Northern Prairie Wildlife Research Center
To date, most information gained about migrating whooping cranes has been derived from non-standardized, incidental observations, such as those occurring in the 2 databases used here. Observational data can be biased by a number of factors. Because confirmed whooping crane sightings consist of a chance observation made by a variety of constituents (i.e., farmers, ranchers, rural mail carriers, and biologists) and documented by a knowledgeable observer, the sightings database can be biased by the detectability of whooping cranes in different habitats and regions in the flyway and the availability of a knowledgeable observer in the region. For example, whooping cranes that stop in an area with a higher density of farmers and ranchers (e.g., loess region of Nebraska) are more likely to be detected than cranes that stop in less populated ranching areas like the Nebraska Sandhills region. Compared to other places in the flyway, whooping cranes may be more commonly reported on refuges, state management lands, or other conservation areas because biologists are actively looking for birds or are more available for confirmation of citizen-reported sightings. Level of interest and effort also may vary among states. For example, the numerous papers published in Nebraska Bird Review and proceedings of crane workshops indicate that biologists in Nebraska have long had a strong interest in recording whooping crane occurrences. Seasonal and yearly biases in observation data also exist. For example, many fall sightings for North Dakota are reported by hunters (S. Kohn, North Dakota Game and Fish Department, Bismarck, ND, personal communication); areas and habitats frequented by hunters likely differ from those frequented by farmers. As areas have become known over the years as whooping crane "use areas," observers have focused increased attention to these regions for further sightings. Landscape patterns also may influence detectability. For example, farmers tend to spend more time monitoring their croplands than wetlands, hence increasing the probability of seeing cranes in crop areas. Furthermore, visibility of areas used by whooping cranes may be obstructed (e.g., heavily forested river edges vs. open reaches, hills that isolate wetland areas from roads). Such spatial and temporal factors will influence the detection of whooping cranes and therefore will bias the data so that particular regions and habitat types may be over- or under represented relative to actual use by migrants. Therefore, these observational data are not appropriate for use in assessing habitat preferences or to address questions such as whether cranes shifted their distribution or habitat use during drought years. Objective evaluation of whooping crane distribution or habitat use patterns, such as determining preferential use of wetland types or feeding habitats, would require targeted studies, preferably using marked or radio-marked cranes.
Although the whooping crane database was collected in a manner that precludes statistical analyses and limits interpretation, it is important to recognize that investigations of any species' habitat use, especially during migration, are subject to many difficulties. Biases associated with detectability and survey effort commonly are discussed in literature relating to monitoring and research. Unlike nesting birds, migrants usually are present in an area for only a short time, and the timing of that occurrence may differ from year to year, depending on abiotic and biotic conditions on their breeding, wintering, or migration areas. For a species with a small population like the whooping crane, it often is very difficult to gather a suitable sample size over a reasonable area of study because migrants often are difficult to locate and occur in small numbers. Solitary or non-gregarious species, such as whooping cranes, are less likely to be detected than large flocks of migrants (i.e., geese, sandhill cranes). Considering the extent of the whooping crane's migration path relative to their non-gregarious nature, the species' sensitivity to human presence, and its small population size, it is easy to understand the difficulty in obtaining a sample size large enough for use in inferential analyses.
It is advantageous that the whooping crane sightings database was collected over a long time period and large geographic region, therefore creating a relatively robust sample size. To date, it is the most inclusive collection of information on general migration patterns of wild, naturally-occurring whooping cranes. The cooperative monitoring program provided a cost-effective method for collecting flyway-wide data, in a manner that was minimally intrusive to whooping cranes. In the discussion below, we consider the research value and limitations of the sightings database that was collected during 1943-99.
