2010). These physical attributes are often correlated with biological assessments to establish relationships with habitat preference (Johnson 1961, Jennings et al. 1999, Lewin et al. 2004, Newbrey et al. 2005, and Sass et al. 2006a), predator-prey interactions (Sass et al. 2006b), as well as population dynamics and food web responses (Sass et al. 2006b). Habitat characterization is also important in identification of critical habitat areas for aquatic biota (Cunningham 2008) and has been used to develop habitat fingerprints (Schmidt 2010). Side-scan sonar (SSS) is an innovative technology that offers continuous bottom coverage information that may replace traditional, manual methods of habitat mapping. It has been used to detect features in the underwater environment including sediment distribution (Kaeser and Litts 2010, Anima et al. 2007, Bates and Oakley 2004), boundaries of submerged aquatic vegetation such as seagrass beds (Ardizzone et al. 2006), and CWH (Kaeser and Litts 2008). The purpose of this study was to determine if the low-cost, Lowrance side-scan sonar LSSS could be used to quantify in-lake habitat sediment distribution and CWH counts in nine inland lakes located in Eastern Marathon County, Wisconsin. Mapping lake habitat attributes such as CWH and sediment distribution has been historically accomplished with a multitude of methods including transect lines and/or quadrats (Dombeck 14
et al. 1984, Everitt and Ruiz 1993, Johnson 1961, Schmidt 2010), underwater video imaging, and more recently with low-cost side-scan sonar image interpretation (Edsall et al. 1989, Matarrese et al. 2004, Kaeser and Litts 2008, Kaeser and Litts 2010). Coarse woody habitat in this study is defined as entirely or partially submerged wood pieces with diameter 10 centimeters (McHenry et al. 1998, Newbrey et al. 2005, Kaeser and Litts 2008) and length 1.5 meters (Kaeser and Litts 2008) for practicality of CWH identification limitations when using LSSS. In previous studies, similar definitions exist with slight variations in naming including coarse woody habitat (sticks and logs with >5 centimeter diameter) (Sass et al. 2006b), coarse woody structure (tree or tree parts with 10 centimeter diameter anywhere along their length) (Newbrey et al. 2005), and coarse woody debris (pieces of wood with 2-5 centimeter diameter or larger with branching) (Everitt and Ruiz 1992, Christensen et al. 1996). Coarse woody habitat has been previously studied because it is one of the most important attributes of the aquatic ecosystem that influences the distribution of aquatic biota (Everitt and Ruiz 1993, Sass et al. 2006b). Subsequently, CWH is also attractive to predators as foraging areas in lakes (Newbrey et al. 2005). Coarse woody debris with complex branching and increased surface area may lead to greater insect density and diversity (Schmude et al. 1998). Epibenthic fish and shrimp abundances in a Maryland river were strongly influenced by CWH (Everitt and Ruiz 1992). Common Perch (Perca fluviatilis) and common Roach (Rutilis rutilus) were found to prefer woody structured habitats over reed both diurnally and nocturnally. Both species had increased abundances with CWH structural complexity (Lewin et al. 2004). Aquatic organism community structure and diet can be altered by the removal of CWH (Sass et al. 2006a, Sass et al. 2006b). When the majority of CWH was removed from the treatment 15
basin on Little Rock Lake in Wisconsin, Yellow Perch (Perca flavescens) populations rapidly declined and Largemouth Bass (Micropterus salmoides) diet changed from predominantly perch (>60%) to a diet with an average of 14% perch and 51% to 55% terrestrial prey (Sass et al. 2006a). Additional research showed prey fish including Yellow Perch, Pumpkinseed (Lepomis gibbosus), Bluegill (Lepomis macrochirus), and the Fathead Minnow (Pimephales promelas) had decreased predation risks when located within CWH habitat areas (Sass et al. 2006b). It has been shown that as human structural development around a lake increases the amount of CWH significantly decreases, which could negatively impact lake ecosystems (Christensen et al. 1996). Inputs and decay rates of CWH are altered when development increases (Christensen et al. 1996). Intolerant fish species [darters Etheostoma spp., Mottled Sculpin Cottus bairdii, Smallmouth Bass (M. dolomieu), and Rock Bass (Ambloplites rupestris)] in Wisconsin lakes were found in significantly greater abundances in less disturbed lakes, and were rare or not present in developed lake systems (Jennings et al. 1999). Substrate Distribution Defined Substrate Classification schemes used in categorizing substrate types for lake studies are varied. Some have modified the Wentworth scale for substrate size when locating optimum and good trout spawning habitat in northern Lake Michigan (Edsall et al. 1989). The WI DNR substrate categories include marl, detritus, clay, silt, sand, fine gravel, coarse gravel, rubble/cobble, small boulder, large boulder, and bedrock (Cunningham, 2008). Saunders et al. (2002) classified substrate as fine organic material, silt, sand, gravel, cobble, rubble, small boulders, large boulders, and bedrock. Other classifications include sandy, rocky fine, rocky boulder, limerock fine, and limerock boulder (Kaeser and Litts 2010). Defining substrate classes is project specific 16
and is determined by the questions being asked. Herein we adopted the substrate categories of muck, sand, gravel, cobble, mixed (sand/gravel, sand/gravel/cobble, gravel cobble, sand/cobble), and boulder (Table 2). Substrate Distribution Significance Mapping substrate in lakes has many practical applications for resource managers. Spawning and cover habitat for fishes is often controlled by distributions of substrate (Cunningham, 2008). For example, Walleye (Sander vitreus) prefer to spawn on gravel versus sand areas; however, if gravel areas are not available Walleye will utilize muck and sand areas, but egg survival rates are best on gravel-rubble bottoms (Johnson 1961). Smallmouth Bass construct nests with gravel and flat rocks, and it has been shown that nests built with more uniform substrate particle diameter and are more likely to be spawned (Winemiller and Taylor, 1982). More recent work identified gravel substrate abundance (>40% in 1m 2 ) as a key feature to modeling successful Smallmouth Bass nesting habitat in two Wisconsin lakes (Bozek et al. 2002). Some submersed aquatic vegetation growing patterns are correlated with bottom substrate type. For example, muskgrass (Chara spp.) was associated with peat sediments in a shallow Florida lake (Havens et al. 2002). Eurasian watermilfoil (Myriophyllum spicatum) (EWM), a nuisance aquatic invasive, prefers moderately dense, fine-textured substrate and conversely does not grow well on coarse substrates such as sand and gravel (Smith and Barko, 1990). Curly-leaf pondweed (Potamogeton crispus), another nuisance aquatic invasive, prefers soft substrates (Nichols, 1999). Substrate preferences associated with aquatic plant growth patterns can be used to monitor for early detection of invasive plants in lakes. Resource 17
managers use information about substrate in a wide variety of practical management activities and have a continued need to employ time-efficient means of obtaining that information in the field. Habitat Mapping: Alternatives to Side-Scan There are many different methods for collecting and quantifying lake habitat information. The Wisconsin Department of Natural Resources (WI DNR) has guidelines for identifying critical habitat areas in Wisconsin lakes which includes CWH sampling and measurements using transect methods (Cunningham, 2008). This transect method requires intensive evaluation of critical area features including log length, diameter, orientation, and branchiness which requires a lot of man-hours (Cunningham, 2008). In addition, the WI DNR uses labor-intensive transect sampling, pebble counts, and ocular estimates to map substrate distribution (Cunningham, 2008). The amount of effort required to assess habitat can be reduced by reducing the number of transects surveyed, but with a loss of accuracy when quantifying whole-lake habitat. Furthermore, Schmidt (2010) states than many man-hours would still be required to map substrate. Prior to the combination of GPS location information with side-scan imaging, side-scan images had to be manually laid out and trimmed by hand before interpretation was possible which required additional man-hours (Edsall et al. 1989). Another shortfall of subsampling by using transects is that coverage is not continuous and certain important substrate features may be missed. Continuous habitat coverage mapping via traditional methods of transects and quadrats is usually impractical due to time and money constraints. 18
Overview of Side-Scan Sonar Previous methods used to map lake habitat do not collect continuous information over an area of interest as does side-scan sonar. Side-scan uses sonar technology which measures soundwave travel time. Sonar waves reflect differently off surfaces with different densities, and sidescan sonar can usually detect those differences under water. Side-scan images are a collection of stacked line image swaths oriented perpendicular to the boat path that are processed but not rectified; each line represents a scan of data collected from one ping (Cervenka and de Moustier 1993). Side-scan images resulting from a stacked collection of multiple pings are distorted for several reasons including interference (noise) from other devices, deviating from a straight path or incidental change in boat heading, and beams that are not uniform across the bottom (Cervenka and de Moustier 1993). To correct for distortion, side-scan images are processed, rectified, and mosaicked together, then laid out for manual interpretation with a variety of software or algorithms such as SonarWeb Pro (Bates and Oakley 2004), Coda (Ardizzone et al. 2006), Dr. Depth 4 (BT+SS Version), and SonarTRX v10.13 (Leraand 2012). Others have developed their own processing technique to spatially reference and display sidescan images (Kaeser and Litts 2008, Cervenka and de Moustier 1993). Validating SSS interpretation can be accomplished by manual ground truthing (Kaeser and Litts 2008, Kaeser and Litts 2010) or with the use of an underwater video camera (Edsall et al. 1989, Matarrese et al. 2004). Applications of Side-Scan Sonar Side-scan sonar may offer researchers an alternative to expensive, labor-intensive, manual methods for identifying and mapping key habitat features such as substrate type and woody 19
habitat. Side-scan sonar was used to locate coarse woody habitat, including deadhead logs, in streams of Southwest Georgia (Kaeser and Litts 2008). Although reviewers of SSS images underestimated actual CWH presence by approximately 50 percent, the underestimations were consistent and highly correlated with CWH counts from ground truth surveys. Kaeser and Litts 2008 suggest time spent in the field quantifying CWH could be dramatically reduced (29 manhours per kilometer with field surveys versus 2.5 man-hours per kilometer using SSS image interpretation). A follow-up study (Kaeser and Litts 2010) revealed how low-cost SSS can be used to map stream substrate into five distinct classes. Stretches of streams in southwest Georgia were imaged with a Hummingbird 900-series side-scan (Humminbird, 678 Humminbird Lane, Eufaula, AL 36027). Images were then mosaicked together and rectified to create a georeferenced continuous surface. Manual digital interpretation of the georeferenced imaging yielded map accuracies around 75 percent. Researchers at the University of Florida (UFL) are currently developing a new technique that uses side-scan image and depth information to estimate aquatic plant volumes. Algorithms are currently being developed to process raw data collected from a Lowrance LSSS into plant volumes that are displayed visually on lake maps, showing concentrated, dense growth information that aquatic plant managers would consider using for plant volume controls (Mike Netherland, personal communication, November, 2011). The UFL is hopeful this technology will successfully detect plant volume changes, between before and after applied treatments to control plant growth, that are greater than five percent. 20
Low-cost side-scan sonar technology has not been used to evaluate habitat information in inland lakes of northern Wisconsin. The objective of this study is to determine how well LSSS evaluates CWH abundance and substrate sorting for nine lakes in eastern Marathon County, and to identify if correlations exist between habitat classes and fish species diversity. Based on previous studies, we believe interpretation of LSSS can yield accurate substrate information and correlated CWH density estimates for our study area. 21
METHODS There were nine lakes in the study area located in Eastern Marathon County, Wisconsin (Figure 1). These surface waters fall into one of two distinct river basins: The Wisconsin River basin and the Wolf River Basin. These lakes are predominantly located in the eastern third of the county in the towns of Bevent, Norrie, Elderon, and/or Reid. The lakes are kettle lakes in hilly topography and sandy soils that were deposited as a result of glacial till. Surface water in this region is primarily groundwater fed with surface runoff inputs often originating nearby. The lakes include seven seepage lakes: Bass, Big Bass, Lost, Mayflower, Mission, Mud, Norrie, and Wadley Lake, two groundwater drainage lakes, Lilly and Rice Lakes, and one drainage lake, Pike Lake. Surface areas of the study lakes range from 10 to 82 hectares (Table 1) with shorelines ranging from 1.8 to 4.5 kilometers. The nine study lakes were mapped for in-lake habitat features using LSSS. Littoral bottom habitat was sampled with LSSS to create maps of bottom substrate classes and CWH abundance (log diameter >10 centimeters and length >1.5 meters) extending from the shoreline to a range of 30.5 meters. The following substrate classes are modifications to classes found in the Wisconsin DNR critical habitat designation manual (Cunningham 2008): muck, sand, gravel, cobble, mixed (sand/gravel, sand/gravel/cobble, gravel cobble, sand/cobble), and boulder (Table 2). Classifications were determined by interpreter ability to distinguish between different substrate types. Previous work shows side-scan sonar images can be successfully interpreted and mapped into substrate classes and CWH abundance in river systems (Kaeser 22
