Nest-Site Selection Patterns of Coexisting Sandhill and Whooping Cranes in Wisconsin

Whooping cranes Grus americana and sandhill cranes Antigone canadensis have similar evolutionary lineage and ecological traits. They represent the only two crane species native to North America, are slow to reach reproductive maturity (i.e., typically ≥3 y), have low annual reproductive potential (i.e., <2 young annually), and are subject to high juvenile mortality rates (Kuyt and Goossen 1987; Tacha et al. 1989; Drewien et al. 1995). Further, both species rely on shallowly flooded wetlands for nesting and roosting, and forage in both wetland and upland habitats (Walkinshaw 1973; Urbanek and Bookhout 1992; Meine and Archibald 1996; Timoney 1999). Widespread wetland destruction and unregulated hunting contributed to dramatic declines in the abundance and breeding range of both whooping cranes and sandhill cranes until the early 1900s, but populations of both species have increased over the past 75 y (Meine and Archibald 1996).

Protected as a Federally Endangered species (ESA 1973, as amended), whooping crane populations are slowly recovering from an historical low in 1941 of <25 known individuals remaining in the wild (Meine and Archibald 1996; CWS and USFWS 2007). To reduce risk of extinction and enhance recovery, the U.S. Fish and Wildlife Service (USFWS) and its partners have attempted to establish several spatially distinct breeding populations of whooping cranes from captive-bred stock (CWS and USFWS 2007). In 2001, the first whooping cranes were released at the Necedah National Wildlife Refuge (NWR) in central Wisconsin to form the eastern migratory population (EMP) of whooping cranes in the eastern United States. Necedah NWR was selected as the breeding location for reintroduced whooping cranes, in part because of the presence of successfully breeding sandhill cranes (Cannon 1999; USFWS 2001).

Despite 15 y of effort and continued releases of captive-reared whooping cranes, the EMP has not yet realized reproductive rates adequate for positive population growth (Servanty et al. 2014). Furthermore, low reproductive success appears to be the EMP's primary limiting factor (Servanty et al. 2014). From 2005 to 2015, EMP whooping cranes have laid eggs in 175 nests (excluding 22 experimental nests during 2014–2015), of which 38 produced a viable chick (R. P. Urbanek, Necedah NWR, unpublished report). The 22% apparent nest survival rate is markedly lower than those rates reported for sandhill cranes in Michigan (i.e., 63%; Urbanek and Bookout 1992), Minnesota (i.e., 76%, DiMatteo 1992; 56%, Maxson et al. 2008), and Wisconsin (i.e., 84%, Howard 1977; 83%, Bennett 1978). During the 2014 and 2015 breeding seasons, apparent nest success of whooping cranes at Necedah NWR was 29% (n = 17) and 38% (n = 13), respectively; while sandhill crane apparent nest success was 56% (n = 16) and 51% (n = 35) during the same years (B. N. Strobel, Necedah NWR, unpublished data).

Several hypotheses have been developed to explain the low reproductive success of the EMP, including 1) limited forage resources in wetlands, 2) poor body condition of returning adults, 3) predation of chicks, 4) harassment of incubating birds by biting insects, and 5) inappropriate nest-site selection (Urbanek et al. 2010; Runge et al. 2011). Of these, recent research supports the hypothesis that whooping crane nest failures were related to seasonal abundance of ornithophilic blackflies (Diptera: Simuliidae; Converse et al. 2013). Sandhill cranes are also known hosts for Simulium spp. black flies (Anderson and DeFoliart 1961; Urbanek et al. 2010), yet sandhill crane reproduction at Seney NWR in Michigan did not appear to be affected by abundant black fly populations (Urbanek and Bookhout 1992). Investigating differences in nesting habitat selected by sandhill cranes and whooping cranes is the first step toward determining if one more of these mechanisms may account for apparent differences in nest survival between species.

