breeding ecology of the spotted barbtail (premnoplex

BREEDING ECOLOGY OF THE SPOTTED BARBTAIL (PREMNOPLEX BRUNNESCENS): A JOURNEY INTO THE UNKNOWN WORLD OF A TROPICAL, UNDERSTORY FURNARIID.

Harold F. Greeney Yanayacu Biological Station & Center for Creative Studies, Cosanga, Napo, Ecuador. c/o 721 Foch y Amazonas, Quito, Ecuador. E-mail: [email protected]

Submitted in partial fulfillment of requirements for: PhD. at the University of Wroclaw, Museum of Natural History, Wroclaw, Poland. Promoter: Prof. Dr. Hab. Tadeusz Stawarczyk

2

TABLE OF CONTENTS Page 3. Page 5. Page 13. Page 20. Page 24. Page 33. Page 35. Page 36. Page 40. Page 45. Page 48. Page 62. Page 65. Page 70. Page 86. Page 87.

Dedication Introduction. Site description and study methods. Results: General natural history. Results: The nest: architecture, placement, and construction. Results: Nest orientation. Results: Nesting seasonality. Results: Incubation rhythms. Results: Nestling growth and plumage development. Results: Nesting success. Discussion. Closing remarks on barbtail conservation. Summary Literature Cited Acknowledgements Publications resulting from this study

3

“The very strangeness of some of the birds which, like the Spotted Barbtail, live obscurely in the dim depths of the dripping mountain forests challenges us to cultivate their acquaintance; but often they are so retiring, the range of their activities and vocalizations appears to be so narrow when compared with those of birds of sunnier habitats, they seem so deficient in character or ‘personality,’ that we feel poorly rewarded for our strenuous efforts to know them more intimately.” Alexander F. Skutch, 1967

FIGURE 1. An adult Spotted Barbtail removed from the stomach of an Oxyrhopus cf. petola (Colubridae) snake found in a barbtail nest at 1150 m elevation, Mushullacta, eastern Ecuador. Also removed from the stomach were two eggs.

4

DEDICATION This study is dedicated to my mother, my father, and my sister, all of whom put up with my jars of slugs, boxes of bugs, “escaped pets” in the house, and animal skull collections for many, many years.

5

INTRODUCTION Neotropical ovenbirds and woodcreepers (Furnariidae sensu Remsen et al. 2008) represent one of the most diverse avian radiations in the New World. In addition to remarkable morphological, ecological, and behavioral diversification, this group is characterized by the explosive evolution of diverse nest architectures (Collias 1997, Zyskowski & Prum 1999). Certain evolutionary innovations in nest placement and structure in the ovenbirdwoodcreeper clade are thought to have facilitated its diversification into new habitats and to have encouraged the evolution of new morphological specializations (Irestedt et al. 2006). Unfortunately, however, a thorough understanding of furnariid nest evolution has been impeded by a lack of data on nest architecture and nesting behavior for several key species and genera (Zyskowski & Prum 1999, Remsen 2003, Irestedt et al. 2006, Greeney & Zyskowski 2008a, b). While nest descriptions are now available for additional species (e.g., Dobbs et al. 2003, Greeney & Zyskowski 2008b), there are still few available data concerning the natural history and reproductive biology of most members of this enigmatic family (Remsen 2003). With this study, on a poorly known, yet widely distributed species, I contribute the most complete look at the breeding ecology for any species of furnariid.

FIGURE 2. Adult Spotted Barbtail on its way to feed nestlings. Photo by Murray Cooper.

6

The Spotted Barbtail haunts the undergrowth of humid, montane, neotropical forests from 600-3000 m (Hilty & Brown 1986, Ridgely & Tudor 1989, Ridgely & Greenfield 2001, Hilty 2003, Remsen 2003). While its behavior makes it inconspicuous, it is a strikingly patterned bird (Figure 2), overall rusty-brown with a whitish to tawny-colored throat, and a breast bearing oblong tawny-ocraceous spots. The spots lengthen posteriorly to a dull brown belly (Figures 2 & 28.). The tail is dark brown to rufous, with the central rectrices slightly stiffened, and all rectrices lacking barbs on the terminal 3-6 mm, giving the tail a spiny appearance. Vocalizations range from a high, thin song, trilling at the end eep, eep, eep ti’ti’ti’titititi (Hilty & Brown 1986), to a high pitched pssiit call, and a cricketlike, high-frequency trill slowly descending in pitch and tempo, ti-ti-ti-ti-ti-ti-ti-ti-ti-ti (Areta 2007). Spotted Barbtails forage in the understory to mid-story, creeping along horizontal and vertical branches, only occasionally using their barbed tails as support. Adults glean and probe for a variety of arthropods on branches, bark crevices, epiphytes, dead leaves, and tree trunks, often hanging upside down (Wetmore 1972, Remsen 2003, Areta 2007). They may forage alone, in small groups, or occasionally with mixed flocks (Hilty & Brown 1986, Stiles & Skutch 1989, Ridgely & Greenfield 2001, Remsen 2003). Premnoplex is currently considered to be comprised of only two species; Spotted Barbtail (P. brunnescens) and White-throated Barbtail (P. tatei). The systematic position of Premnoplex has long been debated, and while previous classifications have subsumed the genus into Margarornis (e.g., Meyer de Schauensee 1966, Vaurie 1980), Premnoplex is now recognized as a valid genus (e.g., Remsen 2003, Irestedt et al. 2006, Remsen et al. 2008). Based principally on size, wing shape, tarsal proportions, hindlimb musculature, and plumage pattern, Premnoplex was considered part of the “Margarornis assemblage,” consisting of Margarornis, Roraimia, Premnornis, and Premnoplex (Vaurie 1980, Rudge & Raikow 1992a, 1992b). Within this group, however, relationships were unclear due to the conflict between anatomical features, molecular data, and natural history (Dobbs et al. 2003, Irestedt et al. 2006, Areta 2007). While Rudge & Raikow (1992b) suggested that Premnoplex and Premnornis should be considered sister taxa, Dobbs et al. (2003) suggested that the divergent nature of Premnornis nest structure should exclude it from the Margaronis group. Indeed, recent DNA data suggest that Premnornis forms a separate

7

clade with Pseudocolaptes, and is only distantly related to the Premnoplex-Margarornis clade (Irestedt et al. 2006). This is further supported by similarities in nest architecture between Premnornis and Pseudocolaptes, both building loose-cup nests of tree-fern scales inside tree cavities (Skutch 1969, Dobbs et al. 2003, Solano-Ugalde & Arcos-Torres 2007). Using the limited data available on nest architecture in the Furnariidae, Zyskowski & Prum (1999) suggested that Premnoplex belonged to a clade of “pensile nest” builders which included Margarornis, Siptornis, Cranioleuca, & Thripophaga. Recently, Areta (2007) agreed with previous studies that, in light of the convergence prone nature of characters associated with scansorial habits (e.g., hindlimb musculature) (e.g., Feduccia 1973, Irestedt et al. 2004), as well as differences in nest architecute and foraging behavior, Premnoplex and Margarornis should be considered sister taxa, with Roraimia provisionally considered related but with Premnornis excluded from this clade.

