The Auk 113(4):802-810, 1996

INFLUENCEOF THE TRAILING-EDGENOTCH ON FLIGHT PERFORMANCE OF GALLIFORMS

SERGEI V. DROVETSKI • BurkeMuseum and Department of Zoology,University of Washington,Seattle, Washington 98195, USA

ABsTRACT.--Trailing-edgenotches, formed by shortenedfirst secondaries,characterize the wings of most galliforms. I investigated the function of these notches with comparative measurementsof notch size taken from extended-wing specimensand with experimental studiesof model wings of four representativespecies. Pheasants, quail, and turkeys,all of which use flight to escapepredators, have wide wings and deep notches.Grouse with dark flight muscleshave long, narrow wings with small trailing-edge notchesand typically fly relatively long distancesfrom one foraging site to another. Grousewith light coloredflight muscleshave short,broad wings with large trailing-edgenotches and mostlyfly from ground to canopyor from branchto branch to reachtheir food. Model wings of two pairsof galliforms with different wing shapeswere used in the experiments.White-tailed Ptarmigan(Lagopus leucurus)and Sage Grouse(Centrocercus urophasianus) have small notches,high aspectratios, relatively heavy wing loadings,low maximum lift coefficients,and dark pectoralmuscles. In contrast,Wild Turkey (Meleagrisgallopavo) and California Quail (Callipeplacalifornica) have deep notches,low aspectratios, relatively light wing loadings,high lift coefficients,and light colored pectoral muscles.Experiments using model wings in a water flow tunnel show that the trailing-edgenotch increasesthe maximum lift-to-drag ratio and stabilizesairflow around the wing, but reducesthe maximumlift coefficient.Thus, the trailing-edgenotch increases performancein vertical and slow flight but reducesefficiency in level flight. Sucha function is consistentwith the suite of differencesthese birds show in musclecolor, wing shape,and predominantmode of flight. Received17 November1995, accepted 27 February1996.

BIRD WINGSEXHIBIT morphological adapta- ed. However, only Shestakova(1971) has noted tions for different kinds of flight (Rayner 1988, the existenceof this notch in variousgalliforms Norberg 1989).Regardless of the species,slots and some gulls. between the primaries increasethe lift-to-drag In this paper, ! presentcomparative data on ratio by increasinglift and reducing drag (es- the size of the trailing-edge notch for 31 species peciallyinduced drag) that resultsfrom deflec- of galliforms. I also present resultsof experi- tion of an air streamaround a tilted wing. Such mentsthat showhow this notchinfluences glid- conclusionsabout the functionalsignificance of ing performancein the wings of four species wing slotscome primarily from the established with different wing shapes. correlation between deep slotting and slow, level flight that characterizesbirds with wings MATERIALs AND METHODS that are broad at the tip. Wind-tunnel experi- ments supportthese conclusions(Vinogradov Measurementsof wing length, wing breadth, and 1951; Hofton 1978; Tucker 1993, 1994). trailing-edgenotch size were taken from 220 extend- ed-wing specimensof 31 galliform species(Table 1). Galliformshave deepslots between their pri- I calculated notch size (in %) as: maries and a large alula, but they rarely have been used as subjectsfor aerodynamicstudies (WS - WSs•)/WB, (1) (Shtegman 1953, Shestakova1971). Further- where WBs•is the distancebetween the leading edge more, no study of the aerodynamicsof galli- of the wing and the tip of the first secondary,and forms hasaddressed the functional significance WBis wing breadth measuredfrom the leading edge of the large notch on the trailing edge of the of the wing to the tip of the fifth secondary.I chose wing that is formed by a shortfirst secondary. the fifth secondaryfor these measurementsbecause This feature is exhibited by all galliforms and in somebirds (e.g. turkeys)the wing notch is formed is quite evident when the wing is fully extend- by severalshortened secondary feathers. Wing length is the distancefrom the proximal end of the humerus to the tip of the longest primary, measuredalong a line parallel to the leading edge of the wing. • E-mail: [email protected] Throughout this paper, meansare reported + I SD. 802 October1996] FlightPerformance and the Wing Notch 803

