Ecological Limits of Hydrologic Alteration in Dane County Streams

Transcription

1 Fish Community Status Ecological Limits of Hydrologic Alteration in Dane County Streams Final Report April 24, 2014 Matthew Diebel, Aaron Ruesch, and Diane Menuz Wisconsin Department of Natural Resources 100% 80% 60% 40% 20% 0% 100% 80% 60% 40% Flow Remaining 20% 0%

2 Abstract Stream flow regime is a primary determinant of aquatic and riparian ecological structure and function. Water use, landscape modification, and climate change can all modify flow regimes, but there are few tools available to predict the effects of these changes on ecological conditions. The objective of this analysis is to predict the response of stream fishes to changes in stream flow that are expected to occur by 2035 due to changes in land use and groundwater use in Dane County, Wisconsin. The results of this analysis can be used to identify streams where mitigation of flow changes should be focused in the near future. Models were also used to predict the response of fish communities in each stream to a uniform 20% change in measures of spring high flow and summer low flow. This response is a measure of the ecological sensitivity of streams to flow changes, and can be used to identify sensitive areas in long range plans. By 2035, significant changes to fish communities are expected to occur in about 5% (34 miles) of the stream length in Dane County due to reduction in summer baseflow. These streams are primarily headwaters in or near Madison and the Yahara River downstream of Lake Waubesa. If stormwater controls are not implemented on new development, the resulting increase in runoff is predicted to cause significant changes to fish communities in about 2.5% (16 miles) of the stream length in Dane County, primarily in and around Fitchburg, Cottage Grove, and Sun Prairie. Streams that are most sensitive to a 20% reduction in summer low flow include headwaters throughout the county and the mainstem Yahara River and Koshkonong Creek. Streams that are most sensitive to increases in stormwater runoff are primarily headwaters in the eastern half of the county. To help decision makers, these models are integrated in a spreadsheet template which may be used to develop stream-specific curves describing fish community response to flow alterations. 1

3 Introduction It is widely recognized that stream flow regime is a primary determinant of aquatic and riparian ecological structure and function (Poff et al. 2010). Water use, landscape modification, and climate change can all modify flow regimes, but there are few tools available to predict the effects of these changes on ecological conditions. Scientifically-based water management requires methods for assessing the tradeoffs between alteration of natural water flow patterns and consequent changes in ecological health. These relationships can serve as the foundation of social and political processes for balancing the benefits of water use and landscape modification with those that derive from healthy, functioning ecosystems. In addition, they can help design management strategies that are robust to potential climate changes. The Ecological Limits of Hydrologic Alteration (ELOHA) framework provides an approach for determining environmental flows the quantity, timing, and quality of water flows required to sustain freshwater ecosystems and the human livelihoods and well-being that depend on these ecosystems (Arthington et al. 2006, Poff et al. 2010). ELOHA is designed to be applied at regional scales in cases where site-specific studies cannot be performed for all streams in a region. ELOHA synthesizes existing hydrologic and ecological databases from many streams in a region to generate flow alteration-ecological response relationships that can then be used to simulate the ecological effects of various human activities that alter stream flows. For example, Michigan is using this approach to support review of groundwater withdrawal permit applications (Hamilton and Seelbach 2011). ELOHA frameworks may be used in many stages of decision making, including broad discussions of issues, long-range planning, land use zoning, public land acquisition, and permit review. From a technical perspective, an ELOHA framework is composed of hydrologic and ecological models. The hydrologic models relate metrics of climate, natural landscape characteristics, and human activities to temporal and spatial variation in stream flows. The ecological models relate hydrologic and other stream characteristics to one or more ecological characteristics, such as fish species composition. When linked, these models can not only be used to simulate how various human activities, such as groundwater withdrawal, will affect stream flows, but how those changes in flow will be manifested in ecological characteristics. In other words, how much change in flow can a stream tolerate before an unacceptable ecological impact occurs? 2

