The Auk 114(2):249-262, 1997

MOLT AND MIGRATION IN THE NORTHERN ROUGH-WINGED

TAMAKI YURI1 AND SIEVERTROHWER BurkeMuseum and Department of Zoology,Box 353010, University of Washington,Seattle, Washington 98195, USA

ABSTRACT.--Wedescribe the prebasicmolt of Northern Rough-wingedSwallows (Stelgidop- teryxserripennis) using museumspecimens. from easternNorth America initiate their flight-feathermolt about40 daysearlier than birds from west of the RockyMountains, sug- gestingthat ecologicaldifferences between the two populationsaffect their molt schedules. In both populations,juveniles start their flight-feathermolt 1 to 4 weekslater than adults,but the timing of body molt doesnot differ for adultsand juveniles.Molt occurssimultaneously with the fall migrationin both populations.However, eastern birds interrupt their migration when they reachthe northerncoast of the Gulf of Mexico.Here, they spendabout two months finishingtheir flight-feather molt beforecrossing the Gulf in Octoberand November.Molt and migrationoften are concurrentin diurnallymigrating , but the two activitiesdo not occursimultaneously when migration must be sustainedwithout interruption. Received 15 May 1996,accepted 6 December1996.

MOLT IS AN ESSENTIAL COMPONENT in the an- gration, and is completedon the wintering nual cycleof birdsbecause worn feathersdetri- grounds(Niles 1972).Finally, some mentallyaffect survival and reproduction(Jenni from western North America and some trans- and Winkler 1994).Because the timing of molt Saharanmigrants from Europebegin fall migra- rarelyoverlaps with otherdemanding activities, tion prior to molting,but interruptmigration to suchas reproductionand migration,it is often molt before continuingon to their wintering consideredenergetically "expensive." However, grounds (Rohwer and Manning 1990, Young effortsto measurethe energeticcosts of molt 1991,Jenni and Winkler 1994).Common to these vary dependingon studymethods and the spe- strategiesis thetemporal separation of moltand ciesstudied (Perekand Sulman 1945,Wallgren migration. 1954, Lustick 1970, Gavrilov 1974, Gavrilov and Molt and migrationoccur simultaneously in Dolnik 1974, Chilgren 1975, Wijnandts 1984, many speciesof swallows (Niles 1972, Cramp Dietz et al. 1992,Lindstr6m et al. 1993).Murphy 1988, Jenni and Winkler 1994). However, be- (1996)argues that the protein and energyre- causefew North Americanspecies have been quirementsof molt do not appear to be great studied in detail (Niles 1972, Stutchburyand enough to pose significantnutritional chal- Rohwer 1990), it is difficult to make compari- lengesand suggeststhat the schedulingof molt sonsamong swallows. Here, we reportthe tim- in the annualcycle is a selectivecompromise as- ing of prebasicmolt in the Northern Rough- sociated with various non-nutritional demands. wingedSwallow ( serripennis) and The timingof molt relativeto migrationvar- the extentto which this molt overlapswith the ies amongspecies. This variationmay be asso- fall migration.We alsodocument differences in ciatedwith the durationand physiologicalcosts the schedulingof molt and migrationin North- of moltingand migrating,as well aswith social ern Rough-wingedSwallows breeding in east- and ecologicalrequirements. In most north- ern and western North America. Differences in temperatepasserines, a completepostbreeding the timing of molt and migrationin thesetwo molt of body and flight feathersoccurs on the populations elucidate factors that affect the breedinggrounds just prior to fall migration,or schedulingof molt in the annualcycle. on the winteringgrounds following migration (Pyleet al. 1987).Less often, this molt beginson METHODS the breedinggrounds, is suspendedduring mi- We examined477 museumspecimens of Northern Rough-wingedSwallows collected from throughout • Presentaddress: Museum of Zoologyand Depart- theirwinter and summerranges. Specimens were ex- mentof Biology,University of Michigan,Ann Arbor, aminedfrom all timesof theyear except late April and Michigan48109, USA. E-mail:[email protected] May, with an emphasison specimenstaken from late 249 250 YURI AND ROHWER [Auk, Vol. 114

TABLE1. Seasonaldistribution of all specimensexamined.

Easternpopulation Westernpopulation Adult Juvenile Age ? Adult Juvenile Age ? Date M a F ? M F ? M F ? M F ? M F ? M F ?

