Steven I. Apfelbaum1, John. D. Eppich2 and James A. Solstad3
　　1. Applied Ecological Services, Inc., Chairman/Founder and Senior Ecologist, Wisconsin 53520, USA
　　2. Waterflow Consultants, Environmental Engineer, Illinois 61820, USA
　　3. Minnesota Department of Natural Resources, Division of Waters, Technical Analysis Program Hydrologist, Minnesota 55155, USA
　　Received: March 12, 2011 / Accepted: July 19, 2011 / Published: January 20, 2012.
　　Abstract: The authors assessed if wetlands can contribute to flood damage reduction in the Red River Basin, Minnesota, by providing reliable flood water storage. Hydrology and biodiversity in 28 natural and restored wetlands suggested uncontrolled natural wetlands provided the highest mean annual flood storage at 15 cm of runoff while single and 2-stage outlet controlled wetlands provided 3.0 and 8.1 cm of runoff control. Natural controlled wetlands, followed by 2-stage and single stage outlet controlled restorations provided 10.2, 6.6, and 2.2 cm of storage for early summer storm events. Two years of recorded water levels and a 20-year continuous meteorological record were used to model “temporary water level increases” in each wetland. Species diversity, hydrology, and watershed land use variables are inversely related where high quality and diverse wetlands had the lowest amplitude and frequency of water level increases, while low quality wetlands had the highest. Uncontrolled natural wetlands had the highest biological diversity and the lowest frequency and magnitude of temporary water levels increased. A significant biodiversity declines were measured where water level increases were greater than 2.7 meters. Strong multi-linear relationships between watershed land uses and watershed/wetland ratio explained wetland hydraulic performance and biodiversity relations (r2 ranging from 0.6-0.8). Non-native wetland plant diversity increased with greater water level dynamics.
　　Key words: Diversity and hydrology, floodwater in wetlands, water level dynamics.
　　 1. Introduction
　　The Minnesota, USA portion of the Red River of the North basin includes 4.5 million hectares of glacial moraines, glacial lake plains, and beach ridges. Major flooding has frequently occurred along the Red River and its major tributaries. Levees now protect most urban areas, and efforts to minimize flood damages to rural areas include building impoundments and improving drainage. However, vast areas of productive agricultural lands are still subject to periodic flooding.
　　Local watershed districts have constructed numerous flood damage reduction (FDR) projects in the basin. However, during the 1990s, growing environmental concerns about the cumulative impacts of these projects, especially on-channel impoundments, led to gridlock in the permitting of new projects. In 1997, the Minnesota Legislature funded a mediation process to reconcile differences among key stakeholder groups, including state and federal agencies, local watershed districts, environmental groups, and individual citizens. A Flood Damage Reduction Working Group (FDRWG) comprised of agency and citizen participants, and a Technical and Scientific Advisory Committee (TSAC), comprised of engineers and biologists, were formed to address the future of flood management in the Red River Basin. TSAC provided technical support to address specific issues raised by the FDRWG.
　　One key issue was the potential use of wetland restorations for providing flood protection, while enhancing biodiversity. These wetlands may temporarily store excess runoff, reducing downstream flooding. Similar to impoundments, the greater the amount of runoff detained by a wetland, the greater the potential flood damage reduction. For any given wetland restoration project, potential runoff storage during a flood event could be increased by allowing a greater temporary water level increase.
　　On the other hand, wildlife managers generally try to minimize temporary water level increases when restoring wetlands due to concerns about nesting birds and possible negative impacts to other biota. Recent land-use and cultural changes in the basin have biological consequences [1-3], and flood behavior of the Red River and the strategies to reduce damage have also changed [4-8]. Elsewhere in the U.S. an increasing magnitude and frequency of water level deviation and flooding duration from storm events is believed to be associated with declining diversity in some biotic groups [9-12]). Riparian wetlands are also impacted by many variables, such as urbanization, contaminants from highway runoff, fertilizer inputs, and thermal changes. In many wetlands native biodiversity has decreased and invasive species have increased [13-17] with greater floods from urban and agricultural land-use changes. However, a detailed literature review and TSAC workshop series revealed that no suitable empirical data were available for evaluating wetland hydrology and biodiversity relations in the Red River basin. This investigation was designed to begin to address concerns of wildlife managers and conservation biologists and the uncertainty of the engineering community on these issues. Because the same data gaps are found globally in other large river basins, the results may provide an understanding that is widely useful.
　　The engineering and biological literatures use many hydrology terms such as hydroperiod, water level deviation, storm event fluxuation, and seasonal water level change. Because none described the water level changes the authors measured, the term “temporary water level increase” was used and defined as the measurable change in wetland water level amplitude(depth) over the duration of normal wetland inflows from discrete measured storm events.
　　The authors investigated the following questions: (1) How much floodwater storage is available in the Red River Basin wetlands? (2) What relationships exist between temporary water level increases and wetland outlet control? (3) What relationships exist between wetland hydrology, watershed characteristics, and wetland biodiversity and outlet characteristics? (4) In joint FDR and natural resource (NRE) designed wetland restorations, what are the key technical understandings for establishing and sustaining biodiversity? and (5) Can the unitized measurements from the study wetlands be used to address FDR benefits at the basin scale?
　　 2. Methods
　　The locations and characteristics of 28 wetlands, selected throughout Minnesota’s Red River Basin, are depicted in Fig. 1 and summarized in Table 1. Wetland size, watershed tributary area, watershed area to wetland area ratio, and watershed cover types were treated a priori as continuous variables. The water outlet control methods in the wetlands were evaluated as discrete variables. Wetland and watershed characteristics, including watershed areas, wetland size and storage volumes, and land cover within the watersheds, were obtained from U.S. Geological Survey 7.5’ topographic maps, aerial photographs, field surveys, and available GIS datasets. During field surveys, outlet type, size, and invert elevations were confirmed by observation, measurement, and survey. Water levels and rainfall were also measured at each wetland using staff and rain gages from early Spring 2000 to the late fall winter freeze-up in 2001. Daily rainfall data were obtained from the Minnesota state climatologist.
　　


