Click here to load reader

ROOSTS AND MIGRATIONS OF SWALLOWS

  • View
    216

  • Download
    1

Embed Size (px)

Text of ROOSTS AND MIGRATIONS OF SWALLOWS

  • 2006 ROOSTS AND MIGRATIONS OF SWALLOWS 85

    ROOSTS AND MIGRATIONS OF SWALLOWS

    DAVID W. WINKLER

    Cornell University Museum of Vertebrates, Department of Ecology and Evolutionary Biology,Cornell University. Ithaca, NY 14853, USA. [email protected]

    ABSTRACT. Swallows of the north temperate zone display a wide variety of territorial behaviourduring the breeding season, but as soon as breeding is over, they all appear to adopt a pattern ofindependent diurnal foraging interleaved with aggregation every night in dense roosts. Swallowsgenerally migrate during the day, feeding on the wing. On many stretches of their annual jour-neys, their migrations can thus be seen as the simple spatial translation of nocturnal roost siteswith foraging routes straightening out to connect them. However, swallows that must make longjourneys over ecological barriers clearly fly at night as well as in the day, and many suggestionsindicate that there is considerable complexity in the altitude and bearing of flights even duringthe day. There are especially intriguing indications that much swallow migration may take placehigh out of sight of ground observers with movements near the ground often associated withforaging in passage. Provided that roost sites can be reliably found, swallow migration can beextremely flexible, and there are interesting contrasts in the biogeography and phenologicalflexibility of swallows compared to other passerine birds. Even within the swallows, there isconsiderable interspecific and intraspecifc variability in the distances of their annual migrations,and we are only just beginning to understand the biological causes and consequences of thisvariation. The profusion of Doppler weather radar stations in the eastern United States has allowedthe characterization in considerable detail of the North American distributions of Tree Swallows(Tachycineta bicolor) and Purple Martins (Progne subis) throughout the non-breeding season.Evaluating the relative roles of movements and mortality in creating these patterns remains animportant challenge for further research.KEY WORDS: diurnal, Hirundinidae, martins, migration, nocturnal, roost, swallows.

    RESUMEN. DORMIDEROS Y MIGRACIONES DE GOLONDRINAS. Las golondrinas de la zona templada deAmrica del Norte poseen una amplia variedad de comportamientos territoriales durante laestacin de cra, pero ni bien culmina la reproduccin todas parecen adoptar un patrn comn,alternando la alimentacin diurna independiente con el agrupamiento en populosos dormiderosdurante la noche. Las golondrinas generalmente migran durante el da, alimentndose en vuelo;sus migraciones, entonces, pueden ser vistas como si fuesen un simple traslado entre distintossitios que poseen dormideros nocturnos, con las rutas de alimentacin conectndolos directa-mente. Sin embargo, las golondrinas que deben realizar largos viajes cruzando barreras ecolgicasvuelan tanto de noche como de da, y hay evidencias que indican que hay una considerablecomplejidad en la altitud y en las caractersticas de los vuelos an durante el da. Hay evidenciasespecialmente interesantes de que la mayor parte de la migracin de las golondrinas puede tenerlugar a una altura tal que no es advertida por los observadores en tierra, pero con movimientoscercanos al suelo a menudo asociados con la alimentacin. Si los sitios con dormideros puedenser encontrados con certeza, la migracin de las golondrinas sera extremadamente flexible, yexisten interesantes contrastes en la biogeografa y la flexibilidad de la fenologa de las golondri-nas en comparacin con otros paseriformes. Entre las golondrinas mismas hay una considerablevariabilidad inter e intraespecfica en la distancia de migracin anual, y estamos an empezandoa entender las causas y las consecuencias biolgicas de esta variacin. La creciente disponibilidadde estaciones con radares climatolgicos Doppler en el este de Estados Unidos ha permitido lacaracterizacin, con un considerable detalle, de las distribuciones norteamericanas de la Golon-drina Bicolor (Tachycineta bicolor) y la Golondrina Purprea (Progne subis) durante la estacin noreproductiva. La evaluacin del papel relativo que juegan los movimientos y la mortalidad en laconformacin de esos patrones es un importante desafo para futuras investigaciones.PALABRAS CLAVE: diurno, dormidero, golondrinas, Hirundinidae, migracin, nocturno.

    Received 13 March 2006, accepted 26 December 2006

    Hornero 21(2):8597, 2006

  • 86 WINKLER Hornero 21(2)

    The spatial biology of swallows and martins(Hirundinidae) in the breeding season isextremely diverse, ranging from dense andlarge colonial aggregations to dispersed terri-torial breeding (Turner and Rose 1989). Mostspecies are central-place foragers spendingmost of their foraging time well away fromthe immediate vicinity of the nest (Bryant andTurner 1982, Turner 1982, McCarty andWinkler 1999). The diversity of summer soci-ality has engendered a rich literature in thecomparative behavioural ecology of breedingsocial systems (e.g., Hoogland and Sherman1976, Mller 1987, Brown and Brown 1996,Safran 2004). The plethora of data on breed-ing biology and ecology (e.g., Winkler andAllen 1996, Burness et al. 2001, Hasselquist etal. 2001, Ardia et al. 2003, Czarnowski et al.2004, Lombardo et al. 2004, Safran andMcGraw 2004, Safran et al. 2005) contrastssharply with the relatively few studies thathave been published about the biology ofswallows during migration and in the non-breeding season (e.g., Lyuleeva 1973, Meadand Harrison 1979, Szep 1995). My goal hereis to review what we know about swallowmigration and wintering biology and presenthypotheses to explain some of the patternsthat are beginning to appear. I must empha-size that we know far too little, and this reviewis thus intended to serve as an invitation tofurther research rather than a compendiumof established facts.