1-15 June 6 1 1 15 13 2 1 16-30 June 3 I I I 8 5 3 3 1-15 July 7 7 6 2 1 16 4 4 2 16-31 July 4 3 6 3 4 13 4 1 5 9 3 1-15 August 3 2 6 4 1 3 2 13 11 1 16-31August 1 2 6 3 1 6 1 23 12 2 1-15 September 4 3 2 1 1 1 7 4 2 16-30 September 1 1 1 3 1 1 2 1 1-15 October 3 3 1 16-31 October 1 1 1 4 1 1 1-15 November 1 2 2 1 2 2 16-30 November 1 1-15 December 5 16-31 December 3 1-15 January 2 3 1 2 1 16-31 January 4 5 1 1 1-15 February 2 I 1 5 1 16-28February 2 2 5 1 1-15 March 6 6 7 1 16-31 March 4 1 21 7 1-15 April 10 5 17 10 a M = male; F = female; ? = sex unknown. summerthrough early winter (Table1). Our sample northern Mexico, the United States, and Canada are consisted of 60.2% males, 35.6% females, and 4.2% of migratory(Miller 1957).We excludedspecimens col- unknown sex.Juveniles were identifiedby the cinna- lectedsouth of theMexican states of Sinaloa,Durango, mon color on their upperpartsand on the edgesof and Coahuila from June through August (most of their wing covertsand inner secondaries(Pyle et al. which would have been residents) but included those 1987).After the firstprebasic molt, first-yearbirds be- collectedafter August.We have assumedthat includ- come indistinguishablefrom adults; these birds, ing specimensfrom the residentraces would be less mostlytaken in late winter, were of unknown age.In harmful to our resultsthan excludingmigrants from our specimens,31% were adults, 35% were juveniles, North America, most of which should have arrived in and34% were of unknownage. Birds taken in or west theseareas by the end of August. of Alberta,Montana, Wyoming, Colorado, and New Specimenswere examined using a magnifyinglamp Mexico, or west of Tabascoand Chiapasin Mexico, lighted with an incandescentbulb. The molt of flight were defined as the westernpopulation; birds taken feathers was scored as described in Newton (1966). elsewhere in the United States or Canada were defined Each of the 18 primaries,18 secondaries,and 12 rec- as the easternpopulation. triceswas scoredas: 0 (no molt), 1 (missingor small The two recognizedspecies of Rough-wingedSwal- pin), 2 (large pin or brush), 3 (brush to one-half lows (Stelgidopteryx)breed from southernCanada to grown), 4 (one-halfto three-quartersgrown), or 5 Argentina (Stiles1981) and are divided into numer- (three-quartersto fully grown).Birds with a complete ousraces. Southern Rough-winged Swallows (S. ruff- setof new flight feathersreceived a scoreof 240. collis)are residentfrom easternHonduras to Argen- In summarizingmolt progression,we comparedthe tina, and are distinguishedfrom Northern Rough- intensityof molt amongmolt seriesfor specimensin winged Swallowsby their blackercrowns, brighter which the outermostgrowing primary was the same. throats, paler rumps, yellowish bellies, and boldly The intensityof flight-feathermolt was assessedboth black-tippedundertail coverts. Southern races of the asthe averagenumber of the feathersgrowing simul- Northern Rough-wingedSwallow that breed from taneouslyand as the averageamount of missingfeath- Mexicoto CostaRica (e.g.fulvipennis, stuarti, and ridg- ers in linear millimeters. We converted the raw molt wayi)are practicallyindistinguishable from popula- scoresto fractionsof missingfeather as follows:1 = tionsnorth of Mexico.These races vary from resident 100%missing, 2 = 87.5%,3 = 62.5%,4 = 37.5%,and to partlymigratory and may havemolt schedulesthat 5 = 12.5%. The amount of each feather that was miss- differ from racesto the north. All populationsof ing was calculatedby multi-plyingthe fractionmiss- Northern Rough-winged Swallows that breed in ing by the feather's aver-age length. The average April1997] Moltand Migration in Swallows 251 length of eachof the primaries,secondaries, and rec- eru population(Table 1). Therefore,in addition to sta- triceswas obtainedby measuringfeathers from speci- tisticalanalyses, we relied extensivelyon graphical mens(four of eachsex) that containeda completeset summarizations of our results. of fully grownflight feathers. Body molt was scoredfor six non-overlappingre- REPLACEMENT RULES gions:crown, back, rump, chin, breast,and belly (see Rohwer 1986).Four to eight featherswere lifted with A molt seriesis a set of remigesor rectricesthat is forcepsat five to eight pointsin eachregion to check replacedaccording to a singleset of rules (Langston for pin feathersor for growingfeathers with sheaths. and Rohwer 1995). Because the direction of feather re- Then, eachregion was scoredon an ordinal scaleof 0 placementmay not reliablyidentify a molt series(e.g. to 4 for the proportionof activelymolting feathers, inner secondaryseries of passerines;Jenni and Win- where 0 = none, 1 = less than 25%, 2 = 25-50%, 3 = kler 1994),attention to the temporal sequenceof re- 50-75%,and 4 = morethan 75%.The maximumbody placementis essentialin defining a molt series.We molt scorefor the total of the six regionswas 24. treated primaries and secondariestogether because We analyzedfrequency data usingthe G-testwith the inner primariesand outer secondariescan be part Williams' correction (Sokal and Rohlf 1995). Collection of a single molt series(e.g. Langstonand Rohwer dateswere divided into half-monthblocks (e.g. early 1995).Following Langston and Rohwer(1995), we cat- and lateSeptember). For body molt data,we usedlog- egorizedeach growing primary, secondary, and rectrix linear models (Sokal and Rohlf 1995) to test the inde- accordingto the directionof molt, and whetherit was pendenceof four factors(population, age, sex, and the featherthat startsor stopsa molt series.This is body region)in a three-wayinteraction with time and doneby placingeach growing (or focal)feather into frequencyof molting birds. Also, using G-testswith oneof the four categoriesdiscussed below. log-linear models, we tested the independenceof Proximalto distal replacement.--This occurs when the flight-feathermolt stage(i.e. no molt, first half of molt, focalfeather is lessadvanced in growth than the next and second half of molt) in an interaction with lati- proximalfeather, and more advancedthan the next tude (dividedinto 5ø blocks)and frequencyof speci- distalfeather. For P9 and S9,comparisons can be made mens. onlywith a singleadjacent feather. To avoid inferences We used linear regressionto estimatethe mean aboutdirection based on incompleteinformation, and startingand completiondates, and the mean duration to eliminatedouble tallying, such feathers are classi- of flight-feathermolt as describedin Pimm (1976). fied either as nodal or terminal (see below). For ex- Thismethod treats molt scoreas the explanatoryvari- ample,in many passerinesP9 can only be compared able and date as the dependentvariable so that mean with P8 (because P10 is not functional and much re- startingand completiondates of molt are estimated ducedin size),so it is assignedas the terminal feather for individualsrather than populations.Although this if it is shorter than P8. method has been criticizedfor violating assumptions Distalto proximalreplacement.--This occurs when the of the linear regressionmodel, it is computationally focalfeather is more advancedin its growth than the simple and used more often than the likelihood nextproximal feather and lessadvanced than the next methoddeveloped by Underhilland Zucchini(1988). distalfeather. Again, at thebeginning or end of a molt Moreover, we were not able to convert molt scores to series,where comparisonscan be made only with a percentagefeather mass grown as describedin Under- singleadjacent feather, feathers are classifiedeither as hill and Zucchini (1988) because masses for individual nodal or terminal (see below). flight feathersof Northern Rough-wingedSwallows Nodeof initiation.--Nodalfeathers mark siteswhere were not available. molt is initiatedin the presentepisode of molt. They To test for differencesin flight-feathermolt sched- are typicallynewer or more advancedin their growth ules between populations,and among age and sex thaneither of the adjacentfeathers (e.g. P1). Nodes of- groupswithin populations,we performedthree-way ten markthe beginningof a molt series,but theyfail analysesof covariance(ANCOVA). Regressionanaly- to do so when molt within a single seriesis inter- sesand ANCOVA were performedusing Minitab 8.21 ruptedin one episodeof moltingand resumedin the (MinitabInc. 1991).In assessingthe timing of molt,we next (seeLangston and Rohwer1995). When a focal included only birds replacingflight feathersand ex- featherhas but oneadjacent feather (e.g. P9 in many cluded birds for which locality,age, or sex were un- passerines),it is nodalif it is lostbefore its neighbor. known.Significance level was set at P < 0.05for all Amongpasserines, this condition should only existin tests. Muscicapastriata, which replacesits primaries from Specimenstaken by various collectorsover long outer to inner (Stresemann1963). When two adjacent time periodsprobably do not representa random or feathersare at the samestage of development,both systematicpopulation sample. Collecting effort may shouldbe tallied as nodesof initiation,even though be biasedby sexor age class,time of year,plumage, onemay not mark the beginningof a molt series.For molt stage,and geographiclocation. Indeed, we found example,P1 and P2 sometimesare lost nearly simul- few specimensin later stagesof molt from the west- taneously in passerines,thus causing both to be 252 YURIAND ROHWER [Auk, Vol. 114