Many of the results presented here concur with earlier findings about whooping crane migration. The flyway used by whooping cranes migrating from Aransas NWR through North Dakota is more clearly defined by overlaying 57 years of data, but it remains essentially similar to that outlined by Allen (1952), Johnson and Temple (1980), Armbruster (1990), and in the current recovery plan (U.S. Fish and Wildlife Service 1994). This distribution largely correspond to Allen's (1952:2) grama grass-antelope biome. The apparent association of the migration path with rivers, particularly along the Missouri River in the Dakotas, supports the idea that whooping crane movements during migration are at least partly directed by recognition of landscape features such as stream and wetland mosaics (Gill 1990). The migration path defined from tracking radio-marked cranes (Kuyt 1992: Figure 9) is generally similar to that described here but it did not include locations in North and South Dakota east of the Missouri River or portions of the Platte River or Rainwater Basin east of Kearney, Nebraska as found in the present study. There does not appear to have been any shifts in the spring or fall migration route over the 57 years of data. However, these comparisons were limited by the relatively few observations during the earlier period (1943-74), prior to changes in habitat related to more recent dam building and conversion of grasslands to cropland. Timing of spring and fall migrations also appears similar to that first described by Allen (1952), and no changes in the timing of migration are apparent.
Early studies describing roost sites were generally limited to riverine sites (Aronson and Ellis 1979; Shoemaker et al. 1982; Lingle et al. 1984, 1986), especially along the Platte River and other Nebraska sites. Studies of broader geographical scope have consistently demonstrated the significance of palustrine wetlands for roosting and foraging habitat (Howe 1987, Johns et al. 1997, Richert 1999, this study). The present study showed that riverine roost sites were common only in Nebraska, primarily on the Platte, Niobrara, Middle Loup, and North Loup rivers. The higher use of riverine roosts in Nebraska may be related to the relatively unique geomorphic characteristics of rivers there, which include shallow, relatively slow-moving channel flows and sand bars with little vegetative cover. The other 2 studies examining flyway-wide habitat use also reported high use of palustrine wetlands. Radio-marked cranes roosted primarily on palustrine wetlands in most areas, and only 2 sites used by cranes in the United States were described as riverine (Howe 1987). In Saskatchewan, 84% of observational records were on palustrine wetlands (Johns et al. 1997). In our study, palustrine wetlands were used by all social groups of whooping cranes for both roosting and feeding. However, most of the whooping cranes found on riverine roosts were single cranes or nonfamily groups, particularly on the Platte, although social groups did not differ on feeding or dual-use sites. Richert (1999), using a subset of these data for Nebraska to assess habitat use at various landscape scales, noted that nonfamily groups were primarily the social groups associated with the Rainwater Basin and Platte River areas whereas family groups were more commonly associated with the Table Playa area in Custer County, Nebraska. This area contained a much larger proportion of grassland at both local and landscape scales than did the Rainwater Basin or Platte River areas. Further investigation of other regions of the flyway is needed to determine whether grassland is an important landscape feature for use by family groups.
Most palustrine wetlands used for roosting were defined as seasonal or semipermanent wetlands; feeding sites also included many temporary palustrine wetlands. Howe (1987) reported radio-marked cranes used intermittently-exposed and semipermanent wetlands more than any other regimes for both feeding and roosting; temporarily-flooded wetlands often were used in fall. In Saskatchewan, migrant cranes were most frequently observed on seasonal and temporary wetlands in spring and on semipermanent and permanent wetlands in fall (Johns et al. 1997). Differences among areas, years, or studies likely were affected at least in part by availability of wetland regimes, in response to climate variation on seasonal and yearly basis. Differences found in whooping crane habitat use among states or regions in this study also may reflect differences in how the information on wetland classification was obtained. Based on discussions with W. Jobman, S. Kohn, and other biologists who provided information, it is apparent that the derivation of wetland classification data varied. Most of the Nebraska wetland classification data, particularly for records after the early 1980s when NWI maps became readily available, were derived from the information directly on NWI maps, which were frequently used to map crane locations. In such cases, class and modifiers (water regime) appear to be derived from the map polygon rather than from the deepest water regime mapped for that wetland basin or from field observations. Biologists in other states, however, seemed to have relied primarily on field observations to report wetland class data, and reported water regime where the most permanent water regime applied to the entire basin. Although this would not affect wetland system (lacustrine, palustrine, riverine), it probably did affect whether subclass and class were recorded, and how water regime was characterized. We caution that observers should not rely on NWI maps because 1) some errors do exist in the original NWI maps, 2) NWI maps are now >10 years old, and 3) wetland characteristics particularly regime may have changed (e.g., additional drainage efforts, change in water regime due to prolonged drought or flooding). We recommend water regime instead be determined using judgment in the field at the time of observation, based on the deepest regime of the entire wetland.
