Integrative and Comparative Biology Advance Access originally published online on June 2, 2008
Integrative and Comparative Biology 2008 48(1):134-151; doi:10.1093/icb/icn044
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
This article appears in the following Integrative and Comparitive Biology issue: Aeroecology: Probing and Modeling the Aerosphere–The Next Frontier [View the issue table of contents]
Wingbeat frequency and flap-pause ratio during natural migratory flight in thrushes
,
*Ecology and Conservation Science Division, Illinois Natural History Survey, Champaign-Urbana, IL 61820, USA; Division of Biological Sciences, 32 Campus Dr., University of Montana, Missoula, MT 59812, USA; Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ 08544, USA; Max Planck Institute for Ornithology, D-78315 Radolfzell, Germany
Correspondence: 1E-mail: wikelski{at}princeton.edu
 |
Synopsis |
|---|
Top
 Synopsis IntroductionMethodsResultsDiscussionAcknowledgmentsReferences |
|---|
Powered flapping flight has evolved independently in many different taxa. For flapping fliers, wingbeat parameters such as frequency and amplitude are the primary determinants of these animals’ energetic expenditure during flight. Here we present data on wingbeat frequency and amplitude for three New World thrush species during 15 entire nocturnal migratory flights over the Midwestern United States. Using continuous (non-pulsing) radio transmitters, we were able to measure wingbeat frequency and relative amplitude of wingbeats as well as the characteristics of flap-pauses. Contrary to previous telemetric findings, all of the individuals we followed used both flapping-only and flap-pause flight. During migratory flights, wingbeat frequency, effective wingbeat frequency, and amplitude were highest during initial ascent. Effective wingbeat frequency and amplitude were lowest during final descent. We show that identification of species based solely on characteristics of the wingbeat e.g., during radar studies, can be difficult because variables such as wingbeat frequency and amplitude, wingbeat pausing, and pattern of beats and pauses vary between individuals of the same species and even within individual flights. We also show that observed wingbeat frequencies were lower than those predicted by theoretical models. We speculate that this may be because theoretical predictions are generally based on (1) data from larger birds and (2) data from diurnal flights. We found that diurnal wingbeat frequencies of thrushes were generally higher than were those during nocturnal migratory flight. Finally, we suggest that rather than remaining at a single altitude during flight or climbing slightly as theoretical models predict, thrushes often moved up and down in the air column, perhaps searching for favorable atmospheric conditions.
|
Introduction |
|---|
|
Top
SynopsisIntroductionMethodsResultsDiscussionAcknowledgmentsReferences |
|---|
Flapping flight is an energetically expensive form of locomotion, at least compared to gliding or soaring (Schmidt-Nielsen 1997
), and is comparatively difficult to study in the wild (Liechti and Bruderer 2002). Nocturnal migratory flight in particular poses logistical problems for researchers as the birds are moving in the dark above the surface of the earth. However, Lord et al. (1962) developed a method of remotely sensing wingbeat frequency using radio telemetry. Alternatively, radars have been used for many years to measure the wingbeat frequency of passing birds (for recent summaries see e.g., Bruderer 1997; Alerstam and Hedenström 1998).
Wingbeat frequency is not the only variable that determines the expenditure of energy during flight. Among several other parameters, amplitude of wingbeat should also affect the power required to fly (Norberg 1990). However, of these variables researchers have focused primarily on wingbeat frequency because of the relative ease of measuring it via radar-tracking methods (Liechti and Bruderer 2002). The importance of wingbeat frequency during the migration of songbirds has recently been confirmed in the wild; wingbeat frequency is correlated with the use of energy during natural migratory flight (Bowlin et al. 2005).
Because of its importance for bird flight, wingbeat frequency has received a lot of theoretical attention. Pennycuick (1990) developed an equation to predict the wingbeat frequencies of birds from dimensional reasoning and later modified it based on empirical data (Pennycuick 1996, 2001). He argued that wingbeat frequency should depend on an individual bird's mass, wing area, wingspan, and the density of the air in which it is flying. Rayner (1995) used another approach, fitting equations to empirical data. His equation for predicting wingbeat frequency from body mass, wingspan and wing area is similar but not identical to that of Pennycuick (2001). Rayner (1985) also created theoretical models describing intermittent flight in birds based on the idea that muscles have optimal contraction frequencies and that this could lead to small birds varying their airspeeds and altitudes by increasing or decreasing the amount of time they stop flapping (pause) during flight, rather than slowing down or speeding up their wingbeat frequencies.
Most data on wingbeat frequency during natural nocturnal migration come from radar-tracking studies (Bruderer and Steidinger 1972; Bruderer 1997; Bruderer et al. 2001). The advantage of the radar method is that data from thousands of birds can be gathered in a single season (Bellrose and Graber 1968; Liechti and Bruderer 1995). Moreover, with automation (Bloch et al. 1981), these large data sets require less labor than the radio telemetry method, which is also limited to comparatively small samples. However, radar tracks of naturally migrating individuals are brief (typically < 5 min) and thus provide no information about variation in migratory behavior during entire flights whereas radio telemetry data are often from entire flights. Furthermore, with thoughtful but only circumstantial evidence, researchers often assign categories of radar signatures to classes of birds such as large, small, or medium passerines or waders and waterfowl (Bloch et al. 1981; Liechti and Bruderer 1995; Schmaljohann et al. 2007). In contrast, the characteristics of subjects studied by radio telemetry are known. The same is true for birds captured for release and subsequent observation by radar, but the methods used to release these birds influence their behavior (Emlen 1974; Vaughn 1974). The recorded flights are also not only short in terms of natural migratory flights but are not at times or under conditions chosen by the birds. The effect of carrying a radio transmitter on a bird's behavior is unknown (Cochran 1972); birds tracked by radar are unencumbered.
One conclusion derived from radar studies is that most passerine migrants except swallows use a style of flight in which individuals regularly alternate between short (ca. a few wingbeats) flapping and pausing phases (Bruderer et al. 2001). The opposite conclusion was reached for two passerine species in a radio telemetry study that reported only continuous flapping flight (Diehl and Larkin 1998). One of our goals was to determine whether Catharus and Hylocichla thrushes use flap-pause flight, flap continuously, or use both types of flight. To our knowledge, the latter possibility has not been previously considered in the wingbeat frequency literature.
