Consolidation Drainage and Climate Change May Reduce Piping Plover Habitat in the Great Plains

Many waterbird species utilize a diversity of aquatic habitats worldwide, but there is global concern about whether increasing anthropogenic needs to manage and manipulate water regimes will degrade these habitats and negatively affect waterbird populations (Ma et al. 2009; Bellio and Kingsford 2013). Wetlands, in particular, provide a range of important resources for waterbirds, but have been widely degraded across the world (Ma et al. 2009). Over half of the wetlands in the United States have been drained or converted to other land uses, and, not surprisingly, about half of the U.S. bird species listed as threatened or endangered, pursuant to the U.S. Endangered Species Act (U.S. Endangered Species Act [ESA 1973, as amended]), are known to occupy such habitats (Dahl 1990; Erwin et al. 2000). More subtle wetland alterations, including hydrological modifications, may also detrimentally affect wetlands and the birds they support (Anteau 2012; Anteau et al. 2012a; Roche et al. 2012; Bellio and Kingsford 2013). Thus, maintaining the ecological function of remaining wetlands may be as important a conservation strategy for shorebirds as restoration of previously lost wetlands (Finlayson et al. 2011; Bellio and Kingsford 2013).

The piping plover Charadrius melodus is one such shorebird that utilizes a range of shoreline habitats, including wetlands. Piping plovers (hereafter plovers) breed on shoreline habitat in three regions across North America including the northern Great Plains, the Great Lakes, and the Atlantic coast (Roche et al. 2010). The northern Great Plains population of plovers breeds in three principal habitats: riverine sandbars; reservoir shorelines; and shorelines of prairie wetlands, although reservoirs were not historically available (Prindiville Gaines et al. 1988; Anteau et al. 2012a, 2012b; Shaffer et al. 2013). Plovers nest on unvegetated shorelines, which are created and maintained by fluctuations in water levels (Burger 1987; Prindiville Gaines et al. 1988; Elliott-Smith and Haig 2004; Sherfy et al. 2009; Anteau et al. 2012a, 2014). Plovers in the northern Great Plains population, and others throughout North America, have declined over the last century (Haig and Oring 1985). Although the reasons for the population declines are not fully understood, it has been suggested that habitat loss, nest depredation, human disturbance, or management for high water levels at reservoirs and rivers may be to blame (Haig and Plissner 1993; Plissner and Haig 2000; Root and Ryan 2004; Catlin 2009; Anteau et al. 2012a). Less attention has been given to the loss of prairie wetland habitat, which may play an important role in maintaining plover populations in the northern Great Plains. We suspect that agricultural drainage and consolidation of wetlands used by plovers for breeding in the Prairie Pothole Region (PPR) have potentially reduced the amount of available breeding habitat (Anteau 2012).

The PPR experiences frequent interannual fluctuations in precipitation (hereafter wet–dry periods) that can vary in magnitude across the region (Niemuth et al. 2008, 2010; Forcey et al. 2011), and influence water levels of prairie wetlands (Euliss et al. 2004; Johnson et al. 2004; McCauley et al. 2015). Accordingly, availability of plover breeding habitat in this region historically varied across the landscape, with different locations experiencing decreases in water levels (hereafter drawdown) at different times (Anteau 2012; Post van der Burg et al. 2015). These spatiotemporal fluctuations could have potentially helped to provide plovers with available habitat, especially when conditions were unsuitable elsewhere in the region. However, ≥ 60% of wetlands within the PPR have been drained for agriculture since the mid-19th century (Dahl 2014). Although a great deal of the agriculturally related wetland drainage in the United States has been through drain tile (subsurface soil drainage), wetland drainage in the PPR, particularly in North Dakota, was primarily facilitated by draining smaller seasonal and temporary wetlands into larger wetlands and shallow lakes via ditches, a process termed consolidation drainage. Those drained wetlands were then presumably used for agricultural production. Consolidation drainage increases water levels in larger wetlands and shallow lakes (McCauley et al. 2015), which makes those wetlands progressively fuller with each wet–dry cycle (Wiltermuth 2014). This could have the effect of reducing the frequency with which those larger wetlands draw down enough to expose sufficient shoreline for plover breeding (Anteau 2012).