    SWALLOWS AS NOCTURNAL ROOSTERS

    At least in the Northern Hemisphere,hirundinids outside the breeding season tendto spend every evening in large dense roosts.These roosts can sometimes be very large (wellin excess of a million birds; van den Brink etal. 2000, Burney 2002, Bijlsma and van denBrink 2005), and there are interesting differ-ences in roosting behaviour among species.Barn Swallows (Hirundo rustica) and TreeSwallows (Tachycineta bicolor) prefer to roostin reed or cane beds usually over water (vanden Brink et al. 2000, Burney 2002), occasion-ally using trees or wires (e.g., Moreau 1972).In contrast, the larger martins in the genusProgne prefer to roost in shrubbery or trees oron ledges under bridges and in large indus-trial buildings (Oren 1980, Russell andGauthreaux 1999, Purple Martin Conservation

    Association 2005). All species appear to preferto roost over water, although all will roost intheir favoured substrate over dry land onoccasion (e.g., Skutch 1960). Species in thegenus Progne are variable in their approachto roosts, in some locations flying into theroost in small parties as sunset approaches(Russell and Gauthreaux 1999) and, in others,funnelling-in in one huge concentration (Oren1980). By contrast, Barn, Tree and rough-winged swallows of the genus Stelgidopteryxgenerally enter their roosts in very large, tightand short-duration streams.

    One factor determining why Progne speciesprefer to roost in woody vegetation and onfirm structures may be that their larger massserves as a deterrent to perching in the tips ofherbaceous marsh vegetation. This mayexplain why Progne species are less tied toroosting over water than are the smaller spe-cies. The fact that these species enter the roostin a less organized fashion than do the smallerswallow species may also indicate that thesmaller species may have had their roostinghabits most strongly molded by the threat ofpredation. The smaller swallow species tendto come to roost some time in the hour aftersundown. Our observations of Tree Swallowroosts (Burney 2002, pers. obs.) appear to befairly typical of these smaller species (seeSkutch 1960 on Stelgidopteryx species): theswallows display their normal reluctance tosettle into vegetation, and the birds congre-gate a few hundred meters above the roostsite, milling around the site in an increasinglylarge and dense cloud of birds. Finally, as thelast daylight fades, a few courageous birdsmake the plunge downward into the reeds ofthe roost site, followed immediately by a swirl-ing stream of birds pouring into the vegeta-tion, with hundreds of thousands of birdssettling in only a few minutes time. Few whohave witnessed these spectacles can avoiddescribing the descending cloud as a tornado.Both the late hour of gathering and whatseems to be the birds extreme reluctance toventure solo into the vegetation suggest thatthe principal selective force molding thisbehaviour has been the risk of predation.Observations on Barn Swallows being preyedupon by aerial raptors as they come to roostin Africa (Bijlsma and van den Brink 2005)support this interpretation. This concernabout predation may also explain why the

  • 2006 ROOSTS AND MIGRATIONS OF SWALLOWS 87

    birds appear to gather higher over the rooston clearer evenings (Skutch 1960).

    The truly large roosts of hirundinids occurin migratory species, with the largest roostsgenerally being reported in wintering areasthat support birds that bred over much largerareas in the north. It remains to be seen towhat extent the non-migratory or short-dis-tance species visit roosts in the non-breedingseason. Resident Southern Rough-wingedSwallows (Stelgidopteryx ruficollis) appeared tojoin roosts of migrant Northern Rough-winged Swallows (Stelgidopteryx serripennis)in cane when they were in the vicinity, butthey also roosted in smaller aggregations inthe same site when the migrants were gone(Skutch 1960). Skutch (1960) also describes asmall group of seven Black-capped Swallows(Notiochelidon pileata) roosting together everynight after the breeding season in cold cloudforest habitat, in a burrow that was not usedfor nesting. He also reports that the Blue-and-white Swallow (Notiochelidon cyanoleuca) canroost year-round in and near its nest; thus,these species apparently do not join largepost-breeding roosts. I have seen a White-winged Swallow (Tachycineta albiventer) pair

    settle down for the night with a fully volantjuvenile on small branches over a runningstream near Eldorado, Misiones Province,Argentina, on a date (9 March 2001) thatappeared to be outside the local breeding sea-son (Belton 1985). Roosts of 200010000 swal-lows (mostly the migrants Hirundo rustica andRiparia riparia in southern summer andTachycineta leucorrhoa and Tachycineta meyeni insouthern winter) occur regularly in the stateof Rio Grande do Sul in Brazil (R Dias and CFontana, pers. com.). And all the South Ameri-can Progne species appear to occur in largeroosts at some time in their annual cycles(Oren 1980, Hill 1995). Further reports on thesizes and locations of the roosts of tropicalswallows and data on whether they live theirmore sedentary lives without visiting largenocturnal roosts, would be very interesting.Such data would help us understand to whatextent nocturnal roosting aggregations aretied with migratory behaviour.

    ROOSTS AND RADAR

    Roosts generally empty out a few minutesbefore sunrise the next morning, and the birdstend to leave the roost once again en masse,retracing their flight upward to even higheraltitudes before flaring out on foraging flightsin all directions (Skutch 1960). This upwardflight, followed by the near-symmetric exodusof foraging birds in all direction, makes theselarge roosts of swallows apparent in the Dop-pler weather radar now deployed through-out the United States (Russell and Gauthreaux1999, Burney 2002; Fig. 1). The resulting roostring-echoes provide the prospect of a syn-thetic view of the size and seasonal distribu-tion of swallow roosts throughout NorthAmerica (Russell and Gauthreaux 1998,Russell et al. 1998, Burney 2002). These remotemethods are being supplemented by orga-nized citizen science projects in both NorthAmerica (Purple Martin Conservation Asso-ciation 2005) and the United Kingdom (BritishTrust for Ornithology 2006) from which a greatdeal is being learned about comparative roost-ing biology.

    One of the interesting aspects of roostingbiology is that roost sites vary considerablyin how consistently they are used from yearto year. For instance, only a few coastal sites,such as the one at Icklesham, Sussex, are reli-

    Figure 1. How a flock of swallows emerging in earlymorning from their roost (R) climbs into the alti-tudes being scanned by weather radar (above),and spreading out from there in foraging flights(below) leaves a distinctive ring echo in the radarimage before it spreads out enough that the den-sity of flying swallows becomes so low as to beundetectable by the radar.