Figure 1: Map of study area in eastern Marathon County, Wisconsin consisting of 9 lakes with surface area ranging from 17 to 82 hectares. 23
and Litts 2008; Kaeser and Litts 2010). This study brought LSSS to inland lakes located in Central Wisconsin. Sonar video imagery collected in these study lakes was georeferenced using SonarTRX v10.13 (Leraand 2012) and spatially displayed using Esri s ArcGIS 10.0 mapping software. Habitat substrate was manually interpreted where images were obtainable and interpretable (Table 3) inside the 30.5 meter range of shoreline. A minimum mapping unit (mmu) of 78.5 square meters (a circle with 5 meter radius) was used to delineate substrate habitat (Kaeser and Litts 2010). Coarse woody habitat was also manually interpreted and abundance was calculated as logs per kilometer of shoreline. Ground truth sampling locations for substrate type were randomly assigned using ArcMap 10.0 (ESRI 2012). All similar habitat classes were combined among lakes, buffered by 4 meters internally to account for GPS positional error, and assigned 40 approximately random sampling points (Table 3). Random points were equally distributed among different substrate classes despite large variations in percent cover between classes. Habitat Survey Methods Data was collected during the 2012 sampling season (May-October) using a Lowrance HDS5 side-scan sonar. The LSSS was front-mounted to a 6hp Jon boat at approximately 15 centimeter below the water surface. The LSSS left/right viewing range varied between 15.24 and 24.38 meters. Image contrast was set between 50 and 65 percent during recording. A boat speed of 8 kilometers per hour or less was maintained and data was collected perpendicular to the shoreline at a distance of approximately 15 meters where accessible. Multiple passes around 24
Table 1: Size (hectare) and perimeter (kilometer) of nine study lakes in eastern Marathon County, WI ranging in size from 19-84 hectare and 1.9-4.5 km of shoreline. Area Perimeter Lake ha km Bass 32.3 3.1 Big Bass 72.8 3.4 Lilly 34.5 2.7 Lost 17.2 1.8 Mayflower 39.0 4.5 Mission 44.0 3.7 Mud 28.4 2.6 Norrie 40.3 2.6 Pike 83.8 4.2 Rice 10.2 2.3 Wadley 19.1 1.9 Table 2: Substrate classifications and descriptions for eastern Marathon County lake substrate maps, modified from the Wisconsin Department of Natural Resources critical habitat designation manual (Cunningham 2008). Class Acronym Description Muck M >50% particle size is <0.06mm. Sand S >50% particle size range is 0.06mm - 2mm. Gravel G >50% particle size range is 2mm - 64mm. Cobble C >50% particle size range is >64mm - 256mm. Mixed X S/G, G/C, S/C, or S/G/C combinations. Boulder B >50% particle size range is >256mm. Table 3: Total near-shore habitat out to 30.5m available for side-scan sonar scanning in each lake and total interpreted near-shore area, substrate points, and coarse woody habitat plots that were attained in 2012 field season. Lake Total near-shore Habitat Area Total interpreted Near-shore Area Percent Coverage Substrate Sampling CWH Sampling Plots ha ha % Points Wadley 5.6 5.6 100.0 3 16 Bass 8.9 8.4 94.9 21 5 Mud 7.1 6.6 92.6 18 6 Big Bass 9.8 8.1 82.5 20 8 Rice 6.7 5.4 80.2 3 7 Mission 10.4 6.7 64.3 9 1 Pike 12.6 6.8 54.0 121 4 Norrie 7.3 3.6 49.5 48 6 Mayflower 13.3 6.4 47.9 121 15 Total 81.9 57.7 * 364 68 25
each shoreline were collected to give the interpreter additional images to verify habitat features, to pick up areas missed in the first pass, or increase the extent of viewing outward to 30.5 meters. Sonar imagery was rectified using SonarTRX v10.13 (Leraand 2012) software and then spatially displayed using Esri s ArcGIS 10.0 mapping software. Habitat substrate was manually interpreted where images were obtainable and interpretable (Table 3) inside the 30.5 meter range of shoreline, and boundaries were digitized with polygon shapefiles in ArcMap 10.0 (ESRI 2012). A minimum mapping unit (mmu) of 78.5 square meters (a circle with 5 meter radius) was used to delineate substrate habitat (Kaeser and Litts 2010). Percent coverage of the desired near-shore area among lakes ranged from 48-100 percent. The total area of each substrate habitat classification was calculated in acres and percent cover evaluations were provided as pie charts for each lake with respect to substrate classes on the final habitat maps. Coarse woody habitat was manually interpreted and marked with point locations in ArcMap 10.0 (ESRI 2012). Ground-truth Verification Ground truth sampling locations for each substrate type were randomly assigned using ArcMap 10.0 (ESRI 2012). All similar habitat classes were combined among lakes, buffered by 4 meters (2 meters on either side of each boundary) to account for GPS positional error, and assigned approximately 40 random sampling points. Random points were equally distributed among different substrate classes despite large variations in percent cover between classes. Ground-truthing was completed at the randomly assigned substrate point locations described above for each lake using a combination of underwater cameras, visual observations, wading, 26
and dredging. A Trimble survey-grade GPS receiver was used to identify the point locations which employed an R6 GPS receiver and TSC2 personal data assistant. Instantaneous positional corrections were obtained from the Wisconsin Department of Transportation s Wisconsin Continuously Operating Reference Stations (WISCORS) which were accessed through a US- Cellular Blackberry phone modem. The WISCORS differential positioning corrected horizontal (latitude/longitude) positions to sub-centimeter accuracy during navigation to the points. Samples were grabbed by hand or with the use of an Ekman dredge and manually handtextured to determine grain size. At each point location, the substrate type was documented and verified against the field technician notes to ensure good agreement of substrate type. Interpreted results at the sampling points were blinded from the interpreter and field technician. All ground-truth verifications were done in the same sampling season that images were taken. Coarse woody habitat abundance was also verified with ground-truth sampling. All interpreted shorelines among all nine study lakes were grouped and split into 48.8 meter sections. Each section was designated a code of 0 to 2 depending on abundance of CWH (0 = no CWH, 1 = 1-3 logs or moderate abundance, 2 = 4+ logs or dense abundance). Forty sections of 0 and 2 codings were randomly selected for ground-truthing in the field. Using the Trimble surveygrade GPS receiver to navigate to start and end points in the field with WISCORS correction, waders and boats were used to locate logs within the first 30.5 meters of shoreline. Every log observed was measured for length and diameter using calipers and measuring tape. Logs that were too deep to be assessed with waders were omitted from measurements. A GPS location 27
was also marked for each log, compared with the interpreted results, and point locations added to the final maps. RESULTS The substrate type of the near-shore habitat (within 30.5 meters from shoreline) was assessed in each lake. The eleven lakes varied in near-shore substrate coverage, varying from soft muck (or marl) to a boulder bottom, but all lakes were dominated by softer substrates (Table 4). User interpretation accuracy of LSSS substrate habitat classification was determined using an error matrix (Table 5). A total of 364 points were assessed across the six substrate categories. Interpreted accuracy was best with boulder (77%) and worst with gravel (0%) classification. Overall accuracy was 53 percent. User interpretation accuracy of LSSS coarse woody habitat classification was also determined with an error matrix (Table 6). Sites showing absence of CWH (code = 0) were interpreted with 94% accuracy; whereas sites with dense abundances of CWH (code = 2) were interpreted with 78% interpreted accuracy. Overall accuracy was 74 percent. Coarse woody habitat abundance was calculated as logs per kilometer of shoreline mile for study lakes. Log abundance ranged from 0 to 98 logs per kilometer. 28
Table 4: Bottom substrate percent cover by lake in 2012. Surveys of in-lake habitat cover the near-shore area extending out 30.5 meters from the shoreline. Lake Muck Sand Sand/Gravel Gravel/Cobble Sand/Gravel/Cobble Cobble Boulder Sand/Cobble Gravel Scattered Wood (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) Bass 96.4 1.7 * 0.9 1.0 * * * * * Big Bass 77.8 22.2 * * * * * * * * Mayflower 82.4 6.7 1.1 3.7 5.6 * 0.5 * * * Mission 89.3 10.5 * 0.2 * * * * * * Mud 95.0 3.4 * * 0.4 * * 1.2 * * Norrie 80.3 5.2 * * 0.9 * 0.3 5.9 * 7.4 Pike 55.6 26.4 1.6 0.2 15.0 0.6 0.6 * * * Rice 100.0 * * * * * * * * * Wadley 87.1 10.5 * * 0.1 * * 2.2 0.1 * 29
Point Sampling data (field sampling) Row Column Total 45 44 36 162 39 38 364 Table 5: Error matrix for substrate map classification values in nine eastern Marathon County study lakes (2012). Interpreted accuracy is total verified as true during ground-truth divided by the total random sample locations per category. User s accuracy is determined from total true per category divided by total found in category among all random sites. Overall accuracy equals total number correct divided by the total number of random sites. Interpreted data M S G MIX CO BO Muck (M) 31 3 3 12 0 0 49 63% Sand (S) 10 29 21 34 7 15 116 25% Gravel (G) 0 0 0 5 0 0 5 0% Mixed (MIX) 4 12 10 108 23 1 158 68% Cobble (CO) 0 0 2 2 4 1 9 44% Boulder (BO) 0 0 0 1 5 21 27 78% Interpreted Accuracy 69% 66% 0% 67% 10% 55% Total User's Accuracy Overall Accuracy 53% 30
Point Sampling data (field sampling) Row Table 6: Error matrix for coarse woody habitat map classification values in nine Marathon County study lakes (2012). Interpreted accuracy is total verified as true during ground-truth divided by the total random sample locations per category. User s accuracy is determined from total true per category divided by total found in category among all random sites. Overall accuracy equals total number correct divided by the total number of random sites. 0 (no CWH) 1 (1-3 logs) 2 (4+ logs) 0 (no CWH) 29 6 6 41 71% 2 (4+ logs) 2 4 21 27 78% Column Total 31 10 27 68 Interpreted Accuracy 94% 78% Overall Accuracy 74% Interpreted Data Total User's Accuracy 31
Table 7: Interpreted coarse woody habitat (CWH) logs per kilometer values in nine eastern Marathon County study lakes (2012). CWH Lake Logs/km Bass 1.0 Big Bass 8.6 Mayflower 34.4 Mission 3.8 Mud 34.6 Norrie 53.6 Pike 1.7 Rice 0.0 Wadley 97.9 32
DISCUSSION This study attempted the use of LSSS to determine CWH abundance estimates and classifications of substrate distribution on small inland lakes located in Central Wisconsin. Previous studies show successful LSSS can detect CWH that is greater than 1 meter in length and 10 centimeters in diameter (Kaeser and Litts 2008). However, determining actual logs per kilometer is limited by factors such as log orientation, size, and blockage from other objects and bottom variation (Kaeser and Litts 2008). Vegetative cover was also problematic in this study when determining actual logs per kilometer and substrate boundaries, distorting sonar imagery and blocking habitat nearest to shore. Our approach to quantify CWH abundance as absent, moderate, or dense abundance was successful (74% overall accuracy) when compared to other LSSS work (Kaesar and Litts 2008, Kaesar and Litts 2010, Kaesar et al. 2012). This method may help determine presence/absence of CWH structure in lake systems quickly and economically. Classification of lake bottom substrate distribution was not as successful (overall accuracy = 53%) as other studies reaching interpretation accuracies of 77 to 84 percent (Kaesar and Litts 2010. Kaesar et al. 2012). Areas of lakes that were densely vegetated were difficult for obtaining clear images and most imagery was collected during mid-summer. Distinction of gravel and cobble substrates were most problematic. Sites interpreted as gravel were mostly sandy areas; whereas, sites interpreted as cobble were mostly a mixture of substrate types. It is possible gravel-sized pieces of substrate existed in interpreted gravel sites, and ground-truth methods failed to properly analyze the sites. Most ground-truthing of near-shore areas was done by hand, without the use of a sieve or lab analysis. 33
Lake-bottom hardness can be an important component to a lake s fishery. Some fish (game and non-game species) prefer soft substrates for spawning/habitat including Bullhead (Amieurus sp.), Catfish (Ictalurus sp.), Muskellunge (Esox masquinongy), and Warmouth (Lepomis gulosus). Other fish species prefer harder substrates (such as sand, gravel, or cobble) for spawning and habitat but have also been observed spawning over soft muck bottoms. These species include non-game fish (Tadpole Madtom (Noturus gyrinus), Golden Shiner (Notemigonus crysoleucas), Spotfin Shiner (Cyprinella spiloptera), Fathead Minnow (Pimephales promelas) and game-fish including Black Crappie (Pomoxis nigromaculatus), Bluegill, Largemouth Bass, Northern Pike (Esox lucius), Pumpkinseed, and Walleye. Other fish species prefer only hard bottoms for spawning and habitat (such as sand or gravel) including Bluntnose Minnow (Pimephales notatus), Central Mudminnow (Umbra limi), Rock Bass, Smallmouth Bass, Mottled Sculpin (Cottus bairdii), Yellow Perch, and darter species. Although game-fish are more desirable for angling, successful reproduction of non-game species is equally important in a lake ecosystem. To anglers, non-game fish are an essential food source to large, desirable sport fish. The continued successful reproduction of different species of fish that are already present in the system is needed to maintain fish diversity and ecosystem health. Determining bottom habitat needs (substrate coverage) for each individual lake s fish species would be advantageous for understanding ecological requirements for a sustainable fish community. Introduction of harder substrates such as gravel and/or cobble may improve fish habitat used for spawning activities in certain systems. There was little CWH observed in most of the lakes. Coarse woody habitat provides forage areas for predators (Newbrey et al. 2005), increases insect diversity (Schmude et al. 1998), 34