On the basis of the best available records, plovers historically nested at wetlands throughout the western PPR, and it is possible that before detailed records were kept they nested throughout the entire PPR (Prindiville Gaines et al. 1988; Anteau 2012). Presently it appears that plovers are restricted to larger, more alkaline wetlands of the Missouri Coteau where wetland drainage and other landscape modifications are less pronounced than they are in the Glaciated Plains (Figure 1). We know that consolidation drainage is making remaining wetlands in the PPR larger (McCauley et al. 2015). It is important to understand if consolidation drainage is occurring in similar amounts around wetlands used by plovers as it is in other wetlands in the region and if so, if consolidation drainage and fuller wetlands are reducing plover presence at those wetlands. This is important to understand because the loss of prairie wetlands as plover habitat could potentially alter the breeding range, population structure, dispersal, or refuge habitats and ultimately influence the viability of the northern Great Plains population of plovers. To begin to understand whether consolidation drainage could potentially reduce plover use of prairie wetlands, we assessed available records documenting the presence of plovers on these wetlands through time. Our objective was to evaluate the hypothesis that there will be a decline in the presence of plovers in wetlands with more consolidation drainage in their catchment and in wetlands that are fuller.

Our study area was the PPR of North Dakota (Figure 1, inset), which overlapped the known distribution of the northern Great Plains population of plovers. We concentrated on prairie wetlands in the Missouri River Coteau–Coteau Slope and the Glaciated Plains physiographic regions of North Dakota because many wetlands in this area were known to have plovers historically present (Figure 1; Haig and Oring 1985). Plover surveys have been conducted in this region from 1979 to present by the U.S. Fish and Wildlife Service and the North Dakota Game and Fish Department. There were ∼200 wetlands surveyed throughout the survey years; however, most wetlands were not surveyed every year. We limited our analysis to the wetlands that were consistently surveyed (1979–2011) and historically used (used in > 50% of years before 2000), a total of 32 wetlands (Figure 1). We assumed that most of the drainage in this region occurred from 1970 to 2000 via consolidation drainage, as subsurface tile drainage in this area was uncommon at that time (Anteau 2012; McCauley et al. 2015). Thus, we chose wetlands with plovers present before 2000 (while drainage was potentially still ongoing) and assumed they were historically used. Approximately half of the chosen wetlands remained frequently used by plovers after 2000, whereas the other half was seemingly no longer used (either not used at all or only during extreme drought). Table S1 in Supplemental Material lists the 32 sample wetlands and years of survey data that were available for each wetland, a total of 178 site-year combinations. Those 32 wetlands represent the full set of wetlands in this region with sufficient data available to evaluate our hypotheses and we assumed that they are representative of the other wetlands used by plovers in the region without sufficient data available.

We indexed plover presence using surveys for adult plovers conducted once on each wetland every June in North Dakota from 1979 to 2011. Data included surveys conducted annually and surveys conducted every 5 years (beginning in 1991) for the International Piping Plover Census (Murphy et al. 1999; Elliott-Smith et al. 2009). Methodologies between surveys were similar and we chose only wetlands that were surveyed in both the annual and international surveys to maximize the number of years of available data. We identified the location of these wetlands with a wetland map that was based on the National Wetlands Inventory (U.S. Fish and Wildlife Service 2003a). Surveys for 2002 and 2008 were unavailable.

Wetland catchments were generated for each sample wetland using high-resolution elevation data (light detection and ranging- and interferometric synthetic aperture radar-derived digital elevation models; 3-m and 5-m cell sizes respectively) and methods used in McCauley and Anteau (2014). We used digital elevation data to identify the direction of water flow across the landscape and delineate wetland catchments. We defined a wetland catchment as the portion of the landscape in which water flows into a subject wetland and is essentially a watershed-derived wetland complex that includes other smaller wetlands and their catchments if they would fill and spill into our focal wetlands during a 10-day, 100-year flood event.

