Resource Use Overlap by Sympatric Wintering American Black Ducks and Mallards in Tennessee

An underlying reason for studying resource use and potential partitioning among species is to understand the influences of intra- and inter-specific competition and how the extent of resource partitioning may influence the number of species that can coexist in shared space (i.e., sympatry; Hutchinson 1958; Schoener 1974; Nudds and Kaminski 1984; Gurd 2008). Sympatry has ecological ramifications potentially manifested through exclusion, avoidance, territoriality, or possible extinctions of less adaptive species (Arlettaz 1999; Steen et al. 2014; Estevo et al. 2017). Nonetheless, mechanisms that decrease niche overlap in habitat, diet, and temporal space use (MacArthur 1958; Mahendiran 2016) usually enable persistence of similar sympatric species (Hardin 1960; Schoener 1974; Estevo et al. 2017). Examining such mechanisms may also elucidate ability of a species to range-shift from historical locations amid changes in habitat quality and abundance (Perry and Deller 1996), presence of competing conspecifics and congeners (Gurd 2006), or changes in climate at greater spatial scales (Schummer et al. 2017; Meehan et al. 2021).

Resource partitioning among the 13 species of North American dabbling ducks (Anatini) often occurs spatially within or among habitats or associated intrinsic resources (e.g., in or among wetland complexes; DuBowy 1988; Nudds 1992; Nudds et al. 2000). For example, water depths and vegetation composition or structure can influence species distribution in wetlands given dietary preferences (e.g., herbivorous gadwall Mareca strepera vs. omnivorous American green-winged teal Anas crecca, or mallards Anas platyrhynchos; Miller 1975; Gross et al. 2020). Interspecific and intersexual competition is expected to be most intense between morphologically similar congeners and conspecific sexes (e.g., Nudds and Kaminski 1984; Graves and Gotelli 1993; Nudds and Wickett 1994). Although observed patterns in species' co-occurrence or resource use alone do not necessarily refute a lack of competition or resource partitioning (Schluter 1984; Gurd 2008), morphologically and behaviorally similar species predictably may segregate in some way in time or space to avoid competition for resources (Hardin 1960).

In North America, two populations of American black ducks Anas rubripes (hereafter, black duck), are recognized—an eastern population in the Atlantic flyway and a midcontinent population that inhabits parts of the Mississippi flyway (Baldassarre 2014; USFWS 2019). Both populations experienced steep declines from the 1950s to the 1990s (Devers and Collins 2011; USFWS 2019). The U.S. Fish and Wildlife Service surveys breeding black ducks and other species in the Atlantic flyway, with recent surveys indicating that breeding black ducks in this region were ∼20% below the long-term average from 1998 to 2018 (USFWS 2018, 2019). Estimates do not exist for breeding grounds in the Mississippi flyway, but estimates of black ducks on wintering grounds in the Atlantic and Mississippi flyways also show continued declines (USFWS 2015, 2019), coincident with apparent shifts of this species' range to more northern latitudes (i.e., fewer black ducks are found in Alabama and Mississippi; Meehan et al. 2021). An average of 21% of black ducks in the Mississippi flyway have historically wintered at Tennessee National Wildlife Refuge (TNWR) in western Tennessee (USFWS 2019, 2020). However, the number of black ducks at TNWR has declined 98% between 1990 and 2019 (35,200 to 807; Fronczak 2020).

Pre–20th century mallards predominantly occurred west of the Appalachian Mountains, in the Mississippi, Central, and Pacific flyways (Heusmann 1974). However, during the 20th century, mallards dispersed into the range of black ducks, and by the late 1960s, there were more mallards than black ducks in the Atlantic flyway (Ankney et al. 1987; Mank et al. 2004; Baldassarre 2014). Several theories exist regarding range expansion by mallards into historical ranges of black ducks: 1) forest loss and increase in agricultural landscapes (Heusmann 1974; Ankney et al. 1987; Petrie et al. 2012), 2) competitive exclusion (Petrie et al. 2012), and 3) hybridization between the species (Dwyer and Baldassarre 1994; Morton 1998; Petrie et al. 2012). While mallards and black ducks are genetically related, and moderate levels of hybridization occur (∼25%; Lavretsky et al. 2019), these species may have strategies to limit hybridization (e.g., assortative mating, hybrid breakdown; Lavretsky et al. 2020). It is not clear whether competitive exclusion may be contributing to black duck declines on traditional wintering areas, concomitant with potential climate effects on shifting distributions of these birds from the lower Mississippi flyway to the north and east (Brook et al. 2009; Devers and Collins 2011; Meehan et al. 2021).