TABLE2. Frequencyof remigesin activemolt. Remigesare identifiedas directional,nodal, and terminal,us- ing replacementrules described in Methods.

Focal feather

S9 S8 S7 S6 S5 S4 S3 S2 S1 P1 P2 P3 P4 P5 P6 P7 P8 P9

Proximal-to-distal replacement 9 34 35 17 14 14 15 12 Distal-to-proximal replacement 5 8 12 10 11 Nodal feathers 6 32 1 1 44 13 I Terminal feathers 10 7 1 1 12 recorded as nodal in some individuals. This is a con- placedfrom the innermost(P1) outward (Table sequenceof the loss of P2 following the loss of P1 2). P1 and P2 appearto be molted almostsimul- closelyin time. Other specimens,however, show that taneously,because the scoresof thesefeathers only P1 is a nodal feather,because it usually is lost wereeither the same (13 of 35 birdsmolting both before P2. P1 and P2) or differedby a scoreof only one. If S1and P1 arelost simultaneously (they are not in The secondaries are divided into two molt se- most passerines),recording both as nodal would be ries, S1-S6 and S7-S9 (Table 2). Molt of the inner appropriatebecause each marks the beginningof a separatemolt series.When the nodalfeathers of sepa- secondariesbegins with S8,when the outermost rate molt seriesare adjacent(e.g. S1 and P1),then the growing primaries are P2-P4, and proceedsin secondof theseto be replacedwill havea new feather the sequenceS8-S9-S7. Of 22 birdswith only one on one side and an old (or less advanced) feather on of these secondariesin molt, 21 were growing the other side. In this case, the first feather of the sec- S8. In all eight of the birds where only two in- ond seriesto beginmolting would incorrectlybe cat- ner secondaries were in molt, the feathers were egorizedas directional.The node of the secondseries S8 and S9. Molt in the outer series of secondar- canbe identifiedby examiningthe associationin tim- ies (S1-S6)begins with S1, after the molt of the ing of replacementwith nearbyfeathers. For example, inner secondarieshas been initiated, and pro- P2 is replacedshortly after P1 in passerines,but S1 may be lost when any of severalinner primariesare ceedsproximally (Table2). Molt of the rectrices is divided into an inner growing,indicating that it is nodal. Terminalfeathers.--When the focal featheris lessad- series (R1-R5) and an outer series (R6; Table 3). vancedin its growth than adjacentfeathers it is un- Rectrixmolt beginswith the lossof R1 (which ambiguouslyterminal, unless feathers of a seriesare occurs when either P3 or P4 is the outermost moltedin a nonlinearsequence. When a focalfeather growingprimary) and proceedsoutward to R5. has but one adjacentfeather (e.g. P9 in many pas- We included R5 in the inner series because in serines),it may be identified as terminal if it is lost most cases R5 is lost after R4 (in two cases R5 afterits neighbor.When a focalfeather shows one ad- was molted before R4 on one side but not on the jacentfeather in a moreadvanced state of growth,and other),whereas timing of the lossof R6 varied the otheradjacent feather is fully grown,then the fo- calfeather is terminal,and the growingfeather is part considerably.R6 was moltedbefore R4 in three of the molt series of the terminal feather. The new of seven cases, between R4 and R5 in three of featheris usuallythe nodalor terminalfeather of the seven cases, and after R5 in one of seven cases. adjacentmolt series.When two molt seriesproceed in oppositedirections to meetat a centrallylocated pair TABLE3. Frequencyof rectricesin active molt. Rec- of feathers,only one of thesetwo featherscan be iden- tricesare identified as directional,nodal, and termi- tified as a terminalfeather by thisrule. The othermust nal, usingreplacement rules described in Methods. be identifiedusing information on directionand tim- ing of replacement(e.g. S7 in this data set). Focalfeather R1 R2 R3 R4 R5 R6

RESULTS Proximal-to-distal replacement 12 9 7 2 Flight-feathermolt.--The sequence of primary Distal-to-proximal and secondarymolt in Northern Rough-winged replacement Nodal feathers 19 2 6 Swallowsis typicalof mostpasserines. The pri- Terminal feathers 1 5 1 maries form a single molt seriesand are re- April1997] Moltand Migration in Swallows 253

Eastern Birds

• [] E]E]•E]•D •D ßß xo x xx ß • nR E] •( XX X X

ß •]0 X I I I I I I I I I I I I 0 30 60 90 120 150 180 210 240 270 300 330 360

0 • []n • XX X XX I I I I I I I I I I I I iI0 30 m•e•ee 60 90 120 150Back 180 210 240 270 300 330 360

Chin

x

x

xx x x I I I I I I I I 60 90 120 150 180 210 240 270 300 330 360

1'33 •$•3• ß X

[] • • []]•ß•1• XX X XX I I I I I I I I I I I I 30 ma60o• •390 120 150 Breast 180 210 240 270 300 330 360

EI• []

I I I I I I I I ] I I I iI0 30m •=•e•60 90 ß 120 150 Belly 180 210 240 270 300 330 360

[] [3• •1ß;• XX X XX I I ] I I I I I I I I I Julian Dateii 0 30[] •60 eeß 90 120 150Rump 180 210 240 270 300 330 360

Date 1 Jun 1 Aug 1 Oct I Dec I Feb I Apr FIG.1. Temporaldistribution of body molt in crown,back, chin, breast,belly, and rump for easternpopu- lationsof Northern Rough-wingedSwallows. Only data for specimensundergoing body molt are shown.Data for westernpopulations not shown.See Methods for descriptionof molt score.Julian (day 1 = 1 June)and calendardates in bottom panel. 254 YURIAND ROHWER [Auk, Vol. 114

24- [] 22- [] Eastern Adult ß [] [] o Eastern Juvenile 20- [] [] [] •, EasternAge-unknown 18- [] [] [] ß Western Adult [] Q•3 o 16- []ß ß Western Juvenile (3•D o 14- ß Western Age-unknown ß o 0 12- o oo o [] ß 10- oß ß ß o 8- o o oo [] [3ß 0D ß o

0 •--•, 0 2'0 40 6'0 8'0 1•)0a120 ' 1'•0 160 180 200 220 240

FlightFeather Molt Score FIG.2. Timingof body molt comparedwith flight-feathermolt for NorthernRough-winged Swallow un- dergoingflight-feather molt.