In the most recent molecular

analysis, Irestedt et al. (2006) concur that Margarornis and Premnoplex are sister taxa, and suggest that together they form a clade with Pygarrhichas. Within Premnoplex, the distinction between P. brunnescens and P. tatei is also debated, with some authors having considered them conspecific (Hellmayr 1938, Vaurie 1980). Others, however, separate them, but consider evidence warranting this separation to be weak (Remsen 2003). Areta’s (2007) recent ecological studies of the two species, including observations on vocalizations, behavior, and habitat use, suggest that these taxa are indeed deserving of species rank. Premnoplex brunnescens, originally described in the genus Margarornis, was later subdivided to exclude Premnoplex tatei by Chapman (1925), and this distinction is generally upheld (e.g., Ridgely & Tudor 1994, Remsen 2003, Remsen et al. 2008). While Premnoplex tatei is a range-restricted species endemic to Venezuela and considered vulnerable to extinction due to rapidly increasing habitat destruction (Remsen 2003, BirdLife International 2004), P. brunnescens is more broadly distributed from Costa Rica southward to central Bolivia.

Currently, P. brunnescens includes five recognized

subspecies (Remsen 2003). Remsen (2003) provided the currently accepted subspecific breakdown as follows: ssp. brunneicauda, Costa Rica and western Panama; ssp. brunnescens, extreme eastern Panama into western Venezuela and southward through all three ranges of the Colombian Andes through Ecuador to Cuzco, Peru; ssp. coloratus,

8

endemic to the Santa Marta Mountains of northern Colombia; ssp. rostratus, endemic to the costal mountains of northern Venezuela; and ssp. stictonotus, Andes of southern Peru south into Western and Central Bolivia as far south as Cochabamba. Despite its wide distribution, there is relatively little published on the breeding biology of the Spotted Barbtail. Most recently, Remsen (2003) summarized available nest descriptions as “massive ball[s] c. 30 cm in diameter.” Skutch (1967) provided the first nest description from a single nest (ssp. brunneicauda) found at Agua Buena, Costa Rica, at an elevation of 1,250 m.

This nest was built into a natural cavity in a fallen log,

suspended over a stream.

Subsequently, Stiles & Skutch (1989) expanded upon this

description of Costa Rican nests (ssp. brunneicauda), describing them as “more or less globular structure[s] of green mosses or liverworts, bound together with fine, dark-colored rootlets,” with a tubular, downward oriented entrance leading to a chamber lined with thick mosses and liverworts. They describe nests as being attached to “rough or pitted surface[s] of upright or fallen trunk[s] or vertical rocky cliff[s], or in a crotch, often near tip of drooping branch[es].”

Areta (2007), in Venezuela (ssp. rostratus), describes nests

similarly, as mossy balls with an entrance “usually facing downwards” and leading to an inner platform with an elevated front portion to contain the eggs. Nests (n = 35+) in Venezuela were built “attached to rocks, trees, or roots, usually in dark, hidden, and extremely wet areas, but never in what would strictly be considered a cavity” (Areta 2007). Areta (2007) also mentions the propensity of nests to be situated next to, and in the spray of, small waterfalls. He states that they are “usually” placed in sites above fast-flowing water, but confirms that, indeed, all nests were directly over water (Areta pers. com.). Nests in Colombia (Hilty & Brown 1986) (ssp. brunnescens) are described as being “wedged in crevice[s] between logs, bark, or rocks, and usually low.” In Ecuador (Marín & Carrión 1994, Greeney & Nunnery 2006, Greeney & Gelis 2007) (ssp. brunnescens), nests have been described in a generally similar fashion. Marín & Carrion (1994) provided the first Ecuadorian descriptions from four active nests and over 100 inactive nests in the northwest. Unlike previous descriptions, however, they mention that nests were lined with “fine rootlets and vegetable fibers.” Like other authors, they note the wide variety of substrates to which nests are attached, adding to previous descriptions that some are attached to the terminus of hanging clumps of tangled vegetation. Additional descriptions of Ecuadorian nests from the northwest and northeast generally agree with these

9

descriptions, but describe nest linings as simply “pale fibers” (Greeney & Nunnery 2006, Greeney & Gelis 2007). Ecuadorian nests are (with one exception, see Greeney & Gelis 2007) all described as being directly over water, often near small waterfalls, and always in extremely humid situations. Only two nest illustrations have been published; Skutch’s (1967) original, detailed illustration (reproduced in Skutch 1996), and a photograph provided by Marín & Carrion (1994). Of all described nests, Skutch’s Costa Rican nest is the only one which was not described as a completely formed ball, but instead built of several sections of moss which partially filled an inverted cavity and used part of the log itself to complete the nest’s globular structure. Skutch’s (1967, 1996) figure also clearly shows the lack of an internal lining, though it does illustrate the raised anterior portion of the egg cup, forming a lip to prevent eggs from falling out. Without exception across the Spotted Barbtail’s range, clutch size is given as two, and eggs are described as immaculate white (Skutch 1967, Stiles & Skutch 1989, Hilty & Brown 1986, Marín & Carrión 1994, Greeney & Nunnery 2006, Greeney & Gelis 2007). Mean dimensions (± SD) of eggs described in the literature (n = 22, all from Ecuador, Marín & Carrión 1994, Greeney & Nunnery 2006, Greeney & Gelis 2007) are 21.5 ± 0.5 by 16.7 ± 0.4 mm. Almost nothing has been described of Spotted Barbtail nestling care. Areta (2007) provides the only published information, noting only that nestlings defecate through the nest entrance in the absence of adults, the fecal sacs presumably removed by the running water below. The breeding season of the Spotted Barbtail in Costa Rica lasts from March-June (Stiles & Skutch 1989), and occurs at least during these months in Venezuela (Areta 2007). In Colombia, birds in breeding condition and nests with nestlings are reported from MarchAugust (Hilty & Brown 1986, Fjeldså & Krabbe 1990), with one report of a juvenile in November (Fjeldså & Krabbe 1990). In northwestern Ecuador, active nests have been found in nearly all months, suggesting year-round breeding in this area (Marín & Carrión 1994, Greeney & Nunnery 2006). Fewer nests are reported from northeastern Ecuador, but several active nests were observed in March, April, and one in December (Greeney & Gelis 2007), again suggesting a prolonged breeding season. Too few nests have been reported, however, to accurately access breeding seasonality in any particular geographic region. No breeding data have been published from either Peru or Bolivia.