4 This report describes an ELOHA analysis for streams in Dane County, Wisconsin. The analysis was conducted using a set of hydrologic and ecological models developed for the entire state of Wisconsin by the Wisconsin Department of Natural Resources (WDNR) (Diebel et al. 2014), and from a groundwater model of Dane County developed by the Wisconsin Geological and Natural History Survey (WGNHS). The ecological models use fish species composition as a surrogate for overall biological integrity. The main objective of the analysis is to predict the response of stream fishes to changes in stream flow that are expected to occur by 2035 due to changes in land use and groundwater use. The results of this analysis can be used to identify streams where mitigation of flow changes should be focused in the near future. The same models were also used to predict the response of fish communities in each stream to a uniform 20% change in measures of spring high flow and summer low flow. This response is a measure of the ecological sensitivity of streams to flow changes, and can be used to identify sensitive areas in long range plans. To help decision makers, these models are integrated in a spreadsheet template which may be used to develop stream-specific curves describing fish community response to flow alterations. Methods The analyses described in this report were conducted using a set of hydrologic and ecological models developed for the entire state of Wisconsin. The reader is referred to the report for this statewide project (Diebel et al. 2014), called the Wisconsin ELOHA project herein, for details of these models. Briefly, the hydrologic models (flow models) are a set of regression models that relate 23 stream flow metrics at U.S. Geological Survey stream gages to weather data and watershed characteristics. The models can be used to predict stream flow at any stream segment (average length = 900 m) in Wisconsin s 1:24,000-scale hydrography. The flow predictions for each metric (e.g., August median flow) are the geometric means of annual predictions from , and therefore represent the typical flow across a range of weather conditions. The ecological models are a set of stream fish species distribution models that relate the presence/absence of 79 fish species in WDNR fish surveys to a large set of stream and watershed characteristics, including modeled stream flow and temperature metrics. The flow metrics were predicted using the flow models described above. The temperature metrics were predicted with an artificial neural network (ANN) model that links measured daily water 3

5 temperatures to geology, topography, climate, and land cover variables. This model was described in more detail in Roehl et al. (2006), Stewart et al. (2006) and Lyons et al. (2009). The fish models can be used to predict the probability of occurrence of each species in any Wisconsin stream. The fish models can also be used to simulate how changes in flow will change the habitat suitability for each species. A habitat suitability function is a graphical depiction of how the probability of occurrence of a species changes with variation in one of the model variables, while holding all other variables constant at their average values. Habitat suitability functions were created for three variables (Appendices A, B, and C). April 10% exceedance flow is a measure of the high flow during a period when spawning activities (e.g., migration, nest construction) for many fishes may be influenced by stream flow. April 10% exceedance flow is also correlated with other high flow metrics, including annual 5% exceedance flow (r=0.86), spring 10% exceedance flow (r=0.85), and summer 10% exceedance flow (r=0.64), so it can be considered a general indicator of runoff potential. August median stream flow is a standard measure of baseflow or low flow in streams, and is considered a limiting factor for many stream fishes (Zorn et al. 2008). To separate the effects of stream size from flow, these metrics were expressed as flow yields by dividing discharge by drainage area. July mean water temperature is a standard measure of water temperature during the warmest time of the year and is used to classify streams into thermal classes (Lyons et al. 2009). Baseflow reductions increase stream temperature during periods when air temperature is warmer than water temperature (i.e., summer) because the reduced water depth and velocity increase the rate of heat exchange with atmosphere and the time available for radiative heating. We used an approach similar to the one used by Michigan (Zorn et al. 2008) to estimate the change in July mean temperature that would result from a given reduction in August median flow yield (see Diebel et al for details, July and August flow and temperature conditions were assumed to be equivalent and were used interchangeably). Scenarios We used the models described above, with a few modifications, to predict fish response to hydrologic alterations that are projected to occur in the next 20 years in Dane County. We organized these projections into scenarios, which are summarized here and described in detail below. The 2035 Low Flow Scenario is based on changes in stream baseflow that are projected 4

6 to result from projected groundwater withdrawal and recharge rates in The effect of these changes on stream base flows was modeled by WGNHS with a MODFLOW groundwater model. Low Flow Sensitivity illustrates the relative sensitivity of stream fishes to a uniform 20% reduction in base flow in each stream. The No-Controls Stormwater Scenario is based on changes in spring high flow that are projected to occur if existing stormwater management regulations are not implemented on new development. The With-Controls Stormwater Scenario is based on changes in spring high flow that are projected to occur if existing stormwater management regulations are implemented on new development. Stormwater Sensitivity illustrates the relative sensitivity of stream fishes to a uniform 20% increase in spring high flow. The effects of each future scenario on stream fishes are represented as changes from the Present Scenario, which represents conditions in Stream Map The spatial extent of the study area is all streams in Dane County, Wisconsin. Streams were initially mapped with Wisconsin s 1:24,000-scale flowlines. The flowlines were manually cross-referenced with surface water routing cells in the groundwater model. Flowlines whose downstream ends corresponded with a surface water routing cell with zero flow in the Present Scenario were removed from the stream layer because they were assumed to not be fish habitat. Flowlines included in the final layer can generally be assumed to flow perennially. The final stream layer contains 953 stream segments, which have REACHID as a unique identifier. The stream spatial data layer also contains segments that connect flow paths through lakes, but no lake level or lake fish community projections were made by this study. Dane County streams include examples of all classes of stream natural communities that exist in Wisconsin (Table 1 and Appendix D3). The most common stream types are coldwater and cool-cold headwater, which are commonly thought of as trout streams. 5

