Abstract Sex differences in the behaviors underlying avian protandry, where males arrive at breeding areas earlier than females, are still poorly understood for most species. We tested for sex differences in stopover behavior, refueling rates, and postdeparture movements during spring migration in 2 consecutive years in wood-warblers (Parulidae) at a coastal site on Lake Erie, Ontario, using automated radio telemetry (black-throated blue warblers Setophaga caerulescens and magnolia warblers Setophaga magnolia) and analysis of plasma metabolites as indicators of refueling (magnolia warblers, American redstarts Setophaga ruticilla, and common yellowthroats Geothylpis trichas). We found no differences between sexes in stopover duration or refueling index, although we did find subtle sex differences in the onset and end of diel activity. Males of both species began to forage earlier in the morning than females in 2015, and adult males of both species ended diel activity later in the evening than adult females in 2014 and 2015. More obvious were annual differences in stopover duration and the timing of diel activity, with shorter stopovers and an earlier onset of diel activity in the year with a warmer spring. We also did not find any evidence that sexes differed in their postdeparture ground speeds or migration routes. In wood-warblers, males and females can differ in some aspects of their stopover ecology, but these differences are likely context dependent and likely do not drive protandry in a consistent way. INTRODUCTION Protandry, the earlier arrival of males to breeding areas than females, is a common reproductive strategy among migratory birds with typical sex roles (Newton 2008; Morbey et al. 2012), but we do not know the precise behavioral mechanism or mechanisms used or if different species use similar combinations of mechanisms to achieve protandry (Coppack and Pulido 2009). Several recent studies show that males depart from wintering areas before females (Marra et al. 1998; Stanley et al. 2012, Fig. 1a in McKinnon et al. 2016; Schmaljohann et al. 2016), however, additional factors may be at play. For example, some studies show that males refuel more quickly than females at migratory stopover sites (Seewagen et al. 2013) and in captive studies (Schmaljohann et al. 2016). Specifically in sparrows and finches, males often winter farther north than females, and so can get a head start on migration (e.g., Woodworth et al. 2016). Our incomplete knowledge about sex differences in timing and movement can be partly attributed to the challenges of tracking individual birds over great distances and at multiple stopover sites. Nearctic-Neotropical wood-warblers (Parulidae) are mostly small, sexually dimorphic, insectivorous, and medium or long distance migrants. Based on analysis of data for Parulidae (Francis and Cooke 1986; Kissner et al. 2003), protandry at migratory stopover averages 4.2 days (n = 37 estimates from 22 species, range = 0.7–7.8 days; also see Bauböck et al. 2012). A number of studies have observed sex differences in the migratory behavior of Parulidae warblers. For example, Marra et al. (1998) re-sighted marked birds and observed earlier departure of males than females in Jamaica in 1 of 2 years and in 1 of 2 habitat types. Using plasma metabolite analysis, Seewagen et al. (2013) observed higher rates of refueling in males than in females at stopover sites in New York City, and suggested that faster refueling by males might be an important mechanism underlying protandry. Other studies have observed sex differences in foraging behavior or habitat use at stopover sites (Yong et al. 1998; Smith et al. 2007). The majority of studies, however, have not found any evidence of sex differences in stopover duration or patterns of mass gain (Otahal 1995; Morris and Glasgow 2001; Morris et al. 2003; Smith et al. 2007; MacDade et al. 2011; Dossman et al. 2015). Together, these observations suggest that sex differences in migratory behavior might depend on site- and year-specific factors. In the current study, we tested for sex differences in the behavior of several sexually-dimorphic warblers at a spring stopover site on the north shore of Lake Erie. We focused on the general hypothesis that males migrate faster than females, and predicted that males would start diel activity earlier in the day, refuel at a higher rate, depart the stopover site earlier, and have higher ground speeds than females. We used automated radio telemetry to characterize the timing of diel activity, stopover duration, and postdeparture flight paths in black-throated blue warblers Setophaga caerulescens and magnolia warblers Setophaga magnolia on their journey north to breed. We used plasma metabolite analysis to assess sex differences in refueling rates in a different sample of magnolia warblers and in 2 other species (American redstarts Setophaga ruticilla and common yellowthroats Geothlypis trichas). These species were chosen because they were sufficiently abundant at our site (yet there were too few black-throated blue warblers to also measure refueling). Consistent with other Parulidae, all 4 species exhibit protandry at stopover sites with estimates ranging from 2.9 to 6.1 days (Francis and Cooke 1986; Kissner et al. 2003). METHODS Study area We used data on birds captured at Long Point Bird Observatory (Old Cut Research Station; 42.584°, −80.397°) situated near the base of the Long Point peninsula on Lake Erie, Ontario, Canada. Radio tagged birds were subsequently detected to the north by automated receiving towers placed strategically throughout the landscape in southern Ontario as part of the Motus Wildlife Telemetry System (www.motus.org), a coordinated system of receiver towers (Taylor et al. 2017). Most towers had 3 nine-element directional Yagi antennas pointing in different directions, depending on local factors and topography. Under optimal conditions, radio transmitters are detected up to 20 km away, but in practice, the properties and orientations of antennas, uneven terrain, weather, and obstruction with vegetation reduce detection probability. At the time of the study, there were 49 (2014) and 74 (2015) automated towers broadly covering the geographical area. We obtained wind data from the NCEP/DOE Reanalysis 2 data set (http://www.esrl.noaa.gov/psd/data/gridded/data.ncep.reanalysis2.html) which provides meteorological data 4 times per day at a 2.5° × 2.5° spatial resolution (Kanamitsu et al. 2002). We also collected air temperature data about 6.5 km northwest of Old Cut Research Station at Bird Studies Canada Headquarters (42.615°, −80.458°). Site-level protandry To assess protandry in the 4 study species (black-throated blue warblers, magnolia warblers, American redstarts, and common yellowthroats) at Old Cut Research Station in the 2 years of our study, we used first capture dates (1 = May 1) of individual birds processed through the migration monitoring program at Long Point Bird Observatory (LPBO 2005). We used species-specific general linear models (GLMs) to analyze first capture date as a function of the factors sex, age group (SY = second year; ASY = after second year), and year. To test for the inclusion of 2-way interactions, we used likelihood ratio tests (LRT) to compare nested models with and without interactions (e.g., Myers 1990). Data analyses were performed in R 3.3.1 (R Core Team 2016) or SAS v. 9.3 (SAS Institute 2011). Stopover duration To test for sex differences in the minimum length of stay (hereafter stopover duration) of warblers at stopover, we used the Motus Wildlife Telemetry Array to track black-throated blue (2014: n = 33; 2015: n = 21) and magnolia