10

The term life history is typcially defined as “a set of evolved strategies, including behavioral, physiological and anatomical adaptations, that more or less influence survival and reproductive success directly” (Ricklefs & Wikelski 2002), and understanding the evolution of life history traits has been the focus of much research (e.g., Roff 1992, Stearns 1992, Charnov 1993, Ricklefs 2000, Ricklefs & Wikelski 2002). In particular, because of their relatively well known taxonomy and general biology, factors driving avian nesting strategies have been widely investigated (e.g., Martin 1996).

Factors such as food

availability, weather, fecundity, longevity, predation, and parasite loads are all thought to influence avian nesting strategies in predictable ways (e.g., Conway & Martin 2000b, Ghalambor & Martin 2001, Martin et al. 2001. Martin 2002, Chalfoun & Martin 2007). For example, during incubation adult birds are thought to balance trade-offs such as egg temperature and development time, with predation risk and energy expenditure. Thus, the patterns of egg attendance exhibited by a wide variety of avian species are shown to be affected by complex interactions of multiple ecological, physiological, and evolutionary traits.

Yet, while comparing patterns among similar species may be very useful in

elucidating the underlying mechanisms driving incubation rhythms, often it is the extreme examples which provide us with the most significant insight (Lill 1979, Booth & Jones 2002, Carey 2002). So few data are available for tropical birds, however, that comparisons are difficult, and as rhythms vary across taxa, there is no real definition of “normal incubation.” In particular, there is currently no complete, or even extensive, dataset on incubation behavior in any tropical, understory furnariid. In this study I investigate the incubation rhythms of Spotted Barbtails and provide the most complete understanding of incubation behavior for any furnariid. Studies of avian nest-site selection have repeatedly shown there to be a great deal of variation in nest orientation preferences, as well as variability in the strength of patterns between species and/or nesting guilds (i.e., open cup, domed nest, primary/secondary cavity nesters). While some studies reveal little or no pattern of nest orientation (e.g., Albano 1992, Rendell & Robertson 1994, Tarvin & Smith 1995), many groups, representing a variety of nest-architectures, do show strong preferences in the orientation of their nests (Austin 1974, Inouye 1976, Walsberg 1981, Martin & Roper 1988, Balgooyen 1990, Bergin 1991, Nores & Nores 1994, Hooge et al. 1999, Mezquida 2004, Ardia et al. 2006).

11

Typically, nest orientation has been found to be correlated with either prevailing winds (Facemire et al. 1990, Norment 1993, Mezquida 2004) or sun exposure (With & Webb 1993, Yanes et al. 1996, Hartman & Oring 2003, Burton 2006), both of which may have a direct effect on the microclimate of the nest (Hartman & Oring 2003, Ardia et al. 2006). Thus, nest orientation and placement are important aspects of avian reproduction, and may affect embryonic development, hatching success, and nestling growth (Austin 1974, Viñuela & Sunyer 1992, Cook et al. 2003, Lloyd & Martin 2004, Burton 2006), the energetics of attending adults (White & Kinney 1974, Vleck 1981), and/or overall nesting success (Martin & Roper 1988, Filliater et al. 1994, Rauter et al. 2002). Most species of birds show a unidirectional pattern of nest orientation (e.g., Viñuela & Sunyer 1992), occasionally bidirectional (e.g., Bergin 1991). Sometimes, within-species nest orientation may shift during a breeeding season to capitalize on shifting weather or solar radiation patterns (Ricklefs & Hainsworth 1969, Finch 1983). Due, in part, to predictable shifts in abiotic factors, some birds have have even been found to vary nest orientation latitudinally (Burton 2007). I was able to find no studies, however, which document preferences in nest orientation for any tropical, understory species which regularly nests in deep shade within the protective confines of dense tropical vegetation. In this study I will investigate the nest orientation preferences of the Spotted Barbtail in northeastern Ecuador, providing the largest dataset for any ecologically or phylogenetically similar species. Despite the apparent aseasonality of many tropical regions, especially when compared to temperate latitudes, most tropical organisms show fairly pronounced seasonal shifts in abundance, foraging, movement patterns, and reproduction (e.g., Davis 1945, Medway 1972, Saul 1975, Wolda & Fisk 1981, Loiselle & Blake 1991, Flecker & Feifarek 1994).

Understanding the factors driving reproductive seasonality in tropical avian

communities involves distinguishing among complex interactions between weather, resource abundance, hormones, behavior, and other life history traits (Hau 2001, Wikelski et al. 2003).

While there is a great deal of evidence that many tropical birds breed

seasonally (e.g., Marchant 1959, 1960, Miller 1963, Snow & Snow 1964), we still understand little of the ecological factors driving the observed patterns (Wikelski et al. 2003). Most studies have suggested that avian breeding seasons in tropical regions are resource driven (e.g., Skutch 1950, Snow & Snow 1964, Poulin et al. 1992, Young 1994), and a few have shown correlations between the timing of breeding and the avoidance of