To index consolidation drainage, we digitized all wetlands within the catchment of each focal wetland where drainage was visible. To determine whether a wetland was partially or completely drained, we identified ditches using three sources: high-resolution elevation models, ArcHydro-generated water-flow accumulation lines, and aerial photographs from the National Agricultural Imagery Program for the years 2003–2010. ArcHydro is a GIS-based tool that delineates water flow across the landscape using digital elevation data and ArcHydro-generated water-flow accumulation lines display the precise location and direction of water flow on the landscape. Historical aerial photographs from the years 1937–1969 were also collected for each sample wetland from U.S. Geological Survey Earth Explorer, U.S. Fish and Wildlife Service offices, and U.S. Department of Agriculture Natural Resources Conservation Service Farm Service Agency county offices to identify predrainage wetlands. We digitized all wetlands that were either visible on historical photographs taken during wet years, visible on current aerial photographs or in National Wetlands Inventory data, or a depression (identified with digital elevation data) with hydric soils indicative of a wetland (McCauley and Jenkins 2005). We classified the digitized wetlands as drained if ditches were present and flow-accumulation lines showed the direction of water flow going out of the wetland. Thus, drained wetlands could include either wetlands that were completely drained or those that had a ditch that had potential to drain all or part of it. We digitized all drained wetlands within each catchment of our focal wetlands and calculated the proportion of catchment area that was drained wetland for each available year (hereafter drained wetlands index; McCauley et al. 2015).

We calculated how full each wetland was by dividing the water surface area in each year by the maximum wetland size. We calculated maximum wetland size by using digital elevation models to find the spill point and spill elevation of each wetland. Wetlands are located in a depression in elevation and if a wetland is filled up with enough water, it will eventually reach the point where the water begins to spill out of the depression. The point at which the water begins to spill out of the depression's perimeter is called the spill point and the elevation at which this occurs is called the spill elevation. If a contour line is created at the spill elevation of a wetland, it represents the maximum size for that wetland.

To calculate water surface area in each year, we manually digitized surface water areas for our sample wetlands for the years 2001–2011 following McCauley et al. (2015). All digitization was done at the scale of 1:2,500. Water surface area was directly digitized from aerial photographs when the water boundary was readily visible or could easily be estimated on the photograph. When the water boundary was hidden by emergent vegetation, we calculated the water surface area as: area of visible water + (area of emergent vegetation/2). This method interpolated the water boundary as the halfway point between the visible water boundary and the outer edge of the emergent vegetation. The years 2001 and 2011 were complete plover survey years, but there was no aerial photography available for those years to allow us to digitize water surface areas. Thus, for 2011, we substituted the near-infrared band from SPOT 5 imagery because, although the resolution was coarser (10-m pixel size vs. ∼1-m pixel size of photographs), the near-infrared band made water boundaries readily visible. We used the same methods as with aerial photographs and manually digitized water surface areas for that year. For 2001, we substituted cloud-free Landsat5 thematic mapper imagery (U.S. Geological Survey Earth Explorer). We were concerned that, with the yet coarser resolution of Landsat imagery (30-m pixel size), manual interpretation would be more difficult. Thus, we performed an unsupervised classification with 10 classes in ArcGIS v10.1 to identify water boundaries. Water boundaries were clearly defined by one class and we exported that class as a vector data set to calculate surface water areas for 2001. We compared water surface areas that were derived through remote sensing with those we independently digitized from aerial photography on all 32 selected wetlands during 2003, 2006, and 2010 using a linear regression with no intercept. Water surface areas digitized from aerial photography and remotely sensed from Landsat were essentially the same (slope = 1.00; r² > 0.99).

