The wintering period is often a limiting time for waterfowl. To understand the behavioral dynamics of Atlantic brant Branta bernicla hrota wintering along coastal New Jersey, USA, we conducted observations across the full 24-h diel period in an effort to construct an accurate time-budget model for the wintering population. In most behavioral studies, it is only possible to collect diurnal and crepuscular behavior data, forcing the assumption that these data are representative of nocturnal behavior in order to model the full 24-h diel period. We collected behavior data in 5,902 instantaneous observational scans across 4 time periods (morning crepuscular, diurnal, evening crepuscular, and nocturnal) from the third week in October to the third week in February 2009–2010 and 2010–2011. Brant primarily allocated time toward swimming (43.5%), feeding (26.4%), resting (15.4%), and flying (7.7%); these proportions differed significantly across times of day. Brant exhibited decreased flight (4.8% vs. 9.3%) and feeding (22.3% vs. 29.6%) and increased resting behavior (24.4% vs. 10.5%) nocturnally compared with diurnal periods. We further modeled explanatory environmental variables, hunting effects (open vs. closed seasons, locations open vs. closed to hunting), and time of day (diurnal and nocturnal only) on wintering behaviors. Feeding, resting, and swimming behavior presence were most influenced by a predictive model of (Hunt Season × Hunt Location × Period) + (Tide × Period). Flight behavior presence were most influenced by a predictive model of (Hunt Season × Hunt Location × Period) + (Tide × Temperature). There is an interactive effect of hunting pressure and period of day on observed activity; therefore, our results demonstrate that not accounting for nocturnal variation in behavior can lead to biases when extrapolating to energy expenditure models. Additionally, hunting areas proved to be nocturnally valuable because these areas contain valuable energy resources that may be unavailable diurnally, and our observations show that brant will shift their activities around hunting pressures to make use of these areas.
Quantifying associations between waterfowl and daily temporal cycles is often accomplished through studies of behavior (Altman 1974). Traditionally, these behavioral studies have been limited to diurnal periods. Jorde and Owen (1988) noted that a lack of sufficient equipment required to collect nocturnal data, limited information describing effective methodology, and the reluctance or inability of biologists to conduct field research during nocturnal periods have historically hindered nocturnal observations of behavior. Previous research suggests a limited recognition that waterfowl engage in nocturnal activity (e.g., Tamisier 1974, 1976; Albright 1981; Morton et al. 1989a; Percival and Evans 1997; Anderson and Smith 1999; Guillemain et al. 2002). Consequently, there is value in accounting for nocturnal behavior to construct accurate time-energy budgets (Baldassarre et al. 1988; Jorde and Owen 1988; McNeil et al. 1992; Henson and Cooper 1994; Baldassarre and Bolen 2006), because diurnal research that assumes nocturnal behavior is representative of the entire 24-h period (Ladin et al. 2011; Cramer et al. 2012; Jones et al. 2014) might be biased.
The Atlantic brant Branta bernicla hrota (hereafter, brant) is a regionally important gamebird on its wintering grounds in the Atlantic Flyway. This species has large annual population fluctuations (Roberts and Padding 2018) related to productivity and survival; therefore, harvest regulations (daily harvest and season length) and population numbers on the wintering grounds can vary year-to-year. Little is currently known about how brant interact with available wintering habitat throughout the daily cycle and across the wintering period. Although previous efforts have quantified diurnal behavior of brant (Ladin et al. 2011), research has yet to effectively examine their nocturnal behavior during winter.
