UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Biomechanics and behavior of hummingbird molt Christensen, Beth A. 2001

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_2001-0014.pdf [ 4.09MB ]
JSON: 831-1.0099515.json
JSON-LD: 831-1.0099515-ld.json
RDF/XML (Pretty): 831-1.0099515-rdf.xml
RDF/JSON: 831-1.0099515-rdf.json
Turtle: 831-1.0099515-turtle.txt
N-Triples: 831-1.0099515-rdf-ntriples.txt
Original Record: 831-1.0099515-source.json
Full Text

Full Text

Biomechanics and Behavior of Hummingbird Molt by Beth A. Christensen A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS OF THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Departmentiof Zoology) We accept this thesis as conforming to thej;equi/erJ standard THE UNIVERSITY OF BRITISH COLUMBIA November 2000 © Beth Anne Christensen, 2000 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The University of British Cqlumbia Vancouver, Canada Department of DE-6 (2/88) Abstract This thesis explores how hummingbirds cope with molt, a costly component of their annual cycle, on two very different time scales. My first experiment explores aerodynamic explanations for the unique, non-sequential primary molt pattern followed by all hummingbirds, with the molt sequence of the two outermost primaries reversed in order compared to other birds. I simulated both the ancestral, sequential molt pattern and the non-sequential pattern in rufous hummingbirds (Selasphorus rufus) by removing primary 9 or 10, respectively. I used high speed video (500 fps) to film hovering events before wings were altered, immediately after I removed either feather, and approximately one week later. From the video, I estimated wingbeat kinematics of birds and used oscillator and aerodynamic theories to predict and interpret results. Results suggest possible aerodynamic benefits of the pattern followed by hummingbirds compared to the basic pattern they have evolved away from. My second study investigates how captive molting hummingbirds alter their behavior during molt. I used focal animal sampling to record frequencies and durations of flights, feeding bouts and aggressive encounters before, during and after natural molt. Molting birds flew less, fed less frequently, and engaged more often in aggressive encounters during molt than non-molting periods. These behavioral changes may be a mechanism to partially or entirely offset costs of the molting process. Natural selection has resulted in diverse ways to reduce the costs of molt, including how birds molt. Hummingbirds strictly adhere to a unique primary molt pattern, and results of this study show evidence of possible benefits. Changing behavior allows individual birds a means to compensate for the costs of molt on a daily basis. Overall, these are only two ways that hummingbirds cope with a necessary component of their annual cycle. TABLE OF CONTENTS Abstract ii Table of Contents iii List of Tables iv List of Figures v Acknowledgements viiii Chapter One: General Introduction 1 Chapter Two: Biomechanics of Hummingbird Molt Introduction 6 Materials and Methods 26 Results 35 Discussion 56 Chapter Three: Behavior of Captive Hummingbirds Introduction 70 Materials and Methods 73 Results 77 Discussion 87 Chapter Four: Conclusions and General Discussion 92 Literature Cited 94 iii LIST OF TABLES Table 1: A test of the blade element theory using data from the literature 20 Table 2: Morphological measures of experimental birds 36 Table 3: Kinematic results of experimental birds 37 Table 4: Results of repeated-measures ANOVA of kinematics 58 Table 5: Summary of hummingbird activity budgets from the literature 81 iv LIST OF FIGURES Chapter 1 Figure 1: Primary flight feathers of a hummingbird wing 4 Chapter 2 Figure 2: Relative hand areas of swifts and hummingbirds 8 Figure 3: The horizontal figure-8 wingstroke of a hummingbird 10 Figure 4: The wing disc area of a hovering bird 11 Figure 5: A chordwise element at distance r along a wing 16 Figure 6: Changes in lift requirements to changes in V 2 using 22 data from Chai & Millard (1997) Figure 7: Changes in lift requirements to changes in using data 23 from Wells (1993b) Figure 8: Planview of the video room 28 Figure 9: An example of the fit of a sine curve to the angular data 38 from 10 wingbeats Figure 10: Graph of average stroke amplitude and wingbeat 40 frequency for each of the 10 experimental birds Figure 11: Stroke amplitude and wingbeat frequency across the 41 three experimental periods for individual birds within treatment groups Figure 12: A hummingbird wing divided into 5 elements, with 42 relative area and lift. Figure 13: Changes in residual standard error of the fit of the 43 sine curve to the data across the three experimental periods v Figure 14: Changes in wingbeat frequency in the P10 treatment 45 group across experimental periods Figure 15: Changes in stroke amplitude in the P10 treatment group 47 Figure 16: Changes in velocity-squared in the P10 treatment group 48 Figure 17: Frame from sideview high-speed video (500 fps) of Y05 50 showing the gap created in the wing by removing primary 9 (P9) Figure 18: Gap in X75's wing with P9 removed, midway through the 50 backstroke, from overhead video (500 fps) Figure 19: The size of the gap in Y05's right wing for days 1 through 51 7 following the removal of P9 Figure 20: Changes in wingbeat frequency in the P9 treatment group, 52 as compared to the P10 group Figure 21: Change in stroke amplitude in the P9 treatment group, 54 compared to the P10 group. Figure 22: Changes in velocity-squared in the P9 treatment group, 55 compared to the P10 group. Figure 23: Changes in angle of attack at the wingbase during the 57 backstroke across the three experimental periods. Figure 24: Pressure gradient on an airfoil 60 Figure 25: Examples of hummingbird wings in various stages of 68 primary molt, following both the sequential and non-sequential patterns. Chapter 3 Figure 26: Plan of observation room (2.5x2.2x2.5m) and observing room 74 Figure 27: Average time budgets of the experimental birds before, 79 during and after molt vi Figure 28: Graphs of the proportion of time spent flying, frequency 82 of flights and average duration of flights before, during, and after molt. Figure 29: Graphs of the proportion of time experimental birds spent 84 feeding at feeders, the frequency and average duration of feeding bouts before, during and after molt Figure 30: Graphs of the average number of times per hour birds left 85 their perches to search for and/or catch live Drosophila before, during and after molt Figure 31: Graphs of the proportion of time birds spent engaged in 86 aggressive encounters, the frequency and average duration of aggressive encounters before, during and after molt. vii Acknowledgements I would like to thank the following people for their support and help throughout my experience at UBC: Dr. Lee Gass for his guidance and insight, Dr. John Gosline for his assistance above and beyond with the biomechanics experiment, Committee members, Drs. Ron Ydenberg and Andrew Trites for their assistance and advice, Dr. Douw Steyne for advice about sine curves, Members of the Gosline Lab for their generous support, The guys in the electrical and wood shops for the help building items from walls to snake head apparatuses, Christianne Wilhelmsen and Darin Bennett for sharing the experience, And Jay Christensen for his continued encouragement and support. Chapter 1: General Introduction Birds are unique in many ways. They are the only animals with feathers, which are their most characteristic feature both visually and biologically (Amadon 1966). Feathers have several diverse functions, including thermal regulation, courtship displays, protection from abrasion and the sun, and flight (Jenni & Winkler 1994, Proctor & Lynch 1993, Ginn & Melville 1983, Amadon 1966). Feathers are dead, temporary structures made of keratin (Jenni & Winkler 1994, Amadon 1966). They differ from other keratin structures such as hair, fingernails and claws in that they are not continuously renewed from the base, and consequently, must be replaced periodically. This process of new, developing feathers pushing out old, worn feathers is called molt (Payne 1972, Jenni & Winkler 1994). Molt has two functions. Damaged or lost feathers must be replaced to maintain plumage function. Also, birds must periodically adjust to new requirements, such as seasonal changes or reproductive needs (Jenni & Winkler 1994). Since feathers are vitally important to birds, it is better for them to be replaced before they wear significantly (Payne 1972, Amadon 1966). Molt is an energetically expensive process (Jenni & Winkler 1994, Payne 1972), as evidenced by high basal metabolic rates (Dietz etal. 1992, see reviews in Walsberg 1983, King & Farner 1961), increased body temperatures (Newton 1968), and increased oxygen consumption (Schieltz & Murphy 1995., Dietz etal. 1992, Walsberg 1983, Lustick 1970) in molting compared to non-molting birds. 1 High protein turnover (Murphy & Taruscio 1995), changes in bone metabolism (Murphy etal. 1992), and other physiological changes (Chilgren 1977, King 1981) have been reported in molting birds, indicating that there is much still unknown about the molt process. Direct energetic costs of molt include the energy required to synthesize new feathers, i.e. for keratin deposition. Indirect costs of molt are associated with the decline in function that occurs from the time a feather drops to when it is fully replaced (Jenni & Winkler 1994). One significant indirect cost of molt may be related to flight, for a molting bird must fly with compromised wings (Chai 1997, Jenni & Winkler 1994, Tucker 1991). Some molting birds such as penguins (Cooper 1978), geese (Halse & Dobbs 1985), grebes (Piersma 1988) and some ducks (DuBowy 1985, Douthwaite 1976), gain mass in anticipation of molt, then voluntarily reduce food intake and use the stored fat to meet nutritional requirements (Murphy 1996). But most birds, including hummingbirds, lose mass before or during molt (Chai 1997, Swaddle & Witter 1997, Hiebert 1993, Tucker 1991, Seel 1976, Newton 1969, personal observation). This may be a strategic reduction of the overall maintenance costs of a smaller body (Ankney 1979), or of the increased costs of flying with impaired wings (Swaddle & Witter 1997, Chai 1997, Douthwaite 1976). Reduced mass may also minimize other ecological and aerodynamic consequences of molt. Natural and simulated molt results in reduced flight speed, take-off ability, and aerial maneuverability, all of which likely increase predation risk (Swaddle et al. 1999, Chai etal. 1998, Swaddle & Witter 1997, Swaddle etal. 1996). In fact, 2 wild pied flycatchers with experimentally removed primaries experienced higher rates of disappearance than control birds, presumably due to increased predation from sparrow hawks (Slagsvold & Dale 1996). It is likely that these ecological consequences of molt are directly related to the compromised aerodynamic performance of molting wings. Tucker (1991) was the first to explore the effects of molt on aerodynamics when he found gliding performance degraded during primary molt in a Harris' Hawk (Parabuteo unicinctus). Hummingbirds' hovering ability is also compromised during wing molt, especially at the wingtips, which are important in lift generation (Chai 1997). Natural selection has resulted in many ways to minimize the costs of molt, by altering its timing within the annual cycle, and by changing the sequence of feather loss and regrowth. The timing of molt within the annual cycle is called a molt strategy (Jenni & Winkler 1994, Ginn & Melville 1983, King 1974). Within the molt period, feather loss and regrowth begins at one or more foci, then follows regular sequences along feather tracts. This sequence, called a molt pattern, may minimize the effect of molt on feather function (Jenni & Winkler 1994, Ginn & Melville 1983, Payne 1972, Stresemann 1966, Stresemann 1963). The flight feathers, i.e. remiges on the wing, and the tail feathers, display the most conspicuous molt patterns (Ginn & Melville 1983, Payne 1972, Stresemann 1967, Stresemann 1966, Stresemann 1963). Due to their importance in flight, it is reasonable to assume that molt patterns of flight feathers, especially the primary remiges on the hand area of the wing (Fig. 1), should be strongly shaped by 3 selection (Stresemann 1963). Primary molt patterns appear to be very strict within species (Jenni & Winkler 1994), and are the central focus of Chapter Two. Chapter Two explores the aerodynamics of the non-sequential molt pattern of hummingbirds. Unlike all other birds, hummingbirds reverse the molt sequence of their two outermost primary flight feathers. Using rufous hummingbirds (Selasphorus rufus) as a model species, I simulated both their natural molt pattern and the molt pattern they evolved away from by removing one of the two outermost primaries. I used oscillator and aerodynamic theories to predict and interpret kinematic changes during hovering in the two treatment groups and used high-speed video to measure those changes. In Chapter Three, I report on an experiment that investigates behavioral changes in captive molting hummingbirds, given that adjusting various activities that require flight is one way that birds can compensate for the increased energetic demands of molt. I used focal animal sampling (Altmann 1974) to Figure 1: The ten primary remiges of a hummingbird wing. Drawing modified from Aldrich (1956). 4 estimate the amount of time birds spent flying, feeding and in aggressive encounters and compared results before, during and after natural molt. Chapter Four offers some broad conclusions to findings of this thesis, which sheds light on why hummingbirds molt their primaries as they do, and demonstrates various ways captive birds compensate for the energetic costs of molt. I also expose difficulties with using current theory to estimate hovering costs with impaired wings and suggest several potential topics for future research. 5 Chapter 2: Biomechanics of Hummingbird Primary Molt Introduction Natural selection has produced many ways to minimize the costs of avian molt, and the resulting diversity of molting patterns may be viewed in this light. Feathers are lost and regrown in sequences along feather tracts, called molt patterns, which are thought to minimize the negative effects of molt on feather function. The flight feathers, i.e. remiges on the wing, and the tail feathers, display the most conspicuous molt patterns (Jenni & Winkler 1994, Ginn & Melville 1983, Payne 1972, Stresemann 1967, Stresemann 1966, Stresemann 1963). Due to their importance in flight, it is reasonable to assume that molt patterns of primary flight feathers should be strongly shaped by selection (Stresemann 1963). Primary molt patterns appear to be very strict within species (Jenni & Winkler 1994), and are the central focus of this report. The generalized ancestral primary molt pattern is symmetrical on both wings and sequential from the innermost to the outermost primary. This basic sequence preserves the aerodynamic function of the flight feathers during molt (Jenni & Winkler 1994, Ginn & Melville 1983) and may support the growing primaries against aerodynamic forces (Noordhuis 1989). Most bird species still follow this pattern, with the majority of exceptions occurring in non-passerines (see Jenni & Winkler 1994 for a complete review of primary molt patterns, Ginn & Melville 1983, King 1974). One unique primary molt pattern is seen in hummingbirds (Order 6 Trochiliformes; Family Trochilidae). The ancestral sequential pattern is followed until primary 8 (P8). When P8 is fully regrown, they lose the outermost primary, P10. They drop primary 9 last, well before P10 has completely regrown (Baltosser 1995, Stiles 1995a, Stresemann & Stresemann 1966, Aldrich 1956, Williamson 1956, Wagner 1955, personal observation). This atypical sequence appears to be nearly invariant among hummingbirds. Stiles (1995a) found that 99% of 642 records for 13 species followed this sequence, and Baltosser (1995) found that 88% and 94% of ruby-throated (Archilochus colubris) and black-chinned (Archilochus alexandri) hummingbirds, respectively, also followed it. Contradicting reports of this pattern in hummingbirds are from Wagner (1946 as cited in Stiles 1995a) and Ruschi (1962 as cited in Stiles 1995a), but these discrepancies have been attributed to small sample sizes, methodological differences and other issues (see Stiles 1995a for details) and are most likely exceptions or inaccurate. Why do all hummingbirds, and only hummingbirds, follow this unique, non-sequential molt pattern? Examination of hummingbirds' genetic and ecological relatives may provide some insight. Swifts (Order: Apodiformes) are the closest genetic relatives to hummingbirds (Sibley & Alhquist 1990). Both groups comprise small birds that use flight as their sole means of locomotion and have small feet unsuitable for walking. They feed on the wing, although on very different food types, with hummingbirds feeding primarily on flower nectar and swifts on flying insects. Perhaps the most striking similarity between these two groups is in their wing morphology, especially their relatively larger hand areas compared to other birds (Fig 2.; Chantler 2000, 7 a. d. Figure 2: Drawings of wings of a hummingbird (a), swift (b), grouse (c), and gull, with hand areas shaded. All drawings scaled to allow comparisons. Drawings modified from Aldrich (1956), Chandler (2000). 