Bodymolt.--Molt of non-flightcontour feath- in smaller gaps in the total surfaceof the pri- ers occurs at about the same time in each of the maries when the longer and more important six body regions(Fig. 1). The timing of molt is feathersof the wing are replaced.The most in- not significantlydifferent among body regions tenseperiods of inner and outer secondarymolt for all birds examined, or for the western and do not overlap (Table4). Body molt is most in- eastern populations analyzed separately (G- tense during the time that P3 to P6 are being tests, P > 0.30 in all cases). For this reason, we grown,and is mostlycompleted by the time that presentthe data only for easternbirds; the rela- P8 is growing. tionshipbetween molt and timing for western Timingof molt.---On average, eastern birds ini- birds is similar in all respects,except that west- tiate flight-feathermolt in early July and com- ern birds initiate molt later (seebelow). In juve- pleteit by late December(Fig. 3). Westernbirds niles, crown and chin feathers tend to molt later initiate flight-feathermolt in mid-August and and with less intensity than in adults. Spring completeit by early January.Analyses of the ef- specimensof unknown age showed light molt fects of population (western or eastern), age, on their crowns (10 of 11) and occasionallyon and sex on the timing of molt, with the stageof their chins (2 of 11). Among adults, body and the flight-feathermolt held constantas a covari- flight-feather molt begins at about the same ate, revealedthat populationand age, but not time, but the body molt is completedbefore all sex,affect the timing of flight-feathermolt sig- flight feathershave been replaced(Fig. 2). Out nificantly(population: F = 25.7, P < 0.001;age: of 27 adults with flight-feather molt scoresof F = 10.9, P < 0.01; sex: F = 0.62, P > 0.4; df = 1 more than 160, 17 had already finished their and 84 in eachcase). However, when only west- body molt. Juvenilesbegin their body molt be- ern populationswere examined,no significant fore the onset of their flight-feathermolt and, differencein flight-feathermolt scheduleswas like adults,complete it well beforereplacing all found betweenage classes (F = 0.61,df = 1 and flight feathers. 25, P > 0.4), but a significantdifference was Intensityand relative scheduling of molt.--When found between sex classes (F = 4.45, df = 1 and molt intensityis measuredas the mean number 25, P < 0.05). This differenceis not apparent of feathersgrowing simultaneously,the inten- when observedgraphically, and the resultsfor sity of the primary molt peaksat P5 (Table4). western populations may be due to small However,when measuredas the meanlength of samplesizes and the potentiallyspurious effects missingfeather(s), molt intensityis more scat- of outliers. tered, and the overall trend is a decline in in- Easternadults begin molting flight feathersin tensity from P2 to P9. This pattern may result earlyJuly, about 26 daysearlier than easternju- April 1997] Moltand Migration in Swallows 255

TABLE4. Summaryof the intensityand schedulingof molt in differentfeather tracts.

Outermostgrowing primary 1 2 3 4 5 6 7 8 9 Samplesize 8 26 21 11 7 8 11 8 12 Mean no. of feathersgrowing per side Primaries 1.00 1.96 2.24 2.36 2.43 1.75 1.82 1.38 1.25 Inner secondaries (7-9) 0.31 0.95 1.36 1.43 1.00 Outer secondaries (1-6) 0.05 0.86 1.00 1.82 1.00 0.75 Inner rectrices (1-5) 0.13 0.24 0.64 1.57 2.00 2.56 0.63 Outer rectrices (6) 0.64 0.13 Totals 1.13 2.27 3.48 4.36 6.29 5.75 6.82 3.13 2.00 Mean mm of feathers missing per side Primaries 44.6 79.6 66.3 100.4 85.2 67.9 74.2 72.5 56.7 Inner secondaries (7-9) 8.1 23.2 28.6 25.4 20.9 Outer secondaries (1-6) 2.2 31.6 33.6 57.3 17.9 15.4 Inner rectrices (1-5) 5.9 11.0 29.3 56.0 48.4 70.3 18.9 Outer rectrices (6) 19.7 6.4 Totals 50.6 87.8 102.8 158.3 198.1 170.7 221.6 115.7 72.1 Mean body-moltscore 5.25 4.46 11.52 15.73 17.86 10.63 5.82 0.75 0.33 veniles(Fig. 4A, Table5) and about40 daysear- October.Peaks are late July in easternadults and lier than western adults (Table 5). The data in- mid-Augustin westernadults (Fig. 5), a sea- dicate that western juveniles initiate flight- sonal difference that is highly significant feathermolt about 10 days later than western (Gaaj = 23.3,df = 4, P < 0.001).Eastern birds adults(Fig. 4B, Table 5), althoughthe difference show no significantdifferences between age is not significant.Estimates of the mean dura- and sexgroups in thetiming of bodymolt (age: tion of flight-feathermolt are 94 and 105 days Gaaj = 2.29,df = 4, P > 0.65;sex: Gaa j = 6.73, for easternadults and juveniles,respectively. df = 5, P > 0.2). Samplesizes for the western The estimatesfor westernpopulations are not populationare too small for analysis. discussedbecause of the largevariability in the Molt andmigration.--Non-molting birds tend data. to be found in the north, whereas birds in the Bodymolt occurs from late June through early later stagesof flight-feathermolt are concen-

240'

o o 200-

o o• 160- [] [] ß [] o • 120- • o

o u_ 80- [] Eastern Adult ß Western Adult o Eastern Juvenile ß Western Juvenile 40- A EasternAge-unknown ß WesternAge-unknown