Four speciesof galliformswere chosenfor flight- TABLE1. Trailing-edge notch size and wing length- performanceexperiments based on extremesof wing to-breadth ratio of galliforms. Values are means shape (see Fig. 1) and body size within the order: (with SD in parentheses). White-tailed Ptarmigan ([Lagopusleucurus]; Tetraoni- nae; subadultmale specimen);Sage Grouse ([Centro- Wing cercusurophasianus]; Tetraoninae; adult male); Wild Notch size length-to- (% of wing breadth Turkey ([Meleagrisgallopavo]; Meleagridinae; adult fe- Species n breadth) ratio male); and California Quail ([Callipeplacalifornica]; Odontophorinae;adult male).I refer to themas "mod- Turnicidae el" species.Other speciesof galliformsprovide more Turnix sylvatica 3 12.36 (4.93) 1.77 (0.04) extreme examplesof wing-shapevariation, but my Phasianinae choiceswere limited by the availability of frozen Tetraogallushimalayensis 1 18.14 1.81 specimensat the University of Washington Burke Alectorischukar 3 36.62 (3.54) 1.69 (0.01) Museum for use in creating modelsof wings. Francolinussephaena 3 35.45 (1.60) 1.62 (0.02) I measuredflight performanceby steady-state(con- F. africanus 4 24.98 (2.29) 1.86 (0.05) stant velocity) lift and drag characteristicsof various F. swainsonii 2 47.34 (5.64) 1.57 (0.01) wings.Active flight in thesespecies may not be char- Perdixperdix 6 30.17 (2.34) 1.91 (0.01) acterized by this particular dynamic condition, but P. dauurica 1 36.36 1.78 any differencesin the coefficientsassociated with wing Coturnix coturnix 3 21.57 (2.04) 2.25 (0.05) Lophuraignita 1 35.71 1.58 lift and drag are presumedto apply to flapping, at Gallusgallus 3 38.11 (2.67) 1.48 (0.07) least for fast forward flight. Lift (C•) and drag (Cd) Phasianuscolchicus 13 30.86 (3.20) 1.71 (0.10) coefficientsare defined by the following relation- ships: Odontophorinae Colinusvirginianus 15 32.49 (3.76) 1.61 (0.05) C, = L/(0.5 p SU2), and (2) Callipeplasquamata 2 35.06 (3.46) 1.63 (0.07) C. californica 9 31.84 (3.76) 1.66 (0.05) cd = D/(0.5 p STY2), (3) C. gambelii 2 37.37 (0.58) 1.62 (0.01) where L is the measured lift force, D is the measured Meleagridinae drag force,• is the densityof the fluid medium, S is Meleagrisgallopavo 4 31.40 (2.04) 1.47 (0.10) the plainform area of the wing, and U is the fluid Tetraoninae(light and intermediate velocity relative to the wing. Lift and drag coefficients flight muscles) were measured for model wings. Models were cut from 0.2-mm thick brass foil and soldered to a metal Dendragapusfalcipennis 2 26.50 (3.23) 1.95 (0.11) D. canadensis 10 21.82 (4.29) 1.95 (0.09) rod 2.5 mm in diameter and about 130 mm long. All D. obscurus 19 23.24 (3.31) 1.90 (0.07) modelshad a width of 30 mm (length varied from 46 Bonasabonasia 14 24.52 (2.67) 1.94 (0.07) to 62 mm depending on species).Each model was B. umbellus 22 28.94 (2.49) 1.88 (0.10) slightly bent around the axis of the metal rod, and Tetraoninae(dark flight muscles) the spacebetween the leading end of the model wing Lagopuslagopus 21 22.09 (1.89) 2.09 (0.05) and the rod wasfilled with glue. The bendingcreated L. mutus 14 20.68 (3.08) 2.35 (0.08) camberin the models.The top of the curvature was L. leucurus 8 19.60 (2.86) 2.12 (0.09) 10 mm behind the leadingedge of a modelwing, and Tetrao tetrix 1 18.87 2.08 the height of the curvature,measured as the distance T. urogallus 3 20.67 (1.29) 2.05 (0.16) from a line connectingthe leading and trailing edges T. parvirostris 10 15.17 (2.52) 2.24 (0.11) of a model wing to the top of the curvature, was Centrocercusurophasi- approximately4 mm. The patternsused for preparing anus 10 18.06 (3.54) 2.24 (0.16) the modelswere images(reduced by photocopy)of Tympanuchusphasianel- lus 4 19.25 (2.49) 2.16 (0.05) freshly thawedwings that had beenpinned in a fully T. cupido 7 20.76 (3.17) 2.05 (0.10) extendedposition. Lift and drag coefficientsare functionsof the Reyn- olds number: tunnel. The flow tunnel provided a uniform flow of Re = UL/•, (4) relatively low turbulence intensity (ca. 5%). The working section of the tunnel was square in cross where U is definedas above,L is the length of the section,with sides of 0.2 m and a length of 1.2 m. wing chord, and v is the kinematic viscosityof the The flow in the working sectionhad all four bound- fluid medium of the wing. Consequently,scale mod- aries fixed. During the measurementsI placed each elswere prepared so that in the testingapparatus (see model such that its center was in the center of the below),each model had a Reynoldsnumber relatively working section. Apart from the model, there was similar to that of real wings. only a piece of rod approximately7 cm long in the Lift and drag forceswere measuredin a water flow flow. This rod connected the model to the force trans- 804 Sm•G•V. DItOVETSKI [Auk,Vol. 113