8 Johnsgard 1997, Hertel 1963). Despite these similarities, primary molt patterns differ between the two groups. Adult swifts molt their primaries in a slow basic pattern, from P1 to P10 (Chantler 2000, Mees 1985, Stresemann 1963, DeRoo 1966). Even though hummingbirds and swifts are related genetically, hummingbirds are more related ecologically to other nectarivorous birds in other parts of the world (Wolf & Hainsworth1975, Skead 1967). Hummingbirds hover at flowers to feed on nectar, and are important pollinators in the Americas. Sunbirds (Nectariniidae), sugarbirds, and honeyeaters hold this niche in other parts of the world (Pyke 1980, Wolf et al. 1975, Skead 1967), but obtain nectar in varying ways. Most non-hummingbird nectarivores perch while foraging (Pyke 1980, Wolf et al. 1975, Wolf 1975, Skead 1967). Some species of sunbirds hover at flowers, but only for short periods and at great energetic cost (Norberg 1996, Skead 1967). Honeyeaters, sugarbirds and most sunbirds follow the basic sequential molt pattern (Hanmer 1981, Dow 1973, DeSwardt 1990), although 3 of the 104 species of sunbird often, but not always, molt P10 before P9 (Hanmer 1981, Skead 1967). The adaptive significance is probably very different than the non-sequential pattern in hummingbirds since sunbirds' P10 is a tiny feather (Hamner 1981) contributing little to wing area and nothing to wing length. The differences in primary molt patterns between hummingbirds and their relatives, either genetic or ecological, might be explained by differences in their flight. Swifts and the nectarivores other than hummingbirds fly with flexed wings and forward thrust similar to all other birds. Some passerines, including sunbirds, 9 may hover for brief periods of time using an inclined stroke plane, with lift created primarily on the downstroke (Norberg 1990, Ellington 1984a, Weis-Fogh 1972, Skead 1967). On the upstroke the wing flexes and twists, avoiding negative lift and letting air through the wing (Norberg 1990, Ellington 1984a). This is called asymmetrical hovering (Weis-Fogh 1973). In contrast, only hummingbirds engage in symmetrical, or normal, hovering that more resembles that of insects than of birds (Ellington 1984a, Weis-Fogh 1972 & 1973). In normal hovering, the body is held in a vertical position. Flexible shoulder joints permit rigid wings to move in a horizontal figure-8 wingstroke that creates equal lift on the forestroke and backstroke (Fig. 3; Wells 1993a, Ellington 1984a, Weis- Fogh 1973, Greenewalt 1990). It seems reasonable to expect that hovering abilities may underlie the evolution of hummingbirds' unique primary molt pattern. I am unaware of any studies of hummingbirds' non-sequential molt pattern, and few have speculated about its origin or possible adaptive significance. Williamson (1956) stated that this pattern may reduce the effect of molt on hummingbird maneuverability, but no one has yet tested this possibility. Epting Figure 3: Horizontal figure-8 wingstroke of a hummingbird. Drawing from Ellington (1984a). 10 (1980, page 356) suggested that the pattern "should minimize further loss of integrity of the wing disc during molt and result in the most rapid restoration of the normal wing disc diameter." Wing disc area is defined as the area swept out by the wings (Fig. 4; Pennycuick 1969) and is a function of wing length. This measure is important in many models of hummingbird flight (Norberg 1990, Ellington 1984a & 1978, Pennycuick 1975 & 1969). The power output needed for hovering is directly proportional to wing disc loading, or the ratio of body weight to wing disc area (Epting & Casey 1973). Since the wing disc is governed by wing length, I interpret Epting's (1980) comments as: hummingbirds' non-sequential molt pattern keeps the wings longer and restores wing length faster than the ancestral sequence would, therefore reducing the power for flight during the molt of the two outermost primaries. Although Epting's (1980) idea about maintaining longer wings may have merit, he didn't specify how the wing disc is restored faster. P10 is slightly longer than P9 in most species of hummingbirds (Johnsgard 1997), and therefore provides the longest wings and the largest wing disc. P10 wing disc area Figure 4: Wing disc area of a hovering bird. Drawing from Norberg (1990). 11 would be missing or in some stage of regrowth for the same period of time regardless of which molt pattern is followed. In both Allen's {Selasphorus sasin) and rufous (Selasphorus rufus) hummingbirds, P9 is slightly longer than P10, but because P9 is inserted more proximally than P10, both feathers protrude equal distances (Aldrich 1956, unpublished data). How does the non-sequential molt pattern restore the wing length faster than the basic pattern? I suggest that the adaptive value of this molt pattern is related to gaps created by molting in wings. Given the large hand area of hummingbird wings (Fig. 2; Hertel 1963, Johnsgard 1997), the loss of P9 before P10 in the ancestral pattern would create a large gap in the most aerodynamically important part of the wing, the wingtip (Chai 1997). The gap would be directly perpendicular to the airflow on the wing, and would likely increase drag and therefore the power required to fly. I hypothesize that the non-sequential molt pattern reduces the energetic costs of flying with molting wings relative to the ancestral sequential pattern. To test this idea, I created two treatment groups using adult male rufous hummingbirds (Selasphorus rufus). In one treatment group, I simulated the sequential primary molt pattern by removing P9, and in the other group, removed P10 to simulate the non-sequential pattern. I compared kinematic changes using high-speed digital video (500 fps) of hovering bouts before, immediately after, and approximately one week after feather removal. Because complex wing morphologies were created in this study, I needed a way to compare power requirements between the two treatment groups. 12 Mechanical power required for flight equals the sum of inertial power, which is required to accelerate and decelerate the wing mass, and aerodynamic power. Aerodynamic power is required to overcome drag and consists of three components; induced, profile and parasite power, only two of which are important in hovering (Wells 1993a, Norberg 1990 & 1996, Alexander 1988, Ellington 1984f, Pennycuick 1969). Induced power is required to produce lift. Since no kinetic energy is stored in hovering, induced power is the major power drain during hovering. Profile power is required to overcome the friction and pressure drags of the wings, in moving the wings through the viscous medium, air, and is thought to be a small component of hovering costs. Parasite power is required to move the non-lift producing parts of the body (i.e. all but the wings) through the air. Since the body is stationary in hovering, parasite power is assumed to be negligible (Wells 1993a, Norberg 1990, Ellington 1984f, Weis-Fogh 1972). Inertial Power There is evidence that hummingbirds' flight mechanisms work as resonant systems, or oscillators. Hummingbirds move their wings in simple harmonic motion within a narrow range of frequencies determined by the mass, shape and area of the wing (Pennycuick 1996 & 1990, Wells 1993a & b, Rayner 1985, Greenewalt 1990 & 1975, Corben 1983, Hagiwara et al. 1968). Like those of oscillators, hummingbird wingbeat frequencies are strongly related to wing length (Chai & Dudley 1999, Pennycuick 1996 & 1975, Chai & Dudley 1995, Wells 1993a & b, Corben 1983, Greenewalt 1990 & 1975). In addition, several studies have 13 demonstrated that hummingbirds typically conserve frequency, producing extra lift by increasing amplitude (Chai & Dudley 1999 & 1995, Chai etal. 1999, Chai etal. 1998, Chai 1997, Chai etal. 1997, Chai & Millard 1997, Chai etal. 1996, Wells 1993a & b, Hagiwara etal. 1968). Finally, and perhaps most importantly, Wells (1993a) found evidence of considerable elastic storage of inertial energy in hummingbirds during hovering. As the wings decelerate during the second half of the wingstroke, kinetic energy is stored as elastic potential energy. This energy is then released during the first half of the next wingstroke and accelerates the wing. Assuming perfect elastic storage, with no energy loss during storage, the net inertial power requirements would be zero and power requirements for flight would be only aerodynamic power, i.e. induced and profile powers. More likely, some energy is lost during release or storage, or that energy storage is not perfect. Also, energy is required to overcome the damping that results from frictional losses, due to the viscous medium, air. Regardless, a flight system with elastic storage would result in large inertial energy savings over a system without storage, and should be strongly selected for in animals using energetically expensive modes of locomotion, such as hovering (Chai etal. 1998, Wells 1993a, Alexander 1988, Weis-Fogh 1973 & 1972). Inertial costs depend also on the moment of inertia of the wings, which equals: I = E rrij r 2 i 14 where m equals the mass of particle i and r equals its distance from the pivot point of the wing. Therefore, the distal areas of a hummingbird's wing contribute much more to the moment of inertia than proximal areas. Perhaps to minimize inertial power, 50% of a hummingbird's wing mass is found within 10% of the distance from the shoulder joint (Wells 1993a). If hummingbird wings act as resonant systems, I can use oscillator theory to predict changes in wingbeat frequency. If the wing is shortened or mass is removed, the natural frequency will increase, whereas increasing either mass or length will result in a lower natural frequency. Predicting changes in other kinematics will require aerodynamic theory. Aerodynamic Power Propeller theories, such as the momentum jet and blade element theories, have been used to estimate hovering costs in hummingbirds, who like helicopters, use horizontal movement of wings to accelerate air downward (Norberg 1990, Greenwalt 1990, Ellington 1984a, Pennycuick 1969, Weis-Fogh 1973 & 1972, Hertel 1963). These analyses assume steady or quasi-steady state conditions, i.e. that instantaneous forces on a flapping wing equal those for steady motion at the same instantaneous speed and angle of attack (Ellington 1984a). The validity of this assumption is controversial, especially for slow or hovering flight with large stroke amplitudes (Ellington 1984a, Weis-Fogh 1972). The momentum jet theory estimates induced power. It disregards the lift generating mechanism and is based solely on the jet of air pushed downward by 15 forces on the wings (Norberg 1990, Ellington 1978). The area of the jet is determined by the size of the wing disc, or the area swept by the wings (Fig. 4; Pennycuick 1969) and is a function of wing length. From this, Epting & Casey (1973) found that the power output required to hover is directly proportional to the wing disc loading, or the ratio of body weight to wing disc area. Simple aerodynamic theory states that: L = 1/ 2 p S V 2 C L where L is the lift required to support the weight of the bird, p is the density of air, S is the wing area, V equals the velocity of the distant air relative to the wing surface, and C L is the lift coefficient. C L is a measure of the lift generating capacity of the wing and depends on its shape and angle of attack. The blade element theory is a modification of aerodynamic theory that has been most widely used to estimate costs of animal flight (Norberg 1990, Ellington 1978 & 1984a). It assumes that forces are generated by the wings acting as airfoils. To calculate these forces, the wing is divided into a number of Figure 5: An element of a hummingbird's wing at r distance away from the wingbase and of area dr. Drawing from Norberg (1990). 16 independent elements, or chordwise strips (Fig. 5). The forces acting on each element vary during the wingbeat cycle, and the total forces acting on the wing equal the sum of the forces acting on all wing elements, or n L = 2 (1/2 p S i Vj 2 C L ) i=1 where Sj is the area of the element, Vi equals the velocity of the distant air relative to the surface of element i, and d is the lift coefficient as above. Velocity is proportional to the product of wingbeat frequency and stroke amplitude and varies along the wing length, with the wingtip and the wing base moving the fastest and the slowest, respectively. Again, a disadvantage of this method is the assumption of quasi-steady state conditions. However, its advantage is that it directly examines the forces of the lift generating mechanism, i.e. the wing. The vortex theories of Ellington (1984a-f) and Rayner (1979a & b) are the most recent and most accurate methods to estimate induced power, but they must be supplemented with aerodynamic theory to estimate profile and inertial powers. Vortex theories consider lift to result from air circulation around the wing and free vortices released from the wing into the passing air (Rayner 1979a, Ellington 1984e). One benefit of vortex theory is that it can be used in non-steady conditions. A disadvantage is that the wings must conform to laws of shape (Ellington 1984c), but molting wings directly violate this assumption. The blade element theory may provide the best alternative to examine kinematic changes in wings with altered morphology, since it is the only theory that can reflect changes in wing area. The assumption of quasi- or steady-state conditions is most likely violated to some degree (Ellington 1984a, Rayner 1979a), 17 however, hummingbirds most likely do not use complex lift-generation mechanisms reported in some insects (Dudley 1999, Dickinson et al. 1999, Ellington 1984c, Weis-Fogh 1973) that would grossly violate these assumptions. Although the theory likely oversimplifies airflow patterns about hovering wings (Norberg 1990, Ellington 1984a & 1978, Weis-Fogh 1973 & 1972), I can use principles of this theory to make simple, semi-quantitative predictions of kinematic changes in my treatment groups. Although I change the wing shape of experimental birds by removing primaries, I assume the majority of compensation for reduced wing area will be via frequency and amplitude rather than other kinematics, i.e. via adjustments of the V 2 term rather than via adjustments of CL. Overall, I can use both oscillator and blade element theories to predict and interpret kinematic changes in hummingbirds that result from changes in wing morphology, from natural or simulated molt. To test the validity of using these theories for hummingbird hovering flight, I can use data from experiments in the literature. Testing theories using data from the literature Oscillator theory There are several investigations of kinematic changes in hummingbirds in the literature, most with findings consistent with oscillator theory (Chai & Dudley 1995 & 1999, Chai etal. 1996, 1997 & 1998, Chai & Millard 1997, Wells 1993a & b). Hummingbirds tend to conserve wingbeat frequency when lift demands increase, and generate incremental lift by increasing stroke amplitude. Frequency 18 increases only after stroke amplitudes reached maximum values of near 180 degrees, and then only by 4-10%. In possible contradiction to oscillator theory, Chai (1997) found that molting birds significantly decreased wingbeat frequency. Chai was unable to explain this result, since wings with missing feathers should have a lower mass and therefore a higher natural frequency. I will expand on this in the Discussion. There is good evidence, although not entirely conclusive, that hummingbird wings act as resonant oscillators. I will assume that wings of experimental birds resonate in this report. Blade element theory Again, blade element theory states that total lift equals the sum of the lift created at each element of the wing, or: n L = Z (1/2 p S i V j 2 C L ) i=i with symbols as previously defined. As I don't have wing information for birds from these experiments and cannot examine individual elements, I will examine the forces on the entire wing. All results from this section are summarized in Table 1. Chai & Dudley (1995) measured kinematics of 4 hovering ruby-throated hummingbirds {Archilochus colubris) while decreasing air density (p) until the birds could no longer fly. Birds compensated for a 54% average decrease in air density 19 CD o to 0) CD O) ^ c c CO — x: CO o CD T3 j z CD co 2 co CO "D CO 2 0) CO -Q 5-o o CD > c CO co — c c CO CD 0 -C ZJ £ O ^ E • £ w re "F -O — T3 CD CD ^ * * C Q. c « E ® -g re E ^ CD CD T3 i 2 "CD O O CD CD T J _C0 CO 3 CO co CD • £ O T3 O ) * - -E -2 °-'•ff. re CD CD « £ F -. • .E o ^ CO CD • 5 ; c re -9 to =• tC •= D" I- CD to CM > 0) c (0 x: O o > O (0 (0 c v 0) Q. > I -3 (0 O DC (U u 3 o CO CD D)| V— CD > < >^  0 T3 Q o3 'co .c O co CO E > < CO 0) CO E CD H— > < CO CD CO CD > < CD CD CD CO GO CD CD CD CD CD CD CD CD CD CD CO CO CD O) CO 1 CD > < 20 by increasing amplitude and frequency, resulting in an average increase in V of 62%. This is in reasonable agreement with expectations from the theory (Table 1)-Chai era/. (1996) again decreased p as in Chai & Dudley (1995), but added the computation of the lift coefficient, as defined by Ellington (1984f). Considering only normoxic treatments and assuming that lift requirements remained constant, kinematic changes are in accord with expectations from aerodynamic theory (Table 1). Results from loading experiments agree well with aerodynamic theory, whether they test maximum loading (Chai era/. 