PLUMAGE DEVELOPMENT AND MOLT IN LONG-TAILED MANAKINS (CHIROXIPHIA LINEARIS): VARIATION ACCORDING TO SEX AND AGE

Abstract Lek-mating Long-tailed Manakins (Chiroxiphia linearis) exhibit an unusual pattern of delayed plumage maturation. Each year, males progress through a series of predefinitive plumages before attaining definitive plumage in their fifth calendar year. Females also exhibit variation in plumage coloration, with some females displaying male-like plumage characteristics. Using data from mist-net captures in northwest Costa Rica (n = 1,315) and museum specimens from throughout the range of Long-tailed Manakins (n = 585), we documented the plumage sequence progression of males, explored variation in female plumage, and described the timing of molt in this species. Males progressed through a series of age-specific predefinitive plumages, which enabled the accurate aging of predefinitive-plumaged males in the field; this predefinitive plumage sequence is the basis for age-related status-signaling in these males. Females tended to acquire red coloration in the crown as they aged. However, colorful plumage in females may be a byproduct of selection on bright male plumage. Females exhibited an early peak of molt activity from February to April, little molt from May through July, and a second, more pronounced peak of molt activity in October. By contrast, males in older predefinitive-plumage stages and males in definitive plumage exhibited comparable unimodal distributions in molt activity beginning in June and peaking between July and October. Our data are consistent with selective pressure to avoid the costs of molt-breeding overlap in females and older males. Our findings have important implications for social organization and signaling in Long- tailed Manakins, and for the evolution of delayed plumage maturation in birds. Desarrollo del Plumaje y Muda en Chiroxiphia linearis: Variación de Acuerdo al Sexo y la Edad

DќѢѐђѡ ђѡ юљ. 30 [Auk,Vol. 124 Iћ њюћѦ яіџёѠ, males do not assume defi nitive, adult-like plumage for one or more years aĞ er hatching. In some species, this delay in plumage maturation is accompanied by a delay in sexual maturation (Lawton and Lawton 1986). In many sexually dichromatic, northtemperate passerines, however, males do not aĴ ain their defi nitive plumage until aĞ er their fi rst potential breeding season, despite having reached sexual maturity (Rohwer et al. 1980, Lyon andMontgomerie 1986). The adaptive signifi cance of this type of delayed plumage maturation has been the focus of extensive research in recent decades (e.g., Rohwer et al. 1980, Lyon and Montgomerie 1986, Hill 1996, with most studies investigating the one-year delays typically exhibited by north-temperate passerines. However, many tropical passerines deviate from this paĴ ern. In some species, such as bowerbirds and birds of paradise, males molt into the same predefi nitive plumage for several years before assuming defi nitive plumage (Frith andBeehler 1998, Frith andFrith 2004). In other species, for example manakins, Darwin's fi nches, paradise fl ycatchers, Hawaiian honeycreepers, and monarch fl ycatchers, males progress through a transitional series of different predefi nitive plumage stages before assuming defi nitive plumage (Foster 1987;McDonald 1989aMcDonald , 1993aLepson and Freed 1995;VanderWerf 2001;Mulder et al. 2002;DuVal 2005). Investigations of the adaptive signifi cance of delayed plumage maturation, and of the signal function of variation in plumage, require a thorough understanding of plumage development and molt. Here, we investigate age-and sex-related plumage variation in the Long-tailed Manakin (Chiroxiphia linearis).
Long-tailed Manakins have a lek-based mating system (Foster 1977;McDonald 1989a, b). Males gather at lek sites, where they establish agegraded dominance hierarchies. The two most dominant males at each lek, the alpha and beta males, perform vocal duets and elaborate, dualmale dance displays for females visiting the leks. Females usually copulate only with alpha males (Foster 1977, McDonald 1989b, McDonald and PoĴ s 1994. As in other lekking species (Höglund and Alatalo 1995), females are solely responsible for rearing off spring (Foster 1976).