Several studies have described the nesting habitat of sandhill cranes and whooping cranes. Baker et al. (1995) found that sandhill cranes nesting in the Upper Peninsula of Michigan selected nest sites with less forested upland vegetation than was randomly available. Sandhill crane nest sites in Wisconsin (Howard 1977) and Michigan (Walkinshaw 1965) were typically surrounded by dense vegetation providing visual obscurity. In contrast, sandhill cranes in northwestern Minnesota (Maxson et al. 2008) nested in less dense vegetation than was available. Whooping cranes nesting in Wood Buffalo National Park selected “visually open” nest sites with large amounts of open water (Timoney 1997). If present, species-specific patterns in nest site characteristics may explain some of the differences in reproductive success in Necedah NWR. Comparing the characteristics of coexisting sandhill and whooping crane nest-site selection may shed light on the low reproductive success of the EMP. Our objectives were to 1) compare nest-site characteristics between whooping cranes and sandhill cranes, 2) compare characteristics of nesting territories between whooping cranes and sandhill cranes, and 3) estimate nest-site selection by comparing characteristics of nest sites from those potentially suitable locations within nesting territories.

We collected data for this study from the Necedah NWR and the Meadow Valley State Wildlife Area in Juneau County, Wisconsin. Combined, these areas covered approximately 44,515 ha in the central sands region of Wisconsin (Figure 1). The region is characterized by deep coarse sandy soils that remain from wind-sculpted sand dunes that formed after glacial lake Wisconsin receded 13,000 y ago (Clayton and Attig 1989). Vegetation communities included sand prairies, oak–pine Quercus spp.–Pinus spp. barrens, mixed forests, sedge Carex spp. meadow wetlands, and impounded wetlands. Water levels in most of the impounded wetlands were managed through a network of conveyance ditches and water control structures that were originally created during the early 1900s. Precipitation accounted for 85% of the annual water inflow to the Necedah NWR, and averaged 80 cm annually (Hunt et al. 2000).

We located whooping crane nests using a combination of ground-based and fixed-winged aerial monitoring, both facilitated by very high frequency radiotelemetry. We conducted ground-based surveys daily and aerial surveys approximately weekly. The intensive monitoring allowed us to determine locations of whooping crane nests and date that incubation was initiated for each nest. However, the more cryptic coloration of sandhill cranes and lack of very high frequency radiotransmitters required different methods to locate nests. Therefore, we located sandhill crane nests using helicopter-based aerial searches throughout sedge meadows and impounded wetlands in areas known to be previously occupied by nesting cranes. An observer located a crane nest and then recorded the location using a photo or a Global Positioning System location. We monitored the status of all known nests at least weekly throughout the nesting period. We visited each nest to measure attributes of the nest site after incubation ended and we no longer observed cranes in the vicinity. To compare the conditions at nest sites with those of sites presumed available to the cranes, we used ArcGIS (ESRI Inc., Redlands, CA) to randomly position three points between 25 and 200 m from each nest. We used a maximum distance of 200 m because a previous study was unable to detect sandhill crane nest-site habitat selection beyond that scale (Baker et al. 1995). We constrained random locations to those that occurred in standing water <1 m deep.

We recorded site characteristics at 1-m intervals along four 5-m transects oriented in the cardinal directions from each nest site or random location (i.e., plot center). At each of the resulting 20 measurement locations, we used a staff to measure water depth, and a 0.32-m cord to demarcate a 1-m² circular area around the staff. Within each circular area we obtained an ocular estimate of percent coverage of the two most abundant cover types; and, if the cover type was vegetation, we measured its height (Table 1). We did not identify vegetation to species; instead, we used four general categories representing distinct structural growth forms, including grass (e.g., sedges, wool-grass Scirpus cyperinus, cattail Typha spp., reed-canary grass Phalaris arundinacea), emergent forb (e.g., water lily Nymphaea spp., meadowsweet Spiraea alba), shrub (e.g., willow Salix spp., dogwood Cornus spp.), and open water (e.g., the absence of rooted vegetation cover).

We determined visual obstruction surrounding each nest site and randomly located site using methods developed by Broadway (2015). We captured digital images of a red cylinder (30.5 cm long and 3.8 cm diameter) using a Canon PowerShot A1200 camera (Canon USA Inc., Arlington, VA) set to 10× zoom and held at 1.5 m (i.e., approximate head height of a standing crane) above the substrate at the end of each 5-m transect. We imported each photo into the GNU Image Manipulation Program (GIMP version 2.8.16, gimp.org). We used the color selection tool within GIMP to select all of the visible (i.e., unobscured) red pixels within the image. We divided the number of visible red pixels by the average number of pixels from photos of unobstructed cylinders and subtracted this value from 1 to index the percent visual obstruction in each photo.