alters fish diet (Sass et al. 2006a), and reduces predation risks for prey fish (Sass et al. 2006b). Since the CWH input rate into aquatic ecosystems is a slow process (Guyette and Cole 1999), abrupt removal can negatively alter habitat for long-term periods. Tree-lined shores in drainage lakes, which tend to maintain more constant water levels, should be protected and trees that fall into the water should be left in place to provide cover. Some lakes had more natural shorelines and associated cover than others. Mission, Mud, Lilly, Lost, Bass, Rice, and portions of Mayflower Lake had more minimally developed shorelines than most other lakes and the shoreline areas of these lakes could serve as models of a more natural lake setting that would benefit the aquatic ecosystem as a whole. Norrie Lake CWH abundances are partially supported by years of inputs from the saw mill that once functioned near the lake where saw timber scrap was thrown into the near-shore environment. Many Wisconsin lakes continue to face developmental pressures that disturb and negatively alter littoral habitat by increasing embeddedness of substrate materials and decreasing CWH abundance (Jennings et al 2003). Pike and Big Bass Lakes have more developed shorelines and less in-lake CWH than other lakes with less development and more structure. Humans remove trees both on the landscape and in near-shore aquatic habitats, reducing CWH inputs (Christensen et al. 1996) and significantly lowering sequestration of carbon by CWH (Guyette et al. 2002). Downed trees in littoral areas represent the most permanent and often only yearround cover for fish. Fish populations in most Eastern Marathon County lakes could benefit from the addition of woody cover below the lowest reported water levels where it would remain continuously submerged. 35
CHAPTER 2: FISH SPECIES RICHNESS AND LAKE HABITAT EVALUATION IN ELEVEN WISCONSIN LAKES INTRODUCTION Lake physical attributes such as coarse woody habitat (CWH) and substrate distribution are often correlated with biological assessments to establish relationships with habitat preference (Johnson 1961, Jennings et al. 1999, Lewin et al. 2004, Newbrey et al. 2005, and Sass et al. 2006a), predator-prey interactions (Sass et al. 2006b), as well as population dynamics and food web responses (Sass et al. 2006b). Habitat characterization is also important in identification of critical habitat areas for aquatic biota (Cunningham 2008) and has been used to develop habitat fingerprints (Schmidt 2010). Lake habitat features such as substrate, coarse woody habitat (CWH), depth, and vegetation are important components of the ecosystem, and are used by fish and other aquatic organisms for foraging, refuge, and spawning (Becker 1983). Previous studies show CWH is one of the most important attributes of the aquatic ecosystem that influences the distribution of aquatic biota (Everitt and Ruiz 1993, Sass et al. 2006b). Subsequently, CWH is also attractive to predators as foraging areas in lakes (Newbrey et al. 2005). Common perch (Perca fluviatilis) prefer woody structured habitats (Lewin et al. 2004) in European lakes. Aquatic community structure and diet can be altered by the removal of CWH (Sass et al. 2006a, Sass et al. 2006b). When the majority of CWH was removed from the treatment basin on Little Rock Lake in Wisconsin, Yellow Perch (Perca flavescens) populations rapidly declined and Largemouth Bass (Micropterus salmoides) diet changed from predominantly perch (>60%) to a diet with an average of 14% perch and 51% to 55% terrestrial prey (Sass et al. 2006a). 36
Additional research shows prey fish (Yellow Perch, Pumpkinseed (Lepomis gibbosus), Bluegill (Lepomis macrochirus), and the Fathead Minnow (Pimephales promelas) predation risks decreased when located within CWH habitat areas (Sass et al. 2006b). As human structural development around a lake increases the amount of CWH significantly decreases, which could negatively impact lake ecosystems (Christensen et al. 1996). Inputs and decay rates of CWH are altered when development increases (Christensen et al. 1996). Intolerant fish species (darters (Etheostoma spp.), Mottled Sculpin (Cottus bairdi), Smallmouth Bass (Micropterus dolomieu), and Rock Bass (Ambloplites rupestris)) in Wisconsin lakes were found in significantly greater abundances in less disturbed lakes, and were rare or not present in developed lake systems (Jennings et al. 1999). Lake bottom substrate hardness is also an important habitat component. Spawning and cover habitat for fishes is often controlled by distributions of substrate (Cunningham, 2008). Walleye (Sander vitreus) prefer to spawn on gravel versus sandy areas; however, if gravel areas are not available Walleye will utilize muck and sand areas, but egg survival rates are best on gravelrubble bottoms (Johnson 1961). Smallmouth Bass construct nests with gravel and flat rocks, and it has been shown that nests built with more uniform substrate particle diameter are more likely to be spawned (Winemiller and Taylor, 1982). More recent work identified gravel substrate abundance (>40% in 1m 2 ) as a key feature to modeling successful Smallmouth Bass nesting habitat in two Wisconsin lakes (Bozek et al. 2002). Some submersed aquatic vegetation growing patterns can be correlated well with bottom substrate type. Eurasian watermilfoil (Myriophyllum spicatum) or EWM, a nuisance aquatic invasive, prefers moderately dense, fine-textured substrate and conversely does not grow well 37
on coarse substrates such as sand and gravel (Smith and Barko, 1990). Curly-leaf pondweed (Potamogeton crispus), another nuisance aquatic invasive, prefers soft substrates (Nichols, 1999). Substrate preferences associated with aquatic plant growth patterns can be used to monitor for early detection of invasive plants in lakes. Eleven lakes in eastern Marathon County, WI were sampled to determine richness of fish species using fyke nets, conventional seines, and boom-shocking. In addition, lake habitat features including CWH and substrate type were quantified using side-scan sonar and groundtruth methods. Fish species sampling was summarized using linear regression analysis (LOG weight vs. LOG length) and histograms by gear type (length versus frequency). Results from this study were compared to fish community and stocking information documented at the Wisconsin Department of Natural Resources (WDNR) dating back to 1938 in some lakes. Three of the eleven lakes sampled (Mission, Pike, and Wadley) were boom-shocked, and detailed habitat information was recorded to determine the relationship between habitat condition and fish species richness. Habitat variables include depth (m), vegetative percent cover (± 10%), substrate type (muck, sand, gravel, cobble, mixed, boulder), slope (%), azimuth (angle), CWH (presence/absence), and water temperature (Celsius). Boom-shock GPS locations were recorded at precise latitude and longitude coordinates using a Trimble R-6 GPS receiver and Wisconsin s Continuously Operating Reference Station (WISCORS) corrections. Historical overview Organized efforts to manage fish in the eastern Marathon County lakes dates back to the 1930s based on WDNR records (WI DNR 2012b). With the exception of Lilly Lake, historical records indicate that all of the lakes have been periodically stocked with a variety of native and non- 38
native fishes as far back as 1938. In the past, indiscriminant stocking of fish had been a common management practice of resource agencies around the country and to some extent continues today (WI DNR 2012b). The assumption was that natural reproduction was limited and any stocked fish would supplement natural populations. Studies in recent years in Wisconsin and throughout the country have shown that stocking of native species into waters where they were already present does not consistently have a measurable impact on numbers of adult fish or angler success (Jennings et al 2005, Boxrucker 1986). Popular game fish stocked that have been supported by stocking efforts include Walleye and Muskellunge (Esox masquinongy) (Hauber 1983, WI DNR 2010b). Two eastern Marathon County lakes had their native fish populations removed by whole-lake poisoning in an attempt to establish a modified fishery including Big Bass (c. 1957) and Mud (c. 1979) Lakes (WI DNR 2012b). This practice was common when fisheries management in Wisconsin was synonymous with trout management. A more holistic management philosophy is practiced today that focuses more on preserving the biological integrity of the ecosystem as a whole, particularly in systems with lower disturbance and absent of invasive flora and fauna. The loss of fish from winterkill events in the eastern Marathon County lakes has also been documented frequently (WI DNR 2012b). In most cases, winterkill events were the result of low dissolved oxygen (DO) concentrations in the water. In an effort to minimize the impact of winterkill events, dip-netting (the practice of using a hand-held net with a handle to harvest fish) was occasionally permitted and opened to the public for short periods of time to salvage desirable fish certain to meet death if not taken by fishermen (WI DNR 2012b). Dip-netting approval in eastern Marathon County by the WDNR has not been documented since 1982 and 39
is not likely to be approved in future events. Winterkill can also affect certain species more than others, noticeably leaving fish such as bullhead to dominate after a tragedy hits. Many lakes in eastern Marathon County have been managed historically for Northern Pike (Esox lucius) as they tend to tolerate low dissolved oxygen (DO) conditions better than other game fish. To increase over-winter survival rates of fish, several lakes have installed temporary or permanent aerators that continue to supply oxygen to the lake during winter months. On occasion, WDNR records (2012b) included information about the construction of boat launches and road access to a few lakes in eastern Marathon County. In the case of Mud Lake, records show that average size of desirable Yellow Perch (Perca flavescens) declined after boat and road access was constructed. This insight shows managers the pressure these fish communities battle by year-round angling demand in these small, and typically, enclosed water bodies. Our objective for this study was to determine if a relationship exists between habitat condition and fish species diversity and richness. 40
METHODS Site Description Eleven lakes in eastern Marathon County, Wisconsin were included (Figure 1). These surface waters fall into one of two distinct river basins: The Wisconsin River basin and the Wolf River Basin. These lakes are predominantly located in the eastern third of the county in the towns of Bevent, Norrie, Elderon, and/or Reid. The lakes are kettle lakes in hilly topography and sandy soils that were deposited as a result of glacial till. Surface water in this region is primarily groundwater fed with surface runoff inputs often originating nearby. The lakes include eight seepage lakes: Bass, Big Bass, Lost, Mayflower, Mission, Mud, Norrie, and Wadley Lake, two groundwater drainage lakes, Lilly and Rice Lakes, and one drainage lake, Pike Lake. Surface areas of the study lakes range from 10 to 82 hectares (Table 1) with shorelines ranging from 1.8 to 4.5 kilometers. Nine of the eleven study lakes were mapped for in-lake habitat features using a Lowrance HDS5 with side-scan sonar (LSSS). Littoral bottom habitat was sampled with LSSS to create maps of bottom substrate classes and CWH abundance (log diameter >10 centimeters and length >1.5 meters) extending from the shoreline to a range of 30.5 meters. The following substrate classes are modifications to classes found in the Wisconsin DNR critical habitat designation manual (Cunningham, 2008): muck, sand, gravel, cobble, mixed (sand/gravel, sand/gravel/cobble, gravel cobble, sand/cobble), and boulder (Table 2). Classifications were determined by interpreter ability to distinguish between different substrate types. 41
Table 8: Size (hectare) and perimeter (kilometer) of eleven study lakes in eastern Marathon County, WI. Area Perimeter Lake ha km Bass 32.3 3.1 Big Bass 72.8 3.4 Lilly 34.5 2.7 Lost 17.2 1.8 Mayflower 39.0 4.5 Mission 44.0 3.7 Mud 28.4 2.6 Norrie 40.3 2.6 Pike 83.8 4.2 Rice 10.2 2.3 Wadley 19.1 1.9 42
Figure 2: Map of study area in Eastern Marathon County, Wisconsin consisting of 11 lakes with surface area ranging from 10 to 82 ha. 43