We controlled for climate variation among years and wetlands with an index that was developed specifically for hydrologic effects of drought on permanent and semipermanent wetlands of the PPR (McCauley et al. 2015; Post van der Burg et al. 2015). This index (hereafter drought index) was developed using the standard precipitation–evapotranspiration index (Beguería and Vicente-Serrano 2012). We calculated the index with monthly precipitation and temperature from the parameter-elevation regressions on independent slopes model (PRISM Climate Group–Oregon State University 2014). We used a weighted average of the previous 10 years of monthly climate data to best approximate the influence of drought on wetland water surface areas (McCauley et al. 2015; Post van der Burg et al. 2015). Drought index values for our focal wetlands ranged from −2.2 to 2.4, with negative values indicating drier conditions and positive values indicating wetter conditions.

We used a mixed-effects logistic regression model (Bates et al. 2013; R Development Core Team 2015) to predict plover presence at sample wetlands as a function of the drained wetlands index, percent fullness of the wetland, and drought index. We assumed that plovers had a greater probability of presence on wetlands during drier periods (because more shoreline habitat would be available) and thus, included the drought index to control for the variation in plover presence among wetlands and between drier and wetter periods. We included the identity of the sample wetland as a random effect because each wetland had multiple years of plover presence–absence data (Table S1). We developed models for all possible combinations of the a priori selected variables and used Akaike's information criterion for small sample sizes (AIC_c) to rank the models. Goodness of fit of the final model was computed using marginal (R²_GLMM[m]) and conditional (R² _GLMM[c]) coefficients of determination (Nakagawa and Schielzeth 2013). The R²_GLMM(m) shows the proportion of the variance explained by the fixed effects, whereas the R²_GLMM(c) shows the proportion of variance explained by the fixed and random effects. Correlation between fixed effects was low (maximum correlation coefficient = 0.21). All data used in analyses are available in the Supplemental Material.

Consolidation drainage (percentage of the catchment area being drained wetland) in the catchments of our sample wetlands ranged from 0 to 11.5%, with an average 1.9% ± 0.02% (SD) and a median drainage amount of 1.3% (25th percentile: 0.7%; 75th percentile: 2.6%). The top model accounted for about 60% of the model weight and was chosen as the final model because the next best model (23% weight) had the same parameters plus the drought index variable, which appeared uninformative (Table 1; Arnold 2010). The final model was comprised of our drained wetlands index (β̂:3.20; SE: 1.56; 85% CI: −5.85 to −1.15). and percent fullness of the wetland (β̂:17.28; SE: 4.64; 85% CI: −24.53 to −11.07). The final model had a marginal R² value of 0.33 and a conditional R² value of 0.87. We found that probability of plover presence declined as the proportion of drainage in the catchment of the sample wetland increased (Figure 2). Probability of plover presence declined appreciably by the time 2% of the catchment had been drained. The probability of plover presence was 99.6% greater in wetlands with no drainage when compared with wetlands with 10% of the catchment drained. In the half of our sample wetlands that remained used by plovers, the proportion of drainage in the catchment was 1.4%, but in wetlands that were no longer used by plovers, the proportion of drainage in the catchment increased to 2.7%.

Our results also suggested that the probability of plover presence declined as the wetland became fuller (Figure 3). Plover presence declined when a wetland was about 80–90% full and the average fullness of our sample wetlands over all the years was 89.5% ± 11% (SD). In drier years (drought index < 0.5) the average fullness of our sample wetlands was 79.6% ±22% (SD) and in wetter years the average fullness or our sample wetlands was 92.3% ± 0.09% (SD). The probability of plover presence was 90.1% higher in wetlands that were half full when compared with wetlands that are completely full. On average, wetlands used by plovers were 88% full, whereas wetlands no longer used by plovers were 92% full.