We excluded data from two nests in areas where habitat management actions (i.e., prescribed fire, wetland draw-down) had modified vegetation or water conditions after nest completion but prior to our data collection. We constructed covariate values for each nest site and random location by averaging values from the 20 observations collected. To meet objectives 1 and 2 we used the t.test function in the stats package of Program R (R Core Team 2015) to conduct Welch's 2-sample tests of the mean covariate values between species (i.e., SACR, WHCR) at the 5-m and 200-m scales (Table 2). To meet objective 3, we used the glmer function and the logit link to conduct a mixed-effects logistic regression on the binomial response variable of type (i.e., 0 = random site, 1 = nest site). We constructed six models using four fixed-effects covariates to describe biotic and abiotic characteristics of the sites (Table 1). We also included six additional models in the model set that were analogous to the first six but also included the fixed-effect covariate of species (Table 3). We controlled for interterritory variability by specifying an arbitrary territory ID as the random effect in all models. We evaluated all models using Akaike's Information Criterion corrected for small sample size (AIC_c; Burnham and Anderson 2002).

We collected data at 35 sandhill crane nests, 20 whooping crane nests, and 164 randomly selected locations from 19 May to 25 July, which averaged 53.7 d (SD = 13.6, N = 49) after cranes initiated incubation at nests (Data S1, Supplemental Material). At the 5-m scale, whooping crane nests were surrounded by slightly greater percent coverage of open water than were sandhill nests (Table 2). We did not detect any additional differences between sandhill and whooping crane nests at the 5-m scale. Similarly, we were unable to detect a difference between any of the characteristics that we measured at whooping crane and sandhill cranes nests at the 200-m scale.

Of the 12 mixed-effects models evaluating the interterritory nest-site selection patterns, 2 received 99% of the total weight (Table 3). Both of the top models included the polynomial effect of water depth around the nest site (β₀ = −2.79, β₁ = 14.56, β₂ = −22.29; Figure 2). The second-highest ranked model also included the binomial species covariate, although it did not appear to be significantly different from zero (β = −0.21, P = 0.51). The average water depth observed at nest sites was 0.29 m (SD = 0.13). Throughout the model set, models that included the species covariate received lower weight than the analogous model that did not include the species covariate (Table 3).

Despite evaluating habitat characteristics at two scales, we detected only subtle differences between nest sites of sandhill and whooping cranes. Differences between sandhill and whooping crane nests sites may have been obscured by our averaging characteristics by nest site prior to univariate comparisons. Sandhill crane nests may have been surrounded by slightly less open water and slightly greater woody vegetation coverage than whooping crane nests at the 5-m and 200-m scales, respectively. However, because these differences were slight, it seems unlikely that they could result in a species-specific difference in nest success on our study site. Evaluating the within-territory mixed-effects models further suggests that nest-site selection patterns of whooping cranes and sandhill cranes were similar. Overall, we were unable to detect substantial differences in the characteristics of sandhill and whooping cranes nest sites on the Necedah NWR.

Water depth has been discussed as a potentially important attribute for nest site selection of sandhill cranes in other studies. Preference for specific depths may be the result of a trade-off between the benefits of deeper sites (e.g., increased isolation from predators) and the efficiency of shallower sites (e.g., shorter and more stable nest platforms) to reproductive success. Taken together, sandhill and whooping cranes selected nest sites with an average water depth of 0.29 m (0.20–0.37 interquartile range). However, some evidence indicated that the average water depth at sandhill cranes was more variable than at whooping crane nests (Figure 2). The average water depth of both species' nest sites at Necedah NWR was deeper than most other average nest-site depths. Howard (1977) found that sandhill crane nests in central Wisconsin were in an average depth of 0.12 m. Similarly, Urbanek and Bookhout (1992) found that sandhill on Seney NWR nested in comparatively shallow water depths of 0.07 m (±0.08 m). Walkinshaw (1965) reported the water depth averaged 0.21 m at sandhill nests in Michigan. It is unclear if the relatively deeper depth we found at nest site on Necedah NWR is the product of regional differences in preference by cranes, differences in available water depths, or a product of the timing of our data collection.