In this article, we analyze the flight behavior of radio-tagged Catharus and Hylocichla thrushes. The analyses include variations in the frequency and intermittency of wingbeats within and between individuals. There are data gaps, but for the most part entire migratory flights were sampled with many periods of 40–60 min when all wingbeats and pauses between wingbeats were measured and counted. We hope that our detailed approach will help settle some long-standing issues and provide baseline data that will prove helpful for all researchers interested in the aeroecology of migrating animals.
|
Methods |
|---|
|
Top
SynopsisIntroductionMethodsResultsDiscussionAcknowledgmentsReferences |
|---|
Birds and radio-telemetric procedures
We obtained wingbeat (Fig. 1) and other data from nine Swainson's thrushes (Catharus ustulatus, S1–S9), two veery (Catharus fuscescens, V1 and V2) and two wood thrushes (Hylocichla mustelina, W1 and W2) during nocturnal migratory and diurnal foraging flights in 1975, 1981, 1994, 1997, 1998, and 2001 (for details see Table 1, Fig. 2). The migratory flights took place in the Midwestern United States over areas where Catharus thrushes are seen only during migration, but where wood thrushes breed. Birds were mist-netted, measured, and weighed. We attached the transmitters to the birds’ backs, between the shoulders near the center of gravity, with eyelash adhesive, and a cotton cloth buffer (Raim 1978
). Catharus thrushes with spotted or streaked wing coverts were classified as immature in both spring and autumn; absence of such markings indicated adult status in the autumn. The absence of spots on Catharus thrushes in the spring suggests that they were adults, but this is not conclusive so we classified these as unknown (Payne 1961, Pyle 1997). During migratory flight thrushes were followed with a radio-tracking vehicle and flight paths (Fig. 2) and altitude were obtained using methods described in detail elsewhere (Cochran and Kjos 1985). General characteristics of the two flights in autumn (Fig. 2G) and 13 in the spring were typical of results previously published about flights of these species (Cochran et al. 1967, 2004; Cochran 1972, 1987; Cochran and Kjos 1985; Diehl and Larkin 1998; Wikelski et al. 2003; Cochran and Wikelski 2005).
|
|
|
In order to obtain predicted wingbeat frequency from the theoretical models of Pennycuick (2001
) and Rayner (1995) for the experimental birds, we needed measurements of body mass, wingspan, and wing area. However, we measured only the length of the folded wing and body mass of the experimental birds prior to release. Therefore, in spring 2003–2005 we mist-netted 90 Swainson's thrushes in Champaign-Urbana, Illinois, near the site where the experimental birds were caught (for details see Bowlin 2007). We measured these birds’ folded-wing lengths, wingspans, and wing areas according to Pennycuick (1989) and used these data to create regressions for Swainson's thrushes, predicting wingspan and wing area from wing length. We did not capture enough wood thrushes or veeries for reliable regressions. The predicted wing area and wingspan of each experimental Swainson's thrush was then placed into the equations of Pennycuick (2001) and Rayner (1995) to obtain the bird's predicted wingbeat frequency.
Recording and analysis of transmitter signal
Wingbeat modulation of the back-mounted continuous transmitters was in the form of transmitter frequency variations associated with proximity to the birds’ bodies, which in turn varied with accelerations of the body associated with wing flapping. The varying transmitter frequency, through conversions in the receiver, was recorded as an audio tone whose mean frequency (pitch) was typically between 1800 and 2800 Hz, depending on the receiver tuning (see Supplementary Material S1). Tone variations provided a relative index of wingbeat amplitude. The function relating tone frequency to wingbeat amplitude depends on (1) frequency sensitivity of the transmitter to its proximity to the bird's body and (2) the variation in proximity allowed by the attachment. Both varied from individual to individual and the latter was subject to changes over time due to attachment loosening within an individual. Thus, the measured peak-to-peak variation of tone (Figs 1 and 3) provided only a short-term index (approximately several hours) to changes in forces produced during wingbeat cycles. Because we could not calibrate the amplitude against standards, the differences in amplitude indices, even from the same individual between beginning to end of flight, must be interpreted cautiously.
Cassette recorders were used to record the signal prior to 1994. To mitigate the poor time regulation of this type of recorder, we recorded a 1000 Hz reference tone along with the signal tone (Diehl and Larkin 1998). Data from 1994 onward were recorded on digital tape recorders (Sony DAT recorder, model TCD-D7) whose precise timing made an external reference unnecessary.
All transmitters had a working life of about 11 days. Transmitters used for V1, V2, S1, and S3 weighed 1.4 g with 26 cm antennas and were relatively insensitive to wingbeat modulation and thus provided wingbeat data only when we were close to the birds and the signal was strong. We used transmitters weighing 1.3 g with 16 cm antennas on S2, W1, W2, and S4–S9; these were of an improved design that provided data from all but the weakest signals. Some of the latter transmitters (W1, S4–S9) included a microphone (www.sparrowsystems.biz) for the purpose of counting flight calls (WW Cochran, unpublished data). Microphones picked up varying wind noises associated with flapping which ceased during wingbeat pauses. Wind noises from flapping sounded like a steam locomotive and their correlation with the variations in tone from the continuous frequency transmitter confirmed that the latter represented wingbeats (see Supplementary Material S2).
Analyses of wingbeats
Tape recordings were transferred to wave computer files (see Supplementary Material) in 90 s segments using analog-to-digital conversion (8 bit, 44.1 KHz sampling). Zero-crossing times in wave files were converted to frequency at 1 ms intervals and saved in frequency files. The latter were presented as cyclic temporal variations of the audio tone, i.e., signal frequency (Figs 1 and 3) by a program written for that purpose. The program included semi-automatic measurement of wingbeat frequency, pause duration, and relative amplitude. As an independent check of the counting method, a sample of these files was checked by a spectrum-analysis program (Cooledit, Syntrillium Software Corp., Phoenix, AZ). Records for diurnal flights, although they rarely exceeded 15 s, were analyzed as above.
