Seasonal movements and habitat use of a threatened rail among fragmented riparian tidal wetlands and impoundments: management implications

1 Introduction

Wetlands support 40% of Earth’s plant and animal species but experience disproportionate negative impacts of anthropogenic land conversion for the purposes of agriculture, silviculture, and development (Dahl, 1990; Wilen and Frayer, 1990; Dahl and Johnson, 1991; USFWS, 2023). In the conterminous United States, the average loss of wetlands between the 1780s and 1986 was calculated to be 53% (Dahl, 1990). Yet, about 50% of all listed threatened or endangered species reside in the remaining wetlands (USFWS, 2023), that are increasingly subject to human modification.

The king rail (Rallus elegans), a secretive marsh bird, is an obligate freshwater wetland specialist that was historically found throughout the eastern United States (Meanley, 1969; Cooper, 2008). However, inland wetland loss has resulted in king rail populations being distributed mostly toward the coastal margins of their prior range. Decreasing population trends have led to the king rail being designated globally Near Threatened (BirdLife International, 2021). The king rail was identified as a focal species for the U.S. Fish and Wildlife Service in 2008 (Cooper, 2008), and more recently federally designated as a bird of conservation concern (USFWS, 2021). Yet, king rails are difficult to study because of their secretive nature. Moreover, they live in concealing habitats and are therefore rarely seen and more often detected by voice (Schroeder and McRae, 2019, 2020). In saltmarshes, king rails are replaced by their sister species, the eastern clapper rail (Rallus crepitans) (Maley and Brumfield, 2013). The species’ ranges may overlap in brackish marsh, and hybridization between the species has been reported (Maley and Brumfield, 2013; Elizondo et al., 2025). Hybridization poses a risk to the integrity of the king rail gene pool which, in addition to population declines, adds to the vulnerability of the species.

King rails’ bodies are laterally compressed as an adaptation to navigate densely vegetated marsh habitat. Their diet is mostly carnivorous and they feed primarily on decapods (Crustacea), especially crayfish, though they will eat crabs, minnows, insects, bivalves and other prey caught with their long slender bills while wading in shallow water (Meanley, 1969). In early Spring, males foray out from their wintering home ranges (Kolts and McRae, 2017) and broadcast loud, pulsed kek kek kek mating calls, sometimes continuously for extended periods (Schroeder and McRae, 2019), to attract a mate with whom they will defend a breeding territory. Mates share parental duties. Both participate in building a well-concealed nest in reeds or rushes (Meanley, 1969; Clauser and McRae, 2017). Both parents incubate for about 21–24 days (Taylor and van Perlo, 1998), with males performing the nocturnal shift (Clauser and McRae, 2017). Females lay clutches of up to 12 eggs (Meanley, 1969; Taylor and van Perlo, 1998), but more commonly 8–10 eggs (Johnson and McRae, 2022). Atlantic coastal plain king rails initiate nests through July, but nest predation rates are high (Clauser and McRae, 2016; M. Linke and S. McRae, unpublished). Though renests are regularly observed, there have been no reports of double-brooding. Chicks are downy and nidifugous, able to run, swim and thermoregulate within a day of hatching, but they are reliant on parental feedings for about 4 weeks (Meanley, 1969; Taylor and van Perlo, 1998). Brood-rearing extends for about 9 weeks post-hatch at which time the offspring become volant.

Remaining strongholds for breeding king rails tend to be in areas with large expanses of freshwater marsh. Along the Gulf Coast of the United States, king rails inhabit impounded rice fields and crayfish ponds (Pierluissi and King, 2008) as well as large natural emergent marshes along the coastal plain (Perkins, 2007; Perkins et al., 2010; Maley, 2012; Pickens and King, 2013) and Lower Mississippi valley (Krementz et al., 2016; Roberts et al., 2026). Along the mid-Atlantic coast, king rails were historically found in the floodplains of rivers feeding the Chesapeake Bay (Meanley, 1969). However, they have become scarce: surveys conducted in Virginia from 2014 to 2016 failed to detect them (unpublished, described in Stiffler et al., 2018). A 12-year study conducted at Mackay Island National Wildlife Refuge (NWR) along the intracoastal waterway of North Carolina monitored king rails nesting in vast, natural, fresh to oligohaline emergent marshes, and foraging in drawn-down impoundments in the summer (Brackett et al., 2013; Schroeder and McRae, 2018; Johnson and McRae, 2022; Gause and McRae, 2025). Migratory populations breed in marshes along the Upper Mississippi Valley (Darrah and Krementz, 2011; Bolenbaugh et al., 2012) and along the southern shore of Lake Erie where hemi-marsh impoundments (50% water: 50% vegetation) are managed for waterfowl (Kane et al., 2019; Brewer et al., 2023).

In coastal South Carolina, where the king rail is considered a species of greatest conservation need (SCDNR, 2015), the undeveloped coast is dominated by unsuitable saltmarshes and beaches, so much of the king rail population resides in managed impoundments and patches of natural marsh along tidal rivers. These tidal freshwater wetlands are dramatically different in structure and hydrology from other coastal habitats where resident king rails have been studied (but see McGregor (2007) and Ricketts (2011) for surveys conducted in and around the ACE basin (a delta at the confluence of the Ashepoo, Combahee, and Edisto Rivers, Colleton County, between Charleston and Beaufort, SC)).

Wetlands in the human-modified fluvial systems just north of Charleston, SC, are highly fragmented. The landscape was historically modified for the Carolina Gold rice industry beginning in the early 1700s (Tibbetts, 2006). Plantations were established along the sinuous tidal rivers of the coastal plain. Commercial rice cultivation and export was discontinued in the region in 1911, after the loss of knowledgeable enslaved laborers, and subsequent costly devastation from several major hurricanes (Morgan et al., 1975; Tibbetts, 2006).

Since the collapse of the industry, the former rice impoundments have been subject to irregular management, resulting in the formation of a mosaic of wetland units with habitat at different stages of ecological succession. Of the remnant and derelict rice plantations, only 48,785 acres (15% of the total) in South Carolina remain as managed impounded marsh (Devoe and Baughman, 1986). Recent restoration of water control structures has enabled impoundment managers to resume rice cultivation during the warmer months to provide winter forage for waterfowl. An initiative to revisit the planting of Carolina Gold rice was undertaken by refuge staff at Waccamaw NWR for the dual purposes of educating the public about historical land use and benefiting declining wading birds including the king rail (USFWS, 2022).