warblers (2014: n = 32; 2015: n = 69). Birds were captured in mist nets at Old Cut Research Station (LPBO 2015), and were selected for the study to balance sample sizes by species, sex, and day of year. Despite setting a high density of nets daily in a small site, we cannot know with certainty when birds actually arrived. Birds were initially sexed and assigned to an age group following Pyle (1997), measured for body weight (g), tarsus length (mm), and wing length (mm), and to minimize handling time were then scanned once in a quantitative magnetic resonance (QMR) body composition analyzer to measure fat mass and lean mass (Guglielmo et al. 2011; Kennedy et al. 2016). The second inner-most tail feather was taken and archived. Due to ambiguous sexing of magnolia warblers by plumage in the field, we later extracted DNA from these feathers and developed novel PCR primers to unambiguously assign sex (Supplementary Material). Each individual was banded with a standard aluminum band issued by the US Fish and Wildlife Service and affixed with a uniquely encoded 0.29 g radio transmitter (model NTQB-1, Lotek Wireless, www.lotek.com) with a 12.7 s pulse rate on the Motus frequency (166.380 MHz). Radio transmitters were attached using a leg-loop harness (Rappole and Tipton 1991) with elastic string that deteriorates after a few months. Birds were released immediately after tagging. To determine if capture dates of birds in the radio telemetry study were similar between sexes as intended, we used species- and year-specific GLMs with the factors sex and age. We also assessed whether body condition (fat mass, lean mass, and weight) differed between years after controlling for sex and age using species-specific GLMs with the factors year, sex, and age. At the end of the summer, detections were downloaded from each tower in the Motus Wildlife Telemetry Array. Each detection included the transmitter id, site (location of the receiver tower), antenna number (1–3), and signal strength. In wood-warblers, landscape-level stopover movements can occur prior to a long-distance migratory movement (Taylor et al. 2011; Dossman et al. 2015). We considered the stopover region for a given individual to be all receiving stations within 20 km of Old Cut Research Station. This area allowed for some landscape-level relocations and movement away from the site of initial capture. The night of migratory departure was thus considered to be the last night a bird was detected at a tower within the stopover region (i.e., the Old Cut, Lower Big Creek, Bird Studies Canada, or Long Point Eco-Adventures tower). For departures from Old Cut, departure time was assigned as the last time an individual was detected at Old Cut on its departure night. Departure time relative to sunset was calculated using sunset times from the National Oceanic and Atmospheric Administration’s solar calculator (http://www.srrb.noaa.gov/highlights/sunrise/sunrise.html). Departure times could not be reliably or consistently assessed for birds that relocated away from Old Cut before departing the stopover region. Following Dossman et al. (2015), stopover duration (departure date − capture date + 1) was compared between sexes using survival analysis. We applied Cox semiparametric regression in a proportional hazard modeling framework using the PHREG procedure in SAS. Proportional hazard models generally assume that the covariates affect the hazard multiplicatively and, unlike accelerated failure time models, have the advantage of allowing for time-dependent covariates (Allison 1995). PHREG provides different options for handling ties in event times. We specified the TIES = EXACT option which assumes a true but unknown ordering of tied stopover durations. In addition to sex, we included in the models the factors: species, year, age, and migratory departure type (Old Cut vs. postrelocation), the covariate day of year, and tailwind component as a time-dependent covariate. Tailwind component was included because warblers generally prefer to migrate in the evening with tailwinds (e.g., Dossman et al. 2015). To model a time-dependent covariate, each bird was assigned the value of tailwind on each day it was at risk of departure (Allison, 1995, p. 144). Using package RNCEP in R (Kemp et al. 2012), we first extracted the u and v wind components (wind speeds in m·s−1 in the easterly [90°] and northerly direction [0°], respectively). This was done for the 925 mb pressure level (750–850 m altitude; realistic for migrating warblers [e.g., Mitchell et al. 2015]). Using the function NCEP.interpol, we used linear interpolation to interpolate 925 mb wind conditions at the coordinates at Old Cut with an assumed departure time of 21:00. We estimated the tailwind component as Vw × cos(β), where Vw is wind speed (m·s−1) and β is the difference between flight bearing and wind direction (Safi et al. 2013). We assumed a flight bearing of 21°, which was the average of year-specific estimates of inter-tower flight bearings (see Results section). Fuel stores have the potential to influence length of stay at stopover (Schaub et al. 2008; Goymann et al. 2010; Dossman et al. 2015), and so we tested whether any single morphometric covariate (weight, tarsus, wing, fat mass, and lean mass) was significant when added to the PHREG model. We also used principal component analysis on the standardized values of weight, tarsus, wing, fat mass, and lean mass to obtain independent indices of size (PC1M; M refers to morphology) and/or shape (PC2M) at capture (Tabachnick and Fidell 2007). We tested for significant effects of PC1M and PC2M in the PHREG model. Finally, we used the LIFETEST and LIFEREG procedures in SAS to evaluate the shape of the time to event function (e.g., exponential, Weibull, or gamma) using methods outlined in Allison (1995). Postdeparture movement We used detections from Motus receiving stations to map flight paths and determine the number of days birds stopped over within the array following their departure from the Old Cut region. We also used these detections to estimate inter-tower (i to i + 1) ground speeds (flight distance·flight time−1). In most cases, postdeparture detections occurred at night, and the relationship between signal strength and time showed a characteristic parabola shape indicative of a tagged bird flying through the beam of a directional antenna. We selected the time of maximum signal strength at a tower as the time of passage. In some cases, we had to filter out longer periods of detection. We estimated the bearing and distance moved between towers using functions in the R package argosfilter. We filtered out flight times >0.5 days to exclude stopovers between flights. To remove unrealistic ground speeds, we filtered out flight distances <20 km, which were within the detection radius of multiple towers, and ground speeds >100 m·s−1. Ground speed was compared between species and sexes using linear mixed models (the MIXED procedure in SAS) with id as a random factor and the fixed factors sex and species. We also estimated tailwind component (m·s−1) of these flights using package RNCEP as described above, but using the latitude and longitude at tower i, time at tower i, and the observed bearing from tower i to tower i + 1. Air speed was calculated as ground speed − tailwind component. Timing of diel activity We developed a method to assign the onset and end of diel activity based on variability in signal strength. Inactivity can be identified by low temporal variability in signal strength, and typically occurred throughout the night (Figure 1). The standard deviation of signal strength (sig) was first estimated for each antenna and every 15 min period