For each 90 s segment of flight, the durations of pauses were measured and recorded and percent time paused (pause%) was calculated from the ratio of total time paused to the total noise-free time that wingbeats and pauses could be detected. Wingbeat counts over inter-pause periods were recorded and mean wingbeat period for segments calculated as the total number of wingbeats divided by the cumulative inter-pause time. Effective wingbeat period was similarly obtained from the total number of wingbeats divided by the total time interval (Liechti and Bruderer 2002). Mean wingbeat frequencies were calculated as the inverse of mean wingbeat period of the 90 s segments. For segments with no pauses, wingbeat frequency and effective wingbeat frequency were identical. Accuracy of mean wingbeat frequency was better than 1% due to the large number of wingbeats averaged (> 100). Indices of amplitude were also measured for four flights (Table 1–4) that provided nearly continuous and noise-free data.
We analyzed migratory flight data separately for the initial ascent phase after take-off, the final descent phase prior to landing, and the intervening period that we call the "cruise phase." Transition to the cruise phase was defined as the time when wingbeat frequency abruptly stopped decreasing or decreased when averaged over three consecutive 90s segments. We defined the start of the final descent phase as the last time in a flight when a sustained increase in wingbeat frequency began. Short-term variations in wingbeat frequency often precluded clear timing of a transition to the final descent phase, but in some flights (e.g., S8, W2, W2b, S5) an abrupt increase in wingbeat frequency clearly signaled the start of this phase. Our estimates of altitude were not frequent enough to determine whether or not the initial ascent phase, defined by decreasing wingbeat frequency, represented continuous ascent or that ascent completely ceased at its end, or similarly, whether or not the start of the final descent phase, defined by an increase in wingbeat frequency, coincided with the beginning of descent or that descent was continuous throughout that phase. For flights that ended during morning twilight (S5, S9, V2), the end of the descent phase consisted of a period of high wingbeat frequency at an altitude of < 100 m, presumably a period when birds are seeking habitat in which to land.
Experimental protocol
All experimental protocols were approved by the University of Illinois and Princeton University's Office of Laboratory Animal Research committee and comply with the "Principles of Animal Care" (Publ. 86-23, NIH) and with current U.S. laws.
Analysis of data
Data were statistically analyzed with SPSS 10.0 (Chicago) for Windows. Significance was accepted at the P = 0.05 level. For each bird we calculated average wingbeat frequency over 90 s flight segments that contained at least 30 s of undisturbed signal recordings (Fig. 3). We then averaged these 90 s values to calculate the overall average wingbeat frequency, and report it as mean ± SD. The main reason for binning wingbeat frequency into 90 s intervals was to create a less auto-correlated measure of overall individual performance. However, our analysis would have been qualitatively similar had we used even longer time bins. We also calculated average wingbeat pause% for each of the flight phases (initial ascent, cruise, and final descent). We present means ± SD for all data except when indicated otherwise. General Linear Models (GLM) were used to test for differences in wingbeat frequency both between genera (Catharus and Hylocichla) and between the three flight phases. However, we log-transformed the frequency of wingbeat pauses because they were not normally distributed and used Kruskal–Wallis ANOVAs separately for Catharus and wood thrushes to test for differences in pause% between flight phases. We also used Wilcoxon signed-rank tests to test for significant differences between empirically determined Swainson's thrush wingbeat frequency and wingbeat frequency predicted by theoretical models (Rayner 1995; Pennycuick 2001).
|
|
Results |
|---|
|
Top
SynopsisIntroductionMethodsResultsDiscussionAcknowledgmentsReferences |
|---|
General flight observations
The duration of flights of Catharus thrushes ranged from 121 to 640 min (n = 11, mean 325 ± 171, Table 1); the three wood thrush flights lasted 210, 367, and 450 min. One flight (S2b, Fig. 2F) was under complete overcast with intermittent rain from prior to departure until after landing. Another (W1, Fig. 2A) began under clear sky and ended under overcast and rain; all other flights were under clear or mostly clear skies. Estimated cruising altitude varied from as high as 1500 m under clear sky (S1) to as low as 180 m under overcast with rain (S2b). Differences between wingbeat patterns of foraging and the three phases of migratory flight (initial ascent, cruise, and final descent) were evident in signal-frequency plots (Fig. 1). Wingbeat pauses were common in all types of flight except the initial ascent phase. The durations of wingbeat pauses were usually close to the durations of a single wingbeat, but sometimes as long as two or three wingbeats, especially during the final descent phase.
Diurnal flights
Large changes in amplitude index and intermittent flapping at high and varying wingbeat frequency characterized diurnal foraging flights (Fig. 1A). Samples of diurnal flights of three Swainson's thrushes and one wood thrush (S1, S2, S6, W1) showed varying frequency and duration of wingbeats, often broken into bursts of 3–10 wing flaps separated by pauses of 0.2–1 s. Wingbeat frequencies of Swainson's thrushes ranged from 11.9 to 19.6 Hz (15.1 ± 1.57 Hz, N = 258 single flights of 3 individuals) and duration of their flights ranged from 0.2 to 15.1 s (median 1.6 s). Wingbeat frequency of wood thrushes ranged from 10.1 to 18.6 Hz (13.4 ± 1.38 Hz, N = 69 flights of one individual) and duration of flight ranged from 0.3 to 12.0 s (median 2.5 s). These variable flight patterns are consistent with the authors’ visual impressions of the flight maneuvers of thrushes during short ascending, descending, and level flights and during longer flights through vegetation.