  • 88 WINKLER Hornero 21(2)

    able Barn Swallow migratory stopover sitesfrom year to year in the United Kingdom,while all others being monitored so far in thiscountry seem to move around from year toyear (Griffin et al. 2005). Similarly, a largeswallow roost seems to occur every fall in themarshes of Montezuma National WildlifeRefuge near Ithaca, New York, but its preciselocation can vary by 5 km or so from year toyear (Burney 2002, Winkler and Haber,unpublished data). As we learn more abouthirundinid roosting biology, it will be interest-ing and of considerable conservation impor-tance to try to understand just what makes asite attractive from year to year: it may actuallybe the case that unpredictability per se is anadvantage in reducing the familiarity of roostsites to local predators.

    ROOSTS AND MIGRATION

    Over much of their migratory and winter-ing ranges, hirundinid roost sites appear tohave fairly consistent roost spacing. Roost sitesin North America tend to be about 100 to150 km apart, though there are certainly con-secutive roost sites in prospective northboundor southbound directions that are farther apart(Fig. 2). Spacing at distances approximating100 km is very similar to the scale of move-ments reported for roosts of migrant speciesof the genus Hirundo in South Africa (Oatley2000). By contrast, roost sites in Britain are

    often much closer together, averaging as littleas 2530 km apart (Ormerod 1991), and thewave of Barn Swallows returning to Britainmoves about 50 km/day (Huin and Sparks1998). This suggests that migrating hirundi-nids in Britain may actually have the luxuryof hopping over some roost sites in their jour-neys. However, even in North America, thedistances between many of the roost sites arewell within the range of a days fairly leisurelyflight. Flight speeds of migrating hirundinidsare in the range of 4080 km/h (Lyuleeva1973), and even if the speeds of leisurelyforaging birds are half those, most roostspacings would appear to accommodate anunhurried itinerary from site to site.

    These observations suggest an interestingmodel for swallow migration, with birdsmigrating through the day linking every nightto the next site in their successive chain ofnocturnal roost sites. Wintering swallows arereasonably seen as classical central place for-agers, with birds radiating out from the roostsite in early morning and most often return-ing to the same roost the following evening.Thus, migration for swallows may often be asimple extension of the loops of daily centralplace foraging into line segments of approxi-mately the same length linking up to the nextroost site (Fig. 3). It would be interesting toexplore how these connections to the nextroosts are made. Individual birds may set outfrom a roost in the morning with a memory

    Figure 2. A compilation of weather radar datathroughout the southeastern United States for anearly morning in late summer. The image showsthe clear signatures of roost ring-echoes spacedfairly regularly at about 100 km throughout theregion.

    Figure 3. A simple graphical interpretation of howmovements by foraging swallows between roostsites may involve very little if any increase in thedistance flown per day (right) relative to birds re-turning to the same roost site in succeeding nights(left).

  • 2006 ROOSTS AND MIGRATIONS OF SWALLOWS 89

    of the next site on their migratory chain, orthey may simply forage more in the preferredmigratory direction and then be recruited tothe next roost site by aggregating near the endof the day with birds that used the next sitefor their own roost the night before. Thesebehavioural mechanisms of roost joiningwould also presumably illuminate the mys-tery of which sites are used from year to yearand serve as a clue to the causes of the inter-annual ebb and flow of roost size.

    HIRUNDINIDAE MIGRATORY BEHAVIOUR

    This view of swallow migrations as journeysstitched together by roost sites suggests thatthe exertion of migration on many parts oftheir journeys may not differ substantiallyfrom that expended during the daily and rou-tine foraging movements out from and backto a nocturnal central place. Thus, most move-ments of swallows appear to take place duringthe day with these fairly routine movementsfrom one roost site to the next. Yet, no matterwhat the details of how roosts and migrationare integrated, it is clear that not all parts ofall hirundinid migrations are a simple redi-rection of routine foraging patterns. There arerecords of marked Barn Swallows covering12000 km in 34 days (320 km/day) and 3028 kmin 7 days (433 km/day; Turner 2004). Theserates of movement are much greater thanthose likely to be achieved by birds makingmore leisurely movements from roost to roost.In addition, many of the longest and fastestpassages made by swallows are those madeby the three long-distance European swallowmigrants as they cross the Mediterranean andSahara. Like most passerines that make thistrip twice annually, or die trying, it seemslikely that the vast majority head south fromstaging areas in Spain or Italy (Rubolini et al.2002b), then cross directly to Africa, someprobably stopping to coast along theSaharan verge of the sea and others probablycontinuing directly to areas south of theSahara (reviewed by Moreau 1961). Hirundorustica is the most commonly seen bird cross-ing the Sahara, and, especially in spring, it isoften seen flying against, and not uncom-monly succumbing to, the northerly windsthat prevail at that season (Moreau 1961). Aselsewhere in their annual cycle, Hirundo spe-cies tend to fly lower than other hirundinids,

    probably making them more visible thanDelichon and Riparia species during passage(Lyuleeva 1973). Moreau (1961) estimated thatall the passerines making the non-stop trans-Saharan trip in spring must fly 5060 h non-stop against head-winds. Recent radarevidence (Schmaljohann et al. 2007) suggests,however, that the 21002400 km trip betweenEurope and tropical Africa is not made non-stop by most European passerines that crossthe Sahara and Mediterranean. Much moredetail is needed on the biology of these mi-grants. Nevertheless, the sudden and sporadicappearance of large numbers of all threeEuropean swallow genera at low elevationsduring migration, together with correlationsbetween the intensity of observed migrationsand local weather, suggest the hypothesis thathirundinids under very favourable winds flyout of sight high above the ground and thatthey fly nearer the ground when they arefaced with head winds or the need to refuelon insects en route. Flying insects are gener-ally more abundant near the ground (e.g.,Glick 1939, Taylor 1974), and it appears likelythat, when insects are relatively abundantover the terrain being over-flown, hirundinidsdescend to lower levels (i.e., nearer theground) to refuel, returning, once fed, tohigher elevations to cover much larger dis-tances when and where the winds arefavourable. This possible alternation betweenhigher altitude cruising and lower altitudeforaging flight could also explain: (1) whymigrating swallows often do not appear atmigration observation stations until hoursafter sunrise (Lyuleeva 1973) when they pre-sumably leave the roosts many hours earlier,and (2) why adults and juveniles appear tofly for different amounts of the day (e.g.,Gatter and Behrndt 1985). Nearly exclusivehigh altitude foraging may also explain theextreme sparseness of observations of somecommon species (e.g., Delichon urbicum;Moreau 1972) on the wintering grounds.