FIG.1. Extended-wingspecimens used as models for the experiments.White-tailed Ptarmigan (Lagopus leucurus;upper left); Sage Grouse (Centrocercus urophasianus; upper right); c31ifornia Quail (Callipepla californica; lowerleft); and Wild Turkey(Meleagris gallopwoo; lower right). Scale under each wing is in inches(upper) and cm (lower).

ducer.Use of a waterflow tunnelprovides high res- ficient remains stable and constant across Re values olutionof lift and dragforces, which, as long asthe of 5 x lip to 5 x l0 s(Vogel 1983). Reynoldsnumber is similar(see below), gives useful Forceswere measured separately with a strain-gauge datafor lift anddrag coefficients. Water speed in the transducer(sensitivity 2 x 10-3 N) as one arm of a flow tunnel was 0.5 m/s. Wheatstonebridge (see Lanyon 1976). The output was If modelsare perfectlyscaled, then the Reynolds amplifiedand collectedby a PC computer.For lift numbersof the modelsand the wingsthey represent measurements,the wing was mounted to the force will be equal. Due to the limited size of the flow tank, transducer with its sensitive axis normal to the flow; constantflow velocity,and the need to keepmodel for drag, the sensitiveaxis was parallel to the flow. geometrysimilar to the real wing geometry,models For all models,lift and dragforces were measuredat were madesuch that the greatestbreadth of eachwas six anglesof attack:0, 5, 10, 15, 20, and 25 degrees. 30 mm. Thus,the Reynoldsnumber of all modelswas At eachangle of attack,2,000 samples were measured the same, viz. Re = 1.5 x l0 s. This is somewhat lower over 10 s and averaged.To reduce error, five mea- thanthe Reynolds numbers for the real wings (White- surements were made before water flow was started, tailedPtarmigan, Re = 12.5 x lOS;Sage Grouse, Re = five during the stableflow, and five after the flow 19.5 x lOS;Wild Turkey,Re = 33.5 x lOS;California was stopped.The mean of the 10 electricpotential Quail, Re = 9.0 x l0 s)at a presumedflight speedof measurements (five measurements each before the 15 m/s, a value availablefor the Ring-neckedPheas- flow was startedand after it was stopped)was con- ant (Phasianuscolchicus; Rayner 1985). Despite the dif- sidered to be the "0 force" condition. This mean value ferencesbetween these figures, the modelsare still was subtracted from the mean of the five measure- useful for the experimentsbecause Re = 1.5 x 1OSis ments that were taken when flow was stable, and the within theperformance range of realwings. Reynolds difference between these two means was then con- number increasesproportionally with flight speed vetted to a force.By measuringthe "0 force"condi- betweentakeoff and fastforward flight. Also, as long tionbefore and after measuring the lift anddrag forc- as geometricsimilarity is maintained,the drag coef- es, errors in measurements were reduced. Measure- October1996] FlightPerformance andthe Wing Notch 805