In contrast to similar habitat selection studies on sandhill cranes (Baker et al. 1995) and whooping cranes (Timoney 1999), we were not able to detect patterns in nest site selection based upon vegetation structure. This does not imply that whooping cranes or sandhill cranes are selecting nest sites randomly, only that they do not appear to be selecting nest sites based upon the metrics we used. Defining the breadth of resources available to an individual is an important and challenging aspect to use–availability studies (Manly et al. 2002). The scale at which Baker et al. (1995) and Timoney (1999) defined habitat availability was different from the spatial scale we selected for our study. Timoney (1999) considered available resources across Wood Buffalo National Park and excluded only closed-canopy forests and deep open water from the analysis. Baker et al. (1995) used circle plots of increasing radii around nests and were unable to detect habitat selection beyond the 200-m scale. Despite restricting our vegetation measures to within 200 m of the nest sites, the differences in the vegetation characteristics at nest sites and territories of sandhill and whooping cranes were negligible.

Data from 2014 and 2015 preliminarily suggested sandhill cranes may experience higher nest survival rates than whooping cranes on our study site. However, to account for interannual variation in reproductive success, our research is ongoing. If future data continue to indicate nest survival is species-specific, it does not appear that nest-site habitat selection is a causal factor. Other possible mechanisms that could cause differences between the reproductive success of sandhill cranes and whooping cranes include unintentional consequences of captive propagation or rearing techniques, inherent differences in the reproductive strategies between the species, and the species-wide loss of traits (i.e., genetic bottleneck) important to successful reproduction outside of Wood Buffalo National Park, Canada. However, it is also possible that reproductive success is not species-specific, and both crane species have consistently low reproductive success on our study site relative to breeding sandhill cranes elsewhere in the Midwest. Indeed, the apparent nesting success of sandhill cranes on our study site was lower than that reported by many other studies in the region. If reproductive success of sandhill cranes and whooping cranes are correlated, a priori monitoring of sandhill crane breeding ecology could increase the probability of success for future whooping crane reintroductions.

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.

Data S1. Data set containing biotic and abiotic characteristics measured at 35 sandhill crane Antigone canadensis nests, 20 whooping cranes Grus americana, and 164 randomly selected locations between 25 and 200 m from nests, on the Necedah National Wildlife Refuge and Meadow Valley State Wildlife Area in Juneau County, Wisconsin, during 2015.

Found at DOI: http://dx.doi.org/10.3996/032016-JFWM-025.S1 (18 KB TXT).

Reference S1. Bennett AJ. 1978. Ecology and status of greater sandhill cranes in southeastern Wisconsin. Master's thesis. Stevens Point: University of Wisconsin.

Found at DOI: http://dx.doi.org/10.3996/032016-JFWM-025.S2 (6558 KB PDF).

 Reference S2. Cannon JR. 1999. Wisconsin whooping crane breeding site assessment. Final report. Submitted to The Canadian–United States Whooping Crane Recovery Team, Front Royal, Virginia, USA.

Found at DOI: http://dx.doi.org/10.3996/032016-JFWM-025.S3 (3196 KB PDF).

Reference S3. Howard TJ. 1977. Ecology of the greater sandhill crane in central Wisconsin. Master's thesis. Stevens Point: University of Wisconsin.

Found at DOI: http://dx.doi.org/10.3996/032016-JFWM-025.S4 (3846 KB PDF).

Reference S4. Hunt RJ, Graczyk DJ, Rose WJ. 2000. Water flows in the Necedah National Wildlife Refuge. Middleton, Wisconsin: U.S. Geological Survey. Found at DOI: http://dx.doi.org/10.3996/032016-JFWM-025.S5 (587 KB PDF); also available at wi.water.usgs.gov/pubs/FS-068-00/ (file type).

Reference S5. Timoney K. 1997. The summer breeding habitat of whooping cranes in Wood Buffalo National Park, Canada. Report. Sherwood Park, Alberta, Canada: Treeline Ecological Research. Found at DOI: http://dx.doi.org/10.3996/032016-JFWM-025.S6 (6351 KB PDF).

We thank A. Lacy, E. Szyszkoski of the International Crane Foundation, B. Paulen of the Wisconsin Department of Natural Resources, and B. Lubinski of the U.S. Fish and Wildlife Service for assistance in locating crane nests. We also thank A. Enderle, J. Jaworski, R. Voorhorst and S. Wendel-Hernandez for assisting with the field data collection.

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