    J. Cobb (in Griffin et al. 2005) reported thatHirundo species lured down to nets near aroost site in the United Kingdom by playbackof swallow song in the afternoon, well beforeswallows would be coming to roost, werenever captured in the roost later in the dayand averaged about 2 g heavier than thosecaptured later in the evening at the same site.This seemed to happen after warm days, sug-

  • 90 WINKLER Hornero 21(2)

    gesting interesting environmental effects onthis behaviour; however, it also suggests notonly that high-flying birds may have greaterenergy stores but also that, even in the UnitedKingdom, there may be a diversity of migra-tory strategies in the Hirundo population andthat some birds might even be migrating atnight and making much more direct over-landprogress over the United Kingdom as a result.The generality and interpretation of this resultwould be strengthened enormously by play-back experiments well in advance of the roost-ing hour near hirundinid roost sites in otherplaces.

    Any journey of two days or more requiresboth nocturnal and diurnal migrants to aban-don their preferred hours of flight, and it isclear that hirundinids fly both day and nightduring these long migratory passages (re-viewed in Moreau 1961). It remains an openquestion whether and how much hirundinidsfly at night outside the most extreme legs oftheir journeys, but it seems safe to continueto designate them as diurnal migrants.

    In the Western Hemisphere, observations ofenormous numbers of migrant swallows pass-ing through Central America (Table 1; Brownand Brown 1999, Ruelas Inzunza et al. 2005),suggest that the bulk of North Americanswallows access wintering ranges south of theGulf of Mexico by an overland route; but thepresence of some individuals from all NorthAmerican species in Caribbean islands andmigrating over water in the Gulf confirm thatat least some individuals attempt the over-water route to the Southern Hemisphere

    (Hailman 1962, Yunick 1977, Russell 2005).Given what closely related birds are able todo in the Eastern Hemisphere, it would beinteresting and important to try to determinethe actual proportions of birds using the over-land vs. direct aquatic route over the Gulf ofMexico, especially given that north-boundbirds appear to take more varied routes overland during spring migration than fall (RuelasInzunza et al. 2005, E Ruelas Inzunza, pers.com.).

    BIOLOGY OF THE MIGRANTS

    Southbound Hirundo species in Europe canaccumulate up to about 40% of their mass asfat (Rubolini et al. 2002a) preparing for theMediterraneanSaharan journey, and theyhave different fat dynamics if heading overGibraltar vs. Italy (Rubolini et al. 2002b). Bycontrast, Hirundo species in fall appear to gainonly 12 g (

  • 2006 ROOSTS AND MIGRATIONS OF SWALLOWS 91

    actually quite heterogeneous, a careful studyof how changes in body mass and individualbody components relate to the migratorybehaviour of individual birds could be bothextremely difficult and enlightening.

    The biology of moult and how it relates tohirundinid breeding and migration appearsto be equally diverse. Among species, there isa gradient from species (such as Tachycinetabicolor; Stutchbury and Rohwer 1990) thatcomplete a full moult of body feathers, rec-trices and remiges before and during migra-tion; through populations that moult bodyfeathers as they initiate migration, stoppingto complete moult of rectrices and remigesbefore making a long-distance migration (e.g.,eastern Stelgidopteryx serripennis; Yuri andRohwer 1997); through species moulting bodyfeathers before and during migration and notinitiating or interrupting early flight feathermoult until reaching the wintering ground

    (apparently all trans-Saharan species: Ripariaand Delichon species, Cramp 1988; Hirundorustica, Cramp 1988, van den Brink et al. 2000,Rubolini et al. 2002b, Griffin et al. 2005). Onceagain, there is ample evidence of flexibility inmoult schedules within species, with WesternStelgidopteryx serripennis adopting a strategymore like Tachycineta bicolor, moulting while itmigrates without interruption, in contrast tothe moult migration of the Eastern popula-tion (Yuri and Rohwer 1997).

    WHY DO SOME MIGRATE SO MUCHFARTHER THAN OTHERS?

    Winkler (2000) suggested that there weremultiple selective reasons that have madechanges in swallow nest-type less evolution-arily flexible than other aspects of the life-his-tory, and much of the spring life history ofthese birds can be seen as adjustments to the

    Figure 4. The breeding (right diagonal hatching) and wintering (left diagonal hatching) ranges of six east-ern North American swallow species relative to the type of nest they build at their northern nesting sites.

  • 92 WINKLER Hornero 21(2)

    mode of nesting. The swallow fauna of easternNorth America is distinctive in the presenceof six obligate migrant species with diversenesting mode among them (Fig. 4). Of these,three are cavity adopters, reliant on other spe-cies or processes to create a nest-cavity inwhich they build a grass-nest cup. It seemsmore than coincidence that these three cavity-adopting species migrate the shortest dis-tances of the lot, a feature that likely allowsthem to return to the breeding grounds ear-lier in spring to compete for and secure a nest-cavity, which, unlike the burrowing andmud-nesting species, they cannot make forthemselves (see also Rubolini et al. 2005).

    HOW PHYLOGENETICALLY FLEXIBLE ISMOVEMENT IN SWALLOWS?

    An alternative hypothesis to explain themore northerly wintering of the cavity-adopters is that patterns of movement maybe nearly as difficult to evolve as are patternsof nest construction, and that all the longest-distance migrants in the New World hirun-dinid fauna (Riparia riparia, Hirundo rustica andPetrochelidon pyrrhonota; Fig. 4) are derivedfrom Old World forms (Sheldon et al. 2005)that presumably colonized the Western Hemi-sphere relatively recently. However, availableevidence in support of this phylogeneticinertia hypothesis is less than compelling.First, there are two tropical species in the NewWorld fauna, Petrochelidon fulva and Petro-chelidon rufocollaris, that appear to havederived from Petrochelidon pyrrhonota in theNew World, and both are short-distance- ornon-migrants. Second, Riparia riparia, togetherwith its short-distance congener Ripariapaludicola from the Old World, appears to besister group to the genus Tachycineta and muchmore closely related to the endemic NewWorld swallows than they are to other OldWorld forms. Thus, the short-distance or non-migratory habits of Tachycineta species andRiparia paludicola appear not to have beenconstrained by the habits of their long-dis-tance close relative Riparia riparia. Finally,Hirundo rustica has begun breeding in SouthAmerica, on what were solely its winteringgrounds, within the past 30 years (Martnez1983). Though the movement patterns of thissmall breeding population are not known, itappears extremely unlikely that these birds