While information exists on black duck space use of historical wintering grounds, such as the TNWR (Newcomb et al. 2016; Monroe et al. 2021), little is known on how black ducks and mallards co-exist, or potentially partition the use of different landcover types, on wintering areas in the Mississippi flyway; this is information that could provide insight into factors that might exacerbate wintering black duck declines. Monroe et al. (2021) reported that black ducks used all discernible landcover types within the TNWR study area during winter, but particularly emergent wetlands. We examined space and resource use of black ducks and mallards at TNWR to investigate the degree of resource partitioning between these species. Black ducks and mallards have similar morphology and ecology, so we predicted that they would exhibit little partitioning 1) in space use and 2) in croplands where both species forage on grains (Jorde et al. 1983; Monroe et al. 2021). However, a lesser amount of overlap 3) would occur in other landcover types, such as palustrine wetlands (i.e., forested or emergent wetlands) given black duck affinity for these landcover types (Monroe et al. 2021). Knowledge of resource exploitation could yield important guidance for future landcover conservation and management for black ducks and mallards on the TNWR and other similar wintering refuges in the Mississippi flyway.

We collected data primarily in and around the Duck River Unit (DRU) of TNWR in western Tennessee (35°57′30″N, 87°57′00″W; Figure 1) in winter 2011–2012. The DRU was the largest (10,820 ha) of 3 wetland complexes comprising TNWR (20,784 ha) and generally had the greatest abundances of black ducks and mallards on TNWR (T. Littrell, TNWR, unpublished data). The DRU and the adjacent lands comprised uplands dominated by oak Quercus spp. and hickory Carya spp.; scrub-shrub wetlands; hardwood bottomlands; corn and soybean croplands; seasonally flooded, emergent herbaceous (i.e., moist-soil) wetlands; and open water, which included parts of Kentucky Reservoir and the Duck River (Newcomb et al. 2016; Monroe et al. 2021). No waterfowl hunting was permitted on the DRU but surrounding private and public lands were hunted. The study region receives an average of 380 mm of precipitation during winter months (December–February) and temperature averaged 4.4°C (monthly average range: 3.1–5.1°C; taken from climate data; NOAA 2020). In winter 2011–2012, our study area received 305 mm of winter precipitation and mean daily temperature was 6.3°C (mean daily temperature range: 0.2–12.6°C; Newcomb 2014).

Capture and radiotracking procedures for ducks in this study were published previously (Newcomb et al. 2016; Monroe et al. 2021). Briefly, we trapped female black ducks (ABDU; n = 49) and mallards (MALL; n = 17) from November 2011 through early February 2012 at the DRU using baited rocket-nets and swim-in traps. We banded all ducks with U.S. Geological Survey standard aluminum tarsus bands, and aged birds by wing plumage characteristics (Carney 1992; Ashley et al. 2006). We then weighed females and females ≥ 800 g were fitted with a 23-g (∼3% of body mass) harness-type very-high-frequency transmitter (Model A1820; Advanced Telemetry Systems, Isanti, MN). We only radiotagged pure black ducks or hybrids that were considered black duck dominant based on morphological characteristics (Table S1, Supplemental Material; Devers et al. 2021). Once captured ducks were processed, we left them undisturbed in covered crates for 1 h before they were released back to their original capture location. We then commenced data collection on the third day postrelease to avoid resource use bias associated with acclimation to the transmitter (Dugger et al. 1994). Our capture and handling protocols were approved by the Institutional Animal Care and Use Committee at Mississippi State University (#10-070).