Migratory flights
The initial ascent phase
The average duration of initial ascent was 10.9 ± 6.8 min (3–27 min) for 11 flights of Catharus thrushes and the durations of initial ascent phases of three wood thrush flights were 39, 14, and 16 min (Table 2). Wingbeat frequency and amplitude were highest immediately after take-off (Fig. 1B); mean wingbeat frequency for the first 5 s was 14.3 Hz ± 0.38 (13.5–15.0) for seven Catharus flights and 10.8 and 11.6 Hz for two wood thrush flights. Wingbeat frequency decreased rapidly for the first several seconds and then more gradually throughout the remaining ascent. The average wingbeat frequency for the ascent phase was 11.2 ± 0.5 Hz for Catharus thrushes and 9.1, 9.8, and 9.5 Hz for three wood thrush flights. The ratios of these initial ascent wingbeat frequency means to cruise-phase wingbeat frequency means ranged from 1.22 : 1 to 1.53 : 1 for individual Catharus thrushes and 1.27 : 1 to 1.40 : 1 for wood thrushes. Wingbeat pauses were few or absent during the initial ascent (Figs 4–6), mean pause % ranged from 0% to 3.0% for Catharus thrushes and 0–6.9% for wood thrushes.
|
|
|
|
The cruise phase
The average cruise duration was 307 ± 173 min (104–621) for 9 flights of Catharus thrushes and the durations of the cruise phases of three wood thrushes’ flights were 183, 336, and 404 min (Table 3). Varying wingbeat frequency and amplitude characterized episodes of both flapping-only and flap-pause flight that lasted from minutes to hours (Figs 4–6
). Average wingbeat frequency of Catharus thrushes (10.3 ± 0.5 Hz) was higher than that of wood thrushes (8.6 ± 0.1 Hz) while their mean pause%, 8.7 ± 5.3 and 8.2 ± 5.1%, respectively, did not differ significantly. During the cruise phase, the change in mean wingbeat frequency per hour was very small, ranging from –0.06 Hz (–0.01) to 0.42 Hz (0.007) for 12 Catharus flights and –0.01 Hz (–0.0018) to 0.1 Hz (0.002) for three wood thrush flights; change in pause% per hour for the 12 Catharus flights ranged from 0% to 0.06% and there was virtually no change for the three wood thrush flights.
|
The final descent phase
Wingbeat frequency was usually higher and pause% greater in the final descent phase than during the cruising phase (Figs 1F and 7; Table 4). The average duration of the final phase of 10 flights of Catharus thrushes was 23.4 ± 17.1 min (6–53 min). For the three wood thrush flights, the final descent lasted 13, 13, and 14 min. This final flight phase was usually characterized by a decreasing amplitude index and by high and/or increasing percent wingbeat pausing (Fig. 7). Average wingbeat frequency of Catharus thrushes (11.1 ± 0.5 Hz) was higher than that of wood thrushes (9.5 ± 0.3 Hz) while the mean pause% for Catharus thrushes (19.4 ± 6%) was less than that for wood thrushes (41 ± 8%) (Table 4). Mean wingbeat frequency during the last 10 s before landing was 12.7 Hz (12.5–13.0 Hz) for four Catharus thrushes, 20% > 10.16 Hz mean wingbeat frequency during the cruise phase for the same four flights. The wingbeat pause% during the landing phase of the Catharus flights was 20.4 ± 10.4% (7.2–42.5%), 2.3 times higher than during the cruise phase. For the three wood thrush flights the mean pause% was 41.0% (33.1, 41.9, and 48.0%), five times higher than pause% during the cruise phase.
|
|
Wingbeat parameters during different flight phases
We detected a significant difference in average wingbeat frequency among the three flight phases and between the two genera across all fifteen flights (Fig. 8A; GLM, F (5, 35) = 16.86, P < 0.001; for genus: P < 0.001; for flight phase: P = 0.002; for genus x flight phase: P = 0.96); post hoc Scheffe tests determined that birds beat their wings faster during initial ascent and final descent than they do during the cruise phase. Pause% was very low during initial ascent, increased during cruise and peaked during final descent in both Catharus and wood thrushes (Fig. 8B; Kruskal–Wallis ANOVA for Catharus,
2 = 6.3, P = 0.01; for wood thrushes, 2 = 2.9, P = 0.05).
|
Comparisons with theoretical predictions
The regressions for predicting wing area and wingspan from folded-wing length of Swainson's thrushes were as follows. Wing area, m2 = –0.006 + 0.228 (folded wing length, m); [F (1, 89) = 105.524, P < 0.001, R2 = 0.542]. Wingspan, m = 0.0923 + 2.055 (folded wing length, m); [F (1, 88) = 154.22, P < 0.001, R2 = 0.637]. Average wing area was therefore 0.0165 ± 0.0007 m2 and average wingspan was 0.295 ± 0.007 for the nine Swainson's thrushes in this study. To calculate the body masses of these thrushes for use in the theoretical models, we added 1.4 g, the average weight of the transmitters, to the masses given in Table 1.
Compared with our measured wingbeat frequencies during the cruise phase, the predicted wingbeat frequencies from Pennycuick's (2001) equation 10 were consistently higher (Wilcoxon signed ranks test, Z = 2.803, P = 0.005, difference 0.816 ± 0.325 Hz). Likewise, predicted wingbeat frequencies from Rayner's (1995) equation 8 (which includes wing span and wing area) were higher (Wilcoxon signed ranks test, Z = 2.668, P = 0.008, difference 1.082 ± 0.290 Hz). However, wingbeat frequencies predicted using Rayner's (1995) equation 7, which depends only on mass, were not statistically different (Wilcoxon signed ranks test, P > 0.05) from our observations. For Pennycuick's (2001) model, we used a power fraction of 1; power fractions should only be included for birds using a bounding style of flight and we could not be certain (see discussion) that the thrushes were in fact bounding. If we assumed that they were bounding, however, and used the average power fraction for each individual [1–(pause% / 100)], predicted wingbeat frequencies were even higher than those given above (mean 11.58 ± 0.53 Hz).
|
Discussion |
|---|
|
Top
SynopsisIntroductionMethodsResultsDiscussionAcknowledgmentsReferences |
|---|
Aeroecology, by its very nature, will often, if not always, necessitate remote sensing or "biotelemetry" of animals (Cooke et al. 2004
). We furthered the methods of Lord et al. (1962) and show that we can use biotelemetry to measure wingbeat parameters including frequency, relative amplitude and pause% throughout the nocturnal migratory flights of small migratory passerines as well as during short flights in stopover habitat (Bowlin et al. 2005).