Previous radiotelemetry studies followed king rails in vast, open marshes (Pickens and King, 2013; Kolts and McRae, 2017; Kane et al., 2019; Brewer et al., 2023; Schmit, 2024). Little is known about the movement ecology of rails in other habitats due to the difficulty of following them in densely vegetated environments. This site provided us a unique opportunity to investigate habitat use and movements in a human-modified landscape where impoundments and tidal marshes are intermixed with wooded wetlands, upland woodlands and developed residential areas. The principal aim of this study was to learn how king rails navigate this fragmented landscape of lowland tidal rivers shaped by a history of sporadic management. Specific aims were (1) to characterize king rail home ranges in the breeding and non-breeding seasons, (2) to document movement patterns within and between seasons, and (3) to characterize king rail habitat use and microhabitat selection during the breeding season. Habitat types were defined as units such as impoundments, emergent tidal marsh or flooded forest. We considered microhabitat to be smaller scale areas within units that had unique characteristics (e.g., topographic or vegetation structure). Our findings provide insights to guide management practices in support of conservation of king rails and other wading birds.

2 Methods

2.1 Study area

We studied king rails in South Carolina Lowcountry near Georgetown between March 2023 and February 2025. King rails were observed in impoundments and natural wetlands near the confluence of the Great Pee Dee River, Thoroughfare Creek, and the Waccamaw River (33.541424, -79.172160), in the vicinity of the Hasty Point Plantation tract of Waccamaw National Wildlife Refuge (NWR). The refuge was established in 1997, and this tract was added in 2021. The region is comprised of flooded forest, fringe marsh, and impoundments that include former rice fields, many with broken water control structures creating deteriorated irrigation channels with tidal influence.

The focal research area included five historic impoundments within three managed study sites (Figure 1). Two sites were managed by Waccamaw NWR staff: Hasty Point and Pee Dee East, positioned on opposite sides of the Great Pee Dee River near its difluence into Thoroughfare Creek. The third site, Birdfield, located ~6 km South along the Great Pee Dee River, was privately owned but managed similarly. Pee Dee East and Birdfield were each comprised of two impoundments with independent water control structures. Salinity measurements taken at least monthly throughout the year in impounded and non-impounded units at different tide levels were consistently in the range of 0–0.1 ppt.

At each study site, impoundments were flooded in October for overwintering waterfowl. Shortly after the waterfowl hunting season closed (January 31), the impoundments were drained. Vegetation was removed with a combination of herbicide (Rodeo^® or AquaNeat^®), prescribed burns, plowing, or mechanical flattening with heavy machinery, and the impoundments were planted with rice and/or millet. Once the rice reached ~0.5 m tall, the fields were flooded to suppress weeds. Within each study site, a portion of the impoundment was left unmanipulated, aside from water level changes necessary for cultivating rice in the rest of the impoundment. Unmanipulated sections of the impoundments were dominated by giant cutgrass (Zizaniopsis miliacea). Birdfield had the largest percentage of unmanipulated (Salix spp.) area within impoundments and included more willow trees interspersed with the giant cutgrass as well a forested section dominated by cypress (Taxodium spp.) and tupelo (Nyssa spp.) trees.

2.2 Field data collection

Starting in early February, king rails actively search for mates with kek mating calls, particularly in the hours following dawn and preceding dusk. Capitalizing on their readiness to approach rivals, recorded conspecific vocalizations were broadcast to lure the breeders into the target zone of a whoosh net (Kolts and McRae, 2017). Bungies were released to deploy the net over the bird, briefly trapping them beneath, before retrieving them manually. In smaller areas located in the interior of the marsh near open mudflats a walk-in trap was used with an audio lure. The trap was a ~1.2 × 1.2 m cage, 0.5 m high with 16-gauge green vinyl coated wire fencing on all sides. On two sides of the trap were funneled entry ways, triangular in cross-section, with wire bent inwards at the end to prevent escape of the birds before retrieval.

King rails were banded with a U.S. Geological Survey aluminum band and a unique combination of three Darvic flat spiral color bands (A.C. Hughes, U.K.); the latter were sealed using a portable soldering iron to reduce the possibility of removal. Standard morphometric data of weight (600-g Pesola, ± 5 g), wing chord (wing rule ±1 mm) and tarsus length (calipers, ± 0.1 mm) were collected. For molecular sex determination, up to 50 μl of blood from the bird’s brachial vein was collected and preserved in 100% ethanol (for sex diagnosis protocol, see Kolts and McRae, 2017).

Select king rails were fitted with a small VHF backpack transmitter (Advanced Telemetry Systems (ATS), model A1080, Isanti, MN). A leg-loop harness was made using a single strand of 1 mm diameter beading cord (Stretch-Magic Inc, Sonoma CA), a lightweight material providing expansion capability. The transmitter was positioned on the sacral region of the birds’ back with the antenna angled at 45° (Brewer et al., 2023). The ends of the Stretch-Magic cord were clamped together on either side of the transmitter with crimp tubes (Beadalon #4, 1.8 mm internal diameter, Valley Township, PA). Together the transmitter (3.9 g) and the harness (~0.4 g) weighed 4.3 g, well below the recommendation of <3% of the bird’s body weight (Fair et al., 2023). A previous study (Kolts and McRae, 2017) found that backback-mounted VHF transmitters did not deter king rails from breeding, a legitimate concern previously raised with respect to transmitters on birds (Barron et al., 2010).

Triangulation was used to obtain king rail locations. Bearings were taken from multiple geolocated positions, stations along the dikes surrounding the impoundments or from a boat on the river, using a handheld receiver (ATS model R410) and a three-element Yagi antenna. From each GPS position, a compass bearing was taken in the direction of the loudest signal heard at the lowest gain. A confidence level was recorded for each bearing reported, this corresponded with how wide the audible range of the signal was at the gain the bearing was recorded at (High = 0°–20°, Medium = 21°–60°, or Low >61°). For each bird, two to six fixes were obtained within a 30-min window to minimize error attributed to movement by the bird over time (Kolts and McRae, 2017). Coordinates for each bearing intersection were calculated using maximum likelihood with the software Location of a Signal (hereafter, LOAS; Ecological Software Solutions, LLC, v. 4.0).