    have retained the long-distance migratoryhabit of their immediate ancestors and per-form a long migration all the way to NorthAmerica in the austral winter (GH Huber, pers.com.). Hirundo rustica is known for the com-plexity of movement patterns within its largepopulations in both the Eastern and Westernhemispheres (Moreau 1972, Phillips 1986,Cramp 1988), with some resident forms inter-acting during parts of the year with both long-and short-distance forms passing through.Furthermore, Hirundo rustica is closely relatedto a large number of other species of the genusHirundo in sub-Saharan Africa, many of whichare residents or short-distance migrants (forall phylogenetic statements in this paragraphsee Sheldon et al. 2005). Finally, patterns ofmovement within the New World endemichirundinids are very flexible. Though thereare no long-distance migrants that go as faras Petrochelidon, Hirundo or Riparia species doin the New World, there is a great deal ofdiversity in distances travelled in the generaTachycineta, Progne and Stelgidopteryx. In short,the distance of movement seems to be quiteflexible in swallows (see also Turner and Rose1989), and the patterns observed above for thesummer and winter ranges of North Americanbreeding species seem more readily explainedby nest-site competition on the breedinggrounds than by phylogenetic conservatismof the traits.

    FLEXIBILITY WITHIN SPECIES INWHETHER TO MIGRATE

    There are at least two observations that sug-gest considerable flexibility in migratoryscheduling within populations of hirundinids.First, many first-year male Progne subis returnto the breeding grounds even though they donot generally breed their first year. Thissuggests that the observation of first-yearmales in the wintering grounds during thenorthern breeding season (Hill 1995) may ac-tually be the result of a strategy by theseyounger birds to forego the return to breed-ing areas until their chances of breeding arehigher (see also comment by Oren 1980). Itwould be interesting to investigate the prop-erties of those birds that stay in the south theirfirst year and compare them to those that com-plete the annual migratory circuit without at-tempting breeding on the northern end.

  • 2006 ROOSTS AND MIGRATIONS OF SWALLOWS 93

    The other exceptional indicator of flexibilityin breeding and migration is provided bythose northern migrants that have begunbreeding in the southern wintering grounds.Delichon urbicum has bred and produced fledg-lings intermittently in South Africa, withindividual colonies sometimes lasting forsome years, but no sustained population hasbeen established (McLachlan and Liversidge1957, Maclean 1988:465). By contrast, and asmentioned above, a breeding population ofHirundo rustica in Buenos Aires Province,Argentina, colonized in the early 1980s(Martnez 1983) and is still slowly growing (PPetracci and GH Huber, pers. com.). It is notyet clear whether this population is being sus-tained by local production of offspring or con-tinued recruitment from the migrant pool. Inany event, these cases of re-distribution ofmajor life events in the annual cycle suggesta flexibility in swallows that is equalled onlyby nomadic passerines of boreal forest ortropical deserts, and swallows appear to bethe only long-distance migrant passerineswith such flexibility. Just how, physiologically,this readjustment of the life history is per-formed is a fascinating area for on-goingresearch.

    IMPLICATIONS OF DIURNAL MIGRATION

    Unlike the bulk of other migrant bird spe-cies, the Hirundinidae are cosmopolitan. Fewother bird groups have achieved this breadthof distribution, and the migrations of swal-lows seem analogous with other broadlydistributed migrants, such as Apodidae,Accipitridae and Anatidae, with thrivingNorthern Hemisphere populations linked bymigration to habitats further south in thenorthern winter. All of these groups also sharethe property of being diurnal migrants, andother aerial insectivore long-distance migrants(e.g., Tyrannidae in the Western Hemisphereand Muscicapidae in the East) are nocturnaland are limited to one hemisphere or the otherin their distributions. Of course, there arenumerous insectivorous diurnal migrants thatmigrate smaller distances and do not currentlyoccur in both hemispheres (e.g., Meropidae,Artamidae). To test the possibility thatdiurnality of migration per se is associatedwith the broad-scale distributions of birds atthe family level, I categorized each avian

    family into residents and migrants. I furtherdivided the migrants into diurnal, nocturnal,or both, and recorded whether these types aredistributed solely in the Eastern or WesternHemispheres or both (a much more detailedspecies-level treatment of migration modesworld-wide is forthcoming; Farnsworth et al.,unpublished data). In characterizing distribu-tions, I ignored single species in one hemi-sphere that otherwise were endemic to theother in four families (Alaudidae, Troglodyt-idae, Timaliidae and Sylviidae). The resultinganalysis (Table 2) shows, not surprisingly, thatresidents are much more likely to be endemicto one hemisphere or the other (i.e., there arefar fewer distributed in both than expectedby chance). By contrast, diurnal migrants aresignificantly more likely to occur in both hemi-spheres. Exactly how this pattern arises isunclear, but the broad array of visual cuesavailable to diurnal migrants may lend themincreased flexibility and the ability to exploreroutes for migratory possibilities that changewith changing Earth climate and that may notbe available to a more pre-programmed navi-gationorientation system in nocturnal mi-grants. Hirundinids are clearly one of the mostflexible groups both in their migratory biol-ogy and breeding schedules, and it is perhapsa bit surprising that there are so many tropi-cal forms in this group with such limited dis-tributions (e.g., Hirundo megaensis, Tachycinetastolzmanni, Tachycineta cyaneoviridis, Tachycinetaeuchrysea, Notiochelidon pileata).

    THE WAY FORWARD:MIGRATION AND MORTALITY

    The exceptional flexibility of hirundinid life-cycles raises many interesting possibilities forresearch, many of which will become avail-able as soon as digital telemetry tags and data-loggers become small enough to deploy onswallows. These technological developmentsraise the prospect of following individualsthrough their annual cycles and relating theirpersonal histories and physiological states tothe movements that they undergo. Such tagsmay also eventually lead to a refinement ofour estimates of where in the life cycle mostswallows die and why.