mentswere first made on wing modelswithout notch- 48- es. Then, notches were cut in each model, and the measurementcycles were repeated.The areaof each NS = 74.33 - 25.29 (WL / WB) cut piecewas calculated by weighing it (+ 0.001mg) and using foil thicknessand specificgravity of the metal. I used differencesin pectoral muscle color as an index of differencesin power output and resistance to fatigue among galliforms.White muscleshave higher poweroutput but are lessresistant to fatigue. Galliform muscles that are white or intermediate in 24' WildTurkey-• _ color are composedpredominately of glycogen-load- 20 ed white fibers(Norberg 1989).For example,glyco- gen-loadedwhite fibersconstitute > 80%of all muscle fibersin the pectoralmuscles of Ring-neckedPheas- 16 -etailedPtarmigan e•e use antsand SpruceGrouse (Dendragapus canadensis). This 12 ß •.4 1:6 •:s 2:0 2:2 2:4 similaritysuggests that the proportionof red and in- termediatefibers varies among galliformsby about Wing length-to-breadthratio only 20%.Because galliforms are thought to be mono- F]c. 2. Relationshipbetween notch size and wing phyletic, all specieswithin the order are likely to length-to-breadthratio for 31 galliforms.Circles are share similar types of red and intermediatemuscle tetraonidswith dark pectoralmuscles, diamonds are fibers.When musclefiber typesare the same,differ- tetraonidswith light pectoralmuscles, and squares encesin musclecolor might correlatewith differences are other galliforms. in the proportionsof thesefiber typesand, thus,dif- ferences in muscle function that are related to the ecologyof the respecitvespecies. that undergoeslong, strait flights. The Hima- While preparing museum specimensI recorded layan Snowcockis characterizedby downhill pectoralmuscle color according to a visualscale. This gliding insteadof activeflapping flight. Despite scalehad three grades:dark or red (e.g. Lagopus),in- theseexceptions, I found a strong negativecor- termediate(e.g. Dendragapus), and light or white (e.g. relation between notch size and the ratio of Bonasa). wing length to wing breadth (r = -0.76, n = 31, P < 0.001; Fig. 2). Galliforms with short, RESULTS wide wings have deeper notches. DIFFERENCES AMONG GALLIFORMS IN WING The differencesin wing shapeand notch size GEOMETRY AND NOTCH SIZE between grouse and other galliforms could be related to differencesin the frequency and du- Wing narrowness (length-to-breadth ratio) ration of flights performed by the two groups. and notch size vary widely among the 31 spe- Flight is a critical element of foraging behavior cies studied (Table 1). Ptarmigan and grouse in grouseand ptarmigan, especiallyin winter, (Tetraoninae)have narrower wings and smaller when they feed on buds and foliage in trees. notchesthan other galliforms.The rangeof wing To reach these foods, forest grousemust make length-to-breadth ratios for grouse and ptar- verticalflights to limbsand frequentshort flights migan is 1.88to 2.35,compared with 1.47to 1.91 from limb to limb, and ptarmigan must fly from in all other speciesexcept Common Quail (Co- one foraging site to another. Other galliforms turnixcoturnix), which is migratory and has long, fly lessoften, usingflight primarily to flee from narrow wings (length-to-breadthratio of 2.25). predators. Notch size varies between 15.17% and 28.94% of wing breadth in the tetraoninae and between MORPHOLOGICAL FLIGHT PARAMETF, RS OF THE 24.98%and 47.34%in most other galliforms ex- MODEL BIRDS cept three species:Small Buttonquail (Turnix sylvatica), Common Quail, and Himalayan Several morphological measurementsare Snowcock(Tetraogallus himalayensis; Table 1).The usefulin comparativeanalyses of flight perfor- systematicposition of buttonquailsis not clear. mance in birds (Rayner 1988, Norberg 1989). Most authorstreat them as galliforms,but some Among the most important are wing loading place them with Gruiformes and others in the and aspectratio. Wing loading (in N/m 2) is de- Turniciformes.The Common Quail is a migrant fined as: 806 SERGEIV. DROVeSKI [Auk,Vol. 113

TABLE2. Wing loading, aspect ratio, and trailing- 125- edge notch size for the specimensused to make the model wings. 115- ßSageGrou•

Notch z size m•'105- • Wingloading=91.4 M0.201 Wing (% of loading Aspect wing • 85 / WildTunkey, Species (N/m 2) ratio area) Lagopusleucurus 74.12 9.36 1.57 8575 //y/White-tailed Ptarmigan Centrocercusurophasianus 121.52 5.79 1.64 /. CaliforniaQuail 55 Meleagrisgallopavo 87.19 4.23 4.11 o.o ,.0 2:0 d.5 3.0 ;.0 Callipeplacalifornica 59.43 3.67 2.16 Bodymass (kg) FIG.3. Plot of wing loading vs. body massfor the wing loading = Mg,/S, (5) four model speciesrelative to expectedrelationship (curve)based on an allometricequation calculated for where M is body massin kg, g, is acceleration galliforms by J. M. V. Raynet (pets. comm.). due to gravity (i.e. 9.81 m/s2), and S is the total wing area (the planar area of both wings and that part of body between the leading and trail- the first type are red, whereasfibers of the sec- ing edges of the wings in m2). Wing loading ond type are white. These types of fibers have determinesimportant flight characteristicssuch different biochemistry that results in differ- asspeed and maneuverability.Birds with heavy ences in contraction frequency and speed as wing loading fly at high speedand have a large well as resistanceto fatigue. Red fibers have turning radius. Aspect ratio is defined as: lower frequencyand speedof contraction,but much higher resistanceto fatigue than white AR = b:/S, (6) fibers (Norberg 1989). Therefore, muscleswith where b is wingspan(in m), and S is total wing a high proportion of red fibers are suited for area (in m2). Aspect ratio determines cost of sustained(horizontal) flight, but white muscles transport, or the ratio of power to speed.Birds are adaptedfor short, powerful bursts(take off, with a high aspectratio useless energy per unit verticalflight). Wild Turkey and CaliforniaQuail of distance flown (i.e. less power) to generate have much lighter coloredpectoral muscles than the samehorizontal speedcompared with birds do White-tailed Ptarmigan and Sage Grouse. with low aspectratio (Raynet 1988). Grouseand ptarmigan with red pectoralmus- The White-tailed Ptarmigan has the highest cles (i.e. Lagopus,Tetrao, Tympanuchus, and Cen- aspectratio of the four modelsand a fairly heavy trocercus)have smaller trailing-edge notches wing loading (Table 2). The aspectratio for Sage than thosewith white or intermediate pectoral Grouseis lower, but it still is much higher than muscles(i.e. Dendragapusand Bonasa;œ = 19.46 that of the Wild Turkey and Common Quail. + 2.01%, n = 9 vs. • = 25.04 + 2.79%, n = 5, Wing loading of the SageGrouse is exception- respectively;t = 4.32, df = 13, P < 0.001) and ally high, whereasthe California Quail hasthe wings with larger length-to-breadthratios (œ= lightestwing loading. Wild Turkeyshave some- 2.15 + 0.10 vs. œ = 1.92 + 0.03; t = 4.76, df = what greater wing loading than White-tailed 13, P < 0.0005). Ptarmigan.However, when size is taken into The four model speciesform two pairs, each account,wing loading changesproportionally with different wing characteristicsand pectoral to M ø':ø•in galliforms (J. M. V. Raynet pets. musclecolors. White-tailed Ptarmiganand Sage comm.).Thus, Wild Turkeysactually have much Grouse have high wing loading, high aspect lower wing loadingthan expectedby allometry ratio, and red pectoralmuscles, which servetheir (Fig. 3). long, straightflights at high speeds.In contrast, Another important morphologicalparameter Wild Turkey and CaliforniaQuail have low wing that can be usedfor comparativestudy of close- loading, low aspect ratio, and white pectoral ly related speciesis the color of the pectoral muscles that are better suited for slower, more muscles.In galliforms it may indicate relative powerful and maneuverableflights. Their short, abundanceof fast-twitchoxidative-glycolytic fi- broad wings are better suited to frequent short bers and fast-twitch glycolytic fibers. Fibers of (and vertical)flights becausethey generatemore October1996] FlightPerformance andthe Wing Notch 807