Many mechanisms have been hypothesized to influence waterfowl nocturnal behavior including tide, moon phase, daylight length, food availability, predator avoidance, and anthropogenic disturbance. For example, greylag geese Anser anser and barnacle geese Branta leucopsis have been shown to move their foraging activities to nocturnal periods when moon phase is at or near full (Lebret 1970; Ebbinge et al. 1975; Ydenberg et al. 1984). Common eiders Somateria mollissima and spectacled eiders Polysticta stelleri commenced nocturnal feeding when shortened day length, along with low energy densities of preferred food items, did not allow for meeting daily energy requirements with diurnal feeding alone (Systad et al. 2000; Systad and Bustness 2001). Natural and anthropogenic sources of diurnal disturbance also can reduce diurnal feeding opportunities and increase nocturnal activity (Owens 1977; Ydenberg et al. 1984; Bélanger and Bédeard 1990; Ward et al. 1994; Kahlert et al. 1996; Riddington et al. 1996). Additionally, hunting disturbance has the potential to shift geese feeding away from areas open to hunting during the day (Owens 1977; Stock 1993; Ward et al. 1994), while using areas open and closed to hunting nocturnally (Burton and Hudson 1978; Cox and Afton 1997; Dooley et al. 2010).
Daily and seasonal climatic patterns have the potential to influence waterfowl diurnal and nocturnal behaviors. By shifting from diurnal to nocturnal activity, birds generate behavior-related heat energy that offsets costs of thermoregulation (Tamisier 1976; Jorde et al. 1984; Turnbull and Baldassarre 1987; Baldassarre et al. 1988; Thompson and Baldassarre 1991; Baldassarre and Bolen 2006). Waterfowl may increase foraging activity during severe winter weather (Tamisier 1972), but they also have been shown to curtail energetically costly behavior (e.g., flight) while simultaneously increasing energetically conservative behaviors (e.g., loafing while exploiting diurnal microclimates and radiant heating) in an effort to cope with colder temperatures. For example, Canada geese Branta canadensis wintering in Illinois greatly reduced or stopped foraging flights to agriculture fields when temperatures fell below −9°C (Raveling et al. 1972). Albright et al. (1983) noted reduced activity of American black ducks Anas rubripes along coastal Maine as birds sought thermally advantageous roosting sites when temperatures fell below −20°C. Mallards Anas platyrhynchos wintering in Nebraska exploited roost sites that offered a diurnal thermal advantage and only commenced feeding activities around evening crepuscular periods when this realized thermal advantage lessened (Jorde et al. 1984), creating behavior-specific heat energy and reducing the costs of thermoregulation as temperatures decreased outside of the diurnal period.
Our objectives were to 1) quantify the behaviors of wintering brant throughout the 24-h diel period inclusive of morning crepuscular, diurnal, evening crepuscular, and nocturnal periods along southern coastal New Jersey, USA, 2) determine if environmental variability (e.g., freezing temperatures, wind, snow cover, ice cover, etc.), and anthropogenic disturbance (e.g., hunting) affect time spent in different behaviors between diurnal and nocturnal time periods, and 3) qualitatively assess if observed flock size (as a secondary behavior) could be influenced by time of day and anthropogenic disturbance.
Our study area was located in Atlantic and Ocean counties just north of Atlantic City on New Jersey's (USA) southeastern shore. Encompassing over 65,000 ha (Figure 1), this area is primarily public land comprising Edwin B. Forsythe National Wildlife Refuge (EBFNWR; 39°36′N 074°17′W) and Great Bay (39°31′N 074°22′W) and Reeds-Absecon Bay (39°25′N 074°27′W) Wildlife Management Areas, which are owned and managed by the New Jersey Division of Fish and Wildlife and lie within New Jersey's Coastal Waterfowl Hunt Zone. Based on midwinter waterfowl inventory counts (U.S. Fish and Wildlife Service, 1955–2010; https://migbirdapps.fws.gov/), this area holds approximately 44% of the Atlantic Flyway population of wintering brant. Large numbers of migrating and wintering waterbirds utilize this area, so the Ramsar Convention designated EBFNWR properties as Wetlands of International Importance in 1986 (Ramsar Convention Secretariat 2006).
Habitat in the study area falls within a tidal estuarine system protected by barrier beaches. These areas encompass large expanses of natural and anthropogenically altered salt marshes (i.e., open marsh water management), brackish wetlands, managed tidal and freshwater impoundments, freshwater wetlands, and woodlands. Anthropogenic marsh alterations for mosquito abatement are usually conducted by county-level Mosquito Control Commissions, under permit from the landowner and regulatory agencies. Open marsh water management alters the marsh by creating permanent pools that support mosquito larvae eating fish (James-Pirri et al. 2009). Tidal flux promotes four major estuarine ecosystems: high marsh, low marsh, mudflat, and subtidal areas (Tiner 1987, 2009).