12

predation (Morton 1971, Dyrcz 1983, Ueta 2001). From an abiotic standpoint, the period during which birds nest is largely thought to be affected by seasonal weather patterns, generally due to the congruence between breeding periods and distinct shifts in temperature or rainfall. Some species, despite their near-equatorial breeding habits, have also been shown to physiologically respond to small shifts in daylength (Hau et al. 1998, Wikelski et al. 2000). Due, in part, to the difficulties of finding large sample sizes of nests in tropical forests, most studies of tropical species focus on one, or only a few, species (e.g., Miller 1959, Morton 1971, Gradwohl & Greenberg 1982, Young 1994, Wikelski et al. 2000, Ahumada 2001). Additionally, many rely on secondary evidence of breeding activity, such as reproductive condition or molting (e.g., Levey 1988, Loiselle & Blake 1991). Those studies which present direct data at a community/regional level (e.g., Marchant 1959, 1960, Snow & Snow 1964, Greeney & Nunnery 2006, Greeney & Gelis 2007, 2008a), are generally far from complete in their taxonomic breadth. Indeed, most of these are also anecdotal, and do not provide sufficient data on any one species such that concrete hypotheses concerning the causes of seasonality may be explored. Due to this paucity of information for most tropical species, to get a picture of avian breeding seasonality in any one species, we are forced to piece together scattered information from across a species’ range (for example, see data and references in Hilty & Brown 1986, Stiles & Skutch 1989, Fjeldså & Krabbe 1990, Hilty 2003). Such data suggest that breeding seasons within a species vary geographically, as would be expected, usually correlated with local rainfall patterns. Additionally, for those tropical studies which address seasonality at a local level, most have focused on locations with fairly extreme temporal changes in rainfall (Lack 1950, Voous 1950, Marchant 1959, 1960, Ramo & Busto 1984, Cruz & Andrews 1989, Poulin et al. 1992, Best et al. 1993, 1996), and comparatively few have looked at relatively aseasonal low-latitude locations (but see Miller 1963, Moreau 1950). In summary, our understanding of the seasonal breeding patterns of most equatorial, evergreen forestdwelling species is greatly hindered by the lack of large data sets, and the propensity of researchers to focus on highly seasonal habitats where large samples may be gathered more quickly.

13

For the few species that have been studied on the east slope of the equatorial Andes, which is relatively aseasonal when compared to most tropical habitats (Stireman et al. 2005), almost all show fairly strong monomodal preferences in their breeding activity (e.g., Greeney 2007, Greeney & Gelis 2008, Greeney & Halupka 2008, Halupka & Greeney in review, Greeney et al. 2008a). Interestingly, there is an emerging pattern that suggests that the microhabitat where breeding occurs may influence the timing of a species’ breeding efforts, exemplified by the markedly seasonal breeding habits of the taxanomically diverse community of species nesting in montane bamboo thickets (Greeney & Sornoza 2005, Greeney et al. 2005, Gelis & Greeney 2006, Martin & Greeney 2006, Greeney et al. 2007, Greeney & Miller 2008, Juiña et al. 2008). In the case of this guild of habitat specialists, it has been suggested that avian breeding may be timed so as to coincide with the least amount of vegetative bamboo growth, thus insuring a less dynamic (i.e., more structurally sound) microhabitat and nesting substrate (Greeney & Miller 2008, Greeney et al. 2008b). As the species represented in this “bamboo-nesting guild” include a wide range of dietary and foraging specialists, from granivores to insectivores and frugivores (see references above), it thus seems unlikely that the observed reproductive seasonality is a result of shifting resource abundance. While a few studies have pointed out slight variation in the within-species initiation of breeding based on microhabitat (e.g., Wikelski et al. 2003), we know very little about how micro-habitat choice for nesting may affect (or be affected by) breeding seasonality. In this study I explore the seasonal breeding patterns in a 0° latitude population of Spotted Barbtails, which show very specific microhabitat nesting preferences (strictly riparian). I examine the results in the context of other avian species in the area, in particular other riparian-nesting specialists. While general life-history theory predicts that organisms should grow at an infinite rate (Sterns 1992), a more realistic view acknowledges that physiological limitations constrain species to certain maximum rates (West et al. 2001). Within these limitations, however, in order to maximize offspring output, we would predict that organisms should grow at the fastest rates possible (Ricklefs et al. 1998). Nevertheless, organisms face a variety of trade-offs with rapid growth, often having to balance growth rates against a variety of other physiological costs (e.g., Metcalfe & Monaghan 2001). The degree to which avian developmental rates are affected by other aspects of a species’ life history, however, remain somewhat unclear (Ricklefs et al. 1998, Remeš & Martin 2002). Remeš

14

& Martin (2002) showed that growth rates of altricial birds were correlated with ecological and life history traits such as predation rates, mode of foraging, and clutch size. Additionally, there is a well established pattern of latitudinal variation in growth rates, with those of tropical species being generally slower than those of their temperate counterparts (see data and references in Ricklefs 1968, 1973, 1976, Oniki & Ricklefs 1981, Remeš & Martin 2002). Despite the relative abundance of data for growth rates of altricial birds in general (see references in Remeš & Martin 2002), and the numerous hypotheses which have been put forth to explain variation in avian developmental strategies (Stark & Ricklefs 1998), emperical data for many tropical species are lacking. In particular, we know almost nothing of developmental rates in the family Furnariidae, and there are no quantified data available for any species. Additionally, despite their value to field researchers and students of comparative life history, detailed descriptions of plumage development for tropical birds are few, and most species’ growth curves are based on very small sample sizes. This is likely a result of prior constraints imposed by the costs of field photography and of publishing colored plates through traditional methods, in conjunction with a recent disregard of descriptive natural history as a valuable contribution to the scientific process. With recent technological advancements, most especially the development of digital photography and the on-line publication of a variety of journals, a few studies have begun to document and illustrate the details of nestling feather development in wild birds (e.g., Greeney 2007, Greeney et al. 2008b). As photographic and quantified descriptions of plumage development and weight gain are invaluable field tools for research on comparative life history strategies and basic studies of natural history, I hope that this may become a more frequently investigated aspect of tropical avian biology. In this study I present the first developmental growth data for any species of furnariid, as well as present the most detailed account of ontogenetic feather development in the family. As mentioned above, an important goal of recent ecological research has been an attempt to understand why organisms vary in their life history traits (e.g., Roff 1992, Stearns 1992, Charnov 1993, Ricklefs 2000, Ricklefs & Wikelski 2002). Much of this work has focused on sources of stage-specific mortality (e.g., Charlseworth 1980, Reznick & Bryga 1987, Reznick et al. 1990, McCleery et al. 1996), and how this affects tradeoffs in foraging, development, and behavior.