Habitat Survey Methods Data was collected during the 2012 sampling season (May to October) using a Lowrance HDS5 with side-scan sonar. The LSSS was front-mounted to a 6-horsepower Jon boat at approximately 15 centimeters below the water surface. The LSSS left/right viewing range varied between 15.24 and 24.38 meters. Image contrast was set between 50 and 65 percent during recording. A boat speed of 8 kilometers per hour or less was maintained and data was collected perpendicular to the shoreline at a distance of approximately 15 meters where accessible. Multiple passes around each shoreline were collected to give the interpreter additional images to verify habitat features, to pick up areas missed in the first pass, or increase the extent of viewing outward to 30.5 meters. Sonar imagery was rectified using SonarTRX v10.13 (Leraand 2012) software and then spatially displayed using Esri s ArcGIS 10.0 mapping software. Habitat substrate was manually interpreted where images were obtainable and interpretable (Table 3) inside the 30.5 meter range of shoreline, and boundaries were digitized with polygon shapefiles in ArcMap 10.0 (ESRI 2012). A minimum mapping unit (mmu) of 78.5 square meters (a circle with 5 meter radius) was used to delineate substrate habitat (Kaeser and Litts, 2010). Percent coverage of the desired near-shore area among lakes ranged from 48-100 percent. Habitat Ground-truth Verification Ground truth sampling locations for each substrate type were randomly assigned using ArcMap 10.0 (ESRI 2012). All similar habitat class polygons were combined among lakes, buffered by 4 meters (2 meters on either side of each boundary) to account for GPS positional error, and 44
assigned approximately 40 random sampling points. Random points were equally distributed among different substrate classes despite large variations in percent cover between classes. Ground-truthing at the randomly assigned point locations was completed using a combination of underwater cameras, visual observations, wading, and dredging. A Trimble survey-grade GPS receiver was used to navigate to the point locations which employed an R6 GPS receiver and TSC2 personal data assistant. Instantaneous positional corrections were obtained from the Wisconsin Department of Transportation s Wisconsin Continuously Operating Reference Stations (WISCORS) which were accessed through a US-Cellular Blackberry phone modem. The WISCORS differential positioning corrected horizontal (latitude/longitude) positions to subcentimeter accuracy during navigation. At each point location, the substrate type was documented and verified against the field technician notes to ensure agreement of substrate type (>80 percent agreement estimated). Interpreted results at the sampling points were blinded from the interpreter and field technician. Ground-truth verification was completed in the same field season. Coarse woody habitat abundance was also verified with ground-truth sampling. All interpreted shorelines among the nine scanned study lakes were grouped and split into 48.8 meter sections. Each section was designated a code of 0 to 2 depending on abundance of CWH (0 = no CWH, 1 = 1-3 logs, 2 = 4+ logs). Forty sections of 0 and 2 codings were randomly selected for ground-truthing in the field. Using the Trimble survey-grade GPS receiver to navigate to start and end points in the field with WISCORS correction, waders and boats were used to locate logs within the first 30.5 meters of shoreline. Every log observed was measured for length and diameter using calipers and measuring tape. Logs that were too deep to be assessed with 45
waders were omitted from measurements. A GPS location was also marked for each log, compared with the interpreted results, and point locations added to the final maps. 46
Table 9: Substrate classifications and descriptions for Eastern Marathon County lake substrate maps. Class Acronym Description Muck M >50% particle size is <0.06mm. Sand S >50% particle size range is 0.06mm - 2mm. Gravel G >50% particle size range is 2mm - 64mm. Cobble C >50% particle size range is >64mm - 256mm. Mixed X S/G, G/C, S/C, or S/G/C combinations. Boulder B >50% particle size range is >256mm. Table 10: Total near-shore habitat out to 30.5m available for side-scan sonar scanning in each lake and total interpreted near-shore area, substrate points, and CWH plots that were attained in 2012 field season. Lake Total near-shore Habitat Area Total interpreted Near-shore Area Percent Coverage Substrate Sampling CWH Sampling Plots ha ha % Points Wadley 5.6 5.6 100.0 3 16 Bass 8.9 8.4 94.9 21 5 Mud 7.1 6.6 92.6 18 6 Big Bass 9.8 8.1 82.5 20 8 Rice 6.7 5.4 80.2 3 7 Mission 10.4 6.7 64.3 9 1 Pike 12.6 6.8 54.0 121 4 Norrie 7.3 3.6 49.5 48 6 Mayflower 13.3 6.4 47.9 121 15 Total 81.9 57.7 * 364 68 47
Boom-shocking and habitat information of shocking locations was recorded in Mission, Pike, and Wadley Lakes. Selected locations were surveyed for fish abundance and diversity using electroshocking around the perimeter of each lake (Table 1). Lakes were chosen for accessibility with a shocking boat, differences in shoreline development, and variation in bottom type. Shock locations varied in vegetative cover (%), CWH presence/absence, depth, azimuth, and distance from shore. Expected fish species were defined based on preferences detailed in Fishes of Wisconsin (Table 4) (Becker, 1983). Habitat variables were also recorded at each of the shocking locations. Depth, azimuth, and slope were determined using bathymetry mapping results from surveys conducted in 2012 (Koeller Chapter 3, 2014). Vegetative percent cover was visually estimated to plus or minus 10 percent. Substrate type was determined from hand-texturing grab or Ekman dredge samples when possible. Alternatively, substrate type was estimated with the use of underwater cameras when dredging and grab sampling was not possible. The presence or absence of CWH was also visually documented. Fish Survey Methods All eleven lakes were surveyed for fish in 2011 or 2012 using a variety of gear. Lakes were set with fyke nets for a three-night period. Eight of the eleven lakes were traditionally seined (Bass, Rice, Mud lakes excluded). Three lakes were surveyed with boom-shocking (Pike, Mission, Wadley). Survey periods are listed in Table 3. 48
Family Common Species Preferred Substrates Vegetation Woody Habitat Depth (m) Water Clarity Aphredoderidae Pirate Perch Family endangered Pirate Perch Aphredoderus sayanus sand/soft muck/organic Dense macrophytes brush piles <1 NA Centrarchidae Sunfish Family Black Crappie Pomoxis nigromaculatus Sand/Mud/Gravel abundant NA varies clear Bluegill Lepomis macrochirus Sand/Gravel/Mud moderate, rooted NA varies clear Largemouth Bass Micropterus salmoides Sand/Gravel/Mud sparse to dense NA <1.5 clear-slightly turbid Pumpkinseed Lepomis gibbosus Sand/Gravel/Mud dense NA <1.5 clear-slightly turbid Rock Bass Ambloplites rupestris Sand/Gravel some NA NA clear Smallmouth Bass Micropterus dolomieui Sand/gravel Rooted vegetation rare-uncommon Warmouth Lepomis gulosus Soft mud/sand/gravel Dense May nest near stumps/roots White Crappie Pomoxis annularis Sand/mud/gravel/rubble/clay /silt NA lakes with depths >6 clear NA muddy-turbid sparse NA varies slightly turbidturbid Cottidae Sculpin Family Mottled Sculpin Cottus bairdi Sand/Gravel any cover NA 0.1-0.5 clear-slightly turbid Cyprinidae Minnow and Carp Family Blacknose Shiner Notropis heterolepis NA sparse-dense NA 0.1-1.5 clear Bluntnose Minnow Pimephales notatus Gravel/sand submerged NA varies clear-slightly turbid Brassy Minnow Hybognathus hankinsoni Sand/gravel some NA 0.1-1.5 clear-slightly turbid Common Carp Cyprinus carpio Sand/gravel Abundant NA 0.6-1.5 turbid common-abundant Fathead Minnow Pimephales promelas Sand/rubble/gravel floating/submerg ed algae NA up to 1.5 varying Golden Shiner Notemigonus crysoleucas Sand/mud/gravel moderate-dense NA NA clear invasive Goldfish Carassius auratus NA dense NA shallow NA common/tolerant Spotfin Shiner Notropis spilopterus Sand/mud/gravel with/without NA 0.1-0.5 clear-turbid Esocidae Pike Family Muskellunge Esox masquinongy spawn over muck/detritus numerous submerged beds NA <5 NA common Northern Pike Esox lucius Sand/mud light-dense NA NA clear-slightly turbid Table 11: Fish species preferences of substrate, vegetation, woody habitat, depth (m), and water clarity of all expected fish species in the study area (Becker, 1983). 49
Family Common Species Preferred Substrates Vegetation Woody Habitat Depth (m) Water Clarity Gadidae Cod Family Burbot Lota lota Mud/sand/rubble/boulders/sil t/gravel patches (juvenile only) NA >1.5 NA Gasterosteidae Stickleback Family Brook Stickleback Culaea inconstans Sand/gravel moderate to dense NA up to 1.5 clear-slightly turbid Ictaluridae Bullhead Catfish Family most common bh Black Bullhead Ictalurus melas Sand/gravel nest in matted vegetation nest under woody debris <1.5 varies Brown Bullhead Ictalurus nebulosus sand/rock/mud/silt vegetated Nest shelter logs shallow tolerant of high turbidity Channel Catfish Ictalurus punctatus Mud/sand/clay NA NA NA turbid Tadpole Madtom Noturus gyrinus Sand/Gravel/mud thick, submergent dense branches 0.1-1.5 clear-slightly turbid Yellow Bullhead Ictalurus natalis mud/gravel weedy spawn near/under logs 0.6-1.5 clear Percidae Perch Family Yellow Perch Perca flavescens Sand/gravel modest NA shallow clear-slightly turbid Johnny Darter Etheostoma nigrum Sand/Gravel NA NA <0.5 clear-turbid Logperch Percina caprodes Sand/gravel NA NA 0.6-1.5 clear Walleye Stizostedion vitreum sand/gravel/mud NA NA NA clear Iowa Darter Etheostoma exile sand/gravel submergent/filam NA <1.5 clear-slightly turbid. Algae Umbridae Mudminnow Family Central Mudminnow Umbra limi Gravel/sand moderate-dense NA up to 0.5 clear Table 11 (cont): Fish species preferences of substrate, vegetation, woody habitat, depth (m), and water clarity of all expected fish species in the study area (Becker, 1983). 50
Table 12: Survey periods by gear type (fyke net, seine, boom shock) of eastern Marathon County lakes in 2011 to 2012. Lake Fyke Net Survey Period Seining Survey Period Boom Shock Survey Period Bass 10-12 July 2012 NA NA Big Bass 19-22 April 2012 16 August 2012 NA Lilly 6-9 June 2012 28 August 2012 NA Lost 21-24 October 2011 21 October 2011 NA Mayflower 12-15 June 2012 11 September 2012 NA Mission 5-8 April 2012 21 August 2012 1 June 2012 Mud 19-22 June 2012 NA NA Norrie 27-30 June 2012 24 July 2012 NA Pike 29 March - 1 April 2012 18 August 2012 7 June 2012 Rice 1-4 October 2011 NA NA Wadley 12-15 April 2012 1 August 2012 14 June 2012 51
A total of seven nets in three different sizes were soaked for three nights in each lake. Net sizes and quantities included: o o o Two: mesh, ½ ; trap end, 4 x 6 ; pot, 4 dia. hoops; leads, ½ to 1 mesh, 40 to 70 long Two: mesh, ½ ; trap end, 4 x 4 ; pot, 4 dia. hoops; leads, ½ mesh, 40 to 70 long Three: mesh, ¼, trap end, 2 x 3 ; pot, 2 dia. hoops; leads, ¼ mesh, 40 to 70 long Nets were checked daily and taken down on day 4. All fish captured were measured (mm) and tallied. Up to 100 fish per species were weighed (g); weights came from 100 fish spanning the entire length interval observed. Fish age was not estimated. Up to five fish from each species in each lake were kept for later potential DNA analysis. Other individual fish from each species were kept and deposited in the Becker Memorial Ichthyological Collection of the UWSP College of Letters and Science for permanent record. Generally, these were the individuals that died as a result of stress or injury incurred as a result of the netting process. As such the number of individuals kept per species per lake varies. Exhaustive seining was done on one day for each lake (Lost, Mud, and Bass were excluded due to vegetation, growth, soft substrates and/or sloping bathymetry close to shore). These methods consisted of a crew of two to four persons seining intensely for 6 to 8 hours (10 hour day including driving to and from lake). Crew members would stop periodically to assess catch (ID, weigh, measure etc). Seines had a mesh size of 3.2 to 1.6 millimeters, ranged in length from 1.8 to 12.2 meters, and ranged in depth from 0.9 to 2.1 meters. All fish caught were measured (mm), and tallied. Fish species that were weighed, kept for DNA analysis, and/or deposited in the ichthyological collection had taken into account the individual fish species already sampled and kept during the fyke net survey. Up to 100 fish per species were weighed (g); weights came from 100 fish spanning the entire length interval observed. Up to five fish 52
from each species per lake were kept for later potential DNA analysis. Individual fish from each species were kept and deposited in the ichthyological collection as a permanent record of their occurrence. The number of individuals kept per species per lake varied. No age estimation was undertaken. Three of the study lakes (Mission, Pike, Wadley) were shocked with a boom shocker June (one day per lake) for approximately five hours. Shocking efforts targeted the near-shore, island, and/or sandbar areas were focused on during each survey. Shock locations in 2012 can be seen in Figure 1. Each shocking location was marked with a Trimble GPS receiver obtaining WISCORS corrections real-time. The number of shock locations varied per lake (Mission = 73, Pike = 93, Wadley = 85). Each fish dipped during shocking was measured and identified to species, then released. These data were used in multiple regression models to determine if relationships existed between habitat characteristics and fish species richness and diversity. Fish netting, seining, and shocking data were used to create length (centimeters) versus frequency (n) histograms of common species by lake (Appendix A). Data were also used for length (LOG length) versus weight (LOG Weight) simple linear regression models for each species by lake (Appendix B). No among lake comparisons were conducted due to variation in the sampling season and lack of multiple-year data, varying lake morphological characteristics, and historic anthropogenic pressures such as whole-lake poison applications (Mud and Bass lakes), inconsistent chemical control of aquatic nuisance plants, and inconsistent presence of invasive flora and fauna (not explored in this study). Forward stepwise logistic regression models were used to determine if relationships existed between measured habitat variables and fish diversity and measured habitat variables and 53