　　 Fig. 1 The Red River basin and wetland study locations, Minnesota. Readers are referred to Table 1 for physical descriptions of each wetland.
　　The two years of growing season hydrologic and climatic data were used to compare wetland performances. During this period, some wetlands experienced extreme runoff events, while others did not. Computer simulations were therefore used to standardize the analyses for wetlands, using longer-term climate records. A water budget analysis was completed for 24 of the 28 wetlands using WATBUD, a program developed by the Minnesota Department of Natural Resources [18]. WATBUD is a physical model capable of calibrating water-balance parameters by comparing simulated and measured wetland water levels. A separate model for each wetland was developed using the wetland water-level data and corresponding precipitation and temperature data with land use, precipitation, and wetland water discharge measurements. Calibration is a formal process of using measured variables for specific storm events (e.g., measured rainfall, wetland-surface water discharges, and time). The authors used documented land use/cover types and their acreages and matched surface-water-discharge hydrographs for each wetland by adjusting the primary unknown variables (e.g., ground-water inflow or outflow) to calibrate the modeled hydrographs with the measured hydrographs for the same storm events. Once calibrated, models were then extrapolated to a 20-year (1982-2001) precipitation and temperature record from Fergus Falls, Detroit Lakes, and Red Lake Falls, Minnesota. Daily wetland water-level changes were simulated from daily inputs of precipitation and temperature; inputs of runoff, pan evaporation, or ground-water exchange; and estimates of runoff, evaporation, or ground-water exchange.
　　Table 1 Physical characteristics of each wetland.
　　


　　a Age in years since restoration began from agency records and dated aerial photographs.
　　b The watershed to wetland area ratio is the direct watershed area, also used in the water budget analysis.
　　Projected hydrograph performance and water-level changes (maximum temporary water level increases in height and duration) for the 20-year simulation period were summarized in 15-cm increments. Averaged monthly water levels were tabulated to predict available storage as: (1) cm of temporary water level increase; and (2) cm of runoff. Three categories of water storage, based on the outlet control engineering definitions, were used in this study: (1) permanent storage was available below the run out elevation(volume of runoff totally removed from the downstream hydrograph); (2) outlet controlled storage was temporary runoff storage between the run out elevation and an emergency spillway; and (3) spillway surcharge storage was temporary runoff storage above the spillway elevation. Storage data were tabulated separately for spring snowmelt runoff and early summer flooding resulting from heavy rainfall. Uncontrolled wetlands (natural) and four types of controlled outlets were studied, including ditch plugs(ditches that had been backfilled to match historic grades), single and 2-stage outlets (culverts with one or two outlet elevations or orifice sizes to manage normal water levels and high flood flows), and operable controls (having manual closures for flood management).
　　Richness, diversity, and relative abundance for breeding birds, vascular plants, and macroinvertebrates were measured to evaluate their response to temporary water-level increase dynamics. Biological data were summarized using H’ = -sum pi loge pi , the Shannon-Weiner diversity index [19] by computing the individual species fractional percentage of the total abundance of all species found in a site expressed as a natural log, then summed among all species present for birds, vascular vegetation, and macroinvertebrates separately, and then as a combined H’. This procedure was conducted to enable a discussion among policy, regulatory, and conservation professionals on acceptable wetland diversity and wetland quality expectations.
　　Breeding bird richness and relative abundance were surveyed at three random points in each wetland during June 2000 and 2001. Calling or observed birds were recorded until no additional species were found for a total of approximately two survey hours per wetland. Only breeding birds in wetland habitats were included, and field sampling was modified after the variable circular-plot method of estimating bird numbers [20] and Breeding Bird Atlas program criteria from the Illinois Department of Conservation[21]. Relative abundances were categorized as follows:(1) only 1 individual heard or observed; (2) 2-10 individuals present; (3) 11-20 individuals present; (4) 21-50 individuals present, and (5) > 50 individuals present.
　　Wetland vegetation was mapped, and plant species richness and relative abundance were sampled using Timed Meandered Searches (TMS) [22] that followed the topographic survey-cross sections from the highest elevation wetland vegetation zones through emergent and aquatic vegetation zones and concluded in each zone after no additional new species were identified. Plant species abundances were categorized as: (1) rare;(2) localized but widespread; (3) isolated dominant colonies or patches; and (4) abundant. Wet prairie, sedge meadow, emergent, and aquatic zones [23, 24] were found in each of the wetlands.
　　Macroinvertebrates were sampled at a minimum of three collection locations in June 2001 and 2002 in each wetland. All substrates and the water column were sampled using “D” frame sweep nets with a 0.6-mm mesh, and then field washed to remove large debris, with retained specimens bottled in 70 percent ethanol. In the laboratory, macroinvertebrates specimens were sorted to order or family and tallied following Eddy and Hodson’s taxonomic keys to the common animals of the north central states [25] and Pennak’s Fresh-Water Invertebrates of the United States Protozoa to Mollusca [26].
　　Distributions were evaluated for normality, homogeneity, and colinearity of variances [27]. As a conservative measure of robustness, statistical runs included all outliers. Normally distributed data were reviewed for linear relationships using bivariate and multi-linear regressions [28]. Step-wise, interactive, linear-regression analysis was used to test relationships among all variables with one dependent variable. For example, when using plant H’ diversity as the dependent variable, and a combination of temporary water level increase metrics, watershed-land cover type, and wetland and watershed/wetland ratios as independent variables, final models having the highest regression coefficient and greatest significance for each variable were derived [28]. In this analysis, visual examination of biological data was used to define wetland ecological quality that was then assessed using H’ to initiate a discussion among policy, regulatory, and conservation professionals on acceptable wetland diversity and wetland quality expectations where FDR use of wetlands is contemplated. Regression analyses and student “t” tests [27] were tested at P ≤ 0.05 and are reported as significant if P ≤ 0.05, or highly significant if P ≤ 0.01. All plots from regressions analyses are provided in TSAC 2004 [29] and only a selection of the analyses are presented here.
　　