As they grow older, male Long-tailed Manakins progress through a series of transitional, predefi nitive plumage stages before aĴ aining defi nitive plumage in their fi Ğ h calendar year (Foster 1987;McDonald 1989aMcDonald , 1993a. This delay is unusually long for such a small (15-21 g) passerine (Lawton and Lawton 1986). Female Long-tailed Manakins also vary in plumage color, with some females developing red or tawny crown feathers, a trait believed hechas con redes de niebla en el noroeste de Costa Rica (n = 1,315) y especímenes de museo de todo el rango de distribución (n = 585), documentamos la secuencia de plumajes progresivos de los machos, exploramos la variación en el plumaje de las hembras y describimos el momento en que tiene lugar la muda en C. linearis. Los machos progresaron a través de una serie de plumajes pre-defi nitivos específi cos para cada clase de edad, lo que permitió determinar la edad de los machos sin plumaje defi nitivo con exactitud en el campo. Esta secuencia de plumajes predefi nitivos es la base para la señalización del estatus entre los machos, que está relacionada con la edad en esta especie. Las hembras tendieron a adquirir coloración roja en la corona a medida que envejecieron. Sin embargo, el plumaje colorido en las hembras podría ser un subproducto de la selección por plumaje brillante en los machos. Las hembras exhibieron un pico temprano de actividad de muda entre febrero y abril, poca muda entre mayo y julio y un segundo pico más pronunciado de actividad de muda en octubre. En contraste, los machos en estadíos de plumaje predefi nitivo más avanzados y en plumaje defi nitivo exhibieron distribuciones unimodales de actividad de muda comparables comenzando en junio, con un pico entre julio y octubre. Nuestros datos concuerdan con la existencia de una presión selectiva para evitar los costos del sobrelapamiento de la muda de las hembras y de los machos más viejos. Nuestros hallazgos tienen implicaciones importantes relacionadas con la organización social y la señalización entre individuos de C. linearis, y también para la evolución de la maduración retardada del plumaje en las aves.
to be associated with age (McDonald 1989b). Such extensive age-and sex-based variation in plumage may present unique signaling opportunities, particularly in species with complex social organization (Lawton and Lawton 1986). In Long-tailed Manakins, for example, a male's position within a lek hierarchy may be largely infl uenced by his age. Thus, transitional plumages that reliably signal a male's age or status in the hierarchy would likely reduce the occurrence of costly, escalated encounters between males of low and high status (Foster 1987, McDonald 1993a. Foster (1987) described predefi nitive male plumages in Long-tailed Manakins as transitioning through several stages along a continuous spectrum, with second-year males being mostly green with a red crown patch and, occasionally, some black on the face, coverts, and body and fl ight feathers, and third-and fourth-year males having a mixture of red, green, black, and blue plumage. McDonald (1989aMcDonald ( , 1993a proposed an alternative sequence, whereby males progress through a transitional series of age-specifi c predefi nitive plumages; in this proposed sequence, males develop a red crown in their second year, a black mask in their third year, and a blue mantle in their fourth year. Using recapture and resighting data from three study populations, together spanning 27 years, we reexamined the plumage sequences proposed by Foster (1987) and McDonald (1989aMcDonald ( , 1993a to determine whether predefi nitive male plumage can reliably signal age in Long-tailed Manakins. Additionally, we investigated paĴ erns of variation in female plumage and described sex-and age-based variation in the timing of molt.

MђѡѕќёѠ
We studied Long-tailed Manakins at three sites in northwestern Costa Rica: from 1971Rica: from to 1974Rica: from and in 1977 at the Enrique Jiménez Nuñez Experiment Station (10°20'N, 85°8'W); from 1981 to 1999 in Monteverde (10°18'N, 84°48'W); and from 2003 to 2005, as well as for a few weeks in 1986, in Santa Rosa National Park, Guanacaste Conservation Area (10°40'N, 85°30'W). The Monteverde site is located in premontane tropical moist forest (Holdridge 1966), and the Santa Rosa and Jiménez Station sites encompass both evergreen boĴ omland moist forest and areas of tropical dry forest on surrounding hillsides (Stiles and Skutch 1989). All three sites exhibit marked seasonality, with a dry season extending from approximately January through April, and a rainy season extending from May through December. Male Long-tailed Manakins display actively from February through September, with a pronounced peak in activity from March through June (Foster 1977;McDonald 1989aMcDonald , b, 1993b. Active nests have been discovered from March to July (Foster 1976; S. M. Doucet unpubl. data), though the nesting season likely extends until September (Foster 1976).
We captured 1,315 Long-tailed Manakins using mist nets and fi Ĵ ed each individual with a unique combination of plastic colored leg bands; birds at Monteverde and Santa Rosa also carried a numbered aluminum leg band. Of 20 birds banded as nestlings, only 2 were recaptured or resighted in following years (one male and one female). To determine the plumage sequence followed by birds as they aged, we recorded detailed descriptions of each bird's plumage coloration each time it was captured. Whenever possible, we also recorded plumage descriptions of marked individuals seen during behavioral observations at lek sites or encountered opportunistically at the study sites. Over the course of the study, R.P.C. and M.S.F. noted that second-year birds of both sexes could be identifi ed by the presence of retained juvenal wing feathers; R.P.C. also discovered that the mouth-lining color of second-year birds was diagnostic (Clay 2001). Thus, we examined the mouth-lining color and the amount of wear on wing feathers of all green-plumaged birds captured from 1997 to 1999 and in 2005. Only a subset of the data we present here was described briefl y elsewhere (plumage sequence information from 56 males; McDonald 1989aMcDonald , 1993a.