During the breeding season, king rail positions were obtained 1–7 times per week (most often 5–7 times per week). Non-breeding season locations were determined at least twice per month. If the signal came consistently from the same location, wellness checks (walk-ins) were performed via direct tracking with a Telonics RA-24k VHF flexible H antenna for visual confirmation.

Nest searching was conducted during the breeding season within and around impoundments by technicians standing at arms-length apart combing through the habitat. Nests were backdated and an estimated hatch date was determined by floating the eggs to estimate the stage of egg development (Rush et al., 2007). Nests were monitored every three days, until the nest fate could be determined.

Motion-activated infrared professional trail cameras (Reconyx, Hyperfire 2) custom-focused to 1.8 m were placed at the edge of small clearings (mudflats or pools) in the marsh within territories. Cameras were used to resight banded and instrumented individuals to supplement our location data and movement analysis. These cameras were also used to track juvenile development and brood survival through the breeding season.

Habitat surveys were completed during the 2024 breeding season at points of king rail detection (telemetry locations), and at a paired random point 50 m away in the direction of a randomly generated bearing. Habitat surveys were conducted at telemetry locations separated by at least 20 m from other telemetry locations. For locations <20 m apart, the location with the highest confidence level reported by the observer, and in the point of intersection determined via LOAS, was selected for the habitat survey.

At each habitat survey plot, the distance to nearest habitat features (forest, river, rice field), percentage of land cover types, and general plot characteristics were recorded (Supplementary Table 1). A Robel pole (Robel et al., 1970) marked with 21 alternating 10 cm sections of white and black was used to record the height at which vegetation density obscured the pole at a distance of 5 m from the center of the plot (Brewer et al., 2023). If no sections of the Robel pole were visible, the maximum Robel pole reading of 21 was reported. This was repeated for all four cardinal directions. At the same time, water depth was recorded in each cardinal direction as well as at the center point, with depths >1 m recorded as 1 m (Krementz et al., 2016; Brewer et al., 2023; Schmidt, 2024). The number of stems, leaves, seed heads or flowers touching the Robel pole in each 10-cm section were reported, sections that were >50% occluded by water were reported as falling under water and stems in these sections were not counted. Sections were considered saturated if more than 10 stems touched the pole, truncating counts to a maximum of 10 stems per section. The number of stems reported for each section were summed such that a single stem count was recorded at each plot. Interspersion was defined by the number of edges where clumps of vegetation met either open water or exposed mudflat. The number of edges at the ground/water level that the 5 m length of paracord intersected was counted in each of the four cardinal directions and averaged.

2.3 Home range analysis

The breeding season for king rails at this study site extended from ~March 1 to August 15, estimated from back-dating the initiation of the earliest nest to the estimated date the last chicks observed would have become flighted. However, there was considerable variation among pairs in the initiation and conclusion of breeding activities. Therefore, dates for the breeding periods of marked adults were determined a posteriori. For breeding range calculation, the nest initiation date, or date of first telemetry point if the breeding period was already underway, marked the beginning of an individual’s breeding period. The conclusion of an individual’s breeding period was marked by late season nest failure, the observation of an individual in its end of breeding season flightless molt, the observation of the individual with flighted chicks, or by the estimated date the chicks would have become flighted based on the most recent observation (Meanley, 1969). For individuals where the end of the breeding period could not be determined precisely, the mean for tagged individuals with known end dates was imputed (June 23, n = 2).

A variogram was created for each individual based on locations determined throughout its breeding and non-breeding periods prior to calculating home ranges. Variograms measure the semi-variance between every point, against the length of time between them (Calabrese et al., 2016). When the semi-variance reaches an asymptote, it can be inferred that the home range has stabilized. Therefore, home ranges were produced only for datasets with variograms that reached an asymptote to ensure that individuals were tracked long enough to identify their full home range (Calabrese et al., 2016).

Area-corrected Autocorrelated Kernel Density Estimated (AKDEc) home ranges were mapped using the web application ctmmweb (version 0.2.10; Calabrese et al., 2021), for the R ctmm: Continuous-Time Movement Modelling package (version 1.2.0; Fleming and Calabrese, 2023). All possible movement models were analyzed, including those that added a diffusion factor to account for autocorrelation in the dataset, since treating autocorrelated data as independent and identically distributed can underestimate home range size (Fleming et al., 2015; Noonan et al., 2019; Calabrese et al., 2021). All models were ranked via AICc to ensure parsimony and account for small samples sizes (Fleming et al., 2015; Calabrese et al., 2016; Silva et al., 2022). The model that provided the lowest ΔAICc value was selected, and 95% AKDEc home range contours and 50% AKDEc core home range contours were populated. Home ranges were exported as shapefiles and mapped in ArcGIS (ESRI, Redlands, CA).

Model bandwidth optimizers adjust the kernel size around each point to reduce possible error between the actual distribution and the estimation due to directionality (anisotropy) in the bird’s movement resulting from boundaries like rivers or fences (Noonan et al., 2019). Applying best practices to further limit home range calculations outside of suitable habitat areas (Pickens and King, 2013), sections of home ranges that overlapped with landcover types deemed unsuitable for rails were removed. This included areas that overlapped with deep river channels, as well as houses and their immediate vicinity. The removal of sections of residential properties only affected the seasonal home ranges of two individuals (0.1 ha in each case, representing 7.6% and 4.3% of their 95% kernel densities and not affecting their core 50% kernel density home ranges). These areas were traced and converted into a polygon feature class in ArcGIS. The ArcGIS erase tool from the Geoprocessing Toolbox was used to remove the areas where these two polygons overlapped from the home range shapefiles. The remaining area was reported as the bird’s home range.