For studies of avian life-history evolution in

particular, predation has long attracted a great deal of attention (e.g., Skutch 1949, Case

15

1978, Dyrcz 1983, Martin 1992, 1995, 2002, Martin et al. 2000, Chalfoun & Martin 2007). Predation rates, especially on nests, have been implicated as driving forces in the evolution of a number of life-history traits such as development rates, nesting cycle duration, nest placement, clutch size, and parental care behaviors (e.g., Ricklefs 1969, Bosque & Bosque 1995, Martin 1995, 1996, Blanco & Tella 1997, Bogliani et al. 1999, Conway & Martin 2000b, Ghalambor & Martin 2001, Remeš & Martin 2002). Often, however, theoretical debates are hindered by lack of empirical data (Robinson et al. 2000) and, while multiple studies have shown that parent birds are extremely sensitive to the presence of predators (e.g., Briskie et al. 1999, Ghalambor & Martin 2002, Fontaine & Martin 2006, Fontaine et al. 2007), not all theories on how predation drives life history characteristics are supported by models and empirical data (e.g., Ricklefs 1977, Young 1996, Styrzky et al. 2005). In particular, data concerning nesting success in most tropical species remains incomplete, retarding our ability to make solid comarisons between and among species varying in life history strategies (Robinson et al. 2000). While all aspects of the breeding ecology of Spotted Barbtails have undoubtedly evolved under the independent influences of phylogenetic constraints (Skutch 1996, Irestedt et al. 2006), physiological limitations (Ricklefs & Wikelski 2002, Wikelski et al. 2003), and multiple ecological and abiotic pressures (e.g., Moreau 1944, Lack 1947, Skutch 1949, Ashmole 1963, Stark & Ricklefs 1998, Ardia et al. 2006, Chalfoun & Martin 2007), in this study I will focus on describing their basic breeding ecology and documenting predation rates and causes of nest failure. Given that Spotted Barbtails have low nesting success (22%; see Nesting Success), I will test a priori predictions concerning their expected reproductive strategies.

In particular, within a theoretical avian life history

evolution framework, I will discuss nest placement, architecture, construction behavior, and orientation, as well as incubation behavior, timing of reproduction, and nestling growth and development.

16

SITE DESCRIPTION AND STUDY METHODS Study site. I studied the nesting of Spotted Barbtails (here after barbtails) along streams in the vicinity of the private reserve of Cabañas San Isidro, adjacent to the Yanayacu Biological Station & Center for Creative Studies (00°36’ S, 77°53’ W, 1900-2400 m). Yanayacu lies 5 km west of Cosanga, just north of the Huacamayos Ridge, where the Cosanga River valley makes a sharp northward turn from its easterly direction. Scant data are available from the immediate area, but Ecuadorian government rainfall data from nearby Baeza, and reports from previous studies in the area (Valencia 1995, Stireman et al. 2005, Greeney et al. 2006b, Guayasamin et al. 2006), indicate that rainfall in the area ranges from 100 to 300 mm per month, with a drier season lasting roughly from September to February (Figure 3). Total annual rainfall ranges from 2300 to 3500 mm, and mean monthly temperatures range from 15 to 17°C. While there are no climatological data from the area on the south side of the Huacamayos Ridge (7 km east of Yanayacu), daily observations during the past 9 years suggest that this area receives higher levels of precipitation, and that my study area may be inside a small rain shadow.

FIGURE 3. Rainfall data from the Baeza-Cosanga area of Napo Province, Ecuador. Historical data are compiled from various internet sources for a total of 19 years from 1953-1982. Data are presented in a unitless form, showing rainfall relative to other months.

Forest in the area (Valencia 1995, Velencia et al. 1998, Greeney et al. 2004, 2006b, J. Homeier pers. comm.) is typical of lower montane forests or mid-elevation “cloud

17

forests” on Ecuador’s northeast Andean slopes (Harling & Sparre 1979, Neill 1999). The canopy in the region ranges from 20 to 35 m, and is dominated by Erythrina (Leguminaceae), Hyeronima (Euphorbiaceae), Meriania, Miconia (Melastomataceae), Ocotea, Nectandra (Lauraceae) (Neill & Palacios 1997).

The specific area where I

collected most of my data is a largely flat area to the north of the Cosanga River, bordered on the opposite side by the steep slopes of the Antisana Volcanoe (Figure 4). This flatter area is floristically less diverse than surrounding slopes (J. Homeier pers. comm.), and dominated by Vismia (Clusiaceae), Nectandra (Lauraceae), Critoniopsis (Asteraceae), Miconia, and Erythrina.

The understory is composed predominantly of Genoma

(Aracaceae), Piper (Piperaceae), along with various small trees and shrubs in the families Solanaceae and Rubiaceae. Epiphytic growth is abundant, and most plants host thick layers of mosses and liverworts (Figure 5).

FIGURE 4. View from the Yanayacu Biological Station showing large flat area where my studies were conducted.

18

FIGURE 5. Photograph of the interior of the forest at Yanayacu.

General field work. I searched for nests and observed adult barbtails by walking in, and along, over-20 km of small streams that wind their way through a roughly 400 ha patch of primary forest, that is relatively flat by Andean standards (Figure 4). I located nests by either systematically searching, accidentally flushing adults, or by observing adult behavior and following adults to their nests. I marked each nest individually with unique voucher numbers, with flagging tape placed 2-4 meters from the nest. Once I encountered and marked a nest, it retained its original voucher designation throughout the study, regardless of how many times it was reused. I flagged all nests, even those that were not active at the time of discovery. At each visit to a nest, I recorded the time to the nearest 15 min., the date, and any pertinent observations on the nests’ contents and architectural status (i.e., with lining or without, intact or falling apart). I first wrote the details of each nest visit into a field notebook, and later transcribed them into nest-card computer files, which were maintained throughout the “life” of the nest. In this way, I was able to quickly view the entire history of a nest, including the details and fate of each breeding attempt.

19

Using calipers accurate to the nearest 0.1 mm, I measured eggs upon their initial discovery. I used only eggs within 1 day of laying to calculate mean fresh weights, weighing eggs to the nearest 0.001 g. At each weighing, in order to account for uneveness, I calibrated the scale with a standard 50g weight. I then placed each egg on the scale 3 times, and took the mean of all three weighings. I removed eggs every 4-5 days to visibly monitor their development by holding them up to the light. Evaluation of nesting success. In order to calculate overall nesting success, I included only nests in which I observed the entire nesting event, beginning with laying of the eggs and ending with failure or fledging (n = 50 nests). To compare nesting success between various samples (i.e., years, seasons, etc.), I calculated daily survival rates (DSR) for each sample using methods first developed by Mayfield (1975). I checked active nests every 2-3 days, and any nests which I was unable to monitor this frequently were dropped from the analysis. DSRs were calculated by using the following formula: DSR = 1-(F/T) where F is the number of nests which failed during the observation period and T is the sum of the number of days that all nests within the sample survived. I compared DSRs between samples using methods developed by Johnson (1979). I calculated predicted fledging success by raising the DSR of a sample to the power equal to the duration of the entire nesting cycle (2 days laying. 28 days incubate, 20 days nestling; see General Natural History). Upon encountering a nest that had failed, I carefully searched the surrounding area and the inside of the nest for any remains of its contents (i.e., blood, egg shells, feathers). I also inspected the nest and recorded any visible signs of disturbance. Quantification of nest site selection and nest architecture. I include observations on 340 nests, found from May 2001 to December 2007 at my principle study site. In addition, I use information from nests at the following localities: Sumaco Volcano (1700-2400 m), Napo Province, northeastern Ecuador (n = 5; Greeney & Gelis 2007); Mushullacta Community Reserve (00°50’ S, 77°34’ W, 1000-1150 m), Napo Province, northeastern Ecuador (n = 2); Tandayapa (00º00’ N, 78º41’ W, 1350 m), Pichincha Province,