Julian Date 0 60 120 150 180 210 240 270 300 330 360

Date 1 Jun 1 Aug 1 Oct 1 Dec 1 Feb 1 Apr FIG.3. Progressionof flight-feathermolt for adultand juvenile Northern Rough-winged Swallows in east- ern and westernpopulations. Molt scoreof 0 indicatesno molt,and 240indicates completion of molt.Note the generaltemporal order (eastern adults, eastern juveniles, western adults, and western juveniles). 256 YURIAND ROHWER [Auk, Vol. 114

180. A Eastern Birds winged Swallowsis typical of most passerines except for the outermostrectrices (R6). R6 is moltedbefore R5, and oftenbefore R4, a pattern o also found in several other passerines(e.g. Eu- ropeanBarn Swallow [Hirundo rustica], Eurasian Crag-Martin [Hirundorupestris], and Meadow Pipit [Anthuspratensis]; Jenni and Winkler1994). Severalspecies of Motacillidaealso feature an early loss of R6 (Jenniand Winkler 1994). We suspectthat closerattention to establishingthe rules of flight-featherreplacement may reveal 180' B Western Eirds that the rectricesare typically organized into 1 Nov• 150' two molt series, as we have documented for Northern Rough-wingedSwallows. 1 Oct •120. The intensity of molt is relatively constant 1 Sep• 90' throughoutmuch of the molt, which may reflect

I Aug• 60' the fact that theseswallows are under pressure to maintainhigh flight efficiency.Intensity of the I Jul -•' 30' primary molt peaksearly and then steadilyde- I0 AdultOJugnile ] 1 Jun • clinesas the molt proceedstowards the longest

Dale Julian Date FlightFeather Molt Score and most important primaries. This phenom- (Day 1 = 1 Jun) enonis muchmore evident when molt intensity FIG. 4. Least-squareslinear regressionsof collec- is measuredas the meanlength of missingfeath- tion dates againsttotal flight-feathermolt scorefor ers,rather than as the meannumber of growing adultand juvenileNorthern Rough-winged Swallows feathers,which suggeststhe importanceof the separatedby region of collection.Birds collectedin former measurementof intensity. winter are of unknown age and thereforeexcluded. Timingof molt in adultsrelative to breeding.- Easternadults: y - 0.439x+ 34.0 (F = 55.6, df = 1 Adults beginreplacing body and flight feathers and29, P < 0.001,R 2 = 0.657);eastern juveniles: y = 0.393x + 60.0 (F = 125, df = I and 34, P < 0.001, about40 days later in the west than they do in R2 = 0.786);western adults: y = 0.294x+ 73.8(F = the east (Table 5, Fig. 5). Breedingschedules 15.0,df = I and15, P < 0.01,R 2 = 0.499);eastern ju- also differ between western and easternpopu- veniles:y = 0.432x + 83.3 (F = 16.2, df = 1 and 13, P lations; egg dates for northwesternbreeding < 0.01, R2 = 0.555).Only specimensundergoing localitiesrange from early June to early July, flight-feathermolt areanalyzed. Solid regression lines with young fledging in late July (Weydemeyer are for adults,dashed lines for juveniles. 1933, Bent 1942, Jewett et al. 1953, Burleigh 1972, Cannings et al. 1987). In contrast, egg dates for northeasternbirds range from mid- trated in the south(Figs. 6•8). Thereis a signifi- May to mid-June,with youngfledging in early cant difference in the latitudinal distribution of July (Bent 1942, Lunk 1962, Mumford and locations where specimens were collected Keller 1984, Bull 1985, Bohlen 1989). The initia- amongdifferent molt stages (Gaa • = 108.1,df = tion of molt coincideswith the fledging of 10, P < 0.001).Clearly, Northern Rough-winged young in both populations,beginning in early Swallowsare migratingsouthward at the same July in eastern adults and early August time that they are molting.Map summariesin- in western adults. Both male and female dicatethat westernbirds probablymigrate con- Northern Rough-winged Swallows provision tinuously through the western United States their young, which may explain why the sexes and Mexicowhile they are molting (Figs.6-8). are similar in their molt schedules. By contrast, eastern birds move southward TreeSwallows ( bicolor) also initiate while they are moltingonly until they reachthe flight-featherand body molt when their young Gulf of Mexico (Fig. 8). are fledging (Stutchburyand Rohwer 1990). However,Tree Swallowscomplete their flight- DISCUSSION feather and body molt at about the sametime, whereas Northern Rough-winged Swallows Sequenceof flight-feathermolt.--The sequence complete their body molt well before their of flight-feather molt in Northern Rough- flight-feathermolt is finished(Fig. 2). Thisis due April 1997] Moltand Migration in Swallows 257

TABLE5. Estimatesof startingand completiondates, and durationof flight-feathermolt + 95% confidence intervalusing the least-squareslinear regression model. Dates are in Juliandate (Day 1 = 1 June),with calenderdates in parentheses.

Startingdate Completiondate Duration(days) Easternpopulations Adults 34 (4 Jul) + 13 139 (17 Oct) _+20 105 + 29 Juveniles 60 (30 Jul) + 6 154 (1 Nov) + 14 94 -+ 17 Westernpopulations Adults 74 (13 Aug) + 14 144(22 Oct) + 31 71 + 39 Juveniles 83 (22 Aug) -+ 11 187(4 Dec) + 52 104 + 56 to the shorterduration of body molt in North- and the molt scoresfor crown and chin in juve- ern Rough-winged Swallows (about three nilesgenerally are low. Interestingly,the limited months;see Fig. 5) comparedwith Tree Swal- springmolt is restrictedalmost entirely to feath- lows (about four months). ers of the crown and chin, suggestingthat birds Timingof moltin juveniles.--JuvenileNorthern moltingin springmay be second-yearbirds. Rough-wingedSwallows delay flight-feather Timingof moltrelative to migration.--Northern molt but replacebody feathersat the sametime Rough-wingedSwallows molt during much of as do adults.Juvenile European Barn Swallows their fall migration,which occursfrom July to also delay the initiation of flight-feathermolt Octoberin easternpopulations and from August (Cramp 1988).Adult Northern Rough-winged to November in western populations (Bent Swallowsinitiate flight-feather molt whenjuve- 1942). Two recoveries of banded birds ( nilesare only a few weeksold and may still be BandingLaboratory) indicate the timing of post- growingtheir juvenal flight feathers (three early breedingmigration: one was bandedon 3 July juvenilesfrom Utah werestill growingouter pri- 1950in Michiganand recapturedon 13 August maries).Juveniles may delaytheir flight-feather 1950 in Tennessee,and the other was banded on molt until they becomeexperienced at foraging 24 June 1945 in Ontario, Canada and shot on 29 and escapingpredators. Body molt may start August1945 in Louisiana. earlierbecause it hasless influence on flight per- Many other swallowsmolt during migration formance.Fewer juvenilesare molting crown (PurpleMartins [ subis], Niles 1972;Crag and chin feathersthan in other body regions, Martins, Elkins and Etheridge 1977; European