To assess the timing of molt, we examined all birds captured from 1971 to 1977, some birds captured in 1986 and 1987, and all birds captured since 1995 for signs of molt (i.e., sheathed feathers), recording whether birds were molting at time of capture and noting in which regions of which feather tracts they were molting. Occasionally, we noted extremely asymmetric molt or molt of only a single, isolated feather on some individuals. We assumed in both instances that the molt was adventitious and did not include these individuals in our calculations of proportions of birds molting (Pyle 1997, VanderWerf 2001. Although most birds were DќѢѐђѡ ђѡ юљ. 32 [Auk,Vol. 124 captured and observed between March and July, we collected molt information on wild birds in all months except September and February. To obtain additional molt data spanning the calendar year, we also examined 585 Longtailed Manakin specimens in museum collections (see Acknowledgments). We recorded feather regions and tracts with molting feathers and, if discernible, the plumages from which and into which a bird was molting. Many museum specimens were also examined for the presence of juvenal remiges and coverts. Taken together, our data from museum specimens and wild birds spanned the calendar year, though some months are considerably beĴ er represented than others. Molt and plumage terminology follow the Humphrey-Parkes (H-P) system (Humphrey and Parkes 1959) as summarized in Pyle (1997). A recent review (Howell et al. 2003) recommends modifi cations to H-P terminology. The complex nature of molt in Long-tailed Manakins, however, makes it diffi cult to assign consistent molt terminology to diff erent sex and age classes under the proposed changes. Thus, for simplicity, we opted for traditional H-P terminology. We also follow Pyle's (1997) age terminology. Thus, a hatch-year is a bird in its fi rst calendar year (until 31 December of the year it fl edged), a second-year bird is in its second calendar year (from 1 January to 31 December of the year following fl edging), a third-year bird is a bird in its third calendar year, and so on.
Wild birds in all-green or primarily green plumage were identifi ed as females if they had a vascularized brood patch, were recaptured or resighted in green plumage in multiple years, were aged as aĞ er-second-year birds on the basis of the absence of retained juvenal wing feathers (see above), exhibited female-like behavior during dance displays by males (McDonald 1989a), or were identifi ed as such either by genetic sexing (Griffi ths et al. 1998) or by laparotomy. Birds in green plumage were identifi ed as males if they were recaptured or resighted in predefi nitive or defi nitive male plumages in subsequent years or by laparotomy. For museum specimens, green birds were identifi ed as females or males only if the specimen tag indicated the presence of ovaries or testes, respectively. Birds in green plumage that did not meet these criteria were considered of unknown sex and were not included in our analyses.

RђѠѢљѡѠ
Molt.-All Long-tailed Manakins underwent a prebasic (postbreeding) molt each year. The olive-green juvenal plumage grown in the nest was identical in males and females. The fi rst prebasic molt began within four months of fl edging and was a partial molt: hatch-year birds of both sexes retained some of their juvenal wing feathers. Thus, second-year birds had two generations of wing feathers: retained juvenal remiges and distal greater coverts, and new lesser and median coverts and proximal greater coverts. Additionally, the mouth linings of second-year birds were bright orange to orange-yellow, similar to those of nestlings (Foster 1976), whereas the mouth linings of older birds were paler and more pinkish in color. All subsequent prebasic molts were complete.
Male plumage sequence.-Overall, we captured and marked 653 individuals in Monteverde, 280 individuals at the Jiménez Station, and 382 individuals in Santa Rosa. Many of these birds were recaptured or resighted in one or more subsequent years, which allowed us to document 562 plumage transitions from a total of 235 males. Of these transitions, 343 were successive molts in defi nitive plumage, whereas 219 were molts from one predefi nitive plumage to another or from predefi nitive to defi nitive plumage. Of these 219 informative transitions, we have records for two diff erent plumage stages for 98 males, three diff erent plumage stages for 32 males, and four diff erent plumage stages for 19 males. One male was sighted in all fi ve plumage stages.
Males progressed through the following plumage sequence ( Fig. 1 and Tables 1 and 2). They acquired their juvenal plumage in the nest, which, like the plumage of females, was olive green above with a paler wash below. Within four months of fl edging, males initiated their fi rst prebasic molt. Males in their fi rst basic plumage, which we term "red-cap plumage," were olive green throughout with a small red crown patch. The amount of red in the crown was highly variable, and males oĞ en had two strips of red feathers on the outer edges of the crown rather than a contiguous red crown patch ( Fig. 2A). A limited number of males in red-cap plumage also had some black feathers on the face. These black feathers, or additional ones, were acquired along with an expanded red crown in some males during a partial molt (either a limited prealternate molt or the early onset of the second prebasic molt; see below) in March and April. In addition, some males undergoing the fi rst prebasic molt replaced their central, but no other, rectrices (M. S. Foster unpubl. data). These rectrices were darker and longer than those retained from the juvenal plumage.