Material and Methods
We selected 44 observation locations—20 locations on EBFNWR lands closed to hunting and 24 locations on state and EBFNWR lands open to hunting. We ensured that each observation location contained several habitat types to avoided biased behavioral expression. We conducted behavioral observations between the third week of October and the third week of February 2009–2010 and 2010–2011 to effectively sample the wintering period of brant within the study area based on long-term anecdotal observational data (P.M. Castelli, New Jersey Division of Fish and Wildlife, personal communication). We compared observations over a series of hunting and nonhunting periods to evaluate the possible effects of increased anthropogenic disturbance on behavior (2009–2010 season: 44 hunting-days between 7 Nov–14 Nov and 26 Nov–14 Jan and 33 nonhunting-days between 23 Oct–6 Nov, 15 Nov–25 Nov, and 15 Jan–21 Feb; 2010–2011 season: 44 hunting-days between 6 Nov–13 Nov and 25 Nov–25 Jan and 39 nonhunting-days between 19 Oct–5 Nov, 14 Nov–24 Nov, and 26 Jan–18 Feb). There was no Sunday hunting in New Jersey; therefore, observation weeks ran from Tuesday to Saturday to capture behavior on days of increased hunting pressure during weekends that fell within New Jersey's Coastal Zone seasons. Depending upon the location, behavioral observations occurred from a carry-in pop-up blind, a permanent elevated blind, or a marked state or federal truck. We selected daily observation locations at random 1 wk in advance.
We collected behavioral data through instantaneous scan samples (Altmann 1974; Baldassarre et al. 1988) conducted every 10 min over 6-h observation periods (0300–0900 hours, 0900–1500 hours, 1500–2100 hours, and 2100–0300 hours) using 8× binoculars diurnally and a light-intensifying 6× night-vision scope nocturnally (Generation 3 Pinnacle: Morovision MV 760 with an advertised recognizable range of 400 m; Table S1, Supplemental Material). We classified 10-min scans as diurnal, nocturnal, morning crepuscular, and evening crepuscular. We defined a crepuscular period as occurring 30 min before and after sunrise or sunset. Six-hour observation blocks alternated biweekly: one week focused on diurnal and nocturnal periods and the following week focused on morning and evening crepuscular periods. This resulted in paired 6h daily observation sessions occurring 12 h apart, thus being matched over that day's tide cycle. Arrival to observation locations occurred ≥30 min in advance of the observation session to allow adequate time for waterfowl acclimation after possible disturbance of arriving to observation areas (Jeske and Percival 1995).
We recorded behavior of all brant within a 200-m radius of the observation location because that was the effective distance where designation of species and behavior could be determined under inclement weather conditions and/or poor ambient-light conditions utilizing the night-vision scopes (Allison and DeStefano 2006). We trained observers on observational distances of known objects prior to completing data collection. We assigned behaviors to one of the following eight categories: feeding (head down while actively grabbing and ingesting vegetation), resting (sleeping and loafing), comfort (preening), swimming, alert, flying, agonistic (inter- and intraspecies aggression), and walking (Johnsgard 1965; Paulus 1988). If a brant was exhibiting more than one behavior during a given instantaneous scan (e.g., motile sleeping) we recorded the most energetically costly behavior as the bird's behavior for that particular scan. If a disturbance occurred during or between a scan(s) that could have impacted behavior, we recorded and categorized the type of disturbance.