20

northwestern Ecuador (n = 1); vicinity of Mindo (00º04’ S, 78º43’ W, 1600-1800 m), Pichincha Province, northwestern Ecuador (n = 5; Greeney & Nunnery 2006); Mindo Loma (00º00’ S, 78º44’ W, 1750-1800 m), Pichincha Province, northwestern Ecuador (n = 4); Tapichalaca Biological Reserve (04º30’S, 79º10’W, 1800-2500 m), Zamora-Chinchipe Prov., southeastern Ecuador (n = 2).

FIGURE 6. Photograph of a nest of Spotted Barbtail which has been cut in half to show the inner lining of pale fibers. The inner chamber and entrance tube of the nest have been outlined in black. Inset shows the measurements taken while describing barbtail nests: a, outside height; b, entrance height; c, entrance tube length; d, entrance hood overhang.

Whenever possible, I recorded the following measurements of each nest: height above the water’s surface (n = 304), outside width (n = 50), outside height (Figure 6a; n = 50), entrance width (n = 42), entrance height (Figure 6b; n = 42), entrance tube length (Figure 6c; n = 40), entrance overhang (Figure 6d; n = 42). In addition I recorded the primary and secondary substrates to which nests were attached. For example nest “A” was attached to vines (2° substrate) hanging below a horizontal tree trunk (1° substrate). Often nests were attached directly to a primary substrate. I also scored the mobility of each nest as “hanging” or “fixed.” If I was able to easily change the orientation of the nest entrance by more than 10°, then I considered the nest as “hanging” and, if not, I considered it

21

“fixed.” At each nest I also described what portions of the nest were attached to a substrate (i.e., attached by top and one side, or by back and top). Incubation rhythms.

At every visit to barbtail nests during incubation (n = 344), I

recorded the time of day, to the nearest 15 min, and noted the presence or absence of an incubating adult. Subsequently, at 3 nests, I used HOBO dataloggers (Onset, Pocasset, Massachusetts) to monitor the entire incubation period. Probes were placed 1-2 mm below the inner surface of nest linings, while the external probe was placed inside a nearby cavity, under leaf litter, or in a similarly sheltered location. I programmed loggers to record temperatures every 2 min and, by comparing the difference between internal and external probe measurements, I recorded the duration of adult presence in, and absence from, the nest. At several nests that I attempted to monitor using this method, the eggs dissappeared during the observation period, and for quantification of incubation rhythms I used only the three nests for which I obtained complete data. Using tripod-mounted cameras placed 5-8 m from nests, I videotaped adult activity during the morning and evening on 3 days at each of 4 nests, while simultaneously recording nest lining temperatures using a data logger. I used video data to directly record presence of adults in the nest, allowing me to corroborate and interpret logger data. I inserted loggers into nests either just before clutch initiation, or soon after clutch completion. Of the three nests monitored through the entirety of incubation, one held eggs from 20 September to 19 October 2006, one from 5 November to 6 December 2006, and one from 22 October to 20 November 2007.

As a measure of diurnal % incubation, I

included the period between 06:00 and 18:00, roughly sunrise to sunset at my study site, and comparable to other work on avian incubation in the area (Greeney 2006a; Greeney et al. 2006a, 2008d). Nest orientation. Within my principle study area, I recorded the orientation of the nest entrance (hereafter nest orientation) at 196 barbtail nests. At each nest, I recorded the compass orientation of the nest entrance, as well as the direction of stream flow, to the nearest 5°. I took orientation readings by holding a compass directly below the nest, and orienting with the nest entrance or with the flow of water directly below. Thus, the direction recorded for water flow did not always correspond to the overall direction of the

22

stream, as it was often affected by a bend in the stream or large boulders upstream from the nest. A wide variety of both circular and linear statistics have been used in previous studies of nest orientation (e.g., Ricklefs and Hainsworth 1969, Austin 1974, Norment 1993, Burger & Gochfeld 2005, Ardia et al. 2006). As circular distributions do not always conform to the assumptions of many of these tests (Cain 1989, Zar 1999), most recent studies of circular data rely on one of several analyses. The three most widely used circular statistics show marked differences in their ability to detect non-uniform dispersions with varying sample sizes, vector dispersions, and directional modality (Bergin 1991). Therefore, I tested uniformity of direction around 360° using three separate tests. Raleigh’s test (Z: Batschelet 1981, Zar 1999) and Watson’s test (U2: Batschelet 1981, Zar 1999) are frequently used statistics for analyzing patterns of nest orientation (Rafael 1985, Ardia et al. 2006, Burton 2006). Both of these tests, however, were designed to test unimodal patterns, whereas Rao’s test (U: Rao 1976, Batschelet 1981) performs much better with polymodal distributions (Bergin 1991). Because nest orientation in relation to the cardinal directions showed a weak unimodal distribution with high vector dispersion, while nests showed strong polymodality in relation to stream flow, and the stream showed strong unimodal directionality, I feel it is appropriate to report all three tests. For each analysis I present mean vector length (r). Mean vector length is a unitless measure (0-1) of the dispersion of the data, with a value of 0 being widely dispersed (uniform) and 1 being tightly concentrated. I used Oriana 2.0 (Kovach, Wales, U. K.) for calculating all circular statistics.

Breeding seasonality. For each active nest I recorded the date, status, and contents of the nest. While my effort was not constant every year, across all 7 years effort was evenly distributed across all months. Thus, while I cannot make between-year comparisons, my data represent true intra-annual patterns. I found 166 active barbtail nests during this study. I treated each breeding attempt, either in the same nest or in a freshly constructed one, as an individual data point, using the observed or calculated date of clutch initiation to place each nest within a specific twoweek period of the year. Combined with extensive experience at aging eggs and nestlings

23

of this and other tropical birds, I used an incubation period of 28 days (plus 3 days of laying) and a nestling period of 20 days to estimate clutch initiation for all nests found after incubation had begun. Using these same estimates, I predicted hatching and fledging dates for each nest. Because date is actually a circularly distributed type of data, and thus may not conform to the assumptions of many of linear tests (Cain 1989; Zar 1999), I used Oriana 2.0 (Kovach, Wales, UK) for my calculations as described above under “Nest orientation.”