In the following year, males underwent their second prebasic molt. Males in their second basic plumage, which we term "black-face plumage," were mostly olive-green with a small red crown patch and a black facial mask or hood. In this plumage, the red crown patch was always contiguous and was larger than that of red-cap males but smaller than that of males in defi nitive plumage. The amount of black on the face varied from a small black facial mask to a full black hood. Some black-face males also had some black in their coverts and fl ight feathers, and a blackish tinge to their body plumage. 34 [Auk,Vol. 124 Rarely, males in black-face plumage had a limited number of blue feathers on their mantle.
Males in their third basic plumage, which we term "blue-back plumage," had a mixture of green and black body and fl ight feathers, a full red crown patch, and some blue feathers on the mantle. Birds in blue-back plumage exhibited the greatest range of variation in proportions of plumage colors, though all exhibited some blue in the back, and some black ventrally. OĞ en, blue-back males had body feathers that were mostly black with a tinge of olive-green and fl ight feathers that were black with green edging. Some blue-back males had very liĴ le green at all in their plumage and might be mistaken for defi nitive-plumaged males in the fi eld. Other blue-back males had substantial amounts of green in their black body plumage, giving them an almost grayish appearance.

35
Finally, males in defi nitive-basic plumage had entirely jet-black body and fl ight feathers, a sky-blue mantle, and a bifi d, red crown patch. A small number of males molting from blueblack plumage to defi nitive plumage retained minimal amounts of green on the rump, fl anks, or undertail coverts. This green disappeared during the following prebasic molt.
Four lines of evidence allow us to assign ages to these sequences. First, one male banded as a nestling was recaptured or observed in three subsequent years: in his fi rst prebasic molt, he acquired the red-cap plumage; in his second prebasic molt, the black-face plumage; and in his third prebasic molt, the blue-back plumage. (Unfortunately, this male was not resighted again until his sixth calendar year, when he was in defi nitive plumage.) Second, all red-cap males captured between 1997 and 1999 at Monteverde (32 individuals) and in 2005 at Santa Rosa (11 individuals) were identifi ed as second-year birds on the basis of retained juvenal remiges and distal greater coverts and mouth lining color. Third, we documented the full, four-year progression of 17 males from red-cap plumage to defi nitive plumage. Finally, all males observed in red-cap plumage invariably had black feathers on the face in the following year. Similarly, all males observed in black-face plumage had blue feathers on their mantle the following year. No male observed in defi nitive plumage was ever observed to have green feathers in subsequent years. Our observations suggest that this plumage sequence is unidirectional, nonreversible, and remarkably age-specifi c.
Of the 562 plumage transitions we recorded, 465 were sequential (i.e., the males were recaptured or resighted in the following year). These sequential transitions allowed us to assess the frequency of unexpected plumage transitions. We documented only four unusual plumage transitions. Two males were fi rst captured in all-green plumage and both were recaptured in black-face plumage the following year (Table 2). Because these males were captured in March and April as green males but in the following breeding season as black-face males, it is possible that they showed red in the crown during some of the interval between these records. A third otherwise all-green male was molting in a limited black mask and red crown when it was captured in April. A year later, aĞ er a single prebasic molt, it was in blue-back plumage. A fourth presumably aberrant bird, captured in red-cap plumage and without any black on its body, had a larger red crown, a well-developed black hood (but no other black body feathers), and a small area of blue feathers on the back the following year. Thus, ≤0.7% of plumage transitions deviated from the expected paĴ ern. Even if we exclude transitions between defi nitive plumage stages, these anomalies account for only 1.8% of plumage transitions. These anomalies invariably involved an accelerated plumage-maturation process, because no male ever remained in the same plumage stage for two subsequent years. In 96 instances, more than one year elapsed between recaptures or resightings of particular males. Even among these nonsequential transitions, males were always in the expected plumage category when they were eventually recaptured or resighted, as estimated from the number of years separating recaptures or resightings.
Female plumage.-Female Long-tailed Manakins are typically olive-green above with a paler wash below. However, of the 649 confi rmed females examined (including museum specimens), 145 (23%) had variable amounts of tawny or red feathers on the crown, ranging from a single feather to a full tawny or red crown. The presence of red in the crown of females may make it diffi cult for inexperienced observers to diff erentiate them from young males in redcap plumage. However, the red crown feathers of females can be distinguished from those of males by one or more of the following characteristics (Fig. 2). First, the red feathers of females oĞ en had a tawny or rusty appearance, a feature never observed in males. Second, the red color was oĞ en present on only some of the barbs of each feather in females, while the remainder of the feather remained green, thereby creating a slightly streaked appearance ( Fig. 2A). By contrast, the red feathers of young males usually had red distal barbs, and red, orange, or yellow central and proximal barbs, regardless of the number of red feathers on the crown (Fig. 2B). Third, the red feathers of females were usually the same length as the other (green) feathers on the crown, whereas the red feathers of males were usually longer than the green feathers and tended to increase in length with age ( Fig. 2; S. M. Doucet pers. obs.). Fourth, the distribution of red feathers on the crown diff ered between females and males. In females, red feathers could be found in the front, center, or rear of the crown. By DќѢѐђѡ ђѡ юљ. 36 [Auk,Vol. 124 contrast, among males in red-cap plumage, the red feathers tended to grow on the outer edges of the crown, oĞ en resulting in a split, rather than contiguous, red crown ( Fig. 2A).