After initially checking for birds in flight within the 200-m observation radius, instantaneous flock scans, randomly selected to start from left or right, focused on birds on the ground or in the water. When large flocks (n ≥ 500) were present, we recorded groups of birds exhibiting the same behavior in multiples of 10 individuals to expedite the process of recording behavior for the entire flock before a given 10-min scan window elapsed. We recorded environmental variables including temperature, wind speed, cloud cover, ice coverage percentage, and precipitation hourly directly at the observation location. We measured temperature and wind speed using a handheld anemometer (Kestrel 1000 series). We incorporated hourly water equivalent precipitation rate, recorded at Atlantic City International Airport, (18.7 km SSW of the center of our study area) into each observation to the nearest scan(s) (National Climatic Data Center 2011). We recorded cloud cover into 4 quartiles spanning a 25% coverage window before moving to the next higher category. We determined tide height utilizing the New Jersey Tide Telemetry System (U.S. Geological Survey; https://www2.usgs.gov/science/cite-view.php?cite=2648), which records tidal height readings at 6-min intervals at various locations within the study area. We calculated tidal delays between each observation location and tide telemetry station to more accurately capture tide effects on behavior; we incorporated these data into the data set to the nearest 10-min scan. We also recorded the observation location, regular hunting season designation (open vs. closed of a combination of brant, Canada goose, and/or duck in the Coastal Zone of New Jersey; 2009–2010: 7–14 Nov, 26 Nov–26 Jan; 2010–2011: 6–13 Nov, 25 Nov–25 Jan), hunting area designation (observation location open vs. closed to hunting), sunrise and sunset times, and moon phase. In an attempt to correlate nocturnal behavior with available ambient light, we combined cloud cover and moon phase to estimate a categorized light variable (low, medium, high) to model against observed nocturnal behavior
We proportioned all behavior data to allow for comparison of percent time spent in each behavior between the four observation periods. We only collected data for 2 y, so we did not examine year effects but combined behavioral data to be better representative of a range of values across temporal and environmental variables. We considered 10-min scans as independent units based on two pieces of information. First, previous research has shown that black duck behavior becomes statistically independent with a time lag of 5 min between scans (Morton et al. 1989b). Second, to independently test Morton and colleagues' (1989b) finding, we used a semivariogram method to determine what time lag would limit effects of pseudoreplication of our data set. A semivariogram tests for the amount of semivariance between pairs of observations over a series of consecutive time lags (in this case multiples of 10 min), starting at time zero by calculating squared differences between all of the available paired observations and obtaining half of the average for all observations separated by that lag (Lloyd 2007). When semivariance no longer increases between consecutive time lags (<15% change) the effects of pseudoreplication are limited because behavior at some future observation is less likely to be the same as the time-zero behavior (Boyce et al. 2010). Based on this analysis, we found that 10-min scans were independent of one another.
We compared the proportion of time spent in each behavioral state during morning crepuscular, diurnal, evening crepuscular, and nocturnal periods. Sample size was large but the data were nonnormally distributed as a result of high levels of kurtosis from scans that had either 0% or 100% of a certain behavior; therefore, we used the nonparametric Kruskal–Wallis H test (α = 0.05) to compare each behavior across the four time periods (Khan and Rayner 2003). Mann–Whitney U post hoc pair-wise comparisons were used to compare behavior between periods (IBM Corp. 2012).
We further used binary logistic general linear models and Akaike's Information Criterion corrected for small sample size (AICc; SPSS, version 25; Burnham and Anderson 2002; Table S2, Supplemental Material) to determine the likelihood that a behavior's presence or absence within a 10-min observation period was impacted by temporal and environmental variables. To test the assumption that diurnal behavior is representative of nocturnal behavior, and to assure sample size was robust for model comparison, we only modeled the most common behaviors (feeding, resting, swimming, flying that represented >95% of observed behavior) for diurnal and nocturnal periods against temporal variables. Specifically, we examined 50 a priori models with additive and multiplicative variable combinations investigating influences of temperature, ice cover, hunting season (open or closed), hunting location (if observation location was physically open or closed to hunting), wind speed, tide height, precipitation, nocturnal brightness (a function of moon phase and cloud cover), diurnal cloud cover, and time period (diurnal verses nocturnal) on the presence or absence of the most common behaviors (feeding, resting, swimming, flying) within any 10-min scan.