　　Fig. 2 Mainstem River(s) Hydrograph and Wetland TRA1 Hydrograph Response to Spring 2001 Flood.
　　 3. Results
　　3.1 Potential Flood Storage Availability in Wetlands
　　The timing of the permanent and temporary wetland flood water storage compared to downstream main stem river flooding was assessed to understand each wetlands flood water storage effectiveness, and downstream flood reductions. The authors provide an example of a typical measured wetland response (Fig. 2) where the hydrographs for Twelve Mile Creek, Rabbit River, two Bois-de-Sioux River tributaries (Fig. 3), and the river connected wetland TRA1 indicate that this wetland stored and held the “early” spring runoff throughout the flood event, with a permanent storage approximating 1.7 cm of runoff from its watershed (Fig. 2). Monthly available runoff storage averaged over the 20-year simulated period, documents differences between wetlands (Figs. 3 and 4). Example typical wetland flood storage responses is illustrated using wetland GRA4 which was restored with a simple ditch plug and provided modest runoff storage except for a short period following spring runoff and wetland GRA3 which provided a similar amount of permanent storage as GRA4, but the more restrictive outlet of GRA3 also provided considerable temporary outlet controlled storage.
　　 3.2 What Are the Relationships between Temporary Water Level Increase and Outlet Control?
　　


　　 Fig. 3 Averaged monthly water storage from 20-year WATBUD projection, in centimeters of runoff from tributary watershed, for Wetland GRA3.
　　


　　 Fig. 4 Averaged monthly water storage from 20-year WATBUD projection in centimeters of runoff from tributary watershed, for wetland GRA4.
　　The amount of permanent storage provided by natural summer and fall drawdown was predicted for each wetland using the 20-year simulated water level record (Table 2) and compared levels as recorded in this table using Student “t” tests at 95% probability levels. Natural wetlands had significantly higher permanent storage for capturing early spring and summer runoff from their watersheds than 1- and 2-stage outlet controlled wetlands. Natural wetland stored 10-15 cm of total annual runoff from their watersheds while single stage approximated 2.5 cm, and 2-stage outlet controlled wetlands, approximated 7.5 cm of runoff control by storing the water temporarily or permanently. Temporary water level l increases, while highest in 2-stage controlled wetlands, were not significantly different from 1-stage and natural outlet controlled wetlands (Table 2).
　　Table 2 Available flood water storage based on 20 year WATBUD simulations and spring and summer storms.
　　


　　


　　Simulated water levels and average values for all 20-years that included non-flood years, flood years(1989, 1996, 1997, and 2001) and the record flood of 1997 are presented in Fig. 5 and Table 3. There was no difference in the average fall drawdown(approximately 30 cm) for wetlands with single-stage outlets when comparing all years. Approximately half of the permanent storage in a single-stage outlet controlled wetland was available prior to the record 1997 flood. Single-stage controlled wetlands experienced over 30 cm of natural drawdown below their inverts during the summer and fall months. Larger wetlands with a 2-stage outlet had significantly less natural drawdown than 1-stage and natural wetlands (Tables 2 and 3). Spring storm and fall drawdown rates varied from 0.3 cm/day to 1.5 cm/day(Table 3). Mean drawdown rates varied from 0.9 to 8.8 cm/day for all control outlet types. Naturally controlled wetland drawdown rates were significantly higher than 2-stage wetlands but not significantly different from wetlands with 1-stage outlet controls.
　　Using the physical characteristics of each wetland(Table 1) and the total range of recorded water levels, moderate linear relationships were found with watershed to wetland ratio (y = 0.0798x + 1.6509 r2 = 0.4, N = 28). The magnitude of the recorded temporary water level increases during the 2001 spring runoff varied among wetlands (Fig. 6). Summer temporary water level increase events (May through August) were summarized in 0.15, 0.3, 0.6, and 0.9 m magnitude classes (Table 4). Frequencies for all temporary water level increase categories were statistically inseparable. The study found the following relations between the number of temporary water level increase events and the watershed/wetland area rations for temporary water level increase events > 0.15 m, > 0.3 m, and > 0.6 m:
　　


　　Fig. 5 Available permanent storage on November 1st from the 20-year WATBUD simulation.
　　Table 3 Fall drawdown below wetland runout elevations (cm) using the 20-year WATBUD simulation.
　　