When considering our results, note that added mass and drag caused by the transmitter almost certainly affected the wingbeat patterns of these thrushes. Theoretical models predict that birds carrying extra weight will increase their wingbeat frequency, wingbeat amplitude, and/or decrease their pause% in order to produce more power to offset the weight increase. For our Swainson's thrushes, Pennycuick's (2001) model predicts an increase of 
0.16 Hz due to the added mass of the transmitter. The models do not predict how wingbeat amplitude or pause% change with mass; we saw no relationship between the birds’ body mass (or size-corrected body mass) and pause% or between pause% and the percent of each bird's body mass the transmitter represented. An additional limitation of our wingbeat data is that they are for three species migrating over land in the Midwestern USA; the type of flight these thrushes use for long overwater flights could be different. Finally, great caution should be observed in extrapolation of our thrush flight behavior to other species.
Wingbeat pauses
Our result from Catharus thrushes, showing pauses in all 12 of their flights, contrasts with the results of Diehl and Larkin (1998) who stated that, "Regardless of phase of flight, the examined receiver signal showed no indication of flap-coasting ... or pausing ..." R Larkin and A Raim collaborated with the present author Cochran in gathering data for two flights (our S1 and V2 are the same as Swth1 and Veer3 in Diehl and Larkin 1998). Our analysis revealed that both birds paused (Figs 1E, 4 and 6); we cannot explain the discrepancy because pauses are both visually and aurally detectable.
We cannot be certain whether or not the pauses we observed were due to the bird holding the wings closed ("bounding"), partially closed ("partial bounding"), or outstretched ["undulating," currently called flap-gliding (Tobalske 2007)]. Pauses were sometimes flanked by periods of flapping with constant wingbeat frequency and constant indices of wingbeat amplitude, suggesting abrupt cessation of flapping with wings against or close to the body (bounding or partial bounding). At other times pauses were preceded by declining frequency and amplitude and followed by large amplitude and increased frequency, a sequence suggesting a gradually shallowing wing stroke ending with the wings partially or fully extended during the pause, followed at the end of the pause with higher amplitude and frequency to begin a new flap-pause cycle.
Two lines of evidence suggest that the pauses we observed primarily, if not exclusively, represented bounding or partial-bounding flight. First, flap-gliding is typically used only by larger birds or during slower flight than that used during migration (Rayner 1985), although swallows and swifts are notable exceptions to this rule (Bruderer et al. 2001). More importantly, using high-speed video of a Swainson's thrush in a wind tunnel, we did not observe pauses with wings extended, only pauses with wings tucked. Because the camera had a lateral view of the bird we could not quantify how close its wings were to its body, but it often appeared that they were not completely tucked in (Bowlin et al. unpublished data).
Wingbeat parameters and flight mode
Bruderer et al. (2001) showed that climbing flight in passerines can be achieved by prolonging flapping phases and/or reducing flapping pauses, while descending is mainly achieved by prolonging the pauses. Of the three flight phases, the initial ascent phase had the highest wingbeat frequency and amplitude, and fewest pauses, indicating that during this phase the thrushes were using more power than they did during cruise or descent. Birds are thought to use maximum or near-maximum power during climbing flight (Hedenström and Alerstam 1994), and our results are consistent with this hypothesis. Wingbeat frequencies during the final descent phase were about 8% higher than during cruise phase. Pause%, however, was also higher during the final descent phase, making the effective wingbeat frequencies lower than their cruise phase values (6% lower for Catharus thrushes and 29% lower for wood thrushes). Energy use presumably correlates more strongly with effective wingbeat frequency than with actual wingbeat frequency (Cochran et al. unpublished data), which means that flight power generally declined from one flight phase to the next. Effective wingbeat frequency also decreased throughout the cruise phase, suggesting that, as Pennycuick (1978) predicted, power output decreases during the cruise phase of long flights. Finally, while pause% was high during the final descent phase, it was not close to 100%—i.e., the birds did not simply glide down to the ground. This suggests that theoretical models which do not consider the costs of descent (Hedenström and Alerstam 1994) may require revision.
Rayner (1985) proposed that varying the amount rather than the frequency of flapping would be beneficial for small birds trying to decrease aerodynamic power output because their smaller size should mean lower muscle diversity. Lower muscle diversity would narrow optimal frequencies of muscle contraction, which in turn would mean a small bird would have a comparatively narrow range of optimal wingbeat frequencies. Since straying outside the bounds of that range would be energetically suboptimal, Rayner (1985) argued that small birds should stop flapping for short periods of time when they wanted to decrease power output, thereby lowering effective wingbeat frequency while keeping wingbeat frequency within optimal bounds. Pennycuick (1996) later made a similar argument. Our results are consistent with the idea that pausing is an effective way for small birds to lower their aerodynamic power output, either to descend or to decrease their horizontal airspeed.
Flight levels and change in altitude
The premise that there is a major initiation of nocturnal migration during the hour or so after sunset and a major termination during the hour or two before sunrise is clearly supported for Catharus thrushes (Fig. 9) and for an unknown mix of species by radar observations as well (Bellrose and Graber 1968; Graber 1968; Gauthreaux 1971; Bruderer 2003). Telemetry data on thrushes (Fig. 9) also show that departures and landings seldom occur in the interim between dusk and dawn, a finding not revealed by radar observations that show only that ascending and descending birds are found in similar numbers throughout the night (Bruderer et al. 1995). Ascents and descents without departures and landings can be explained if, after an initial major ascent, a large majority of individual migrants repeatedly sequence through descending, level, and ascending flight.