From November 2011 to April 2012, we determined daily locations of radiomarked females (Data S1, Supplemental Material) once diurnally (i.e., dawn-to-dusk) and once nocturnally (i.e., dusk-to-dawn) within a 24-h cycle, 6 d/wk for 87 d during the above period. However, if there were too many ducks to track within a 12-h period or other logistical constraints limited tracking capabilities, we randomly chose a subsample (∼20 ducks) for tracking. The same randomly selected ducks were tracked during both diurnal and nocturnal periods, and remaining ducks were tracked within 1–3 d. We recorded duck locations using vehicles with roof-mounted, 4-element, null-peak antenna systems (Advanced Telemetry Systems), electronic compasses (Azimuth 1000R; KVH Industries, Middletown, RI), and Location of a Signal software (LOAS 4.0.3.8; Ecological Software Solutions, Hegymagas, Hungary; Cox et al. 2002; Davis and Afton 2010). We obtained at least three compass bearings to estimate locations and 95% confidence ellipses in LOAS; additional bearings were obtained until confidence ellipses were within one landcover type (see landcover map, Figure 1; Davis et al. 2009). Additionally, we conducted two aerial surveys to locate radiomarked ducks not detected via ground tracking and continued to monitor them with ground tracking after they were located.

We created a landcover map with ERDAS Imagine 2010 (ERDAS, Norcross, Georgia) using supervised classification on contrasting seasonal Landsat-5 Thematic Mapper (TM) images with 30-m resolution and < 10% cloud cover (images from 16 March 2011) obtained from the U.S. Geological Survey Earth Resources Observation and Science Center data archives (http://glovis.usgs.gov/; Monroe et al. 2021). We classified landcover as open water, forested wetland, cropland, emergent or scrub-shrub wetland, human developed areas, upland hardwood, and pine (Pinus spp.; Monroe et al. 2021). We were unable to distinguish between emergent and scrub-shrub wetlands as images did not discern relative height and structural differences of vegetation types (hereafter, emergent wetlands; Monroe et al. 2021). To assess accuracy of our landcover classification, we determined landcover type for 70 random points within each of 7 landcover classes (n = 488 useable locations). In addition to on-the-ground knowledge, we classified our random points using Google Earth®, National Wetlands Inventory data, 2010 National Agriculture Imagery Program digital orthophotos, and Geographic Information System (GIS) layers created by refuge biologists (Bahadur 2009; Aguirre-Gutiérrez et al. 2012; Churches et al. 2014; Rwanga and Ndambuki 2017; Lossou et al. 2019). We assessed accuracy using the Kappa statistic (κ) whereby comparison between the produced classified map and random locations indicates the level of classification error (Lillesand et al. 2008; Monroe et al. 2021). Classes with κ > 0.8 are deemed very accurate, and ranges 0.4–0.8 were also deemed acceptable (Pope et al. 2005; Zohmann et al. 2013). Our overall classification accuracy was κ = 0.88, and we achieved excellent to moderate accuracy for all classes: open water, upland hardwood, and pine (κ = 1), forested wetland (κ = 0.97), cropland (κ = 0.98), emergent wetlands (κ = 0.63), and human developed (κ = 0.64).

We created total home ranges defined as the 90th isopleth (Börger et al. 2006) for each individual black duck and mallard, using biased random bridges (R package adehabtatHR; Calenge 2006), which accounted for clustered vs. long-distance movements of individuals (Benhamou 2011). For each individual, we first created home range estimates with an increasing number of locations to obtain the point whereby an increase in the number of locations resulted in limited changes in home range size (i.e., home ranges were stable), and found that home range size increased, on average, < 7% as a result of additional location information of > 20 locations (Figure S1, Supplemental Material). As a result, we removed individuals that had < 20 locations (6 black ducks, 4 mallards) and subsequently used all location data for each individual to create total home ranges; this resulted in the creation of home ranges for 43 black ducks and 13 mallards (Table S1, Supplemental Material). Preliminary analysis indicated that black ducks and possible black duck–mallard hybrids behaved similarly (Table S2, Supplemental Material), so we pooled them in all analyses. To understand whether sampled mallards were using the same space as sampled black ducks (i.e., to examine spatial segregation of home ranges), we subsequently merged all individual home range boundaries together to get the total area used for each species and compared the total area overlap using the intersection function in ArcGIS (ESRI 2019).