Our third objective was to qualitatively assess whether flock size (as a secondary behavior) could be influenced by time of day or anthropogenic disturbance. We did not include flock size as a predictor variable in the behavioral models above because of the potential bias in seeing the entire flock. Oftentimes, as a result of tide stage, portions of flocks would be obscured by habitat features (e.g., edge of marsh or tidal creek, behind islands) and unable to be observed for behavioral recording during a given scan or out of range of the night-vision capability to determine individual birds. By averaging flock size across the observation period, we attempted to limit bias by not comparing flock size per scan, which introduces the assumption that each scan was representative of the entire flock present. Adding a categorical “Yes” or “No” variable of whether or not the entire flock was seen per scan would have increased the accuracy of this analysis and reduced potential bias of not observing the entire flock. We compared average observed flock size between open and closed hunting seasons both grouped by diurnal and nocturnal times of day as well as in open and closed hunting lands using independent sample t-tests (P < 0.05).
Brant behavioral data were recorded in 5,902 instantaneous scans during the morning crepuscular (n = 258), diurnal (n = 3,200), evening crepuscular (n = 350), and nocturnal periods (n = 2,094). Recorded disturbance numbers were low (n = 89) and occurred during 1.6% of morning crepuscular scans (n = 4), 2.4% of diurnal scans (n = 77), 0% of evening crepuscular scans (n = 0), and 0.4% of nocturnal scans (n = 8; Table 1). Disturbance was not included in any behavioral models because of these low numbers.
All behaviors had significant differences between periods (Kruskal–Wallis: H > 31.46, P < 0.01). Nocturnal and morning crepuscular feeding (16–22%) was lower than diurnal and evening crepuscular periods (29–30%, Mann–Whitney: U > 6.53, P < 0.01; Table 2). Nocturnal resting (24%) was greater than all three other periods (9–11%, U > 2.99, P < 0.02). Comfort behavior was low (3–4%), but diurnal rates were greater than morning crepuscular and nocturnal and evening crepuscular rates were greater than nocturnal (U > 3.29, P < 0.01). Swimming behavior was greater during the morning crepuscular period (59%) compared with the other three periods (43–44%, U > 4.42, P < 0.01). Alert behavior was low (1–2%), but morning crepuscular was lower than the other three periods (U > 3.00, P < 0.02). Nocturnal flying behavior (5%) was lower than the other three periods (7–12%, U > 4.21, P < 0.01). Walking behavior was lower during nocturnal and morning crepuscular periods (0.24–0.33%) than diurnal and evening crepuscular periods (2%, U > 3.90, P < 0.01). Lastly, agonistic behavior was lower during nocturnal and morning crepuscular periods (0.00–0.05%) than diurnal and evening crepuscular periods (0.08–0.18%, U > 2.90, P < 0.01).
When testing the environmental and temporal explanatory variable effects on feeding behavior presence within a 10-min scan, the best predicted model was (Hunt Season × Hunt Location × Period) + (Tide × Period) (wi = 1.00; Table 3). The presence of feeding was more likely to occur diurnally than nocturnally despite temporal or spatial hunting season openness and availability (Table 4). When the hunting season was closed there was little difference in diurnal feeding presence between open and closed locations (71% vs. 69%) but nocturnal feeding presence was lower in closed versus open locations (41% vs. 62%). When the hunting season was open, there was increased diurnal feeding presence in closed versus open locations (71% vs. 63%) but nocturnal feeding presence was similar between open and closed areas (46% vs. 42%). There was an additional interactive effect between tide height and time of day. During both the diurnal and nocturnal periods, the presence of feeding being observed increased with lower tides (diurnal: β = −0.107 + 0.026, Wald χ2 = 16.46, P < 0.01; nocturnal: β = −0.416 + 0.039, Wald χ2 = 115.31, P < 0.01).
When testing the environmental and temporal explanatory variable effects on resting behavior presence within a 10-min scan, the best predicted model was also (Hunt Season × Hunt Location × Period) + (Tide × Period) (wi = 1.00; Table 3). When the hunting season was closed, the presence of resting occurring in a 10-min scan was similar between diurnal and nocturnal periods but open locations had lower presence than closed locations (open locations: 32% diurnal vs. 34% nocturnal; closed locations 42% diurnal vs. 40% nocturnal; Table 4). However, when the hunting season was open there was increased presence of resting in closed locations as well as during nocturnal periods (closed diurnal = 39%, closed nocturnal = 48%, open diurnal = 21%, open nocturnal = 35%). There was an additional interactive effect between tide height and time of day. During both the diurnal and nocturnal periods, the presence of resting being observed increased with higher tides (diurnal: β = 0.176 + 0.026, Wald χ2 = 44.11, P < 0.01; nocturnal: β = 0.235 + 0.038, Wald χ2 = 39.0, P < 0.01).