from 0:00 to 23:59. The onset of diel activity and the end of diel activity was assigned as the first (or last) period when the standard deviation of signal strength exceeded 2 on at least 1 antenna. To avoid spurious standard deviations resulting from low sample sizes, we only calculated these metrics when there were more than 20 observations per period both before and after the onset (or end) of diel activity. The onset of diel activity was standardized relative to civil twilight, a time typically associated with the onset of avian activity (i.e., the dawn chorus); the end of diel activity was standardized relative to sunset, because the end of avian activity (i.e., the dusk chorus) tends to be more variable and not necessarily associated with twilight (Da Silva et al. 2014; Gil et al. 2015). The onset and end of diel activity were analyzed using linear mixed models (the MIXED procedure using maximum likelihood) with id as a random factor, the fixed factors sex, species, year, and age, and the covariate day of year. To test for the inclusion of 2-way interactions, we used LRTs to compare nested models with and without interactions. Figure 1 View largeDownload slide Detections of a single black-throated blue warbler on 3 antennas at the Old Cut receiver station during its stopover at Long Point, Ontario, showing low signal strength variability at night and high signal strength variability during the day. On each day the bird was detected, time of day is shown as a proportion. Occasional gaps in the activity traces can also be seen. Figure 1 View largeDownload slide Detections of a single black-throated blue warbler on 3 antennas at the Old Cut receiver station during its stopover at Long Point, Ontario, showing low signal strength variability at night and high signal strength variability during the day. On each day the bird was detected, time of day is shown as a proportion. Occasional gaps in the activity traces can also be seen. Plasma metabolite analysis In refueling birds, triglyceride levels in the blood plasma rise, while β-OH-butyrate levels fall. Thus, levels of these 2 plasma metabolites provide individual-level information about refueling performance (Jenni-Eiermann and Jenni 1994; Jenni and Jenni-Eiermann 1996; Williams et al. 1999; Schaub and Jenni 2001; Guglielmo et al. 2005; Zajac et al. 2006). To test the hypothesis that sexes differ in refueling rate at stopover, we collected blood samples for metabolite analysis from 64 magnolia warblers, 40 American redstarts, and 32 common yellowthroats during May 2014 and May 2015. Birds were captured by mist net at Old Cut Research Station, and were selected for the study to meet target sample sizes by species, sex, and capture date. Open mist nets were monitored continuously, and birds were brought back to a research trailer for immediate blood collection (time elapsed between capture and blood sampling = 7.9 ± 2.2 min [n = 136]). Blood was collected into heparinized capillary tubes via brachial venipuncture with a 26-gauge needle. Birds were measured for body weight (g), tarsus length (mm), and wing length (mm), and were scanned by QMR to estimate fat and lean mass. For common yellowthroats, sex was determined by plumage. Birds were banded with aluminum tags and released. To determine if capture dates of birds in the refueling study were similar between sexes as intended, we used species- and year-specific GLMs with the factors sex and age. We also assessed whether body condition (fat mass, lean mass, and weight) differed between years after controlling for sex and age using species-specific GLMs with the factors year, sex, and age. In the field, blood samples were centrifuged for 5 min and the plasma was stored in cryotubes in a liquid nitrogen dry shipper (Taylor-Wharton CX-100). Later, samples were transferred to a −80 °C freezer for short-term storage. Following laboratory methods outlined in Guglielmo et al. (2005), plasma was analyzed for triglyceride, glycerol, and β-OH-butyrate (mmol·L−1). Triglyceride (after correction for free glycerol) and β-OH-butyrate levels (log[x] + 1, mmol·L−1) were analyzed with principal component analysis to derive a refueling index (PC1R). The refueling index was analyzed using the MIXED procedure with the factors year, species, age, and sex and the covariates day of year, time of day, bleed time (elapsed time between capture and blood collection), and weight. To test for the inclusion of 2-way interactions, we used LRTs to compare nested models with and without interactions with the factors year, species, age, and sex. DNA was extracted from the red blood cells of American redstarts and magnolia warblers, and the new DNA primers were used to genotype sex (Supplementary Material). RESULTS Site-level protandry Based on banding records at Old Cut Research Station, 95% of birds had first capture dates ranging over a 3–4 week period from early May to early June. Males had first capture dates that were significantly earlier than females in black-throated blue warblers, magnolia warblers, American redstarts (in 1 year only), and common yellowthroats (Table 1). In all species and years, ASY birds arrived significantly earlier than SY birds (Table 1). In 2 species, there was evidence of an age × year interaction (magnolia warbler: LRT, likelihood ratio χ12 = 14.9, P = 0.0001; American redstart: likelihood ratio χ12 = 6.0, P = 0.014), so years were analyzed separately in these species. Table 1 Results of GLMs showing sex, age, and year differences in spring first capture dates of black-throated blue warblers (BTBW), magnolia warblers (MAWA, separately by year), American redstarts (AMRE, separately by year), and common yellowthroats (COYE) at Long Point, Ontario Species Sex Age Year BTBW F1,140 = 70.5, P < 0.0001, βM = −6.6 ± 0.8 F1,140 = 33.9, P < 0.0001, βSY = 4.5 ± 0.8 F1,140 = 1.2, P = 0.3, β2014 = 1.0 ± 0.9 MAWA 2014 F1,486 = 14.6, P = 0.0001, βM = −1.9 ± 0.5 F1,486 = 11.7, P = 0.0001, βSY = 1.8 ± 0.5 — MAWA 2015 F1,360 = 26.1, P < 0.0001, βM = −3.7 ± 0.7 F1,360 = 49.2, P < 0.0001, βSY = 5.0 ± 0.7 — AMRE 2014 F1,234 = 7.6, P = 0.007, βM = −2.0 ± 0.7 F1,234 = 22.9, P < 0.0001, βSY = 3.4 ± 0.7 — AMRE 2015 F1,129 = 0.7, P = 0.4, βM = −0.8 ± 1.0 F1,129 = 32.8, P < 0.0001, βSY = 6.0 ± 1.1 — COYE F1,657 = 15.0, P < 0.0001, βM = −2.3 ± 0.6 F1,657 = 33.8, P < 0.0001, βSY = 4.4 ± 0.6 F1,657 = 19.3, P < 0.0001, β2014 = 2.4 ± 0.6 Species Sex Age Year BTBW F1,140 = 70.5, P < 0.0001, βM = −6.6 ± 0.8 F1,140 = 33.9, P < 0.0001, βSY = 4.5 ± 0.8 F1,140 = 1.2, P = 0.3, β2014 = 1.0 ± 0.9 MAWA 2014 F1,486 = 14.6, P = 0.0001, βM = −1.9 ± 0.5 F1,486 = 11.7, P = 0.0001, βSY = 1.8 ± 0.5 — MAWA 2015 F1,360 = 26.1, P < 0.0001, βM = −3.7 ± 0.7 F1,360 = 49.2, P < 0.0001, βSY = 5.0 ± 0.7 — AMRE 2014 F1,234 = 7.6, P = 0.007, βM = −2.0 ± 0.7 F1,234 = 22.9, P < 0.0001, βSY = 3.4 ± 0.7 — AMRE 2015 F1,129 = 0.7, P = 0.4, βM = −0.8 ± 1.0 F1,129 = 32.8, P < 0.0001, βSY = 6.0 ± 1.1 — COYE F1,657 = 15.0, P < 0.0001, βM = −2.3 ± 0.6 F1,657 = 33.8, P < 0.0001, βSY = 4.4 ± 0.6 F1,657 = 19.3, P < 0.0001, β2014 = 2.4 ± 0.6 Shown in each model are the F statistics, P-values, and coefficients (β ± SE) for the effects of sex (first capture dates of males relative to females; that is, the amount of protandry in days), age (SY relative to ASY), and year (2014 relative to 2015). Years are shown separately when there was a significant age × year interaction. View Large Characteristics of birds in the radio telemetry study We assessed sex for all black-throated blue warblers using plumage characteristics. Because of insufficient recovery of DNA from some feathers (mostly those that were stored at room temperature since 2014) only 82% of