7 Table 1. Total length of perennial streams in each natural community class in Dane County. Stream natural communities are defined by flow and temperature. A substantial number of additional intermittent streams exist, but were not evaluated in this analysis. Natural Community Total Length (miles) Macroinvertebrate 19 Coldwater 131 Cool-Cold Headwater 153 Cool-Cold Mainstem 94 Cool-Warm Headwater 95 Cool-Warm Mainstem 51 Warm Headwater 32 Warm Mainstem 56 Total Low Flow Scenario Changes in baseflow in Dane County streams by 2035 were estimated with the MODFLOW groundwater model developed by WGNHS (Bradbury & Parsen in prep.). This model simulates groundwater recharge and discharge to surface waters. It also includes withdrawals from groundwater wells and dicharges from wastewater facilities. Stream flow estimates from this model are baseflow, which is the component of stream flow that derives from groundwater discharge. Groundwater recharge rates were not changed from the present scenario under the assumption that current stormwater management regulations would maintain infiltration rates on newly developed land. Groundwater withdrawal by high capacity wells was modified based on a survey of Dane County water utilities that provided locations and specifications (i.e., depth, pumping rate) of planned wells. Wastewater discharges were maintained at 2013 levels in the 2035 scenario, so flows in streams that receive wastewater effluent may be underestimated. Estimated 2035 wastewater discharges will be added or simulated at a later date, when they are available (M. Kakuska, CARPC, personal communication), demonstrating the utility of the project and models in expected ongoing simulations and resource management being conducted in the region. 6

8 MODFLOW Discharge (cfs) The changes in July mean water temperature that would result from the predicted changes in baseflow were estimated by translating the change in flow into a change in water depth and velocity and consequent increased heating (see Diebel et al for details). We also checked if the baseflow estimates from the MODFLOW model were approximately equivalent to the August median stream flow used in the fish distribution models. We expected there to be differences between the flow estimates at specific locations resulting from the different methods used to generate them, so the goal of the comparison was to verify a lack of bias and a reasonable degree of correlation (Figure 1). Figure 1. Comparison of present baseflow discharge estimated by MODFLOW groundwater model with discharge measured at 143 locations in August Values < 0.1 (including zero values) are plotted as 0.1 cfs to allow presentation on a log scale. 29 sites had both measured and modeled discharge < 0.1 cfs Measured Discharge (cfs) 7

9 The effect of the predicted change in stream baseflow on stream fishes was estimated as follows. First, the species that are predicted to occur in each stream segment were identified as those with a probability of occurrence > 0.5 in the statewide species distribution models. For some species, a higher probability threshold (up to 0.65) was used to improve the correspondence between observed and predicted distributions in Dane County. For each species, the habitat index of each stream segment was calculated as the product of the suitability indices for August median flow yield and July mean temperature (Appendices A and B). MODFLOW baseflow estimates were used for August median flow yield and the ANN temperature model described above was used for July mean temperature. This approach assumes that both temperature and flow must be suitable for overall habitat to be suitable. For each species, the relative habitat suitability of each stream segment in 2035 was calculated as the habitat index in the Present Scenario divided by the habitat index in the 2035 Scenario. Finally, the fish community status for each segment in 2035 was calculated as the average relative habitat suitability for all of the species in that segment. Fish community status and all of its component indices are scaled from 0 to 1 to facilitate comparison among streams and scenarios. For discussion purposes in this report, a significant change in fish community status is designated as a 20% change from the Present Scenario (i.e., fish community status equals 0.8 or less from a baseline of 1). However, in practice, other thresholds may be used, preferably if they are selected with stakeholder involvement. Low Flow Sensitivity The flow changes predicted for 2035 vary a lot, and many streams are predicted to have little to no change in flow. However, because development and water use patterns may play out differently than predicted, and to provide a tool for planning beyond 2035, we evaluated how the fish communities in each stream in the county would respond to 10% increments of flow reduction down to 90% reduction. In this report, we focus on the effects of a uniform 20% flow reduction for simplicity and because that is a feasible outcome in many places. The effects of all flow reduction increments can be reviewed in the spreadsheet accompanying this report (see Appendix E for instructions). This spreadsheet can also be used to create fish response curves for individual streams, which depict the response of the fish community across the full range of potential flow reductions (e.g., Figure 2). We used the same approach with the same set of 8