Nestling growth and development.

I weighed nestlings to the nearest 0.01 g and

photographed them every several days until 1-2 days prior to fledging. I present data for only nests where I observed hatching, and thus was certain of nestling age. I used only nests with two nestlings (the usual clutch) and, to avoid pseudoreplication, used the mean of the weights of nest-mates to generate the growth curve. I fitted the data to a logistic growth equation, relating body mass to age using a least squares method, as proposed by Ricklefs (1967, 1976): mass(age) = A/[1+e(-K(age-t))] In this equation, A, K, and t are the parameters of growth and e is the base of the natural logarithm. I calculated the inverse measure of growth rate, that is, the time required to grow from 10% to 90% of the asymptote, using the formula 4.4/K. In all, I present data from 16 nestlings at 8 nests. Using mist-nets, I captured 21 adults from the same population, also weighing them to the nearest 0.01g.

24

RESULTS

GENERAL NATURAL HISTORY

Here I summarize unpublished information on barbtail reproductive ecology, and fill gaps in our current knowledge. The following observations are meant to provide information necessaray to contextualize the rest of this study. Nest construction, incubation, and nestling and fledgling care are performed by both sexes. Interestingly, successful and unsuccessful nests are reused for multiple breeding attempts, both within and between years. On at least 10 occasions I observed clutch replacement, in the same nest, within 20 days of the loss of the first clutch. Currently, I have four nests which are still active after seven years of observation, and three nests still active after six years. I observed one nest which held at least four clutches during the course of a single year! Two of these four breeding attempts failed, and I do not know the fate of the other two.

In general, it appears that pairs may sometimes produce two

successful broods a year, but my lack of data on banded individuals precludes me from confirming this. Similarly, I do not know if nests that are reused are occupied by only one pair, or if several pairs utilize the same nest. Paul Schwartz (in Skutch 1967) found up to three adults using the same dormitories in Costa Rica, thus it remains to be seen if, and how, breeding nests are shared between individuals.

25

FIGURE 7. A Spotted Barbtail (Premnoplex brunnescens) roosts for the night in a shallow cavity hollowed out in a hanging clump of epiphytes. Photo by Rudolphe A. Gelis.

I also confirmed that this species builds non-breeding dormitory nests in my area (Figure 7), similar to those described by Skutch (1967). They range in complexity from a simple cavity, seemingly hollowed out of existing moss, to partially constructed mossy balls. Often it appears that a natural cavity, usually in a hanging clump of moss and epiphytes, is amplified with a few additions of moss (Figure 7). Dormitories are found in a variety of situations along streams, varying from crevices in tree trunks, rock faces, or dirt embankments, to naturally hanging clumps of moss and epiphytes. They differ from breeding nests in being smaller and lacking any sort of lining or egg cup. Most, however, have a slightly elevated front inner rim, presumably providing a perch for the adult. This rim is often formed by a naturally occurring feature of the attachment point, such as a horizontal stick or a rock lip. While breeding nests are almost always built over running water (see below), dormitory nests are often built up to 10 m from the nearest stream. I found breeding nests exclusively along streams, despite extensive nest searching away from streams.

While all nests were in well-shaded areas, barbtails seem fairly

tolerant of moderate disturbance, so long as most of the canopy remains intact. I found several nests as little as 20 m from forest edges. As expected from previously published

26

descriptions of nest site selection (e.g., Marín & Carrión 1994, Greeney & Nunnery 2006, Areta 2007), I found nests attached to a wide variety of substrates, almost exclusively, however, over water. I found nests attached to substrates ranging from earthen banks and rock ledges to horizontal or vertical tree trunks. Some nests were partially or entirely in an earthen or wooden cavity while others were hanging in the open attached to mossy clumps of vegetation. Nests, especially older ones, are extremely robust and thick-walled, almost impenetrable, even with a sharpened pencil, and the permanency and stability of barbtail nests occasionally led to their use as a nesting substrate by other species.

On two

occasions, the tops of currently inactive barbtail nests were used to support the nests of Slaty-backed Nightingale Thrushes (Catharus fuscater).

One nest was used as an

attachment point for the nest of Speckled Hummingbird (Adelomyia melanogenys), and another supported the nest of a Green-fronted Lancebill (Doryfera ludovicae). On one occasion the nest of a Grey-breasted Woodwren (Henicorhina leucophrys) nest was attached to the outside of a barbtail nest. At another nest, that appeared to have been partially torn open by a predator, the nest was taken over by a pair of Slaty-backed Chattyrants (Ochthoeca cinnamomeiventris), which entered their nest through the partially destroyed entrance of the barbtail nest structure. While it has only been well-documented once (Freeman & Greeney 2008, Greeney 2008a), I suggest that barbtail nests are occasionally usurped by Long-tailed Tapaculos (Scytalopus micropterus) in our area. Specifically, I observed 2 clutches of eggs, in previously confirmed barbtail nests, that were too large to belong to barbtails. While I did not confirm their ownership at the time, in retrospect, I now believe the first 2 clutches also belonged to tapaculos. In general, barbtails appear relatively passive when interacting with other species at their nesting sites. In one case, a pair of barbtails beginning construction of either a dormitory or breeding nest under a horizontal log was displaced by a female Green-fronted Lancebill. The lancebill eventually built her nest hanging from the lip of the partially constructed barbtail nest. Construction by the barbtails was subsequently abandoned and the nest was never built, even after the lancebill had finished nesting. On another occasion a barbtail nest under construction was abandoned when a Rufous-breasted Flycatcher (Leptopogon rufipectus) began constructing its own nest only 20 cm away.

Because

barbtails, Green-fronted Lancebills, and Rufous-breasted Flycatchers in the area favor similar nesting sites (Dobbs & Greeney 2006, Greeney et al. 2006b), the usurpation of

27

barbtail nesting sites by these two species may be more frequent than these two observations suggest. In addition, Highland Motmots (Momotus aequatorialis) and several species of Thripadectes treehunters in the area also prefer nesting in tunnels in streamside banks with overhanging lips (Greeney et al. 2006b, pers. observ.). While I have found many barbtail nests in close proximity to motmot or treehunter tunnels, I never found barbtail nests active concurrently with nests of these other species. Because both motmots and treehunters reuse old nest tunnels or excavate another in close proximity to previous nests (pers. observ.), it is possible that nesting activity of these species may interfere with that of barbtails.