the radio-tracked magnolia warblers could be sexed using molecular genetic markers. We were not able to match capture date by sex in both species and years because of site-level protandry (Table 1) and difficulties in reliably sexing magnolia warblers by plumage in the field. As a result, in black-throated blue warblers captured for radio tagging, males were captured at a similar time to females in 2015 (GLM: βM = −3.9 ± 2.5 days, F1,19 = 2.6, P = 0.13), but earlier than females in 2014 (βM = −2.5 ± 1.2 days, F1,31 = 4.7, P = 0.037). In magnolia warblers, males had capture dates that were similar to females in 2014 (βM = −0.1 ± 0.4 days, F1,14 = 0.08, P > 0.5) but earlier than females in 2015 (βM = −7.8 ± 1.4 days, F1,65 = 31.4, P < 0.0001). Although no effort was made to match capture date by age group, SY birds were captured later than ASY birds only in the 2014 black-throated blue warblers (βSY = 2.6 ± 1.2 days; F1,31 = 5.3, P = 0.029). In black-throated blue warblers, neither fat mass, lean mass, nor weight differed between years after controlling for sex and age (GLM: Ps > 0.5; Supplementary Material Table S1). In magnolia warblers, neither fat mass nor weight differed between years after controlling for sex and age (Ps > 0.090), but lean mass was significantly lower in 2014 than in 2015 (β2014 = −0.24 ± 0.09 g, F1,67 = 6.5, P = 0.013). However, this effect was small in size (i.e., 4% of mean lean mass). Stopover duration Of the 54 radio tagged black-throated blue warblers in the 2 years, 48 had sufficient detection data to be assigned a stopover duration (median = 3.5 days, range = 1–18 days). Of the 101 radio tagged magnolia warblers, 92 could be assigned a stopover duration (median = 3 days, range = 1–13 days). We note that these stopover durations are likely underestimates because of uncertainty in when birds arrived. Departure type (Old Cut vs. Postrelocation) could be assigned for 42 black-throated blue warblers and 83 magnolia warblers, and the frequency of departure type was independent of species ( χ12 = 2.2, P = 0.13). Combining species, 86% of individuals (n = 108) had migratory departures from Old Cut and 14% of individuals (n = 17) relocated within 20 km of Old Cut prior to departing the region. Median departure time from Old Cut was 1.7 h after sunset (n = 86, range = 0.7–5.9 h after sunset). In the proportional hazard (PH) model of stopover duration without body condition, the hazard for departure depended on tailwind at 925 mb, year, age, and departure type, but not on species, sex, or day of year (Table 2). After controlling for these covariates, the hazard of departure increased by 7% for each one m·s−1 increase in tailwind at 925 mb, indicating that birds were sensitive to winds aloft when making departure decisions. The hazard of departure for 2014 birds was 39% of the hazard for 2015 birds, indicating shorter stopover durations in 2015. The hazard of departure for migratory departures following relocation was 47% of the hazard for migratory departures from Old Cut, indicating longer stopovers for birds that first relocated. The hazard of departure for SY birds was 48% of the hazard for ASY birds, indicating longer stopovers by SY birds than ASY birds. When we fit separate PH models of stopover duration for each age class, tailwind selectivity was significant for ASY birds (HR = 1.10, P = 0.032) but not for SY birds (HR = 1.04, P = 0.27). The Weibull model with scale parameter (σ) = 0.60 ± SE 0.05 was the best-fitting function for stopover duration. In this range (0.5 < σ < 1), the hazard for departure increases at a decreasing rate with time since arrival (Allison 1995). Table 2 Parameter coefficients (β ± SE) of the Cox regression model of stopover duration in black-throated blue (BTBW) and magnolia warblers at Long Point, Ontario during spring migration Variable (level) β ± SE χ2 P-value HR Sex (M) 0.07 ± 0.23 0.09 >0.5 1.07 Year (2014) −0.94 ± 0.26 13.7 0.0002 0.39 Day of year −0.00 ± 0.02 0.03 >0.5 1.00 Tailwind 0.07 ± 0.03 6.3 0.012 1.07 Species (BTBW) −0.31 ± 0.26 1.4 0.2 0.73 Age (SY) −0.73 ± 0.21 12.2 0.001 0.48 Departure type (postrelocation) −0.75 ± 0.31 5.8 0.016 0.47 Variable (level) β ± SE χ2 P-value HR Sex (M) 0.07 ± 0.23 0.09 >0.5 1.07 Year (2014) −0.94 ± 0.26 13.7 0.0002 0.39 Day of year −0.00 ± 0.02 0.03 >0.5 1.00 Tailwind 0.07 ± 0.03 6.3 0.012 1.07 Species (BTBW) −0.31 ± 0.26 1.4 0.2 0.73 Age (SY) −0.73 ± 0.21 12.2 0.001 0.48 Departure type (postrelocation) −0.75 ± 0.31 5.8 0.016 0.47 Also shown are the chi-square values, P-values, and hazard ratio (HR). View Large None of the morphometric covariates (including PC1M and PC2M) was significant when added to the model presented in Table 2 (Ps > 0.25). In the PCA of weight, tarsus, wing length, fat mass, and lean mass, the first 2 principal components accounted for 72% and 77% of the variance in each species. PC1M (proportion of variance explained = 45% and 49% in each species, respectively) gave an overall index of body size, with positive loadings for each variable. PC2M (proportion of variance explained = 27% and 28% in each species, respectively) gave an overall index of condition, with negative loadings for weight and fat mass and high loadings for tarsus, wing length, and lean mass. Thus, there was no evidence that fuel stores at capture influenced time to departure from stopover. Postdeparture flight paths appeared to be similar between males and females within years (Figure 2). For birds detected beyond the Old Cut region (n = 124), 67% (n = 83) showed uninterrupted (same night) flights within the array, and the remaining 33% (n = 41) showed interrupted (multi-night) flights within the array, stopping over within the array for an additional 1–11 days (median = 2 days). Flight metrics were calculated for 16 inter-tower movements in 2014, but this was too few to compare between species or sexes by linear mixed models (Table 3). According to nonparametric Wilcoxon rank sum tests, ground speeds were similar between species and sexes in 2014 (Ps > 0.5). Flight metrics were calculated for 69 inter-tower movements in 2015. According to linear mixed models, ground speeds were similar between species (linear mixed model: F1,35 = 3.2, P = 0.08) and sexes (F1,35 = 0.04, P > 0.5). Figure 2 View largeDownload slide Postdeparture flight paths of black-throated blue (BTBW) and magnolia warblers (MAWA) from Long Point, Ontario during spring migration in 2014 and 2015. Black dots show active receiver stations without detections. Birds were detected beyond the Old Cut region up to 11 nights after departure. Figure 2 View largeDownload slide Postdeparture flight paths of black-throated blue (BTBW) and magnolia warblers (MAWA) from Long Point, Ontario during spring migration in 2014 and 2015. Black dots show active receiver stations without detections. Birds were detected beyond the Old Cut region up to 11 nights after departure. Table 3 Median values of inter-tower flight metrics and bearings for radio tagged black-throated blue (BTBW) and magnolia warblers (MAWA) in southern Ontario in 2014 and 2015 Species Year No. flights Ground speed (m·s−1) Tailwind component (m·s−1) Air speed (m·s−1) Bearing (°) BTBW 2014 11 14.4 (10.3, 20.1) 2.3 (1.3, 2.5) 12.3 (8.1, 19.9) 13.0 (−4.2, 31.8) BTBW 2015 m 8 12.4 (7.0, 14.0) 6.1 (3.7, 6.7) 5.4 (1.5, 10.7) 12.5 (7.0, 20.8) 2015 f 6 12.6 (11.8, 13.5) 7.2 (6.8, 8.0) 4.7 (3.7, 6.4) 11.3 (7.1, 15.8) MAWA 2014 5 17.9 (15.1, 26.6) 0.6 (0.5, 2.1) 17.4 (10.0, 24.5) 28.3 (28.3, 29.2) MAWA 2015 m 19 16.1 (14.7, 19.8) 3.4 (2.6, 5.2) 12.4 (11.8, 16.4) 7.2 (2.8, 42.7) 2015 f 36 17.1 (12.9, 22.6) 4.8 (2.3, 8.1) 11.7 (9.6, 16.1) 38.6 (10.8, 62.2) Species Year No. flights Ground