species richness in each lake (Shannon-Weiner diversity) (p-value for entry = 0.25). Data observations varied between lakes (Mission N = 76, Pike N = 89, Wadley N = 86). Analyses were conducted in the Minitab 15 statistical software package (Minitab 15, Minitab Inc, State College, PA). Length and weight for sport fish were also summarized for the 2011 to 2012 sampling period for all eastern Marathon County Lakes (Table 8). We were unable to determine if length/weight relationships differed among lakes with the limited amount of sampling data collected. Length versus weight regressions per species per lake for sport fish are listed in Appendix B (fyke net and seining observations combined). 54
Figure 3: Boom-shock spatial locations for Pike, Mission, and Wadley lakes with total number of shock observations. 55
RESULTS Thirty-seven species of fish have been reported by the WDNR historically from the eleven Marathon County lakes sampled in this study (Table 4) which included survey and stocking data. Twenty-seven of these species were found in the current study. Among all eleven lakes, Warmouth (Lepomis gulosus) was a new occurrence that had never before been documented. Almost every individual lake reported several never-before documented fish species. The species previously reported during fish surveys or stocking efforts that were not found in the present study include Blacknose Dace (Rhinichthys atratulus), Brown Trout, Brook Trout (Salvenis fontinalis), Channel Catfish (Ictalurus punctatus), Emerald Shiner (Notropis atherinoides), Fathead Minnow (Pimephales promelas), Pirate Perch (Aphredoderus sayanus), Rainbow Darter (Etheostoma caeruleum), Rock Bass (Ambloplites rupestris), Slender Madtom (Noturus exilis), and Spotfin Shiner (Cyprinella spiloptera). The absence of these species from the current study may be due to a variety of factors including limitations of sampling gear, environmental disturbance, misidentification of previous reports, and/or simply natural changes that occur in fish communities over long periods of time. The overall sampling effort in any given lake in this study was generally less than the collective effort applied in all previous surveys dating back over 70 years for some lakes, yet new occurrences were documented in 2011 and 2012 (Table 15, Table 21). Although different sampling techniques including seining, fyke netting and electrofishing were used on each lake (with exclusions stated in the introduction) in the present study, none of these techniques effectively sample deep water habitats where species such as Walleye and salmonids are likely to be found in late summer and early fall when some of the sampling took place. In addition, 56
the salmonids and Walleye would not be expected to be present in any lake unless they were recently stocked because they would not normally reproduce or live year round in these limited lake environments. The absence of Brown and Brook Trout is expected since these fish were introduced through stocking efforts but have not been supported with stocking recently. Brown Trout were last stocked in Bass Lake in 1992. Brook Trout have not been stocked in Pike Lake since 1976. The loss of some species such as the Blacknose Dace, Rainbow Darter, and Emerald Shiner could not be verified without voucher specimens; these fish are uncommon in lentic systems and it is possible they were identified incorrectly in the historic surveys. The Pirate Perch, a rare species in Wisconsin, was reported and verified once among lakes (Norrie Lake, 2003). Length and frequency for sport fish in each lake were summarized by sampling gear over the 2011 to 2012 sampling period (Tables 16-18). Length versus frequency regressions per species per lake for seine and fyke net gears were summarized (Table 19). We were unable to determine if length/frequency relationships differed among lakes with the limited amount of sampling data collected. Length versus frequency histograms per species per lake for all sport fish with more than 10 observations are listed in Appendix A (multiple sampling gears are listed on histograms). Forward stepwise regression analyses were inconclusive. Relationships did not exist between measured habitat variables and fish diversity at shock locations; resultant models were not significant in predicting fish diversity (Table 13). Relationships did not exist between measured habitat variables and fish diversity at shocking locations; resultant models were not significant in predicting species richness (Table 14). 57
Table 13: Forward stepwise regression results for measured habitat variables (explanatory variables) versus fish diversity (response variable) (p-value for entry = 0.25) relationship strength (r 2 ) and significance (P). Table 14: Forward stepwise regression results for measured habitat variables (explanatory variables) versus fish species richness (response variable) (p-value for entry = 0.25) relationship strength (r 2 ) and significance (P). 58
Table 15: Fish species occurrence in the 2011 and 2012 study area, historical record, and WDNR stocking records where P = present in the current study, H = present historically, and S = stocked previously. Bass Lake Big Bass Lake Lilly Lake Lost Lake Mayflower Lake Mission Lake Mud Lake Norrie Lake Pike Lake Rice Lake Wadley Lake Summary Black Bullhead PH PH H PH H PH PH H PH Black Crappie P PHS P PHS PH PH PH PHS PHS PH PH PHS Blackchin Shiner P P Blacknose dace H H Blacknose Shiner P P Bluegill PH PH P PHS PH PH PHS PH PHS PH PH PHS Bluegill x Pumpkinseed hybrid P H P P P PH Bluntnose Minnow P P P Brown Bullhead H P P H PH Brook Trout S S Brown Trout S S Central Mudminnow P H PH H P H PH Channel Catfish H H Common Shiner H H PH H PH PH Emerald Shiner H H H Fathead Minnow S S Golden Shiner PH PH H PH H PH P PH Green Sunfish H H P PH PH Iowa Darter P P P PH P P PH Johnny Darter PH PH H PH Largemouth bass HS PHS P PHS PHS PHS PHS PHS PHS PH PHS PHS Least Darter P P Muskellunge S PHS S S PHS Northern pike S HS P PHS PH PHS HS HS PHS PHS PHS PHS Pirate Perch H H Pumpkinseed PH H P PH PH PH H PH PH PH H PH Rainbow Darter H H Rock Bass S H S HS Slender Madtom H S Smallmouth Bass S S PS S PS Spotfin Shiner H H Spottail shiner PH PH Walleye PHS PHS PHS H PHS PHS PHS PHS Warmouth P P White sucker H H H PH PH H PH PH Yellow bullhead P PH PH PH P P PH Yellow Perch H PH P PHS PHS PH H PHS PHS PH PHS PHS 59
Table 16: Average length per species per lake in fyke nets during the survey period. Note survey periods varied among lakes, refer to Table 3 for survey dates. Lake Species Mean Length (cm) Bass Black Bullhead 10.4 Big Bass Black Crappie 14.9 Lilly Black Crappie 18.0 Mission Black Crappie 21.1 Mud Black Crappie 21.1 Norrie Black Crappie 14.7 Pike Black Crappie 24.4 Rice Black Crappie 22.7 Wadley Black Crappie 25.9 Bass Bluegill 12.4 Lilly Bluegill 12.7 Lost Bluegill 17.3 Mayflower Bluegill 8.1 Mission Bluegill 12.4 Mud Bluegill 6.6 Norrie Bluegill 13.5 Pike Bluegill 12.4 Rice Bluegill 14.2 Wadley Bluegill 7.9 Mayflower BluegillxPumpkinseed Hybrid 7.6 Mission Green Sunfish 12.7 Big Bass Largemouth Bass 11.7 Lilly Largemouth Bass 2.8 Mayflower Largemouth Bass 3.0 Mission Largemouth Bass 40.1 Mud Largemouth Bass 3.0 Norrie Largemouth Bass 3.8 Pike Largemouth Bass 35.1 Wadley Largemouth Bass 8.6 Mission Northern Pike 47.5 Pike Northern Pike 56.9 Lilly Pumpkinseed 10.4 Mayflower Pumpkinseed 9.1 Mission Pumpkinseed 10.4 Pike Pumpkinseed 15.2 Norrie Smallmouth Bass 4.1 Norrie Walleye 37.8 Pike Walleye 55.1 Mission Warmouth 16.3 Big Bass Yellow Bullhead 30.0 Lost Yellow Bullhead 34.8 Pike Yellow Bullhead 24.6 Rice Yellow Bullhead 26.9 Wadley Yellow Bullhead 27.4 Big Bass Yellow Perch 24.0 Mission Yellow Perch 13.0 Norrie Yellow Perch 4.3 Pike Yellow Perch 12.2 60
Table 17: Average length per species per lake in seines during the survey period. Note survey periods varied among lakes, refer to Table 3 for survey dates. Lake Species Mean Length (cm) Lilly Bluegill 5.7 Lost Bluegill 15.8 Mayflower Bluegill 9.1 Mission Bluegill 3.2 Norrie Bluegill 8.2 Pike Bluegill 3.9 Big Bass Largemouth Bass 6.6 Lilly Largemouth Bass 8.0 Mayflower Largemouth Bass 5.8 Norrie Largemouth Bass 5.2 Wadley Largemouth Bass 5.6 Mission Pumpkinseed 16.8 Pike Pumpkinseed 14.2 Norrie Smallmouth Bass 5.7 Norrie Yellow Perch 8.6 Table 18: Average length per species per lake during the boom shock survey period. Note survey periods varied among lakes, refer to Table 3 for survey dates. Lake Species Mean Length (cm) Mission Black Crappie 19.8 Pike Black Crappie 15.5 Wadley Black Crappie 18.5 Mission Bluegill 12.2 Pike Bluegill 10.7 Wadley Bluegill 7.4 Mission Largemouth Bass 24.6 Pike Largemouth Bass 31.0 Wadley Largemouth Bass 16.5 Mission Northern Pike 53.1 Pike Northern Pike 46.7 Mission Pumpkinseed 14.7 Pike Pumpkinseed 11.9 Mission Warmouth 14.5 Pike Yellow Bullhead 24.4 Mission Yellow Perch 13.2 Pike Yellow Perch 13.0 61
Lake Species Gear Length Range (cm) N (total number) Y-intercept Slope Standard Error Bass Black Bullhead Fyke 1.2-7.9 124-4.280 2.724 0.020 Mayflower Black Bullhead Fyke and Seine 10.8-14.3 2 * * * Norrie Black Bullhead Fyke and Seine 0.9-14.6 7 * * * Bass Black Crappie Fyke 8.8-12.2 4 * * * Big Bass Black Crappie Fyke and Seine 2.5-7 51-5.185 3.147 0.024 Lily Black Crappie Fyke and Seine 2.5-10.6 8 * * * Mayflower Black Crappie Fyke and Seine 6.8-9.6 4 * * * Mission Black Crappie Fyke and Seine 1.2-13.5 129-4.919 3.026 0.019 Mud Black Crappie Fyke 5.6-12.8 10 * * * Norrie Black Crappie Fyke and Seine 1.1-10.4 23 * * * Pike Black Crappie Fyke and Seine 4.9-15.3 37-4.203 2.714 0.027 Rice Black Crappie Fyke 7.2-11.4 7 * * * Wadley Black Crappie Fyke and Seine 7.7-12.5 9 * * * Lily BLGxPUS Fyke and Seine 4.8-7.8 4 * * * Mayflower BLGxPUS Fyke and Seine 1.9-7.4 38 * * * Bass Bluegill Fyke 1.4-8.3 71-3.272 2.337 0.038 Big Bass Bluegill Fyke and Seine 0.9-10 345-5.291 3.257 0.032 Lily Bluegill Fyke and Seine 1.1-8.6 199-5.148 3.186 0.009 Lost Bluegill Fyke and Seine 0.9-10.1 107-4.445 2.759 0.043 Mayflower Bluegill Fyke and Seine 0.4-7 159-4.880 3.052 0.019 Mission Bluegill Fyke and Seine 0.8-7 566-4.427 2.808 0.043 Mud Bluegill Fyke 1.7-10.1 113 * * * Norrie Bluegill Fyke and Seine 0.9-9.3 157 * * * Pike Bluegill Fyke and Seine 0.8-10.9 177-3.135 2.235 0.095 Rice Bluegill Fyke 1.5-8.5 71-4.211 2.658 0.050 Wadley Bluegill Fyke and Seine 0.9-8.3 294-5.419 3.310 0.023 Rice Brown Bullhead Fyke 10.5-11.5 3 * * * Mission Green Sunfish Fyke and Seine 1.9-7 27-4.420 2.884 0.038 Big Bass Largemouth Bass Fyke and Seine 1.2-14.8 166-4.768 2.875 0.042 Lost Largemouth Bass Fyke and Seine 3.1-17.9 6 * * * Mayflower Largemouth Bass Fyke and Seine 0.7-7.3 95 * * * Mission Largemouth Bass Fyke and Seine 14.6-18.2 7 * * * Table 19: Length range and regression summary per species per lake (n>25) with fyke net and seine gear (Y-intercept, Slope, Standard Error reported). 62
Lake Species Gear Length Range (cm) N (total number) Y-intercept Slope Standard Error Mud Largemouth Bass Fyke 0.9-1.9 26 * * * Norrie Largemouth Bass Fyke and Seine 0.9-4.1 66 * * * Pike Largemouth Bass Fyke and Seine 11-15.6 5 * * * Wadley Largemouth Bass Fyke and Seine 1.4-13.3 37 * * * Mission Muskellunge Fyke and Seine 11.7-33.3 2 * * * Mayflower Northern Pike Fyke and Seine 24.2-28.4 2 * * * Mission Northern Pike Fyke and Seine 13.1-26.4 3 * * * Pike Northern Pike Fyke and Seine 17.3-28.7 29-5.094 2.938 0.013 Rice Northern Pike Fyke 21.9-30.8 3 * * * Wadley Northern Pike Fyke and Seine 12.8-20.9 2 * * Lily Pumpkinseed Fyke and Seine 2.5-6.9 8 * * * Mayflower Pumpkinseed Fyke and Seine 2.4-7.2 55 * * * Mission Pumpkinseed Fyke and Seine 4.1-7.9 44-4.438 2.893 0.018 Norrie Pumpkinseed Fyke and Seine 4.6-7.6 4 * * * Pike Pumpkinseed Fyke and Seine 4.5-7.2 17 * * * Rice Pumpkinseed Fyke 3.9-8.5 6 * * * Norrie Smallmouth Bass Fyke and Seine 1.4-2.8 24 * * * Big Bass Walleye Fyke and Seine 20.6-25.3 3 * * * Lost Walleye Fyke and Seine 20.9-22.8 2 * * * Mayflower Walleye Fyke and Seine 16.3-25.5 3 * * * Norrie Walleye Fyke and Seine 5.8-22.4 16 * * * Pike Walleye Fyke and Seine 14.8-24.6 10 * * * Wadley Walleye Fyke and Seine 22-22.3 2 * * * Mission Warmouth Fyke and Seine 4.5-7.4 11 * * * Big Bass Yellow Bullhead Fyke and Seine 5.1-15.4 21 * * * Lost Yellow Bullhead Fyke and Seine 5-16.5 54-4.203 2.634 0.028 Mission Yellow Bullhead Fyke and Seine 6.7-10.2 7 * * * Pike Yellow Bullhead Fyke and Seine 7.8-13.2 136-4.251 2.744 0.024 Rice Yellow Bullhead Fyke 8.5-15 29 * * * Wadley Yellow Bullhead Fyke and Seine 1.9-14 24 * * * Big Bass Yellow Perch Fyke and Seine 3.9-12.4 17 * * * Mayflower Yellow Perch Fyke and Seine 1.3-6.3 6 * * * Mission Yellow Perch Fyke and Seine 3-8.1 184-5.074 3.039 0.028 Norrie Yellow Perch Fyke and Seine 0.8-4.8 127 * * * Pike Yellow Perch Fyke and Seine 3.2-8.1 107-5.486 3.236 0.039 * = Sample size <25 Table 19 (cont): Length range and regression summary per species per lake (n>25) with fyke net and seine gear (Y-intercept, Slope, Standard Error reported). 63
The number of fish species found in each lake in this study appeared to be correlated with physical characteristics of the lake. The number of species in the eleven lakes was correlated with depth (0.50 correlation coefficient on a scale of 0 to +/- 1) and total volume (0.54 correlation coefficient on a scale of 0 to +/- 1) (Figure 4). Deeper and larger lakes in this study support more species. The eleven lakes sampled varied in size (10 to 84 hectares) and maximum depth (1.5 to 9.5 meters). There was no relationship between species richness and shoreline disturbance (docks per mile) in the Marathon County Lakes (Figure 5). This may be explained by the variation in depth among lakes where shallower lakes tend to support less diversity. Lakes with a permanent inlet or outlet had 56% more species than those without these natural corridors for fish migration. The greatest number of fish species (20) found in the 2011 to 2012 study was from Mission Lake (Table 21). Pike Lake, a drainage lake that receives water from Rice Lake Creek and that is also connected to the Plover River from Pike Lake Creek, had a fish species richness of eighteen. The least number of species in the 2011 to 2012 study were found in Mud Lake (3) and Bass Lake (5). Lakes with the greatest percentage of species lost or not recovered when comparing 2011 to 2012 data to historical records were Mud (70%) and Big Bass (44%). The lakes with the lowest percentage of species lost or not recovered were Mission (9%), Norrie (10%), Rice (17%) and Pike (18%). 64