Development of tawny coloration in the crown of females appeared to be associated with age. Of the females we were able to age as second-year or aĞ er-second-year (on the basis of plumage and mouth lining color), none of 43 second-year females had any trace of red or tawny on the crown, whereas 91 of 189 aĞ ersecond-year females had traces of tawny or red (Fisher's exact test, P <0.00001). Of the females that exhibited changes in plumage color during the course of the study, nine were observed or recaptured suffi ciently frequently to estimate the minimum amount of time elapsed between when the bird was originally captured and when a change in plumage was fi rst noted (mean ± SD = 5.1 ± 1.17 years). Two of these females were fi rst captured as second-year females; one of these developed a tawny crown in her eighth year, and the other developed red crown feathers in her ninth year. Some females never developed red or tawny crowns, including the three oldest, minimum-age females. Two were recaptured in completely green plumage 10 years aĞ er initial capture. A third female, last observed 15 years aĞ er her initial capture, had no discernible red or tawny in her crown.
In rarer instances, females departed from the typical olive-green plumage in other ways. During the course of our study, we captured a few females with black lores or black forehead, cheek, or nape feathers (n = 21); with one or more partially or completely black wing or tail feathers (n = 11); or with blue or black wing or tail coverts or blue mantle feathers (n = 5). Apart from crown color, none of these patches was extensive enough to be seen easily in the fi eld. Females with some tawny or red coloration in the crown were significantly more likely to show these additional malelike plumage tendencies (28 of 147) than females that were otherwise all green (6 of 504; Fisher's exact test, P < 0.00001). Two of the females that had black or blue on the body were recaptured in subsequent years and no longer had black or blue in their plumage. By contrast, all females with a red or tawny crown that were recaptured in subsequent years retained a colored crown.
Timing of molt.-Long-tailed Manakins follow a complex paĴ ern of molt that varies by age and sex. Among females, there is a bimodal distribution of the proportion of birds molting over the course of the year, with an early peak of molt activity in March and a second peak in October (Fig. 3A). The molt occurring between February and April is limited to some head and body feathers, whereas the second peak corresponds to the complete prebasic molt. The early peak varies by age. Of the females examined in March and April that we were able to age (n = 81), 13 of 15 (86%) second-year females showed signs of molt, whereas only 25 of 66 (38%) aĞ er-second-year females showed signs of molt (Fisher's exact test, P = 0.0003). Unfortunately, our data do not allow us to determine whether this early peak corresponds to a limited prealternate molt or whether it is an early beginning to the prebasic molt that is suspended through the reproductive period. This distinction would require confi rmation that newly molted feathers were replaced in the subsequent prebasic molt (in the case of a limited pre-alternate molt) or not replaced in the subsequent molt (in the case of an early suspended prebasic molt).
The molt activity paĴ erns of males in predefi nitive plumage are more complex. We separated males by plumage stage, which generally corresponds to age (see Table 1, and above). As with females, a large proportion of red-cap (second-year) males showed signs of molt in March and April (Fig. 3B). Among these young males, there was a slight decrease in molt activity in May. By July, however, all red-cap males captured or examined showed signs of molt. Of the red-cap males showing signs of molt in March and April (n = 28), 82% were molting only crown or head feathers. Three of these males (11%) were molting both head and body feathers, and two others (7%) were molting only body feathers. Many of these red-cap males were molting in additional red crown feathers and some were molting in black feathers on other parts of the head. In some cases, however, these males were simply molting in green crown or head feathers. The peak of molt in June corresponds to the complete fi rst prebasic molt: males were in a much heavier molt that oĞ en included fl ight feathers. Among black-face (third-year) and blue-back (fourth-year) males, we detected molt in only a small percentage of males (13%) in March and April (Fig. 3C, D). As with red-cap males, this early molt in black-face and blueback males was largely restricted to the crown and head regions. Black-face and blue-back males exhibited a peak of molt activity between June and August, which corresponded to their complete third and fourth prebasic molts, respectively.
Males in defi nitive plumage exhibited the simplest yearly molt activity paĴ ern (Fig. 3E). Very few (<5%) males showed any signs of molt from March to May. By June, more than onethird of defi nitive-plumaged males examined were molting, and molt activity in these males peaked in September.