speed (m·s−1) Tailwind component (m·s−1) Air speed (m·s−1) Bearing (°) BTBW 2014 11 14.4 (10.3, 20.1) 2.3 (1.3, 2.5) 12.3 (8.1, 19.9) 13.0 (−4.2, 31.8) BTBW 2015 m 8 12.4 (7.0, 14.0) 6.1 (3.7, 6.7) 5.4 (1.5, 10.7) 12.5 (7.0, 20.8) 2015 f 6 12.6 (11.8, 13.5) 7.2 (6.8, 8.0) 4.7 (3.7, 6.4) 11.3 (7.1, 15.8) MAWA 2014 5 17.9 (15.1, 26.6) 0.6 (0.5, 2.1) 17.4 (10.0, 24.5) 28.3 (28.3, 29.2) MAWA 2015 m 19 16.1 (14.7, 19.8) 3.4 (2.6, 5.2) 12.4 (11.8, 16.4) 7.2 (2.8, 42.7) 2015 f 36 17.1 (12.9, 22.6) 4.8 (2.3, 8.1) 11.7 (9.6, 16.1) 38.6 (10.8, 62.2) Also shown for each variable is the interquartile range. View Large Timing of diel activity We extracted 226 values for the onset of diel activity from 68 birds in the 2 years. The onset of diel activity depended on the species (βBTBW = −6.3 ± 1.8 min, F1,154 = 12.7, P = 0.001), year (β2014 = 7.8 ± 2.3 min, F1,154 = 8.1, P = 0.005), sex (βM = −7.0 ± 2.5 min, F1,154 = 5.3, P = 0.023), with evidence for a sex × year interaction (LRT, likelihood ratio χ12 = 3.8, P = 0.043). The onset of diel activity did not depend on age (F1,154 = 0.03, P > 0.5) or day of year (F1,154 = 0.4, P > 0.5). In this model, the random factor id accounted for 19.1% of the variance. When we fit separate models for each year, the effect of sex was not significant in 2014 (βM = −1.0 ± 2.2, F1,118 = 0.21, P > 0.5), but males began diel activity earlier than females in 2015 (βM = −8.1 ± 2.7 min, F1,35 = 9.4, P = 0.004; Figure 3). In summary, black-throated blue warblers began diel activity earlier in the day than magnolia warblers, both species began diel activity earlier in 2015 than in 2014, and males of both species began diel activity earlier than females in 2015 but not in 2014. Figure 3 View largeDownload slide A jitter plot comparing the onset of diel activity by sex, year, and species (BTBW = black-throated blue warbler; MAWA = magnolia warbler) at Long Point, Ontario. Figure 3 View largeDownload slide A jitter plot comparing the onset of diel activity by sex, year, and species (BTBW = black-throated blue warbler; MAWA = magnolia warbler) at Long Point, Ontario. We extracted 286 values for the end of diel activity from 92 birds in the 2 years. The end of diel activity depended on the species (βBTBW = 10.3 ± 3.7 min, F1,190 = 10.0, P = 0.002), sex (βM = 15.9 ± 4.9 min, F1,190 = 8.2, P = 0.005), day of year (β = 0.6 ± 0.3 min, F1,190 = 4.5, P = 0.036), and the sex × age interaction (LRT, likelihood ratio χ12 = 6.7, P = 0.010), but not on year (F1,190 = 1.2, P = 0.27) or age (F1,190 = 0.0, P > 0.5). In this model, the random factor id accounted for 14.9% of the variance. When we fit separate models to the 2 age groups, we found no evidence for effects of species, sex, year, or day of year for SY birds (Ps > 0.12). In ASY birds, we found no evidence for an effect of day of year (F1,78 = 2.1, P = 0.15) but evidence for effects of species (βBTBW = 19.3 ± 4.8 min, F1,78 = 16.1, P = 0.0001) and sex (βM = 17.7 ± 4.8 min, F1,78 = 13.8, P = 0.0004; Figure 4). We also found a species × age interaction that improved the model with main effects only (likelihood ratio χ12 = 3.9, P = 0.048), but this interaction did not improve the model described above (likelihood ratio χ12 = 2.2, P = 0.14) and thus was not considered further. In summary, in ASY birds, black-throated blue warblers ended diel activity later than magnolia warblers and females of both species ended diel activity earlier in the day than males. Figure 4. View largeDownload slide A jitter plot comparing the end of diel activity by sex, species (BTBW = black-throated blue warbler; MAWA = magnolia warbler), and age (ASY = after second year; SY = second year) at Long Point, Ontario. Figure 4. View largeDownload slide A jitter plot comparing the end of diel activity by sex, species (BTBW = black-throated blue warbler; MAWA = magnolia warbler), and age (ASY = after second year; SY = second year) at Long Point, Ontario. Characteristics of birds in the refueling study All American redstarts and magnolia warblers in the refueling study were accurately sexed by the molecular genetic method. We were not able to match capture date by sex in each species and year. As a result, male magnolia warblers were captured earlier than females in 2014 (βM = −2.4 ± 1.0 days, F1,41 = 5.7, P = 0.022) and 2015 (βM = −9.1 ± 2.1 days, F1,15 = 18.4, P = 0.0006). Second year magnolia warblers were captured later than ASY birds in 2014 (βSY = 3.1 ± 0.9 days, F1,41 = 10.4, P = 0.002) but not in 2015 (F1,15 = 4.2, P = 0.058). In American redstarts and common yellowthroats, capture dates were similar between sexes in the 2 years (Ps > 0.093), and were similar between age groups in the 2 years (Ps > 0.052). In American redstarts and magnolia warblers, fat mass, lean mass, and weight were similar between years after controlling for sex and age (GLM: Ps > 0.075; Supplementary Material Table S1). In common yellowthroats, lean mass and weight were similar between years (Ps > 0.27), but birds had significantly more fat in 2014 than in 2015 (β2014 = 0.5 ± 0.2 g, F1,25 = 5.1, P = 0.032, Table S1). Refueling rates We obtained triglyceride levels for 63 magnolia warblers (1.70 ± SE 0.10 mmol·L−1), 40 American redstarts (2.01 ± 0.13 mmol·L−1), and 30 common yellowthroats (1.95 ± 0.30 mmol·L−1). We obtained β-OH butyrate levels for 60 magnolia warblers (0.80 ± 0.05 mmol·L−1), 38 American redstarts (0.73 ± 0.08 mmol·L−1), and 29 common yellowthroats (1.00 ± 0.08 mmol·L−1). The refueling index (PC1R) could be derived for 60 magnolia warblers, 38 American redstarts, and 28 common yellowthroats. PC1R accounted for 76% of the variance in metabolites, with a positive loading for log(triglyceride) and a negative loading for log(β-OH butyrate). Due to evidence of a highly significant year × species interaction in the analysis of the refueling index (likelihood ratio χ22 = 16.7, P = 0.0002), we fit separate models for each species. In magnolia warblers, the refueling index was higher in 2015 and increased with day of year (Table 4). There was some evidence for a sex × year interaction (likelihood ratio χ12 = 3.8, P = 0.051), but no sex effect was apparent when separate models were fit by year (2014, t10 = −1.0, P = 0.35; 2015, t4 = −2.3, P = 0.083). Table 4 Summary of the GLM of refueling index in magnolia warblers (MAWA), American redstarts (AMRE), and common yellowthroats (COYE) at Long Point, Ontario Variable MAWA AMRE COYE Year (2014) t50= −3.7, P = 0.0005 t27= −3.8, P = 0.0009 t19= 2.9, P = 0.009 Sex (M) t50 = 1.7, P = 0.10 t27= 2.4, P = 0.027 t19 = 0.68, P > 0.5 Age (SY) t50 = −0.5, P > 0.5 t27 = −1.8, P = 0.089 t19 = 1.0, P = 0.32 Day of year t50= 4.2, P = 0.0001 t27 = 1.5, P = 0.14 t19= 2.2, P = 0.039 Time of day t50 = 1.8, P = 0.077 t27= 6.1, P < 0.0001 t19= 3.5, P = 0.002 Bleed time t50 = 0.6, P > 0.5 t27 = −1.4, P = 0.17 t19 = −1.2, P = 0.26 Weight t50 = −0.3, P > 0.5 t27 = 1.5, P = 0.15 t19 = −1.1, 0.28 Sex × year F1,50 = 3.9, P = 0.055 — — Sex × age — F1,27= 6.8, P = 0.015 — Age × year — F1,27 = 4.1, P = 0.053 F1,19= 4.8,P= 0.041 Variable MAWA AMRE COYE Year (2014) t50= −3.7, P = 0.0005 t27= −3.8, P = 0.0009 t19= 2.9, P = 0.009 Sex (M) t50 = 1.7, P = 0.10 t27= 2.4, P = 0.027 t19 = 0.68, P > 0.5 Age (SY) t50 = −0.5, P > 0.5 t27 = −1.8, P = 0.089 t19 = 1.0, P = 0.32 Day of year t50= 4.2, P = 0.0001 t27 = 1.5, P = 0.14 t19= 2.2, P = 0.039 Time of day t50 = 1.8, P = 0.077 t27= 6.1, P < 0.0001 t19= 3.5, P = 0.002 Bleed time t50 = 0.6, P > 0.5 t27 = −1.4, P = 0.17 t19 = −1.2, P = 0.26 Weight t50 = −0.3, P > 0.5 t27 = 1.5, P = 0.15 t19 = −1.1, 0.28 Sex × year F1,50 = 3.9, P = 0.055 — — Sex × age — F1,27= 6.8, P = 0.015 — Age × year — F1,27 = 4.1, P = 0.053 F1,19= 