|
A probable function of periodically changing altitude throughout the night is to remain in a favorable wind stratum. Radar studies show that birds are more numerous at altitudes with favorable winds (Able 1970
; Bruderer and Steidinger 1972; Bruderer et al. 1995). For Catharus thrushes, Cochran and Kjos (1985) reported: "Some of the thrushes we observed ascended, during the first hour or so of flight, through a variety of unfavorable winds and then descended to an altitude with winds more favorable than those encountered at higher altitudes [eg. S2a, S7, S9] some remained at a higher altitude with a wind no better than one encountered at a lower altitude [eg. S5] still others ascended for a shorter period and leveled off at an altitude with acceptable wind without testing higher altitudes for more acceptable winds [eg. S2b, S6]. We note also that the ascent at the initiation of migratory flight is not repeated later in flight even if winds change significantly. Instead, after the initial ascent, altitude selection is restricted to descent or to small upward adjustments [eg. S4, S5]." It should be noted, however, that the wind vector is not likely to be the only altitude-dependent character of the atmosphere that is important to nocturnal migrants; turbulence, temperature, humidity, and cloud cover may also have varying importance. Liechti et al. (2000), however, reported that temperature and humidity are of minor importance in altitude selection of birds migrating over Israel. Periodic ascents and descents throughout the night may be an adequate explanation for the variations of wingbeat frequency and amplitude, and pause% that we observed during cruise phase (Figs 4–6 and 7B).
Observations from multiple flights
We followed two individuals (S2, W2) for two nights. While little can be said definitively with such a small sample size, we nevertheless wish to call attention to the fact that the wingbeat parameters for different flights made by the same individual, particularly pause%, vary widely enough that they could be from two different individuals. In other words, the inter-flight variation we observed appears to be similar to that of the inter-individual variation. For example, S2 paused very little during its first flight (3.6%), where it had an average wingbeat frequency of 10.57 Hz, but paused frequently during its shorter second flight (15.4%), when it had an average wingbeat frequency of 11.16 Hz (Table 1). High inter-individual variation has been reported from radar studies (summarized in Bruderer 1997), but the brevity of radar observations precluded recording intra-individual variation.
There are several possible explanations for this variation. First, between the two flights, the birds had at least one day to forage, during which time body mass may have changed. Second, the atmospheric conditions were not identical during each set of two flights: S2, for example, flew in clear conditions the first night and flew under an overcast sky with light rain the second night. The resulting difference in variables such as air density, wind vector, temperature, precipitation, and turbulence could have affected wingbeat parameters.
Comparisons with radar studies
Our 2–11 h telemetric observations of wingbeat characteristics of migrating wood and Catharus thrushes (Figs 4–6) provided a whole-flight perspective not available from radar observations. Able et al. (1982) noted some limitations of radar data: "These data were drawn from a swarm of migrants that may contain as many as 75 species and individuals of varying age, experience, and migratory goal," but did not mention that tracking of individuals on radar is limited to several minutes at most, precluding findings about intra-individual variation on time scales of actual migratory flights. Whatever an individual does for several minutes as observed by radar is only anecdotal evidence for what it did before or does after the observation.
For example, Liechti and Bruderer (1995) divided the wingbeat frequency and pause behavior of their radar targets into four classes that ranged from "continuous flapping at 9 Hz or less (big waders and waterfowl) to intermittent flapping at 12 Hz or more (small passerines)". Each of our 90s segments can be thought of as a single radar observation, a series of them for an individual thus providing a test of the efficacy of radar anecdotes in assessing whole-flight behavior. Our thrushes fall into a size gap between "small" and "large" European passerines (B. Bruderer, personal communication); thus, their wingbeat patterns could be expected to fall into both categories. However, if the radar classifications are accurate, they should never fit the small and large "waders and waterfowl" categories. As individuals, one Swainson's thrush and both wood thrushes fit two of Liechti and Bruderer's (1995) non-passerine classes while the other eight Swainson's thrushes and two veery fit one of their non-passerine classes (small waders and waterfowl). Collectively, all three species fit into at least one non-passerine class and fit the correct "passerines" category only 50–70% of the time (Fig. 10). Of course, the average pause percentages of our thrushes may be lower than they would be without the transmitters and thus they may only fit into the non-passerine categories due to our methods; on the other hand the much larger (70g+) passerines also use intermittent flight according to radar data (Bloch et al. 1981), so it is unlikely that increasing the birds’ mass by 1.4 g would have such a large impact on their pause%. Regardless, Emlen (1974) mentioned similar categorization difficulties "These [his radar] results should serve as a warning against over-zealous categorization of radar targets by signature analysis." Similar limitations apply to visual observations, which for a variety of reasons have relatively short observation time.
|
A related issue is the use of anecdotal data as evidence to ascribe particular behaviors to species or classes of birds. For instance, the two-part statement by Bruderer et al. (2001
) that "In long-distance flights, e.g. migration, the majority of small birds use intermittent flight styles, characterized by regularly alternating phases of flapping and resting ..." implies (1) that on long flights most small birds use flap-pause but some do not; i.e., there are two classes of small birds, and (2) that the intermittent flight class shows a regularity in the alternation of phases. Regarding the first point, we note that our thrush observations prove the existence of a behavioral class that alternates between flap-pausing and flapping continuously. Moreover, we note that each of our thrushes would have provided radar observations of flap-pause flight sometimes and continuous flapping at other times, observations that would lead to the correct conclusion that there were two kinds of flight but could also lead to the incorrect conclusion that the two kinds of flights show that there are two different classes of birds. This illustrates one way radar data can be and often have been misinterpreted. A class of birds that only flap-pause and a class that only flaps continuously may exist, of course, but data from the entire flights of individuals are required to determine this.
Before addressing the second point it is necessary to make clear what we call a session of flight with continuous wingbeat flapping, i.e., with no pause interrupting the sequence; in other words a series of wingbeats bounded by wingbeat pauses. While several papers use additional terms, Bruderer et al. (2001), Leichti and Bruderer (2002), Pennycuick (2001), and Rayner (1985) all use the term "flapping phase." Thus, we will follow this terminology.