Our study provides an example of the potential biological impacts of wetland consolidation drainage on wildlife. More specifically, our work shows how consolidation drainage and fuller wetlands have potentially reduced habitat quality of prairie wetlands used by plovers. Previous work on a larger set of wetlands in the region showed that wetlands in highly drained catchments had 197% larger water surface areas than those in catchments with no drainage (McCauley et al. 2015) and that these wetlands get progressively larger with each wetting cycle (Wiltermuth 2014). However, on the basis of McCauley et al. (2015), it was unclear if consolidation drainage was occurring on wetlands used by plovers and whether this would affect plover presence on these wetlands. Our results showed that consolidation drainage was occurring in the catchments of our sample wetlands similarly to other wetlands in the region (McCauley et al. 2015 drainage range 0–13%; this study drainage range 0–11.5%). We found that plovers historically used prairie wetlands as habitat during the breeding season but that plover presence was negatively associated with fuller wetlands and wetland drainage in the catchment. The half of our sample wetlands that was previously used by plovers but is no longer used is fuller and surrounded by more drainage than our sample wetlands that remain used by plovers. Prairie wetlands are less studied as plover habitat when compared with river and reservoir shorelines but survey data suggest that the remaining alkali lakes, mostly in the Missouri Coteau, are habitat for more than half of the plovers in this region (Haig and Plissner 1993; Plissner and Haig 2000). Monitoring data indicate that plovers bred in wetlands throughout the Glaciated Plains in North Dakota. Our results support the possibility that prairie wetlands may have played a role in plover ecology in the region but consolidation drainage may have caused a range change by eliminating some wetlands as habitat and reducing habitat quality at other wetlands. Our finding that fullness of a wetland is a major driver of plover presence also ties the fate of plover habitat to potential changes in climate, even in areas where drainage may be absent.

Other studies have found that consolidation drainage has increased water levels and had a progressive effect upon the hydrology of wetlands that receive surface water from drained wetlands (Wiltermuth 2014; McCauley et al. 2015). Wiltermuth (2014) found that wetlands with about 2% of their catchment drained become larger and will likely continue to increase in size with every wetting period, even without further drainage, until they overspill or stabilize. This was the same level of drainage we observed associated with the greatest rate of decrease in probability of plover presence. In drier years, wetlands could be less full and drainage could have less of an effect on plover habitat in the region. However, future climate change projections in the region predict conditions becoming warmer and wetter, with increases in winter and spring precipitation, the main source of precipitation for prairie wetlands (Millett et al. 2009; Melillo et al. 2014). Accordingly, even wetlands that currently have a high probability of plover presence may not be able to continue to support plover productivity in most years or provide habitat for plovers for dispersal if conditions are unsuitable elsewhere. In this case, the northern Great Plains population of plovers may have already and may continue to become even more dependent upon the highly modified habitats along reservoir shorelines and sandbars of the Missouri River.

Our results also corroborate concerns that habitat loss could be a factor inhibiting recovery of the northern Great Plains population of piping plovers (Prindiville Gaines et al. 1988; Root and Ryan 2004). Clearly, interannual water levels and flow dynamics are responsible for the creation and maintenance of habitat in all three principal breeding areas for plovers in the Northern Great Plains (Prindiville Gaines et al. 1988; U.S. Fish and Wildlife Service 2003b; Anteau et al. 2014), but the bulk of attention about plover conservation has been focused on river and reservoir management (U.S. Fish and Wildlife Service 2003b). Including prairie wetland restoration and management as part of a broader strategy might be necessary for conserving piping plovers over the long term. In the PPR, restoration of wetlands within the catchment of plover-nesting wetlands may partially restore predrainage hydrology (Anteau 2012) and we speculate that this could consequently improve wetland habitat quality for plover habitat and ensure the region continues to be an important area contributing to the persistence of this species. Furthermore, for waterbird conservation efforts to be successful worldwide in the face of changing climate and land-use practices, we may need to consider how alterations in a wetland's catchment influence that wetland's ability to provide appropriate habitat.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Table S1. Wetland identification, year of survey, piping plover Charadrius melodus presence–absence, percent full, proportion of drained wetlands, and drought index data used in regression analyses in North Dakota from 2001 to 2011.

Found at DOI: 10.3996/072015-JFWM-068.S1; (51 KB XLS).

Reference S1. Dahl TE. 1990. Wetlands losses in the United States, 1780's–1980's. Washington, D.C.: U.S. Department of the Interior, Fish and Wildlife Service.

Found at DOI: 10.3996/072015-JFWM-068.S2; also available at http://www.fws.gov/wetlands/Documents/Wetlands-Losses-in-the-United-States-1780s-to-1980s.pdf (2380 KB PDF).