2.4 Movement analysis

Graphical representations of king rail movements were created using net squared displacement plots. These plots depict the squared distances between a bird’s first position and its successive positions over time. To determine distances traveled between consecutive fixes, step length and squared step length were calculated in R (version 4.3.1; R Core Team, 2023) package amt (version 0.2.2.0). Package ggplot2 (version 3.5.1) was used to graphically display the squared distances between points. In accordance with the amt package, telemetry points were reprojected to UTM using the R package sf (version 1.0.19) with the EPSG of 26917, for NAD83 UTM zone 17N.

Step length was used to compare the average distance traveled between breeding and non-breeding periods. We verified that differences in frequency of telemetry fixes would not significantly affect the comparisons by relating the distance data to date. With this concern alleviated, distance data were initially compared between breeding and non-breeding periods using the R package lmerTest (version 3.1.3) with bird identity and year as random effects. Neither random effect was significant, so a standard linear model function was used.

2.5 Habitat use analysis

Generalized linear models (family = binomial) were used to compare the habitat at king rail telemetry locations to random points. Since it was determined that adding a random factor for Bird ID would not significantly affect the analysis (SD = 0, variance = 0), a standard generalized linear model, glm function was selected from within the R package lme4 (version 1.1.35.5).

To prepare the data, variables with multiple readings taken throughout the plot were averaged (Density and Height of emergent vegetation, Interspersion, and Water depth), and then all variables were normalized using the scale function in R. Following a protocol from Butler et al. (2023), nonpredictive and correlated variables were removed from final analyses to reduce the number of input variables to avoid overfitting the model (Burnham and Anderson, 2002).

A few land cover categories calculated as percent total cover were combined based on habitat similarities: Marsh vegetation (summed all vegetation categories that can withstand inundation = Emergent reeds, Mixed herbaceous, and Floating aquatics), Floating mat vegetation (summed vegetation types comprising floating mats = Mixed herbaceous and Floating aquatics), and Woody (summed all woody vegetation = Forest and Shrub). All variables, including combined land cover variables, were individually compared by AICc to the intercept-only null model to determine the best predictors. Variables that performed better than the null model were then examined for collinearity via a Spearman’s rank correlation coefficient (ρ) using the cor function (method = “spearman”) in R. Variables with a |ρ| <0.5 were included in the final model; in scenarios where variables had a |ρ| >0.5, only the variable with the lower model AICc was included in the final model.

A generalized linear model for every possible combination of retained fixed effects added to the final model was conducted using the dredge function in R package MuMIn (version 1.48.4). Models were ranked by AICc and the top models (ΔAICc < 2), if more than one, were averaged using the R package AICcmodavg (version 2.3.3).

3 Results

A total of 27 individual king rails were captured over two years: six at Hasty Point, five at Pee Dee East, and 16 at Birdfield. In 2023, between March 6 and July 2, 18 king rails (16 male, two female) were captured. In 2024, between February 25 and June 9, nine adults (eight male and one female) were captured. Fourteen of the 27 captured king rails were fitted with VHF radio transmitters. In 2023, five king rails were instrumented late in the season between June 15 and July 2, two at Hasty Point and three at Birdfield. Between February 25 and June 9, 2024, nine adult king rails were captured and all were instrumented, two at Hasty Point, three at Pee Dee East and four at Birdfield. The mean weight of radio-tagged king rails was 407 g (range: 355–475 g), resulting in ~1% addition to an individual’s body weight when transmitter and harness was attached.

Although 14 individuals were radio-tagged, one was excluded from analysis because its transmitter fell off soon after deployment. The remaining 13 individuals were tracked for an average (± 1 SD) of 212 ± 203 days (range: 11–596 days) between June 17, 2023 and February 9, 2025 and a total of 660 telemetry locations were obtained. Trail cameras and direct observation of instrumented individuals added an additional 14 locations to the dataset. The mean number of locations per individual was 52 ± 50 (range: 6–160).

For three radio-tagged individuals, too few points were obtained to calculate a home range. One of these was depredated, likely by a bobcat captured on a trail camera at that spot shortly after the carcass was recovered. A second individual dropped the transmitter but was observed nesting near where it was recovered. The third individual’s harness was recovered from the marsh, and the status of the bird remained unknown. Home range sizes were determined for the remaining 10 individuals (Supplementary Table 2) based on a mean ( ± 1 SD) of 65 ± 50 locations per individual (range: 17–160) obtained over a mean time period of 324 ± 205 days per individual (range: 42–596).

Trail cameras captured king rails eating and feeding their young red-jointed fiddler crabs (Minuca minax) in tidal areas as well as red swamp crayfish (Procambarus clarkii) from within impoundments. Other wading birds such as ardeids were also observed feeding on red swamp crayfish within impoundments. Cameras placed at the edges of channels revealed that king rails used trails through the thick mats of primrose-willow (Ludwigia peploides), a floating invasive species, that reached heights of ~0.25–0.75 m over the channels. King rails were also observed standing and preening on platforms within this vegetation.

3.1 Seasonal home range sizes and range shifts

In total, nine breeding season home ranges were calculated, one in 2023 and eight in 2024 (Table 1; Supplementary Table 2). The mean ( ± 1 SD) area of breeding season 95% AKDEc was 4.6 ± 2.6 ha with a mean core 50% AKDEc of 1.1 ± 0.5 ha (Table 2). Seven non-breeding home ranges were also calculated, five in 2023 and three in 2024. The mean non-breeding season 95% AKDEc area was 4.0 ± 2.3 ha with a mean core 50% AKDEc of 1.0 ± 0.5 ha (Table 2). Accounting for bird identity, mean 95% AKDEc’s were not significantly different between breeding and non-breeding periods (GLMM F_1,13.7 = 0.17, p = 0.69). The largest non-breeding home range size belonged to the only hatch-year bird (151.099) captured in June 2023 (see Supplementary Figure 1 for an example of immature plumage compared to adult plumage).

Table 1. Breeding and non-breeding home range sizes. Seasonal 95% and 50% AKDEc home range sizes are reported in hectares (ha).