To examine differences in resource selection, we used resource selection functions (Boyce et al. 2002) that compared landcover characteristics at observed and random locations. We created 10 random locations for every used location for a given duck with random locations distributed within each individuals' home range. We then extracted the landcover type at each used and random location. Using a binomial logistic regression model, which used presence–absence information as the response variable (i.e., 0 and 1, random and used locations), we first examined potential temporal partitioning using species × month (January–March) × landcover type, and species × period of day (day vs. night) × landcover type interactions as our predictor variables. We did not detect any resource selection differences, except for corn fields, for either the species × month × landcover type interactions (z = 0.36–14.4, P > 0.32) nor the species × period × landcover type interactions (z = −0.33–14.2, P > 0.44); therefore, we pooled all data into a binomial logistic regression model that contained a species × landcover type interaction as our predictor variables to explore broader patterns in resource selection between species. We examined availability of landcover types using our random sampling of home ranges, and the results of selection models using beta-coefficients and selection ratios (i.e., the ratio of used to available locations in each landcover type). Given that individuals may avoid a landcover type relative to that landcover's availability but may spend a moderate proportion of time within avoided landcover types, we also calculated proportional use of each landcover type as the number of used locations in each landcover divided by the total number of locations for each animal. Using a linear model, we then examined proportional use (our response variable) in relation to a species × landcover type interaction (predictors). We square-root-transformed proportional use to meet the assumption of normality, which is appropriate to do when numbers are between 0 and 1 (Osborn 2002).

Black duck and mallard home ranges were similar in area, with black duck home ranges ranging from 15 to 77 km² and mallard home ranges ranging from 21 to 72 km² in size (Table S1, Supplemental Material). Black ducks and mallards used a total of 288 and 182 km², respectively, and the total area used by black ducks and mallards overlapped 93% (Figure 2). Indeed, only portions of home ranges for three black ducks were not encompassed by mallard home ranges, indicating great overlap despite the disparity in sampling of each species. Random sampling across black duck and mallard home ranges revealed they contained similar landcover availability, with open water being the most available landcover type (ABDU, 29%; MALL, 25%) and cultivated lands being the second most available landcover type (ABDU, 20%; MALL, 21%) for both species. Forested wetlands (ABDU, 15%; MALL, 18%), emergent wetlands (ABDU, 13%; MALL, 10%), and upland hardwoods (ABDU, 16%; MALL, 18%) comprised most of the remaining available landcover types. For both species, human development (< 6% availability) and pine forests (< 3% availability) were scarce.

Overall, emergent wetlands followed by cultivated lands and forested wetlands were selected by both species, while all other landcover types were avoided to varying extents (Table 1; Figure 3a). There were differences in selection between species, with black ducks having a selection ratio 1.2× greater than mallards for emergent wetlands, while mallards had a selection ratio twice that of black ducks for cultivated lands (Table 1; Figure 3a). Both species avoided open water, but mallards had a selection ratio half that of black ducks (Figure 3a). Both species also avoided upland hardwoods, human developed land, and pine forest to a similar degree (Table 1; Figure 3a). There were minor differences in proportional use between species, with mallards having lesser use of permanent water and greater use of cultivated lands, upland hardwood, and development, compared with black ducks (Table 2; Figure 3b). For both species, proportional use also differed across landcover types (Table 2; Figure 3b). Open water was used the most compared with all other landcover types (Figure 3b). Proportional use was second greatest in cultivated lands, followed by forested wetlands and then emergent wetlands and upland hardwoods (Table 2; Figure 3b). Proportional use was least in development and pine for both species (Table 2; Figure 3b).