1997 and Chai & Millard 1997; Table 1 and Fig. 6) or use relatively small loads (Wells 1993b; Table 1 and Fig. 7). Results of maximum loading experiments are especially notable, since wings reach stroke amplitudes of approximately 180°, and quite possibly interfere with each other. If so, the assumption of quasi-steady state conditions are likely violated, so the results agree perhaps better than expected. Chai (1997) reported kinematic changes in molting hummingbirds. Birds lost an average of 24% of their body mass, and therefore, of lift requirements. In reasonable agreement with predictions from aerodynamic theory, the product of reduced V 2 and smaller wing area (S) decreased by approximately 19%. Overall, the changes in hummingbird kinematics reported in the literature are consistent with predictions from both oscillator and blade element theories. Therefore, these theories may provide useful, semi-quantitative methods to predict and interpret kinematic changes in my experimental subjects. 21 70 90 110 130 150 Change in Lift Requirements (%) 170 190 Figure 6: Comparison of the percent change in the product of velocity-squared and lift coefficient to the change in lift requirements using blade element theory, using data from Chai & Millard (1997). See text for full explanation. 22 80 60 9^ 0 0 5 10 15 20 25 30 Change in Lift Requirements (%) Figure 7: Compensatory changes in velocity-squared (as measured by frequency times amplitude squared) in relation to increases in lift requirements due to loading, using data from Wells (1993b). 23 Predictions of kinematic changes in birds with P10 removed This treatment group, from here on called the P10 group (named after the feather removed), simulates hummingbirds' natural molt sequence of losing P10 before P9. If hummingbirds wings act as oscillators, then: • Frequency should immediately increase with the removal of P10 due to the removal of mass from the wing. • After one week, the frequency should remain at this elevated level, since wing mass and wing length should remain approximately constant. Because P10 is the outermost primary, removing it reduces wing area, but leaves the surface of the airfoil continuous. Keeping all other terms in the equation constant including lift, a decrease in wing area (S) should result in a proportional increase in the square of the product of stroke amplitude and wingbeat frequency. The earlier predicted increase in frequency may partially or completely compensate for the decrease in S, and, if necessary, stroke amplitude should increase to complete the lift requirements. So, following blade-element theory: • Stroke amplitude should increase if the predicted increase in frequency does not fully compensate for the decreased wing area. • Wing velocity should increase after removing P10 to compensate for a reduced wing area. • Both parameters should remain elevated after one week. These predictions are based on the assumption that all other measurements, such as stroke plane angle, body angle and angle of attack, will remain constant over time. 24 Predictions of kinematic changes after removing P9, compared to the P10 group This group, now called the P9 group, simulates the basic, sequential primary molt pattern that hummingbirds no longer follow, where P9 is lost with the old P10 still present. As with the other treatment group, I can use oscillator theory to predict qualitative frequency changes that occur in response to the loss of mass. • Frequency should increase immediately after the removal of P9, to approximately the same extent as P10 birds, as the reduction in mass of the wing should be roughly equal. • Since wing mass should remain constant the week after plucking P9, frequency should remain elevated to the same extent. Removing P9 will create a large gap in the wing between P8 and P10, which should disturb airflow across the surface of the airfoil. If I consider the isolated P10 to be an aerodynamically useless part of the wing, losing P9 first would reduce the aerodynamically effective wing area (S) more than losing P10 first would. If so, blade-element theory predicts that a proportionally larger increase in the velocity-squared term of the lift equation is necessary to maintain lift after plucking P9 than after plucking P10. • Stroke amplitude should increase more in the P9 group than the P10 group, since: • Velocity should increase more in the P9 group than the P10 group, to compensate for the larger loss of effective wing area. 25 • Depending on the compensation mechanisms employed by P9 birds, amplitude, and therefore, velocity could be reduced to P10 values after one week. As with the P10 birds, other kinematic measurements should remain unchanged throughout the experiment. P10 birds may follow the predictions more closely than P9 birds, since wing shape and therefore lift coefficient may change more in the latter. P9 birds may also deviate more from steady-state conditions than P10 birds due to the complex wing morphology (and complex airflow patterns) created by removing P9 first. If results follow predictions, it may indicate higher hovering costs in the treatment group simulating the basic sequential pattern than in the group simulating hummingbirds' non-sequential pattern. To my knowledge, this is the first exploration of the adaptive value and aerodynamic consequences of a primary molt pattern. Materials and Methods Birds I used 10 adult male rufous hummingbirds (Selasphorus rufus) in this experiment. Using adult males avoided the confounding effects of the variation in morphology among the different age/sex classes of this species (Moore 1997, Carpenter era/. 1993a & b, Kodric-Brown & Brown 1978, Stiles 1972, personal observation). Birds were captured between April and June of 1999 at three British Columbia sites; Sumas Mountain, the Sunshine Coast, and near Rosewall Creek 26 on Vancouver Island. All birds were housed communally at the Zoology Vivarium on the UBC Campus in five aviaries, ranging in size from 1.3m X 2.5m X 2.5m to 2.5m X 6.2m X 2.5m, and the photoperiod was adjusted regularly to reflect seasonal tightidark hours. Birds' diet consisted of commercial hummingbird nectar, QuikoNectar (Quiko GmbH, Germany), and was supplemented with live Drosophila. I provided excess feeders to avoid exclusion from feeding of submissive birds by dominant birds. Experiments were run between April and June of 1999. Video Room I constructed a video room measuring 2.5 m long X 2.2 m wide X 2.5 m high (Fig. 8) and suspended a feeder 0.5 m from the ceiling in the center of the room. To monitor birds' weights while in the video room, I placed a perch on a balance at the same height as the feeder nozzle, 0.9m away. A 63 X 57cm mirror was suspended from the ceiling at a 45-degree angle directly above the feeder. I built two camera positions into a wall of the video room so that only the camera lens could be seen by the birds. The lower position was level with the feeder and provided a view of the bird's right side while it hovered at the feeder. The upper position was level with the center of the suspended 45-degree mirror and provided a view of the hovering bird from directly above. I mounted nine 90-Watt halogen lamps 25 cm from the ceiling behind angled plexiglass panels to provide light for filming. Birds were kept in the video room during experiments and occasionally overnight if filming was to continue the following day. 27 Video camera O = ' ' 9 n t iii = feeder = perch on balance Figure 8: Video Room Planview. Room measurements are 2.5mX2.2mX2.5m. Dashed lines represent walls constructed of white ripstop nylon. The feeder is located approximately 0.5 m from the ceiling and 1.6 m from the camera. The lights are 90-Watt halogen spotlights, pointed at the front of the feeder. Camera I used a MotionScope PCI-500 high speed video camera (Redlake Imaging) to film hovering bouts, and used a framing rate of 500 fps for all bouts. The camera was controlled by a computer located outside of the video room, which permitted me to film and view images without disturbing birds in the video room. 28 Training I placed birds in the video room with an elevated perch so they could readily find it. After they did, I gradually lowered the perch until it was level with the feeder. I began filming when the bird had found the feeder and was feeding regularly. Treatments I randomly assigned five birds to each of the two treatment groups. I removed the P10 feather from the P10 group to simulate the natural, non-sequential hummingbird primary molt pattern. In the P9 group, I simulated the basic sequential, ancestral primary molt pattern by removing the P9 feather. Filming protocol I filmed birds as they hovered at the feeder, typically every 6-15 minutes. I filmed ten hovering bouts by each bird before any wing manipulation; five bouts from the side and five from overhead. These bouts provided baseline kinematics and allowed each bird to serve as its own control. I then removed the bird from the video room, plucked either P9 or P10 from both wings depending on the treatment for that bird, and immediately returned the bird to the video room. There were no signs of trauma to any bird at any time, from either handling or from removing the feather, and they usually began feeding regularly within minutes of their returning to the video room. After plucking, I filmed 10 more bouts, 5 from above and 5 from 29 the side, and then returned the bird to an aviary with other birds. After seven to eleven days, I returned it to the video room to film another 10 bouts. When the video camera was stopped, the computer automatically stored the last 1000 frames, or 2 seconds of video. From each of the thirty 2-seconds bouts captured on video, I selected one to two hundred consecutive frames (0.2-0.4 seconds of hovering) for later analysis. By visually examining the video, I attempted to choose 100-200 frames of steady hovering for analysis. Birds' eyes were remarkably motionless, frame-to-frame, and I used this steadiness to assess frames. I also estimated wingbeat frequency by counting the number of frames (1/500 t h of a second) for 10 wingbeats to occur. A sudden increase in wingbeat frequency typically indicated that the bird was preparing to back out of the feeder nozzle (also reported in Weis-Fogh 1972). I used this and any visual cues to avoid inappropriate motion in the saved frames. Frames for analysis were stored on compact disc. Weight Control Due to the importance of body mass in flight mechanics, I carefully monitored birds' weights during filming. Since hummingbirds tend to gain weight during each day and while in captivity (Chai et al. 1999, Wolf & Hainsworth 1977, personal observation), I occasionally had to withhold food until individuals achieved their baseline weights. During each feeding bout, birds typically gained 3 0 - 1 0 0 milligrams, or approximately 1-3% of their mass, so I used weights from before and after feeding to estimate mass for that bout. The maximum difference in 30 mass between any two bouts for any bird was 9% (mean % difference ± SD= 4.5 ± 2.1%). One of five P10 birds (X73) increased his mass by approximately 12% the week after feather removal. I attempted to force his weight down by restricting food availability, but his health appeared questionable and for humane reasons, I didn't record his final hovering events, reducing sample size for that period and treatment to 4. Plucking versus cutting feathers I plucked feathers instead of cutting them at the base because plucked feathers regrow within several weeks. In contrast, cutting leaves the feather shaft in place, and therefore the wing would remain without that feather until the next natural molt. After I released them, the birds would have to migrate to Mexico with compromised wings, which might reduce their chances for survival. Plucking appeared to cause little or no discomfort to the birds. Analysis of video records I made angular measurements from video images stored on compact disk and measured wing lengths and areas from 35mm photographs of outstretched wings taken against a reference grid. All measurements were made using public domain NIH Image software. I assumed that the wings moved symmetrically. I tested this assumption for wingbeat frequency and stroke amplitude on one bird using circular statistics (see below; Batschelet 1981) and found no significant difference in the wing pair 31 between stroke amplitude (p=0.936, n=8 bouts) or frequency (p=0.8155, n=8 bouts). Given these results, I used the wing in the overhead video that provided the best contrast and resolution for analysis. I measured body angle, stroke plane angle and geometric angle of attack from right wings. Wingbeat Frequency and Stroke Amplitude The angular movements of hummingbirds' wings are almost perfectly sinusoidal (Weis-Fogh 1972, Hertel 1963, Stolpe & Zimmer 1939). Based on this, I used two separate methods to estimate wingbeat frequency and stroke amplitude, both using the same angular measurements of 10 consecutive wingbeats from the video records. Both methods fit the measurements to sine curves. One method fits one wingbeat at a time to a single sine wave, and the other fit 10 consecutive wingbeats to a continuous sine wave. I measured the angle of the line from the wingtip to the pivot point at the base of the wing relative to horizontal. For consistency with P9 birds, I considered the tip of P8 to be the tip of the wing for all angular measurements before and after removing P9. Amplitude and frequency estimates correlated well regardless of whether I used the tip of P10 or P8 for measurements (frequency r=0.994, p<0.001, n=8; amplitude r=0.956, p<0.001, n=8). I used periodic regression (Batschelet 1981) to fit 10 consecutive angular measurements of a single wingbeat to a sine curve. This non-linear regression used a method of least squares to fit the sine function to the data, using a given frequency that I estimated from the video recordings (see under Filming Protocol). 32 I then either added to or subtracted from the frequency, in 0.1 Hz increments, until I found the frequency that minimized the sum of squares and considered this the best estimate of frequency for that wingbeat. I calculated stroke amplitude from the curve by doubling its amplitude. This method allowed me to estimate variability in frequency and amplitude within hovering bouts and to corroborate results from the alternate method. I also estimated frequency and amplitude by fitting a continuous sine curve to 90-100 angular measurements from 10 consecutive wingbeats. I specified initial values of frequency, amplitude and phase for that hovering bout, and the computer program used Marquardt's (1963) method of least squares to iteratively find the curve that best fits the data. I used results from this method in all tables and statistical analyses. Body Angle I estimated body angle from side-view frames with the wings in the extreme forestroke position. This was the angle between the line from the base of the neck to the tail, and a straight, vertical line of pixels within the frame. I averaged the approximately 20 estimates per 200-frame video record. Stroke Plane Angle To estimate stroke plane angle, I determined the X and Y coordinates of the tips of P10 (or P8) in the extreme positions of the backstroke and forestroke for each wingbeat in the video record. After removing P9, I used the tip of P8 in the 33 calculations instead of P10, since both measures correlated well (r=0.988, p<0.001, n=21). I took the inverse tangent of the ratio of the change in Y divided by the change in X to estimate the angle of the line between the wingtips and a straight, horizontal line of pixels. I did this for each of the approximately 20 wingstrokes in the record and averaged them. Angle of Attack I estimated geometric angle of attack, defined here as the angle of the wing relative to the horizontal, of the forestroke and the backstroke at two locations on the wing: at the wing base and at approximately 50% of the winglength. I took these measurements from frames in which the wing was perpendicular to the bird's body and pointing at the camera. Typically 5-15 frames were measured and averaged per 200-frame video record to produce one value per hovering bout. Error Correction I hung a string with a small lead weight in the field of the side view to determine the true vertical for each video record. I used this angular measurement, usually less than 2°, to correct angle of attack, stroke plane angle, and body angle measurements for any tilt or rotation of the camera. Blade Element Theory Using photographs of experimental birds' wings, I used NIH Image to calculate wing areas before and after removing primary feathers. For three P10 34 birds, I used the same methods to divide wings into 5 individual elements and measure wing areas of each, both before and after removing a feather. From these data, I used the equations of blade element theory to calculate the relative amounts of lift generated by each element. Statistics Since all baseline values were different and