Confidence intervals are included in parentheses. All birds are labeled by signal frequency and sex, determined by molecular diagnostics. Seasonal ranges are labeled by year. For non-breeding periods, this was the year of initiation. If the non-breeding season spanned the subsequent year, the two-digit year is included on the end date. For most birds, the simplest model, assuming no autocorrelation with a diffusion factor of zero for independent identically distributed (IID) was selected. Some birds displayed directional movements accounting for anisotropic movements. For a few birds, Ornstein-Uhleneck (OU) or OU foraging (OUf) with independent position and velocity correlation models were selected.

Table 2. Mean seasonal home range sizes.

Seasonal ranges are reported as mean (± 1 SD) 95% AKDEc and mean (± 1 SD) 50% AKDEc separately among years because some individuals are represented in each category. There was only one breeding season home range in 2023, so the confidence interval is given parenthetically instead of the SD.

Five king rails (four male and one female) were observed making large home range shifts during the breeding season. These shifts occurred immediately following nest failure (n = 1), suspected brood loss (n = 1), successful conclusion of the brood-rearing period (n = 2), or the loss of nesting habitat in their home range as a result of management (n = 1). These home range shifts were recorded between May 26 and July 18. Notably, all four adult males that shifted their range were lighter on average than the two adult males that did not (Mean ± SE = 386 ± 14 g for those that underwent a range shift; Mean = 445 ± 28 g for those that did not; Mann-Whitney U = 0, p = 0.1). The males that did not shift their ranges, also experienced nest failure (n = 1) and brood success (n = 1). Specific cases for select individuals are discussed below (referenced by their unique transmitter frequencies), providing the timing and context for a selection of home range shifts.

Among the five individuals that made large home range shifts, 150.701 shifted its range at the end of each breeding season and had the smallest recorded breeding season home range (Figure 2a). Its breeding territory in 2023 was in the northern tip of the slender Hasty Point impoundment, with mowed lawns and houses on the north and west sides, and a dike separating the impoundment from the deep Great Pee Dee River to the east. Its core home range was dominated by primrose willow. Initially captured on May 8, 2023, it was recaptured and radio-tagged ~7 weeks later on June 25 just 205 m away while caring for a ~6-week-old chick. Observations of the individual rearing its single chick continued into late July when the bird’s position became less predictable. On July 20, 150.701 embarked on a ~6.5 km journey south along the Great Pee Dee River that would take ~8 days (Figure 2c). During its journey, the bird was observed foraging and preening in a fringe marsh with an immature rail, likely its now flight-capable offspring. With the offspring no longer dependent, the end of the adult’s breeding period coincided with the home range shift.

The non-breeding ranges of individuals that made range shifts tended to occur along the edges of a tidal river, in both fringe marsh and forested habitat. The non-breeding range for 150.701 was in a narrow strip of tidal emergent marsh dominated by giant cutgrass along the river’s edge (Figure 2b). It abutted an expanse of forested wetlands with remnant irrigation creeks running through it. In early April 2024, the bird traveled back northward but remained on the east side the river, occupying a different breeding season home range 6 times larger than the previous years’ and dominated by flooded forest habitat near the river’s edge (Figure 2a). By May 30, after the midpoint of the breeding season, the bird traveled back down to the same non-breeding home range it occupied the year before and remained there for the rest of the season (Figure 2b).

Two larger males concurrently occupied the small Hasty Point impoundment in the 2023 breeding season; however, too few points were collected to accurately map a breeding season home range for either, since the end of their breeding period occurred shortly after capture (Figure 2a). One male (151.000) was nesting at the time of capture. When its nest failed, this large male remained within the impoundment for the duration of the non-breeding season. The second male (150.801), at the time of capture, was rearing no fewer than five, approximately 8-week-old juveniles. Upon capture on July 2, 150.801 had lost a significant number of remiges as part of the post-breeding molt. This individual had been monitored visually on the breeding ground prior to capture and also remained at the south end of the impoundment during the non-breeding season.

Male 150.202 provided another example of a home range shift. When captured on March 16, 2024, this individual occupied the rice field at Pee Dee East. However, after management began, it avoided the impoundment where rice was planted, occupying the bank of the dike surrounding the impoundment, floating aquatics covering the channel, and tidal forested area dominated by cypress and tupelo. The first observation for this individual foraging in rice occurred on June 23, when the rice was approximately 0.75–1 m tall. It subsequently flew approximately 20 m across the channel accompanied by two immatures that were already capable of short flights. Two days later, 150.202 was detected via telemetry at Hasty Point, ~800 m across the river from its previous location. Shortly after, two immatures, likely its offspring, were spotted near the signal from 150.202’s transmitter. Thus, the family moved together from Pee Dee East across the Great Pee Dee River to Hasty Point Impoundment: the offspring were observed at Hasty Point Impoundment with the female near the telemetry signal from the radio-tagged male. Though the nest had not been monitored, we backdated its nesting stage based on offspring development to the month of March when it was captured. The estimated breeding periods and dates of management actions within rice fields were overlain on the individual’s nest squared displacement plot (Figure 3). This revealed that the nest likely hatched before the field was treated with herbicide and planted with rice. During the non-breeding period, 150.202 remained within the northern tip of the Hasty Point impoundment until late January 2025 when it returned to the vicinity of its previous breeding season territory at Pee Dee East (for positions of sites, see Figure 1). It was depredated shortly after.

Where tagged king rails’ breeding territories abutted one another, there were areas of overlap in the 95% AKDEc, but the core 50% AKDEc never overlapped between individuals (Figure 4). Five proximate home ranges at Birdfield during the 2024 breeding season illustrate that core 50% kernel densities did not overlap despite their 95% AKDEc overlapping (Figure 4). Exceptionally, locations from a single individual, 150.302, whose home range did not stabilize, represent the only example of points falling within another breeding king rail’s 50% AKDEc (150.281) (Figure 4). However, these points were recorded a month prior to 150.281’s occupancy of this corner. Therefore, the points did not overlap temporally.