Regarding point two, Bruderer et al. (2001) said that the duration of the flapping phase of the two species of hirundines they studied was "highly variable" while Liechti and Bruderer (2002) showed a distribution for the two species with virtually all the samples of flapping phase durations between 0.07 and 0.4 s, a range at least 100 times smaller than the flapping phase range of under 0.3–1000 s that we show with detailed measurements for the flights of four Swainson's thrushes (Fig. 11A). The wide ranges of pause% of Swainson's, veery, and wood thrushes (Figs 4–6) show that the variation of flapping phases was also quite variable for the other thrushes in our study. The argument of Bruderer et al. (2001), that a lack of regular patterns would distinguish hirundine echo signatures from those of other passerines, does not hold for Swainson's, veery, and wood thrushes. If anything, the relatively regular pattern of the hirundines could distinguish them from the more irregular pattern of the thrushes. The duration of pauses during the flights of Swainson's thrushes varied over a range of about 0.75–3 s (Fig. 11B), with a distribution similar to that reported for European passerines by Renevey (1981).
|
The flapping phase and pause durations we report (Fig. 11) are mean values for the 90 s data segments; thus variation within the segments is missed (Fig. 3). It is extraordinary that transitions from continuous to flapping flight and the reverse have not to our knowledge been reported for radar observations. The 90 s sample of Fig. 3 is not unusually long for a radar observation and the transition to pausing is abrupt enough that it shows up convincingly in 10 or 20 s. However, transitions are not plentiful, averaging one every hour or two for the thrushes (1 or 2% of our 90 s segments); such small percentages may have been regarded as anomalies during analyses of radar data.
Comparisons with theoretical predictions
For each of our thrushes, the wingbeat frequencies predicted from the equations of Pennycuick (2001) and Rayner (1995, eqn. 8) were circa 1 Hz higher than we measured during the cruise phase (Table 3). The predicted frequencies should have been lower than our measured values if the difference was due to the added mass or drag of a transmitter. While the 10% difference between our measured values and predicted values was statistically significant, given the broad interspecific nature of the theoretical equations a 10% difference may indicate decent agreement with predictions. Regardless, the differences we observed may be explained in several ways. For example, the empirical data on wingbeat frequency used for Rayner's (1995) and Pennycuick's (2001) equations were necessarily for diurnal flight and may be biased slightly, probably towards the high side, for the same reason that wingbeat frequencies of thrushes flying by day are 50% higher than their nocturnal wingbeat frequencies during the cruise phase of migratory flight. In fact, the wingbeat frequencies of the two bounding species Pennycuick (1996) observed during the day are remarkably similar to the wingbeat frequencies of our thrushes during diurnal flight. Furthermore, the body mass of thrushes is near the low end of the range of masses of birds used to generate these equations. We might therefore expect to see a greater discrepancy between actual and predicted wingbeat frequencies for thrushes than for medium-sized birds.
Interestingly, Rayner's (1995) simpler equation 7, based solely upon mass, predicted the observed wingbeat frequencies in thrushes much more accurately than did the two more complicated equations (Rayner 1995, eqn. 8; Pennycuick 2001, Table 3). We do not know if this is due to chance or if there is an alternative explanation for this observation, but we feel that additional data collected on migrating passerines may help resolve this issue.
Supplementary material are available at ICB online.
|
Acknowledgments |
|---|
|
Top
SynopsisIntroductionMethodsResultsDiscussionAcknowledgmentsReferences |
|---|
We wish to thank Arlo Raim for invaluable help in the field (radio tracking, driving, netting birds, attaching transmitters). We also thank Ron Larkin, Peter Lazarevich, Larry Pater. James Cochran, Michael Anderson, and Robb Diehl for help during the study. Judy Montgomery kindly helped count wingbeat patterns. Finally, Bruno Bruderer provided very helpful comments on an early draft of this article.
|
Footnotes |
|---|
From the symposium "Aeroecology: Probing and Modeling the Aerosphere—The Next Frontier" presented at the annual meeting of the Society for Integrative and Comparative Biology, January 2–6, 2008, at San Antonio, Texas.
|
References |
|---|
|
Top
SynopsisIntroductionMethodsResultsDiscussionAcknowledgmentsReferences |
|---|
Able KP. A radar study of passerine migration. Bird Banding (1970) 41::282–90.[Web of Science]
Able KP, Bingman VP, Kerlinger P, Gergits W. Field studies of avian nocturnal migratory orientation in white throated sparrows (Zonotrichia albicollis) released aloft. Anim Behav (1982) 30::768–73.[CrossRef][Web of Science]
Alerstam T, Hedenstrom A. Optimal migration. J Avian Biol (1998) 29::339–40.
Bellrose FC, Graber RR. A radar study of the flight directions of nocturnal migrants. Proc XIII Int Ornithol Congr (1968) 362–89.
Bloch R, Bruderer B, Steiner P. Flugverhalten naechtlich ziehender Voegel - Radardaten ueber den Zug verschiedener Vogeltypen auf einem Alpenpass. Vogelwarte (1981) 31::119–49.
Bowlin MS. Sex, wingtip shape and wing-loading predict arrival date at a stopover site in the Swainson's Thrush (Catharus ustulatus). Auk (2007) 124::1388–96.[CrossRef][Web of Science]
Bowlin MS, Cochran WW, Wikelski M. Biotelemetry of new world thrushes during migration: physiology, energetics and orientation in the wild. Integr Comp Biol (2005) 45::295–304.
Bruderer B. The study of bird migration by radar. Part 2; Major achievements. Naturwissenschaften (1997) 84::45–54.[CrossRef][Web of Science]
Bruderer B. The radar window to bird migration. In: Avian Migration—Berthold P, Gwinner E, Sonnenschein E, eds. (2003) Berlin: Springer-Verlag.
Bruderer L, Liechti F, Bilo D. Flexibility in flight behaviour of barn swallows (Hirundo rustica) and house martins (Delichon urbica) tested in a wind tunnel. J Exp Biol (2001) 204::1473–84.[Abstract]
Bruderer B, Steidinger P. Methods of quantitative and qualitiative analysis of bird migration with a tracking radar. In: Animal orientation and navigation—Galler SR, Schmidt-Koenig K, Jacobs GJ, Belleville RE, eds. (1972) 262:. Washington, DC: National Aeronautics and Space Administration Special Publication. 151–67.