Whooping cranes were observed on a wide range of wetland sizes in both spring and fall. We found no real pattern of use by social groups among the different sizes of wetlands. Cranes often were observed roosting on large managed wetlands (e.g., moist-soil units, impoundments) on state or federal lands in fall, but large lakes and natural wetlands also were used in both seasons. Radio-tracked cranes (Howe 1987) also were located on a range of wetland sizes, but over 50% were located on wetlands ≤1 ha. Unfortunately, wetland sizes were not consistently recorded for all wetland sites in that study (Armbruster 1990:9). Although there was no consistent pattern suggesting cranes usually used smaller wetlands for feeding sites, dual-use sites usually were small (<2 ha) wetlands; the latter might reflect lack of availability of larger wetlands for roosting in those areas. Investigation of wetland densities and size classes available around sites, using archival remote sensing data, could reveal a clearer picture of site-use patterns.
Water depths were recorded for either the entire wetland used during a stopover or for the coordinate location within the wetland where the cranes had been observed roosting or foraging. Unfortunately, there were no records where both were recorded. The significance of shallow water sites for both whooping and sandhill cranes was discussed by Armbruster (1990:8). Average water depths at specific sites within roost wetlands and feeding wetlands were similar to those reported earlier (Lingle et al. 1984, 1986; Howe 1987; Ward and Anderson 1987; Johns et al. 1997) but toward the high end of Johnson and Temple's (1980) optimum water depth of 7.6-20.3 cm (2.2-8.0 inches).
Results of this study also concur with previous findings that cranes usually were associated with sites having scattered or no vegetation (Johnson and Temple 1980, Howe 1987, Johns et al. 1997). Riverine roost sites and dual-use sites were consistent in their lack of vegetation but feeding sites tended to have more vegetation. Most of the commonly occurring vegetative types were of low stature and thus would not likely obstruct visibility for cranes. Unfortunately, willow, which is of interest relative to island management on the Platte River, was not a defined category, so we are unable to evaluate presence or distribution of willow in these data except for a few scattered occurrences when willow was specifically denoted under "other" vegetation.
Whooping cranes appear similar to sandhill cranes in their frequent use of cropland for feeding, particularly corn and wheat stubble (Howe 1987, Johns et al. 1997, this study). However, data from dual-use sites indicated that wetlands may provide important feeding areas for some whooping cranes. Howe (1987) did not distinguish between feeding-only and dual-use sites for radio-marked whooping cranes. He noted that the importance of cropland for feeding-only sites was likely higher than the 42% he reported because many feeding sites were actually categorized as roost sites. That is consistent with the frequent use of permanent or seasonally-flooded wetlands for dual-use sites in this study. The similarity of results between roost and dual-use sites in this study suggest the 2 site uses could be merged for this database. However, we suspect closer examination of sites (i.e., longer observations to verify roost-only or roost-and-feeding activity) may reveal important differences between sites used exclusively for roosting and those used for both feeding and roosting.