3.2 Movement analysis

We investigated whether king rails moved greater distances when they were not breeding compared to when they were. In spite of the frequency of fixes being reduced in the winter due to logistical constraints, date of telemetry had no significant effect on the distances moved (F_1,1186 = 0.296, p = 0.586, n = 558). Nor was there an effect of month of telemetry, pooling years, on the distances moved (F_1,831 = 0.208, p = 0.649). We compared distances traveled between consecutive telemetry locations for breeding and non-breeding king rails (Summary of step lengths in Supplementary Table 3). We initially included random factors of Bird identity and Year, but although the effect of season remained significant, neither random factor was (Wald’s p = 0.555 and 0.709, respectively). Distance traveled between fixes was significantly greater on average for breeding king rails (Mean ± SD = 85 ± 36 m, n = 247) than for non-breeding king rails (63 ± 12 m, N = 311; F_1,556 = 12.80, p = 0.0005).

3.3 Habitat use during the breeding season

Of a total of 358 king rail telemetry locations recorded during the breeding season from February 25 to July 22, 2024, habitat surveys were conducted at 130 (36%) of the locations for nine individual rails. Habitat surveys were completed between May 16 and August 1, 2024, at points where birds were located and an equivalent number of random points (260 habitat surveys for 130 bird locations).

The AICc value for a generalized linear model conducted individually for each variable was compared to the AICc value of the null model (intercept only). The six variables that performed better than the null model were retained for further analysis (Table 3). Retained variables were distance to channel (m), stem count, as well as percent land cover of herbicide treated vegetation, marsh vegetation, floating mat vegetation, and open water/mudflats.

Table 3. Habitat parameters retained based on univariate generalized linear models that performed equivalently or better than the null model, ranked by AICc (ΔAICc > 2 denoted by asterisks).

The analysis compared 130 points with king rail locations to a matched sample of random points.

Only two retained variables were found to be significantly correlated based on Spearman’s rank correlation coefficients, percent land cover of open water/mudflats and marsh vegetation (|ρ| = 0.74; Figure 5). Of the two, percent marsh vegetation was retained as it had a lower AICc value.

Percent marsh vegetation was a combined variable that included percent floating mat vegetation. Therefore, due to its lower AICc value, the former was retained in the model and the latter excluded.

A final model was created with the remaining four variables: distance to channel, stem count, percent herbicide treated vegetation, and percent marsh vegetation (Table 4). The AICc value of the 16 possible models using every combination of these variables was calculated. Comparing the AICc between all models, the removal of any variable(s) resulted in an increase in AICc >2. Therefore, the top model containing all four variables was used to explain king rail habitat use. Estimates for each variable in the top model were plotted with 95% confidence intervals on a forest plot (Figure 6). Confidence intervals for each variable did not overlap with zero, indicating a significant effect on king rail habitat use.

Table 4. Significant variables characterizing king rail habitat use, as determined by generalized linear modeling.

Variables are displayed with their estimate (± SE) and p-value. King rails were significantly more likely to be in marsh vegetation, and in areas where the percentage of herbicide treated vegetation was greater, closer to channels, and where summed stem counts were lower.

King rail home ranges frequently spanned the dredged channels that circumscribed impoundments (Figure 4). These contained abundant prey that was easily accessible to king rails from the floating mats of vegetation that covered them: species such as invasive primrose willow provided dense cover and refuge in winter when impoundments were flooded for waterfowl. An abundance of historic irrigation channels within impoundments as well as remnant irrigation channels in natural marshes pervaded the wetlands in this region, and king rails foraged for prey along their shallow-sloped edges allowing access at ideal water depths while remaining close to emergent cover. The adjacency of these habitat features appeared beneficial to rails and emphasizes why distance to channel was a key predictor of king rail locations.

During the breeding season, king rails at Waccamaw NWR were more likely to be found in areas with a higher percentage of wetland vegetation. Dominant emergent plant species capable of withstanding periods of inundation in this region included giant cutgrass, smartweeds (Polygonum spp.), sedges (Carex spp.), coffeeweed (Sesbania herbacea), and primrose-willow. Robel pole stem count measurements revealed that king rails selected areas with moderate densities of emergents. The highest stem counts were associated with marsh vegetation that was impenetrable.

Herbicide treatments generally occurred before and in association with rice planting in the impoundments. Their positive association with king rail habitat use does not imply that birds sought treated vegetation, only that they did not abandon their breeding territories after herbicide application. Typically, the rails would be found at the edge of treated impoundments where tufts of vegetated cover were left near the channels. King rails avoided impoundments after tilling and planting, when vegetation was temporarily removed completely.

4 Discussion

We tracked 13 king rails over 2 years and calculated nine breeding and eight non-breeding home ranges. Though variance was high among individuals, we found no significant difference in mean home range size between seasons. We report some of the smallest seasonal home ranges for king rails, likely influenced by high prey densities and a concentration of breeders in relatively small impoundments. Over half of the individuals used disparate breeding and non-breeding home ranges. These were determined to be the lighter males and females, whereas the heaviest males remained on their breeding territories throughout the winter. King rails traveled slightly greater distances on average between locations when breeding than when not breeding. King rails used microhabitat closer to channels in areas with higher percentages of marsh vegetation.

4.1 Home range size

Breeding home range sizes at Waccamaw NWR fell at the lower end of the scale of this study and were smaller on average than breeding season home range sizes in South Carolina (McGregor et al., 2009), North Carolina (Kolts and McRae, 2017) and along the shores of Lake Erie (Brewer et al., 2023). Waccamaw impoundments were small in area compared to sites in these other studies. Hasty Point Impoundment in particular was only 13 ha and supported three breeding territories in 2023. However, crustacean prey were abundant and accessible in these wetlands in the breeding season (S. McRae, pers. obs.), likely contributing to the high density of territories in the impoundments. A previous study in coastal Texas found that king rail breeding home range sizes in managed areas were highly variable. Smaller home ranges, most similar to the mean home range size at Waccamaw NWR, were found in high quality habitat having a high proportion of open water for foraging, whereas large home ranges were reported for a small number of individuals navigating poorer habitat, some of which moved between two sites (Pickens and King, 2013). Prior to our study, only a small number of king rail home ranges had been described along tidal rivers in the South Carolina Lowcountry. In the ACE Basin NWR, ~150 km south and closer to the mouth of the river, three king rail breeding home range sizes reported in Appendix A of Ricketts (2011) were similar in size to what we observed (0.7, 7.0 and 9.9 ha, the smallest belonging to a male and the largest to a female). A king rail occupying a series of oligohaline impoundments (4–5 ppt, within king rail level of tolerance for salinity) was found to have the largest home range for a king rail reported in the literature (overall 95% kernel density = 31 ha, but with a core range of 10 ha in one impoundment between May and July). Though classified as ‘adult’ at capture (McGregor, 2007), it was suspected of possibly having been a non-breeder or first-year bird (McGregor et al., 2009). The largest home range we observed at Waccamaw NWR was that of a first-year male in its first winter, suggesting that, when possible, age should be considered in interpreting home range size data.