Figure 4: Relationships between species richness (Y) and lake physical characteristics of maximum depth and volume. Reported values are coefficient of determination (R 2 ). Figure 5: Relationships between species richness (Y) and shoreline development (docks per mile). Reported values are coefficient of determination (R 2 ). 65
Table 20: Species richness in the 2011 and 2012 study, species previously documented in WDNR records, newly documented species in 2011 and 2012, lost or undocumented species, total species richness (past/present), and percentage of species lost or undocumented. 66
The substrate type of the near-shore habitat (within 30.5 meters from shoreline) was assessed in each lake. The eleven lakes varied in near-shore substrate coverage, varying from soft muck (or marl) to a boulder bottom, but all lakes were dominated by softer substrates (Table 10). User interpretation accuracy of LSSS substrate habitat classification was determined using an error matrix (Table 22). A total of 364 points were assessed across the six substrate categories. Interpreted accuracy was best with boulder (77%) and worst with gravel (0%) classification. Overall accuracy was 53 percent. User interpretation accuracy of LSSS coarse woody habitat classification was also determined with an error matrix (Table 23). Sites showing no CWH present (code 0) were interpreted with 94% accuracy; whereas sites with abundant CWH (code 2) were interpreted with 78% interpreted accuracy. Overall accuracy was 74 percent. Coarse woody habitat abundance was calculated as logs per kilometer of shoreline mile for participating study lakes. Log abundance ranged from 0 to 98 logs per kilometer. 67
Table 21: Bottom substrate percent cover by lake in 2012 surveys of near-shore, in-lake habitat out to 30.5 meters of shoreline. Lake Muck Sand Sand/Gravel Gravel/Cobble Sand/Gravel/Cobble Cobble Boulder Sand/Cobble Gravel Scattered Wood (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) Bass 96.4 1.7 * 0.9 1.0 * * * * * Big Bass 77.8 22.2 * * * * * * * * Mayflower 82.4 6.7 1.1 3.7 5.6 * 0.5 * * * Mission 89.3 10.5 * 0.2 * * * * * * Mud 95.0 3.4 * * 0.4 * * 1.2 * * Norrie 80.3 5.2 * * 0.9 * 0.3 5.9 * 7.4 Pike 55.6 26.4 1.6 0.2 15.0 0.6 0.6 * * * Rice 100.0 * * * * * * * * * Wadley 87.1 10.5 * * 0.1 * * 2.2 0.1 * 68
Point Sampling data (field sampling) Row Column Total 45 44 36 162 39 38 364 Table 22: Error matrix for substrate map classification values in nine Marathon County study lakes (2012). Interpreted accuracy is total verified as true during ground-truth divided by the total random sample locations per category. User s accuracy is determined from total true per category divided by total found in category among all random sites. Overall accuracy equals total number correct divided by the total number of random sites. Interpreted data M S G MIX CO BO Muck (M) 31 3 3 12 0 0 49 63% Sand (S) 10 29 21 34 7 15 116 25% Gravel (G) 0 0 0 5 0 0 5 0% Mixed (MIX) 4 12 10 108 23 1 158 68% Cobble (CO) 0 0 2 2 4 1 9 44% Boulder (BO) 0 0 0 1 5 21 27 78% Interpreted Accuracy 69% 66% 0% 67% 10% 55% Total User's Accuracy Overall Accuracy 53% 69
Point Sampling data (field sampling) Row Table 23: Error matrix for coarse woody habitat map classification values in nine Marathon County study lakes. Interpreted accuracy is total verified as true during ground-truth divided by the total random sample locations per category. User s accuracy is determined from total true per category divided by total found in category among all random sites. Overall accuracy equals total number correct divided by the total number of random sites. 0 (no CWH) 1 (1-3 logs) 2 (4+ logs) 0 (no CWH) 29 6 6 41 71% 2 (4+ logs) 2 4 21 27 78% Column Total 31 10 27 68 Interpreted Accuracy 94% 78% Overall Accuracy 74% Interpreted Data Total User's Accuracy 70
Table 24: Interpreted Logs per kilometer values in nine Marathon County study lakes. CWH Lake Logs/km Bass 1.0 Big Bass 8.6 Lilly 10.2 Lost * Mayflower 34.4 Mission 3.8 Mud 34.6 Norrie 53.6 Pike 1.7 Rice 0.0 Wadley 97.9 71
DISCUSSION The fish communities in several of the eastern Marathon County lakes have undergone substantial change. Due to anthropogenic pressures, such as whole-lake poisonings, and morphological differences, such as lake type, relationships between habitat characteristics and fish species richness and diversity were unable to be determined. Species richness was lowest in Mud and Bass Lakes which were both poisoned and reset (restocked) according to historic WDNR management records in an effort to local desires for sport fishering. Reintroduction of extirpated species could increase diversity in these systems and potentially improve ecosystem health. Pike Lake, the only drainage lake, had one of the highest species richness values which may be driven by migration of fish through tributaries from other systems. Mission Lake reported the highest species diversity. Mission Lake shows signs of intermittent connection to surrounding water bodies including the Plover River, which may contribute to higher diversity than fully isolated systems; however Mission Lake is also uniquely undeveloped when compared to other Marathon County Lakes similar in size. Most lakes in the study are small, isolated, easily over-fished, and may never support a selfsustaining high-intensity hook and line consumptive fishery for large predatory fish. The greatest fishery value of some lakes may be the genetic diversity preserved in species of fish. Mission Lake has an interesting fish community. Populations of Warmouths are rare in glacial lakes this far north. Warmouths are being reported in this document for the first time; however, the origin of the Warmouth into Mission Lake is unclear and more investigation is needed to determine origin. Blacknose Shiners (Notropis heterolepis), and Blackchin Shiners (Notropis heterodon) also co-occur in Mission, a rare combination. These two species are 72
mainly found in lakes as opposed to other shiners that prefer flowing waters. The Least Darter (Etheostoma microperca) is a species of special concern in Wisconsin and was also reported from Mission Lake for the first time; it has been reported in the nearby Plover River. The combination of Iowa Darters (Etheostoma exile), Least Darters, Blacknose Shiners and Blackchin Shiners are an indicator of relatively intact and undisturbed near shore habitat (Wang et al 2003). Mission Lake species richness is significantly higher than other seepage lakes reviewed in this study, and maintaining the biological diversity is beneficial to the fish community and ecosystem. Norrie Lake also had a couple noteworthy fish. Pirate Perch, a threatened species in Wisconsin, has been previously reported in Norrie Lake; however, no Pirate Perch were documented in this study. Norrie Lake s CWH abundance and tanic waters support Pirate Perch habitat. Pirate Perch are most commonly associated with woody backwaters of slower moving larger rivers in the south and the west of the state. Juvenile Smallmouth Bass (Micropterus dolomieu) were also observed in Norrie which are rarely found in smaller glacial lakes in this part of the state. Most of the lakes have extensive beds of submersed aquatic vegetation. Submerged vegetation provides essential cover for young fish, but excessive vegetation along with too few large predators may contribute to overpopulation and stunting of fish size. Small Bluegill dominated the samples of fish populations in these lakes and large predators were less commonly documented. The most common predator observed in Marathon County lakes is the Largemouth Bass (Micropterus salmoides). Although abundant in some lakes, few individuals greater than 13 inches were sampled; although shocking is generally required for population sampling since Largemouth Bass can evade traditional fyke nets and seines. Another large 73
predator that could help reduce sunfish populations, the Northern Pike, was uncommonly documented as well which may be a reflection of sampling seasonal and gear biases. Although the sampling intensity was insufficient to make specific sport fish management recommendations on individual lakes, management strategies to increase the number and size of large game fish could enhance angler experience (Carpenter et al 1994; Wilde 1997). Strategies to increase fish size could include some form of catch restrictions or size limits and their potential use should be explored further on a lake by lake basis. Lake-bottom hardness can be an important component to a lakes fishery. Some fish (game and non-game species) prefer soft substrates for spawning/habitat including bullhead, catfish, Muskellunge, and Warmouth (Becker 1983). Other fish species prefer harder substrates (such as sand, gravel, or cobble) for spawning and habitat but have also been observed spawning over soft muck bottoms. These species include non-game fish (Tadpole Madtom, Golden Shiner (Notemigonus crysoleucas), Spotfin Shiner, Fathead Minnow) and game-fish (Black Crappie (Pomoxis nigromaculatus), Bluegill, Largemouth Bass, Northern Pike, Pumpkinseed, and Walleye) (Becker 1983). Other fish species prefer only hard bottoms for spawning and habitat (such as sand or gravel) including Bluntnose Minnow (Pimephales notatus), Central Mudminnow (Umbra limi), Rock Bass, Smallmouth Bass, Mottled Sculpin, Yellow Perch, and Darter species (Becker 1983). Although game-fish are more desirable for angling, successful reproduction of non-game species is equally important in a lake ecosystem. To anglers, non-game fish are an essential food source to large, desirable sport fish. Most of the eastern Marathon County lakes are isolated water bodies, so species diversity in these systems would only improve if stocking of 74
fish was attempted. However, the continued successful reproduction of present species is needed to maintain fish diversity and ecosystem health. Introduction of harder substrates such as gravel and/or cobble may improve natural reproduction in certain systems. There was little CWH observed in most of the lakes. Coarse woody habitat provides forage areas for predators (Newbrey et al. 2005), increases insect diversity (Schmude et al. 1998), alters fish diet (Sass et al. 2006a), and reduces predation risks for prey fish (Sass et al. 2006b). Since the CWH input rate into aquatic ecosystems is a slow process (Guyette and Cole 1999), abrupt removal can negatively alter habitat for long-term periods. Tree-lined shores in drainage lakes, which tend to maintain more constant water levels, should be protected and trees that fall into the water should be left in place to provide cover. Some lakes had more natural shorelines and associated cover than others. Mission, Mud, Lilly, Lost, Bass, Rice, and portions of Mayflower Lake had more minimally developed shorelines than most other lakes and the shoreline areas of these lakes could serve as models of a more natural lake setting that would benefit the aquatic ecosystem as a whole. Many Wisconsin lakes continue to face developmental pressures. Pike and Big Bass Lakes have more developed shorelines and less in-lake CWH than other lakes with less development and more structure. Humans remove trees both on the landscape and in near-shore aquatic habitats, reducing CWH inputs (Christensen et al. 1996) and significantly lowering sequestration of carbon by CWH (Guyette et al. 2002). Downed trees in littoral areas represent the most permanent and often only year-round cover for fish. Fish populations in most Eastern Marathon County lakes could benefit from the addition of woody cover below the lowest reported water levels where it would remain continuously submerged. 75
There were many reports of winterkill in the eastern Marathon County lakes over the last 60+ years. Winterkill events are typically a function of low dissolved oxygen (DO) concentrations. Low DO conditions occur more frequently in shallow lakes or lakes with high organic matter that consumes oxygen during decomposition. Lakes with below average species richness (10 or less) and frequently reported winterkill events may best be managed as Put-and-Take fisheries with supporting enhancement and restoration techniques (Eades et al 2008). Although no relationships could be identified between measured habitat variables and fish diversity and species richness, previous studies have shown an important link between bottom type, CWH, and species dependency (Becker 1983, Bozek et al. 2002, Johnson 1961, Sass et al. 2006b). 76