T^BLœ3. Maximumlift coefficientsof the modelwings, maximum lift-to-drag ratios (L/D), anglesof attack at which they occur,and increasein lift-to-dragratios with wing notching.

Max. lift coefficient Max. L/D Angle of attack' Species No notch Notched No notch Notched No notchNotched increase b Lagopusleucurus 0.69 0.61 3.96 5.98 5 5 33.8 Centrocercusurophasianus 0.64 0.56 2.40 2.61 10 10 8.0 Meleagrisgallopavo 0.80 0.76 4.57 5.23 5 10 12.6 Callipeplacalifornica 0.70 0.65 4.81 5.12 5 10 6.1 At maximum lift-to-drag ratio (L/D). Increasein maximumL/D with wing notching. power on the upstroke, facilitating takeoff, were the same as those for lift measurements. landing, and steepclimbs at low forward speed. In addition,notching decreased the CV for mea- The wing notch sizes of White-tailed Ptar- surementsof drag to a different extent in dif- migan and SageGrouse are similar (Table 2). ferent species(White-tailed Ptarmigan,70.2%; This suggeststhat such a structureis lessim- SageGrouse, 25.1%; California Quail, 21.3%; and portantfor galliformbirds with relativelyhigh Wild Turkey, 0.1%). aspectratios and heavy wing loading. In con- The inverse relationshipsbetween angle of trast, the wing notch sizesof California Quail attack and the CV for lift and drag measure- and Wild Turkeys are much larger. These com- mentsmay be relatedto the meansof the forces parisonssuggest that birdswith short,rounded increasingwith an increasein angle of attack. wingsshould derive more benefit from the wing notch than do longer-winged species. 1.0- 1,0 ÷ While-tailed ptarmigan Sage Grouse

0.8 LIFT-TO-DRAG COEFFICIENT RATIOS IN WINGS nn WITH AND WITHOUT NOTCHES 0.8- The trailing-edge notch increasedthe maxi- 0.4- 0.4 mal lift-to-drag ratio in all the model wings 0.2- 0.2 (Fig. 4). However, this increasewas small in 0.0-0.8- • 0.0 SageGrouse, Wild Turkey, and CaliforniaQuail but reasonablylarge in White-tailed Ptarmigan -0.2 - -0.2 ß 0.0 0.2 0.4 0.6 ø'80.0 ••0.2 0.4 2n 0.6 (Table 3). The angle of attack of maximal lift- to-drag ratio did not change in White-tailed Drag coefficient Drag coefficienl Ptarmigan and SageGrouse, but it increasedin Wild Turkey and California Quail after notches 1.0- 1.0 were createdexperimentally. California Quail Wild Turkey All wing models also had smaller standard 0.8- deviationsof lift and drag coefficientsafter the trailing-edge notcheswere created (Fig. 4). I 0.6- • 0.6 testedfor the effectof the notch, angle of attack, 0.4- • .4 and species,and for the interactionof species 0.2- ..3 0.2 and notch, on coefficients of variation (CV) of the lift and drag measurements(Table 4). Co- 0.0- 0.80.0 •n efficients of variation for measurements of lift -0.2- -0.2 were significantlyaffected by only two factors: 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 (1) increasingthe angle of attack from 0ø to 25ø Drag coefficient Drag coefficient progressivelyreduced the CV, and (2) notches Fla.4. Plot of lift vs.drag coefficients for the mod- reducedthe CV by 53.4%regardless of species. el wings. Eachcurve is basedon independentmea- Coefficients of variation for measurements of surements(indicated by standarddeviations) made at drag were significantlyaffected by all four fac- six anglesof attack.From left to right: 0ø, 5', 10ø, 15 ø, tors (Table 4). The relationships of CV to the 20%and 25ø;nn is model with no notch, wn is model angle of attack and the presenceof the notch with notch. 808 SERGE•V. DROVErrSK/ [Auk, Vol. 113