10 Fish Community Status indices as for the 2035 Low Flow Scenario, including predicting the changes in July mean temperature that would result from each flow reduction increment. Figure 2. Example fish response curve for Black Earth Creek upstream of Cross Plains, WI (Reach ID ). 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 100% 90% 80% 70% 60% 50% 40% Flow Remaining 30% 20% 10% 0% Stormwater Scenarios Stormwater runoff from impervious surfaces increases peak flows in streams, which can harm fish through several mechanisms. During high flow events, increasing flow velocity and turbidity make it difficult for fish to hold position and find food, and can wash away or bury incubating eggs. In addition, high flow events can cause long-lasting changes in channel form and substrate, which may make habitat unsuitable for some species. A significant amount of urban development is expected to occur in Dane County in the near future. Current regulations require that all new development in the county include 9

11 April 10% Exceedance Flow stormwater runoff controls that will capture and retain 90% of the annual runoff volume. CARPC staff provided GIS layers of Dane County parcels with impervious percentages that represent planned development 1 with and without stormwater controls that meet current standards specified in Dane County Ordinances Chapter (Table 2). These layers were used to create values of adjusted soil permeability (SSURGO soil permeability (100 - percent impervious) and predict changes in April 10% exceedance flow for each stream reach in the county for each scenario (Figure 3). To evaluate the effect of these flow changes on fish communities, we used the same approach as for the 2035 Low Flow Scenario, except with the habitat suitability functions for April 10% exceedance flow for each species (Appendix C). Figure 3. Relationship between adjusted soil permeability and relative change in April 10% exceedance flow based on model described in Diebel et al For example, if a watershed initially had an average soil permeability of 4 in/hour, and new impervious surface reduced it to 2 in/hour, the April 10% exceedance flow would increase by (111% - 96%) / 96% = 16%. 140% 130% 120% 110% 100% 90% 80% 70% Adjusted Soil Permeability (in/hour) 1 The map of planned development for Dane County was created by merging many local plans developed by municipalities. The planning horizon year for local plans varies significantly, depending upon when the plan was prepared. The map used in this study was current as of 5/2013. The current map, titled Existing and Planned Future Land Use is available from: 10

12 Table 2. Percent impervious surfaces by land use category with and without stormwater controls currently required on new development in Dane County. Percent Impervious Land Use Category With Without Controls 2 Controls Open Land 0 0 Woodland 0 0 Commercial Forest 0 0 Agriculture 2 2 Cemetery 2 2 Extractive 2 2 Outdoor Recreation 2 2 Under Construction 2 2 Vacant 2 2 Single-Family Residential Communications or Utilities Institutional or Governmental Two-Family Residential Industrial Right of Way Transportation Commercial Sales or Services Multi-Family Residential The percent impervious with controls column in table 2 is representative of the how a developed lot would function if the stormwater were sent to infiltration devices designed to Dane County infiltration standards. As long as the treatment device is designed to the Dance County infiltration standards, it will function similarly to a parking lot utilizing 15% of the land regardless of the type of development draining to the treatment device. For example, a pasture in Dane County will only have 1.72 inches of runoff thorough out the year. If that same pasture were developed with dense residential houses without treating the stormwater, the amount of runoff would increase to inches per year. However, if the storm water were treated with infiltration devices designed to meet the Dane County Standard the amount of runoff would be 4.4 inches. The sizing of treatment devices is site specific but since all new development in Dane County is required to meet the same standard (except Right of Way), it is possible to represent treated impervious area as a smaller untreated impervious area on the same sized parcel. In the above example, the 4.4 inches of runoff is equivalent to constructing a parking lot on 15% of the original pasture. Stormwater Sensitivity As with baseflow, high flow characteristics are only expected to change in the few streams whose watersheds are expected to experience significant development in the near future. To provide a consistent measure of sensitivity to changes in high flows, we evaluated how the fish communities in each stream in the county would respond to a uniform 20% increase in the April 10% exceedance flow, using the same approach as for the Stormwater Scenarios. 11