Bruderer B, Underhill LJ, Liechti F. Altitude choice of night migrants in a desert area predicted by meteorological factors. Ibis (1995) 137::44–55.[CrossRef]
Cochran WW. Long-distance tracking of birds. In: Animal orientation and navigation—Galler SR, Schmidt-Koenig K, Jacobs GJ, Belleville RE, eds. (1972) 262:. Washington, DC: National Aeronautics and Space Administration Special Publication. 39–59.
Cochran WW. Orientation and other migratory behaviors of a Swainson thrush followed for 1500 Km. Anim Behav (1987) 35::927–9.[CrossRef][Web of Science]
Cochran WW, Kjos CG. Wind drift and migration of thushes: a telemetry study. 125 years of biological research 1858-1983: A Symposium. Nat Hist Surv Bull (1985) 33::297–330.
Cochran WW, Montgomery GG, Graber RR. Migratory flights of Hylocichla thrushes in spring: a radiotelemetry study. Living Bird (1967) 6::213–25.
Cochran WW, Mouritsen H, Wikelski M. Migrating songbirds recalibrate their magnetic compass daily from twilight cues. Science (2004) 304::405–8.
Cochran WW, Wikelski M. Individual migratory tactics of New World Catharus thrushes: current knowledge and future tracking options from space. In: The ecology and evolution of migration—Greenberg R, Marra P, eds. (2005) Baltimore, MD: Johns Hopkins University Press. 274–9.
Cooke SJ, Hinch SG, Wikelski M, Andrews RD, Kuchel LJ, Wolcott TG, Butler PJ. Biotelemetry: a mechanistic approach to ecology. Trends Ecol Evol (2004) 19::334–43.[CrossRef][Medline]
Diehl RH, Larkin RP. Wing beat frequency of Catharus thrushes during nocturnal migration, measured via radio telemetry. Auk (1998) 115::591–601.[Web of Science]
Emlen ST. Problems in identifying bird species by radar signature analyses: intra-specific variability. In: Proceedings of the Conference on the Biological Aspects of the Bird/Aircraft Collision Problem—Gauthreaux SA, ed. (1974) Clemson, SC: Clemson University. 509–24.
Gauthereaux SA. A radar and direct visual study of passerine spring migration in southern Louisiana. Auk (1971) 88::343–65.[Web of Science]
Graber RR. Nocturnal migration in Illinois – different points of view. Wilson Bull (1968) 80::36–71.
Hedenstrom A, Alerstam T. Optimal climbing flight in migrating birds – predictions and observations of knots and turnstones. Anim Behav (1994) 48::47–54.[CrossRef][Web of Science]
Liechti F, Bruderer B. Direction, speed, and composition of nocturnal bird migration in the south of Israel. Israel J Zool (1995) 41::501–15.
Liechti F, Bruderer L. Wingbeat frequency of barn swallows and house martins: a comparison between free flight and wind tunnel experiments. J Exp Biol (2002) 205::2461–7.
Liechti F, Klaassen M, Bruderer B. Predicting migratory flight altitudes by physiological migration models. Auk (2000) 117::205–14.[CrossRef][Web of Science]
Lord RD, Bellrose FC, Cochran WW. Radio telemetry of the respiration of a flying duck. Science (1962) 137::39–40.
Norberg UM. Vertebrate flight (1990) Berlin: Springer-Verlag.
Payne RB. Age variation and time of migration in Swainson's and Gray-cheeked Thrushes. Wilson Bull (1961) 73::384–6.
Pennycuick CJ. Fifteen testable predictions about bird flight. Oikos (1978) 30::165–76.[CrossRef][Web of Science]
Pennycuick CJ. Bird flight performance: a practical calculation manual (1989) Oxford: Oxford University Press.
Pennycuick CJ. Predicting wingbeat frequency and wavelength of birds. J Exp Biol (1990) 150::171–85.
Pennycuick CJ. Wingbeat frequency of birds in steady cruising flight: new data and improved predictions. J Exp Biol (1996) 199::1613–8.[Abstract]
Pennycuick CJ. Speeds and wingbeat frequencies of migrating birds compared with calculated benchmarks. J Exp Biol (2001) 204::3283–94.
Pyle P. Identification guide to North American birds, part 1 (1997) Bolinas, CA: Slate Creek Press.
Raim A. A radio transmitter attachment for small passerine birds. Bird Banding (1978) 49::326–32.
Rayner JMV. Bounding and undulating flight in birds. J Theoret Biol (1985) 117::47–77.[CrossRef][Web of Science]
Rayner JMV. Flight mechanics and constraints on flight performance. Israel J Zool (1995) 41::321–42.
Renevey B. Étude du mode de battements d'ailes d'oiseaux migrateurs nocturnes à l'aide d'un radar. Revue Suisse Zool (1981) 88::875–86.
Schmaljohann H, Liechti F, Bruderer B. Songbird migration across the Sahara: the non-stop hypothesis rejected! Proc R Soc B (2007) 274::735–9.[Medline]
Schmidt-Nielsen K. Animal physiology (1997) Cambridge, UK: Cambridge University Press.
Tobalske BW. Biomechanics of bird flight. J Exp Biol (2007) 210::3135–46.
Vaughn C. Intraspecific wingbeat rate variability and species identification using tracking radar. In: In: Proceedings of the conference on the biological aspects of the bird/aircraft collision problem—Gauthreaux SA, ed. (1974) Clemson, SC: Clemson University. 443–7.
Wikelski M, Tarlow EM, Raim A, Diehl RH, Larkin RP, Visser GH. Costs of migration in free-flying songbirds. Nature (2003) 423::704.[CrossRef][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  H. Schmaljohann and F. Liechti Adjustments of wingbeat frequency and air speed to air density in free-flying migratory birds J. Exp. Biol., November 15, 2009; 212(22): 3633 - 3642. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Coppack, S. F Becker, and P. J.J Becker Circadian flight schedules in night-migrating birds caught on migration Biol Lett, December 23, 2008; 4(6): 619 - 622. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||