TABLE 4. Effectof wing notching and angle of attackon the CV of lift and drag measurementsof model wings.

Notching Angle of attack Species Notchingand species F P F P F P F P

Lift 7.77 0.008 30.43 <0.001 2.81 ns 0.23 ns Drag 20.48 0.001 34.09 <0.001 3.91 0.016 8.94 <0.001

Because the CV is a ratio of standard deviation during take off. Therefore, notches may help to the mean, increasing the means will lower such birds to ascend faster. the CV if the SD doesnot change.Drag increas- es through the whole range of anglesof attack, DISCUSSION and lift increasesthrough most of the range, and then levels off or slightly decreases.Also, My comparativedata show that galliforms the rate of changein lift and drag is not constant display a wide variety of trailing-edgenotch through the range of angles of attack (Fig. 1). sizesand wing shapes(Fig. 2). In general, gal- For example, a minor error in positioning the liforms of all groups except Tetraoninae have model wing for a small angle of attack (closeto the shortest wings and deepest notches found 0ø) will have a large effecton lift measurements in the order. Bonasaand Dendragapushave in- and a small effecton drag measurements.When termediate notch sizes and intermediate wing the angle of attack is closeto 25ø , the opposite length-to-breadthratios. Other grouse(Lagopus, is true. Tetrao,Tympanuchus and Centrocercus)and the Despite other factors, wing notches signifi- migratory Common Quail have the longest cantly reduced the CV of lift and drag mea- wings and smallestnotches found in the order, surements.This suggeststhat wings with trail- and all, except the quail, have dark pectoral ing-edgenotches should be morestable in flight, muscles,whereas the other 22 specieshave light and that their performanceshould be more pre- or intermediate colored pectoral muscles. dictable.Trailing-edge notchesalso reduced the Although the flight performancedata were maximum lift coefficients (Table 3). The maxi- collectedfor fixed wings in steadyflow, differ- mum lift coefficients of models without notches encesin lift and drag characteristicsshould ap- were inversely related to the amount of their ply to fast forward flight with reduced fre- reduction (r = -0.84, n = 4, P = 0.16). This quencies(see Norberg 1989) less than 0.5 Hz. indicatesthat long wings that produce lesslift The reducedfrequency parameter (2a-f L / U) for (e.g. SageGrouse and White-tailed Ptarmigan) these birds in forward flight is about 0.65 (cal- loseproportionately more lift with the addition culated for Ring-neckedPheasant, f = 9 Hz; of a notch than do short,round, high-lift wings Greenewalt 1962:table 13). Despitea rough ap- (e.g. turkey and quail). However, they alsolose proximation (the pheasant has an intermediate proportionatelymore drag, so the lift-to-drag body mass,so its speedis probablyhigher than ratio of long wings increasesmore than that of the speed of California Quail and White-tailed short wings. Therefore, trailing-edge notches Ptarmigan but lower than that of Wild Turkey make takeoff easier by improving the lift-to- and SageGrouse; the oppositeis true for wing- drag ratio, but they also increasethe energetic beat frequency), the calculated reduced fre- cost of forward flight by decreasinglift, thus quency parameter is only slightly higher than forcing birds to fly faster. the generally acceptedcutoff. Thus, the results The angle of attack at maximum lift-to-drag of the experimentsstill are likely to apply to ratio is probably stable regardlessof the pres- the flight of model species.However, in take enceor absenceof a notch in long-winged birds off or landing, somecaution should be usedin (Sage Grouse, White-tailed Ptarmigan), but it interpreting differencesin steady state aero- increasesin short-winged birds (Wild Turkey, dynamic coefficients. California Quail). The increasein angle of at- The differencesin wing geometryand notch tackin birds with short,rounded wings should size are related to the ecologyand behavior of increaselift (at the expenseof some increasein differentspecies. Except for the grouseand ptar- drag) and, thus, increase vertical acceleration migan, which use flight between bouts of for- October1996] FlightPerformance andthe Wing Notch 809 aging, most galliforms use flight primarily to other branch in the same tree, or to a neigh- flee from predators.Thus, becausethe main el- boring tree in the same canopy. During one ement of their flight is vertical (takeoff and as- foraging bout a Hazel Grouse visits 6-12 cending), level flight is not so important. In