When testing the environmental and temporal explanatory variable effects on swimming behavior presence within a 10-min scan, the best predicted model was once again (Hunt Season × Hunt Location × Period) + (Tide × Period) (wi = 1.00; Table 3). When the hunting season was closed, the presence of swimming occurring in a 10-min scan was greater diurnal than nocturnally. Diurnally, there was little difference between open and closed locations (76% vs. 73%); however, nocturnally, there was a higher presence of swimming occurring in open vs. closed locations (67% vs. 40%; Table 4). When the hunting season was open, there was a greater presence of diurnal swimming in open vs. closed locations (81% vs. 75%) as well as nocturnal open vs. closed locations (82% vs. 65%). There was an additional interactive effect between tide height and time of day. During both the diurnal and nocturnal periods, the presence of swimming being observed decreased with higher tides (diurnal: β = −0.129 + 0.029, Wald χ2 = 20.19, P < 0.01; nocturnal: β = −0.284 + 0.042, Wald χ2 = 46.38, P < 0.01).
When testing the environmental and temporal explanatory variable effects on flying behavior presence within a 10-min scan, the best predicted model was (Hunt Season × Hunt Location × Period) + (Temperature × Period) (wi = 1.00; Table 3). When the hunting season was closed, the presence of both diurnal and nocturnal flying was greater in closed vs. open locations (diurnal: 21% vs. 16%, nocturnal: 18% vs. 6%). When the hunting season was open, there was little difference between open and closed areas; however, flying presence was much higher diurnally vs. nocturnally (15–18% vs. 4–5%; Table 4). Unlike the other behaviors that were influenced by tidal height, the presence of flying was also influenced by the interactive effect between temperature and time of day. During both the diurnal and nocturnal periods, the presence of flying decreased with colder temperatures (diurnal: β = 0.017 + 0.007, Wald χ2 = 5.15, P = 0.02; nocturnal: β = 0.082 + 0.015, Wald χ2 = 29.78, P < 0.01).
During closed hunting seasons, flock sizes were similar between nonhunting (x̄ = 46 ± 2) and hunting (x̄ = 46 ± 3) locations during the diurnal period (t1,1274 = 0.76, P = 0.94) as well as between nonhunting (x̄ = 18 ± 2) and hunting locations (x̄ = 22 ± 2) during the nocturnal period (t1,712 = 1.62, P = 0.11). During the hunting season, average flock sizes tended to be similar between nonhunting (x̄ = 45 ± 2) and hunting (x̄ = 45 ± 3) locations during diurnal period (t1,1851 = 0.22, P = 0.83); however, they were much larger on hunting (x̄ = 79 ± 5) compared with nonhunting (x̄ = 29 ± 2) locations during the nocturnal period (t1,1383 = 7.63, P < 0.01).
Period of day was an important variable in our behavior modeling, and the results of our time budget show that brant wintering in New Jersey display different proportions and presence of behaviors during different periods of the day. An increase in the observed proportion of feeding during diurnal and evening crepuscular periods, and a reduction during nocturnal and morning crepuscular periods has previously been documented for barnacle geese in Britain (Owen et al. 1992). Decreased feeding during the morning crepuscular period, along with the increase in flying and swimming, was most likely a result of the birds either being disturbed by increased anthropogenic disturbance or more likely moving to diurnal foraging areas. During morning crepuscular periods, we observed brant flocks dispersing from large nocturnal roosting bays to feed along mudflats, in shallow subtidal areas where Ulva is more prevalent and attainable, or in manicured grass areas because the energy quality of terrestrial grasses is higher than submerged aquatic vegetation, saltmarsh cordgrass Spartina alterniflora, or eelgrass Zostera marina (Ladin et al. 2011). Nocturnal feeding was never observed in terrestrial habitats, likely because of associated predation risk of these areas. Nocturnal feeding occurred almost exclusively in subtidal and mudflat areas in association with ebbing or low tides, similar to observations by Tinkler et al. (2009). Tide height was an important interactive explanatory variable with period of the day and likely helped explain the observed amounts of nocturnal feeding. Tide cycle has previously been shown to influence foraging patterns (Johnson and Rohwer 2000), especially as it relates to nocturnal feeding (Paulus 1988; Morton et al. 1989a; Thompson and Baldassarre 1991; Wilson and Atkinson 1995).