4.8,P= 0.041 t-Statistics are shown for each factor (level) and continuous variable (to show directionality), whereas F-statistics are shown for significant interactions based on LRTs. Significant effects are shown in bold. View Large In American redstarts, the refueling index was higher in 2015, higher in males, and increased with time of day. However, a model with a sex × age interaction improved the model without interactions (likelihood ratio χ12 = 6.9, P = 0.009), and the former model was further improved by including a year × age interaction (likelihood ratio χ12 = 3.9, P = 0.048). When we fit separate models for each age group, an effect of sex was not apparent (SY: t12 = 0.8, P = 0.45; ASY: t11 = −1.6, P = 0.15). When we fit separate models for each year to further explore the year × age interaction, SY birds had a higher refueling index than ASY birds in 2014 (t23 = 2.9, P = 0.009); sample sizes were insufficient to test for age effects in 2015. In common yellowthroats, the refueling index was lower in 2015, increased with day of year, and increased with time of day. A model with an age × year interaction improved the model without interactions (likelihood ratio χ12 = 4.4, P = 0.036). When separate models were fit for each year, SY birds had a lower refueling index than ASY birds in 2014 (t10 = −2.6, P = 0.027), but a higher refueling index than ASY birds in 2015 (t4 = 3.10, P = 0.036). Summarizing main effects across the 3 species, the refueling index was higher 2015 than in 2014 for the 2 species with larger sample sizes (magnolia warblers and American redstarts). Sex and age effects were generally absent or inconsistent. DISCUSSION Our study suggests that sex differences in stopover behavior are subtle, context dependent, and may not play a consistent or dominant role in determining protandry, at least not in wood-warblers. Sexes were similar in many aspects of their stopover behavior in 2014 and 2015. In both black-throated blue warblers and magnolia warblers, males and females had similar departure probabilities and stopover durations. In magnolia warblers, American redstarts, and common yellowthroats, plasma metabolite analysis suggest similar refueling rates between sexes. Others have suggested that earlier departure from wintering areas may be the most important mechanism of protandry (Coppack and Pulido 2009; Dossman et al. 2015) and indeed, there is evidence that earlier departure of males is genetically determined (Maggini and Bairlein 2012). From an evolutionary point of view, perhaps protandry in a single event (departure) can reliably achieve an evolutionarily stable degree of protandry at breeding areas. In contrast, reliance on intrinsic sex differences in refueling rates and/or stopover duration throughout migration, especially when stopover sites vary in quality and whether they are used, might result in high variability in protandry among years and therefore mismatched timing relative to optimal protandry. This does beg the question, however, of why males would sometimes refuel more quickly than females during migratory stopover. We detected some sex differences in the timing of diel activity. Male black-throated blue and magnolia warblers began diel activity 8 min earlier than females in 2015, and adult males ended diel activity 14 min later than adult females in both years. Given the different weather at Long Point in 2014 and 2015, these results suggest that sex differences in stopover behavior might depend on environmental conditions. For example, better weather conditions in 2015 may have been more permissive for an earlier onset of foraging in males than females. In other studies, sexes have been shown to differ in their habitat use during stopover possibly due to weather or other factors (Yong et al. 1998; Smith et al. 2007), and it seems possible that habitat-specific prey availability also could vary among years. In the future, the role of local environmental conditions on sex-specific patterns of habitat use and foraging behavior warrants further study if we are to understand when, how, and why sexes differ in migration. Our study does not rule out sex differences in stopover behavior earlier in migration, or sex differences in others aspects of migration. Indeed, site-level protandry indicates that males and females must have differed in their departure timing or their behavior en route to Long Point. Coppack and Pulido (2009) suggest several mechanisms of protandry in addition to the main 3 (sex-specific departure timing, wintering latitude, and migration rate), including sex-specific flight speed, flight distance, and flight route. The importance of sex-specific flight speed is evident in northern wheatears (Oenanthe oenanthe), in which differences in protandry between subspecies are associated with differences in the degree of sexual dimorphism in wing length and shape (Schmaljohann et al. 2016). In black-throated blue warblers and magnolia warblers, we found no evidence that the sexes differed in their postdeparture migration routes or flight speeds within the confines of the array (about 200 km distant) and given the error associated with estimating flight distances with automated radio telemetry data. However, males have longer wings than females of similar weight or size (tarsus length; results not shown), which is consistent with sex-specific flight speeds that might be realized over longer migratory flight distances. The 2 years of the study differed in weather and in several aspects of warbler stopover behavior, suggesting that the 2 are linked. During the study, average hourly temperatures in the area (measured at Bird Studies Canada Headquarters, about 5 km N of the capture site) were warmer in 2015 (April: 6.5 °C, May: 15.4 °C) than in 2014 (April: 6.0 °C, May: 13.3 °C). This reflected a broader regional and seasonal trend of a warmer spring in 2015. For example, average March–May temperatures in Ohio, the US state immediately to the south of Lake Erie, were higher in 2015 (10.4 °C) than in 2014 (9.4 °C; NOAA National Centers for Environmental information, Climate at a Glance, http://www.ncdc.noaa.gov/cag/). Lake Erie ice out was also earlier in 2015 (28 April) than in 2014 (11 May; NOAA Great Lakes Environmental Research Laboratory, https://www.glerl.noaa.gov/data/ice/). A warmer spring is consistent with an earlier seasonal availability of prey, both through direct effects on insect activity and through earlier green-up of deciduous trees. In insectivorous species, there is good evidence that long-distance migrants adjust their spring migration rate flexibly by tracking green-up and responding to local weather conditions (Schmaljohann et al. 2017; Thorup et al. 2017). In addition, warmer spring temperatures are associated with faster migration by song birds over long distances in North America (Marra et al. 2005). Thus, birds might be expected to have shorter stopovers in a warmer year. Compared to 2014, birds in 2015 had shorter stopovers (black-throated blue and magnolia warblers), higher refueling rates (magnolia warblers and American redstarts), and earlier onsets of diel activity (black-throated blue and magnolia warblers). Otherwise, the years were quite similar in terms of arrival timing and body condition (weight, fat mass, and lean mass) at arrival. Differences in behavior between years suggest that year-and site-specific environmental conditions may have played a role in influencing stopover decisions in the species we studied. For example, if there was more abundant insect prey in 2015 than in 2014, this might explain why some birds began foraging earlier in the day, refueled more quickly, and had shorter stopover