branches.Flights within foragingbouts are short contrast,flight is an important part of the for- and slow, and a flock does not move more than aging behavior of grouse and ptarmigan, and 300 m during a single bout of foraging. Thus, the relative importance of level flight varies the majority of the energy Hazel Grousespend dramatically in specieswith dark versus light in foraging flights is investedin vertical move- (or intermediate)pectoral muscles. Grouse with mentsand in the frequent takeoffsand landings a higher proportion of red fibers in the pector- required to move from branch to branch. Fur- alis majorfly relatively long distancesfrom one thermore, Hazel Grouse land and feed on such foraging site to another, but then they either small branchesthat they can not jump from walk on the ground or perch on branches to them with their feet to aid in takeoff. This places feed. Thesegrouse use level flight primarily to even greater demands on their wings for ac- reachfood patches.However, grousewith light celeration.For theseshort burstsof flight, white or intermediate pectoral muscle fibers do not muscle fibers; short, broad wings; and deep fly long distances.Rather, they usevertical flight trailing-edge notchesare the best compromise. primarily from ground to canopy,or slow flight I usedmodel wings of two pairsof galliforms from branch to branch, to reach their food. The with different flight characteristicsin my ex- grouse with white and intermediate fibers and periments. White-tailed Ptarmigan and Sage short, broad wings have deep trailing-edge Grouse represent specieswith small notches notchesthat increaseboth their lift-to-drag ra- (Fig. 2). They have high aspectratios, heavy tio and the angle of attack at which their wings wing loading for their size, low maximum lift are more efficient. coefficients,and dark pectoral muscles(Tables For example, the Willow Ptarmigan (Lagopus 1-3). Theseare all adaptationsfor relativelylong, lagopus),which has red pectoral muscles,feeds fast, straightand efficient flight (Rayner 1988, on willows (Salix schwerinii,S. rorida) and cho- Norberg 1989).In contrast,the Wild Turkey and senia (Choseniaarbutifolia) twigs and buds dur- California Quail, which represent specieswith ing winter in northeastern Russia (Andreev deep notches(Fig. 2), have low aspectratios, 1980, Drovetski 1992b). The willow and cho- relatively light wing loading, high lift coeffi- senia thickets where ptarmigan forage are sit- cients,and light pectoralmuscles (Tables 1-3). uated on river islands and along the banks of Thesefeatures are adaptationsto powerful short braided river channels. During foraging birds flights and maneuverability(Rayner 1988,Nor- walk along patchesof thickets and pick twigs berg 1989). and buds from bushesand snow. After passing Short, wide wings can generateas much lift through a patch of willows, a flock of ptarmigan aslong, narrow wings of the samearea only by must fly to reach another willow patch, which increasingthe angle of attack (Vogel 1983). usually will be situated on another river island However, higher anglesof attack generatein- or along another channel. Thus, Willow Ptar- creasesin both profile and induced drag. My migan perform several flights of several hun- results show that the trailing-edge notchesin dred metersa day. They fly through open places galliform wings increasethe maximumlift-to- at a fast pace and at low altitude. Becausethese drag ratio by improving air flow around the flights are relatively long, the most expensive wing. Notches had a similar effect on all four componentis not takeoff and landing, but rath- species,i.e. increasingthe maximumlift-to-drag er level movement. ratio and stabilizingair flow around the wing. In the same region, the Hazel Grouse (Bonasa However, the trailing-edge notch slightly re- bonasia),which haswhite pectoralmuscles, feeds ducedthe maximum lift coefficient(Fig. 2). This in trees but in denser, more continuous forests reduction in maximum lift is presumablymore (Andreev 1980; Drovetski 1992a, 1992b). Unlike detrimental to speciescharacterized by rela- Willow Ptarmigan,Hazel Grouseforage mostly tively long flights than to speciesthat spenda by perching on branchesin the canopy, and high percentageof their flight time in landings extremely rarely by walking to branches that and takeoffs, times when the angle of attack canbe reachedfrom the snow. Eachbird spends must be great. 