24- [] [] 22- o [] Eastern/•dult [] ß [] o Eastern Juvenile 20- o •. Eastern Age-unknown 18- [33[3 ß Western Adult [] E]O 0 •16- [] ß © Western Juvenile o o o o ß Western Age-unknown

o

•10-

o

4-

2-

Julian Date 0 3'0 6'0 9'0 120 150 180 210 240 270 300 3:•0''3•0

Date 1 Jun 1 Aug 1 Oct 1 Dec 1 Feb 1 Apr FIG.5. Progressionof body molt for adultsand juvenilesin easternand westernpopulations of Northern Rough-wingedSwallows. Body-molt score of 0 indicatesno molt or completionof molt. On average,eastern adultsand juveniles undergo molt before western adults and juveniles. 258 YURIAND ROHWER [Auk, Vol. 114

NoFlight Feather MoltI

ß Adults & Subadults ß Juveniles

FIG.6. Geographicaldistribution of NorthernRough-winged Swallow specimens collected before flight- feathermolt. Only specimenscollected between July and November are included. Numbers represent the num- ber of specimenscollected at the samelocality.

Barn Swallows,Cramp 1988; Tree Swallows, many individualsare probablywell into migra- Stutchburyand Rohwer 1990).Niles (1972) ar- tion by that time (cf. Figs. 6 and 7). Northern gues that swallowscan meet the energy de- populationsof Bullock'sOrioles (Icterusbul- mands of molt during migrationby migrating lockii)also migrate to the southwesternUnited during daytime with frequentinterludes for for- Statesbefore they begin prebasicmolt and are aging. This pattern contrastswith most other thoughtto do so to escapethe dry late-summer passerinemigrants, which migrate at night. For conditionsof the breeding grounds (Rohwer manyswallows, the instantaneouscost of flight- and Manning 1990,Rohwer and Johnson1992). feathermolt is further reducedby its long du- In contrast,Baltimore Orioles (I. galbula)un- ration (four to eight months;Niles 1972,Pyle dergo the prebasic molt on the breeding 1987,Stutchbury and Rohwer 1990). groundsprior to migration. The 40-day differencein timing of flight- Easternbirds may initiate flight-feathermolt feathermolt betweeneastern and westernpopu- earlybecause of their needto completeit before lationscannot be fully explainedby differences the trans-Gulf migration. Most eastern speci- in breeding schedules,because eastern birds menstaken in later stagesof flight-feathermolt fledgetheir young less than a monthearlier than were collected in southern Alabama, Missis- westernbirds. Western birds may delay initiat- sippi,and Louisiana(Fig. 8). Burleigh(1944:407) ing their molt until they have passedthrough documentedthe early departure of Northern the northernparts of their breedingrange where Rough-wingedSwallows from the breeding late-summerconditions are dry and unproduc- rangein the easternUnited Statesand their con- tive. The mean starting date of flight-feather centrationalong the Gulf coastof Mississippifor molt for western populations of Northern over two months. He noted that flocks of North- Rough-winged Swallows is mid-August, and ernRough-winged Swallows assemble along the April1997] Moltand Migration in Swallows 259 Fir;tHalfStages ofFlight Feather Molt

-Juv

FIc. 7. Ceo•aphica] dis•ibuQono• No•he•n Rou•h-win•ed Swallow spedme• collecteddu•in• •he fi•s[ ha]• o• Qi•h•-•ea•he•molt. Only specimenscollected between July and Novembe•a•e included.Numbers •ep- •esen••he numbe•o• specime• collecteda• •he same

Gulf coastfrom the middleof Julythrough early apparentlyhad interruptedher flight-feather August,gradually disappearing only after the molt just beforeits completion,as P9 had not first of October.The timing of their disappear- been replacedon either wing. We found no ance coincideswith the completionof their specimencollected along the Gulf coastof Texas flight-feathermolt. Thus, eastern populations of and northeasternMexico (Figs. 6-8). NorthernRough-winged Swallows may initiate In summary,evidence suggests that earlyfall their molt earlier to accommodate their trans- arrivals of eastern Northern Rough-winged Gulf migration,which they apparentlydo not Swallowsgather at the northernGulf coastto undertakeuntil flight-feathermolt is completed. completetheir flight-feathermolt before con- We found no published documentationof tinuingtheir southward migration. A late trans- a trans-Gulf migration in Northern Rough- Gulf migrationapparently follows the comple- wingedSwallows, but thereis indirectevidence tion of thismolt. Able (1972)suggests that opti- for sucha migration,including several winter mum conditionsfor directbird flight from the recordsfor the WestIndies (Bond 1956, Faaborg northernGulf coastto the tropics(i.e. following and Terborgh1980, Voous 1983) and Guatemala the passageof a coldfront far into the Gulf) oc- (Griscom1932). Northern Rough-winged Swal- cur in early to mid-October.Thus, weather con- lows arrive in Yucatan in mid-October (Howell ditions and the late completionof the flight- 1989). Northern Rough-wingedSwallows are feather molt could favor a late migration of not recordedin any of the publishedaccounts Northern Rough-wingedSwallows over the of trans-Gulfmigrants dating from late August Gulf. through early September(e.g. Paynter 1953). Easternjuveniles complete the flight-feather One specimenwas collectedin Quintana Roo, molt in late November. If juveniles migrate Mexico on 29 October 1952, an adult female who acrossthe Gulf with adults in late October, then 260 YURIAND ROHWER [Auk, Vol. 114