　　November 1st data for 1988, 1995, 1996 and 2000;
　　Indicates significant differences between 1-stage, 2-stage outlet controlled wetlands water level drawdown means; Fall drawdown is the available depth of storage available for winter snowmelt on November 1.
　　Temporary water level increase events > 0.15 meter = 0.4117 (watershed/wetland area ratio) + 3.5176;
　　Temporary water level increase events > 0.30 meter = 0.2453 (watershed/wetland area ratio)-0.2325;
　　Temporary water level increase events > 0.60 meter = 0.0867 (watershed/wetland area ratio)-0.3807.
　　Lower correlations between the watershed/wetland area ratio and the number of temporary water level increase events were found as temporary water level increase height increased from > 0.15 meters (r2 = 0.42) to > 0.6 meters (r2 = 0.287). Outlet type influence on temporary water level increase level and frequency, were not significant when comparing the single and two-stage outlets (Table 4).
　　3.3 Watershed Characteristics and Wetland Biodiversity
　　Land-use moderately correlated to measured biotic diversity. Watershed percent in agricultural production showed a moderate inverse relationship with avian H’(y = -0.0082x + 3.2319, r2 = 0.52, N = 26) and vegetation H’ diversity (y = -0.01x + 4.2675, r2 = 0.63, N = 26) for older wetland restorations. Vegetation H’ diversity correlated to watershed/wetland ratio for all age restorations (y = -0.0213x + 4.1555, r2 = 0.48, N = 26).
　　Water-level drawdown rates and biodiversity showed consistent moderate polynomial relationships with H’ diversity of macroinvertebrates, birds, and plants. For plants, mid-range water-level drawdown rates 3-6 cm/day) were associated with higher biotic diversity than either higher or lower rates (y =-76.202x2 + 23.028x + 2.5084, r2 = 0.44, N = 13). The diversity index means for birds, plants, and macroinvertebrates were significantly higher for drawdown rates between 1.5 and 6 cm/day than for drawdown rates above or below these values(Table 5).
　　


　　Water storage is quantified as cm of runoff from wetland’s tributary watershed
　　Fig. 6 Total water storage provided and temporary water level increase experienced by study wetlands during spring 2001.
　　


　　** Indicates significant differences between 1-stage, 2-stage outlet controlled wetlands water levels.
　　Consistent inverse linear regressions showed that frequent and higher amplitude temporary water level increase events were associated with wetlands having lower H’ diversity for bird and plant species. Using combined H’ diversity for all taxa, the strongest inverse relationship was with the number of temporary water level increase events > 0.9 m in height (y =-0.0507x + 3.732, r2 = 0.77, N = 27), and for the number of temporary water level increase events < 0.8 m in height (y = -0.0462x + 2.4042, r2 = 0.52, N = 24).
　　Non-native plant species H’ diversity and the number of non-native plant species were significantly related to maximum wetland temporary water level increase levels when comparing wetlands that experienced > 0.9 and < 0.9 m, and when comparing wetlands with ≤ 0.3 and > 0.3 m temporary water level increase events (Table 6). Wetlands with over 0.9 m of maximum temporary water level increase and wetlands with less than 0.3 m of maximum temporary water level increase both showed an increased percentage of non-native plant species in the wetlands.
　　Table 5 Wetland drawdown rate vs. diversity index mean.
　　


　　 Table 6 Temporary water level increase level vs. percentage of non-native plant species.
　　


　　3.4 FDR/NRE Wetland Restoration Criteria for Sustaining Biodiversity
　　Visual examination of the combined strata H’diversity suggested three wetland quality categories-low, intermediate, and high. The categories for bird, vascular plants and macroinvertebrates where H’~<= 2.6 was defined as low quality wetlands; high quality wetlands had an H’ of > 3.2; and intermediate quality fell between these values (Table 5). Mean watershed characteristics and temporary water level increase frequency statistics indicated no significant difference in the number of summer temporary water level increase events between the low and high quality wetlands. Higher quality wetlands were found to have a significantly smaller watershed/wetland ratio mean of 5:1, while low and intermediate quality wetlands had inseparable means and watershed/wetland ratios of ~15:1. Land cover in watersheds of the higher quality wetlands averaged ~20% row crops, which was significantly lower than the means of 30% in intermediate, and 60% in low quality sites. All means were significantly different from each other (Table 7). H’ for birds, plants, and macroinvertebrates in wetlands with temporary water level increases > 0.9 m were significantly lower than in wetlands bouncing < 0.9 meters. H’ for plant and macroinvertebrate communities in wetlands with temporary water level increases less than and greater than 0.3 m were also significantly different. Wetlands with greater (or less) than a 0.6 m amplitude show no differences in H’biodiversity levels.
　　Step-wise, interactive, linear-regression analysis was used to test relationships among all variables with one dependent variable. This analysis suggested that plant H’ diversity and watershed variables were very highly significantly related to watershed/wetland ratio, significantly to the wetland area (ha), and not significantly related to row-crop agricultural areas (ha)(P~0.1), hay and pasture area (ha) (P~0.16), and bog and marsh area in the wetlands’ watershed (P~0.1), but together, all these variables had a very highly significant relationship with plant H’ diversity (y =-0.027 ratio + 0.0004 wetland area + 0.0086 row-crop agricultural area + 0.007 hay and pasture area + 0.0102 bog and marsh area + 2.150, r2 = 0.736, F = 10.62, P~0.0001, N = 24). Avian H’ diversity showed weaker and not significant relationships (r2 = 0.21); probabilities varied from P~0.1 to P~0.8 for the wetland area (P~0.1) agricultural, or hay-pasture, forested land, brush lands, water, or bog and marsh lands in the watersheds, number of temporary water level increase events > 0.15 m, and total temporary water level increase amplitude. Macro invertebrate H’ diversity also showed not significant and moderate-to-weak correlation coefficients with these same watershed variables (r2 = 0.36, y = 0.0012 row crop agricultural area + 0.0115 bog-marsh area + 1.311, F = 2.3; P~0.12, N = 24 and was most strongly related to percent agriculture in each watershed(P~0.3), and percent in bog and marsh cover types(P~0.04).
　　Table 7 Summary statistics for wetland quality categories.
　　