the changes were small, all kinematics were tested as percent deviation from their baseline values. I used univariate repeated-measures analysis of variance to test changes in the P10 treatment group across periods. To compare treatment groups, I used a mixed-model repeated-measures analysis of variance. The analysis considered 8 dependent kinematic variables in relation to one independent variable: period.. Throughout this report, I will use Period 1 to refer to birds with unperturbed wings. Periods 2 and 3 will be used to refer to immediately after removing the feathers and approximately one week later, respectively. Results General results Morphological results for the 10 experimental birds are summarized in Table 2. Kinematic results for all 10 experimental subjects can be found in Table 3. All fits of sine curves to angular movements of 10 consecutive wingbeats were good (residual standard errors of all flights across 3 periods, mean = 5.0 degrees ±0.15 S.E. ; see Fig. 9 for an example). The correlation between the results of this 35 Table 2: Morphological measures of the 10 experimental birds ± 1 standard deviation. Treatment 9 represents the birds with P9 removed, simulating the basic, sequential molt pattern, and Treatment 10 represents birds with primary 10 removed, simulating the non-sequential molt pattern followed by hummingbirds. Period 1 is with unperturbed wings. Periods 2 and 3 are immediately and one-week after plucking, respectively. Wing lengths in parenthesis are based on P8 rather than P10. lird ID Treatment Wing Length Period # flights (n) Mass (g) (cm) 1 5 overhead 2 side 3.37 ± 0.06 Y95 10 4.10 2 5 3.32 ± 0.08 3 5 3.36 ± 0.06 1 5 3.14±0.02 Y96 10 4.10 2 5 3.04 ± 0.01 3 5 3.07 ± 0.03 1 5 3.31 ± 0.02 X71 10 4.25 2 5 3.21 ± 0.03 3 5 3.24 ± 0.03 1 5 3.42 ± 0.02 X73 10 4.03 2 5 3.39 ± 0.02 3 0 * 1 5 3.53 ± 0.02 X76 10 4.13 2 5 3.52 ± 0.03 3 5 3.52 ± 0.02 1 5 3.35 ± 0.03 X74 9 4.21 2 5 3.32 ± 0.02 (3.96) 3 5 3.34 ± 0.02 1 5 3.18 ±0.02 X72 9 4.25 2 6 3.13 ±0.03 (3.89) 3 5 3.18 ±0.02 1 5 3.50 ± 0.02 X75 9 4.17 2 5 3.48 ± 0.01 (3.86) 3 5 3.52 ± 0.01 1 5 3.40 ± 0.03 Y05 9 4.02 2 5 3.31 ± 0.02 (3.89) 3 4 3.29 ± 0.05 1 5 3.40 ± 0.03 Y62 9 4.19 2 5 3.39 ± 0.03 (3.91) 3 5 3.38 ± 0.07 36 CD . C -t—• by CD MO E O CO C T 3 CD 0 > CO O a. 9 remi 1 molt ctively Q. co 0 .c ->—1 en es| rds seq c c o CD o cz CD Q. ents i _ 0 ents a> H— CO co c CD 4—* _ca 0 a. ^ 0 CD E 5 i CD CO eu C T 3 o CD E > c5 CD O CO E >> i h-i _ 0 O c T— atio ary ime > E E CD "D Q. Q) "D CO m wit co "D wit "D c CO C CO "co bird CO CM-(/) co +1 •+—> "D rds isen erio CD » QL Q. CD co o> CD O E CD CD 9-1 rbed CD CO o CD , _ £ pe CD c * J C E CO -•—1 o c g CD CO pa CO CD o mol erio o_ CO CO E uenti CO ine uenti T 3 i _ cr !a 0 cri CO c: CD o 'E CO E CO CO 1 - .O SZ CD < CD LO -2-^ CM S <o 5 ^ ^ CO T— CO CM" ^ +1 +1 +1 j^- m to CD CO CD ^ -5 ^ CO CO ID CO CO CO +! +1 +1 W N N CO N N Tj" Tj" O C D +1 +1 . O) LO CD CO co LO r--+i +i +i T- O O) r- W O CD h- CO -H +1 +1 CO CO CO c\i cd cvi LO co +i +i +i LO C\l Tt S lS 5 ° * r** o> +i +" + ! in i- CO CO t It CD CD LO +1 +1 +1 CM T - ^ f i l l O (D C 03 CD CO N T— c\i oj +1 +1 +1 cn to CD CM O CO rs.rs.co LO i ~ co c\i c\i co +i -H +i N ro en o rs.rs.rs. Tj" T- csj •r^  c\i T -+i +i +i CO o o r- CO O N CO N co •<* CM d +1 +1 . CO CO d K h- co CO s s i- d +i +i +i C M oo r--+i +i +i CO CO CM cb co CD CO 1 -T - CM CO +1 +1 +1 OJ CO CD CO CO CM CM CO CM +1 +1 +1 O LO CO -tf O 'tf 1— CM CO +1 +1 +1 CM O O CM CO CM CM O +1 +1 +1 LO O) CM CM cd d N CO N CD m rji c C OT O < CD ~ ^ £ CD CD ° S? O) n_ CDLO -2 0) CM f^r CM CM CM +1 +1 +1 N O CO C7) oi CO J^- ^ ID i— O CO C M r^' CM +1 +1 +1 C M N C D d LO C D LO LO Is-. LO T - CO CM +1 +1 +1 CO CO oi d N LO CD LO CD - ^ J -T - ' CM +1 +1 H CD LO o in N T - CM CM 00 ^ CM LO CO +! +1 +1 CD CM r- ^ CO CM CM CO +1 +1 +1 CM r- LO O) O CO CO CO CO +1 +1 +1 ^ CM LO O CO CO CO CO CM +1 +1 +1 CO r r oo ^ LO LO LO | S 8 » 2< s $ 8 ° gg1 i ? s. CD -^CM CM CM +i +i +i t T~ IS. d T— co LO in LO r-- co co CM T~ T-+i +i +i CO O CO CO i— o T— CM T— +1 +1 +1 CM CD lO CO oS CO LO LO LO +1 +1 , LO CD T - CD LO CO N N CM T - T -+1 +1 +1 LO CO ^ "5 LO LO LO r*- CM T - ^J- ' CM +1 +1 +1 o •<* CO CO CM CO CO y-i— CM CO +1 +1 +1 CO CD Tj-S LO N i- CO h-CM CM i— +1 +1 +1 O O rj-co cvi L  O CO LO LO LO LO LO LO CM CM CO +1 +1 +! CO CD CO LO O T~ +1 +1 +1 CO ^ CO o 8 LJ- CD d) "° O 03 < lO o to o C) o +1 +1 +1 T - LO lO co to CO 0 CN] T - C M +1 +1 +1 T o> -* ° h~ L O L O +1 +1 ^ o ° ° co m o c i -+1 +i +i ^ O T-O C O C D •>- C O L O d ^ +1 +1 +1 L O L O C M cd ^ cd C O C O o d d T-' +i +i +i T— L O C O C M O "r^  +1 +1 T o ° oi 6 C O C O •r^  -r^  O +1 +1 +1 co co cd d d CT 2! c co < 2 "D CD m E < o CO CO o LO CM CO CD rs- LO CO CM CM CO rs. o CO CO CO LO LO CD o CO LO CO d CM CM CO CM CO 1-' CO LO CM CO CM CM cb 1-' d d CM cb cb +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 « +1 +1 +1 +1 -H +1 +1 +1 4-1 +1 +1 +1 +1 +1 +1 +1 +1 +1 LO CM CM CO LO CO CD f- o CD CD IS. CM LO o CO •st CM CD 1— LO CO q CO CD CM o LO LO LO CO d CD CD CD d d CM CM CO d CO CO LO LO ib ib CO cb CO CO LO LO LO LO LO LO LO LO 3- LO LO •^ t LO •^ r LO CO CO o o LO CD o CM CO LO LO is. CM rs. LO CD LO CO LO CM CM CM LO CO CO CO CO CM CO CO CM 1— T - 1— CM CM CM LO cb CM CM cb CO CO CO CM cb +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +! +1 +1 +1 +1 +1 CO LO f-. CO CO CO LO h- * CO fs. CM L0 o CD o CM CO CM 03 rs. CO CM CO rs. q CO CO cd CM oi CO CO CO 1—^ co cb cb cb cb 1—^ d cb LO cb oi OJ 1—' CM CM " T _ CM , ~ CM CM CM CM CM CO CO CO CM ^~ CO CM CM CM CO T - CM CM ,— o CO -t; ^ o co o O >. T- T- r-o oi d ^ +i +i +i C D C O C O C D C O C O o d d co r~ d d +1 +1 , C D O L O ^ L O d d +i +i +i CD co r— C O T~ +1 +1 +1 C O C O o +1 +1 +1 co o o o i- i-+1 +1 +1 +1 +1 +1 n o oi. s w t -CO T - CO O +1 +1 +1 co co co CD CD a LO LO s i - C M C O 1 - C M C O T - C \ J C O - ^ C M C O i - C M C O - ^ C M C O - r - C N J C O T - C M C O T - C N C O T - C M C O CD — L0 o >-o 00 o co LO o o co o LO o o CO o LO o o o CD CO CD E (S99j6ap) UO!;OLU 61NM jo 9|6uv o = 1 £ 5 03 CO O CD J= > - II (0 "D *—1 CO «_T CD *E -o ic-CS CO -I CD • — II CX >-CD " CO c ^ 0 3 O 3 a -~- CD 8 « e o> SJi CL . O CD "a c S CD CL E g CD CO != II Jo « CO CD CD E ® « 5 O) + s ? O CD r E c ' w CO * CD II £ t « CO CD o _co CD o CD c 'co CO "5 o ^ £ N o CD S ' Q . CD < § 3 LO LL > CD "a CO " O c CO -t—' CO "co •g 'co CD 38 method and those of the alternate method using circular statistics (Batschelet 1981) was excellent for both wingbeat frequency (r=0.998, n=46, p<0.001) and stroke amplitude (r=0.993, n=46, p<0.001). Results from birds with unperturbed wings Variation of frequency and amplitude was low within hovering bouts, as evidenced by the closeness of the fits of the data and the sine curve. Between-bout frequencies within individuals varied around the mean by an average of 4.9% (range 2.4-8.9%), and amplitudes varied 6.3% (range 2.8-11.7%). The relationship between wing length and wingbeat frequency was insignificant (r=-0.54, n=10, p=0.11). Stroke amplitude and wingbeat frequency of unperturbed wings varied inversely between individuals (Fig. 10; r = -0.70, p= 0.02). Individual birds' wingbeat frequencies and stroke amplitudes in both treatment groups across experimental periods can be found in Figure 11. Based on my analysis using blade element theory, the outer 20% of the wing is responsible for over 30% of all of the lift produced by the wing (Fig. 12). The lift distribution along the wing, along with relative wing areas, can be found in Figure 12. General results after feather removal Fits of sine curves to the data deteriorated slightly (by approximately 1%) after I removed either P9 or P10, with significantly greater residual standard errors 39 141 3 1 3 6 T3 50 52 54 56 58 60 62 Wingbeat Frequency (Hz) Figure 10: Average wingbeat frequency (Hz) and stroke amplitude (degrees) from the 10 experimental birds with unperturbed wings (solid line; r= -0.699, p= 0.024, n=10). The dotted line represents the line if a. 15% change in one parameter would result in a 15% change in the other. Birds'masses ranges from 3.04-3.53 grams. 40 P10 Birds I -i 1 1 1 1 1 1 1 1 H 1 1 1 i 1 2 3 1 2 3 1 2 3 1 2 1 2 3 Y 9 5 Y 9 6 X71 X 7 3 X 7 6 7 0 £ 65 >. o c CD cr « 60 CD C 55 P9 Birds o / / / V 0 \\ \ \J / / S s \ h »\ r f r \ 1 1 2 3 X 7 4 2 3 X72 2 3 X 7 5 2 3 Y 0 5 2 3 Y 6 2 Figure 11: Wingbeat frequency (closed circles and solid line) and stroke amplitude (open circles and dotted line) for individuals in P10 (top) and P9 (bottom) treatment groups. Period 1 is baseline, with unperturbed wings, and Periods 2 and 3 are immediately after and one week after plucking. Individuals' indentifications are in the x-axis, below period labels. In all figures, values are means ± 1 S.E.M., n=5. 41 T3 CD > O E 0 CO CD CM CO CM CM CM CO CD c d co CM CM CM 0 0 h-• i - CO CD - i -CO CD CO CD C 0 CD o d o d CO d T 5 CD > o E CD rr CO CD 8 ^ -o 0-C0 CD CD 15 E | 2 CD CO ^ T " 0 > 0 > < g jjg .= O) co x : ±z D ) Q - S c JO < 5 © E o o i o o "O CO co fc 0 3 o> ~ CO JO 2 2 E 0 p CO Q £ 0 0 • > CJ5 0 > CO -C .<—*-• (0 0 •*-» t : > c co 0 E c 0 p 0 I 0 02 L O j 2 - g O r> j o 0 0 is W £ 13 D ) 0 — •E "55 g b o * C Q . O | o ' « E •= 0 3 Eg c < I ™ CM a? = ^ - 0 § CD _£Z . 2 3 JC i3 O ) * ; 0 ~ 5 OC 42 2.0 1.5 1.0 0.5 0.0 Figure 13: Average relative changes from baseline (period 1) in the residual standard errors of fits of sine curve to angular wingstroke data between treatment groups, P9 (open circles, dotted line) and P10 (closed circle, solid line). Period 2 is immediately following feather removal and period 3 is 7-11 days after feather removal. In all figures, values are means ± S.E.M., and n=5 for each treatment group. 43 in Periods 2 and 3 than Period 1 (Fig. 13; repeated-measures ANOVA: period F 2,15=4.7, p=0.03; treatment Fi,8=0.24, p=0.64; interaction F2,15=0.63, p=0.55). Mean errors remained elevated in Period 3, although they decreased insignificantly in the P9 group. This observation alone suggests an increase in hovering costs during wing molt. Results from the P10 treatment group Changes in wing morphology Removing P10 left wing length, and therefore wing disc loading, unchanged since P9 and P10 are of equal length in the outstretched wing. Overall wing area decreased 5.3% ± 1.0 S.E. after removing P10, but the airfoil was left essentially intact. The wingtip area was reduced by more than 21%, with all of the removed area from the outer 3/5 , h of the wing (Fig. 12). After one week, wings appeared morphologically the same as immediately after plucking. Are frequency changes in the P10 group consistent with oscillator theory? Only three of five individuals in the P10 treatment group significantly increased wingbeat frequency with the removal of P10 (Fig. 11). However, if the data are examined as a group, wingbeat frequency increased significantly in Period 2 (Fig. 14; repeated-measures ANOVA F2,7=26.3, p<0.001), which is consistent with the prediction from oscillator theory for a decrease in the mass of the wing. Frequency increased again in Period 3 (Fig. 14), unexpectedly since the wing mass should have remained the same during that time. 44 * 6 CD CT LL CO n 4 CD c?2 CO Figure 14: Change in wingbeat frequency of P10 treatment birds, as percent change from baseline (Period 1). Periods are as defined in Figure 13. 45 Are kinematic changes in the P10 group consistent with blade element theory? Stroke amplitude was increased significantly immediately after removing P10, and remained elevated in Period 3 (Fig. 15; repeated-measures ANOVA F2,7=6.93, p=0.02). P10 birds increased both wingbeat frequency and stroke amplitude with the removal of P10, which resulted in a significant increase in the velocity-squared term in Period 2 and another in Period 3 (Fig. 16; repeated-measures ANOVA F2,7=43.72, p<0.01). The Period 2 increases were predicted with a reduction in wing area. The additional increase in V 2 in Period 3 was unanticipated, since wing area remained the same since plucking. Using blade element theory, wingbeat frequency, stroke amplitude, or a combination of both, needed to increase by 5% to compensate for the decrease in area (S). Average frequency increased by 2.7% when P10 was removed, and consequently, amplitude may have been increased to complete requirements. According to these calculations, a 10% increase in the velocity-squared term was necessary, and it increased by an average of 14.6%. Results from the P9 treatment group, compared to the P10 group Changes in wing morphology Wing area was reduced 8.7 % ± 4.5 S.E. by removing P9, and as in the P10 group, all of the area removed was from the outer 3/5 t h of the wing. In sharp contrast to the P10 group, removing P9 resulted in a large disruption of the wing surface. P10 was supported only at the base of the shaft, and during hovering, it 46 Figure 15: Change in stroke amplitude in P10 treatment group, as measured by percent change from baseline (Period 1). Periods are as defined in Figure 13. 47 25 0 1 2 3 Period Figure 16: Changes in velocity-squared term, as measured by the product of wingbeat frequency and stroke amplitude squared, in the P10 treatment group. Results are measured as percent change from baseline values (Period 1). Periods are as defined in Figure 13. 48 separated from the rest of the wing, creating a large gap between P8 and P10. P10 was pushed down and behind the wing as the wing began moving upward towards the end of the backstroke (Fig. 3). The gap was widest as the wing flipped over and started to accelerate at the beginning of the forestroke (Fig. 17; average gap size = 8.3 mm ± 0.6 S.E.). It persisted throughout the wingstroke and gave the wing a dragonfly-like two-part appearance in the overhead view (Fig. 18). With P10 effectively disengaged from the wingtip, the area that produces most of the lift (Fig. 12) decreased more dramatically than in the P10 birds. Remarkably, all P9 birds greatly reduced or entirely closed the gap within a week. I filmed one P9 bird, Y05, for 7 consecutive days after feather removal and measured gap sizes at the beginning of the forestroke. Mean gap size decreased gradually at first, then narrowed to less than half its width between days 4 and 5 (Fig. 19). Comparing P9 frequency changes to the P10 group, and are they consistent with oscillator theory? I compared changes in frequency between treatment groups and found significant period and treatment effects (repeated-measures ANOVA using arcsine-transformed data: period F2,15=46.80, p<0.001; treatment Fi,s=5.75, p=0.04). The interaction between period and treatment was also significant (F2,15=4.94, p=0.02). P9 birds increased frequency more than P10 birds in Period 2, but the difference between groups disappeared in Period 3 (Fig. 20). Although 49 P10 from left wing Figure 17: Frame from sideview video (500 fps) of Y05 showing the gap created in the wing by removing primary 9 (P9). Figure 18: Gap in X75's wing with P9 removed, midway through the backstroke, from overhead video (500 fps). 50 1 2 3 4 5 6 7 Days after removing P9 Figure 19: The gap size in Y05's right wing for days 1 through 7 following the removal of P9. 51 1 2 3 Period Figure 20: Mean wingbeat frequency changes of the P9 (closed circles and solid line) and P10 (open circles and dotted line) treatment groups. Period 1 represents birds with unperturbed wings. Periods 2 and 3 are immediately after, and approximately 1 week after feather removal, respectively. In all figures, values are means ± 1 S.E.M. 52 the increase in frequency is consistent with oscillator theory, I had predicted the same change in frequency in both treatment groups, due to approximately equal amounts of mass removed from the wings. Comparing kinematic changes of P9 birds to P10 birds and are they consistent with aerodynamic theory? Stroke amplitude varies insignificantly with treatment or period, and there was no interaction (period F2,15=3.26, p=0.07; treatment Fi,s=1.31, p=0.29; interaction F2,15=2.81, p=0.09), although trends between the two treatment groups were apparent. In Period 2, P9 birds decreased stroke amplitude insignificantly, compared to the increase in P10 birds (Fig. 21). This reduction in amplitude was unexpected, as I had predicted the P9 birds to increase amplitude more than the P10 birds to compensate for a smaller effective wing area. Mean changes in amplitude disappeared one week later due to a large increase in the P9 group (Fig. 21). I had predicted this result provided the P9 birds could compensate somehow for the gap in the wing. V 2 (the square of the product of frequency and amplitude) varied significantly with period (period F2,15=53.34, p<0.001), but not with treatment, and there were no interaction effects (treatment Fi,8=0.17,p=0.69, interaction F2,15=0.10, p=0.91). Kinematic increases in both treatment groups resulted in the same increase in velocity-squared in Period 2, and again in Period 3 (Fig. 22). I had predicted a larger increase in velocity-squared in the P9 group than in the P10 53 Figure 21: Mean stroke amplitude percent changes from baseline. Symbols and Periods as defined in Figure 20. 54 1 2 3 Period Figure 22: Percent change in the velocity squared from baseline of the two treatment groups. V was measured as the product of stroke amplitude and wingbeat frequency squared. Periods and symbols are as defined in Figure 20. 