    Improved information on mortality coulddramatically increase our understanding ofthe lives of these resourceful birds and give

  • 94 WINKLER Hornero 21(2)

    us powerful new knowledge to help sustaintheir populations on Earth. It appears thatmost adults die in non-breeding quarters andjuvenile survival is most affected by condi-tions on or near the breeding grounds (Szep1995, Stokke et al. 2005). The huge prepon-derance of juveniles in roosts in northern lati-tudes (Griffin et al. 2005), their variableproportions in African roosts from year to year(van den Brink et al. 2000), and the frequentobservation of dead and dying Hirundo rusticaduring trans-Saharan migration (Moreau1961, 1972) all indicate that mortality can beswift and severe at times.

    Understanding the causes of mortality inthese long-distance migrants is a fascinatingand important challenge, but it may also beextremely interesting to apply this increasedknowledge to interpretation of patterns ofseasonal movement. If gathering the traces ofroost ring-echoes from weather radar fromacross North America can be automated, thereis the prospect of understanding the changingdistributions of a very large fraction of theworlds population of at least Tachycinetabicolor through the annual cycle. When view-ing the changing distributions and sizes ofthese roosts, it is tempting to interpret the

    Table 2. Summary of the distributions of families of birds of the world as they relate to the migratorymodes of their members. For each cell, the number of families is followed (in parentheses) by the devia-tion of the observed cell frequency from that expected from the marginal totals. Global analysis:26 = 66.091, P < 0.001. The two cells that contribute the most to the observed

    2 are indicated by anasterisk. Four families with unknown major migratory mode (Acanthizidae, Eopsaltridae/Petroicidae,Chionididae and Thinocoridae) were left out of this analysis.

    Hemisphere of distribution

    Migratory mode Eastern Both Western Total

    Diurnal 8 (-2.175) a 28 (4.099 *) e 3 (-1.978) i 39 Nocturnal 11(-1.268) b 17 (1.284) f 9 (0.191) j 37 Both 4 (-1.247) c 11 (2.203) g 2 (-0.955) k 17 Resident 57 (2.772) d 6 (-4.454 *) h 28 (1.586) l 91 Total 80 62 42 184 a Alaudidae (one species Holarctic), Anseranatidae, Artamidae, Coliidae, Meliphagidae, Meropidae, Ploceidae, Sturnidae. b Campephagidae, Cisticolidae, Monarchidae, Muscicapidae, Oriolidae, Pittidae, Prunellidae, Pycnonotidae, Remizidae,

    Sylviidae, Upupidae. c Coraciidae, Dicruridae, Glareolidae, Zosteropidae. d Acanthisittidae, Aegithinidae, Apterygidae, Atrichornithidae, Balaenicipitidae, Batrachostomidae, Brachypteraciidae,

    Bucerotidae, Callaeatidae, Casuariidae, Cinclosomatidae, Climacteridae, Corcoracidae, Cracticidae, Drepanididae, Dromadidae, Estrildidae, Eurylaimidae, Grallinidae, Hemiprocnidae, Hypocoliidae, Ibidorhynchidae, Indicatoridae, Irenidae, Leptosomatidae, Lybiidae, Maluridae, Megalaimidae, Megapodiidae, Melanocharitidae, Menuridae, Mesoenatidae, Musophagidae, Nectariniidae , Numididae, Orthonychidae, Otididae, Pachycephalidae, Paradisaeidae, Paramythiidae, Pardalotidae, Passeridae, Pedionomidae, Philepittidae, Phoeniculidae, Picathartidae, Podargidae, Pomatostomidae, Pteroclididae, Ptilonorhynchidae, Rhynochetidae, Sagittariidae, Scopidae, Struthionidae, Timaliidae, Turnicidae, Vangidae.

    e Accipitridae, Alcidae, Anhingidae, Apodidae, Bombycillidae, Ciconiidae, Corvidae, Dendrocygnidae, Diomedeidae, Falconidae, Fringillidae, Gaviidae, Hirundinidae, Hydrobatidae, Laridae, Motacillidae, Pandionidae, Pelecanidae, Pelecanoididae, Phaethontidae, Phalacrocoracidae, Phoenicopteridae, Procellariidae, Psittacidae, Spheniscidae, Stercorariidae, Sternidae, Sulidae.

    f Burhinidae, Caprimulgidae, Certhiidae, Charadriidae, Cinclidae, Cuculidae, Haematopodidae, Jacanidae, Picidae, Podicipedidae, Rallidae, Recurvirostridae, Regulidae, Rostratulidae, Sittidae, Strigidae, Tytonidae.

    g Alcedinidae, Anatidae, Ardeidae, Columbidae, Emberizidae, Gruidae, Laniidae, Paridae, Scolopacidae, Threskiornithidae, Turdidae.

    h Aegithalidae, Aegothelidae, Fregatidae, Heliornithidae, Phasianidae, Trogonidae. i Cathartidae, Cotingidae, Trochilidae. j Cardinalidae, Mimidae, Parulidae, Pluvianellidae, Polioptilidae, Thraupidae, Troglodytidae, Tyrannidae, Vireonidae. k Icteridae, Rynchopidae. l Anhimidae, Aramidae, Bucconidae, Capitonidae, Cariamidae, Conopophagidae, Cracidae, Dendrocolaptidae, Dulidae,

    Eurypygidae, Formicariidae, Furnariidae, Galbulidae, Momotidae, Nyctibiidae, Odontophoridae, Opisthocomidae, Oxyruncidae, Phytotomidae, Pipridae, Psophiidae, Ramphastidae, Rheidae, Rhinocryptidae, Steatornithidae, Thamnophilidae, Tinamidae, Todidae.

  • 2006 ROOSTS AND MIGRATIONS OF SWALLOWS 95

    changes in distributions as representing themovements of birds up and down the northAtlantic coast. But the survival rates ofTachycineta bicolor at our Ithaca study site are,if anything, lower than those of Hirundo rustica(unpublished data). Thus, if we attribute shiftsin distribution to movement, we may forgetthat a large number of the birds detected inthe radars will not be returning to the breed-ing grounds. Just as dispersal from the natalgrounds confounds estimates of survival,movement and mortality may well be con-founded in the non-breeding season as well.When large roosts on the north Atlantic coastdisappear from the radar in autumn, some ofthe animals have no doubt migrated to thesouth, but a large proportion of them mayhave died instead. Short-distance facultativemigrants retain the possibility of getting backto the breeding grounds earlier, but the mor-tality cost of doing so may be larger than theextreme costs incurred by their cousinsmigrating longer distances.

    ACKNOWLEDGEMENTS

    Many thanks to Ernesto Ruelas Inzunza, CurtBurney, Eduardo Iigo-Elias, Rafael Dias and CarlaFontana for sharing unpublished data with me andaiding their interpretation; to Andrew Farnsworthfor sharing his widespread knowledge of themigratory habits of the worlds birds; to AndrewFarnsworth, Jamie Mandel, Gernot Huber, andDave Cerasale for providing me with lots ofinsights and perspectives on bird migration; toTheunis Piersma and Andrew Farnsworth forvaluable comments on the manuscript; and toBenjamin Shattuck and Noah Hamm for help withthe figures. This material is based upon work sup-ported by the National Science Foundation undergrants No. IBN-013437 and 9207231, and a Coop-erative Agreement with the National Center forEnvironmental Assessment, EPA (CR 829374010).

    LITERATURE CITED

    ARDIA DR, SCHAT KA AND WINKLER DW (2003) Repro-ductive effort reduces long-term immune functionin breeding tree swallows (Tachycineta bicolor).Proceedings of the Royal Society of London B270:16791683

    BELTON W (1985) Birds of Rio Grande do Sul, Brazil.Part 2. Formicariidae through Corvidae. Bulletin ofthe American Museum of Natural History 180:1241

    BIJLSMA RG AND VAN DEN BRINK B (2005) A BarnSwallow Hirundo rustica roost under attack: timingand risks in the presence of African Hobbies Falcocuvieri. Ardea 93:3748

    VAN DEN BRINK B, BIJLSMA RG AND VAN DER HAVE TM(2000) European Swallows Hirundo rustica inBotswana during three non-breeding seasons: theeffects of rainfall on moult. Ostrich 71:198204

    BRITISH TRUST FOR ORNITHOLOGY (2006) SwallowRoost Project. (URL: http://www.bto.org/ringing/swallowroostproject/index.htm)

    BROWN CR AND BROWN MB (1996) Coloniality in theCliff Swallow. University of Chicago Press, Chicago

    BROWN CR AND BROWN MB (1999) Barn Swallow(Hirundo rustica). Pp. 132 in: POOLE A AND GILL F(eds) The birds of North America. Academy of Natu-ral Sciences and American Ornithologists Union,Philadelphia and Washington DC

    BRYANT DM AND TURNER AK (1982) Central placeforaging by swallows (Hirundinidae): the questionof load size. Animal Behaviour 30:845856

    BURNESS GP, YDENBERG RC AND HOCHACHKA PW (2001)Physiological and biochemical correlates of broodsize and energy expenditure in tree swallows.Journal of Experimental Biology 204:14911501

    BURNEY CW (2002) A study of swallow roosts found inthe Eastern United States. MSc thesis. Cornell Uni-versity, Ithaca

    CRAMP S (1988) The Birds of the Western Palearctic.Oxford University Press, Oxford

    CZARNOWSKI MR, LOMBARDO MP, POWER HW ANDHASSELQUIST D (2004) Sudden energetic deficit inmale tree swallows (Tachycineta bicolor) has negli-gible effect on immunocompetence and gluco-corticosteroid levels. Integrative and ComparativeBiology 44:543543

    GATTER VW AND BEHRNDT M (1985) Unterschiedlichetageszeitliche Zugmuster alter und junger Vogel amBeispiel der Rauchschwalbe (Hirundo rustica).Vogelwarte 33:115120

    GLICK P (1939) The distribution of insects, spiders andmites in the air. United States Department of Agricul-ture Technical Bulletin 673:1150

    GRIFFIN B, CLARK J AND ROBINSON RE (2005) SwallowRoost Project Newsletter. British Trust for Ornithology,Thetford

    HAILMAN JP (1962) Direct evidence for trans-Caribbean migratory flights of swallows anddragonflies. American Midland Naturalist68:430433

    HASSELQUIST D, WASSON MF AND WINKLER DW (2001)Humoral immunocompetence correlates with dateof egg-laying and reflects work load in female treeswallows. Behavioral Ecology 12:9397

    HILL JR III (1995) Thousands of martins killed by hailin Brazilian tornado. Purple Martin Update 6:23

    HOOGLAND JL AND SHERMAN PW (1976) Advantagesand disadvantages of Bank Swallows (Riparia riparia)coloniality. Ecological Monographs 46:3358

    HUIN N AND SPARKS TH (1998) Arrival and progres-sion of the Swallow Hirundo rustica through Britain.Bird Study 45:361370

  • 96 WINKLER Hornero 21(2)

    LOMBARDO MP, GREEN ML, THORPE PA, CZARNOWSKIMR AND POWER HW (2004) Repeated samplingaffects Tree Swallow semen characteristics. Journalof Field Ornithology 75:394403

    LYULEEVA DS (1973) Features of swallow biologyduring migration. Pp. 219272 in: BYKHOVSKII BE (ed)Bird migrations: ecological and physiological factors. JohnWiley & Sons, New York

    MACLEAN GL (1988) Roberts birds of southern Africa.New Holland, London

    MARTNEZ MM (1983) Nidificacin de Hirundo rusticaerythrogaster (Boddaert) en la Argentina (Aves,Hirundinidae). Neotropica 29:8386

    MCCARTY JP AND WINKLER DW (1999) Foraging ecol-ogy and diet selectivity of tree swallows feedingnestlings. Condor 101:246254

    MCLACHLAN GR AND LIVERSIDGE R (1957) Roberts birds ofSouth Africa. South African Bird Book Fund, Cape Town

    MEAD CJ AND HARRISON JD (1979) Sand Martin move-ments within Britain and Ireland. Bird Study 26:7386