　　** Indicates significant differences between 1-stage, 2-stage outlet controlled wetlands water levels and temporary water level increase events.
　　The maximum amplitude, duration, and number of temporary water level increase events in > 0.15 m, > 0.3 m, and > 0.6 m and > 0.9 m categories showed strong multi-linear relationships between temporary water level increase metrics in wetlands and land use in watersheds. Temporary water level increase amplitude showed strong and very highly significant relationships with the direct and total tributary areas, study site wetland area, tributary agricultural area, hay and pasture area, and brush land area and was significant for bog and marsh area (r2 = 0.604, N = 24). Temporary water level increase durations were also very significantly related to many of the same watershed physical variables as temporary water level increase amplitude (r2 = 0.756, N = 24).
　　Temporary water level increase events greater than 0.15 m were highly significantly correlated to direct and total tributary areas, wetland area, tributary agricultural area, hay and pasture area, brush land area, and significantly to bog or marsh area (r2 = 0.646). Temporary water level increase events greater than 0.9 m had an overall correlation coefficient of 0.706 and very significant relationships to these same variables. This study provided general guidance (Table 8) to consider in the design of wetland restorations and expectation for such restorations in Red River Basin future FDR and NRE projects. Watershed context, location, and physical characteristics were found more important than outlet design as most important variables in the wetland restoration design process.
　　Table 8 General design and siting criteria for wetland restorations in the Red River Basin, MN.
　　