Past research has noted that decreased activity in waterfowl may be a result of individuals avoiding unfavorable thermoregulatory conditions and reducing the amount of resources needed to maintain body temperatures (Paulus 1988). Reduced feeding in relation to falling temperatures is consistent with reports by Raveling et al. (1972), Albright et al. (1983), and Jorde et al. (1984). Below a threshold temperature, reduced feeding and increased resting may be thermally advantageous; however, this value is unknown and varies based on environmental variables. Despite these past findings, temperature was not a predictive variable in our top model of feeding, swimming, or resting behavior, although we acknowledge it did emerge in the second models and thus it may remain a small predictor.
Despite our findings of feeding, swimming, and resting not being affected by temperature, we did see colder temperature significantly reduce flying presence in our scans. Flight is the most energetically expensive behavior (Ladin et al. 2011); thus, this finding supports the past research. Biologically, reduced flight behavior and associated energy expenditure during colder temperatures has previously been recorded for Canada geese wintering in Illinois (Raveling et al. 1972) and for black ducks wintering in Maine (Albright et al. 1983). We do acknowledge this result could be a result of methodological bias. Although we were able to capture an equal or greater proportion of flight behavior compared with previous waterfowl behavior studies (Burton and Hudson 1978; Afton 1979; Gauthier et al. 1984; Morton et al. 1989a; Cramer et al. 2012), our instantaneous-scan sampling method remains relatively ineffective at capturing flight behavior data nocturnally because of the limited field of view of night-vision scopes used (field of view = 8°, as compared with 360° diurnally). Thus, we encourage future research to use Global System for Mobile communications radiotelemetery with built-in accelerometers (Weegman et al. 2017) to better quantify flight behavior during nocturnal periods.
Brant exhibited a significant increase in resting behavior nocturnally when compared with other observation periods. Additionally, our predictive modeling found that during the hunting season, observed presence of resting behavior was higher nocturnally for hunting and nonhunting areas compared with diurnal periods. As a result of reduced visibility, brant were most likely less affected by visual risk factors (e.g., raptors, helicopters) that are more prevalent diurnally, even though increased moonlight allows some degree of detection (Kahlert et al. 1996; Tinkler et al. 2009). Most sources of observed diurnal disturbance were anthropogenic in nature (n = 62 of 77 diurnal disturbance events; Table 1) and largely absent outside of the diurnal and crepuscular periods; therefore, we also hypothesize that brant were generally not as susceptible, or simply not exposed, to disturbance during the nocturnal period and thus could allocate greater time toward resting behavior. Additionally, brant may have been able to meet daily energy requirements with diurnal and evening crepuscular feeding, which may explain the reduced amounts of feeding and locomotive behaviors and increased resting observed nocturnally.
An increase in diurnal and nocturnal swimming and a decrease in flying during the open hunting seasons may also suggest that brant shift their main locomotive activities to compensate for harvest risk. Although swimming could possibly reduce access to preferred food items, based on location of feeding areas in relation to roosting areas, it is arguably less dangerous than flying. Our model selection analysis also revealed that during the hunting season there was a reduction in the presence of observed diurnal feeding in open hunting locations compared with closed refuge locations. This supports previous research on brant and brent B. bernicla geese during the diurnal period (Owens 1977; Stock 1993; Ward et al. 1994), where brant fed more on nonhunting areas compared with hunting areas during open hunting seasons, presumably because of the presence of hunting disturbance in open hunting areas. There was no significant difference in observed amounts of feeding in these same areas nocturnally. Based on observed proportions of feeding, brant exploited hunting and nonhunting areas equally at night although at a lesser degree than diurnally. Past research suggests that birds will disperse into hunting areas in the absence of associated hunting disturbance (Burton and Hudson 1978; Cox and Afton 1997; Dooley et al. 2010). We did not see increased presence of feeding in open hunting locations at night, but it was noteworthy and perhaps contradictory that the flock sizes in open locations at night drastically increased, indicating birds were taking advantage of a resource in those lands in the absence of anthropogenic disturbance.