55 group because of a smaller effective wing area, nor did I anticipate the additional increase from Period 2 to Period 3. Other kinematic measurements in both treatment groups Contrary to my predictions, the geometric angle of attack during the backstroke, when measured at the wing base, decreased significantly with period (F2,15=10.29, p<0.01) but not with treatment (Fi,8=0.34, p=0.58), and there was no interaction (F2,15=0.24, p=0.79). The angle of attack decreased in both treatment groups after plucking and remained lower than baseline values for at least one week (Fig. 23). As predicted, there were no significant changes in body angle, stroke plane angle, or other angle of attack measurements. See Table 4 for a summary of these analyses. Discussion I had predicted that P10 birds would increase frequency with the removal of P10, and additionally increase amplitude to complete the compensation for a reduced wing area. In contrast, I predicted P9 birds would increase frequency to the same extent as the P10 birds, but increase amplitude more, to compensate for a smaller effective wing area. These results would indicate higher energetic costs of hovering in the P9 group, simulating the basic molt pattern, compared to the P10 group, simulating the hummingbirds' non-sequential pattern. Although results 56 -8 H 2 Period 3 Figure 23: Relative changes from baseline in backstroke angle of attack at the wingbase. Periods and symbols as defined in Figure 20. CD _sz O ) cc CD != CD W CD « C D £ c CC CD CL E o o > ? CO CD Q -CO & CD CO C CO CD o ' _ £ = CD 2 . 2 0 E *- o < .fc > co O CD ^ g> < CO S o CO CO co CD CO E 0 5 1 co -a Jg 0 co -•—» CO o CD CLQ_ CD ~" >- T3 v_ CO *" Q. co CD CO =3 Q . t l Q_ CT . . C CD © E J D CO CO CD CD CO CD CO E o "0 CD CD LO O O o CO CO CD .9-cb «3 CD C Q CO CO B £ CO ZS CO CD CD C CO o 0 c *i CO g "co O c0 9-'-*= CL co CO CO CO CO co CD o c CD T3 T3 O o Q. ^ * m c -CD CM E w CO CD T3 ^ O LO I S , c E oo' +J CO T~ CD CO CO E (0 i _ (0 a. c 0 Z J a l CD CO CD . Q CD •a Z ! -t—• t < CD o -*—» CO CD CD < CD C c < | CD o m LO CO LO CM CO d LO CM d LO CM Si 00 o d LO co CM LO CO o CM LO CD CO ' f t c 1 o a c 1 < o » CD E o CD CD CO CD O « CO 0 cm c 0 O LO CO CD O t_ -•—< CO CD 0 CO CO . Q D5 C 0 CO c col , co o '•g CD cr-O LO -4—' CO , ® o -4—" co o CO CO from the P9 group were unexpected, net findings may suggest higher costs of hovering in the P9 group, as predicted. Immediate results After losing their P10 feathers, hummingbirds changed wingbeat kinematics in conventional and predictable ways. Wing areas were reduced, yet wings maintained the smooth, continuous surface important for them to function as airfoils (Muller & Patone 1998). According to my analysis based on blade element theory, P10 birds increased wingbeat frequency and stroke amplitude more than necessary to maintain lift. Interestingly, they also decreased the angle of attack of their backstroke, which would decrease their coefficient of lift. The net change in kinematic response in this treatment group appears to compensate for the loss of wing area, closely adhering to the principles of blade element theory. Removing P9 simulated the basic molt pattern that hummingbirds evolved away from, and P9 birds changed their wingbeat kinematics in unconventional and unexpected ways. Removing P9 left P10 with no support from inner feathers and separated from the rest of the wing by a gap, which persisted through the wingstroke, disrupting the smooth surface of the airfoil. This altered wing morphology unquestionably changed the airflow circulating around the wing, but how? Conceivably, the isolated P10 acted as a small airfoil and aided in lift generation, similar to two-winged dragonfly flight. More likely, isolating P10 affected flight performance negatively. The gap was located in a section of the airfoil with very low pressure on both the upstroke and downstroke (Fig. 24). This would allow air to escape through the gap, probably increasing profile drag (Tucker 59 Figure 24: Pressure distribution across a cross section of an airfoil. Drawing from Muller & Patone (1998). 1991) and decreasing lift. In addition, lift production was likely severely affected by removing P9, as this would effectively decrease the wingtip as if both feathers were gone (Fig. 12). After P9 was removed, some birds appeared to hover unsteadily in front of the feeder, as if off balance. Occasionally, I had to reject entire P9 hovering bouts for analysis because of large variations in wingbeat frequency during the 2-second video sample. Assuming resonance, there are several possible explanations for the larger increase in wingbeat frequency by P9 birds than P10 birds. Forces bent and isolated the P10 feather throughout the wingstroke, so effective wing length was likely based on the shorter P8, which would indicate a higher frequency. Another possibility is that the mass of P10 may have effectively disengaged from the mass of the wing, resulting in a larger effective decrease in wing mass than P10 birds (two feathers lost rather than one) and a larger increase in frequency. Yet another explanation is that birds "chose" to increase frequency above resonance. 60 Birds can control their wing kinematics (Pennycuick 1996, Wells 1993b). Wells (1993b) found strong evidence of this when his experimental hummingbirds reduced stroke amplitude when approaching a large-faced feeder, probably to avoid damaging their wings, and compensated by increasing frequency. Four of five P9 birds in my study reduced stroke amplitude after I removing P9 (Fig. 11). These P9 birds may have "chosen" to beat their wings at a frequency above the natural resonant frequency to compensate for lowering their amplitude, which if true, would increase flight costs (Wells 1993b, Greenewalt 1990, Pennycuick 1996). In mammals, higher stride frequencies are more energetically expensive than lower rates (Heglund & Taylor 1988), so why would birds in this study opt for this tactic? Perhaps the P10 feathers sent turbulent signals through their shafts to the central nervous system, and birds reduced amplitude to calm the signals. Most birds have feathers called filoplumes around the bases of their remiges with sensory corpuscles at their bases that detect subtle movements, and these are thought to help birds judge the position of their flight feathers (Clark & DeCruz 1989). Additionally, there is evidence that birds can detect subtle changes in airflow and airspeed on the wing surface via mechanoreceptors on or near feather follicles (Brown & Fedde 1993). Using either or both of these mechanisms, birds should be able to detect an erratically moving feather, and they could alter kinematics in response. According to vortex theory, a decrease in stroke amplitude results in a smaller vortex sheet, requiring greater air circulation to maintain lift, and therefore, a greater lift coefficient (Rayner 1979b). This might explain why both treatment 61 groups increased velocity equally (Fig. 22). To cope with the unnatural perturbation I imposed, P9 birds may have adjusted kinematics in ways I did not analyze or could not detect using my methods. Despite a decreased backstroke angle of attack, changes in other kinematics could have resulted in an overall elevated lift coefficient. One P9 bird moderately increased both amplitude and frequency after plucking, more resembling the changes of a P10 than a P9 bird, despite a gap comparable in size to the other treatment birds. Interestingly, that bird (Y05) hovered with a lower baseline stroke amplitude than any of the other 9 individuals, so perhaps a reduction wasn't aerodynamically feasible. There were no other apparent characteristics to explain why he compensated in this manner. Chai (1997) found that naturally molting ruby-throated hummingbirds reduced frequency an average of 4.8%, but did not significantly change amplitude. This reduction in frequency was unexpected, since an increase in frequency is expected following a reduction in wing mass (from missing feathers), and Chai could offer no explanation. Chai studied naturally molting birds that were undergoing many physiological changes including a significant loss of body mass. Also, birds were tested during molt of the inner primaries, which contribute little to the moment of inertia of the wing (Wells 1993a). The tissues that support the growth of the new outermost primaries would already be forming, which may contribute more to the moment of inertia of the wing due to their more distal position on the wing. An analysis of the mass distribution along molting wings may help explain Chai's results. The only other published data on frequency changes 62 during molt, to my knowledge, is from Sachs (1968), who reported a 17% increase in frequency in molting curlews to attain flight speeds typical of non-molting birds. One week later At first glance, the increased frequency in P10 birds during the week after plucking is puzzling. Wing mass and therefore frequency should have remained approximately constant during this period. But kinematic changes made just after natural or manipulated changes to the wing could be an immediate but inefficient response (Swaddle etal. 1999). Within a week, birds could have learned or otherwise developed a more effective and efficient way to regain performance. Chai (1997) found similar results in one naturally molting hummingbird whose wingtips he cut off. This bird initially increased wingbeat frequency and amplitude by 4.2% and 10.5%, respectively, resulting in a 32% increase in velocity-squared. At the same time, this bird also increased its lift coefficient by more than 30%. Eight days later, although overall wing area hadn't changed, this bird's frequency and amplitude were an additional 6% and 2% higher, respectively. An aerodynamic trade-off may have occurred, for this added increase in velocity-squared corresponded with a 30% decrease in the lift coefficient, or a return to pre-manipulation values. This would suggest that over time, changing kinematics that increase the lift coefficient is a less efficient way of regaining lift performance than increasing frequency and amplitude. The ability of P9 birds to reduce or close the gap within one week is impressive, but how they mechanically pulled P10 in to join the rest of the wing is 63 unclear. It appeared highly effective, however, for stroke amplitude and frequency changes of the P9 group were virtually identical to those of the P10 group in Period 3 (Figs. 21 & 20). The decrease in gap size that occurred between days 4 and 5 after plucking (Fig. 19) was remarkable. There are two primary muscular systems in the skin that can move feathers (see Stettenheim 1972), so it is possible that P9 birds employed muscular control at the base of their primaries to close the gap. But, muscle development is typically slow and gradual. Another possibility is that birds employed a change in kinematics to close the gap. This requires a further look. Historically, aerodynamic and vortex theories have assumed a constant angle of attack (Chai 1997, Wells 1993a & b, Ellington 1984a, Hertel 1963). The assumption of an unchanging angle of attack may be accurate for intact wings, but not for molting or damaged wings, and warrants further investigation. The importance of gaps in wings The ecological and aerodynamic importance of gaps in wings is becoming increasingly evident. Holmgren etal. (1993) found that large gaps, partly due to rapid primary molt, were correlated with lower body masses in migrating dunlins (Calidris alpina). Lower body mass may strategically reduce flight costs, or may be a consequence of increased flight costs. Furthermore, migrating dunlins have adapted a slower primary molt that results in smaller wing gaps compared to resident sub-species. 64 There have been recent reports of declines in other flight performance measures caused by gaps in wings. Slagsvold & Dale (1996) simulated molt gaps in pied flycatchers and found higher predation levels than birds with non-manipulated wings. Gaps may have compromised turning ability and therefore, the ability to avoid predation because of reduced wing area (also see Hedenstrom 1998 and Howland 1974). Swatter & Witter (1997) found that wing gaps negatively effected flight and take-off speeds, and take-off trajectory. These ecological consequences of gaps probably act as strong selective forces. One adaptive benefit of the "stepwise" primary molt pattern may involve gaps. Terns, shags, and most pelecaniformes (pelicans, gannets, and others) follow this pattern, where two or three molt waves start simultaneously at different loci on the wing and progress distally (Rasmussen 1988, King 1974, Potts 1971, Ashmole 1968). Gaps created by several feathers at one location on the wing are larger than gaps created by the loss of single primaries, which will be almost covered by adjacent feathers. It is therefore aerodynamically beneficial to molt equal numbers of primaries at several locations rather than at one location along the wing (Prevost 1983, Ashmole 1968). In several species, including hummingbirds, primaries are dropped less frequently as the molt wave reaches the wingtip (Chai 1997, Chai etal. 1999, Baltosser 1995, Jenni & Winkler 1994, Spina & Massi 1992, Stiles & Wolf 1974, Haukioja 1971, Newton 1967). One explanation is that feathers on other parts of the body begin molting at the same time the molt wave reaches the outer wing, resulting in a slower pace because of metabolic constraints (Spina & Massi 1992). 65 However, this deceleration is also seen in birds with prolonged molt periods and therefore low metabolic stress (Jenni & Winkler 1994). Also, spotted flycatchers (Muscicapa straita) that follow a reverse primary molt also molt their outer primaries slower, when other feathers are not growing (Dorka 1971, as cited in Jenni & Winkler 1994). A more likely explanation is that gaps in outer primaries are more detrimental to flight ability than gaps in inner primaries, and slower molt would minimize gap sizes near the wingtip (Tucker 1991, Gwinner 1966). Hedenstrom & Sunada (1999) studied the aerodynamic effects of wing gaps using a rectangular model wing and found that both gap size and position reduced aerodynamic performance. As expected, larger gaps had a greater effect on performance than smaller gaps. Gaps in the middle of the model wing, simulating gaps created by the loss of inner primaries, resulted in the highest reduction in flight performance. Interestingly, gaps in the outer area of the model wing, i.e. the wingtips, actually increased performance. One major difference between the model wing and a hummingbird wing is that the wingtip gaps are parallel and perpendicular to the airflow, respectively. The anecdotal and experimental evidence discussed previously suggests that gaps in the wingtip impair flight performance more than gaps in the inner wing. Wingtips are especially important in hovering (Chai 1997, Ellington 1984b, Weis-Fogh 1973, Epting 1980, Fig. 12). Chai (1997) found lift production severely reduced after cutting the wingtips of a hummingbird. The power required to hover greatly increases with change to primaries 8, 9 or 10 via molt or breakage (Epting 66 1980). Also, shorter wings result in higher wing disc loading, which increases induced power (Ellington 1984d, Epting & Casey 1973, Hainsworth & Wolf 1972). Wingtip shape and wing length change constantly during the molt of the 2 outermost primaries of a hummingbird wing. Comparing wing lengths following the basic pattern and the hummingbirds' non-sequential pattern leads to interesting findings. Following the basic sequence, the tip of P8 would define the effective wing length for a longer period of time than the non-sequential pattern, since when P9 drops, the unsupported P10 is largely unable to contribute to the wing disc (Fig. 25a). Following the non-sequential pattern, P8 is the leading feather only after both P10 and P9 are dropped, but P10 is already 50% regrown (Fig. 25e) and soon will be entirely replaced. The new P10 would be better able to withstand aerodynamic forces, its proximal half supported by the growing, stiff P9 feather (Fig. 25f). Considering this, I agree with Epting's (1980) comments that the non-sequential molt pattern of hummingbirds maintains a longer wing length compared to the basic pattern, but results of this study suggest the adaptive benefit of the non-sequential molt pattern is more complex. Darwin (1859) assumed that natural selection preserves individuals with characteristics that best suits them for life in their particular environment, and eliminates those less equipped. The selection of this primary molt pattern in all 319 species of hummingbirds, and that it is so strictly followed to by individuals within species (Stiles 1995a, Baltosser 1995), both strongly suggest that this pattern provides a substantial benefit to all hummingbirds relative to the basic pattern. Results from this study suggest that this unique pattern is, at least in part, 67 c » CD .E n - £ CD 9- > co o ^ > n -o - 0 CD ^ Q -§ I i ° - 2 . § -11 _ o) P « 2 « 2 § ? a c > E w © g co P. 'I 5 i — c o> © CO T3 — C W 0. C . CD CD u. co -rr-—jr $ S . S | E? co * O ) " c C O - O - f l O L -— -a t c a (B Q . ' _ £ , X > g CD • -2 CD £ w c a. 8 -CD CD _ - -x: x: S j E ** > 9 CD -2-0 ) > -fe t i r ' c .