Second Half Stages of Flight Feather Molt

•3

ß Adults & Subadults ß Juveniles

FIG.8. Geographicaldistribution of NorthernRough-winged Swallow specimens collected during the sec- ond half of flight-feathermolt. Only specimenscollected between July and Novemberare included.Numbers representthe numberof specimenscollected at the samelocality. In comparingfigures 64, notethat speci- mensin moreadvanced stages of molt are collectedfarther south. they would still be growingflight feathersdur- and migrationin NorthernRough-winged Swal- ing their trans-Gulfmigration. The frequencyof lows. northerly,post-frontal winds, and particularly Conclusions.--Becausetemperate swallows those penetratingto Yucatanand beyond,is are both diurnal migrantsand aerial foragers, greaterin Novemberthan in Septemberor Oc- theyhave less need to separatemolt and migra- tober (Buskirk 1980). Thus, weather conditions tion than do other passerines.When swallows are suitablefor a late trans-Gulfmigration in ju- migrateacross deserts and large bodies of ocean veniles. However, such a late-fall passagemay unsuitablefor foraging and obtaining water, result in young-of-the-yearbeing confronted they have differentways of avoidingthe con- with wintering habitatsthat are alreadyfilled flictsof simultaneouslymolting and migrating. with earlier migrants. For this reason,we sus- North American Barn Swallows and Purple pectit will proveto be late-moltingyoung birds Martins interrupt their prebasicmolt for the that makeup the flocksof Rough-wingedSwal- trans-Gulfmigration (Niles 1972).Similarly, Eu- lows wintering near the SabineNational Wild- ropeanpopulations of BarnSwallows, which are life Refugeon the Louisiana-Texasborder (Root trans-Saharanmigrants, interrupt their prebasic 1988). Investigationsof: (1) the progressionof molt during the passagefrom southernEurope flight-feathermolt in adultsand juvenilesalong to tropicalAfrica (Cramp 1988). In contrast,east- the Gulf Coast,(2) seasonalchanges in age ra- ern breeding populationsof Northern Rough- tios of fall birds taken along the Gulf, and (3) winged Swallowsinterrupt their migrationin the age classesof individualswintering on the order to completemolt beforemaking their fall United States side of the Gulf of Mexico will fur- passageover the Gulf of Mexico. ther elucidatethe relationshipsbetween molting Why shouldNorthern Rough-winged Swal- April 1997] Moltand Migration in Swallows 261 lows have opted for delaying their migration Universityof WashingtonBurke Museum, Washing- while they finishmolting? Barn Swallows, Cliff ton State University Conner Museum, U.S. National Swallows (Hirundopyrrhonota), Bank Swallows Museum of Natural History. Bandingrecovery data (Ripariariparia), and Purple Martins winter in were kindly suppliedby the Bird BandingLaboratory, National BiologicalSurvey. D. M. Niles scoredmany SouthAmerica, where they occupya relatively of thespecimens for molt.R. B.Payne, J. S. Albert, and large land area. All of thesespecies molt most J.M. Swalesprovided valuable comments on the early of their flight feathersafter arriving in their draft of this manuscript.C. Thompsonhelped with winter quarters (Pyle et al. 1987). In contrast, referenceson the trans-Gulfmigration. Northern Rough-wingedSwallows, Tree Swal- lows, and Violet-green Swallows (Tachycineta thalassina)winter in the southern United States, LITERATURE CITED Mexico, and Central America,where they are ABLE,K. P. 1972. Fall migrationin coastalLouisiana concentratedinto a much smallerand perhaps and the evolutionof migrationpatterns in the less productiveland area. These speciescom- Gulf region.Wilson Bulletin 84:231-242. plete their annualmolt by late fall, eitheron the BENT,A. C. 1942. Life histories of North American breedinggrounds or during southwardmigra- flycatchers, larks, swallows and their allies. tion in North America.Similar patterns are also United States National Museum Bulletin No.179. found in many Europeanspecies; those winter- BOHLEN,H. D. 1989. The birds of Illinois. Indiana UniversityPress, Bloomington. ing in southernEurope molt on their breeding BOND,J. 1956. Check-list of birds of the West Indies, grounds,whereas many trans-Saharanmigrants 4th ed. Academyof Natural Sciencesof Philadel- molt in southern Africa (Jenni and Winkler phia,Philadelphia, Pennsylvania. 1994). BULL,J. 1985. Birds of New York State. Cornell Uni- If a winter molt is energeticallyor nutrition- versityPress, Ithaca, New York. ally difficult for swallows that winter in the BURLEIGH,T. D. 1944. The bird life of the Gulf coast Northern Hemisphereon a relativelysmall land region of Mississippi.Occasional Papers of the area, then one would expect that Northern Museumof ZoologyLouisiana State University Rough-wingedSwallows from western North No. 20. BURLEIGH, T. D. 1972. Birds of Idaho. Caxton Print- Americaalso should pause in the southwestern ers, Caldwell, Idaho. United Statesto completetheir molt (or at least BUSKIRK,W. H. 1980. Influenceof meteorologicalpat- migratevery slowly throughthe southwest)be- terns and trans-Gulfmigration on the calendars fore migratingto their wintering groundsin of latitudinalmigrants. Pages 485-491 in Migrant southern Mexico and Central America. Unfortu- birdsin the Neotropics:Ecology, behavior, distri- nately,we couldnot evaluatethis predictionbe- bution, and conservation(A. Keast and E. S. Mor- causeof the dearthof specimenstaken in west- ton, Eds.). Smithsonian Institution Press, Wash- ern North Americaduring the later stages of pri- ington, D.C. mary molt. CANNINGS,R. A., R. J. CANNINGS,AND S. G. CANNINGS. 1987. Birds of the OkanaganValley, BritishCo- lumbia. Royal BritishColumbia Museum, Victo- ACKNOWLEDGMENTS ria, British Columbia. CHILGREN,J. D. 1975. Dynamicsand bioenergeticsof We are very gratefulto the followinginstitutions post-nuptialmolt in captiveWhite-crowned Spar- and their curatorsfor accessto specimens:Museum rows (Zonotrichialeucophrys gambelii). Ph.D. dis- of NorthernArizona, University of Arizona,Califor- sertation,Washington State University, Pullman. nia Academyof Sciences,Los AngelesCounty Mu- CRAMP S. (Ed.) 1988. Handbook of the birds of Eu- seumof Natural History,San Diego Natural History rope,the Middle Eastand North Africa.The birds Museum,University of Californiaat LosAngeles, Uni- of the western Palearctic, vol. 5. Oxford Univer- versityof CaliforniaMuseum of VertebrateZoology, sity Press,Oxford. WesternFoundation of VertebrateZoology, Denver DIETZ, M. W., S. DANN, AND D. MASMAN. 1992. En- Museumof naturalHistory, University of Colorado, ergy requirementsfor molt in the Kestrel,Falco FieldMuseum of Natural History,University of Kan- tinnunculus.Physiological Zoology 65:1217-1235. sasMuseum of Natural History,Louisiana State Uni- ELKINS,N., AND B. ETHERIDGE.1977. Further studies versity Museum of Natural Sciences,Museum of of wintering Crag Martins. Ringing and Migra- ComparativeZoology, University of Michigan Mu- tion 1:158-165. seumof Zoology,American Museum of Natural His- FAABORGJ. R., ANDJ. W. TERBORGH.1980. Patternsof tory,Carnegie Museum of Natural History,Philadel- Migration in the West Indies. Pages157-163 in phiaAcademy of NaturalSciences, University of Utah, Migrant birds in the Neotropics:Ecology, behav- 262 YURIAND ROHWER [Auk, Vol. 114