Most locations for both species occurred in similar proportions throughout landcover types, although proportions varied for open water, emergent wetlands, and cultivated lands (Table 3). Black ducks had moderate niche breadths (B_A = 0.47), while mallards had narrower niche breadths (B_A = 0.34). Each mallard–mallard and black duck–black duck pair of individuals had a large degree of intraspecific resource overlap (mean O = 0.63 ± 0.15 [SD], 0.67 ± 0.14, respectively). Additionally, each mallard–black duck pair of individuals had a large degree of interspecific overlap (O = 0.61 ± 0.17). Our analysis identified three principal components, which captured 71% of the variation of black duck and mallard use among seven landcover categories (32, 22, and 17%, respectively). The first component was positively correlated with cultivated lands (r = 0.82). The second component was positively associated with upland hardwoods (r = 0.54), development (r = 0.59), and open water (r = 0.54). The third component was positively correlated with emergent wetlands (r = 0.70) and negatively associated with forested wetlands (r = −0.66). Despite distinct landcover components, there was no detectable clustering of black ducks and mallards in multivariate landcover space (Figure 4).

We found considerable overlap in home range placement, selection and use of various landcover types, and similar niche breadths between black ducks and mallards within and around the TNWR, indicating little resource partitioning between these species based on our metrics. The TNWR complex historically has been managed to provide palustrine emergent, forested, and agricultural lands typical of other refuges in the midsouthern Mississippi flyway. Our study area contained seeds of wetland plants and agricultural crops, other plant material (e.g., stems, tubers), and aquatic invertebrates; these are foods that provide dabbling ducks with winter nutrition (Dugger et al. 2007; Hagy and Kaminski 2015; Hagy et al. 2017; Osborn et al. 2017). It may be that 1) black ducks do not forego use of resources in these areas even if mallards are present (and vice versa; Osborn et al. 2017), 2) the genetic similarity between these birds (< 1.5% genetic difference; Lavretsky et al. 2019) might allow them to tolerate coexistence without competitive exclusion on wintering grounds (as in McAuley et al. 2004), or 3) competition is not enough of an evolutionary force on wintering grounds to create distinct partitioning (as seen in other species; Chatterjee et al. 2020). It is notable, however, that there remains contention on the effect of mallards on black ducks given observations of direct negative interactions (McAuley et al. 1998; Schummer et al. 2020), although this was not observed at TNWR with > 1,000 h of focal observation, Osborn et al. 2021), and mallard replacement of black ducks once the latter populations have receded (Merendino et al. 1993).

Although there was little segregation between species, mallards did use cultivated lands to a greater degree than did black ducks; the latter of which made greater use of emergent wetlands and open water, revealing that there are some (albeit limited) instances where resource use does not completely overlap. Whether these patterns were influenced by competition is unknown, although these trends are not surprising. Emergent wetlands and open water are important parts of historical and contemporary habitat use by black ducks. Indeed, White et al. (1993) found black ducks use open water habitats to a greater degree and agricultural fields less extensively than mallards at the TNWR in winters 1990–1992. Additionally, forage studies in the TNWR show that, despite similarities, black ducks consume a greater amount of natural vegetation (i.e., stems, leaves, and seeds) while mallards consume more grain, a difference that can likely be attributed to the lower use of agricultural fields by black ducks (Byrd 1991; White et al. 1993). Black ducks also exploit cropland resources in winter (Bellrose 1976), but mallards may benefit from feeding on waste grain to a greater degree than black ducks because they may be better adapted to landscape changes associated with agriculture and human development (Baldassarre 2014; Bleau 2018; Droke 2018).

There may be other behavioral means by which black ducks segregate themselves from mallards. For example, in a concurrent study of dabbling ducks at TNWR from 2011 to 2013, Osborn et al. (2021) found that black ducks disproportionately used low-food-density patches within landcover types, such as wooded portions of wetlands and mudflats, suggesting that black duck distribution may be less related to absolute food density on the TNWR. Factors influencing these behaviors were unclear, but black ducks may have foraged on mudflats or in other landcover types with less-than-average seed densities to acquire invertebrates or perhaps to spatially segregate from other species (Osborn et al. 2021). We also do not know whether black ducks attained adequate food from these areas nor how it might have subsequently influenced survival. For example, Newcomb et al. (2016) showed that both daily precipitation and minimum temperatures influenced survival of female black ducks and that almost twice as many mortalities occurred during the colder, drier winter of 2010–2011 versus winter 2011–2012 at TNWR. Although it is possible birds in this second winter may have had access to increased available resources and perhaps decreased thermal constraints, it is also possible that altered foraging strategies in the presence of mallards may have hindered black duck ability to deal with harsher environmental conditions (Newcomb et al. 2016). Given this scenario, understanding how black ducks' use of landcover types such as mudflats and wooded wetlands containing lesser forage density factor into birds' overall survival is relevant (Newcomb et al. 2016).