DіѠѐѢѠѠіќћ
By examining the plumage characteristics of 1,315 color-banded individuals in northwestern Costa Rica and 585 museum specimens, we documented age-and sex-related variation in plumage maturation and molt timing in Longtailed Manakins. Male plumage development progressed in discrete, age-specifi c categories from an all-green juvenal plumage, through three distinct predefi nitive plumages, to a defi nitive adult plumage in fi Ğ h-year and older birds. This represents the longest delay yet reported in any manakin (see Foster 1987, Duval 2005. Although variation in female plumage was more subtle, some females acquired male-like plumage features as they aged. Our fi ndings have important implications for elucidating the adaptive signifi cance of delayed plumage maturation, and the potential signal function of intraspecifi c variation in plumage, in Longtailed Manakins and other species. 38 [Auk,Vol. 124 Male plumages.-Male Long-tailed Manakins progressed through the following plumage sequence. Hatch-year (juvenal) males were olive green throughout; second-year (red-cap) males were olive green with some red on the crown; third-year (black-face) males were olive green with a red crown and a black facial mask or hood; fourth-year (blue-back) males were a mixture of green and black with a red crown and a blue and green mantle; and fi Ğ h-year and older (defi nitive) males were black with a red crown patch and a blue mantle. This plumage sequence can be summarized by the following heuristic: from (1) a green-plumaged bird, (2) add red, (3) add black, (4) add blue, (5) take away green. Our conclusions are consistent with those of McDonald (1989aMcDonald ( , 1993a but differ from those of Foster (1987). In Foster's (1987) sequence, red-cap and black-face males were included in the same age class, whereas blueback males were separated into two age classes. Diff erences in interpretation of the red-cap stage may refl ect the fact that birds in red-cap plumage occasionally have some black feathers on the face, particularly once they begin to molt in the early breeding season. Foster (1987) therefore (incorrectly) included black-face birds in the red-cap plumage stage. The discovery that second-year birds can be aged on the basis of diff erences in wing feathers now makes this distinction infallible (see also Ryder and Durães 2005). Foster's (1987) diff erentiation of males with blue, black, and green plumage into two stages likely refl ects a misinterpretation of the extreme variation in the proportions of these colors in the third basic plumage. Nevertheless, our data from numerous plumage transitions show that the presence of blue mantle feathers and black body feathers (other than those on the head) is a robust criterion for defi ning the third basic (blue-back) plumage.
Although there was considerable variation within plumage stages of male Long-tailed Manakins, there was liĴ le overlap between plumage stages. Two apparent exceptions to the sequence involved males that were fi rst captured in green plumage and recaptured the following year in black-face plumage. These birds were captured before it was discovered that green birds could be aged on the basis of retained juvenal wing feathers, however, and it is likely that these were second-year males without a red crown patch, as opposed to juvenal-plumaged hatch-year males. Indeed, in a separate analysis, Clay (2001) identifi ed two museum specimens as second-year males with all-green crowns. Moreover, the males in our study were captured in March and April, and would have fl edged from exceptionally early nests if they were, in fact, hatch-year males. Two other males apparently molted from a red-cap plumage into a blue-back plumage. If, in fact, these four males represent anomalous sequence transitions, they correspond to accelerations of the typical sequence. However, because of the rarity of these anomalies (≤1.8% of predefi nitive plumage transitions), they probably occur arbitrarily rather than as a response to selective pressure on males to accelerate through the maturation process. Taken together, our fi ndings strongly suggest that plumage coloration is a highly consistent signal of age in young males until they reach their fi Ğ h calendar year. This conclusion has two important implications. First, our data confi rm that young males can be aged reliably in the fi eld on the basis of plumage features. Second, our data support the hypothesis that plumage variation can serve as a social signal of age in young males (McDonald 1993a).
Social signaling in males.-The complex social system of Long-tailed Manakins is dependent on the development of long-term cooperative alliances between males (Foster 1977;McDonald 1989aMcDonald , b, 1993a. Males must spend several years working their way up the dominance hierarchy to eventually have the opportunity to display for females and, in a minority of males, to copulate. The social stability of these mating queues appears to be strictly enforced by female mate choice, given that females will oĞ en leave the dance perch at the fi rst sign of disruption by males other than the dominant males (Foster 1987;McDonald 1989aMcDonald , 1993a. The evolution of predefi nitive male plumages is benefi cial to both subordinate and dominant males in this type of social system. For dominant males, the predefi nitive plumage of young males immediately identifi es them as not posing a threat to their reproductive success (McDonald 1989b, McDonald andPoĴ s 1994), which likely reduces the amount of aggression that young males will receive from older males. Indeed, in the congeneric Blue Manakin (C. caudata), the amount and intensity of aggression shown by dominant males toward subordinate individuals decrease with decreasing age of the target 39 individual (Foster 1987). In addition, the evolution of age-specifi c plumages can provide additional information about the status of males within the dominance hierarchy. In support of this hypothesis, a taxidermy mount experiment revealed that male Long-tailed Manakins responded more strongly to males in defi nitive plumage than to males in predefi nitive plumage (McDonald 1993a). Moreover, responses were oĞ en initiated by non-alpha males, which suggests that the model intruders posed a threat to established male-male alliances rather than a risk of stolen copulations (McDonald 1993a). These observations are paralleled in Blue Manakins, in which aggression directed toward a transgressor was oĞ en initiated by males of intermediate rank and directed toward the individuals nearest them in the hierarchy (Foster 1981). These studies suggest that males are most likely to challenge males of similar rank in a hierarchy and, presumably, against whom they have the greatest probability of success. The obverse of this is that males are also most likely to have to defend their own positions against challenges from members of their own or adjoining age cohorts. A model presentation experiment in another species with a graded, multiyear delay in plumage maturation, the Hawaiian 'Elepaio (Chasiempis sandwichensis), yielded similar fi ndings (VanderWerf and Freed 2003).