We cannot assess the relative value of cropland, wetland, or grassland habitats for foraging cranes with these data because we lack any measure of total time spent feeding in each habitat type. We also do not have adequate data on available habitats around each site. Foraging strategies likely vary depending on season (nutritional needs of cranes, seasonal availability of food), juxtaposition of roost and feeding habitats, availability of habitats, and availability of suitable foods. A more definitive evaluation of the relative use and value of cropland, wetland, and grassland habitats would require a study of color- or radio-marked cranes combined with time-activity budgets, similar to that conducted by Howe (1987) or Lingle et al. (1991). In the latter study, which was conducted in south-central Nebraska, diurnal habitat use was nearly evenly divided between upland and wetland habitats: 37% of bird-hours were on corn stubble, 18% on tilled wetlands, and 17% on natural wetlands. It would be interesting to conduct comparative studies elsewhere in the flyway, particularly in areas with varying proportions of cropland and native habitats. Further examination of the site evaluation data set using GIS also could provide some additional insights into availability of wetland, grassland, or upland habitats relative to site use.
Distance to feeding sites varied with roost type. Palustrine roosts usually were within 0.8 km of feeding sites, as also was reported by Howe (1987). Riverine roost sites, however, tended to be farther from feeding sites. Distances were recorded as categories rather than as a continuous variable, and thus we lack actual maximum distances between roost and feeding sites. Distances between roosts and feeding sites will be influenced by the availability of habitats and foods (e.g., Frederick et al. 1987). On the Platte River, changes in habitat and food availability over time may have increased distances between frequently-used roosts and feeding sites. G. Krapu (Northern Prairie Wildlife Research Center, Jamestown, ND, personal communication) has documented that sandhill cranes roosting on the Platte River in the late 1990s fly longer distances to forage in corn fields than they did 20 years previously; he relates this directly to reduced corn availability in the fields due to improved harvest efficiencies. Palustrine wetlands in the Great Plains often are surrounded by croplands (e.g., Richert 1999, this study). Johns et al. (1997) suggested areas of relatively high wetland density may attract cranes, in particular family groups. We recommend using remote sensing and GIS techniques, similar to the work conducted by Richert (1999) for Nebraska, to examine availability and juxtaposition of habitats relative to roost and feeding sites elsewhere in the flyway.
Horizontal visibility (i.e., having unobstructed view at the level of a crane's head [1.4 m]) has long been considered an important aspect defining optimum and secure habitat for whooping cranes (Shenk and Armbruster 1986, Armbruster 1990). Nearly half of the roost sites and two-thirds of feeding sites, however, were defined as having visibility <0.4 km. These distances are within the range given for sandhill cranes on roosts surrounded by vegetation (140 m) or visible from a road (380 m) (Lovvorn and Kirkpatrick 1981). They suggested that sandhill cranes avoid disturbance by maximizing either distance to human development or visual isolation from human activities. This bears further examination for whooping crane migration habitat, particularly for application to habitat management and interpretive development (e.g., placement and management of crane viewing sites). However, such relationships cannot be adequately examined using the site evaluation database. The scale of measures used here were categorical and relatively coarse (smallest distance to human development was 0.40 km). Over 80% of the sites were within 0.8 km of some human development; this distribution may reflect a relatively high intensity of human development (most likely section roads) and associated human activity, or it may reflect detectability of cranes. For testing an interaction between visibility and distances, however, a better sample size of long distances would be needed. In addition, the type of human development was not defined in the site evaluation database, although it was in the Nebraska forms (Report 6). Cranes' perception and reactions to, or avoidance of, disturbances likely include a combination of factors such as frequency (e.g., number of vehicles passing per hour), noise level, lighting at night, distance to disturbance source, and visibility of the disturbance and surrounding habitat, and in certain areas also may be influenced by the cranes' habituation to disturbances. More detailed examination of types of disturbances or human developments and their relationship to visibility would be valuable. A study combining surveys and behavioral observations, such as used in Europe to examine effects of disturbances to field-feeding geese (e.g., van der Zande et al. 1980), would be feasible on the Platte River and other areas of concern.