> e j 5 * CO Q . -5 "TO .2 Jj- b — ra co S ^ CO x: co CO r- — 0 ) - = CO c 5 c § ^ CT CO CD CO £ i CO C CL O ^ c — CO •— _  iB Q . ^ CD § Q. co xi x: S &££<•» £ ?H « 2 Z> > > 0 > - 0 CO CO o •— O) LI ® Q. g OL CD c ••c o CL CL 3 CO d) c '5 o a> 68 related to hummingbirds' unique hovering ability. This study experimentally examines only one snapshot of the entire molt sequence hummingbirds undergo annually, although it is an important one. It suggests possible adaptive benefits of hummingbirds' unique non-sequential primary molt pattern, and further establishes the ecological and biomechanical importance of gaps in wings. This study confirms the inadequacy of current flight theories for estimating energetic costs of hovering with molting or damaged wings, although further research in this area would likely be difficult and complex. Although my experimental design didn't permit the measurement of the moments of inertia of the wings, it would have allowed quantitative predictions of frequency changes and would be helpful in future work. It would also be useful to estimate power input during hovering by measuring oxygen consumption in the two treatment groups. Finally, further investigation into hummingbirds' flight systems as resonant systems is required. 69 Chapter 3: Behavior of captive molting hummingbirds Introduction Feathers can compose up to 12% of a bird's body mass (Chilgren 1977, Murphy 1996) and must be replaced at regular, usually annual, intervals. Molt is an expensive period of the avian annual cycle. The direct cost of molt is the energy required to produce new feathers. However, there are also indirect costs resulting from the decreased function of missing or partially replaced feathers (Jenni & Winkler 1994) as well as consequences, such as increased risk of predation or starvation. Birds may compensate for these increased costs by increasing food intake, using body reserves and/or decreasing activity (King 1981). Birds have an increased need for protein during molt, especially for sulfur amino acids that are used directly to produce feathers (Murphy 1996, Murphy & King 1982). If necessary, birds can forage selectively to meet nutritional needs (Murphy 1996, Murphy & Pearcy 1993, Murphy & Olson 1991, Murphy & King 1989 & 1987), but little is known about any changes in food quantities ingested during molt or changes in feeding strategies. Many bird species, including hummingbirds, lose mass before or during molt (Chai 1997, Swaddle & Witter 1997, Hiebert 1993, Tucker 1991, Seel 1976, Newton 1969 & 1968, personal observation). This may be strategic to lower flight costs (Swaddle & Witter 1997, Chai 1997, Wells 1993a&b, Hiebert 1993), but other changes may also compensate for the costs of molt. Most birds are less active during molt (Murphy 1996, Jenni & Winkler 1994, Ginn & Melville 1983), whether in captivity (Swaddle & Witter 1997, Wijnandts 70 1984, Newton 1966, Eyster 1954) or in their natural environment (Brown & Bryant 1996, Bryant & Tatner 1988, Bailey 1985, Cooper 1978, Haukioja 1971). Most reports in the literature are anecdotal, with few specific descriptions or quantitative studies. Anecdotal reports have stated that molting birds fly less (Wijnandts 1984, Haukioja 1971), become more secretive (Morton & Morton 1990, Newton 1966), sing less (Morton & Morton 1990), and preen more often (Cooper 1978, Newton 1966). There are two known quantitative experiments measuring behavioral changes during molt. Brown & Bryant (1996) found that wild dippers (Cinclus cinclus) spend more time resting and greatly reduce the amount of time spent flying and diving during natural molt, presumably from elevated flight (and diving) costs. In another experiment, Swaddle & Witter (1997) simulated primary molt in captive starlings (Sturnus vulgaris). Birds with manipulated flight feathers perched significantly more and flew less than control birds, with no significant differences in the time spent vocalizing, preening or in social interactions. This suggests that changes in the amount of time spent flying could reflect the increased costs of flight, but changes in singing, preening and interacting seen during natural molt may be due to the birds' overall physiological condition. The literature on behavioral changes in molting hummingbirds is strictly anecdotal. During molt, both territoriality (Stiles 1973, Williamson 1956, Pitelka 1951) and the frequency of display flights are decreased (Pitelka 1951, Williamson 1956). Contradictory to these reports, Chai et al. (1996) noted an increase in aggression during molt in captive ruby-throated hummingbirds (Archilochus 71 colubris). Also, Stiles (1973) observed one Anna hummingbird {Calypte anna) maintain a high level of territoriality during molt, and suggested that behavioral changes depend on the individual. The purpose of my study was to investigate the effects of molt on the behavior of captive hummingbirds. Timing and pattern of molt in captive rufous hummingbirds {Selasphorus rufus) closely follows that of wild birds, who undergo one complete annual molt while wintering in Mexico (Chai 1997, Hiebert 1993, personal observation). Although details and frequencies can vary, captive hummingbirds demonstrate many of the same behaviors as wild birds (personal observation). Over a 21-week period, I used focal animal sampling (Altmann 1974) to estimate the number and lengths of flights, feeding bouts and aggressive encounters before, during and after the natural molt of 12 individual rufous hummingbirds. I predicted that captive molting hummingbirds would fly less and fight less during molt, reflecting high flight costs and the overall condition of the birds. Changes in feeding tactics during molt, if any, would depend on whether birds preferred to carry more in their crop each bout but less often, or carry less in their crop more often. Any behavioral changes could reflect the increased costs and changes in physiological condition inherent in the molting process Materials and Methods Birds I used 7 male and 5 female rufous hummingbirds (Selasphorus rufus) in this experiment. I captured the birds in June 1998 at two British Columbia sites; 72 Sumas Mountain and Rosewall Creek on Vancouver Island. I housed the birds communally in the Vivarium on the UBC Campus in five aviaries ranging in size from 1.3m wide X 2.5m long X 2.5m high to 2.6m X 6.2m X 2.5 m, and changed the photoperiod routinely to reflect seasonal changes. I fed the birds QuikoNectar (Quiko GmbH, Germany), a commercial hummingbird nectar, supplemented with live Drosophila, and provided feeders in excess to avoid exclusion from feeding of submissive birds by dominant birds. I made all observations between November 1998 and April 1999. I marked some of the experimental birds below the gorget with non-toxic highlighter pens to distinguish between individuals during observations. I noticed no behavioral changes in either marked birds or in other birds in the aviaries and therefore assume in this experiment that the marking did not affect behavior in any way. Observation Room I built an observation room measuring 2.5m wide X 2.2m long X 2.5m high (Fig. 26) and hung 4 feeders and several branches for perching. I hung black plastic sheeting cut into strips from the ceiling to provide places for birds to hide (Newberry & Shackelton 1997), but positioned them as not to obstruct my view from the observing room. I observed from a small adjacent room with a one-way mirror film built into the common wall. 73 ^ =feeder =plastic sheeting Figure 26: Plan of observation room (2.5x2.2x2.5m) and observing room. The observation window was a 0.6m X 1.8m one-way mirror. Bird Rotation I rotated birds through the observation room in groups of four; always 2 males and 2 females, where they remained for up to 7 days. I weighed each bird at the beginning of each rotation. Birds' masses were always obtained in the mornings before birds could feed to ensure clear digestive tracts (Hiebert 1993, Diamond etal. 1986). To minimize the effects of social structure and group dynamics on individual results, I shuffled the combinations of birds with each rotation. 74 Observation Procedure I used focal-animal sampling (Altmann 1974) to record all transitions and their times of occurrence (rounded to the nearest second) for each bird using a DOS computer program running on a laptop computer. I recorded when birds left and returned to perches, and when feeding bouts and aggressive encounters began and ended. A feeding bout began and ended when the bird inserted its beak into or removed it from a feeder nozzle. I defined an aggressive encounter as any of the following: a. One bird displaced, or attempted to displace, another from its perch. b. The bird changed its speed to chase or flee. c. The bird changed its flight path to chase or flee. I calculated frequencies and durations from these data and totaled them to give time invested in each activity. I noted only the frequency and not the duration of flights that involved obvious searches for, pursuits of, or capturing Drosophila. The number of insect bouts cannot be used to estimate the number of insects eaten, since success rates of searches depended greatly on elapsed time since I released Drosophila and the number of insects released, and I could not always know whether attempts were successful. Observations began 1-3 days after I put the birds into the observation room. I observed birds within the first 3 hours of the day, for 1-2 hour sessions, and recorded all of the above activities for each of the 4 birds during 15-minute periods. I observed each rotation (group of 4 birds) for a total of 4 hours. I 75 observed birds in a pre-determined, pseudo-random order with each bird in each of the four 15-minute time slots within that hour. Quantity of Liquid Food Eaten Intermittently during molt and after molt, I weighed the four feeders before I put them into the observation room and again the following day when I removed them. From this, I calculated the difference in weights to calculate the amount of food the four birds had eaten during the previous day. I did this for a total of 14 and 13 days during and after molt, respectively, with several different rotations. Statistics I used one-way repeated-measures ANOVA to compare the effects of three periods: before, during and after molt. Period was the independent variable in the analysis, and the average durations, number per hour, and percent time engaged in behavioral acts were dependent variables. To analyze the quantity of food eaten during and after molt, I used a two-sample t-test. Rejected Data One male bird lost its tail during handling from "fright molt" (King & Murphy 1984, Payne 1972, Stettenheim 1972) on September 18, 1998, and it had grown back entirely by October 22, 1998. When I started data collection in November, this bird had already lost primaries 1 through 5 and mostly regrown them, so no pre-molt data could be collected. This bird then interrupted molt for approximately 76 5 weeks. In exchange for a loss of statistical power in repeated-measures analysis of variance, a missing datum may be estimated from surrounding values, if and only if the sample is representative (Girden 1992). I rejected all data from this bird from analysis because it was unrepresentative. Results All birds molted over a four month period, from late November through late March. Individuals took an average of 87 days (± 2.8 S.E.) to complete molt. The molt pattern of the experimental hummingbirds followed the pattern described for all hummingbirds (Baltosser 1995, Stiles 1995a, Williamson 1956, Aldrich 1956). Primary molt began with innermost feathers and progressed outward in sequence until primary 8 (P8). After P8 was regrown, P10 dropped before P9. Birds lost primaries 1 through 5, 6 or 7 in rapid sequence, so they temporarily flew with only 3 to 5 outer primaries. This has been described previously in both captive and wild hummingbirds (Chai era/. 1999, Chai era/. 1998, Chai 1997, Baltosser 1995, Stiles & Wolf 1974). Primary molt typically lasted the entire molt period, except in male birds, whose iridescent gorget took an additional 10-14 days to complete (also reported in Baltosser 1995, Stiles 1995a). Tail molt generally started with the innermost pair and progressed outward. Birds' masses peaked in early November at 4.5-5.0 grams, approximately 2-3 weeks before molt started, then began to decrease. Masses were at a minimum approximately 2 weeks after the onset of molt, at 3.0-3.2 grams, and 77 increased again only after molt was complete. These seasonal mass changes were typical for hummingbirds (Chai etal. 1998, Chai 1997, Hiebert 1993). One bird's molt sequence was altered by an early loss of his tail from "fright molt" (see Methods). While this bird was in interrupted molt, his mass was between 4.0-4.8 grams, but dropped to 3.4 grams right before molt resumed with the loss of P6. Interestingly, molt resumed at a time that permitted this bird to finish molt at approximately the same time as the other experimental birds. I observed 32 groups of different combinations of four individuals for a total of 107.5 hours; 19 hours before onset of molt, 56.5 hours during molt and 32 hours after molt. In 14 rotations (groups of 4 birds), one individual defended 2 or more feeders, excluding more submissive birds from feeding. I discontinued the rotation if and when the excluded bird's weight dropped to below 3.0 grams, or if I considered its health to be at risk. A total of 6 rotations were stopped; one rotation before any data had been collected and 5 more after partial data had been collected. I extrapolated partial data to estimate the entire rotation. Figure 27 illustrates average time budgets of the experimental birds before, during and after molt. With one exception, the amount of time non-molting, captive hummingbirds spent engaged in recorded activities were within ranges reported for wild hummingbirds (see Table 5; Stiles 1995b, Hixon & Carpenter 1988, Boyden 1978, Stiles 1971, Wolf & Hainsworth 1971, Pearson 1954). Not unexpectedly, captive birds spent only an average of 1.8% of their time feeding, while estimates for wild birds range from 5.7 to 32.8% (Stiles 1995b, Hixon & Carpenter 1988, Boyden 1978, Greenwalt 1975 using data from Pearson 1954, 78 79 Stiles 1971, Wolf & Hainsworth 1971). Captive birds fed from "bottomless flowers", i.e. one location that supplied them with an ad libitum food supply, while wild birds must visit many flowers over a much larger area to meet their energetic requirements. Near the end of February, some male birds began performing mating displays, including shuttling in front of a female or the mirror. During March, males would allow females to feed at their feeders, then shuttle and attempt to mate with them. At the same time, in two rotations, females attempted to protect 2 feeders and defended them aggressively. Time Flying As predicted, experimental birds flew significantly less (about half as much) during molt, than before or after molt periods (Fig. 28; arcsine square root transformed data; F2,2o=13.27, p<0.001). Molting birds significantly decreased the frequency of flights (Fig. 28; F2,2o=11.90, p<0.001), but increased average flight duration (Fig. 28; log transformed data; F2,2o=7.23, p=0.004). Before molt, birds developed stereotypic behavior in the form of repetitive circular or figure-8 flights that made up approximately 13% of their activity budget. This behavior stopped altogether as birds began to molt, but resumed as molt ended at a higher level, approximately 22% of their time. 80 c o CO CO Q. E o o CD i ZJ CO CD CD x: E o CO C 'E E x: § ^ M - CO co ,3 CD I -o « CD ? CL O co .g C CO 2 is CD CD ^ CO TO > - ' H - CO O "0 £ 5 CD CO CO c "D n JO. _ | ^ co o o E . | l 11 CO c < g LO cD-CD > CO CO I— o 'E E E g LI E t> iZ G> ^ CO c 0) . i 2 »-•>. D) c (0 CO o o Q. CO 0 o D o CO LO o p LO d LO CD CD CO _CD CO co CO CO co d d co o o 0 0 CD CD -t—< c CD Q. CO O oS c o X X q LO CO co co d LO CD LO CD CO o CO 0 0 T-o p o p 0 0 CO CO CO h -CD LO CD CNJ oS CO LO LO LO 0 0 CO LO CD CO o 0 0 0 0 CD CD -t—» c CD CL i— CO O ob c o 2 13 co c c CO .CO CO c c CO o ui o 1 ^ CD x: o c CD o CO c o CO CO CD CL co c 'CO X o CD CO _CD CO oo d • LO d CO CO CO LO CM c o CM CM O o • A d T _ d CM d 1 CD d T— T— CM 1 oo CD CO CM 1 h -00 CM CM 1 ^ LO CO CM " I — 0 0 CO CO LO CM d * t CM CM CO CM 1 4 LO CO CO CO CO CO 0 0 d CO LO CO CM CO 1— T CO I 00 LO 2 CO T3 O) C o 4_, w (A 81 Before Molt During Molt Period After Molt Figure 28: The proportion of time spent flying (top), frequency of flights (center) and average duration of flights in seconds (bottom) before, during, and after molt. In all figures, values are means ± 1 S.E.M. 82 Feeding Behavior Feeders I compared the frequency and duration of feeding bouts before, during and after molt. Birds spent significantly less of their time hovering at feeders during molt, than before or after periods (Fig. 29; arcsine square root transformed data; F2,2o=4.15, p=0.031), and birds fed significantly less frequently (Fig. 29; F2,2o=7.01, p<0.005). Feeding bouts were significantly longer during molt than before molt, and remained moderately elevated after molt (Fig. 29; log transformed data; F2,2o=3.58, p=0.047). Molting birds ate significantly less liquid hummingbird food (t=2.18, p=0.029, df=25), and foraged for live Drosophila more often during molt than before or after molt (Fig. 30; F2,2o=13.92, p<0.001). Aggression Against predictions (but see Chai era/. 