Female plumage.-Extreme sexual dichromatism is characteristic of the family Pipridae. Male manakins are usually brightly colored, whereas females are typically olive green (Prum 1997, Doucet et al. 2006. Here, we documented considerable variation in the plumage of female Long-tailed Manakins. Some females developed variable amounts of red or tawny coloration in the crown, reminiscent of the red crowns of males, and a small proportion of females developed black feathers on the face, head, wing, or tail. Older females were more likely than young females to develop red or tawny crowns, and, on average, more than fi ve years elapsed between initial female captures and the appearance of red in their crowns. Female ornamentation may have evolved as a byproduct of genetic correlations between male and female traits (Lande 1987, Amundsen 2000. Strong selection for elaborate ornaments in males, combined with weaker selection against these traits in females, could lead to the expression of ornamental coloration in females, even if it is maladaptive. Because females are solely responsible for parental care in manakins, they should experience strong natural selection for crypsis (Martin and Badyaev 1996), particularly given that rates of nest predation are quite high in the tropics (Stutchbury and Morton 2001). Yet bright female coloration has been reported in a number of manakin genera (Graves 1981) and, in Long-tailed Manakins, females with red crowns were more likely to express other malelike plumage characteristics. Our documented association between ornamental color and age in older females could be proximately mediated by age-related hormonal changes (Kimball and Ligon 1999). In some cases, male-like plumage characteristics other than crown color disappeared aĞ er the following molt, which suggests that they may have resulted from adventitious feather loss and replacement at a time of year when environmental cues, such as photoperiod or daytime light intensity (Gwinner 2003), led to the growth of more male-like plumage.
A less likely possibility is that the development of red crown coloration in older females serves a social signaling function. Because of its association with age, red crown coloration could signal a female's longevity or viability. Such a signal may be useful to males, who could optimize their display intensity or sperm allocation by investing more in females they deem to be of high quality (e.g., Amundsen 2000, Werner andLotem 2003), a paĴ ern that may explain some of the variation in male display intensity in this species. Alternatively, a signal of female age may be useful to young females, whose mate choice decisions could be infl uenced by those of older, experienced females when multiple females observe displaying males together (Dugatkin andGodin 1993, Doucet et al. 2004). Comparative analyses of the relationship between male and female ornamentation, in combination with intraspecifi c empirical investigations of the proximate control and signal function of female traits, will help to identify the relative merit of these hypotheses in explaining the evolution of bright female plumage in manakins.
Timing of molt.-Long-tailed Manakins un dergo a post-juvenal molt within four months of hatching and subsequently undergo a prebasic molt each year. The fi rst prebasic molt is incomplete: birds of both sexes retain their juvenal remiges and distal greater coverts. Second-year birds DќѢѐђѡ ђѡ юљ. 40 [Auk,Vol. 124 also retain the yellow-orange mouth color typical of nestlings of this species (Foster 1976). A similar paĴ ern of delayed maturation in mouth color has been reported in one species of bowerbird (Frith and Frith 2004). Given the considerable number of species that show delayed maturation in iris and soĞ -part color (Lawton and Lawton 1986), delayed maturation in mouth color may be relatively widespread and should be investigated in other species.
We found signifi cant age-and sex-based variation in timing of molt. Females exhibited an early peak of molt activity in February and March. Second-year females were more likely to show signs of early molt than older females. Nevertheless, fewer than half of females showed signs of early molt activity, and this molt was limited to some body and head feathers. Molt activity was low from May through July in females and increased in August to a second peak, corresponding to the complete prebasic molt, in October. The decrease in molt activity from April to July may refl ect evolutionary pressure to avoid the costs of molt-breeding overlap (Foster 1974), because this period corresponds to the peak of female nesting activity (Foster 1976). Molt and breeding are both energetically demanding (e.g., Murphy andKing 1992, Lindstrom et al. 1993). As such, physiological and ecological trade-off s are expected to occur between these two life-history traits (Foster 1975a). Theory therefore predicts that molt and breeding should be temporally segregated, yet molt-breeding overlap has been reported in numerous tropical and temperate passerines (Foster 1975b, Pyle 1997) and, as we have shown here, occurs at the population level in Long-tailed Manakins (Fig. 3). However, our data show that in May, June, and July, <15% of females exhibited any signs of molt. Later in the season, females' primary resource-consuming activity can switch from breeding to molt (see Foster 1975b), which may explain the second, most prominent peak of female molt activity in October.