    MLLER AP (1987) The advantages and disadvantagesof coloniality in the swallow Hirundo rustica. AnimalBehaviour 35:819832

    MOREAU RE (1961) Problems of Mediterranean-Saharan migration. Ibis 103a:373427, 103a:580623

    MOREAU RE (1972) The Palearctic-African bird migrationsystems. Academic Press, London

    OATLEY TB (2000) Migrant European SwallowsHirundo rustica in southern Africa: a southern pers-pective. Ostrich 71:205209

    OREN DC (1980) Enormous concentration of martins(Progne spp.) in Iquitos, Peru. Condor 82:344345

    ORMEROD SJ (1991) Pre-migratory and migratorymovements of Swallows Hirundo rustica in Britainand Ireland. Bird Study 38:170178

    PHILLIPS AR (1986) The known birds of North and MiddleAmerica. Part I. Private edition, Denver

    PILASTRO A AND MAGNANI A (1997) Weather conditionsand fat accumulation dynamics in pre-migratoryroosting Barn Swallows Hirundo rustica. Journal ofAvian Biology 28:338344

    PURPLE MARTIN CONSERVATION ASSOCIATION (2005)Project Martin Roost (URL: http://www.purplemartin.org/roost/)

    RUBOLINI D, MASSI A AND SPINA F (2002a) Replacementof body feathers is associated with low premigra-tory energy stores in a long-distance migratory bird,the Barn Swallow (Hirundo rustica). Journal of Zoology258:441447

    RUBOLINI D, PASTOR AG, PILASTRO A AND SPINA F(2002b) Ecological barriers shaping fuel stores inBarn Swallows Hirundo rustica following the cen-tral and western Mediterranean flyways. Journal ofAvian Biology 33:1522

    RUBOLINI D, SPINA F AND SAINO N (2005) Correlates oftiming of spring migration in birds: a comparativestudy of trans-Saharan migrants. Biological Journalof the Linnean Society 85:199210

    RUELAS INZUNZA E, HOFFMAN SW AND GOODRICH LJ(2005) Stopover ecology of Neotropical migrants incentral Veracruz, Mexico. Pp. 657672 in: RALPH CJAND RICH TD (eds) Bird conservation implementationand integration in the Americas. General TechnicalReport PSW-GTR-191, Pacific Southwest ResearchStation, USDA Forest Service, Albany

    RUSSELL KR AND GAUTHREAUX SA (1998) Use ofweather radar to characterize movements of roost-ing purple martins. Wildlife Society Bulletin 26:516

    RUSSELL KR AND GAUTHREAUX SA (1999) Spatial andtemporal dynamics of a Purple Martin pre-migra-tory roost. Wilson Bulletin 111:354362

    RUSSELL KR, MIZRAHI DS AND GAUTHREAUX SA JR(1998a) Large-scale mapping of Purple Martinpremigratory roosts using WSR-88D weathersurveillance radar. Journal of Field Ornithology69:316325, 69:509

    RUSSELL RW (2005) Interactions between migrating birdsand offshore oil and gas platforms in the northern Gulf ofMexico. Final Report. OCS Study MMS 2005-009, Gulfof Mexico OCS Region, Minerals ManagementService, United States Department of the Interior,New Orleans

    SAFRAN RJ (2004) Adaptive site selection rules andvariation in group size of Barn Swallows: individualdecisions predict population patterns. AmericanNaturalist 164:121131

    SAFRAN RJ AND MCGRAW KJ (2004) Plumage colora-tion, not length or symmetry of tail-streamers, is asexually selected trait in North American barnswallows. Behavioral Ecology 15:455461

    SAFRAN RJ, NEUMAN CR, MCGRAW KJ AND LOVETTE IJ(2005) Dynamic paternity allocation as a functionof male plumage color in barn swallows. Science309:22102212

    SCHMALJOHANN H, LIECHTI F AND BRUDERER B (2007)Songbird migration across the Sahara: the non-stophypothesis rejected! Proceedings of the Royal Societyof London B 274:735739

    SHELDON FH, WHITTINGHAM LA, MOYLE RG, SLIKASB AND WINKLER DW (2005) Phylogeny of swallows(Aves: Hirundinidae) estimated from nuclear andmitochondrial DNA sequences. MolecularPhylogenetics and Evolution 35:254270

    SKUTCH AF (1960) Life histories of Central Americanbirds, II. Family Hirundinidae. Pacific Coast Avifauna34:265286

    STOKKE BG, MLLER AP, SAETHER BE, RHEINWALD GAND GUTSCHER H (2005) Weather in the breedingarea and during migration affects the demogra-phy of a small long-distance passerine migrant.Auk 122:637647

    STUTCHBURY BJ AND ROHWER S (1990) Molt patterns inthe Tree Swallow (Tachycineta bicolor). CanadianJournal of Zoology 68:14681472

    SZEP T (1995) Survival rates of Hungarian sandmartins and their relationship with Sahel rainfall.Journal of Applied Statistics 22:891904

  • 2006 ROOSTS AND MIGRATIONS OF SWALLOWS 97

    TAYLOR LR (1974) Insect migration, flight periodicity andboundary-layer. Journal of Animal Ecology 43:225238

    TURNER A AND ROSE C (1989) A handbook to the swallowsand martins of the world. Christopher Helm, London

    TURNER AK (1982) Optimal foraging by the Swallow(Hirundo rustica, L). Prey size selection. AnimalBehaviour 30:862872

    TURNER AK (2004) Family Hirundinidae (swallowsand martins). Pp. 602685 in: DEL HOYO J, ELLIOTT AAND CHRISTIE DA (eds) Handbook of the birds of theworld. Volume 9. Cotingas to pipits and wagtails. LynxEdicions, Barcelona

    WINKLER DW (2000) The phylogenetic approach toavian life histories: an important complement towithin-population studies. Condor 102:5259

    WINKLER DW AND ALLEN PE (1996) The seasonaldecline in tree swallow clutch size: physiologicalconstraint or strategic adjustment? Ecology77:922932

    YUNICK RP (1977) A Caribbean Barn Swallowrecovery. Auk 94:149150

    YURI T AND ROHWER S (1997) Molt and migration inthe Northern Rough-winged Swallow. Auk114:249262