　　*This study did not examine the relationship between temporary water level increase and nesting success of over-water nesting birds.
　　 4. Discussion
　　4.1 Potential Flood Storage Availability in Wetlands
　　The authors found consistent levels of available permanent storage in most wetlands resulting from measured fall drawdown rates that were higher than anecdotal documentation suggested in the Red River Basin. Spring snowmelt combined with early season rains, filled the study wetlands. Water levels remained until early summer and then declined, falling to seasonal lows in late fall. Wetland response to specific spring snowmelt runoff or heavy summer rainfall events is highly variable. Wetlands that provided available storage for spring runoff had permanent storage available again by mid-summer.
　　4.2 What Are the Relationships between Temporary Water Level Increase and Outlet Control?
　　Wetlands controlled with two-stage outlets provided minimal permanent storage, but high levels of temporary storage of “early hydrograph” runoff. They held some or all of this runoff throughout the flood event. The seasonal drawdown rate for these wetlands ranged from 1.8 to 10.4 cm per week. Although excessive fall precipitation may contribute to spring flooding, these results suggest that a consistent natural drawdown occurred during flood and non-flood years, which provided predictable available flood storage in wetlands. The record flooding of 1997 was no exception when, although wetland storage was reduced, it was still available.
　　Temporary water level increases were influenced by watershed characteristics, particularly the watershed to wetland area ratio and the watershed land-use characteristics, more than by outlet design. Smaller wetlands in the study were restored with a simple ditch plug. These smaller wetlands consistently provided permanent storage whether outlet or naturally controlled. The larger wetlands were restored with a two-stage outlet with a corresponding higher sustained base flow and limited permanent storage. Permanent storage in larger wetland restorations can be increased by outlet design and operation.
　　This study provided important evidence that both controlled and uncontrolled wetlands provide highly predictable flood storage, and that watershed runoff volumes and wetland storage capacities are greater than assumed in previous engineered FDR studies and projects [30]. These data also show a reliable availability of storage in wetlands to capture and hold spring runoff because of the draw down that occurs through fall and winter of the previous year and between storm events. Permanent storage was provided in single stage outlet controlled wetlands, and runoff captured below the invert was 100% removed from the downstream flood hydrograph. More storage was provided for spring runoff than for summer rain events. Using two-stage outlets provided limited permanent storage, more temporary storage, and more total storage than single-stage-outlet controlled wetlands.
　　4.3 Watershed Characteristics and Wetland Biodiversity
　　The number of summer temporary water level increase events that occurred in the study wetlands was highly variable. However, temporary water level increase events greater than 0.6 m were infrequent in all our study wetlands. In Pacific Northwest wetlands, urban land-use changes were found by Azous and Horner [31] to impact diversity at a frequency threshold of three temporary water level increase events per month during the growing season. Because our study wetlands did not experience such frequencies, even annually, the authos were unable to identify a frequency threshold that correlated with diversity. Temporary water level increase alone did not explain the measured variation in wetland biodiversity. Red River Basin wetlands show a wide range of biological diversity, which appears to be at least partially dependent on the extent of existing and historic watershed modifications. The highest diversity vegetation, bird, and macroinvertebrate communities were measured in protected watersheds and wetlands with minimum historic soil and vegetation disturbance. The data suggest that wetlands with either extreme wetland temporary water level increase values or wetlands with limited temporary water level increase but in the most altered watersheds are more likely to have higher non-native plant species present.
　　Joint NRE and FDR wetland restoration projects can use these understandings of trends in decreasing native plant diversity with increasing temporary water level increase height, duration, and frequency of temporary water level increase events. The measured increases in non-native plants with greater temporary water level increase height and frequency provides important design considerations for restorations. This study and the definition of wetland quality using combined H’ diversity provides insight to help clarify regional NRE and FDR issues of concern and conflict. Previously, discussions among policy, regulatory, and conservation professionals on the acceptable wetland diversity and wetland quality expectations where FDR-use of wetlands was contemplated had no basis for understanding the biodiversity likely to result. These analyses suggest that some clear relations exist between a wetlands watershed physical setting, hydrology, and resulting biodiversity (Tables 7 and 8), and these criteria may be useful for evaluating each FDR-wetland project proposed in the Red River Basin.
　　High-quality wetlands consistently supported a diverse fringing sedge meadow that graded with wet prairie, and several emergent vegetation zones in which cattail (T. angustifolia, T. latifolia) and canary grass (Phalaris arundinacea) were absent or minor elements. Most wetlands were classified as intermediate to low quality, dominated by altered plant communities, commonly with cattails and fringe of introduced naturalized canary grass with occasional stands of box elder (Acer negundo) and sand bar willow (Salix interior). Consistent with reports by others [17, 32, 33], the authors found algal blooms and less diverse aquatic macrophyte communities dominated the open water areas.
　　Species and communities are sensitive to ecological stressors [34]; vegetation has been a useful indicator of altered hydrologic regimes [17], and hydrologic stressors such as salinity, sedimentation/turbidity, and excessive nutrient loads/anoxia have resulted in vegetation cover changes [35-38]. In altered wetlands, early invasive species often dominate and tend to form large homogenous stands, resulting in lower species richness [39-43]. Temporary water level increase and watershed/wetland ratio appear to be useful indicators of ecological stress.
　　In Red River Basin, at low levels of watershed conversion to cultivated agricultural land-uses, deeper wetland basins shifted to prevailing monocultures of cattail where diverse emergent vegetation previously grew. Shallow wetlands shifted to reed canary grass monocultures as native plant biodiversity decreased, consistent with other studies [13-15]. Ehrenfeld and Schneider [16] reported declining native plants and increased invasion by invasive species as land use converted to more intense agriculture and urban uses around Atlantic white-cedar (Chamaecyparis thyoides) swamps. Similarly, with greater flood disturbances, plant species richness declined [17].