13 Results Present Scenario A summary of current conditions in Dane County streams can be portrayed with maps of baseflow yield, July mean temperature, and stream natural communities (Appendix D1, D2, and D3). Baseflow yield is discharge per unit watershed area, and is generally higher in the western part of the county. Notable exceptions include Token and Nine Springs Creeks, which are both spring-fed, and Badfish Creek, whose flow is supplemented by discharge from the Madison sewage treatment plant. The pattern in temperature mirrors the pattern in baseflow, with colder temperatures associated with high baseflow yield. Also, most stream systems warm in a downstream direction during the summer because the water temperature has had more time to equilibrate with the atmosphere. Stream size and temperature are the basis for WDNR s stream natural community classification, which indicates the expected biological community (J. Lyons, WDNR, unpublished report) Low Flow Scenario By 2035, projected increases in groundwater withdrawal in Dane County will significantly reduce baseflow in many streams. These streams are mainly located in and around Madison, where the most significant withdrawals are planned (Appendix D4). The largest flow reductions (>50%) are expected in Wingra Creek, Starkweather Creek, and in unnamed tributaries to Upper Mud Lake, Token Creek, and the Yahara River. July water temperature is only expected to increase significantly in a few of these streams, mainly because many of them are already warm or cool-warm streams (Appendix D5). As described in the methods, this scenario does not account for increases in wastewater effluent volume that are likely to occur in some streams by In particular, increased effluent volumes to Badger Mill, Badfish, and Koshkonong Creeks are likely to at least partially offset the predicted reduction in baseflow. However, the increased contribution of wastewater compared to natural stream flow is likely to result in increased nutrient concentrations and temperature, the effects of which are beyond the scope of this study. By 2035, significant changes (> 20%) to fish communities are expected to occur in about 5% (34 miles) of the stream length in Dane County due to reduction in summer baseflow (Table 12

14 4, Figure 5). These streams are primarily headwaters in or near Madison and the Yahara River downstream of Lake Waubesa (Appendix D6). These streams are predicted to support a variety of fish species (Table 3). Figure 4 illustrates how for a given level of flow reduction, the fish communities in different streams respond differently. For example, baseflow in Badger Mill Creek and Fryes Feeder is expected to decrease by 22% by 2035, but the fish community in Fryes Feeder is much more sensitive to that level of flow reduction. This difference is due to differences in the species assemblages of the two streams (Table 3) and to the lower baseline flow yield of Fryes Feeder. Sensitivity to baseflow reduction can even vary substantially among segments of the same stream. For example, in one segment of Starkweather Creek, a 70% flow reduction would not alter the fish community status because this segment is only expected to contain one species, green sunfish, which is not sensitive to this level of flow reduction (Figure 4). In contrast, a 67% flow reduction in another segment of Starkweather Creek would reduce the fish community status to 73% because this segment is expected to also contain bluegill and largemouth bass, which are more sensitive to flow reduction. In any given stream, the species that are most sensitive to reductions in summer baseflow are those for which the existing conditions are near thresholds in their tolerance to flow and/or temperature. For example, in Badger Mill Creek, summer baseflow is expected to decrease by about 21%, which would increase temperature slightly (+0.3 C). The three species that would experience significant declines (American brook lamprey, brown trout, and mottled sculpin) are all sensitive to changes in both flow and temperature in the range of the stream s current conditions. Most of the sensitive species in other streams are tolerant of warm temperatures, so their predicted declines are directly related to expected changes in flow (see discussion for potential mechanisms). 13

15 Table 3. Lists of the most sensitive species to flow reductions in each stream where significant changes to the fish community are expected to occur in the 2035 Low Flow Scenario. Stream Badger Mill Creek Door Creek Fryes Feeder Koshkonong Creek Starkweather Creek Swan Creek Wingra Creek Yahara River Token Creek Trib Upper Mud Lake Trib Species American brook lamprey, brown trout, mottled sculpin Black bullhead, brook stickleback, central mudminnow, fathead minnow Southern redbelly dace, central stoneroller, mottled sculpin Freshwater drum, northern pike Green sunfish, largemouth bass Brook stickleback, central mudminnow, fathead minnow Black crappie, bluegill, largemouth bass Black crappie, bowfin, emerald shiner, freshwater drum, northern hog sucker, northern pike, walleye Central mudminnow Central mudminnow 14