CHAPTER 3: A COMPARISON OF HISTORICAL CHANGES IN LAKE MORPHOLOGY OF SIX INLAND WISCONSIN LAKES ABSTRACT The morphology of lake-bottom changes over time were assessed in six central Wisconsin Lakes. Original bathymetric surveys were conducted with ice-grid surveys in cut holes through ice at intersections of perpendicular transects that were equally spaced over a lake surface. Contours were hand-drawn based on survey depths. The second generation of maps were produced with graphing sounders to traverse manually across perpendicular grid transects. Contour maps were hand-drawn after graphing sounder transects were manually stitched together. In this study, bathymetry maps were created with GPS/Sonar technology with subcentimeter accuracy for comparison with historical maps. ArcMap 10.0 (ESRI 2012) software was used to rectify and build 3D models for comparisons of lake depth and volume changes within lakes over time (H o : X 1 X 2 = 0). Reasons for changes in lake depth and volume were attributed to inaccuracy in historical survey methods, sedimentation and scouring, and rectification errors. This study points out the present need for well-defined base maps with precise (<1cm) elevations so future studies can clearly define bottom changes that may be attributed to natural and anthropogenic factors. INTRODUCTION The morphology of lake bottoms change over time. Sediments from land are picked up during rain events and transported to lakes with runoff that carry attached pollutants and nutrients (Robbins and Edgington 1975). Sediment particle sizes vary greatly, but smaller particles (i.e. 77
silt) can fill in cracks and even cover rocky areas in a lake, changing the habitat structures of the lake bottom that are used to support aquatic life (Jennings 2003). It is important to analyze lake bathymetry since lake shape can vary over time, and current depth acquisition capabilities with GPS/sonar is more detailed and accurate than methods used prior to the 1990 s (Wetzel 2001). Recent depth studies on western Lake Michigan with more accurate sonar/gps information yielded new maps showing more complex bottom structure and relief than previous maps (Waples et al. 2005). Bathymetric maps have many applications including hydrologic modelling, environmental protection, and geographic planning (Gao 2009; Hell et al. 2012). In marine systems, drop-off location and shape are critical for marine species reproduction and foraging (Heyman et al. 2007). Spawning habitat in near proximity to steep drop-offs in freshwater lakes is also preferred by species such as Smallmouth Bass (Micropterus dolomieu) (Pflug and Pauley 1984). Fish distributions are not dependent on productivity alone; lake morphology, primarily depth and surface area, is also related to community structure (Mehner et al. 2005; Olin et al. 2002). Prior bathymetric surveying methods on Wisconsin Lakes included the ice-grid method and graphing sonar method (Hartnett 2012). Ice-grid surveys were primarily conducted prior to 1960 during winter months over an ice surface. Surveyors cut holes through the ice at intersections of perpendicular transects that were equally spaced over a lake surface. A lead weight attached to a measuring string acquired depths that were entered onto an aerial map; then contours were hand-drawn. After 1960, graphing sounders were used to traverse manually across perpendicular grid transects without the aid of GPS accuracy. Contour maps were hand-drawn after graphing sounder transects were manually stitched together. These 78
prior survey methods provided a limited amount of depth information, resulting in contour maps with lower accuracy than recent surveys conducted with free-form transecting using GPS/Sonar technology. However, the historic maps still provide a picture of the lake bottom as it existed many years ago. Understanding how lake bottoms are changing can help predict future changes to a lake. In this study, we determine lake morphological change over time of six kettle lakes in Marathon County, Wisconsin (H o : X 1 - X 2 = 0). The primary object was to determine if the morphology of the lakes differed significantly between the historic surveys and surveys conducted in 2012 with better GPS/Sonar equipment. Using historic maps and precise current data acquired with survey-grade Trimble equipment (variance <1 centimeter), ArcGIS provides a platform to visually see these changes. Study Area The study area consists of six kettle lakes located in eastern Marathon County, Wisconsin (Figure 6). Five are seepage lakes (Big Bass, Lost, Mayflower, Mission, and Wadley) while Pike is a drainage lake. Lake surface areas range from 17 to 84 hectares (ha) (Table 25). Prior bathymetric maps are available through the Wisconsin DNR (WI DNR 2012a) that were surveyed in 1975 or prior (Lost - 1938, Mayflower 1967, Big Bass 1968, Pike - 1968, Mission - 1974, Wadley - 1975). METHODS Historic WI DNR maps were rectified using ESRI s ArcGIS mapping software. The bathymetry maps were first made into vector graphics using ArcMap 10.0 (ESRI 2012) by loading map 79
images as.png files, then rectifying by using the geoprocessing tools (Figure 7) to best-fit the present-day shoreline of aerial imagery (WROC 2012). Contours were then manually digitized. Finally, a triangulated irregular network (TIN) model was constructed for each lake from the digitized contour lines, and each surface was converted to a one-meter raster grid. 80
Figure 6: Six study lakes positioned in eastern Marathon County, WI ranging in size from 17 to 84 hectares. 81
Table 25: Lake area in hectares of study lakes located in Marathon County, WI. Area Lake ha Lost 17.2 Wadley 19.1 Mayflower 39.0 Mission 44.0 Big Bass 72.8 Pike 83.8 Figure 7: Rectified WI DNR 1967 bathymetry map of Mayflower Lake (Marathon County, WI) over present-day aerial imagery (NAIP, 2010). 82
A GPS/Sonar survey was completed for each study lake in 2012 using University of Wisconsin- Stevens Point GIS Center s survey-grade equipment (Ohmex Sonarmite echosounder, Trimble R6 antenna, Trimble TSC2 data collector) (Figure 8). Instantaneous positional corrections were obtained with the WISCORS system accessed through a US-Cellular Blackberry phone modem. The WISCORS differential positioning increased horizontal (latitude/longitude) and vertical (elevation above sea level on Trimble G09-WI Geoid) positions to sub-centimeter accuracy. Therefore, lake-bottom elevations can be calculated. During the survey, grid-like transects spaced no more than 60 meters apart were ran across each lake surface (Figure 9). Near-shore perimeter data also acquired. Latitude, longitude, and depth (XYZ) positions were acquired every second on the TSC2 data collector. Depths were recorded in feet for consistency with other Wisconsin Department of Natural Resources lake maps and comparison to historic models recorded with depths in feet. Lake polygons, shoreline points (depth = 0m), and XYZ data were used to construct the 2012 lake-bottom surface models. Lake shorelines were delineated from WROC 2012 imagery. Shoreline XYZ data points were added to survey data points with depths set to zero. All XYZ data were used to construct a 3D TIN in ArcMap 10.0 (ESRI 2012) (Figure 10) with constraints set to lake-polygon boundaries. TIN models were then converted to one-meter raster grid surfaces. A square 0.4 hectare grid of polygons with one random point per polygon area was created across each lake surface using the Geospatial Modelling Environment (Beyer, 2012) (Figure 11). Depth was acquired at each point from each historic one-meter grid and the corresponding 2012 one-meter grid. 83
Figure 8: Trimble survey equipment (UWSP-GIS Center) attached to the transom of a 6-horsepower Jon boat (UWSP CWSE) during Pike Lake bathymetry survey, Marathon County, WI (2012). Figure 9: Grid transects and near-shore perimeter XYZ data collected on Lost Lake, Marathon County WI (2012). 84
Figure 10: TIN surface of Wadley Lake, Marathon County, WI (2012). Figure 11: Square 0.4 hectare grid with one random point per polygon area on Big Bass Lake, Marathon County, WI (2012). 85
Simple linear regression was used to determine the relationship between lake surface area and change in total lake volume. RESULTS Random-grid surfaces were compared to see if significant differences existed between historic WI DNR lake maps and present-day lake surfaces collected with Trimble survey equipment. A paired T-test 0.05(two-tail),df>50 was conducted on paired data sets for each lake to compare historic versus present-day depths at random points (Table 27). Results show significant differences exist between the historic and present-day lake models for each water body. Spatial changes can also be seen for each lake surface using ArcMap 10.0 (ESRI 2012) mapping software. The historic one-meter raster grids were subtracted from the corresponding 2012 one-meter raster grids to show the spatial change in lakes (Figure 13). With the exception of Wadley Lake, all lakes experienced a decrease in total lake volume over time (Table 28) ranging from 9 to 15 percent. The linear relationship between lake surface area and change in total lake volume shows that lake volumes decreased more as lake size increased (R 2 = 0.62). 86
Table 26: T-test results (t0.05(two-tail),df>50) of paired data on each lake. Results show significant differences exist between historic and present-day surfaces for each lake. Paired T-Test (α = 0.05) Lake P-value t 0.05(two-tail),223 Pike <.0001 t 0.05(two-tail),53 Lost 0.0018 t 0.05(two-tail),103 Mayflower 0.0004 t 0.05(two-tail),58 Wadley <.0001 t 0.05(two-tail),180 Big Bass 0.0001 t 0.05(two-tail),121 Mission <.0001 Table 27: Historic versus present lake volume (acre-feet) as calculated with surface volume 3D-analyst tool in ArcMap 10.0 (ESRI 2012) mapping software. Lake Present Volume Historic Volume % Difference ac-ft ac-ft Pike 2005.9 2337.8-15.3% Mission 1172.9 1283.2-9.0% Lost 432 488-12.2% Wadley 456.3 395.7 14.2% Mayflower 610.4 694-12.8% Big Bass 705.1 782.2-10.4% Figure 12: Relationship between lake surface area of 2012 bathymetry surveys and change in total lake volume from historic documented surveys. 87
Figure 13: Spatial view of morphological lake-bottom between historic WI DNR maps and present-day surfaces. 88
DISCUSSION Understanding how lake bottoms are changing may help predict future morphological changes. We already know lake bottoms change over time (Wetzel 2001) and more detailed bathymetry information is being collected than was previously documented (Waples et al. 2005; Hartnett 2012). In this study, we noticed significant differences between historic and present-day lakebottom surfaces. Although the causes for these results are not clear, they may be attributed to factors such as sedimentation from runoff events, accumulation of organic materials, and/or the inability of the surveyor to properly determine bottom depth in softer sediments with the lead lines used to measure depth through the ice (pre-1960 surveys). In addition, positional accuracy during the historic surveys was visually estimated and manually measured without the use of GPS technology, and fluctuations in lake levels between the two survey periods is not well understood. Mission and Lost Lakes show depth losses concentrated in the near-shore areas where aquatic vegetation die-off may be accumulating. Pike and Big Bass Lake depth losses are concentrated in the central portions of the lake where depth is greater. Differences shown in this study assume good rectification of the historic maps and accurate depth soundings acquired with the Trimble survey equipment. Lake volume change over time was primarily negative, lake volumes decreased between the two sampling periods. The relationship between lake depth and volume change shows larger lakes will lose more total volume due to their size. A correction may need to be applied to volume estimates prior to lake management practices including aquatic plant management chemical application controls. Additional bathymetry map uses include geographical planning and environmental protection (Hell et al. 2012). 89
Accurate positional data is required to create quality bathymetric maps used by lake managers. Hydrologic modelling, environmental protection, and geographic planning (Gao 2009; Hell et al. 2012) all utilize bathymetry data. Drop-off shape and location are critical to marine and freshwater ecosystems for providing foraging and spawning success (Heyman et al. 2007; Pflug and Pauley 1984). Lake depth and size can also be dependent for fish distributions in lakes (Mehner et al. 2005; Olin et al. 2002). Chemical application rates for aquatic nuisance plant control as well as lake nutrient-load budgets will require better volume estimations going forward. There is potential to use this technology and methods presented here to monitor lake-bottom changes over time since vertical and horizontal positioning accuracy is sub-centimeter (Trimble R6 interfaced with WISCORS). Data points can be compared in the future to potentially identify where sediments settle out in lakes. Positional and depth accuracy provided by these techniques provide suitable information to interpret changes in lake morphology over time. 90
REFERENCES Anima, R., Wong, F.L., Hogg, D., and Galanis, P. 2007. Side-scan sonar imaging of the Colorado River, Grand Canyon: U.S. Geological Survey Open-File Report 2007-1087. USGS, 15 p. [http://pubs.usgs.gov/of/2007/1216/]. Ardizzone, G., A. Belluscio, and L. Maiorano. 2006. Long-term change in the structure of a Posidonia oceanica landscape and its reference for a monitoring plan. Marine Ecology 27: 299-309. Bates, C.R., and D.J. Oakley. 2004. Bathymetric sidescan investigation of sedimentary features in the Tay Estuary, Scotland. International Journal of Remote Sensing 25(22): 5089-5104. Becker, G. C. 1983. Fishes of Wisconsin. University of Wisconsin Press, Madison, Wisconsin. Beyer,.L. (2012). Geospatial Modelling Environment (Version 0.7.2.0). (software). URL: http://www.spatialecology.com/gme. Boxrucker, J. 1986. Evaluation of supplemental stocking of Largemouth Bass as a management tool in small impoundments. North American Journal of Fisheries Management 6(3): 391-396. Bozek, M.A., C.J. Edwards, M.J. Jennings, S.P. Newman. 2002. Habitat Selection of Nesting Smallmouth Bass Micropterus dolomieu in Two North Temperate Lakes. American Fisheries Society Symposium 31:135-148. Carpenter, S. R., A. Munoz-Del-Rio, S. Newman, P. W. Rasmussen, and B. M. Johnson. 1994. Interactions of anglers and Walleyes in Escanaba Lake, Wisconsin. Ecological Applications 4(4): 822-832. Cervenka, P., and C. de Moustier, 1993. Sidescan sonar image processing techniques. IEEE Journal of Oceanic Engineering 18(2): 108-122. Christensen, D.L., B.R. Herwig, D.E. Schindler, and S.R. Carpenter. 1996. Impacts of lakeshore residential development on coarse woody debris in north temperate lakes. Ecological Applications 6(4): 1143-1149. Cunningham, P. 2008. Wisconsin s Critical Habitat Designation Model. Madison, WI: WI DNR. Dombeck, M.P., B.W. Menzel, and P.N. Hinz. 1984. Muskellunge spawning habitat and reproductive success. Transactions of the American Fisheries Society 113: 205-216. 91