Though they provide no parental care, male Long-tailed Manakins may expend considerable amounts of energy in breeding-related activities and are, therefore, also expected to avoid molt-breeding overlap. Dominant (alpha and beta) males perform energetically demanding dance displays for females and sing duets at high rates to aĴ ract females to their leks (≤5,000 vocalizations per day; Foster 1977, McDonald 1989b, Trainer and McDonald 1993. By June, there is usually a notable decrease in male display activity, especially at the least successful leks (S. M. Doucet and D. B. McDonald unpubl. data), which, as we have shown here, corresponds to a sharp increase in proportion of males in defi nitive plumage showing signs of molt. Females, on the other hand, are known to nest until at least July, and probably until September (Foster 1976). Male display activity is intrinsically linked to the availability of receptive females. A successful copulation, however, marks only the beginning of a female's parental responsibilities. Thus, the reproductive periods of dominant males and females are inherently staggered, which may explain why the postbreeding molt of males in defi nitive plumage peaks earlier than that of females. Moreover, many defi nitive-plumaged males occupy subordinate ranks at leks. These males are expected to expend much less energy on display than dominant males and to curtail their breeding season activities sooner.
Over the course of the breeding season, young male Long-tailed Manakins in predefi nitive plumage form affi liations with other males at particular lek sites, oĞ en practicing courtship displays on dance perches used by dominant males. Young males are tolerated at dance perches only in the absence of females, however, because only dominant males display for, and copulate with, females (McDonald 1989a). Moreover, young males are resighted at leks less frequently than older males (McDonald 1989a), which suggests that they invest less time and energy in breeding-season activities than older males. Thus, young males can begin molting earlier in the season than older individuals. Correspondingly, blue-back and black-face males exhibited an earlier peak of molt activity than males in defi nitive plumage.
Early molt in red-cap males is more diffi cult to interpret. Many males in red-cap plumage began to molt in March, at the onset of the breeding season. Moreover, this molt was apparently restricted to the crown and head. Indeed, these males oĞ en molted in additional red crown feathers and black-face feathers, thereby accelerating their transition into black-face plumage. Dominance interactions among males of the same age cohort may explain this early molt. If crown and mask color allow males to assert their dominance over members of their own age cohort, then selection might favor the acceleration of such a molt. Males in predefi nitive plumages, particularly younger ones, are subject to opposing selective forces: by signaling their lesser age, the plumages may reduce the aggression directed toward them by older males; at the same time, however, it might benefi t them to develop, as much as possible, the traits that signal their prowess to males of the same cohort (Foster 1987, McDonald 1993a). This confl ict may explain why some species progress through several distinct predefi nitive plumage stages rather than molting into the same predefi nitive plumage for several years, as has been documented in numerous other species (e.g., Frith andBeehler 1998, Frith andFrith 2004).
Implications.-Male Long-tailed Manakins in predefi nitive plumage have testes capable of producing sperm (Foster 1987). Thus, delayed plumage maturation in this species represents an instance of heterochrony and, more specifi cally, of neoteny (i.e., a slowing down of somatic maturation in relation to sexual maturation; Lawton and Lawton 1986). Mean testis size among 60 defi nitive-plumaged males was 29.3 ± 16.16 mm 3 (Foster 1987). Interestingly, the testis size of a beta male known to have been in defi nitive plumage for several years was 75 mm 3 , nearly three standard deviations above the mean (D. B. McDonald unpubl. data), which suggests that plumage maturation likely precedes full testis maturation in this species. Long-tailed Manakins may therefore exhibit both neoteny and progenesis (where somatic maturation precedes reproductive maturation). Lawton and Lawton (1986) proposed that hetero chronic trends in plumage, iris, and soĞpart maturation are associated with complex social organization within the Corvidae and may have played a role in speciation among closely related corvids. As empirical investigations of delayed plumage maturation become more widely available, particularly among tropical species, it will be possible to examine paĴ erns of heterochrony, and the complex mixing of neotenic and progenetic features, in other groups that vary in degree of social organization, including other manakins (Prum 1997). The increasing availability of well-supported phylogenies will greatly facilitate these comparative investigations. Studies of delayed plumage maturation across a wider variety of taxa, particularly species that diff er from the typical north-temperate paĴ ern, are likely to reveal that the adaptive signifi cance of delayed plumage maturation cannot be encompassed by a single hypothesis. Moreover, accumulating evidence of interspecifi c variation in both paĴ erns and consequences of delayed plumage maturation suggest broader evolutionary implications than we originally anticipated.