16 Figure 4. Plot of fish community status versus projected baseflow reduction by 2035, with selected streams labelled as examples. Fish in streams in the lower left and upper right of the plot are more and less sensitive, respectively, to flow reductions. 100% 90% 80% 70% Starkweather Creek Swan Creek Door Creek Badger Mill Creek Koshkonong Creek Swan Creek Badger Mill Creek Yahara River Door Creek Starkweather Creek Yahara River Fish Community Status 60% 50% 40% Door Creek Wingra Creek Starkweather Creek Starkweather Creek Wingra Creek Starkweather Creek 30% Door Creek 20% 10% Fryes Feeder Swan Creek 0% 100% 90% 80% 70% 60% 50% Flow Remaining 40% 30% 20% 10% 0% 15

17 Low Flow Sensitivity Streams that are most sensitive to a 20% reduction in summer low flow include headwaters throughout the county and the mainstem Yahara River and Koshkonong Creek (Appendix D7). A total of 99 stream miles would experience a significant change in the fish community (i.e., > 20% change, Table 4, Figure 5). At the other end of the spectrum, habitat suitability would not change at all for fish communities in 260 stream miles (Figure 5). Table 4. Total length (miles) of perennial streams in each natural community class in Dane County that would experience significant changes in fish community status (>20% change) in each scenario. Natural Community Total Length No-Controls Stormwater 2035 Baseflow Stormwater Sensitivity Baseflow Sensitivity Macroinvertebrate Coldwater Cool-Cold Headwater Cool-Cold Mainstem Cool-Warm Headwater Cool-Warm Mainstem Warm Headwater Warm Mainstem Total Stormwater Scenarios If stormwater controls are not implemented on planned development (i.e., No Controls Stormwater Scenario), April 10% exceedance flow is predicted to increase by as much as 40% in some streams due to the increased impervious surface area (Appendix D8). These flow changes are predicted to cause significant changes to fish communities in about 2.5% (16 miles) of the stream length in Dane County, primarily in and around Fitchburg, Cottage Grove, and Sun Prairie (Table 4, Figure 5, Appendix D9). All of the species that would be significantly affected in these streams are warm water species (Table 5). In contrast, if controls required by current stormwater regulations are implemented on planned development (i.e., With Controls 16

18 Stormwater Scenario), April 10% exceedance flow will change very little, increasing by a maximum of 6% (Appendix D10). These flow changes would have minor effects on fish communities (Appendix D11). Note that both of these findings only apply to the effects of development on the balance between groundwater recharge and runoff, and not to water quality. Table 5. Lists of the most sensitive species to flow reductions in each stream where significant changes to the fish community are expected to occur in the No Controls Stormwater Scenario. Stream Door Creek Koshkonong Creek Nine Springs Creek Swan Creek Maunesha River Tribs Species Central mudminnow Bluntnose minnow, common shiner, largemouth bass, stonecat Largemouth bass, brook stickleback Central mudminnow Central mudminnow, fathead minnow Stormwater Sensitivity Streams that are most sensitive to increases in stormwater runoff are primarily headwaters in the eastern half of the county (Appendix D12). As a whole, the fish communities in Dane County streams are more sensitive to a 20% increase in the April 10% exceedance flow than a 20% decrease in the August median flow; a total of 192 stream miles (30% of stream miles in Dane County) would experience a significant change in the fish community (i.e., > 20% change, Table 4, Figure 5) if April 10% exceedance flows were to increase by 20%. 17

19 Stream Miles Stream Miles Figure 5. Cumulative distribution of fish community status by stream length in four scenarios (see methods for scenario descriptions). Bars at Fish Community Status = 1 are truncated, with labels above bars indicating stream miles where fish community status would not change Low Flow Scenario % 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Fish Community Status (% of Baseline) Low Flow Sensitivity % 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Fish Community Status (% of Baseline) 18

20 Stream Miles Stream Miles Figure 5 (continued) No-Controls Stormwater Scenario % 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Fish Community Status (% of Baseline) Stormwater Sensitivity % 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Fish Community Status (% of Baseline) 19

21 Figure 6. Sensitivity of fish species to a 20% reduction in baseflow in all streams in Dane County. For example, a 20% reduction in baseflow would cause the habitat suitability for brown trout to drop to ~80% of its current value in streams where it currently occurs. 20

22 Figure 7. Sensitivity of fish species to a 20% increase in high flow in all streams in Dane County. For example, a 20% increase in high flow would cause the habitat suitability for brown trout to drop to 73% of its current value in streams where it currently occurs. 21