4.2 King rail movements and home range shifts

Five individuals (one female and four male) of nine for which we had telemetry locations in both the breeding and non-breeding periods, shifted their seasonal home ranges an average of 2.9 km (range: 0.7–7.5 km). Distinct breeding and non-breeding ranges were also observed in female king rails (n = 4) at Mackay Island NWR that overwintered in wooded wetlands (Kolts and McRae, 2017). It was suggested that larger males were able to remain in the same ranges on high quality marsh habitat between seasons, but reduced seasonal food availability could not support additional adults, thus smaller females retreated to more marginal wooded habitat. Our observations at Waccamaw NWR were consistent with this interpretation: breeders with separate non-breeding ranges were lighter than those that remained on their breeding ranges through the winter. King rails that occupied distinct breeding and non-breeding home ranges at Waccamaw NWR used fringe marsh and flooded forest habitats along the tidal Great Pee Dee River in the non-breeding season.

Few other studies have tracked king rails throughout both breeding and non-breeding periods. However, Ricketts (2011) observed a radio-tagged female king rail in the ACE Basin shift its location after nest predation during hatching, to a fringe marsh 900 m away where it remained for the duration of tracking (approximately 3 weeks). Her telemetry studies of eastern clapper rails (Rallus crepitans, a saltmarsh specialist) also included observations of seasonal separation in home ranges. Close to the outlet of the tidally influenced Combahee River in South Carolina, where salinity levels are higher, Ricketts (2011) found pre-breeding home range shifts of four clapper rails captured between February and mid-March. Compared to king rail home range sizes, those of clapper rails were on average small (~1 ha; Ricketts et al., 2020), likely in part due to higher densities. In expansive tidal saltmarshes along the coast of Delaware Bay, only one of 23 tracked clapper rails shifted to a new distinct home range following nest predation in late July (Elizondo, 2023). Thus, in contrast to king rails, home range shifts appear less common in this species, possibly due to the saltmarsh habitat where they were studied being less fragmented.

Similar to other avian tracking studies which reported smaller home ranges for larger individuals, king rails with separate breeding and non-breeding ranges covered greater distances than heavier individuals that remained in the same area year-round. In yellow warblers (Setophaga petechia) overwintering in Mexico, larger males had smaller ranges and females had ranges similar in size to small males (Valdez Juarez et al., 2024). Based on a year-round VHF radiotelemetry study in the Alps of Switzerland and Austria of white-backed woodpeckers (Dendrocopos leucotos), seasonal home range size decreased with increasing body weight and was larger in females (Ettwein et al., 2025). Woodpecker ranges also contracted in the two months before nesting, suggesting an effect of increased territoriality but also potentially greater energetic expenditure in obtaining essential breeding resources.

We considered an alternative explanation for the finding that lighter king rails shift their home ranges: perhaps the energy expenditure required for traveling between sites keeps these individuals more lean. Unfortunately, this explanation does not generate mutually exclusive predictions from the explanation that individuals must expand their non-breeding home ranges to access reduced winter resources and that larger breeders with higher status are more likely to remain and hold their territory. Thus, we are unable to exclude either one.

While range shifts are potentially influenced by population density and rank status of individuals (body size differences), the availability of resources is likely an important factor (Jedlikowski et al., 2016). Food availability is highly influenced by tidal regime and water management in impoundments. At Waccamaw NWR, wetlands along the river hosted abundant red-jointed fiddler crabs and other invertebrate prey that became exposed on mudflats predictably during low tide. The impoundments hosted red swamp crayfish, a preferred food, as well as minnows and other prey. During the breeding season, king rails were frequently observed traversing the dikes separating the tidal wetlands and impoundments. At Mackay Island NWR, king rails also regularly moved between nesting areas in emergent natural marsh and foraging areas in exposed mudflats of managed impoundments (Kolts and McRae, 2017). After impoundments were flooded for overwintering waterfowl, king rails at Waccamaw NWR that did not shift their home ranges also foraged in wooded wetlands adjacent to the impoundments. This suggests that the proximity of these habitat types reduces the necessity for more dominant individuals to shift home ranges during the non-breeding season. The long distances traveled by lighter individuals that shifted their home ranges at this site could be a function of the degree of fragmentation of the landscape. Much of the habitat along the river in between wetlands was unsuitable: upland, residential property or deeply flooded private impoundments.

Home range shifts involved periodic large movements across the landscape. However, king rails traveled much smaller distances as they foraged and moved about their territories. It was somewhat unexpected that king rails moved farther between fixes on average when breeding than when not, but note that the average difference was only about 22 m, and the variances were large. We can only speculate as to the direction of significance, but it likely had to do with the distribution of resources and cover, and the fact that they are interacting more with conspecifics when breeding. When food and cover abound, breeding king rails are free to move around their territories as they engage socially with others. Both king rails (Kolts and McRae, 2017) and clapper rails (Elizondo, 2023) moved more when brood-rearing than when nesting. However, during the non-breeding season, especially in winter months when cover is greatly reduced, it may be safer to find and become familiar with a reliable, protected foraging patch away from others and not stray far from it. Schmit (2024) monitored king rails in a resident population in Arkansas with GPS transmitters, and found average distances traveled per month to be greatest in July and August compared with early in the breeding period and at other times of year, though there was high variance within and between years. Few others have compared breeding and non-breeding movements of resident rails, and further studies are needed.