3 to 8 rain on one branch and then flies to an- Theexperimental data show a tradeoffin flight 810 s•a•i V. DRoversKI [Auk,VoL 113 performance of galliforms associatedwith the with taking picturesof modelwings. I am very grate- trailing-edge notch. The notch reduces maxi- ful to JeremyRayner, Carl Heine, Kenneth Dial, and mum lift, but increasesthe lift-to-drag ratio and an anonymousreviewer for very helpful comments stabilizesairflow around the wing. This means on the manuscript.This study was sponsoredby Mr. that an increasein the ability to take flight and GarretEddy. Thanks also to the BurkeMuseum's Eddy Endowmentfor Excellencein Ornithology for sup- climb fast,and in the ability to maintain height port during theseinvestigations and to the University during a slow flight, may be achieved at a cost of WashingtonBurke Museum and the University of of reduced efficiencyin level flight. Galliforms Puget Sound Slater Museum of Natural History for that experiencerelatively long and fast level useof specimens.The comparativepart of this study flights shouldbenefit more from a high-lift (i.e. could not have been done without the Burke Mu- high-efficiency)wing. In suchspecies the trail- seum'scollections of extendedwings from Asia,North ing-edge notch should be smaller. In short America, and Africa. flights,when verticalmovements become a large componentof eachflight, increasingthe lift-to- LrrERXTUP, E CITED drag ratio becomesmore advantageous.In this circumstanceselection should favor larger trail- ANI)REEV,A. V. 1980. Adaptatsiyaptits k zimnim ing-edge notches. usloviyam subarktiki. "Nauka," Moscow. Comparedwith other volant birds,galliforms DROVETSKI,S. V. 1992a. Materialy po ecologii Ry- spend small amounts of time in the air, rarely abchika (Tetrastesbonasia L.) na Uge Magadan- fly long distances,and have pronounced trail- skoy oblasti v zimniy period. Zoologicheski Zhurnal 71:45-59. ing-edge notches. The comparative analysis within the order supportsthe resultsof exper- DROV'a'TSKI,S.V. 1992b. Novye dannyeo funktsion- al'nomznachenii rogovoy bakhromy na pal'tsakh iments and suggeststhat wing geometry and teterevinykhptits. Zoologicheski Zhurnal 71:100- size of the notch result from of a compromise 109. between what is optimal for level flight and GREENEWALT,C.H. 1962. Dimensionalrelationships what is optimal for frequent takeoffand vertical for flying animals. Smithsonian Miscellaneous flight. Grouse that spend much time in level Collections 144:1-46. flight have small trailing-edge notches,where- HOFTON, A. 1978. How sails can save fuel in the air. as those whose flight is characterizedby fre- New Scientist 78:146-147. quent takeoffs and landings, and by vertical L•Yolq, L.E. 1976. The measurement of bone strain movement, have deep trailing-edge notches. in vivo. Acta Orthopaedica Belgica 42 (Supple- ment 1):98-108. One may ask why ptarmigan and other long- NORI•ERC,U. M. 1989. Vertebrate flight. Zoophy- winged galliformshave any trailing-edgenotch. siology.Springer-Verlag, Berlin. The answer, I think, lies in the fact that even RAYNER,J. M.V. 1985. Speedof flight. Pages225- thosegrouse with relativelylong, narrowwings 226 in A dictionary of birds (B. Campbell and E. fly very little comparedwith other birds. Thus, Lack, Eds.).T. and A.D. Poyser,London. relative to other species,long-winged galli- RAYIqER,J. M.V. 1988. Form and function in avian forms spend a considerable amount of their flight. Current Ornithology 5:1-66. flight time in takeoff, where the notch is help- SHEST.a,KOV.% G.S. 1971. Stroeniekryl'ev i mekhan- ful. ika polyota ptits. "Nauka," Moscow. Whereas the trailing-edge notch improves SHTEC;MAN,B.K. 1953. Osobennostilyotnykh ka- chestv seroy i kamennoy kuropatok.Zoologi- maximum lift-to-drag ratio of the extended cheski Zhurnal 37:677-683. wing, the notch could be even more important TUCKER,V.A. 1993. Gliding birds:Reduction of in- for a powered upstroke. The possibility that it duceddrag by wing tip slotsbetween the primary createsa separatewing should be examinedby feathers.Journal of ExperimentalBiology 180: photography,and, if confirmed,be exploredby 285-310. further work in wind tunnels or flow tanks. TUCKER,V.A. 1994. Drag reductionby wing tip slots in a gliding Harris' Hawk, Parabuteounicictus. ACKNOWLEDGMENTS Journalof ExperimentalBiology 198:775-781. VINOGRADOv,I. N. 1951. Aerodinamikaptits pari- I thank Sievert Rohwer, Thomas Daniel, and Chris teley. DOSARM, Moscow. Wood for their help with this study, and Dmitriy VotEr, S. 1983. Life in moving fluids. Princeton Banin, Richard Droker, and Carol Spaw for the help University Press,Princeton, New Jersey. Associate Editor: K. P. Dial