1996), birds in molt were more aggressive. They spent significantly more time engaged in aggressive encounters than pre- and post-molting birds (Fig. 31; arcsine square root transformed data; F2,2o=12.58, p<0.001). Both the number per hour and the durations of aggressive encounters significantly increased during molt (Fig. 31; log transformed data; F2,2o=26.45, p<0.001 and F2,2o=6.51, p=0.007, respectively). This increased aggression resulted in several discontinued rotations, when I considered an excluded bird's health to be at risk. 83 Before Molt During Molt Period After Molt Figure 29: The proportion of time experimental birds spent feeding at feeders (top), the frequency (center) and average duration of feeding bouts in seconds (bottom) before, during and after molt. Values are means ± 1 S.E.M. 84 Figure 30: The average number of times per hour birds left their perches to search for and/or catch live Drosophila before, during and after molt. Values are means ± 1 S.E.M. 85 Before Molt During Molt After Molt Period Figure 31: The proportion of time birds spent engaged in aggressive encounters (top), the frequency (center) and average duration of aggressive encounters (bottom) before, during and after molt. Values are means ± 1 S.E.M. 86 During molt, more dominant birds engaged more often in a particularly offensive form of aggression towards submissive birds. If a bird was excluded from feeding by the other 3 birds (i.e. one bird protected 2 feeders instead of 1), territory holders would chase the submissive bird immediately before feeding. The aggression typically occurred without any provocation from the submissive bird who was usually sitting in a non-threatening and distant position within the observation room. I coined this "preventative aggression", and will address it further in the Discussion section. Discussion Captive hummingbirds altered several aspects of their behavior during molt. They flew less and left their perches only to feed or fight. They altered their feeding strategy, possibly as not to have to fly as often, ate less liquid food and more insects, and engaged in more territorial aggression. Eating less liquid food may be voluntary to reduce and maintain a lower mass, or to reduce the costs of flying with impaired wings. Swaddle & Witter (1997) found that starlings spent slightly, but insignificantly, less time feeding after they had undergone a simulated molt. Another possibility is that birds compensated for less liquid food by an increased intake of insect matter. Wild, non-molting hummingbirds feed primarily on flower nectar, but also spend up to 6% of their time foraging for insects (Stiles 1995b, Hixon & Carpenter 1988, Wolf & Hainsworth 1971, Stiles 1971, Pearson 1954). According to Brice & Grau (1991), non-breeding (and non-molting) hummingbirds can meet protein 87 requirements in 6 minutes of daily insect foraging in nature. But during molt, birds have an increased need for protein, in particular, the sulfur-based amino acids (Murphy 1996). Although most bird species can meet these demands without selective foraging (Murphy & King 1992), molting hummingbirds may be forced to forage selectively since flower nectar lacks proteins, lipids and other essential nutrients (Baker & Baker 1982, Brice & Grau 1991). My captive molting birds hunted extensively for insects. They searched visually from their perches, hawked flying insects out of the air, or gleaned stationary insects off of walls, floor and ceiling of the observation room. When I brought vials of insects into the aviaries for release, birds would occasionally fly to within a few feet of me to hawk insects as they flew out of the vials. Selective foraging is one way molting birds can adjust to different nutritional needs (Murphy 1996, Murphy 1994, Murphy & Pearcy 1993, Murphy & King 1989 & 1987): they can also reduce activity. Flying less may result in muscle atrophy, which may release amino acids for use in the synthesis of new feathers (Murphy 1996). The decrease in flying time may also offset some of the energy costs of molt (Davis 1955, Lindstrom etal. 1993), or may result from the increased costs of flight incurred while flying with impaired wings (Jenni & Winkler 1994). Although the weight of food in a crop is always important (Tamm 1986, DeBenedictis etal. 1978, Hainsworth 1978), it must be especially so when flying with molting wings. In the observation room, birds typically chose perches close to the feeder they defended, so flying costs to and from foraging sites were minimal. Because there is always some handling time involved in foraging, longer feeding 88 bouts are more efficient than shorter ones (Gass & Roberts 1992, Gass & Montgomerie 1981). This implies that under these circumstances, the marginal cost of harvesting a larger load is less than those of harvesting smaller bouts more frequently. The costs of territory defense would also be low in captivity. Like wild birds, captive rufous hummingbirds set up territories, albeit small ones. They protected feeders as well as surrounding air space and nearby perches, either by defensive chirps or by chasing. Aggression was increased during molt. Chai era/. (1996) also reported increased aggression in captive, molting ruby-throated hummingbirds {Archilochus colubris). Most aggression occurred when one or two birds increased their territories to include two feeders instead of one. This left one or two birds without feeders, and they were attacked wherever they attempted to feed. Is there a motivation to increase territory size during molt? Despite efforts to provide captive birds with complete nutrition, perhaps the diet is not nutritionally complete and birds attempt to increase territories in efforts to complete it. Birds may be driven to increase their foraging area to ensure adequate resources in this demanding part of their annual cycle. Preventative aggression occurred when a territory owner chased another bird without provocation. This usually happened immediately before the owner would feed, commonly in the same flight. Further research is necessary to determine if this increased aggression occurs in wild birds, or if it develops in response to captive living. Stereotypies often develop in animals in captivity, where movement is constrained. They are defined as repetitive and invariant behavioral patterns that 89 have no apparent function (Mason 1991). Although stereotypies are very diverse (Mason 1991), caged birds tend to develop two types. "Route-tracing" is when birds repeated fly along precise paths within the cage, and "spot-picking" is when birds repeatedly touch their beak to a nearby object (Keiper 1969 & 1970). The captive hummingbirds in this study developed both of these forms of stereotypic behavior. Theories about why these behaviors develop are many: stereotypies may develop to alleviate boredom, may be thwarted escape attempts, or may be displacement activities (see Mason 1991 for complete review, Meyer-Holzapfel 1968). Birds may become frustrated being unable to perform behaviors they are internally motivated to do (Mason 1991, Meyer-Holzapfel 1968, Morris 1964), such as protecting territories, or searching for food or mates, and they develop stereotypies as displacement behavior (Morris 1966). I watched birds in the early stages of developing stereotypic flying and agree with the displacement theory. Comparing activity budgets of experimental birds and wild birds further supports this theory. Experimental birds flew approximately 22% of the time stereotypically in the post-molt period, compared to wild hummingbirds, who average of 24% of the time in flight (range 11.1-38.5%, see Table 5). Could it be that non-molting captive birds are "driven" to fly this much of the time, even though foraging and territorial defense require much less flying? Birds who developed stereotypic flying engaged in little aggressive behavior towards other birds in the observation room, despite appearing to be very aware of surrounding events. For example, one bird, while flying stereotypically, deviated from his route to hawk an insect. Birds would also watch others fly above or feed 90 at a feeder, or change flight paths as not to collide with others also flying stereotypically. Stereotypies and aggression don't seem to be strongly related, since some birds engaged in a great deal of stereotypic flying continued to aggressively defend their feeder. Overall, this study has shown that hummingbirds compensate for the costs of molt in several ways. The decrease in flying time is especially important, since feeding hummingbirds hover, which a very expensive form of flight. The results of this study reflect consequences of natural molt, which includes many physiological changes and processes, many we don't yet fully understand. Future research should continue to expose more of the physiological aspects of molt. Other research may include the behavioral changes of the highly-territorial rufous hummingbird during natural molt in Mexico. 91 Chapter 4: Conclusions and General Discussion This thesis examines the consequences of molt in hummingbirds on two very different time scales. Chapter 2 examines short-term changes in response to altered wing morphology to address long-term consequences: why has natural selection, over thousands of years, resulted in this unique primary molt pattern in hummingbirds? In Chapter 3, the scale is greatly reduced to a seasonal and daily level and uncovers how captive hummingbirds compensate for molt by changing their daily behavior. In Chapter 2, I explored possible adaptive benefits of the unique, non-sequential primary molt pattern of hummingbirds. Hummingbird wings are rigid and have relatively large hand areas and primaries, probably related to their hovering abilities. If they followed the ancestral molt pattern a large gap in the surface of the airfoil would be created in the wingtip when P9 was lost, in the most aerodynamically important part of the wing during hovering. In experimental simulations of hummingbirds' natural molt pattern, birds used conventional, predictable mechanisms to increase lift. In contrast, birds resorted to very unconventional mechanisms in simulations of the ancestral molt pattern. The gap created by removing P9 was large and in an aerodynamically vital area of the wing, and probably resulted in high aerodynamic costs. These birds somehow closed or greatly reduced the width of the gap within one week. This experiment demonstrated how hummingbirds cope kinematically with changes in their wing morphology. This is the first 92 known study of possible adaptive and aerodynamic benefits of a primary molt pattern. Chapter 3 is the first quantitative study of behavioral changes during molt in captive hummingbirds. Experimental subjects altered their behavior during molt in many ways, probably to compensate for increased energetic costs and consequent physiological requirements. Molting birds perched more of the time and flew only to fight or feed. Diets changed during molt to include less liquid food and more insects. They probably fed more efficiently by feeding longer at the feeders, but less often. In nature, birds may be motivated to expand territory size during molt, which manifested in the experiment as an increase in territorial aggression, possibly due to demands for resources. Given the high cost of territorial defense, however, this exploration is incomplete. Together, results from these studies reveal different mechanisms to minimize the costs of molt in hummingbirds. The mechanism of changing behavior during molt is a common strategy seen in most, if not all, bird species. The unique primary molt pattern most likely reflects the unique flight mode and morphology of a highly specialized bird. It may provide a way to minimize aerodynamic penalties to birds that depend heavily on flight. 93 Literature Cited Aldrich, E.C. 1956. Pterlyography and molt of the Allen Hummingbird. Condor 58:121-133. Alexander, R.M. 1988. Elastic mechanisms in animal movement. Cambridge University Press, Cambridge. Altmann, J . 1974. Observational study of behavior: sampling methods. Behaviour 49:227-265. Amadon, D. 1966. Avian plumages and molts. Condor 68:263-278. Ankney, C D . 1979. Does the wing molt cause nutritional stress in Lesser Snow Geese? Auk 96:68-72. Ashmole, N.P. 1968. Breeding and molt in the white tern (Gygis alba) on Christmas Island, Pacific Ocean. Condor 70:35-55. Bailey, R.O. 1985. Protein reserve dynamics in postbreeding adult male redheads. Condor 87:23-32. Baker, H.G. and Baker, I. 1982. Chemical constituents of nectar in relation to pollination mechanisms and phylogeny. pp. 131-171 In: Biochemical Aspects of Evolutionary Biology. M.H. Nitecki, Editor. University of Chicago Press, Chicago & London Baltosser, W.H. 1995. Annual molt in Ruby-throated and Black-chinned Hummingbirds. Condor 97:484-491. Batschelet, E. 1981. Circular Statistics in Biology. Academic Press, London. Boyden, T.C. 1978. Territorial defense against hummingbirds and insects by tropical hummingbirds. Condor 80:216-221. Brice, A T . and Grau, C.R. 1991. Protein requirements of Costa's hummingbirds Calypte costae. Physiological Zoology 64(2):611-626. Brown, C.R. and Bryant, D.M. 1996. Energy expenditure during molt in dippers (Cinclus cinclus): no evidence of elevated costs. Physiological Zoology 69(5):1036-1056. Brown, R.E. and Fedde, M.R. 1993. Airflow sensors in the avian wing. Journal of Experimental Biology 179:13-30. Bryant, D.M. and Tatner, P. 1988. Energetics of the annual cycle of dippers Cinclus cinclus. Ibis 130:17-38. 94 Carpenter, F.L., Hixon, M.A., Russell, R.W., Paton, D.C., and Temeles, E.J. 1993a. Interference asymmetries among age-sex classes of rufous hummingbirds during migratory stopovers. Behavioural Ecology and Sociobiology 33:297-304. Carpenter, F.L., Hixon, M.A., Temeles, E.J., Russell, R.W., and Paton, D.C. 1993b. Exploitative compensation by subordinate age-sex classes of migrant rufous hummingbirds. Behavioural Ecology and Sociobiology 33:305-312. Chai, P., Chang, A .C. , and Dudley, R. 1998. Flight themogenesis and energy conservation in hovering hummingbirds. Journal of Experimental Biology 201:963-968. Chai, P., Chen, J .S .C . , and Dudley, R. 1997. Transient hovering performance of hummingbirds under conditions of maximal loading. Journal of Experimental Biology 200:921-929. Chai, P. 1997. Hummingbird hovering energetics during moult of primary flight feathers. Journal of Experimental Biology 200:1522-1536. Chai, P., Altshuler, D.L., Stephens, D.B., and Dillon, M.E. 1999. Maximal horizontal flight performance of hummingbirds: effects of body mass and molt. Physiological and Biochemical Zoology 72(2): 145-155. Chai, P. and Dudley, R. 1995. Limits to vertebrate locomotor energetics suggested by hummingbirds hovering in heliox. Nature 377:722-725. Chai, P. and Dudley, R. 1999. Maximum flight performance of hummingbirds: capacities, constraints, and trade-offs. American Naturalist 153(4):398-411. Chai, P., Harrykissoon, R. and Dudley, R. 1996. Hummingbird hovering performance in hyperoxic heliox: effects of body mass and sex. Journal of Experimental Biology 199:2745-2755. Chai, P. and Millard, D. 1997. Flight and size constraints: hovering performance of large hummingbirds under maximal loading. Journal of Experimental Biology 200:2757-2763. Chantler, P. 2000. Swifts: A guide to the swifts and treeswifts of the world. 2nd edition. Yale University Press, New Haven & London. Chilgren, J.D. 1977. Body composition of captive White-Crowned Sparrows during postnuptial molt. Auk 94:677-688. Clark, G.A. and DeCruz, J.B. 1989. Functional interpretation of protruding filoplumes in oscines. Condor 91:962-965. 95 Cooper, J . 1978. Moult of the black-footed penguin. International Zoo Yearbook 18:22-27. Corben, H.C. 1983. Wing-beat frequencies, wing area and masses of flying insects and hummingbirds. Journal of Theoretical Biology 102:611-623. Darwin, C. 1859. On the Origin of Species by Means of Natural Selection, or the Preservation of Favoured Races in the Struggle for Life. 1st edition. John Murray, London. Davis, E.A. 1955. Seasonal changes in the energy balance of the English sparrow. Auk 72:385-411. DeBenedictis, P.A., Gill, F.B., Hainsworth, F.R., Pyke, G.H. and Wolf, L.L. 1978. Optimal meal size in hummingbirds. American Naturalist 112(984):301-316. DeRoo, A. 1966. Age-characteristics in adult and subadult swifts, Apus a. apus (L.), based on interrupted and delayed wing-moult. Gerfaut 56(2): 113-134. DeSwardt, D.H. 1992. Distribution, biometrics and moult of Gurney's sugarbird Promerops gurneyi. Ostrich 63:13-20. Diamond, J .M. , Karasov, W.H., Phan, D. and Carpenter, F.L. 1986. Digestive physiology is a determinant of foraging bout frequency in hummingbirds. Nature 320:62-63. Dickinson, M.H., Lehmann, F.O. and Sane, S .P. 1999. Wing rotation and the aerodynamic basis of insect flight. Science 284:1954-1960. Dietz, M.W., Daan, S. and Masman, D. 1992. Energy requirements for molt in the Kestral (Falco tinnunculus). Physiological Zoologist 65:1217-1235. Disselhorst G. 1961. Ascendente Handschwingen-Mauser bei Muscicapa striata. Journal fur Ornithologie 102:360-366. Douthwaite, R.J. 1976. Weight changes and wing moult in the Red-billed Teal. Wildfowl 27:123-127. Dow, D.D. 1973. Flight moult of the Australian honeyeater Myzantha melanocephala (Latham). Australian Journal of Zoology 21:519-532. DuBowy, P.J. 1985. Seasonal organ dynamics in post-breeding male blue-winged teal and northern shovelers. Comparative Biochemistry and Physiology 82A(4):899-906. Dudley, R. 1999. Unsteady aerodynamics. Science 284:1937-8. 