4.3 Habitat use among fragmented managed wetlands

During the breeding season, king rails in this fragmented landscape were more likely to be found in areas with a higher percentage of wetland vegetation but with moderate stem densities. Across their range, breeding king rails most often occupy emergent herbaceous wetlands with less woody vegetation (Darrah and Krementz, 2011; Bolenbaugh et al., 2012; Brewer et al., 2023; Gause and McRae, 2025). Prescribed fire is an effective way to achieve moderate vegetation densities and curb woody encroachment (Gause and McRae, 2025).

King rails foraged in the dredged channels surrounding impoundments as well as along remnant irrigation channels in tidal natural marshes. Use of these different microhabitats reflects the positive influence of a high degree of interspersion in both managed and unmanaged wetlands as previous studies have found (Meanley, 1969; Eddleman et al., 1988; Rehm and Baldassarre, 2007; Bolenbaugh et al., 2012; Kolts and McRae, 2017; Brewer et al., 2023). Though the fine measure of interspersion we employed during vegetation sampling (the number of intersections of mudflats/open water with emergent vegetation within 5 m) did not significantly predict king rail presence, measuring habitat features at broader spatial scales has been found to be more likely to predict habitat selection in rails (Jedlikowski et al., 2016). The juxtaposition of impoundments with tidal creeks ensured accessible forage with appropriate water levels on diel and seasonal timescales.

We found that floating aquatic vegetation over the channels provided dense cover for foraging king rails. At Savannah NWR, SC, small area (<2 m²) feeding platforms on floating mats used by king rails suggested a high abundance of forage is available (Meanley, 1969). Floating mats increase the total surface area of the channels available for foraging, and under the weight of a king rail, likely give just enough to allow them to forage at their preferred range of water depths of about 2–10 cm (Meanley, 1969; Brewer et al., 2023). In this study, smaller shallow irrigation channels with tall vegetation on the sides provided cover for king rails. Pierluissi and King (2008) found king rails to prefer the shrubby edges around open water for the same reason. Irrigation channels allow king rails to forage near the edges while reducing predation risk (Ricketts et al., 2020).

Floating mats of vegetation mimic high marsh habitats that are not subject to regular inundation (Adams and Quay, 1958). These provide critical habitat during the non-breeding season when impoundments are flooded for migrating waterfowl. Though it is difficult to observe king rails on these unstable floating mats when they extend far from the raised dikes, drones equipped with infrared have been employed successfully to detect king and clapper rails in Texas (Olsen et al., 2023) and could be used to enhance our understanding of their use.

4.4 King rail navigation of managed wetlands

Along the Great Pee Dee River, many of the historical rice impoundments have reverted to natural wetlands in various stages of ecological succession. Flooded forests and woody swamps are prevalent, with sparsely distributed actively managed impoundments among them. King rails used these wooded wetlands, especially during the non-breeding season. The canopy in forested habitat wetlands offers protection from the elements and from aerial predators (Kolts and McRae, 2017).

Expanses of flooded emergent vegetation within restored impoundments at Waccamaw most resembled the structure of natural coastal marshes where king rails have previously been studied. However, this condition is ephemeral due to the agricultural practices involved in cultivating rice. King rails in our study adjusted their movements during rice cultivation, avoiding impoundments when rendered devoid of vegetation, and returning when the rice was full grown. At two of our three study sites, managers left small strips of vegetation along the edges of the irrigation channels. This provided critical cover for foraging king rails that had established territories prior to the management actions.

4.5 Wetland management recommendations to support king rail conservation

With most refuge system impoundments being managed for wintering waterfowl, understanding how to manage simultaneously for breeding threatened marsh birds is a conservation challenge. Rice cultivation in the south is compatible with breeding king rails, but care should be taken to time major land cover changes to minimize impacts on nesting. When clearing fields for rice, leaving unmanipulated strips of emergent vegetation along the channels and irrigation ditches within rice fields provides critical cover and nesting substrate for king rails. Complete removal of vegetation along a channel’s edge may render the habitat unsuitable. Dormant season burns present an alternative to herbicide treatments for thinning emergent vegetation. Floating vegetation over channels likely increases the total foraging area while also providing cover. King rails used unmanaged tidal wooded wetlands adjacent to impoundments and regularly moved between these, highlighting the importance of the proximity of these habitats for rails and other wading birds. This is particularly important in landscapes where lunar tides alter water levels on a predictable diurnal schedule.

5 Synthesizing findings from telemetry studies of king rails range-wide

King rails are difficult to observe and the severity of their population declines over the last five decades was only recognized when the U.S. Fish and Wildlife Service organized a working group tasked with creating a conservation plan (Cooper, 2008). This motivated studies across the species range to address knowledge gaps, including attempts to understand habitat use and home range size using radiotelemetry (McGregor et al., 2009; Pickens and King, 2013; Kolts and McRae, 2017; Kane et al., 2019; Brewer et al., 2023; this study). Due to the difficulty of capturing and working with a secretive species in challenging terrain, each study tracked a relatively small number of individuals. Yet, when viewed collectively, broader trends emerge. Although they have not always been identified as seasonal home range shifts, king rail movements consistent with those reported here have been reported in multiple resident populations (Ricketts, 2011; Pickens and King, 2013; Kolts and McRae, 2017; this study). These are typically by lighter males and females and result in disparate home ranges covering a larger overall area. Breeding habitat is consistently associated with dense marsh vegetation, though the species of emergent vegetation vary across the range. King rail occupancy is also associated with a high degree of interspersion of vegetative cover with shallow water or mudflats for foraging.

It is noteworthy that each of these studies was centered on public lands that are part of the National Wildlife Refuge system overseen by the U.S. Fish and Wildlife Service. In part, this was to benefit from in kind support for projects on priority species. However, it also highlights the fact that much of the current king rail distribution is on managed public lands. Wildlife refuges are among the only remaining large tracts of wetland that support rails and other secretive wading birds, and the management they provide is critical to species persistence (Gause and McRae, 2025). This has become increasingly urgent recognizing that drastic budget reductions and significant personnel losses have recently occurred in the U.S. Department of the Interior.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcosc.2026.1894344/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Grzegorz Kopij, Wroclaw University of Environmental and Life Sciences, Poland

Reviewed by: Sarah Heidmann, University of the Virgin Islands, US Virgin Islands

Glenn Goodwin, Nova Southeastern University, United States