23 Discussion Effective groundwater management and land use planning require not only understanding how various actions will affect groundwater levels, lake levels, and stream flows, but how ecosystems will respond to those changes. This study integrates detailed models that predict stream flow changes in Dane County with ecological models that predict the response of fish communities. The results indicate that stream fish communities vary in their sensitivity to changes in both baseflow and high flows. This variability in sensitivity is caused by the following set of interrelated factors. First, Dane County streams are highly variable in their physical characteristics, including temperature, flow regime, and gradient. As a consequence, the county s streams contain diverse fish assemblages. Second, fish species vary substantially in their tolerance to flow and temperature. And third, even among streams where a particular species occurs, the streams in which that species is most sensitive to environmental changes are near thresholds in its tolerance. The fish species that shape the overall patterns of flow sensitivity are those that are both common and sensitive to flow changes (Figures 6 and 7). For example, brown trout and mottled sculpin are very common in Dane County, occurring in over 200 stream miles, and are also moderately sensitive to baseflow reduction (Figure 6). In contrast, some of the larger river species, such as quillback, are very sensitive to baseflow reduction, but are only present in a few streams. Brown trout and mottled sculpin are also sensitive to increases in high flows, as is central mudminnow (Figure 7). The predicted responses of fish to flow changes are based on modeled correlations between flow variables and the observed distributions of each species. This approach is based on the hypothesis that flow changes at a site will change the fish assemblage to resemble the fish assemblage at an otherwise similar site that shares its lower discharge and associated hydrologic characteristics (e.g., water temperature, current velocity, or depth). An alternative approach would be to conduct an experiment where flow is reduced and the response (e.g., fish density, growth, reproduction) is observed (see Nuhfer & Baker 2004 for an example). However, an experimental approach would be impractical to use over a large area with a variety of stream types, such as Dane County. The fish distribution models used in this study also control for the influence of other natural and anthropogenic variables that influence fish distributions, such as geology, soils, land cover, and climate, which makes the fish-flow relationships more robust. 22

24 While these relationships do not prove cause-and-effect, there is a lot of other evidence that supports the importance of flow and temperature for fish metabolism, reproduction, survival, distribution, and abundance (Poff and Ward 1989, Lyons 1996, Zorn et al. 2002, Wehrly et al. 2003, Zorn et al. 2004, Lyons et al. 2009). While this study is a useful tool for predicting the ecological effects of stream flow changes, there are several limitations that should be acknowledged. First, the models only predict the probability of occurrence of each species, not the abundance or density. In preliminary analyses, we did not observe strong relationships between abundance of most species and modeled flow or temperature. This finding may have been the result of variation in other stream characteristics that were more important or that obscured these relationships. In contrast, there were strong relationships between flow and temperature and the occurrence (presence/absence) of most species. Second, other stressors may interact with or exacerbate the effects of flow changes, such as habitat and water quality alteration, climate change, invasive species, and fishing pressure, so the predicted fish responses should be considered minimum estimates. Third, while the flow and fish models are relatively accurate, and the best currently available for Dane County, there are certainly prediction errors that mean that many streams are more or less sensitive to flow changes than this analysis suggests. For example, several streams that had measureable flow in August 2011 have zero predicted baseflow in the MODFLOW model (Figure 1). These streams are probably fish habitat at least part of the year, but are not included in this analysis. Also, because there are errors in predicted fish assemblages, fish survey data can be substituted for model predictions where available. In addition, the predicted responses of many large river fish species, such as emerald shiner and freshwater drum, to flow reductions are probably not as robust as those for smaller stream species because there were few large rivers in the fish model dataset that have low baseflow yields (Diebel et al. 2014). Because this issue mostly applies to the Yahara River, it may be worth using a different approach in that system to evaluate the ecological effects of the flow changes that are predicted to occur. This analysis should be considered a first step in understanding and predicting the effects of hydrologic changes on streams in Dane County. There are several areas where improvements could be made in the future. First, future flow predictions for streams that receive wastewater effluent could be improved by estimating how effluent volumes are likely to change in incorporating those inputs into the MODFLOW model. Second, future versions of the fish 23