　　Flood depths, frequency, and durations in Seattle palustrine wetlands strongly related to measured biodiversity [9, 11, 31], as plant species richness was highest with less than three flooding events per month in all water depths. The greatest richness was also found in wetlands with cumulative flood durations above a 0.15 m average depth that was less than six days per month for wetland plant communities in 0.6 m water depths. Azous and Richter [12] found that emergent wetland vegetation zones subjected to greater than 20 cm in fluctuation had significantly lower species richness than emergent zones with less than 20 cm. Washburn [44] documented richer native plant assemblages and wildlife in wetlands with slow, late season drawdowns.
　　Breeding bird species diversity declined with the conversion of shallow basins to reed canary grass and in deeper cattail-dominated basins. Like Washburn[44], the authors also found heaviest-use by waterfowl in wetlands with co-mingled cattail growths and interspersed open water. The causes of bird species diversity and richness decline in urban Seattle area palustrine wetlands [31, 45] were attributable to water-level fluctuations, increasing habitat isolation, and water quality decline, among other factors [46]. Temporary water level increase is strongly suggested to be a key variable in Red River Basin wetlands but alone did not explain all measured variances in bird and plant diversity.
　　Unlike Azous and Richter [12] and Hicks [10], where highest H’ diversity for macroinvertebrates occurred in wetlands with lowest urbanization temporary water level increase amplitudes and frequencies, the authors found in agricultural watersheds that wetland macroinvertebrate H’ diversity weakly increased with higher temporary water level increase frequency and magnitude and durations. Our study results did not corroborate Murkin and Kadlec [47], who documented macroinvertebrate diversity declines after prolonged flooding.
　　4.4 FDR/NRE Wetland Restoration Criteria for Sustaining Biodiversity
　　Most Red River Basin wetland restorations have not considered the effects of an altered watershed or hydrologic context. Inadequate consideration is also commonplace in many wetlands restoration in other geographic settings. As a result, highest quality NRE outcomes and benefits are rare in the Red River basin and elsewhere. Use of wetlands for FDR has been debated for years, and in the absence of data for use in standardized hydraulic models, the regional assumption by the flood management community has been that restored wetlands cannot provide reliable FDR benefits. This study supports a strong contrary conclusion and also provides design guidance to help designers and regulators of restored wetlands provide significant regional and watershed-scale FDR and NRE outcomes. In future wetland restorations, matching project goals to watershed and site characteristics can help guide the design and better predictions of wetland restoration project NRE and FDR benefits and outcomes (Table 8).
　　 5. Conclusions
　　These results suggest that a consistent natural drawdown occurred during flood and non-flood years, which provided predictable available flood storage in wetlands. Smaller wetlands consistently provided permanent storage whether outlet or naturally controlled. Permanent storage in larger wetland restorations can be increased by outlet design and operation. This study provided important evidence that both controlled and uncontrolled wetlands can provide highly predictable flood storage, and that watershed runoff volumes and wetland storage capacities that are greater than assumed in engineering studies in the red river watershed, including reliable availability of storage in wetlands to capture and hold spring runoff because of the draw down that occurs through fall and winter of the previous year and between storm events. Permanent storage was provided in single stage outlet controlled wetlands, and runoff captured below the invert was 100% removed. Wetland restorations can be useful to provide permanent and temporary flood storage and can alter the timing of tributary river hydrographs to reduce downstream main stem river flooding.
　　Plant and bird species diversity and use of wetlands is sensitive to ecological altered hydrologic regimes. In the most hydraulically dynamics wetlands, invasive species and lower biodiversity for the organism groups studied, and a prevalence toward dominance by large homogenous monoculture stands, of such plants as reed canary grass or cattails often prevail. In future wetland restorations, matching project goals to watershed and site characteristics can help guide the design to ensure NRE and FDR benefits and outcomes, and their compatibility.
　　 Acknowledgments
　　The authors appreciate the support by the Working Group members and Minnesota Department of Natural Resources and Applied Ecological Services, Inc staff. Corey Hanson, Doug Mensing, Susan M. Lehnhardt, John Larson, Jason Carlson, and Bill Stoll provided field assistance while Heidy Sowatzke typed numerous drafts. TSAC members Charles Anderson, Larry Lewis, Al Kean, Doug Wells, Henry VanOffelen, Scott Jutila, Brent Johnson, and Tom Groshen contributed to the design and internal peer review of this research. Commentary on an early draft by Joy Zedlar and several anonymous reviewers, and Klaus Richter’s assistance in securing literature were appreciated.
　　 References
　　[1] J.W. Labaugh, T.C. Winter, V.A. Adomaitis, G.A. Swanson, Hydrology and chemistry of selected prairie wetlands in the Cottonwood Lake area, Stutsman County, North Dakota 1979-82, U.S. Geological Survey Professional Paper 1431, Jamestown, ND, USA, 1987.
　　[2] J.W. Labaugh, G.A. Swanson, Algae and Invertebrates in the Water Column of Selected Prairie Wetlands in the Cottonwood Lake Area, Stutsman County, North Dakota, 1984, U.S. Geological Survey, Washington, DC, USA, Open file report 1993, pp. 88-451,
　　[3] K.A. Chapman, M. Kinsella Ziegenhagen, Valley of Grass: Tall Grass Prairie and Parkland of the Red River Region, North Star Press, Inc., St Cloud, MN, USA and the Nature Conservancy, MN, USA, 1998.
　　[4] G.J. Wiche, T. Williams-Sether, Stream Flow Characteristics of Streams in the Upper Red River of the North Basin, North Dakota, Minnesota, and South Dakota, U.S. Geological Survey, Bismarck, ND, USA, Open-file report, 1997, pp. 97-416.
　　[5] McCombs-Knutson Associates, Inc., Water Resources Engineering/Planning Program for the Red River of the North Basin in Minnesota, Prepared for The Lower Red River Watershed Management Board, 1984.
　　[6] U.S. Army Corps of Engineers, Technical Resource Service, Red River of the North, Volume I, Timing Analysis, St. Paul District, Corps of Engineers, St. Paul, MN, USA, 1988.
　　[7] U.S. Army Corps of Engineers, and Federal Emergency Management Agency, Region V and Region VIII, Regional Red River flood assessment report, Wahpeton, North Dakota/Breckenridge, Minnesota to Emerson, Manitoba, 2003.
　　[8] R.W. Maclay, T.C. Winter, L.E. Bidwell, Water Resources of the Red River of the North Drainage Basin in Minnesota, Water-Resources Investigations 1-72, U.S. Geological Survey, 1972.