durations in 2015 than in 2014. The connection between refueling and stopover is consistent with studies of captive birds, where faster refueling is correlated with increased migratory restlessness, a proxy for earlier departure (Eikenaar and Schläfke 2013; Eikenaar et al. 2014). We found that adult (ASY) birds arrived earlier on average than young (SY) birds in the 4 species (Table 1). Adult black-throated blue and magnolia warblers also had shorter stopovers and greater tailwind selectivity than SY birds. Other studies have also found an advancement of arrival timing with age, which was attributed to learning (Hake et al. 2003; Sergio et al. 2014; Schmaljohann et al. 2016), and greater tailwind selectivity among older birds (Mitchell et al. 2015). However, the timing of diel activity was similar between age groups, and SY and ASY magnolia warblers, American redstarts, and common yellowthroats had similar refueling rates. Other studies of Parulidae warblers have not seen age effects in refueling rates or patterns of mass gain (Yong et al. 1998; Woodrey 2000; Morris et al. 2003). Together, these results suggest that during spring migration young birds are capable of refueling as quickly as adults, but are generally less able to select favorable winds at departure or perhaps are under reduced time pressure to depart stopover sites. For birds at Old Cut, the hazard for departure increased at a decreasing rate with time since arrival. An increasing hazard function is expected at stopover, because as time advances birds should be eager to leave and continue migration (e.g., Dossman et al. 2015). However, a third of the birds detected beyond the Old Cut region stopped elsewhere within the array for an additional 1–11 days before continuing their journey. These movements are consistent with the concept of landscape-scale stopover movements (Taylor et al. 2011; Dossman et al. 2015), albeit at a broader scale. Our data also suggest that migratory movements (i.e., flights directed toward breeding areas) by wood-warblers may occur over fine temporal and spatial scales. Such movement patterns suggest that warblers “ride ripples of environmental change” (sensu van Moorter et al. 2013) in addition to riding the wave of green-up. Automated radio telemetry allowed us to obtain detailed information about individual migration behavior, including the timing of diel activity at stopover, stopover duration, departure timing, and postdeparture movements. We show that sexes differ in some aspects of their behavior at stopover, but differences were subtle and context dependent. Protandry must be driven by other sex differences in behavior, such as sex-specific departure timing from wintering areas, flight speeds (i.e., over long distances), flight duration, or migration routes. Tracking birds from their wintering grounds and throughout their migration at an increased spatial and temporal scale will aid in improving our understanding of avian migration behavior, including sex- and age-specific migration strategies. With improvements in tracking technology, such efforts are beginning to reveal extensive individual flexibility in migration behavior (Schmaljohann et al. 2017). SUPPLEMENTARY MATERIAL Supplementary data are available at Behavioral Ecology online. FUNDING Funding was provided by NSERC Discovery Grants to CGG and YEM, the Canada Foundation for Innovation, and the Ontario Research Fund. We would like to thank the staff and volunteers of Long Point Bird Observatory and Bird Studies Canada for their assistance in the field. We also thank additional members of our field crew, Taylor Brown, Jacopo Cecere, Alexander Macmillan, and Taylor Marshall. John Brzustowski provided essential technical assistance with data acquisition, and anonymous reviewers provided insightful comments to help improve the manuscript. Motus is a program of Bird Studies Canada in partnership with Acadia University and collaborating researchers and organizations. Data accessibility: Analyses reported in this article can be reproduced using the data provided by Morbey et al. (2017). REFERENCES Allison PD. 1995. Survival analysis using the SAS system: a practical guide . Cary (NC): SAS Institute, Inc. Bauböck L, Miller-Rushing AJ, Primack RB, Lloyd Evans TL, Wasserman FE. 2012. Climate change does not affect protandry in seven passerines in North America. Wilson J Ornithol . 124: 208– 216. Google Scholar CrossRef Search ADS Coppack T, Pulido F. 2009. Proximate control and adaptive potential of protandrous migration in birds. Integr Comp Biol . 49: 493– 506. Google Scholar CrossRef Search ADS PubMed Da Silva A, Samplonius JM, Schlicht E, Valcu M, Kempenaers B. 2014. Artificial night lighting rather than traffic noise affects the daily timing of dawn and dusk singing in common European songbirds. Behav Ecol . 25: 1037– 1047. Google Scholar CrossRef Search ADS Dossman BC, Mitchell GW, Norris DR, Taylor PD, Guglielmo CG, Matthews SN, Rodewald PG. 2015. The effects of wind and fuel stores on stopover departure behavior across a migratory barrier. Behav Ecol . 27: 567– 574. Google Scholar CrossRef Search ADS Eikenaar C, Klinner T, Szostek KL, Bairlein F. 2014. Migratory restlessness in captive individuals predicts actual departure in the wild. Biol Lett . 10:20140154. Eikenaar C, Schläfke JL. 2013. Size and accumulation of fuel reserves at stopover predict nocturnal restlessness in a migratory bird. Biol Lett . 9:20130712. Francis CM, Cooke F. 1986. Differential timing of spring migration in wood warblers (Parulinae). Auk . 103: 548– 556. Gil D, Honarmand M, Pascual J, Pérez-Mena E, Macías Garcia C. 2015. Birds living near airports advance their dawn chorus and reduce overlap with aircraft noise. Behav Ecol . 26: 435– 443. Google Scholar CrossRef Search ADS Goymann W, Spina F, Ferri A, Fusani L. 2010. Body fat influences departure from stopover sites in migratory birds: evidence from whole-island telemetry. Biol Lett . 6: 478– 481. Google Scholar CrossRef Search ADS PubMed Guglielmo CG, Cerasale DJ, Eldermire C. 2005. A field validation of plasma metabolite profiling to assess refueling performance of migratory birds. Physiol Biochem Zool . 78: 116– 125. Google Scholar CrossRef Search ADS PubMed Guglielmo CG, McGuire LP, Gerson AR, Seewagen CL. 2011. Simple, rapid, and non-invasive measurement of fat, lean, and total water masses of live birds using quantitative magnetic resonance. J Ornithol . 152( Suppl 1): 75– 85. Google Scholar CrossRef Search ADS Hake M, Kjellén N, Alerstam T. 2003. Age-dependent migration strategy in honey buzzards Pernis apiorus tracked by satellite. Oikos . 103: 385– 396. Google Scholar CrossRef Search ADS Jenni L, Jenni-Eiermann S. 1996. Metabolic responses to diurnal feeding patterns during the postbreeding, moulting and migratory periods in passerine birds. Funct Ecol . 10: 73– 80. Google Scholar CrossRef Search ADS Jenni-Eiermann S, Jenni L. 1994. Plasma metabolite levels predict individual body mass changes in a small long-distance migrant, the garden warbler. Auk . 111: 888– 899. Google Scholar CrossRef Search ADS Kanamitsu M, Ebisuzaki W, Woollen J, Yang S-K, Hnilo JJ, Fiorino M, Potter GL. 2002. NCEP-DOE AMIP-II Reanalysis (R-2). Bull Am Meteorol Soc . 83: 1631– 1643. Google Scholar CrossRef Search ADS Kemp MU, Emiel van Loon E, Shamoun-Baranes J, Bouten W. 2012. RNCEP: global weather and climate data at your fingertips. Methods Ecol Evol . 3: 65– 70. Google Scholar CrossRef Search ADS Kennedy LV, Morbey YE, Mackenzie SA, Taylor PD, Guglielmo CG. 2016. A field test of the effects of body composition analysis by Quantitative Magnetic Resonance on songbird stopover behaviour. J Ornithol . 