Nocturnal behavior of brant differs from diurnal and crepuscular behavior and when incorporated into daily time-activity budgets allows further inference about energy expenditure while accounting for various temporal variables. Our behavioral research provides an initial step toward estimating energetic needs of the wintering population of brant. With additional research to investigate landscape carrying capacity across main brant wintering areas, land managers will be able to better understand possible resource limitations affecting the population. Further, failure to account for nocturnal variation when applying behavior proportions will bias energy expenditure models in estimating daily energy expenditure. Last, nonhunting areas played an important role during open hunting seasons. We saw a consistent interactive effect between hunting seasons, hunting locations, and periods of the day on winter behaviors, so brant may not be able to meet existence requirements depending on the hunting pressure on those birds. As a result, nonhunting areas may allow brant to improve foraging and resting needs when hunting disturbance is present on the landscape. However, seasonal and annual variability in food resources mean that these areas alone likely do not always provide such benefit, which may explain the observed amounts of nocturnal feeding in hunting areas and diurnal feeding on manicured grass areas during open hunting seasons. These results show the importance of incorporating ample nonhunting locations across the wintering area in locations where food quantity and quality are beneficial.
Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.
Table S1. Behavioral scan samples of Atlantic brant Brant bernicla hrota during winters of 2009–2011 on properties owned and managed by Edwin B. Forsythe National Wildlife Refuge and the New Jersey Division of Fish and Wildlife, New Jersey, USA.
Found at DOI: https://doi.org/10.3996/032018-JFWM-023.S1 (804 KB XLSX).
Table S2. Full Akaike Information Criterion models determining how the presence of a wintering Atlantic brant Branta bernicla hrota behavior in a 10-min scan (October–February 2009–2010 and 2010–2011, New Jersey, USA) could have been influenced by additive and multiplicative effects of temperature, ice cover, hunting season (open or closed), hunting location (if location was physically open or closed to hunting), wind speed, tide height, precipitation, nocturnal brightness (a function of moon phase and cloud cover), diurnal cloud cover, and time period (diurnal verses nocturnal) on the presence or absence of the most common behaviors (feeding, resting, swimming, flying).
Found at DOI: https://doi.org/10.3996/032018-JFWM-023.S2 (38 KB XLSX).
We would like to thank the organizations and agencies that supported this research: the Atlantic Flyway Council, the Atlantic Coast Joint Venture, the Arctic Goose Joint Venture, the New Jersey Division of Fish and Wildlife, Pittman–Robertson Federal Aid to Wildlife Restoration Grant W-68-R, the University of Delaware, the U.S. Fish and Wildlife Service-Northeast Region, and the Edwin B. Forsythe National Wildlife Refuge. We are grateful for the cooperation and support of Steve Atzert (Edwin B. Forsythe NWR), Ted Nichols (NJ Division of Fish and Wildlife), Orrin Jones, Jaan Kolts, Matthew Carney, Derek Brandt, Taylor Finger, Dane Cramer, Ken Duren, Trevor Watts, Kurt Bonds, Marissa Gnoinski, Andrew Dinges, Andrew Halbruner, Linda Morschauser, and Tom Clifford. We would also like to thank the Journal reviewers and Associate Editor for the constructive feedback on this article through the review process.
Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Citation: Heise JR, Williams CK, Castelli PM. 2019. Factors influencing a 24-hour time-budget for wintering Atlantic brant. Journal of Fish and Wildlife Management 10(1):79–90; e1944-687X. https://doi.org/10.3996/032018-JFWM-023
The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.