　　[9] S.S. Cooke, A. Azous, Vegetation species responses to changing hydrology in urban wetlands, in: Proc. of the Third Puget Sound Res. Meet., Puget Sound Water Quality Authority, Olympia, WA, USA, 1995.
　　[10] A.L. Hicks, Aquatic Invertebrates and Wetlands: Ecology, Biomonitoring and Assessment of Impact from Urbanization, Environmental Institute, University of Massachusetts Press, Amherst, MA, USA, 1996.
　　[11] R. Thom, A. Borde, Influence of urbanization on ecological processes in wetlands, Battelle Marine Sciences Laboratory, Sequim, WA, USA, 2001.
　　[12] A.L. Azous, K.O. Richter, Amphibian and plant community responses to changing hydrology in urban wetlands, in: E. Robichaud (Ed.), Puget Sound Research 1995 Proceedings, Puget Sound Water Quality Authority, Olympia, WA, USA, 1995, pp. 156-162.
　　[13] S. Apfelbaum, Cattail (Typha spp.) management, Natural Areas Journal 53 (1985) 9-17.
　　[14] D. Wilcox, S.I. Apfelbaum, R.D. Hiebert, Cattail invasion of sedge meadows following hydrologic disturbance in the cowles bog wetland complex, Indiana Dunes Natural Lakeshore Wetlands 4 (1985) 115-128.
　　[15] S. Apfelbaum, C. Sams, Ecology and management of reed canary grass (Phalaris arundinacea L.), Natural Areas Journal 72 (1987) 69-74.
　　[16] J.G. Ehrenfeld, J.P. Schneider, Wetlands and Suburbanization: Effects on hydrology, water quality and plant community composition, Journal of Applied Ecology 282 (1991) 467-490.
　　[17] A.G. van der Valk, L. Squires, C.H. Welling, Assessing the impacts of an increase in water level on wetland vegetation, Ecological Applications 43 (1994) 525-534.
　　[18] Minnesota Department of Natural Resources, WATBUD Model: Water Balance Model for Lake Level Simulation[Online], 2008, http://www.dnr.state.mn.us/waters/surfacewater section/lake hydro/watbud.html.
　　[19] C.E. Shannon, A mathematical theory of communication, Bell System Technical Journal 27 (1948) 379-423, 623-656.
　　[20] R.T. Reynolds, J.M. Scott, R.A. Nussbaum, A variable circular-plot method of estimating bird numbers, Condor 82 (1980) 309-313.
　　[21] Illinois Department of Conservation, Illinois Breeding Project: Handbook for Surveyors 1986-1990, Springfield, IL, USA, 1987.
　　[22] F.G. Goff, G.A. Dawson, J.J. Rochow, Site examination for threatened and endangered plant species, Environmental Management 64 (1982) 307-316.
　　[23] L.M. Cowardin, V. Carter, F.C. Goulet, E.T. LaRoe, Classification of Wetlands and Deepwater Habitat of the United States, U.S. Fish and Wildlife Service, Washington DC, USA, 1979.
　　[24] G.B. Ownbey, T. Morley, Vascular Plants of Minnesota: A Checklist and Atlas, University of Minnesota, Minneapolis, MN, USA, 1991.
　　[25] S. Eddy, A.C. Hodson, Taxonomic Keys to the Common Animals of the North Central States, Burgess Publishing Company, Minneapolis, MN, USA, 1961.
　　[26] R.M. Pennak, Fresh-Water Invertebrates of the United States Protozoa to Mollusca, 3rd ed., John Wiley & Sons, Inc., New York, Chichester, Brisbane, Toronto, Singapore, 1989.
　　[27] H.J. Zar, Biostatistical Analysis, 2nd ed., Prentice-Hall, Inc., NJ, USA, 1984.
　　[28] SPSS, Inc., SYSTAT, Version 10. 233 South Wacker Drive 11th Floor, Chicago, Illinois 60606-6307, 2002.
　　[29] Technical and Scientific Advisory Committee (TSAC), Wetland Hydrology and the Effect of Water Level“Bounce” (Temporary Water Level Increase) on Natural Resource Wetland Diversity, Technical Report No. 13, Red River Flood Damage Working Group, Minnesota Department of Natural Resources, St Paul, MN, Red River Basin Institute web site, Moorhead, MN, USA, 2004.
　　[30] Technical Science Advisory Committee (TSAC), An Overview of the Impacts of Water Level Dynamics“Bounce” (Temporary Water Level Increase) on Wetlands, Technical Report No. 1, Red River Flood Damage Working Group, Minnesota Department of Natural Resources, St Paul, MN, Red River Basin Institute web site, Moorhead, MN, USA, 1998.
　　[31] A.L. Azous, R.R. Horner, Wetlands and Urbanization, Implications for the Future, CRC Press, Inc., Lewis Publishers, Boca Raton, FL, USA, 2001, p. 338.
　　[32] D.K. Voigts, Aquatic invertebrate abundance in relation to changing marsh vegetation, American Midland Naturalist 95 (1976) 3136.
　　[33] M.W. Weller, The influence of hydrologic maxima and minima on wildlife habitat and production values of wetlands, in proc. Natl. Wetlands Symp. Wetland Hydrology, in: J.A. Kusler, G. Brooks, (Eds.), Association of State Wetland Managers, Berne, NY, USA, 1988.
　　[34] J.R. Karr, Biological integrity: A long neglected aspect of water resource management, Ecological Applications 1(1991) 66-84.
　　[35] P.R. Adamus, K. Brandt, Impacts on Quality of Inland Wetlands of the United States: A Survey of Indicators, Techniques, and Application of Community Level Biomonitoring Data, U.S. Environmental Protection Agency, Cincinnati, OH, USA, 1990.
　　[36] D.B. Booth, L. Reinelt, Consequences of Urbanization on Aquatic System-Measured Effects, Degradation Thresholds, and Corrective Strategies, King County Surface Water Management Division, Seattle, WA, USA, 1993.
　　[37] D.B. Booth, C.R. Jackson, Effects of Human-Induced Changes on Hydrologic Systems: Urbanization of Aquatic Systems-Degradation Thresholds and the Limits of Mitigation, American Water Resources Association, Seattle, WA, USA, June 1994.
　　[38] P.R. Adamus, Bioindicators for Assessing Ecological Integrity of Prairie Wetlands, U. S. Environmental Protection Agency, National Health and Environmental Effects Research Laboratory, Western Ecology Division, Corvallis, OR, USA, 1996.
　　[39] C. Fritz, A Summary of Wetland Classification Systems and A Strategy for the Development of Water Quality Standards for Wetlands in North Dakota, North Dakota State Department of Health and Consolidated Laboratories, Division of Water Quality Surface Water Quality Program, Bismarck, ND, USA, 1994.
　　[40] C. Fritz, Suggested metric for an index of wetland biological integrity for seasonal and temporary wetlands in North Dakota, MS Thesis, North Dakota State University, Fargo, ND, USA, 1997.
　　[41] S.M. Galatowitsch, A functional assessment of restored wetlands and implication for restoration programs. Abstract in: Prairie Ecosystems: Wetland Ecology, Management, and Restoration, Northern Prairie Science Center, Jamestown, ND, USA, 1993.
　　[42] S.M. Galatowitsch, T.V. McAdams, Distribution and Requirements of Plants on the Upper Mississippi River, Iowa Cooperative Fish and Wildlife Research Unit, Ames, IA, USA, 1994.
　　[43] S. Galatowitsch, J. Tester, D. Whited, S. Moe, A Report on the Development of Indices of Biotic Integrity for Minnesota Wetlands, University of Minnesota, MN, USA, Undated.
　　[44] C.F. Washburn, Timing of managed seasonal drawdowns to promote the establishment of wetland plant species on the Tule Lake National Wildlife Refuge, California, Ph.D. Dissertation, University of Washington, WA, USA, 2001.
　　[45] R.B. Blair, Land use and avian species diversity along an urban gradient, Ecological Applications 6 (1996) 506-519.
　　[46] B.L. Taylor, The influence of wetland and watershed morphological characteristics on wetland hydrology and relationships to wetland vegetation communities, M.S. C. E. thesis, Department of Civil Engineering, University of Washington, Seattle, WA, USA, 1993.
　　[47] H.R. Murkin, J.A. Kadlec, Responses by benthic macroinvertebrates to prolonged flooding of marsh habitat, Canadian Journal of Zoology 64 (1986) 65.