158: 593– 601. Google Scholar CrossRef Search ADS Kissner KJ, Weatherhead PJ, Francis CM. 2003. Sexual size dimorphism and timing of spring migration in birds. J Evol Biol . 16: 154– 162. Google Scholar CrossRef Search ADS PubMed LPBO. 2005. Long point bird observatory: migration monitoring protocol . Bird Studies Canada, Port Rowan, ON, Canada. www.bsc-eoc.org/download/LPBOMigrationProtocol.pdf. Last accessed 8 September 2017. MacDade LS, Rodewald PG, Hatch KA. 2011. Contribution of emergent aquatic insects to refueling in spring migrant songbirds. Auk . 128: 127– 137. Google Scholar CrossRef Search ADS Maggini I, Bairlein F. 2012. Innate sex differences in the timing of spring migration in a songbird. PLoS One . 7: e31271. Google Scholar CrossRef Search ADS PubMed Marra PP, Francis CM, Mulvihill RS, Moore FR. 2005. The influence of climate on the timing and rate of spring bird migration. Oecologia . 142: 307– 315. Google Scholar CrossRef Search ADS PubMed Marra PP, Hobson KA, Holmes RT. 1998. Linking winter and summer events in a migratory bird by using stable-carbon isotopes. Science . 282: 1884– 1886. Google Scholar CrossRef Search ADS PubMed McKinnon EA, MacDonald CM, Gilchrist HG, Love OP. 2016. Spring and fall migration phenology of an Arctic-breeding passerine. J Ornithol . 157: 681– 693. Google Scholar CrossRef Search ADS Mitchell GW, Woodworth BK, Taylor PD, Norris DR. 2015. Automated telemetry reveals age specific differences in flight duration and speed are driven by wind conditions in a migratory songbird. Mov Ecol . 3: 19. Google Scholar CrossRef Search ADS PubMed Morbey YE, Coppack T, Pulido F. 2012. Adaptive hypotheses for protandry in arrival to breeding areas: a review of models and empirical tests. J Ornithol . 153: 207– 215. Google Scholar CrossRef Search ADS Morbey YE, Guglielmo CG, Taylor PD, Maggini I, Deakin J, Mackenzie SA, Brown JM, Zhao L. 2017. Data from: evaluation of sex differences in the stopover behavior and postdeparture movements of wood-warblers. Dryad Digital Repository . http://dx.doi.org/10.5061/dryad.19526. Morris SR, Glasgow JL. 2001. Comparison of spring and fall migration of American redstarts on Appledore Island, Maine. Wilson Bull . 113: 202– 210. Google Scholar CrossRef Search ADS Morris SR, Pusateri CR, Battaglia KA. 2003. Spring migration and stopover ecology of common yellowthroats on Appledore Island, Maine. Wilson Bull . 115: 64– 72. Google Scholar CrossRef Search ADS Myers RH. 1990. Classical and modern regression with applications . 2nd ed. Boston (MA): PWS-KENT. Newton I. 2008. The ecology of bird migration . London (UK): Academic Press. Otahal CD. 1995. Sexual differences in Wilson’s warbler migration. J Field Ornithol . 66: 60– 69. Pyle P. 1997. Identification guide to North American birds, Part 1 . Bolinas (CA): Slate Creek Press. R Core Team. 2016. R: a language and environment for statistical computing . Vienna (Austria): R Foundation for Statistical Computing. Rappole JH, Tipton AR. 1991. New harness design for attachment of radio transmitters to small passerines. J Field Ornithol . 63: 335– 337. Safi K, Kranstauber B, Weinzierl R, Griffin L, Rees EC, Cabot D, Cruz S, Proaño C, Takekawa JY, Newman SHet al. 2013. Flying with the wind: scale dependency of speed and direction measurements in modelling wind support in avian flight. Mov Ecol . 1: 4. Google Scholar CrossRef Search ADS PubMed SAS Institute, Inc. 2011. Base SAS® 9.3 procedures guide . Cary (NC): SAS Institute, Inc. Schaub M, Jenni L. 2001. Variation of fuelling rates among sites, days and individuals in migrating passerine birds. Funct Ecol . 15: 584– 594. Google Scholar CrossRef Search ADS Schaub M, Jenni L, Bairlein F. 2008. Fuel stores, fuel accumulation, and the decision to depart from a migration stopover site. Behav Ecol . 19: 657– 666. Google Scholar CrossRef Search ADS Schmaljohann H, Lisovski S, Bairlein F. 2017. Flexible reaction norms to environmental variables along the migration route and the significance of stopover duration for total speed of migration in a songbird migrant. Front Zool . 14: 17. Google Scholar CrossRef Search ADS PubMed Schmaljohann H, Meier C, Arlt D, Bairlein F, van Oosten H, Morbey YE, Åkesson S, Buchmann M, Chernetsov N, Desaever Ret al. 2016. Proximate causes of avian protandry differ between subspecies with contrasting migration challenges. Behav Ecol . 27: 321– 331. Google Scholar CrossRef Search ADS Seewagen CL, Guglielmo CG, Morbey YE. 2013. Stopover refueling rate underlies protandry and seasonal variation in migration timing of songbirds. Behav Ecol . 24: 634– 642. Google Scholar CrossRef Search ADS Sergio F, Tanferna A, De Stephanis R, Jiménez LL, Blas J, Tavecchia G, Preatoni D, Hiraldo F. 2014. Individual improvements and selective mortality shape lifelong migratory performance. Nature . 515: 410– 413. Google Scholar CrossRef Search ADS PubMed Smith RJ, Moore FR, May CA. 2007. Stopover habitat along the shoreline of northern Lake Huron, Michigan: emergent aquatic insects as a food resource for spring migrating landbirds. Auk . 124: 107– 121. Google Scholar CrossRef Search ADS Stanley CQ, MacPherson M, Fraser KC, McKinnon EA, Stutchbury BJ. 2012. Repeat tracking of individual songbirds reveals consistent migration timing but flexibility in route. PLoS One . 7: e40688. Google Scholar CrossRef Search ADS PubMed Tabachnick BG, Fidell LS. 2007. Using multivariate statistics . 5th ed. Toronto: Pearson Education. Taylor PD, Crewe TL, Mackenzie SA, Lepage D, Aubry Y, Crysler Z, Finney G, Francis CM, Guglielmo CG, Hamilton DJet al. 2017. The Motus Wildlife Tracking System: a collaborative research network to enhance the understanding of wildlife movement. Avian Conserv Ecol . 12: 8. Google Scholar CrossRef Search ADS Taylor PD, Mackenzie SA, Thurber BG, Calvert AM, Mills AM, McGuire LP, Guglielmo CG. 2011. Landscape movements of migratory birds and bats reveal an expanded scale of stopover. PLoS One . 6: e27054. Google Scholar CrossRef Search ADS PubMed Thorup K, Tøttrup AP, Willemoes M, Klaassen RH, Strandberg R, Vega ML, Dasari HP, Araújo MB, Wikelski M, Rahbek C. 2017. Resource tracking within and across continents in long-distance bird migrants. Sci Adv . 3: e1601360. Google Scholar CrossRef Search ADS PubMed van Moorter B, Bunnefeld N, Panzacchi M, Rolandsen CM, Solberg EJ, Sæther BE. 2013. Understanding scales of movement: animals ride waves and ripples of environmental change. J Anim Ecol . 82: 770– 780. Google Scholar CrossRef Search ADS PubMed Williams TD, Guglielmo CG, Egeler O, Martyniuk CJ. 1999. Plasma lipid metabolites provide information on mass change over several days in captive western sandpipers. Auk. 116: 994– 1000. Woodrey MS. 2000. Age-dependent aspects of stopover biology of passerine migrants. Stud Avian Biol . 20: 43– 52. Woodworth BK, Newman AE, Turbek SP, Dossman BC, Hobson KA, Wassenaar LI, Mitchell GW, Wheelwright NT, Norris DR. 2016. Differential migration and the link between winter latitude, timing of migration, and breeding in a songbird. Oecologia . 181: 413– 422. Google Scholar CrossRef Search ADS PubMed Yong W, Finch DM, Moore FR, Kelly JF. 1998. Stopover ecology and habitat use of migratory Wilson’s warblers. Auk . 115: 829– 842. Google Scholar CrossRef Search ADS Zajac RM, Cerasale DJ, Guglielmo CG. 2006. The rapid response of plasma metabolites to changes in feeding rate in a small passerine Wilson’s warbler (Wilsonia pusilla). J Avian Biol . 37: 405– 408. Google Scholar CrossRef Search ADS © The Author(s) 2017. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: email@example.com
Behavioral Ecology – Oxford University Press
Published: Jan 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera