The Phenology of Molting, Breeding and Their Overlap in Central Amazonian Birds

Journal of Avian Biology 43: 141–154, 2012 doi: 10.1111/j.1600-048X.2011.05574.x © 2012 Th e Authors. Journal of Avian Biology © 2012 Nordic Society Oikos Subject Editor: Wolfgang Goymann. Accepted 28 December 2011

The phenology of molting, breeding and their overlap in central Amazonian birds

Erik I. Johnson , Philip C. Stouffer and Richard O. Bierregaard, Jr

E. I. Johnson ([email protected]) and P. C. Stouﬀ er, School of Ren. Nat. Res., LSU AgCenter and Louisiana State Univ., Baton Rouge, LA 70803, USA. EIJ also at: Nat. Audubon Soc., 6160 Perkins Rd., suite 135, Baton Rouge, LA 70808, USA. – R. O. Bierregaard, Jr, Dept of Biol., Univ. of North Carolina, Charlotte, NC 28223, USA.

Th e energetically challenging periods of molting and breeding are usually temporally separated in temperate birds, but can occur simultaneously in tropical birds, a condition known as molt– breeding overlap. Here, we document great variation in the timing and duration of molting and breeding, and in the extent of molt – breeding overlap, among 87 species of understory passerines in central Amazonia. We analyzed molt and breeding from 26 871 birds captured over a 30-yr period near Manaus, Brazil. Although most species typically bred during the late dry season (about October through January), many thamnophilids apparently bred year-round, whereas a few other species from a variety of families bred mainly during the wet season (about January through May). Of all breeding birds with an active brood patch, 12.7% were simultaneously molting. Molt – breeding overlap was more frequently observed among suboscines (13.3%), especially thamnophilids (23.0%), than oscines (6.4%). Some families had Ͻ 5% molt – breeding overlap frequency, including Tyrannidae (4.4%), Tityridae (0.0%), Pipridae (1.5%), Turdidae (0.0%), and Th raupidae (0.0%), indicating that not all tropical species exhibit molt– breeding overlap. Among 31 well-sampled species (n Ն 15 brood patches), variation in molt– breeding overlap frequency was positively correlated with each species’ average duration of fl ight feather replacement (range 98– 301 d). We also measured feather growth rates of individual birds in nine species; in fi ve of these, slower-growing feathers increased with an individual’ s probability of having molt– breeding overlap. Among furnariids, molt– breeding overlap occurred either at the beginning or end of the molt cycle, suggesting that physiological mechanisms typically separate molting from breeding. Th amnophilids showed a much diff erent pattern; molt – breeding overlap occurred at any stage of feather replacement, apparently not regulated to be independent of breeding. Th ese results reveal substantial life-history variation among Amazonian birds. Future work to resolve the physiological regulation of molting and breeding in tropical birds will greatly contribute to understanding these patterns and their relevance to avian diversity.

Annual life history events in birds that demand especially known about the timing of these events in tropical commu- high energy expenditure include breeding, molting, and nities, especially in contrast to temperate communities (Pyle migration, and these are typically separated temporally to 1997, Dawson 2008). minimize their overlap (Drent and Daan 1980, Kjell é n For tropical species that do not migrate, temporal 1994, Murphy 1996). Because energy available to birds is constraints on molting and breeding should be reduced. limiting, trade-off s between successful breeding, molting, Paradoxically, individuals of many tropical species are and migration are often apparent. For example, high repro- thought to regularly molt and breed simultaneously, referred ductive investment can delay molt, decrease feather quality, to as molt – breeding overlap (Snow and Snow 1964, Foster and reduce parental survival (Siikamä ki et al. 1994, Nilsson 1975, Avery 1985, Astheimer and Buttemer 1999, Marini and Svensson 1996, Dawson et al. 2000). Conversely, molt- and Durã es 2001). It is not clear what drives increased ing early can reduce fecundity, but increase parental survival molt – breeding overlap in some tropical birds, but reduced (Morales et al. 2007). Furthermore, costs associated with physiological demands due to a slower-paced lifestyle molting are believed to prevent overlap with migration; some appears to correlate with its occurrence (Foster 1974, migratory species even temporarily arrest molt until migra- Franklin et al. 1999, Wingfi eld 2005). Tropical birds tion is completed (Stresemann and Stresemann 1966, Pyle typically lay fewer eggs per clutch (usually just two; 1997, Leu and Th ompson 2002, P é rez and Hobson 2006). Skutch 1969, Kulesza 1990, Young 1994, Martin et al. Th us, the timing of these life-history events is presumably 2000), have reduced maximum gonad size and hormonal subject to strong evolutionary pressures and is optimized concentrations (Stutchbury and Morton 2001, Wikelski through natural selection to maximize fi tness (Dawson et al. et al. 2003a, Goymann et al. 2004, Hau et al. 2010), 2000, Ricklefs and Wikelski 2002, Moreno 2004). Little is a lower metabolic rate (Ricklefs 1976, Weathers 1979,

141 Wikelski et al. 2003b, Jetz et al. 2008), and a prolonged Methods molt (Helm and Gwinner 1999, Ryder and Wolfe 2009). High nest predation rates resulting in multiple nesting Study site attempts (Skutch 1949, Kulesza 1990, Ferretti et al. 2005, Roper 2005, but see Snow and Snow 1963, Oniki 1979, We conducted our study at the Biological Dynamics of Martin 1995, Robinson et al. 2000) coupled with decreased Forest Fragments Project (BDFFP), which is located seasonality and more constant resource availability (Lack about 80 km north of Manaus, Brazil (2° 30 ′ S, 60 ° W). Th is 1947, Ashmole 1963, Cody 1966, Ricklefs 1980, Hahn landscape contains largely undisturbed terra fi rme lowland et al. 1992, Martin 1996) may at least partially explain tropical rainforest, especially to the north of our study site. this reduced-paced life style with extended molting Th e forest canopy is 30– 37 m tall with emergents reaching and breeding seasons. Under these circumstances, molt – 55 m. Th e understory is relatively open and is dominated by breeding overlap may not severely reduce fecundity or palms. Annual rainfall is approximately 2500 mm (ranging fi tness in tropical species compared to temperate species from 2000 to 3500 mm) with a dry season typically last- (Foster 1974, Slagsvold and Dale 1996, Hemborg and ing from June to November (Stouff er and Bierregaard 1993, Lundberg 1998, Hemborg 1999, Hemborg et al. 2001). Laurance 2001). Th e site has some topography, especially Tropical forest birds show great variation in foraging near streams, and varies in elevation from 50 to 100 m. Soils strategy, social system, and microclimate use (Karr 1971, are generally nutrient-poor sandy or clay-rich ferrasols, typi- Terborgh et al. 1990, Stutchbury and Morton 2001). Th e cal of the region (Sombroek 2000). humid lowland rainforests of the central Amazon contain a Th e BDFFP is well-known for its studies of forest frag- diverse avian community with great variation in life history mentation (Bierregaard et al. 2001) including considerable strategies, providing an ideal opportunity to study within- eff orts to document the natural avian community in mature community variation in molting and breeding phenology undisturbed forest (Cohn-Haft et al. 1997, Johnson et al. (Cohn-Haft et al. 1997, Johnson et al. 2011). For example, 2011), but also includes extensive monitoring and research in a single 100-ha patch of undisturbed central Amazonian at intact continuous forest that act as control sites. Here we forest, over 200 species have been found (Johnson et al. use a long-term mist-netting database of this research from 2011). Such diversity in life history strategies may also result 1979 through 2009 that includes over 60 000 captures and in variation in molt– breeding overlap frequency among spe- 324 860 mist-net hours. Th is database has been accrued from cies, but few large datasets (Poulin et al. 1992, Marini and 11 forest fragments, including 1/4 to 10 yr of pre-isolation Durã es 2001, Wolfe et al. 2009) are available to examine sampling (median 1 yr; Bierregaard et al. 2001), as well as the variation of molting and breeding phenologies within a 34 continuous forest sites. Mist-nets were typically open tropical community. from 08:00 to 14:00 and the number of 12 ϫ 2-m mist In this paper, we present the phenology of breeding nets used at each site was dependent on plot or fragment size using brood patch data, molting using primary feather molt with eight mist nets in 1-ha plots, 16 nets in 10-ha plots, data, and their overlap in a near-equatorial community and 48 nets in 100-ha plots. Beginning in 1991, an addi- of understory forest birds in Amazonian Brazil, as well as tional 16 nets were placed around the borders of fragments explore ecological factors that covary with these patterns. and interior forest plots. Fragments were sampled year- We fi rst predicted that the frequency of molt – breeding round 1 – 16 times (median 5 times) per year in 1979 – 1992, overlap among species would vary among species accord- and 2000, and during the dry seasons of 2007 and 2009. ing to their ecology and taxonomy if this trait was adap- Each continuous forest site has been sampled less con- tive. We then examined correlates of molt – breeding overlap sistently than the fragments, but collectively account for frequencies with the expectation that increased frequency 62% of mist-net hours and 67% of all captures. Although of molt– breeding overlap would correlate with longer molt some uncertainty remains about how fragmentation poten- duration and decreased fl exibility in timing of molt initia- tially impacts the phenology of molting, breeding, and tion. We also predicted that individual birds would increase their overlap (Results), the majority of our data are derived their probability of having molt– breeding overlap when from continuous forest sites. Beginning in 1991 we col- they experienced relatively slow molts. Finally, for two bird lected a rectrix from each bird. Taxonomy follows Gill and families with frequent molt– breeding overlap and large Donsker (2011). sample sizes, we examined when molt – breeding overlap occurs in the molt cycle, providing insight into how molt Assessing breeding, molting, and their overlap and breeding may be regulated. Given that brood patches for single-brooded birds may last no longer than four Since 1979, molt status was determined by the occurrence weeks and that molt often follows breeding (Jones 1971, (yes– no) of primary feather molt, while molt extent was del Hoyo 1992– 2010, Pyle 1997), we predicted molt– determined by the number of the most recently molted breeding overlap to occur only at the beginning of the primary feather (primary 1 to primary 10). Breeding status molt cycle (i.e. when molting primary 1) if a physiologi- of captured birds, defi ned as the presence or absence of an cal mechanism was activated to defer the initiation of molt active brood patch, indicated by a featherless oedematous until after breeding. Alternatively, if molt– breeding over- hyper-vascularized belly region, was assessed since 1982. lap were to occur haphazardly throughout the molt cycle, Th e presence of a vascularized brood patch is indicative of we would instead conclude that molt and breeding are incubation, and typically disappears during the nestling physiologically decoupled, in contrast to conclusions drawn stage (Jones 1971). We excluded captures with asymmet- from many studies of temperate species. rical molt, defi ned as only one wing undergoing primary

142 molt, and molting and breeding records for within-month Frequency of molt – breeding overlap recaptures. We considered molt– breeding overlap to occur when We quantifi ed the proportion of captures with molt – molting primary feathers on the wing occurred simultane- breeding overlap among all 87 passerines by species, sub- ously with an active brood patch. Although other metrics family, family, and suborder. We did not examine non- have been used in the literature to assess molt – breeding over- passerines because they collectively represent few captures lap, including gonad size, body molt, and molt overlapping and because some have ambiguous (e.g. Trochilidae and fl edging (Foster 1975, Nilsson and Svensson 1996, Marini Columbidae) or unknown brood patch development. We and Dur ã es 2001), our assessment is among the most conser- compared molt – breeding overlap frequencies at family- and vative and least subjective. We defi ne the frequency of molt– suborder-levels, and among ecological guilds (Stouff er et al. breeding overlap as the proportion of all breeding birds that 2006) using Chi-square contingency tests (Proq Freq, SAS are simultaneously molting. Th is, by defi nition, eliminates ver. 9.1). age and sex classes that do not develop brood patches from the study sample. Correlates of molt– breeding overlap Ն Taxonomic groupings and guild classifi cation Focusing on 31 species with 15 observed brood patches, we asked four questions to understand correlates in the To test whether there is a phylogenetic component to molt – frequency of molt – breeding overlap among study species. breeding overlap, we grouped passerines into three major categories, the suboscines (suborder Tyranni), 10-primaried 1) Does molt duration affect molt– breeding oscines (suborder Passeri, Muscicapoidea and Corvoidea overlap frequency? [Vireonidae only]), and 9-primaried oscines (suborder We estimated molt duration (MD; the average length of Passeri, Passeroidea; Cracraft and Barker 2009). Although time an individual takes to molt for each species) by examin- 9-primaried oscines are more correctly referred to as oscines ing recaptured individuals as they progressed through their with nine visible primaries and a clear phylogenetic bound- wing molt. By noting which feather was actively molting ary with oscines with 10 visible primaries (Hall 2005), our when captured and dividing the number of feathers molted groupings here are used for convenience as they refl ect three by the number of days between captures, we could therefore distinct phylogenetic groups among our study species. estimate molt duration for each individual. We then averaged Species were also categorized into one of eight ecologi- molt duration among individuals to estimate molt duration cal groups, or guilds, according to their diet and mode of for each species. Th ese data allowed us to test the hypothesis foraging (Stouff er and Bierregaard 1995, Stouff er and that longer average molt duration increases the frequency Borges 2001, Stouff er et al. 2006, Johnson et al. 2011). All of molt – breeding overlap using a simple linear regression. non-forest species were categorized together based on their A potential source of bias with our estimates is that the previously identifi ed dependence on second growth at the primary feather molt duration is not constant throughout study site (Stouff er and Bierregaard 1995). Frugivores were the molting process and slows down as molt completes species that regularly consumed fruit. Insectivores were split because of the increased length of outer primary feathers into six remaining categories: obligate ant-followers (those (Pimm 1976, Summers et al. 1983, Underhill and Zucchini that depended on Eciton army-ant swarms), gap special- 1988, Dawson and Newton 2004), but we expect this to ists (those that were dependent on tree fall gaps), terrestrial have minimal impact on our conclusions. Specifi cally, by (those that foraged by walking or hopping on the ground), using recaptured birds to estimate molt duration, some birds fl ock-specialists (those that spent most of their time among will be equally likely to be early in the molt sequence (with Th amnomanes -led mixed-species fl ocks), fl ock dropouts relatively fast molt progression) as late in the molt sequence (those that often associated with mixed-species fl ocks, but (with relatively slow molt progression), with the most fre- were could establish territories in the absence of these fl ocks), quent estimate of molt duration centering on the middle and other insectivores (those that did not fall into the above of the molt sequence. Further, because our study species fi ve insectivore guilds, but collectively were found in the are non-migratory understory species, the diff erence in forest understory). feather length between the inner- and outer-most primaries We tested the diff erence in molt– breeding overlap fre- is typically relatively small (Th é ry 1997, Lockwood et al. quency among the three taxonomic groups, between sub- 1998, Claramunt et al. 2012). Finally, our use of this metric oscines and oscines, among families within each major is primarily to compare the duration among species, thus taxonomic group, and among ecological guilds using if there is bias in this estimate, it should be similar among Chi-square contingency tests. species.

Molting and breeding phenology 2) Can birds adjust the timing of molt to minimize molt – breeding overlap? We assessed the frequency of captures either breeding or We assessed the variability in the timing of molt initiation molting for each month, pooling across all years and study using the same birds as the previous analysis, but extrapolat- plots. Although we suspected some annual variation in the ing back to estimate the date of wing molt initiation. We timing of molting and breeding, sample sizes for any given also estimated the length of the molting season (MS), species in any given month and year were relatively low, with which we deﬁ ne as the number of months that Ն 16% some years more eﬀ ectively sampled than others. of captures were molting, and the length of the breeding

143 season, defi ned as the number of months that Ն 4% of brood patches according to the extent of primary feather captures were breeding. Th ese molting and breeding season molt (primary 1 to primary 10). If molt extent diff ered metrics indicate seasonality at the population level, whereas on each wing, we added a ‘ half ’ bird to each feather molt molt duration off ers the duration of an individual molt category. Th is tests the hypothesis that molt– breeding within this season (we do not have data to show breeding overlap would predominantly only at the beginning of the duration on the individual level). We then explored how molt cycle (i.e. when molting primary 1) if a physiological variation in molt initiation date, the ratio of molt duration mechanism attempts to defer the initiation of molt until to length of the molting season (MD/MS), and the over- after breeding. Alternatively, if the data reveal that breed- lap over the breeding and molting seasons predicted the ing occurs haphazardly throughout the molt cycle, we would frequency of molt– breeding overlap using generalized linear instead conclude that molt and breeding are physiologically models (simple linear regression and 1-factor ANOVA). decoupled, in contrast to conclusions drawn from many studies of tem perate species. 3) Do feather growth rates of individual birds predict Percentage data were arcsine-transformed and other their probability of exhibiting molt– breeding overlap? variables were log-transformed to meet assumptions of para- For birds with brood patches for which we also collected metric statistics where necessary. We present least-square an outer rectrix, we measured the rate of feather growth means and standard errors unless otherwise stated. and multiplied by the length of the feather to determine the number of days it took to completely grow an outer tail feather. We measured rectrix growth rate by measuring Results the distance between at least six consecutive visible growth bars, each of which are perpendicular to the rachis and indi- Molting and breeding phenology cate 24 h of growth (Michener and Michener 1938, Jovani et al. 2011). We then divided this measurement by the Pooling across all 87 species of passerines, breeding occurred number of growth bars to get an average daily growth rate, year-round at our study sites near Manaus, Brazil, but peaked known as ptilochronology (Grubb 1989). We standard- in the late dry season and early wet season (September – ized feather growth across species by constructing residu- January; Fig. 1). Th e frequency of molt began to increase in als around each species ’ mean feather growth rate and then September and peaked in December to March. Th e propor- pooled residual growth rates at the guild level for a logis- tion of birds molting while breeding (i.e. exhibiting molt – tic regression analysis using nine species, each with at least breeding overlap) among all captures was Ͻ 2% in every 10 feather growth rate samples. Th ese data test whether month and did not correlate with monthly brood patch individual birds that molted feathers more slowly had an frequency (Pearson ’ s r ϭ 0.24, p ϭ 0.46; Fig. 1). However, increased probability of molt – breeding overlap. when brood patch (i.e. breeding) frequency increased, pro- portionally fewer birds were simultaneously molting and 4) When in the molt sequence does breeding overlap? breeding (Pearson ’ s r ϭ Ϫ0.65, p ϭ 0.022). In other words, For species in the Th amnophilidae, Formicariidae, and molt – breeding overlap appeared to be present at low fre- Furnariidae, we counted the number of individuals with quencies throughout the year, but most breeding birds

Figure 1. Th e frequency of captures with active brood patches (open circles), wing molt (dots), and molt– breeding overlap (MBO; bars) pooled across all 87 study species by month near Manaus, Brazil. Months in bold (June– December) indicate the dry season. See Table 1 and 2 for seasonal patterns of 31 common species.

144 apparently avoided simultaneously molting and breeding of molting frequency often followed peaks of breeding during the peak of the breeding season. (Table 1, 2), but many species had individuals molting in Variation was considerable in the breeding and molt- every month of the year, especially within Th amnophilidae, ing phenologies among 31 frequently captured species but also some Furnariidae. Oscines typically had more with Ն 15 individuals captured with active brood patches strongly seasonal breeding and molting seasonality. (Table 1, 2). For many species, the breeding season was strongly seasonal and most pronounced during the dry Molt – breeding overlap by taxonomy and ecological season from June or July through January (e.g. Certhiasomus guild stictolaemus , Xiphorhynchus pardalotus , and Myiobius barbatus ), whereas for other species breeding was occasionally Our dataset includes 26 871 records of 87 species of extended into March or April, well into the wet season (e.g. Passeriformes from 1979 to 2009 for which the presence – Dendrocincla merula and Pipra pipra ). Breeding seasons for absence of wing molt and brood patch was assessed (Sup- some other species peaked during the wet season, between plementary material Appendix 1). Most of these records December and June (e.g. Formicarius colma , Sclerurus were for suboscines (90.4%), with only 6.9% from 10- rufi gularis , and Mionectes macconnelli ). Another strategy primaried oscines and 2.7% from 9-primaried oscines. appeared to be the presence of two breeding peaks, one Th e most frequently captured species were in the during the dry season from about September to December, Th amnophilidae (41.9%), followed by Furnariidae (27.5%), and a second in March and April (e.g. Myrmeciza ferruginea Pipridae (9.1%), and Tyrannidae (7.3%). and perhaps also Percnostola rufi frons ). Other species appeared An active brood patch was observed in 1472 (5.5%) to breed year-round, although breeding peaked at various of captured individuals, and of these, 187 (12.7%) were times of the year depending on the species (e.g. Gymnopithys simultaneously undergoing symmetrical primary feather rufi gula , P. r u fi frons, and Glyphorynchus spirurus ). Peaks molt, which we considered to be molt– breeding overlap

Table 1. The proportion of captured birds (n ϭ number of total captures) in breeding condition by species and month near Manaus, Brazil. The degree of shading corresponds to breeding frequency (none: Ͻ 2.0%, light: 2 – 3.9; medium: 4.0 – 7.9; dark: Ͼ 8.0). Months in bold (June – December) indicate the dry season. Species J F M A MJ J A S O N D n Thamnophilus murinus 8 17 5 7 18 14 8 212 Thamnomanes ardesiacus 9 3 2 3 10 7 13 7 3 13 1002 Thamnomanes caesius 5 2 10 9 6 7 8 13 804 Myrmotherula axillaris 11 13 3 4 4 3 5 2 4 10 2 4 429 Myrmotherula longipennis 13 8 2 8 12 7 5 5 5 10 582 Myrmotherula menetriesii 8 666 5132 319 Hypocnemis cantator 7 3 5 3 13 11 6 7 10 7 6 581 Percnostola rufi frons 3 5 13 6 3 3 5 8 10 6 13 9 813 Myrmeciza ferruginea 31 29 7 6 12 19 31 11 8 31 258 Pithys albifrons 10 3 6 1 2 1 2 4 4 4 1 2086 Gymnopithys rufi gula 5 9 6 6 3 3 4 5 13 13 14 24 1144 Willisornis poecilinotus 12 9 9 4 3 2 5 5 6 8 6 8 1950 Formicarius colma 23 14 10 9 6 4 5 6 7 25 388 Sclerurus rufi gula 13 14 13 19 17 3 15 275 Automolus infuscatus 6 6 3 9 6 12 11 7 13 514 Xenops minutus 13 8 6 3 11 15 286 Certhiasomus stictolaemus 3 13 17 11 20 12 589 Dendrocincla fuliginosa 11 6 6 6 21 14 17 15 6 287 Dendrocincla merula 7 4 7 3 3 3 8 11 17 9 573 Glyphorynchus spirurus 16 10 7 3 4 4 3 2 3 4 7 8 2200 Xiphorhynchus pardalotus 3 2 1 5 15 8 9 5 2 1004 Corythopis torquatus 5 10 6 3 7 15 4 306 Mionectes macconnelli 5 6 13 2 1 3 3 6 1191 Myiobius barbatus 5 4 4 4 4 1 654 Lepidothrix serena 10 6 3 1 2 2 7 3 5 619 Pipra pipra 7 5 2 1 1 3 6 4 9 9 14 1878 Pipra erythrocephala 11 15 24 9 33 209 Hylophilus ochraceiceps 8 5 3 10 7 4 8 357 Microbates collaris 6 961267155912 539 Turdus albicollis 18 3 6 6 2 3 6 9 714 Tachyphonus surinamus 12 16 14 3 15 10 6 297

145 Table 2. The proportion of captured birds (n ϭ number of total captures) with wing molt by species and month near Manaus, Brazil. The degree of shading corresponds to molting frequency (none: Ͻ 4.0%, light: 4– 7.9; medium: 8.0– 15.9; dark: Ͼ 16.0). Months in bold (June – December) indicate the dry season. Species J F M A MJ J A S O N D n Thamnophilus murinus 17 14 50 33 18 28 5 14 24 40 13 53 241 Thamnomanes ardesiacus 65 45 47 49 39 22 14 23 36 44 52 61 1387 Thamnomanes caesius 30 36 52 45 35 26 18 22 31 36 39 37 1058 Myrmotherula axillaris 34 24 26 15 23 10 14 18 31 32 48 33 549 Myrmotherula longipennis 36 17 19 11 10 10 15 30 38 47 53 35 813 Myrmotherula menetriesii 61 43 64 20 38 42 27 19 21 25 18 29 432 Hypocnemis cantator 29 36 26 37 16 20 12 17 26 26 35 37 799 Percnostola rufi frons 46 28 26 49 43 57 48 36 34 33 24 38 1061 Myrmeciza ferruginea 33 16 16 30 50 39 29 33 28 25 31 35 334 Pithys albifrons 68 61 63 63 62 59 49 46 59 60 59 57 2636 Gymnopithys rufi gula 49 54 56 53 59 61 48 39 49 36 47 49 1548 Willisornis poecilinotus 38 40 36 37 36 29 20 8 18 28 31 35 2840 Formicarius colma 5 20 38 46 49 49 35 8 2 565 Sclerurus rufi gula 9 23 18 28 50 38 30 17 4 6 4 450 Automolus infuscatus 23 33 20 17 16 7 11 31 40 37 40 17 708 Xenops minutus 36 38 13 6 2 2 7 3 28 41 29 434 Dendrocincla fuliginosa 19 33 25 7 4 2 20 38 41 44 41 386 Dendrocincla merula 54 56 40 29 24 18 14 17 42 58 48 78 824 Certhiasomus stictolaemus 19 25 22 23 9 15 13 27 46 43 44 33 889 Glyphorynchus spirurus 8 12 32 34 31 33 31 18 13 6 5 4 2771 Xiphorhynchus pardalotus 53 34 24 23 14 11 5 17 46 52 60 60 1300 Corythopis torquatus 43 32 20 10 12 5 3 3 4 13 33 55 538 Mionectes macconnelli 2 14 12 29 18 9 4 1 1 1 1495 Myiobius barbatus 20 15 6 7 10 2 1 8 46 66 59 61 936 Lepidothrix serena 10 25 22 32 8 8 3 1 1 4 706 Pipra pipra 12 14 30 36 14 8 5 1 1 1 3 10 2248 Pipra erythrocephala 13 22 17 10 4 239 Hylophilus ochraceiceps 46 43 43 41 14 8 1 3 13 22 38 582 Microbates collaris 31 33 37 33 19 18 12 6 12 13 26 30 824 Turdus albicollis 2 22 43 51 46 23 9 2 1078 Tachyphonus surinamus 23 43 73 47 38 21 19 6 8 4 8 402

(Supplementary material Appendix 1). Th e occurrence of followed by gap specialists, terrestrial insectivores, other molt– breeding overlap varied among species and higher- insectivores, and fl ock obligates with 13 – 17%. Th e remain- level taxa, and was signifi cantly more frequent in suboscines ing guilds included fl ock dropouts, frugivores, and non- χ2 ϭ ϭ Ͻ (13.3%) than in oscines (6.4%; 1 4.6, p 0.032; forest species, and had 7% molt – breeding overlap Fig. 2). Th e frequency of molt– breeding overlap diff ered frequency. signifi cantly among suboscine families, not including families with Ͻ 15 individuals observed with brood patches χ2 ϭ Ͻ ( 1 98.2, p 0.001; Fig. 2). Among suboscines, molt– Correlates of molt– breeding overlap breeding overlap was most frequent in the Th amnophilidae (23.0%) and least frequent in the Tyrannidae (4.4%), We focus subsequent analyses on 31 species with Ն 1 5 Tityridae (0.0%) and Pipridae (1.5%). Within oscines, observed brood patches. Th e proportion of these popula- molt – breeding overlap was more frequent in 10-primaried tions either breeding or undergoing wing molt in each oscines (7.9%), especially in the Polioptilidae (15.2%), month from 1979 to 2009 indicated substantial variation than 9-primaried oscines (2.5%), but diff erences between in the timing and duration of the molting and breeding χ2 ϭ 10- and 9-primaried oscines were not signifi cant ( 1 1.8, seasons among species (Table 1, 2). Of these 31 species, p ϭ 0.19), nor were pairwise diff erences among oscine fami- 20 (65%) had substantially prolonged breeding and molt- lies (Fig. 2). In only one occasion was molt– breeding overlap ing seasons that each lasted Ն 6 months of the year. Th ese observed in a 9-primaried oscine (Cyanocompsa cyanoides ; prolonged breeding and molting seasons set the stage for Supplementary material Appendix 1). molt – breeding overlap to occur at the individual-level. We Molt – breeding overlap frequency diff ered among eco- asked four questions to understand how the timing and χ2 ϭ Ͻ logical guilds ( 7 81.0, p 0.001; Fig. 3). Ant-followers duration of life history events, like molting and breeding, had the highest frequency of molt – breeding overlap (31%), aff ected molt – breeding overlap frequency.

146 25 A

A 15 A

10 B A B B B B A 5 B Molt-breeding overlap frequency (%) A B A B 81 21 617 30 108 19 26 33 23 29 0 1331303 134 101 40 SuboscinesDendrocolaptinaeFurnariidaeSclerurinaeThamnophilidaeFormicariidaeTyrannidaePipridae 10-primariedVireonidaeTroglodytidae oscinesPolioptilidaeTurdidae9-primariedThraupidae oscines

Figure 2. Frequency of molt– breeding overlap by order (black bars) and family (gray bars) near Manaus, Brazil. Numbers inside bars indicate the sample size (number of individuals captured with brood patches). Letters above bars represent diﬀ erences using post hoc pairwise comparisons; comparisons are made across the three suborders and among families within suboscines and within 10- and 9-primaried oscines.

35 A 30

20 B B 15 B B

10 BC BC Molt-breeding overlap frequency (%) 5 C

178 93 95 340 280 193 33 213 0 Ant-followerGap specialistTerrestrial Otherinsectivore insectivoreObligate flockFlock dropoutNonforest Frugivore

Figure 3. Frequency of molt– breeding overlap for eight ecological guilds near Manaus, Brazil. Letters above bars represent diﬀ erences among post hoc pairwise comparisons and the numbers within the bars represents the sample size (number of individuals with brood patches).

147 0.4 a trend for the greatest variation in initial molt date and a low frequency of molt – breeding overlap (Table 3, Fig. 5), suggesting they were capable of adjusting the timing of 0.3 their molt to avoid overlap with breeding. Second were species with MD more similar in length to MS: Strategy 2. Th ese species still had a shorter MD than MS (0.61 Ͻ 0.2 MD/MS Ͻ 0.84), intermediate variation in molt initiation date, the longest molt duration, and the greatest proportion of birds with molt – breeding overlap (Table 3). Species 0.1 with these ﬁ rst two strategies had a higher frequency of molt– breeding overlap as MD approached MS (R2 ϭ 0.56, ϭ Ͻ

Proportion of molt-breeding overlap F1,16 20.8, p 0.001; Fig. 5). Th ird were species with non- 0.0 overlapping molting and breeding seasons: Strategy 3. 75 100 125 150 175 200 225 250 275 300 325 Th ese species had MD lasting approximately as long as Duration of primary feather molt (days) MS (0.84 Ͻ MD/MS Ͻ 1.16), little variation in molt ini- Figure 4. Th e relationship between the average estimated duration tiation date, and a low frequency of molt – breeding overlap of primary feather molt among individuals for each species and the (Table 3, Fig. 5) because the molting and breeding seasons frequency of molt – breeding overlap in 27 species near Manaus, were relatively distinct (Table 1, 2). Brazil. A simple linear regression line with 95% confi dence inter- Th ese three life history strategies appear to have a phy- vals is also illustrated. logenetic basis. Woodcreepers and antbirds (but also one tanager and one gnateater) typically exhibited the fi rst two strategies, whereas Strategy 3 species were largely fl ycatchers, 1) Does molt duration affect molt– breeding overlap manakins, a vireo, and a thrush, but also two woodcreepers frequency? and an antthrush (Table 3). We used molt duration estimates obtained by documenting molt extent of individual birds recaptured while progress- 3) Do feather growth rates of individual birds predict ing through a complete wing molt. We obtained reason- their probability of exhibiting molt– breeding overlap? able molt duration estimates for 27 of the 31 focal species We used logistic regression to test whether individual (average Ϯ SE: 21.2 Ϯ 7.5 birds examined per species; birds that molted more slowly had an increased probability Supplementary material Appendix 2). For these 27 species, of molt – breeding overlap. Although guilds diff ered in their χ2 ϭ ϭ molt duration positively correlated with molt– breeding response (Wald ’ s 4 15.5, p 0.004), more slowly grown 2 ϭ ϭ Ͻ feathers did not increase the probability of molt – breeding overlap frequency (R 0.69, F1,25 56.9, p 0.001; Fig. 4). χ2 ϭ overlap pooling across all nine species (Wald’ s 1 2.4, 2) Can birds adjust the timing of molt to minimize p ϭ 0.12), although there may have been an interaction χ2 ϭ molt – breeding overlap? between feather growth rate and guild (Wald ’ s 4 7.9, Th ree strategies emerged with respect to variation in molt p ϭ 0.096; Fig. 6). When we conducted the analysis initiation dates. First were species with short molt dura- using only females, the interaction became nearly signifi - χ2 ϭ ϭ tion (MD) relative to their molting season (MS; MD/MS cant (Wald ’ s 4 8.8, p 0.066), indicating guild-specifi c Ͻ 0.61), hereafter called Strategy 1. Th ese species showed responses aff ect the infl uence of feather growth rates

Table 3. Three strategies among average molt duration (MD; the length of time an individual takes to molt), variation in molt initiation date (MI; the variance of when individuals begin molting), and molt– breeding overlap (MBO) frequency near Manaus, Brazil for three strategies. Species are placed in strategies defi ned by the ratio of individual molt duration to the population’ s molting season (MD/MS). Average MD (in days), MI (in days), and MBO (% frequency) are given Ϯ SE, as well as the species comprising each strategy. Each of these three variables was tested for differences among strategies with a 1-factor ANOVA; letters indicate Tukey-adjusted comparisons of variable means across strategies. Strategy 3 Strategy 1 Strategy 2 (distinct molting and Ͻ Ͻ Ͻ (MD/MS 0.61) (0.61 MD/MS 0.84) breeding seasons) F2,24 p MD (days) 149 Ϯ 13 (A) 217 Ϯ 11 (B) 153 Ϯ 12 (A) 11.0 Ͻ 0.001 MI (days Ϯ SE) 25 Ϯ 4 (A 1 ) 22 Ϯ 3 (AB) 15 Ϯ 3 (B1 ) 3.3 0.056 MBO (% Ϯ SE) 4.3 Ϯ 2.5 (A) 21.3 Ϯ 3.3 (B) 2.8 Ϯ 1.0 (A) 18.7 Ͻ 0.001 Species Myrmotherula menetriesii Thamnomanes ardesiacus Formicarius colma Hypocnemis cantator Thamnomanes caesius Dendrocincla fuliginosa Sclerurus rufi gularis Myrmotherula axillaris Glyphorynchus spirurus Automolus infuscatus Myrmotherula longipennis Corythopis torquata Certhiasomus stictolaemus Percnostola rufi frons Myiobius barbatus Dendrocincla merula Myrmeciza ferruginea Lepidothrix serena Tachyphonus surinamus Pithys albifrons Pipra pipra Gymnopithys rufi gula Hylophilus ochraceiceps Willisornis poecilinotus Turdus albicollis Xiphorhynchus pardalotus Microbates collaris 1 Tukey-adjusted A– B difference: p ϭ 0.083.

148 0.4 and Dur ã es 2001, Wolfe et al. 2009). Th is variation in phenology has been demonstrated in other lowland tropical forest communities, suggesting it may be widespread 0.3 (Chapman 1995, Wikelski et al. 2003a, b).

Molt – breeding overlap frequency 0.2 For all captured passerines, 12.7% of individual birds that were breeding were also molting, i.e. were undergoing 0.1 molt – breeding overlap. Th e frequency of molt – breeding overlap varied considerably among well-sampled species, Molt-breeding overlap frequency and was highest in P. albifrons (38.8%), but at or near 0.0 0.0% in many other species (Supplementary material 0.4 0.6 0.8 1.0 1.2 Appendix 1). Antbirds in general were frequently breeding MD/MS and molting simultaneously, whereas oscines, and in par- Figure 5. Th e relationship between molt – breeding overlap fre- ticular thrushes and tanagers, rarely or never were seen with quency and the ratio of average individual molt duration (MD) molt – breeding overlap (Fig. 2). to molting season length (MS). MD/MS indicates the relative Variation in molt – breeding overlap not only varied by amount of time an average individual takes to molt relative to the family but also varied according to foraging guild (Fig. 3). length of time that species ’ population has molting individuals near In particular, ant-followers, gap specialists, terrestrial insec- Manaus, Brazil. Th e simple linear regression line and 95% confi - tivores, other insectivores, and obligate fl ock members had dence interval only includes species with Strategies 1 (black dots; relatively frequent molt– breeding overlap compared to fl ock short individual molts that temporally overlap with the population’ s breeding season) and 2 (gray dots; long individual molts that dropouts, nonforest species, and frugivores. We acknowl- temporally overlap with the population’ s breeding season), i.e. edge confounding phylogenetic eff ects when sorting by for- excluding species with temporally distinct molting and breeding aging guild, because guilds with thamnophilids had greater seasons (Strategy 3; open circles; Table 3). molt – breeding overlap than guilds without antbirds. Even so, many other taxa contributed to at least some of these guilds with frequent molt– breeding overlap (Supplemen- on molt– breeding overlap (Fig. 6). Flock obligates tary material Appendix 1). Future studies should evaluate (Th amnomanes spp.), fl ock dropouts (G. spirurus and variation in molt – breeding overlap among foraging guilds X. pardalotus), and other insectivores (Willisornis poecilinotus ) in the Afrotropics, tropical Australasia, and Oceania to had increasing probabilities of molt– breeding overlap when determine whether our observed patterns are consistent feather growth rates were slower. when taxonomic contributions to each guild are considerably diff erent. 4) When in the molt sequence does breeding overlap? Among Furnariidae with molt – breeding overlap, brood Correlates of molt– breeding overlap patches were observed mainly when molting primaries 1 – 4 and 9 – 10, but among Th amnophilidae with molt – breeding Th e duration of molt at least partially explained molt– overlap, brood patches were observed throughout the molt breeding overlap frequency. Species with greater molt– cycle (Table 4). breeding overlap frequency were those with prolonged molts, often taking at least 150 d to complete primary feather replacement (Fig. 4). From an energy budget per- Discussion spective, it may be particularly costly to molt and breed simultaneously if energetic resources are dedicated to a rapid Breeding and molting seasonality near Manaus molt (Foster 1974). By slowing the rate of molt, daily energetic demands may be reduced and would minimize costs Among all species combined, molt frequency rose and of molt – breeding overlap. A slow molt that decreases gaps fell about one or two months after similar rises and falls among fl ight feathers may also reduce predation risk, espe- in breeding frequency (Fig. 1). Th e breeding season, as cially during take-off , and may ensure greater feather quality measured by brood patches, peaked during the late dry and or allow resources to be diverted to increased immunological early wet season (October – January; Fig. 1) for many spe- vigor (Slagsvold and Dale 1996, Hedenströ m and Sunada cies at our study sites near Manaus, Brazil, similar to other 1999). Gymnopithys rufi gula and P. albifrons represented Amazonian bird communities (Poulin et al. 1992, Marini the extreme in seasonal strategies, with at least 36% in and Dur ã es 2001, Verea et al. 2009). Some species also had G. rufi gula and 46% in P. albifrons molting year-round and a secondary breeding peak during the wet season (often individuals taking 9 – 10 months to complete their wing February– April), whereas a few species bred most frequently molt. Th ese species may not maintain an annual cycle, during the wet season, and others apparently bred year- but instead breed when highly unpredictably local condi- round (Table 1). Typical of most temperate and tropical pas- tions are suitable (i.e. when enough ant swarms in their serines, a complete post-nuptial ( ‘ pre-basic ’ sensu Humphrey home range are simultaneously swarming [J. Chaves- and Parkes 1959, Howell et al. 2003) follows breeding in Campos pers. comm.]). Th ese ant-followers, as well as our study species (Poulin et al. 1992, Pyle 1997, Marini terrestrial insectivores, and fl ock obligates, all are most

149 (a) Insectivore, ant-follower (2 spp.) (b) Insectivore, flock obligate (2 spp.) (c) Insectivore, flock dropout (2 spp.) 1.0 1.0 1.0

0.8 0.8 0.8

0.6 0.6 0.6

0.4 0.4 0.4

Probability of MBO 0.2

Probability of MBO 0.2 Probability of MBO 0.2

0.0 0.0 0.0 –6 –4 –2 0246 –3 –2 –1 0 1 2 3 –6 –4 –2 0 2 4 6 Residual days Residual days Residual days

(d) Insectivore, gap specialist (2 spp.) (e) Insectivore, other (1 spp.) 1.0 1.0

0.8 0.8

0.6 0.6

0.4 0.4 Probability of MBO Probability of MBO 0.2 0.2

0.0 0.0 –4 –2 0 2 4 –3–2–10123 Residual days Residual days

Figure 6. Th e relationship between feather growth rate (residual of the average number of days for a tail feather to complete molting) and the frequency of molt– breeding overlap (MBO) for fi ve guilds representing nine species near Manaus, Brazil illustrated by solid lines (all individuals) and dotted lines (females only) from logistic regression models. Larger residuals represent slower growing feathers. Individuals are represented by dots (females) and open circles (males) with 1.0 ϭ experiencing molting breeding overlap and 0.0 ϭ breeding, but not molting. Species in guilds are as follows: ant followers (Pithys albifrons , Gymnopithys rufi gula ), fl ock obligates (Th amnomanes ardesiacus , T. caesius ), fl ock dropouts (Glyphorynchus spirurus , Myrmotherula axillaris ), gap specialists (Hypocnemis cantator , Percnostola rufi gula ), other insectivore (Willisornis poecilinota ). closely associated with the forest understory among our 1996, Hemborg and Lundberg 1998, Hemborg 1999, study species and had among the highest rates of molt – Hemborg et al. 2001). breeding overlap. Th is microhabitat is climactically the We showed that decreased feather growth rates increased most predictable within tropical forest (Kapos et al. 1993), an individual bird ’ s probability of having molt – breeding thus may have the lowest consequences for molt – breeding overlap in fi ve of nine species (Fig. 6). To our knowledge, overlap from an energetic perspective. Th ere may be addi- this has not been demonstrated before and provides a unique tional advantages to having a prolonged molt in the under- insight into how individual birds may be forced to breed story, such as developing more structurally robust feathers, and molt simultaneously. Decreased feather growth rates which would reduce wear and abrasion (Dawson et al. 2000); can be caused by nutritional defi ciencies (Grubb 1989) sug- this may help explain why gap specialists also have relatively gesting that depressed body condition in some birds can high levels of molt – breeding overlap. increase their probability of having molt – breeding over- When breeding and molting seasons overlap (Strategies lap, which in turn may depress fecundity (Slagsvold and 1 and 2), the ability to vary the timing of molt initiation Dale 1996, Hemborg and Lundberg 1998, Hemborg 1999, appears to correlate with reduced molt– breeding overlap Hemborg et al. 2001). frequency. Other species minimize molt – breeding overlap Individuals of some species, particularly slow-molting by temporally separating the breeding and molting season species like P. albifrons , G. rufi gula, and Percnostola rufi frons , (Strategy 3; Table 3, Fig. 5). Individuals of species that take did not show a relationship between feather growth rates the longest time to molt typically use the entire molting and the probability of experiencing molt– breeding overlap, season to molt and are therefore less fl exible in when they suggesting that molt – breeding overlap is not always a con- initiate molt (Strategy 2). Th is group of species in Strategy sequence of body condition. Many antbirds can breed and 2 also exhibits the most frequent molt – breeding overlap, molt simultaneously at any stage of the wing molt sequence but the trade-off between molt duration and energy expen- (Table 4). Even in some furnariids, molt– breeding overlap diture may minimize fi tness costs otherwise associated with occurred nearly throughout the molting cycle, although molt– breeding overlap (Foster 1974, Slagsvold and Dale more frequently in the beginning or end of the molt cycle.

150 Table 4. The number of birds with brood patches, by stage of primary feather molt (primary 1 to primary 10 [p1– p10]) by species and family. Half birds are a result of asymmetry in the extent of primary molt between wings. Unk indicates birds that exhibited molt– breeding overlap (MBO), but the extent of molt was not recorded. No. MBO is sample size (the total number of birds) used for each species and family. Molting primary feather Family Species p1 p2 p3 p4 p5 p6 p7 p8 p9 p10 Unk No. MBO Thamnophilidae 16 17 17.5 16 10.5 13.5 11 13.5 12 22 19 171 Frederickena viridis 0.5 0.5 1 2 Thamnophilus murinus 0.5 0.5 1 Thamnomanes ardesiacus 2.5 1.5 2 1.5 1 5.5 2.5 0.5 2 1 1 21 Thamnomanes caesius 1 3.5 1 0.5 1 0.5 2.5 10 Epinecrophylla gutturalis 22116 Myrmotherula guttata 0.5 0.5 1 2 Myrmotherula axillaris 112 Myrmotherula longipennis 11 1 1 121 1 9 Hypocnemis cantator 0.5 1.5 1 1 3 3 10 Percnostola ruﬁ frons 2 3 0.5 0.5 6 12 Myrmeciza ferruginea 0.5 0.5 1 2 2 1 1 2 10 Myrmornis torquata 1 1 0.5 0.5 1.5 0.5 5 Pithys albifrons 3 1 4 3.5 2.5 3 3 1 2 2 2 27 Gymnopithys ruﬁ gula 5.5 3.5 5 4 1 4 0.5 2 2.5 6 3 37 Hylophylax naevius 11 Willisornis poecilinotus 1.5 4 1.5 1 0.5 1.5 0.5 0.5 4 1 16 Furnariidae 1.5 4.5 1.5 2.5 1 0 1 0 3 2 6 23 Philydor erythrocercum 1 1 Automolus infuscatus 11 Xenops minutus 0.5 0.5 1 2 Dendrocincla fuliginosa 11 Dendrocincla merula 22 Glyphorynchus spirurus 0.5 1.5 2 3 7 Dendrocolaptes picumnus 1 1 Dendrocolaptes certhia 112 Xiphorhynchus pardalotus 21115 Campylorhamphus procurvoides 11

Brood patches for a single brood may last up to four weeks molting and breeding among all individuals captured or and these taxa take 15– 30 d to grow a primary fl ight feather otherwise sampled (Ralph and Fancy 1994, Yap 2005, Verea (Supplementary material Appendix 2). Th erefore, even et al. 2009), rather than among only individuals in when primary 1 or 2 is molting, the physiological mecha- breeding condition. At our study sites near Manaus, Brazil, nisms that regulate breeding have been long activated, sug- molt – breeding overlap among all individuals using this met- gesting a minimal eff ort to temporally separate molting and ric would be 0.7% (Supplementary material Appendix 1). breeding in contrast to many temperate species (Murphy Th is, however, includes non-breeding-season captures and 1996, Dawson 2006, Hau et al. 2008). demographic groups that do not typically develop brood patches, such as males of some species or subadults, and Across-study comparisons therefore is strongly dependent on the context of the study. Molt – breeding overlap reported as the number of birds Using brood patch and wing molt data from captured birds breeding and molting, divided by the total number of birds is a highly conservative metric and can easily be replicated breeding, as we have done, would facilitate greater compari- across sites, although not all studies have assessed molt – son among studies. breeding overlap in this manner. For example, Foster ’ s (1975) Despite the limitations outlined above, some general seminal paper highlighting the frequency of molt– breeding patterns emerge across studies. First, molt– breeding overlap overlap among tropical birds in Costa Rica was largely is widespread in the Neotropics (Foster 1975, Avery 1985, based on specimens and examination of their reproductive Piratelli et al. 2000, Marini and Dur ã es 2001, Rohwer organs. Variation among tropical birds in the degree to which et al. 2009) and can occur in tropical regions around the they regress their gonads after breeding ranges from nearly world (Moreau 1936, Payne 1969, Ralph and Fancy 1994, complete, as in temperate birds, to nearly absent (Moreau Yap 2005). Second, molt– breeding overlap is apparently 1936, Miller 1962, Snow and Snow 1964, Davis 1971, rarely pervasive in any given species, usually with Ͻ 20% Astheimer and Buttemer 1999, Wikelski et al. 2003a). Th us, of breeding individuals also molting (Marini and Durã es it becomes diffi cult to accurately determine breeding condi- 2001, Supplementary material Appendix 1). Th ird, studies tion across many study species using reproductive organs. conducted in the wet Neotropics suggest some consistency Comparisons to other studies are further complicated in the frequency of molt – breeding overlap. At 18 – 22 ° S, because of inconsistencies in how to molt – breeding over- Marini and Durã es (2001) found 8.3% molt– breeding lap is reported. Its frequency is in some cases measured overlap when expressed as the percentage of breeding birds by calculating the proportion of birds simultaneously simultaneously molting. Th is is consistent with Foster (1975)

151 with 8.1% at 10° N and this study near Manaus, Brazil with Claramunt, S., Derryberry, E. P., Remsen, J. V. and Brumfi eld, 12.7% at 2 °S. Geographical patterns, such as latitudinal and R. T. 2012. High dispersal ability inhibits speciation in a environmental gradients in molt– breeding overlap, remain continental radiation of passerine birds. – Proc. R. Soc. B 279: diffi cult to confi dently discern, however, because so few 1567–1574. Cody, M. L. 1966. A general theory of clutch size. – Evolution 20: community-level datasets are available, and because molt – 174 – 184. breeding overlap frequencies likely depend on the biome, Cohn-Haft, M., Whittaker, A. and Stouff er, P. C. 1997. A region, and taxonomic composition of the avian commu- new look at the ‘ species-poor ’ central Amazon: the avifauna nity (Poulin et al. 1992, Mallet-Rodrigues 2005, Magalh ã es north of Manaus, Brazil. – Ornithol. Monogr. 48: 205 – 235. et al. 2007). For example, we primarily studied understory Cracraft, J. and Barker, F. K. 2009. Passerine birds (Passeriformes). passerines, and molt– breeding overlap frequency of the com- – In: Hedges, S. B. and Kumar, S. (eds), Th e timetree of life. munity may increase or decrease as other members of the Oxford Univ. Press, pp. 423 – 431. community are included. Davis, J. 1971. Breeding and molt schedules of rufous-collared sparrow in coastal Per ú . – Condor 73: 127 – 146. Although the physiology underlying molt– breeding Dawson, A. 2006. Control of molt in birds: association with overlap and the role of the environment on molt– breeding prolactin and gonadal regression in starlings. – Gen. Comp. overlap remains largely untested, our research suggests that Endocrinol. 147: 314 – 322. it is an important component of life history events in some Dawson, A. 2008. Control of the annual cycle in birds: endocrine tropical birds. Future studies documenting molt– breeding constraints and plasticity in response to ecological variability. overlap should report both the proportion of total birds with – Phil. Trans. R. Soc. B 363: 1621 – 1633. molt– breeding overlap as well as the proportion with breed- Dawson, A. and Newton, I. 2004. Use and validation of a molt ing evidence that are simultaneously molting. Th is will facili- score index corrected for primary feather mass. – Auk 121: 372 – 379. tate cross-regional comparisons to understand the frequency, Dawson, A., Hinsley, S. A., Ferns, P. N., Bonser, R. H. C. and extent, and variation of breeding, molting, and their overlap Eccleston, L. 2000. Rate of moult aff ects feather quality: among tropical birds. a mechanism linking current reproductive eff ort to future survival. – Proc. R. Soc. B 267: 2093 – 2098. del Hoyo, J., Elliott, A. and Christie, D. A. 1992 – 2010. Handbook Acknowledgements – We thank the many banders, assistants, to the birds of the World. – Lynx Editions. and mateiros who helped to collect these data. Logistical support Drent, R. H. and Daan, S. 1980. Th e prudent parent: energetic from the staff of the Biological Dynamics of Forest Fragmentation adjustments in avian breeding. – Ardea 68: 225 – 252. Project and the support by Brazil’ s Inst. Nacional de Pesquisas da Ferretti, V., Llambí as, P. E. and Martin, T. E. 2005. Life-history Amaz ô nia and the Smithsonian Inst. made this research possible. variation of a neotropical thrush challenges food limitation Funding for the project has also been provided by the World theory. – Proc. R. Soc. B 272: 769 – 773. Wildlife Fund-U.S., the MacArthur Foundation, the Andrew W. Foster, M. S. 1974. A model to explain molt– breeding overlap Mellon Foundation, U.S. Agency for International Development, and clutch size in some tropical birds. – Evolution 28: U.S. National Aeronautics and Space Administration, Brazil ’ s 182 – 190. Ministry for Science and Technology, U.S. Nation Science Foster, M. S. 1975. Th e overlap of molting and breeding in some Foundation, the Summit Foundation, Shell Oil, Citibank, tropical birds. – Condor 77: 304 – 314. Champion International, the Homeland Foundation, and the Franklin, D. C., Smales, I. J., Quin, B. R. and Menkhorst, P. W. National Geographic Society. Helpful suggestions by W. Goymann, 1999. Annual cycle of the helmeted honeyeater Lichenostomus L. L. Powell, J. V. Remsen and F. C. Rohwer have improved melanops cassidix , a sedentary inhabitant of a predictable the quality of this manuscript. Th is is publication 591 of the environment. – Ibis 141: 256 – 268. Biological Dynamics of Forest Fragments Project Technical Series Gill, F. and Donsker, D. 2011. IOC world bird names (ver. 2.8). Ͻ Ͼ and number 24 in the Amazonian Ornithology Technical Series – www.worldbirdnames.org , accessed 27 May 2011. of the INPA Zoological Collections Program. Th is manuscript Goymann, W., Moore, I. T., Scheuerlein, A., Hirschenhauser, K., was approved for publication by the Director of the Louisiana Grafen, A. and Wing fi eld, J. C. 2004. Testosterone in tropical Agricultural Experimental Station as manuscript number 2012- birds: eff ects of environmental and social factors. – Am. Nat. 241 - 6696. 164: 327 – 334. Grubb, T. C. 1989. Ptilochronology: feather growth bars as indica- tors of nutritional status. – Auk 106: 314 – 320. Hahn, T. P., Swingle, J., Wingfi eld, J. C. and Ramenofsky, M. References 1992. Adjustments of the prebasic molt schedule in birds. – Ornis Scand. 23: 314 – 321. Ashmole, N. P. 1963. Th e regulation of numbers of tropical oceanic Hall, K. S. S. 2005. Do nine-primaried passerines have nine or ten birds. – Ibis 103: 458 – 473. primary feathers? Th e evolution of a concept. – J. Ornithol. Astheimer, L. B. and Buttemer, W. A. 1999. Gonadal and hormo- 145: 121 – 126. nal patterns in the annual cycle of an Australian honeyeater. Hau, M., Perfi to, N. and Moore, I. T. 2008. Timing of breeding – In: Adams, N. J. and Slotow, R. H. (eds), Proc. 22nd Int. in tropical birds: mechanisms and evolutionary implications. Ornithol. Congr., Durban. BirdLife South Africa, Johannes- – Ornitol. Neotrop. 19: 39 – 59. burg, pp. 1768 – 1783. Hau, M., Ricklefs, R. E., Wikelski, M., Lee, K. A. and Brawn, Avery, M. L. 1985. Annual molt pattern in a Malaysian popula- J. D. 2010. Corticosterone, testosterone and life-history strat- tion of fantail warblers (Cisticola juncidis ). – Condor 87: egies of birds. – Proc. R. Soc. B 277: 3203 – 3212. 346 – 349. Hedenstr ö m, A. and Sunada, S. 1999. On the aerodynamics of Bierregaard, R. O., Gascon, C., Lovejoy, T. E. and Mesquita, moult gaps in birds. – J. Exp. Biol. 202: 67 – 76. R. C. G. 2001. Lessons from Amazonia: the ecology and Helm, B. and Gwinner, E. 1999. Timing of postjuvenal molt in conservation of a fragmented forest. – Yale Univ. Press. African ( Saxicola torquata axillaris) and European ( Saxicola Chapman, A. 1995. Breeding and moult of four bird species in torquata rubicola ) stonechats: eff ects of genetic and environ- tropical West Africa. – Trop. Zool. 8: 227– 238. mental factors. – Auk 116: 589 – 603.

152 Hemborg, C. 1999. Sexual diff erences in moult – breeding overlap Morales, J., Moreno, J., Merino, S., Sanz, J. J., Tomá s, G., Arriero, and female reproductive costs in pied fl ycatchers, Ficedula E., Lobato, E. and Mart í nez-de la Puente, J. 2007. Early moult hypoleuca . – J. Anim. Ecol. 68: 429 – 436. improves local survival and reduces reproductive output in Hemborg, C. and Lundberg, A. 1998. Costs of overlapping repro- female pied fl ycatchers. – Ecoscience 14: 31 – 39. duction and moult in passerine birds: an experiment with Moreau, R. E. 1936. Breeding seasons of birds in east African the pied fl ycatcher. – Behav. Ecol. Sociobiol. 43: 19 – 23. evergreen forest. – Proc. Zool. Soc. Lond. 1936: 631 – 653. Hemborg, C., Sanz, J. J. and Lundberg, A. 2001. Eff ects of latitude Moreno, J. 2004. Moult – breeding overlap and fecundity limitation in on the trade-off between reproduction and moult: a long- tropical birds: a link with immunity? – Ardeola 51: 471 – 476. term study with pied fl ycatcher. – Oecologia 129: 206 – 212. Murphy, M. E. 1996. Energetics and nutrition of molt. – In: Carey, Howell, S. N. G., Corben, C., Pyle, P. and Rogers, D. I. 2003. Th e C. (ed.), Avian energetics and nutritional ecology. Chapman fi rst basic problem: a review of molt and plumage homologies. and Hall, pp. 31 – 60. – Condor 105: 635 – 653. Nilsson, J.-A. and Svensson, E. 1996. Th e cost of reproduction: Humphrey, P. S. and Parkes, K. C. 1959. An approach to the study a new link between current reproductive eff ort and future of molts and plumages. – Auk 76: 1 – 31. reproductive success. – Proc. R. Soc. B 263: 711 – 714. Jetz, W., Freckleton, R. P. and McKechnie. A. E. 2008. Environ- Oniki, Y. 1979. Is nesting success low in the tropics? – Biotropica ment, migratory tendency, phylogeny and basal metabolic rate 11: 60 – 69. in birds. – PLoS One 3: e3261. Payne, R. 1969. Overlap of breeding and molting schedules in a Johnson, E. I., Stouff er, P. C. and Vargas, C. F. 2011. Diversity, collection of African birds. – Condor 71: 140 – 145. biomass, and trophic structure of a central Amazonian rain- P é rez, G. E. and Hobson, K. A. 2006. Isotopic evaluation of forest bird community. – Rev. Bras. Ornitol. 19: 1 – 16. interrupted molt in northern breeding populations of the Jones, R. E. 1971. Th e incubation patch of birds. – Biol. Rev. 46: loggerhead shrike. – Condor 108: 877 – 886. 315 – 339. Pimm, S. 1976. Estimation of duration of bird molt. – Condor Jovani, R., Blas, J., Navarro, C. and Mougeot, F. 2011. Feather 78: 550 – 550. growth bands and photoperiod. – J. Avian Biol. 42: 1 – 4. Piratelli, A. J., Siqueira, M. A. C. and Marcondes-Machado, L. O. Kapos, V., Ganade, G., Matsui, E. and Victoria, R. L. 1993. ∂ 13C 2000. Reproduç ã o e muda de penas em aves de sub-bosque as an indicator of edge eff ects in tropical rainforest reserves. na regi ã o leste de Mato Grosso do Sul. – Ararajuba 8: – J. Ecol. 81: 425 – 432. 99 – 107. Karr, J. R. 1971. Structure of avian communities in selected Poulin, B., Lefebvre, G. and McNeil, R. 1992. Tropical avian Panama and Illinois habitats. – Ecol. Monogr. 41: 207 – 233. phenology in relation to abundance and exploitation of Kjell é n, N. 1994. Moult in relation to migration in birds: a review. food resources. – Ecology 73: 2295 – 2309. – Ornis Svecica 4: 1 – 24. Pyle, P. 1997. Identifi cation guide to North American birds, part Kulesza, G. 1990. An analysis of clutch-size in New World I. – Slate Creek Press. passerine birds. – Ibis 132: 407 – 422. Ralph, C. J. and Fancy, S. G. 1994. Timing of breeding and mol- Lack, D. 1947. Th e signifi cance of clutch-size. – Ibis 89: ting in six species of Hawaiian honeycreepers. – Condor 96: 302 – 352. 151 – 161. Laurance, W. F. 2001. Th e hyper-diverse fl ora of the central Ricklefs, R. E. 1976. Growth rates of birds in the humid New Amazon: an overview. – In: Bierregaard, R. O., Gascon, C., World tropics. – Ibis 118: 179 – 207. Lovejoy, T. E. and Mesquita, R. C. G. (eds), Lessons from Ricklefs, R. E. 1980. Geographical variation in clutch size among Amazonia: the ecology and conservation of a fragmented passerine birds: Ashmole’s hypothesis. – Auk 97: 38 – 49. forest. Yale Univ. Press, pp. 47 – 53. Ricklefs, R. E. and Wikelski, M. 2002. Th e physiology life-history Leu, M. and Th ompson, C. W. 2002. Th e potential importance of nexus. – Trends Ecol. Evol. 17: 462– 468. migratory stopover sites as fl ight feather molt staging areas: a Robinson, W. D., Robinson, T. R., Robinson, S. K. and Brawn, review for Neotropical migrants. – Biol. Conserv. 106: 45 –56. J. D. 2000. Nesting success of understory forest birds in Lockwood, R., Swaddle, J. P. and Rayner, J. M. V. 1998. central Panama. – J. Avian Biol. 31: 151 – 164. Avian wingtip shape reconsidered: wingtip shape indices and Rohwer, V. G., Rohwer, S. and Ortiz-Ramirez, M. F. 2009. Molt morphological adaptations to migration. – J. Avian Biol. 29: biology of resident and migrant birds of the monsoon region 273 – 292. of west Mexico. – Ornitol. Neotrop. 20: 565 – 584. Magalh ã es, V. S., de Azevedo, S. M., de Lyra-Neves, R. M., Roper, J. J. 2005. Try and try again: nest predation favors Telino-J ú nior, W. R. and de Souza, D. P. 2007. Biologia de persistence in a Neotropical bird. – Ornitol. Neotrop. 16: aves capturadas em um fragmento de Mata Atlâ ntica, Igarassu, 253 – 262. Pernambuco, Brasil. – Rev. Bras. Zool. 24: 950 – 964. Ryder, T. B. and Wolfe, J. D. 2009. Th e current state of knowledge Mallet-Rodrigues, F. 2005. Molt– breeding cycle in passerines from on molt and plumage sequences in selected Neotropical bird a foothill forest in southeastern Brazil. – Rev. Bras. Ornitol. families: a review. – Ornitol. Neotrop. 20: 1 – 18. 13: 155 – 160. Siikam ä ki, P., Hovi, M. and R ä tti, O. 1994. A trade-off between Marini, M. Â . and Dur ã es, R. 2001. Annual patterns of molt and current reproduction and moult in the pied fl ycatcher: an reproductive activity of passerines in south-central Brazil. experiment. – Funct. Ecol. 7: 476 – 482. – Condor 103: 767 – 775. Skutch, A. F. 1949. Do tropical birds rear as many young as they Martin, T. E. 1995. Avian life history evolution in relation to nest can nourish? – Ibis 91: 430 – 455. sites, nest predation, and food. – Ecol. Monogr. 65: 101 – 127. Skutch, A. F. 1969. Life histories of Central American birds. – Pac. Martin, T. E. 1996. Life-history evolution in tropical and south Coast Avifauna 35: 248 – 269. temperate birds: what do we really know? – J. Avian Biol. 27: Slagsvold, T. and Dale, S. 1996. Disappearance of female pied 1 – 10. fl ycatchers in relation to breeding stage and experimentally Martin, T. E., Martin, P. R., Olson, C. R., Heidinger, B. J. and induced molt. – Ecology 77: 461 – 471. Fontaine, J. J. 2000. Parental care and clutch sizes in North Snow, D. W. and Snow, B. K. 1963. Breeding and the annual cycle and South American birds. – Science 287: 1482 – 1485. in three Trinidad thrushes. – Wilson Bull. 75: 27 – 41. Michener, H. and Michener, J. R. 1938. Bars in fl ight feathers. Snow, D. W. and Snow, B. K. 1964. Breeding seasons and annual – Condor 40: 149 – 160. cycles of Trinidad land-birds. – Zoologica 49: 1 – 39. Miller, A. H. 1962. Bimodal occurrence of breeding in an equato- Sombroek, W. 2000. Amazon landforms and soils in relation to rial sparrow. – Proc. Natl Acad. Sci. USA 48: 396 – 400. biological diversity. – Acta Amazonica 30: 81 – 100.

153 Stouff er, P. C. and Bierregaard, R. O. 1993. Spatial and temporal Underhill, L. G. and Zucchini, W. 1988. A model for avian abundance patterns of ruddy quail-doves (Geotrygon montana ) primary molt. – Ibis 130: 358 – 372. near Manaus, Brazil. – Condor 95: 896 – 903. Verea, C., Solorzano, A., Diaz, M., Parra, L., Araujo, M. A., Stouff er, P. C. and Bierregaard, R. O. 1995. Use of Amazonian Anton, F., Navas, O., Ruiz, O. J. L. and Fernandez-Badillo, A. forest fragments by understory insectivorous birds. – Ecology 2009. Record of breeding and molt activities in some 76: 2429 – 2445. birds of northern Venezuela. – Ornitol. Neotrop. 20: Stouff er, P. C. and Borges, S. H. 2001. Conservation recommenda- 181 – 201. tions for understorey birds in Amazonian forest fragments and Weathers, W. W. 1979. Climatic adaptation in avian standard secondary areas. – In: Bierregaard, R. O., Gascon, C., Lovejoy, metabolic rate. – Oecologia 42: 81 – 89. T. E. and Mesquita, R. (eds), Lessons from Amazonia: the Wikelski, M., Hau, M., Robinson, W. D. and Wingfi eld, J. C. ecology and conservation of a fragmented forest. Yale Univ. 2003a. Reproductive seasonality of seven Neotropical passer- Press, pp. 248 – 261. ine species. – Condor 105: 683 – 695. Stouff er, P. C., Bierregaard, R. O., Strong, C. and Lovejoy, T. E. Wikelski, M., Spinney, L., Schelsky, W., Scheuerlein, A. and 2006. Long-term landscape change and bird abundance Gwinner, E. 2003b. Slow pace of life in tropical sedentary in Amazonian rainforest fragments. – Conserv. Biol. 20: birds: a common-garden experiment on four stonechat 1212 – 1223. populations from diff erent latitudes. – Proc. R. Soc. B 270: Stresemann, E. and Stresemann, V. 1966. Die Mauser der V ö gel. 2383 – 2388. – J. Ornithol. 107: S357 – S375. Wing fi eld, J. C. 2005. Flexibility in annual cycles of birds: implica- Stutchbury, B. J. M. and Morton, E. S. 2001. Behavioral ecology tions for endocrine control mechanisms. – J. Ornithol. 146: of tropical birds. – Academic Press. 291 – 304. Summers, R. W., Swann, R. L. and Nicoll, M. 1983. Th e eff ects Wolfe, J. D., Pyle, P. and Ralph, C. J. 2009. Breeding seasons, molt of methods on estimates of primary moult duration in the patterns, and gender and age criteria for selected northeastern redshank Tringa totanus . – Bird Study 30: 149 – 156. Costa Rican resident landbirds. – Wilson J. Ornithol. 121: Terborgh, J., Robinson, S. K., Parker, T. A., Munn, C. A. and 556 – 567. Pierpont, N. 1990. Structure and organization of an Amazonian Yap, C. A.-M. 2005. Phenology of tropical birds in peninsular forest bird community. – Ecol. Monogr. 60: 213 – 238. Malaysia: eff ects of selective logging and food resources. – MS Th é ry, M. 1997. Wing-shape variation in relation to ecology thesis, National Univ. of Singapore. and sexual selection in fi ve sympatric lekking manakins Young, B. E. 1994. Geographic and seasonal patterns of clutch- (Passeriformes: Pipridae). – Ecotropica 9: 9 – 19. size variation in house wrens. – Auk 111: 545 – 555.

Supplementary material (Appendix J5574 at Ͻ www. oikos oﬃ ce.lu.se/appendix Ͼ ). Appendix 1 – 2.

154