Reference S2. Dahl TE. 2014. Status and trends of prairie wetlands in the United States 1997 to 2009. Washington, D.C.: U.S. Department of the Interior, Fish and Wildlife Service, Ecological Services.

Found at DOI: 10.3996/072015-JFWM-068.S3; also available at http://www.fws.gov/wetlands/Documents/Status-and-Trends-of-Prairie-Wetlands-in-the-United-States-1997-to-2009.pdf (16.9 MB PDF).

Reference S3. Elliott-Smith E, Haig SM, Powers BM. 2009. Data from the 2006 International Piping Plover Census: U.S. Geological Survey Data Series 426, 326 p.

Found at DOI: 10.3996/072015-JFWM-068.S4; also available at http://pubs.usgs.gov/ds/426/pdf/ds426.pdf. (10.3 MB PDF).

Reference S4. Erwin RM., Laubhan MK, Cornely JE, Bradshaw DM. 2000. Managing wetlands for waterbirds: how managers can make a difference in improving habitat to support a North American bird conservation plan. Bonney R, Pashley DN, Cooper RJ, Niles L, editors. Strategies for bird conservation: the partners in flight planning process. Fort Collins, Colorado: U.S. Department of Agriculture, Forest Service, Rocky Mountain Reserach Station.

Found at DOI: 10.3996/072015-JFWM-068.S5; also available at http://www.birds.cornell.edu/pifcapemay/erwin.htm. (340 KB PDF).

Reference S5. Shaffer TL, Sherfy MH, Anteau MJ, Stucker JH, Sovada MA, Roche EA, Wiltermuth MT, Buhl TK, Dovichin CM. 2013. Accuracy of the Missouri River least tern and piping plover monitoring program: considerations for the future. U.S. Geological Survey Open-File Report 2013-1176.

Found at DOI: 10.3996/072015-JFWM-068.S6; also available at http://pubs.usgs.gov/of/2013/1176/ (9409 KB PDF).

Reference S6. Sherfy MH, Stucker JH, Anteau MJ. 2009. Missouri River emergent sandbar habitat monitoring plan—a conceptual framework for ddaptive management: U.S. Geological Survey Open-File Report 2008-1223.

Found at DOI: 10.3996/072015-JFWM-068.S7; also available at http://pubs.usgs.gov/of/2008/1223/pdf/ofr2008-1223.pdf. (13.3 MB PDF).

Reference S7. U.S. Fish and Wildlife Service. 2003. U.S. Fish and Wildlife Service 2003 amendment to the 2000 Biological Opinion on the Missouri River main stem reservoir system, operation and maintenance of the Missouri River bank stabilization and navigation project, and operation of the Kansas River res. Washington, D.C.

Found at DOI: 10.3996/072015-JFWM-068.S8; also available at http://www.fws.gov/feature/pdfs/FinalBO.pdf. (2313 KB PDF).

We thank Alex Lawton and Peter Mockus for collecting aerial photos and providing GIS data assistance. Stuart Blotter, the Farm Service Agency and Natural Resources Conservation Service offices of North Dakota, and the U.S. Fish and Wildlife Service Habitat and Population Evaluation Team, Midwest Region (Sue Kvas) provided historical aerial photographs. Kirsten Brennan, Elise Elliott-Smith, Chris Swanson, and Paulette Scherr all provided piping plover survey data and many staff and volunteers collected piping plover survey data. Support and advice were provided by Mike Symanski, Wes Newton, Terry Shaffer, Jane Austin, Rhianna Golden, Mark Sherfy, and Josh Stafford. The International Water Institute provided light detection and ranging data. We thank D. Haukos and two anonymous reviewers for improvements to this manuscript. This research was funded by the Plains and Prairie Potholes Landscape Conservation Cooperative and U.S. Geological Survey Northern Prairie Wildlife Research Center. None of the project sponsors had any influence on the content of the submitted or published manuscript or required approval of the final manuscript to be published.

Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.