Whooping cranes are commonly associated with sandhill cranes on both palustrine and riverine wetlands (Johns et al. 1997, this study), but the co-occurrence was most frequent for nonfamily groups on riverine sites, primarily on or around the Platte River in spring. These species likely share some preferences for roost habitat, such as shallow water and open visibility for feeding and roost sites (Lovvorn and Kirkpatrick 1981, Armbruster 1990). Single whooping cranes also may be attracted to sandhill crane flocks because their presence would reflect appropriate habitat and they provide additional sentinels for alerting the birds to danger.
Private lands provide the vast majority of cropland and wetland habitats used by whooping cranes during migration (Howe 1987, Johns et al. 1997, this study). However, whooping cranes have been observed on a wide variety of state and federal lands over the years, and some of these areas have received frequent use by cranes. National wildlife refuges, WPAs, and state lands often provide roost locations (often large, shallow natural or managed wetlands), and cranes forage on adjacent private croplands. Three public areas having many observations over the years already have been designated as critical habitat for the whooping crane (Cheyenne Bottoms SWA, Quivira NWR, and Salt Plains NWR). Whooping cranes appear to obtain much of their food on cropland, much like sandhill cranes (Lovvorn and Kirkpatrick 1981, Howe 1987, Johns et al. 1997, this study; but see Lingle et al. 1991). We did not observe a difference among social groups for feeding habitat types as did Johns et al. (1997).
We are reluctant to interpret the results of site security because the meaning of this variable may vary among some observers. For example, S. Kohn (personal communication) had interpreted this term to infer immediate threat to whooping cranes, including the presence of hunters, human disturbances, or threats from utility lines. W. Jobman, however, interpreted this variable to mean that the particular site was threatened with degradation (e.g., drainage, cultivation of wetland or upland habitat). Interestingly, most feeding sites, which largely were composed of private cropland, were considered secure. Although availability of croplands is unlikely to seriously decline in the Great Plains in the foreseeable future, the future quality and security of wetlands used for feeding or roosting are much less clear. Continued loss and degradation of wetlands in intensively-cropped areas of the Great Plains may reduce availability of natural foods and secure roost sites to migrant cranes.
Although these results provide additional insight to distribution and habitat use of whooping cranes during migration, they cannot be used to predict the most suitable habitat for whooping cranes in the proposed Wisconsin-Florida flyway because 1) the data cannot provide an unbiased representation, and 2) most wetland and upland habitats types, and patterns of habitat patches, within the proposed flyway corridor are different from those in the Aransas-Wood Buffalo flyway. As indicated in Richert (1999), conclusions about habitat use are specific to the place of study or to environments with similar habitat composition and landscape pattern. However, trends in habitat use, as found in previous studies and the current investigation, should be considered by those planning the new flyway. For example, studies have consistently found that palustrine wetlands are important for roosting and croplands for feeding. It is likely that these same habitat types will be important to cranes in the new flyway. Although spatial patterns of social groups during migration are difficult to pinpoint, current information suggests there are some differences. Therefore, planners of the new flyway should be attuned to possible differences in habitat needs by different social groups.
Other biologists have stated the need to better understand habitat selection of migratory species (Lingle et al. 1991, Askins 2000), and interests in studies of migration ecology have increased since the application of remote sensing and GIS has become more prevalent within wildlife research (Butler et al. 1995, Farmer and Parent 1997). Further work on whooping crane migration would not only increase the knowledge base about this species but also would contribute to information about migration in general. The works of Lingle et al. (1991), Armbruster (1990), and Richert (1999) suggest that patterns of habitat selection involve recognition of landscape components. Mapped information from observation data also suggests that habitat selection is influenced by landscape structure. For example, North Dakota data suggest a relationship between whooping crane stopovers and the path of the Missouri River and geomorphic features of the Missouri Coteau. We recommend further work using remotely-sensed data and other digital databases, such as the NWI and various data layers created for state GAP analyses, to better understand general migration patterns and to investigate relationships between whooping crane sighting locations and landscape features.