96 Ellington, C P . 1978. The aerodynamics of normal hovering flight: three approaches, pp. 327-345 in: Comparative Physiology: Water, Ions and Fluid Mechanics. K. Schmidt-Nielsen, L. Bolis and S.H.P. Maddrell, Editors. Cambridge University Press, Cambridge Ellington, C P . 1984a. The aerodynamics of hovering insect flight. I. The quasi-steady analysis. Philosophical Transcripts of the Royal Society of London B 305:1-15. Ellington, C P . 1984b. The aerodynamics of hovering insect flight. II. Morphological parameters. Philosophical Transcripts of the Royal Society of London B 305:17-40. Ellington, C P . 1984c. The aerodynamics of hovering insect flight. III. Kinematics. Philosophical Transcripts of the Royal Society of London B 305:41-78. Ellington, C P . 1984d. The aerodynamics of hovering insect flight. IV. Aerodynamic mechanisms. Philosophical Transcripts of the Royal Society of London B 305:79-113. Ellington, C P . 1984e. The aerodynamics of insect flight. V. A vortex theory. Philosophical Transcripts of the Royal Society of London B 305:115-144. Ellington, C P . 1984f. The aerodynamics of hovering insect flight. VI. Lift and power requirements. Philosophical Transcripts of the Royal Society of London B 305:145-181. Epting, R.J. 1980. Functional dependence of the power for hovering on wing disc loading in hummingbirds. Physiological Zoology 53(4):347-357. Epting, R.J. and Casey, T.M. 1973. Power output and wing disc loading in hovering hummingbirds. American Naturalist 107(958):761-765. Eyster, M.B. 1954. Quantitative measurement of the influence of photoperiod, temperature, and season on the activity of captive songbirds. Ecological Monographs 24(1): 1-27. Gass, C L . and Mongomerie, R.D. 1981. Hummingbird foraging behavior: decision-making and energy regulation, pp. 159-194 In: Foraging Behavior. A .C . Kamil and T.D. Sargent, editors. Garland S T P M Press, New York & London. Gass, C L . and Roberts, W.M. 1992. The problem of temporal scale in optimization: three contrasting views of hummingbird visits to flowers. American Naturalist 140(5):829-850. 97 Ginn, H.B. and Melville, D.S. 1983. Moult in Birds Volume 19. The British Trust for Ornithology. Hertfordshire, England. Girden, E.R. 1992. ANOVA: Repeated Measures. Sage Publications, Newbury Park, CA. Greenewalt, O H . 1975. The flight of birds. Transactions of the American Philosophical Society 65(4): 1-65. Greenewalt, C.H. 1990. Hummingbirds. Reprint. Dover Publications, Inc. New York. Gwinner, V .E . 1966. Der zeitliche Ablauf der Handschwingenmauser des Kolkraben (Corvus coraxL.) und seine funktionelle Bedeutung. Die Vogelwelt 87(5): 129-133. Hagiwara, S., Chichibu, S. and Simpson, N. 1968. Neuromuscular mechanisms of wing beat in hummingbirds. Zeitschrift Fur Vergleichende Physiologie 60:209-218. Hainsworth, F.R. and Wolf, L.L. 1972. Power for hovering flight in relation to body size in hummingbirds. American Naturalist 106(951 ):589-596. Hainsworth, F.R. 1978. Feeding: models of costs and benefits in energy regulation. American Zoologist 18:701-714. Halse, S.A. and Dobbs, J .C. 1985. Carcass composition of spur-winged geese with comments on reproduction and molting. African Journal of Ecology 23:171-178. Hanmer, D.B. 1981. Mensural and moult data of nine species of sunbird from Mocambique and Malawi. Ostrich 52:156-178. Haukioja, E. 1971. Flightlessness in some molting passerines in Northern Europe. Ornis Fennica 48:101-116. Hedenstrom, A. 1998. The relationship between wing area and raggedness during molt in the Willow Warbler and other passerines. Journal of Field Ornithology 69(1):103-108. Hedenstrom, A. and Sunada, S. 1999. On the aerodynamics of moult gaps in birds. Journal of Experimental Biology 202:67-76. Heglund, N.C. and Taylor, C.R. 1988. Speed, stride frequency and energy cost per stride: how do they change with body size and gait? Journal of Experimental Biology 138:301-318. 98 Hertel, H. 1963. Structure, Form, Movement. Reinhold Publishing Corporation, New York. Hiebert, S. 1993. Seasonal changes in body mass and use of torpor in a migratory hummingbird. Auk 110(4):787-797. Hixon, M.A. and Carpenter, F.L. 1988. Distinguishing energy maximizers from time minimizers: a comparative study of two hummingbird species. American Zoology 28:913-925. Holmgren, N., Ellegren, H. and Pettersson, J . 1993. The adaptation of moult pattern in migratory Dunlins Calidris alpina. Ornis Scandinavica 24:21-27. Howland, H.C. 1974. Optimal strategies for predator avoidance: the relative importance of speed and manoeuvrability. Journal of Theoretical Biology 47:333-350. Jenni, L. and Winkler, R. 1994. Moult and ageing of European passerines. Academic Press, London. Johnsgard, P.A. 1997. 77?e Hummingbirds of North America. 2nd edition. Smithsonian Institution Press, Washington D.C. Keiper, R.R. 1969. Causal factors of stereotypies in caged birds. Animal Behaviour 17:114-119. Keiper, R.R. 1970. Studies of stereotypy functions in the canary {Serinus canarius). Animal Behaviour 18:353-357. King, J.R. 1981. Energetics of avian molt. Acta International Ornithol. Congress 27:312-317. King, J.R. 1974. Seasonal allocation of time and energy resources in birds, pp. 4-70 In: Avian Energetics. R.A. Paynter Jr., Editor. Nuttail Ornithological Club, Cambridge King, J.R. and Farner, D.S. 1961. Energy metabolism, thermoregulation and body temperature. In: Biology and Comparative Physiology of Birds, Volume II. A . J . Marshall, Editor.: Academic Press, New York, London King, J .R. and Murphy, M.E. 1984. Fault bars in the feathers of white-crowned sparrows: dietary deficiency of stress of captivity and handling? Auk 101:168-169. Kodric-Brown, A. and Brown, J .H. 1978. Influence of economics, interspecific competition, and sexual dimorphism on territoriality of migrant rufous hummingbirds. Ecology 59(2):285-296. 99 Lindstrom, A., Visser, G.H. and Daan, S. 1993. The energetic cost of feather synthesis is proportional to basal metabolic rate. Physiological Zoology 66(4):490-510. Lustick, S. 1970. Energy requirements of molt in cowbirds. Auk 87:742-746. Marquardt, D.W. 1963. An algorithm for least-squares estimation of non-linear parameters. Journal of the Society of Industrial and Applied Mathematics 11:431-441. Mason, G .J . 1991. Stereotypies: a critical review. Animal Behaviour 41:1015-1037. Mees, G.F. 1985. Comments on species of the genus Hirundapus (Aves: Apodidae). Proceedings of The Koninklijke Nederlandse Akademie Van Wetenschappen. Series C: Biological and Medical Sciences 88(1 ):63 Meyer-Holzapfel, M. 1968. Abnormal behavior in zoo animals, pp. 476-503 In: Abnormal Behavior. M.W. Fox, Editor. W.B. Saunders Company, Philadelphia. Moore, J.L. 1997. Ecomorphology of rufous hummingbirds: an investigation of maneuverability and agility in four age-sex classes. University of British Columbia. MSc. Thesis. Morris, D. 1964. The response of animals to a restricted environment. Symp. Zool. Soc. London. 13:99-118. Morris, D. 1966. Abnormal rituals in stress situations. Transcripts of the Royal Society of London B 251:327-330. Morton, G.A. and Morton, M.L. 1990. Dynamics of postnuptial molt in free-living mountain white-crowned sparrows. Condor 92(4):813-828. Muller, W. and Patone, G . 1998. Air transmissivity of feathers. Journal of Experimental Biology 201:2591-2599. Murphy, M.E. 1994. Dietary complementation by wild birds: considerations for field studies. Journal of Bioscience 19(4):355-368. Murphy, M.E. 1996. Energetics and Nutrition of Molt. pp. 158-198 In: Avian Energetics and Nutritional Ecology. C. Carey, Editor. Chapman & Hall, New York. Murphy, M.E. and King, J.R. 1982. Amino acid composition of the plumage of the white-crowned sparrow. Condor 84:435-438. 100 Murphy, M.E. and King, J.R. 1987. Dietary discrimination by molting white-crowned sparrows given diets differing only in sulfur amino acid concentration. Physiological Zoology 60(2):279-289. Murphy, M.E. and King, J.R. 1992. Energy and nutrient use during moult by white-crowned sparrows Zonotrichia leucophyrs gambelii. Ornis Scandinavica 23:304-313. Murphy, M.E. and King, J.R. 1989. Sparrows discriminate between diets differing in valine or lysine concentrations. Physiology & Behavior 45:423-430. Murphy, M.E. and Olson, D.B. 1991. Conrtol of daily protein intake by white-crowned sparrows (WCS). American Zoologist 31:116A. Murphy, M.E. and Pearcy, S.D. 1993. Dietary amino acid complementation as a foraging strategy for wild birds. Physiology & Behavior 53:689-698. Murphy, M.E. and Taruscio, T.G. 1995. Sparrows increase their rates of tissue and whole-body protein synthesis during the annual molt. Comparative Biochemistry and Physiology 111 A(3):385-396. Murphy, M.E., Taruscio, T.G., King, J.R. and Truitt, S .G . 1992. Do molting birds renovate their skeleton as well as their plumages? Osteoporosis during the annual molt in sparrows. Canadian Journal of Zoology 70:1109-1113. Newberry, R.C. and Shackleton, D.M. 1997. Use of visual cover by domestic fowl: a Venetian blind effect? Animal Behaviour 54:387-395. Newton, I. 1966. The moult of the bullfinch Pyrrhula pyrrhula. Ibis 108:41-67. Newton, I. 1967. Feather growth and moult in some captive finches. Bird Study 14:10-24. Newton, I. 1968. The temperatures, weights, and body composition of molting bullfinches. Condor 70:323-332. . Newton, I. 1969. Moults and weights of captive Redpolls Carduelis flammea. Journal fur Ornithologie 110:53-61. Noordhuis, R. 1989. Patronen in slagpenrui: oecofysiologische aanpassingen. Limosa 62:35-45. Norberg, U.M. 1990. Vertebrate Flight. Springer-Verlag, Berlin. Norberg, U.M. 1996. Energetics of Flight. In: Avian Energetics and Nutritional Ecology. Cynthia Carey, Editor. 101 Payne, R.B. 1972. Mechanisms and control of molt. pp. 103-155 In: Avian Biology, Volume II. D.S. Farner and J.R. King, editors. Academic Press, New York and London. Pearson, O.P. 1954. The daily energy requirements of a wild Anna hummingbird. Condor 56(4):317-322. Pennycuick, C .J . 1969. The mechanics of bird migration. Ibis 111:525-556. Pennycuick, C .J . 1975. Mechanics of flight, pp. 1-75 In: Avian Biology, Volume 5. D.S. Farner and J.R. King, editors. Academic Press, London, New York. Pennycuick, C .J . 1990. Predicting wingbeat frequency and wavelength of birds. Journal of Experimental Biology 150:171-185. Pennycuick, C .J . 1996. Wingbeat frequency of birds in seady cruising flight: new data and improved predictions. Journal of Experimental Biology 199:1613-1618. Piersma, T. 1988. Breast muscle atrophy and constraints on foraging during the flightless period of wing moulting Great Crested Grebes. Ardea 76:96-106. Pitelka, F.A. 1951. Ecological overlap and interspecific strife in breeding populations of Anna and Allen hummingbirds. Ecology 32(4):641 -661. Potts, G.R. 1971. Moult in the shag Phalacrocoraxaristotelis, and the ontogeny of the "Staffelmauser". Ibis 113:298-305. Prevost, Y. 1983. The moult of the osprey Pandion haliaetus. Ardea 71:199-209. Proctor, N.S. and Lynch, P.J. 1993. Manual of Ornithology: Avian Structure and Function. Yale University Press, New Haven and London. Pyke, G.H. 1980. The foraging behaviour of Australian honeyeaters: a review and some comparisons with hummingbirds. Australian Journal of Ecology 5:343-369. Rasmussen, P.C. 1988. Stepwise molt of remiges in blue-eyed and king shags. Condor 90:220-227. Rayner, J.M.V. 1979a. A new approach to animal flight mechanics. Journal of Experimental Biology 80:17-54. Rayner, J.M.V. 1979b. A vortex theory of animal flight. Part 1. The vortex wake of a hovering animal. Journal of Fluid Mechanics 91:697-730. 102 Rayner, J.M.V. 1985. Bounding and undulating flight in birds. Journal of Theoretical Biology 117:47-77. Rusch, H.M., Pytte, C.L. and Ficken, M.S. 1996. Organization of agnostic vocalizations in Black-chinned Hummingbirds. Condor 98(3):557-566. Ruschi, A. 1962. A muda nos Trochilidae. Bol. Mus. Prof. Mello-Leitao 38:1-16. Sach, V .G . 1968. Die Masuer des Grossen Brachvogels. Journal Fur Ornithologie 109:485-511. Schieltz, P.C. and Murphy, M.E. 1995. Diurnal variation in oxygen consumption by molting and nonmolting sparrows. Comparative Biochemistry and Physiology 112A(2):265-272. Seel , D.C. 1976. Moult in five species of Corvidae in Britian. Ibis 118:491-527. Sibley, C . G . and Ahlquist, J .E. 1990. Phytogeny and Classification of Birds. Yale University Press, New Haven & London. Skead, C . J . 1967. The Sunbirds of Southern Africa; also the sugarbirds, the white-eyes and the spotted creeper. South African Bird Book Fund, Cape Town. Slagsvold, T. and Dale, S. 1996. Disappearance of female pied flycatchers in relation to breeding stage and experimentally induced molt. Ecology 77(2):461-471. Spina, F. and Massi, A. 1992. Post-nuptial moult and fat accumulation of the Ashy-headed Wagtail (Motacilla flava cinereocapilla) in Northern Italy. Die Vogelwarte 36:211-220. Stettenheim, P. 1972. The integument of birds, pp. 1-45 In: Avian Biology, Volume II. D.S. Farner and J.R. King, editors. Academic Press, New York and London. Stiles, F.G. 1971. Time, energy, and territoriality of the Anna hummingbird {Calypte anna). Science 173:818-821. Stiles, F.G. 1972. Age and sex determination in Rufous and Allen hummingbirds. Condor 74:25-32. Stiles, F.G. 1973. Food supply and the annual cycle of the Anna Hummingbird Volume 97. University of California Press, Berkeley. Stiles, F.G. 1995a. Intraspecific and interspecific variation in molt patterns of some tropical hummingbirds. Auk 112(1):118-132. 103 Stiles, F.G. 1995b. Behavioral, ecological and morphological correlates of foraging for arthropods by the hummingbirds of a tropical wet forest. Condor 97:853-878. Stiles, F.G. and Wolf, L.L. 1974. A possible circannual molt rhythm in a tropical hummingbird. American Naturalist 108(961 ):341 -354. Stolpe, M. and Zimmer, K. 1939. Der schwirrflug des Kolibri in zeitlupenfilm. Journal fur Ornithologie 87:136-155. Stresemann, E. 1963. Taxonomic significance of wing molt. Proceedings of the XIII International Ornithological Congress:171-175. Stresemann E. 1967. Inheritance and adaptation in moult. Proceedings of the 14th International Ornithological Congress: 75-80. Stresemann, E, Stresemann, V. 1966. Die mauser der vogel. Journal fur Ornithologie 107(Sonderheft). Stresemann, V, Stresemann, E. 1960. Die Handschwingenmauser der Tagraubvogel. Journal fur Ornithologie 101:373-403. Swaddle, J .P. , Williams, E.V. and Rayner, J.M.V. 1999. The effect of simulated flight feather moult on escape take-off performance in starlings. Journal of Avian Biology 30:351-358. Swaddle, J .P. , Witter, M.S., Cuthill, I.C., Budden, A. and McCowen, P. 1996. Plumage condition affects flight performance in Common Starlings: implications for developmental homeostasis, abrasion and moult. Journal of Avian Biology 27:103-111. Swaddle, J .P . and Witter, M.S. 1997. The effects of molt on the flight performance, body mass, and behavior of European starlings (Sturnus vulgaris): an experimental approach. Canadian Journal of Zoology 75:1135-1146. Tamm, S. 1986. Behavioural Energetics: acquisition and use of energy by hummingbirds. Stockholm, Sweden: University of Stockholm. Tucker, V.A. 1991. The effect of molting on the gliding performance of a Harris' hawk (Parabuteo unicinctus). Auk 108:108-113. Tucker, V.A. 1995. Drag reduction by wing tip slots in a gliding Harris' Hawk, Parabuteo unicinctus. Journal of Experimental Biology 198:775-781. Wagner, H.O. 1946. Observaciones sobre lavida de Calothorax lucifer. An. Inst. Biol. Mex. 17:283-299. 104 Wagner, H.O. 1955. The molt of hummingbirds. Auk 72:286-291. Walsberg, G.E. 1983. Avian Ecological Energetics, pp. 161-220 In: Avian Biology. Volume VII. Academic Press, New York, London. Watson, G.E. 1968. Synchronous wing and tail molt in diving petrels. Condor 70:182-183. Weis-Fogh, T. 1972. Energetics of hovering flight in hummingbirds and in Drosophila. Journal of Experimental Biology 56:79-104. Weis-Fogh, T. 1973. Quick estimates of flight fitness in hovering animals, including novel mechanisms for lift production. Journal of Experimental Biology 59:169-230. Wells, D.J. 1993a. Muscle performance in hovering hummingbirds. Journal of Experimental Biology 178:39-57. Wells, D.J. 1993b. Ecological correlates of hovering flight of hummingbirds. Journal of Experimental Biology 178:59-70. Wijnandts, H. 1984. Ecological energetics of the long-eared owl (Asio otus). Ardea 72:1-92. Williamson, F.S.L. 1956. The molt and testis cycles of the Anna Hummingbird. Condor 58:342-366. Wolf, L L . 1975. Energy intake and expenditures in a nectar-feeding sunbird. Ecology 56:92-104. Wolf, L.L. and Hainsworth, F.R. 1971. Time and energy budgets of territorial hummingbirds. Ecology 52:980-988. Wolf, L.L. and Hainsworth, F.R. 1977. Temporal patterning of feeding by hummingbirds. Animal Behavior 25:976-989. Wolf, L.L., Hainsworth, F.R. and Gill, F.B. 1975. Foraging efficiencies and time budgets in nectar-feeding birds. Ecology 56:117-128. 105 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items