25 models should attempt to relate inter-annual variability in stream flows to the persistence and abundance of species so that the effects of changes in the frequency and severity of extremely wet or dry conditions can be evaluated. The effects of flow and temperature variability on stream fishes can be buffered if they can take temporarily refuge in connected parts of the stream network. Consequently, barriers to fish movement such as dams and road crossings should also be included in future versions of the fish models. Finally, it would be prudent to establish a set of monitoring sites on streams where flow and fish changes are expected to document whether those changes actually occur. Acknowledgments This report is the product of collaboration between the Capital Area Regional Planning Commission and the Wisconsin Department of Natural Resources, with additional support from the U.S. Geological Survey and the Wisconsin Geological and Natural History Survey. The authors appreciate assistance by Mike Carlson, Troy Zorn, Mike Parsen, Ken Bradbury, Jana Stewart, Steve Westenbroek, and John Lyons. The work that provided the basis for this publication was supported by funding under an award with the U.S. Department of Housing and Urban Development. The substance and findings of the work are dedicated to the public. The author and publisher are solely responsible for the accuracy of the statements and interpretations contained in this publication. Such interpretations do not necessarily reflect the views of the Government. 24

26 References Arthington A.H., Bunn S.E., Poff N.L. & Naiman R.J The challenge of providing environmental flow rules to sustain river ecosystems. Ecological Applications 16: Diebel, M.W., Ruesch, A., & Menuz, D.R Ecological Limits of Hydrologic Alteration in Wisconsin Streams. Project Report to the Wisconsin Department of Natural Resources, Bureau of Drinking Water and Groundwater. Hamilton, D.A., and P.W. Seelbach Michigan s Water Withdrawal Assessment Process and Internet Screening Tool. Michigan Department of Natural Resources, Fisheries Special Report 55, Lansing. Lyons, J Patterns in the species composition of fish assemblages among Wisconsin streams. Environmental Biology of Fishes 45: Lyons, J., T. Zorn, J. Stewart, P. Seelbach, K. Wehrly, and L. Wang Defining and characterizing coolwater streams and their fish assemblages in Michigan and Wisconsin, USA. North American Journal of Fisheries Management 29: Nuhfer, A. J., and E. A. Baker A long-term field test of habitat change predicted by PHABSIM in relation to brook trout population dynamics during controlled flow reduction experiments. Michigan Department of Natural Resources, Fisheries Research Report 2068, Ann Arbor. Poff, N. L., and J. V. Ward Implications of streamflow variability and predictability for lotic community structure: a regional analysis of streamflow patterns. Canadian Journal of Fisheries and Aquatic Sciences 46: Poff, N. L., Richter, B. D., Arthington, A. H., Bunn, S. E., Naiman, R. J., Kendy, E., Acreman, M., Apse, C., Bledsoe, B. P., Freeman, M. C., Henriksen, J., Jacobson, R. B., Kennen, J. G., Merritt, D. M., O keeffe, J. H., Olden, J. D., Rogers, K., Tharme, R. E. And Warner, A The ecological limits of hydrologic alteration (ELOHA): a new framework for developing regional environmental flow standards. Freshwater Biology 55: Roehl, E., J. Risley, J. Stewart, and M. Mitro Numerically optimized empirical modeling of highly dynamic, spatially expansive, and behaviorally heterogeneous hydrologic systems, Part 1. In Proceedings of the iemss Third Biennial Meeting: Summit on Environmental Modelling and Software (Voinov, A., A.J. Jakeman, and A.E. Rizzoli, eds). International Environmental Modelling and Software Society, Burlington, USA, July Stewart, J., M. Mitro, E.A. Roehl, Jr., and J. Risley Numerically optimized empirical modeling of highly dynamic, spatially expansive, and behaviorally heterogeneous hydrologic systems Part 2, In: Hydroinformatics: Proceedings of the 7th International Conference, Nice, France, September

27 Wehrly, K. E., M. J. Wiley, and P. W. Seelbach Classifying regional variation in thermal regime based on stream fish community patterns. Transactions of the American Fisheries Society 132: Zorn, T. G., P. W. Seelbach, and M. J. Wiley Distributions of stream fishes and their relationship to stream size and hydrology in Michigan's Lower Peninsula. Transactions of the American Fisheries Society 131: Zorn, T. G., P. W. Seelbach, and M. J. Wiley Utility of species-specific, multiple linear regression models for prediction of fish assemblages in rivers of Michigan s Lower Peninsula. Michigan Department of Natural Resources, Fisheries Research Report 2072, Ann Arbor. Zorn, T.G., Seelbach, P.W., Rutherford, E.S., Wills, Todd, Cheng, Su-Ting, and Wiley, M.J A landscape-scale habitat suitability model to evaluate impacts of groundwater withdrawals on fish communities in Michigan streams. Michigan Department of Natural Resources Fisheries Research Report 2089, 46 p. 26