Eades, R. T., L. K. Richters, T. Barrow, J. Holz, P. Brakhage, and E. Traylor. 2008. Improving Nebraska Lakes via the Community Lakes Enhancement and Restoration (CLEAR) Program. American Fisheries Society, Symposium 67, Bethesda, Maryland. Everitt, R.A. and G.M. Ruiz. 1993. Coarse woody debris as a refuge from predation in aquatic communities. Oecologia 93: 475-486. Edsall, T.A., T.P. Poe, R.T. Nester, and C.L. Brown. 1989. Side-scan sonar mapping of lake trout spawning habitat in northern Lake Michigan. North American Journal of Fisheries Management 9: 269-279. ESRI 2011. ArcGIS Desktop: Release 10. Redlands, CA: Environmental Systems Research Institute. Gao, J. 2009. Bathymetric mapping by means of remote sensing: methods, accuracy and limitations. Progress in Physical Geography 33(1): 103-116. 2009. Guyette, R.P., and W.G. Cole. 1999. Age characteristics of coarse woody debris (Pinus strobus) in a lake littoral zone. Canadian Journal of Fisheries and Aquatic Sciences 56: 496-505. Guyette, R.P., and W.G. Cole, D.C. Dey, and R. Muzika. 2002. Perspectives on the age and distribution of large wood in riparian carbon pools. Canadian Journal of Fisheries and Aquatic Sciences 56: 496-505. Hartnett, S. 2012. Advances in Lake Survey Techniques. Available at http://www.uwec.edu/hartnesg/hartnett/seanweb/btsur.html (verified 14 May 2012). Hauber, A. Two methods for evaluating fingerling Walleye stocking success and natural yearclass densities in Seven Island Lake, Wisconsin, 1977-1981. North American Journal of Fisheries Management 3(2): 152-155. Havens, K.E., M.C. Harwell, M.A. Brady, B. Sharfstein, T.L. East, A.J. Rodusky, D. Anson, and R.P. Maki. 2002. The Scientific World Journal 2: 949-965. Hell, B., B. Broman, L. Jakobsson, M. Jakobsson, A. Magnusson, and P. Wiberg. 2012. A Journal of the Human Environment 41: 138-150. Heyman, W. D., J.B. Ecochard, and F.B. Biasi. 2007. Low-cost bathymetric mapping for tropical marine conservation a focus on reef fish spawning aggregation sites. Marine Geodesy 30: 37-50. Jennings, M.J., M.A. Bozek, G.R. Hatzenbeler, and E.E. Emmons. 1999. Cumulative effects of incremental shoreline habitat modification on fish assemblages in north temperate lakes. North American Journal of Fisheries Management 19: 18-27. 92
Jennings, M.J, E.E. Emmons, G.R. Hatzenbeler, C. Edwards, and M. Bozek. 2003. Is littoral habitat affected by residential development and land use in watersheds of Wisconsin lakes? Lake and Reservoir Management 19(3): 272-279. Jennings, M.J, J.M. Kampa, and G.R. Hatzenbeler. 2005. Evaluation of supplemental Walleye stocking in northern Wisconsin lakes. Journal of Fisheries Management 25: 1171-1178. Johnson, F.H. 1961. Walleye egg survival during incubation on several types of bottom in Lake Winnibigoshish, Minnesota, and connecting waters. Transactions of the American Fisheries Society 90: 312-322. Kaeser, A.J. and T.L. Litts. 2008. An assessment of deadhead logs and large woody debris using side scan sonar and field surveys in streams of southwest Georgia. Fisheries 33(12): 589-597. Kaeser, A.J. and T.L. Litts. 2010. A novel technique for mapping habitat in navigable streams using low-cost side scan sonar. Fisheries 35(4): 163-174. Kaesar, A.J., T.L. Litts., and T.W. Tracy. In press. Using low-cost side-scan sonar for benthic mapping throughout the lower Flint River, Georgia, USA. River Research and Applications. DOI10.1002/rra.2556. Leraand, T. 2012. SonarTRX: Release 10.13. Honolulu, HI: Leraand Engineering Inc. Lewin, W., N. Okun, and T. Mehner. 2004. Determinants of the distribution of juvenile fish in the littoral area of a shallow lake. Freshwater Biology 49: 410-424. Madsen, J.D. 1999. Point intercept and line intercept methods for aquatic plant management. APCRP Technical Notes Collection (TN APCRP-M1-02). US Army Engineer Research and Development Center, Vicksburg, MS. Marshall, T.R. and P.F. Lee. 1994. Mapping aquatic macrophytes through digital image analysis of aerial photographs: an assessment. Journal of Aquatic Plant Management 32:61-66. Matarrese, A., F. Mastrototaro, G. D Onghia, P. Maiorano, and A. Tursi. 2004. Mapping of the benthic communities in the Taranto Seas using side-scan sonar and an underwater video camera. Chemistry and Ecology 20(5): 377-386. McHenry, M.L., E. Shott, R.H. Conrad, and G.B. Grette. 1998. Changes in the quantity and characteristics of large woody debris in streams of the Olympic Peninsula, Washington, U.S.A. (1982-1993). Canadian Journal of Fisheries and Aquatic Sciences 55: 1395-1407. Mehner, T., M. Diekmann, U. Bramick, and R. Lemcke. 2005. Composition of fish communities in German lakes as related to lake morphology, trophic state, shore structure and humanuse intensity. Freshwater Biology 50: 70-85. 93
Newbrey, M.G., M.A. Bozek, M.J. Jennings, and J.E. Cook. 2005. Branching complexity and morphological characteristics of coarse woody structure as lacustrine fish habitat. Canadian Journal of Fisheries and Aquatic Sciences 62: 2110-2123. Nichols, S.A. 1999. Distribution and habitat descriptions of Wisconsin lake plants. Wisconsin Geological and Natural History Survey, Madison, WI. Olin, M., M. Rask, J. Ruuhijarvi, M. Kurkilahti, P. Ala-Opas, and O. Ylonen. 2002. Fish community structure in mesotrophic and eutrophic lakes of southern Finland: the relative abundances of percids and cyprinids along a trophic gradient. Journal of Fish Biology 60: 593-612. Owens, M., M.A. Learner, and P.J. Maris. 1967. Determination of the biomass of aquatic plants using an optical method. Journal of Ecology 55(3): 671-676. Pflug, D.E., and G.B. Pauley. 1984. Biology of Smallmouth Bass (Micropterus dolomieui) in Lake Sammamish, Washinton. Northwest Science 58(2): 118-130). Robbins, J. A. and D. N. Edgington. 1975. Determination of recent sedimentation rates in Lake Michigan using Pb-210 and Cs-137. Geochim. Cosmochim. Acta 39: 285-304. Sass, G.G., J.F. Kitchell, S.R. Carpenter, T.R. Hrabik, A.E. Marburg, and M.G. Turner. 2006a. Fish community and food web responses to a whole-lake removal of coarse woody habitat. Fisheries 31(7):321-330. Sass, G.G., C.M. Gille, J.T. Hinke, and J.F. Kitchell. 2006b. Whole-lake influences of littoral structural complexity and prey body morphology on fish predator-prey interactions. Ecology of Freshwater Fish 15: 301-308. Saunders, R., Bozek, M.A., Edwards, C.J., Jennings, M.J., and S.P. Newman. 2002. Habitat features affecting Smallmouth Bass Micropterus dolomieu nesting success in four northern Wisconsin lakes. American Fisheries Society Symposium 31: 123-134. Schmidt, S.M. 2010. Development of multi-dimensional littoral zone habitat fingerprints and quantifying structural habitat in lakes. M.S. Thesis. University of Wisconsin-Stevens Point, Wisconsin. 158 pp. Schmude, K.L., M.J. Jennings, K.J. Otis, and R.R. Piette. 1998. Effects of habitat complexity on macroinvertebrate colonization of artificial substrates in north temperate lakes. Journal of the North American Benthological Society 17(1): 73-80. Scribner, N. 2006. Forecasting recruitment of coarse woody structure from the riparian area to the littoral zone of a north temperate lake in Wisconsin. M.S. Thesis. University of Wisconsin-Stevens Point, Wisconsin. 147 pp. 94
Smith, C.S., and J.W. Barko. 1990. Ecology of Eurasian watermilfoil. Journal of Aquatic Plant Management 28: 55-64. Titus, J.E. 1983. Submersed macrophyte vegetation and depth distribution in Chenango Lake, New York. Bulletin of the Torrey Botanical Club 110: 176-183. Wang, L., J. Lyons, P. Rasmussen, P. Seelbach, T. Simon, M. Wiley, P. Kanehl, je. Baker, S. Niemela, and P. M. Stewart. 2003. Watershed, reach, and riparian influences on stream fish assemblages in the Northern Lakes and Forest Ecoregion, U.S.A. Canadian Journal of Fisheries and Aquatic Sciences 60: 491-505. Waples, J.T., R. Paddock, J. Janssen, D. Lovalvo, B. Schulze, J. Kaster, and J. Val Klump. 2005. High resolution bathymetry and lakebed characterization in the nearshore of western Lake Michigan. Journal of Great Lakes Research 31(1): 64-74. Wetzel, R.G. 2001. Limnology: Lake and River Ecosystems. 3rd Ed. Academic Press, San Diego. WI DNR. 2012a. Wisconsin Lake Maps. Available at http://dnr.wi.gov/lakes/maps/dnr/ (verified 14 May 2012). WI DNR. 2012b. Fisheries Management. Marathon County historic lake documents. Marathon County, as of 2012. Wisconsin Department of Natural Resources printed file materials. Accessed May, 2012. Wilde, G. 1997. Largemouth Bass fishery responses to length limits. Fisheries 22(6): 14-23. Winemiller, K.O., and D.H. Taylor. 1982. Smallmouth Bass nesting behavior and nest site selection in a small Ohio stream. Ohio Journal of Science 82(5): 266-273. 95
Appendix A Figure 14: Frequency (n) at Length (cm) histograms for Lilly Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods. Table 28: Total catch and lengths (min/max/average) of species in Lilly Lake during the 2012 fyke net and seining surveys. Min Length Max Length Average Length Total Catch Species (cm) (cm) (cm) n Bluegill 2.8 21.8 8.5 199 Largemouth Bass 2.0 10.2 3.5 103 Iowa Darter 2.7 6.1 4.6 29 Pumpkinseed 6.3 17.5 10.3 8 Black Crappie 6.3 27.0 18.0 8 Bluegill x Pumpkinseed hybrid 12.1 19.7 15.4 4 Yellow Perch 3.7 6.7 5.2 2 Northern Pike 66.0 66.0 66.0 1 96
Figure 15: Frequency (n) at Length (cm) histograms for Mayflower Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods. Table 29: Total catch and lengths (min/max/average) of species in Mayflower Lake during 2012 fyke netting and seining efforts. Min Length Max Length Average Length Total Catch Species (cm) (cm) (cm) n Bluegill 1.1 17.8 8.4 159 Largemouth Bass 1.9 18.5 3.0 95 Pumpkinseed 6.0 18.2 9.0 55 Bluegill x Pumpkinseed Hybrid 4.9 18.8 7.6 38 Yellow Perch 3.2 16.1 5.5 6 Black Crappie 17.3 24.5 20.4 4 Walleye 41.5 64.7 49.9 3 Black Bullhead 27.5 36.3 31.9 2 Northern Pike 61.5 72.1 66.8 2 97
Figure 16: Frequency (n) at Length (cm) histograms for Big Bass Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods. Table 30: Total catch and lengths (min/max/average) of species in Big Bass Lake during the 2012 fyke netting and seining surveys. Min Length Max Length Average Length Total Catch Species (cm) (cm) (cm) n Bluegill 2.2 25.5 9.3 345 Largemouth Bass 3.0 37.5 11.1 166 Black Crappie 6.4 17.7 14.9 51 Iowa Darter 3.4 6.0 5.0 26 Yellow Bullhead 12.9 39.1 30.3 21 Yellow Perch 9.8 31.4 24.0 17 Walleye 52.3 64.3 60.2 3 Black Bullhead 38.6 38.6 38.6 1 Golden Shiner 14.5 14.5 14.5 1 98
Figure 17: Frequency (n) at Length (cm) histograms for Rice Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods. Table 31: Total catch and lengths (min/max/average) of species in Rice Lake during the 2011 fyke net survey. Min Length Max Length Average Length Total Catch Species (cm) (cm) (cm) n Bluegill 3.8 21.5 14.1 71 Yellow Bullhead 21.5 38.1 26.9 29 Golden Shiner 3.9 5.9 5.4 11 Black Crappie 18.4 29.0 22.7 7 Pumpkinseed 9.8 21.6 15.3 6 Largemouth Bass 6.9 8.0 7.5 4 Brown Bullhead 26.6 29.3 27.5 3 Northern Pike 55.7 78.2 68.8 3 Yellow Perch 13.7 16.7 15.7 3 Mudminnow 3.7 6.6 5.2 2 99
Figure 18: Frequency (n) at Length (cm) histograms for Lost Lake in Marathon County, WI with varying sample gear (fyke net, seine). Refer to Table 3 for survey periods. Table 32: Total catch and lengths (min/max/average) of species in Lost Lake during the 2011 fyke netting and seining surveys. Min Length Max Length Average Length Total Catch Species (cm) (cm) (cm) n Bluegill 2.3 25.6 12.2 107 Yellow Bullhead 12.7 41.9 34.8 54 Golden Shiner 4.7 27.5 8.3 15 Largemouth Bass 7.8 45.5 17.7 6 Walleye 53.1 57.9 55.4 2 Black Crappie 16.5 16.5 16.5 1 Iowa Darter 6.1 6.1 6.1 1 Pumpkinseed 20.1 20.1 20.1 1 Yellow Perch 15.8 15.7 15.7 1 Northern Pike 30.7 30.7 30.7 1 100
Figure 19: Frequency (n) at Length (cm) histograms for Pike Lake in Marathon County, WI with varying sample gear (fyke net, seine, shock). Refer to Table 3 for survey periods. 101
Table 33: Total catch and lengths (min/max/average) of species in Pike Lake during 2012 fyke netting and seining efforts. Min Length Max Length Average Length Total Catch Species (cm) (cm) (cm) n Bluegill 2.1 27.7 9.8 177 Yellow Bullhead 19.9 33.5 24.6 138 Yellow Perch 8.2 20.5 12.1 107 Bluntnose Minnow 3.2 5.5 4.5 100 Black Crappie 12.4 38.9 24.4 37 Northern Pike 44.0 73.0 56.8 29 Pumpkinseed 11.5 18.2 15.1 17 Walleye 37.7 62.5 55.0 10 Johnny Darter 5.0 6.0 5.3 7 Largemouth Bass 28.0 39.5 35.2 5 Black Bullhead 17.8 31.9 24.9 2 Golden Shiner 4.4 18.2 11.3 2 Brown Bullhead 31.9 31.9 31.9 1 Iowa Darter 4.8 4.8 4.8 1 Spottail Shiner 6.7 6.6 6.6 1 White Sucker 18.4 18.4 18.4 1 Bluegill x Pumpkinseed hybrid 9.8 9.9 9.9 1 Table 34: Total catch and lengths (min/max/average) of species in Pike Lake during the 2012 boom shocking survey. Min Length Max Length Average Length Total Catch Species (cm) (cm) (cm) n Bluegill 3.556 20.066 10.668 51 Pumpkinseed 5.588 18.542 11.938 38 Yellow Bullhead 19.304 27.432 24.384 14 Black Crappie 8.128 22.86 15.494 10 Black Bullhead 13.208 33.02 27.178 7 Yellow Perch 7.874 17.018 12.954 7 Largemouth Bass 11.43 45.466 30.988 6 Northern Pike 39.878 51.562 46.736 5 Common Shiner 7.62 8.382 7.874 3 Bluntnose Minnow 4.318 6.858 6.604 2 White Sucker 20.574 23.368 22.098 2 Brown Bullhead 27.94 27.94 27.94 1 102
Figure 20: Frequency (n) at Length (cm) histograms for Mission Lake in Marathon County, WI with varying sample gear (fyke net, seine, shock). Refer to Table 3 for survey periods. 103