Annual waterfowl management strategies at TNWR have not changed so significantly through time to warrant the occurring precipitous decline of black ducks. We did expect that considerable overlap would occur in agricultural croplands and that the refuge system would have been able to make the TNWR more attractive to wintering black ducks by implementing management strategies that promote more nonagricultural-based wetlands. Although mallards showed more affinity toward agriculture than did black ducks, the overall difference in agricultural land use by mallards compared with black ducks was minimal (i.e., < 5% difference). Moreover, the difference in proportional use of emergent wetland by the two species was only about 2%, therefore, management to augment the total amount or quality of existing wetlands is likely not a viable option. Additionally, prominent changes in vegetation management (e.g., improved emergent wetland conditions) may not necessarily greatly benefit black ducks at TNWR because they apparently use low-density forage patches (Osborn et al. 2021). However, ensuring resources (e.g., moist-soil plants, submerged aquatic vegetation; Osborn et al. 2021) are dispersed throughout emergent wetlands and open water areas may benefit black ducks during winter at TNWR. Thus, despite the challenges in identifying some optimal wetland complex for black ducks, this work and others (Monroe et al. 2021; Osborn et al. 2021) at a minimum, guides the next steps in investigating how vegetation changes can help management of this species on a historically important shared wintering area.

Going forward, we also suggest that factors influencing black ducks on breeding grounds be investigated given that black duck declines on wintering grounds in the Mississippi flyway could be a consequence of influences occurring on the breeding grounds, especially in Ontario and Quebec, Canada; areas of which have experienced declines to a greater extent than breeding populations to the east (USFWS 2019). Additionally, meta-analyses are needed to examine broader continental changes in black duck distribution, such as possible climate effects (Meehan et al. 2021) and how that might redistribute populations over time. How black ducks and mallards will continue to exploit resources in North America, particularly given black ducks northeasterly retreat (Brook et al. 2009; Meehan et al. 2021) is unknown, given that responses to climate change can be highly variable across bird guilds and, thus, difficult to predict (La Sorte and Jetz 2012). Based on our and previous work (English et al. 2017), there may be great overlap in resource use with mallards in interior wetlands and agricultural complexes of North America. More broadly across Aves, it is likely that the degree of overlap between species' realized and fundamental niches (Soberon and Nakamura 2009; Dormann et al. 2010; La Sorte and Jetz 2012), and the plasticity or adaptive capacity of a species to environmental change will likely be a significant influence on migratory bird abundance and distribution and, subsequently, their management (Skelly et al. 2007; Gienapp et al. 2008).

Although black ducks are a species of Least Concern (Birdlife International 2020), the trend in western Tennessee somewhat harkens to ‘ghosts of birds' past' (Diamond 1990) as the species continues to decline in the south-central Mississippi flyway. Compared with historical segregation of black ducks and mallards provided by eastern forests of North America, the great expansion eastward by mallards into contemporary landscapes has increased interactions between mallards and black ducks. However, it has been demonstrated that black ducks are physiologically better able to tolerate salinity compared with mallards, creating some partitioning of foraging resources in coastal marshes (Barnes and Nudds 1991; Ringelman et al. 2015). Thus, while there is joint use of some coastal environments by the two species, it may be that salt marshes of eastern North America represent the last vestige of resource space disproportionately occupied by black ducks over that of mallards.

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. Raw telemetry data for radiomarked female black ducks Anas rubripes (ABDU) and mallards A. platyrhynchos (MALL) used in examining resource partitioning at the Tennessee National Wildlife Refuge, Tennessee, winter 2011–2012. Provided are the bird identification number (id), species (ABDU or MALL), hybrid identification (ABDX), the x and y coordinates in UTM Zone 16, and the date and time for each location with 1100 hours indicating a diurnal location and 2300 hours indicating a nocturnal location.

Available: https://doi.org/10.3996/JFWM-21-039.S1 (177 KB CSV)

Table S1. Individual radiomarked female black ducks Anas rubripes (ABDU) and mallards A. platyrhynchos (MALL) used in examining resource partitioning at the Tennessee National Wildlife Refuge, Tennessee, winter 2011–2012. The number of transmitter days and locations obtained, the beginning and end dates of collection, and home range sizes (km²) are provided for each individual. Hybrid status for black ducks is indicated.

Available: https://doi.org/10.3996/JFWM-21-039.S2 (133 KB DOCX)

Table S2. Mean (±SE) home range size, selection ratios by landcover type, niche breadth, and intraspecific resource overlap for pure female black ducks Anas rubripes (ABDU), female black duck–mallard hybrids (HY), and mallards A. platyrhynchos; (MALL) at the Tennessee National Wildlife Refuge, Tennessee, winter 2011–2012.

Available: https://doi.org/10.3996/JFWM-21-039.S2 (133 KB DOCX)

Figure S1. Average home range size (km²) in relation to the number of locations used in estimation for radiomarked female black ducks Anas rubripes and mallards A. platyrhynchos at the Tennessee National Wildlife Refuge, Tennessee, winter 2011–2012.

Available: https://doi.org/10.3996/JFWM-21-039.S2 (133 KB DOCX)

Reference S1. Carney SM. 1992. Species, age and sex identification of ducks using wing plumage. Washington, D.C.: U.S. Fish and Wildlife Service.

Available: https://doi.org/10.3996/JFWM-21-039.S3 (11.700 MB PDF)

Reference S2. Devers PK, Collins B. 2011. Conservation action plan for the American black duck. 1st edition. Laurel, Maryland: U.S. Fish and Wildlife Service, Division of Migratory Bird Management.

Available: https://doi.org/10.3996/JFWM-21-039.S4 (5.777 MB PDF)

Reference S3. Fronczak D. 2020. Waterfowl harvest and population survey data. Bloomington, Minnesota: U.S. Fish and Wildlife Service, Division of Migratory Bird Management.

Available: https://doi.org/10.3996/JFWM-21-039.S5 (2.321 MB PDF)

Reference S4.[USFWS] U.S. Fish and Wildlife Service. 2015. Waterfowl population status, 2015. Washington, D.C.: U.S. Department of the Interior.

Available: https://doi.org/10.3996/JFWM-21-039.S6 (3.429 MB PDF) and https://fws.gov/story/2015-07/2015-waterfowl-population-report

Reference S5.[USFWS] U.S. Fish and Wildlife Service. 2019. Waterfowl population status, 2019. Washington, D.C.: U.S. Department of the Interior

Available: https://doi.org/10.3996/JFWM-21-039.S7 (3.957 MB PDF) and https://www.fws.gov/media/waterfowl-population-status-2019

Reference S6.[USFWS] U.S. Fish and Wildlife Service. 2020. Weekly waterfowl population summary for Tennessee National Wildlife Refuge.

Available: https://doi.org/10.3996/JFWM-21-039.S8 (68 KB PDF)

Our project would not have been possible without the support of all TNWR personnel, especially C. Ferrell, R. Wheat, and D. Zabriskie, S. Bard, S. Bearden, R. Corlew, A. Fish, and C. King. T. Peterson provided critical assistance in the field with data collection. We thank the Forest and Wildlife Research Center (FWRC), Mississippi State University for support. R.M. Kaminski was supported by Clemson University's James C. Kennedy Waterfowl and Wetlands Conservation Center during preparation of this manuscript. We thank two anonymous reviewers and the Associate Editor for providing comments that improved an earlier version of this manuscript.

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.