ior, distribution, and conservation (A. Keast and PEREK,M., AND F. SULMAN. 1945. The basal metabolic E. S. Morton, Eds.). Smithsonian Institution Press, rate in molting and laying hens.Endocrinology Washington,D.C. 36:240-243. GAVR1LOV,V. M. 1974. [Metabolismof moltingbirds.] P1MM, S. 1976. Estimation of the duration of bird ZoologicheskiiZhurnal 53:1363-1375. molt. Condor 78:550. GAVR1LOV,V. M, AND V. R. DOLN1K.1974. [Bioener- PYLE,P., S. N. G. HOWELL,R. P. YUNICK,AND D. E DE- getics and regulation of the postnuptial and SANTE. 1987. Identification guide to North postjuvenalmolt in Chaffinches(Fringilla coelebs Americanpasserines. Slate Creek Press,Bolinas, coelebs).]Trudy ZoologicheskogoInstituta Aca- California. demii Nauk SSSR 55:14-61. ROHWER,S. 1986. A previouslyunknown plumage of GRlSCOM,L. 1932. Thedistributionofbird-lifeinGua- first-yearIndigo Buntingsand theoriesof delayed temala. Bulletin of the American Museum of plumagematuration. Auk 103:281-292. Natural History No. 64. ROHWER,S., AND M. S. JOHNSON.1992. Scheduling HOWELL,S. N. G. 1989. Additional information on the differencesof molt and migration for Baltimore birds of the CampecheBank, Mexico. Journalof and Bullock'sorioles persist in a commonenvi- Field Ornithology 60:504-509. ronment. Condor 94:992-994. JENN1,L., ANDR. W1NKLER.1994. Moult and ageing ROHWER,S., AND J. MANNING. 1990. Differences in of Europeanpasserines. Academic Press, London. timing and number of molts for Baltimoreand JEWETT,S. G., W. P. TAYLOR,W. t. SHAW,AND J. W. AL- Bullock'sorioles: Implications to hybrid fitness DRICH.1953. Birdsof WashingtonState. Univer- and theories of delayed plumage maturation. sity of WashingtonPress, Seattle. Condor 92:125-140. LANGSTON,N. E, ANDS. ROHWER.1995. Unusualpat- ROOT,T. L. 1988. Atlas of winteringNorth American ternsof incompleteprimary molt in Laysanand birds.University of ChicagoPress, Chicago. Black-footed albatrosses. Condor 97:1-19. SOKAL,R. R., ANDE J.ROHLF. 1995. Biometry,3rd ed. L1NDSTR•M,•., G. H. VISSER,AND S. DANN.1993. W. H. Freemanand Company,New York. The energeticcost of feathersynthesis is propor- STILES,E G. 1981. The of Rough-winged tionalto basalmetabolic rate. Physiological Zool- Swallows(Stelgidopteryx; Hirundinidae) in south- ogy 66:490-510. ern Central America. Auk 98:282-293. LUNK,W. A. 1962. The Rough-wingedSwallow: A STRESEMANN,V. 1963. Zur Richtungsumkehrder studybased on its breedingbiology in Michigan. Schwingen-und Schwanzmanservon Muscicapa Publicationsof the Nuttall OrnithologicalClub striata.Journal ffir Ornithologie104:101-111. No. 4. STUTCHBURY,B. J., AND S. ROHWER.1990. Molt pat- LUST1CK,S. 1970. Energy requirementsof molt in ternsin the TreeSwallow (Tachycineta bicolor). Ca- cowbirds. Auk 87:742-746. nadianJournal of Zoology68:1468-1472. MILLER,A. H. 1957. FamilyHirundinidae. Pages 110. UNDERHILL,L. G., AND W. ZUCCHINI. 1988. A model 112 in Distributional checkqist of the birds of for avian primary molt. Ibis 130:358-372. Mexico (A. H. Miller, H. Friedmann, L. Griscom, Voous, K. H. 1983. Birds of the Netherlands Antilles, and R. T. Moore, Eds.). Pacific Coast Avifauna No. 2nd ed. De WalburgPers, Zutphen, The Nether- 33. lands. MINITAB INC. 1991. MINITAB statistical software, re- WALLGREN,H. 1954. Energymetabolism of two spe- lease 8.21, Macintosh version. Minitab Inc., State ciesof the genusErnberiza as correlatedwith dis- College,Pennsylvania. tribution and migration.Acta ZoologicaFennica MUMFORD,R. E., AND C. E. KELLER. 1984. The Birds 84:1-110. of Indiana. Indiana University Press,Blooming- WEYDEMEYER,W. 1933. Nesting of the Rough-winged ton. Swallow in Montana. Auk 50:362-363. MURPHY,M. E. 1996. The energeticsand nutrition of W1JNANDTS,H. 1984. Ecologicalenergetics of the molt. Pages158-198 in Avian energeticsand nu- Long-earedOwl (Asiootus). Ardea 72:1-92. tritional ecology(C. Carey, Ed.). Chapman and YOUNG,B.E. 1991. Annual molts and interruptionof Hall, New York. the fall migrationfor moltingin Lazuli Buntings. NEWTON,I. 1966. The moult of the BullfinchPyrrhula Condor 93:236-250. pyrrhula.Ibis 108:41-67. NILES,D. M. 1972. Molt cyclesof Purple Martins AssociateEditor: J. S. Marks (Prognesubis). Condor 74:61-71. PAYNTER,R. g. 1953. Autumnal migrants on the CampecheBank. Auk 70:338-349.