FIGURE 8. Eggs of the Spotted Barbtail from northeastern Ecuador. A, B, & C, are from the Yanayacu Biological Station, 2050 m, while D is from the Parque National Gran Sumaco, 1750 m. They have the following measurements: A = 23.3 x 17.3 mm; B = 24.5 x 17.8 mm; C = 20.3 x 17.3; D = 23.5 x 17.0.

Clutch size was almost exclusively 2 (n = 164 clutches), with only one nest found with three eggs. All eggs (n = 307) were immaculate white, but often slightly stained with vegetative liquids or blood and fluids deposited during laying. Mean dimensions (± SD) of 269 eggs were 22.3 ± 1.1 by 17.1 ± 0.7 mm. Eggs were variable in shape, ranging from nearly round to oblong (Figure 8). Mean fresh weight (before visible development) of 60 eggs was 3.49 ± 0.25 g. Eggs were laid roughly 48 hours apart (n = 18 nests) and always in

28

the morning before 09:00 (n = 27 eggs). There was no difference in size between eggs based on laying order (paired t-test: t = 0.23, df = 10, p = 0.82, n = 11 clutches). Eggs laid during the drier months, however, were significantly heavier (two-tailed t-test: t = 2.16, df = 58, p = 0.035). Incubation period (starting from clutch completion) ranged from 27 (3 nests) to 31 (1 nest) days, with one nest taking 30 days to hatch, one taking 29 days, and 4 nests taking 28 days. In all cases where I observed hatching (n = 24), both eggs hatched on the same day. Nestling barbtails are born with pinkish-orange skin and sparse pale-grey down (See Nestling Growth and Plumage Development). Nestlings fledge after 19-22 days (19 days at 1 nest, 20 days at 4 nests, 21 days at 1 nest, 22 days at 1 nest). Both adults provision nestlings, bringing a variety of invertebrate prey including larval and adult insects, snails, and earthworms. As observed by Areta (2007), in all cases older nestlings defecated out of the nest entrance in the absence of adults. Adults removed or ate the fecal sacs of only very young nestlings too small to maneuver themselves into the position necessary to eject feces out of the entrance. Until ca. 4 days of age, nestlings remain oriented with their heads opposite the nest entrance, presumably to facilitate adult removal of fecal sacs. After this, they sit with their heads oriented towards the entrance, presumably turning about to defecate.

29

THE NEST: ARCHITECTURE, PLACEMENT, AND CONSTRUCTION Nest architecture. Although nest placement and attachment varied considerably (see Nest placement), most nests (n = 287) were of the same basic architectural design. They were large, globular, mossy balls with a downward facing entrance tube (Figures 9-10).

FIGURE 9. An adult Spotted Barbtail peers out of the nest entrance at the Yanayacu Biological Station, Napo, Ecuador, 2000 m.

The downward entrance into the nest chamber was slightly wider than tall, with the entrance flaring laterally, but with the height of the entrance maintained along the length of the tube. The inner chamber held a thick cup of pale fibers with an elevated front lip that served to contain the nest contents as well as to form the inner portion of the entrance tube (Figure 6).

30

Figure 10. An adult Spotted Barbtail delivers and adult owl fly (Ascalaphidae) to nestlings at the Yanayacu Biological Station, Napo, Ecuador, 2050 m.

Although this description applies to the majority of nests, many were variously constructed such that one or several portions of the outer structure were formed by part of the surrounding substrate, similar to the nest illustrated by Skutch (1967, 1996). Thus, large portions of the inner chamber were often devoid of moss, and instead formed by the clay or wood to which the nest was attached (Figure 11a). One extreme example of this was a nest built wedged into a vertical crevice in a horizontal log. Only the front, top, and bottom portions of the nest were composed of moss, whereas both sides and the back were formed by the surrounding wood. Another example was a nest built just inside the entrance to an abandoned Highland Motmot (Momotus aequatorialis) nest burrow. In this case the front of the nest was built as described above, forming a hood over the downward entrance tube, that opened just outside and below the burrow entrance. Other than this, moss had been tightly wedged into the inner portion of the tunnel, effectively blocking it off and creating a nest chamber formed on the sides, top, and bottom, by the earthen walls of the motmot tunnel. On the bottom of the nest chamber, resting directly on the floor of the tunnel, was the typical inner egg-cup of pale fibers. In two nests, both inside shallow, natural, earthen cavities, no moss was used in construction, and the nest consisted of only a thick cup of pale fibers that was entered directly through the cavity opening.

31

FIGURE 11. Drawings in cross-section of Spotted Barbtail (Premnoplex brunnescens) nests illustrating a variety of nest placement strategies: a, fixed nest completely inside an earthen cavity with only sparse construction of the outer mossy shell; b, fixed nest built partially into a shallow cavity in the ceiling of a clay overhang; c, fixed nest built partially into a shallow cavity in the wall of a clay bank; d, hanging nest attached to a thin rootlet under a clay bank.

Using only nests fully formed with moss, mean dimensions (n = 50, cm ± SD, Figure 6) were: outer nest height, 15.3 ± 1.7; outer nest width, 16.1 ± 2.0; entrance width, 5.4 ± 0.8; entrance height 3.1 ± 0.4; entrance hood overhang, 5.9 ± 1.4; entrance tube length, 8.2 ± 1.7; entrance tube width, 3.7 ± 0.6. Because nests were used for multiple breeding attempts, I was able to measure the inner chamber (with lining removed) of only one nest. It measured 11.5 cm tall by 9 cm wide. By carefully removing the lining from 4 nests, I measured the dimensions of the egg cups. These were generally 1-2 cm thick with a mean inner width of 5.3 ± 0.5 cm by 3.0 ± 0.4 cm deep.

32

FIGURE 12. Nesting habitat, nests, and nesting sites of Spotted Barbtail at Yanayacu Biological Station, Napo, Ecuador, 1950-2200 m. White circles surround the nest in each photo: a, nest hanging by a thin vine adjacent to a small waterfall; b, nest hanging by a narrow attachment of moss and vines at the edge of a small rock overhang; c, fixed nest built partially into an earthen cavity of an overhanging bank; d, fixed nest attached from above to an overhanging dirt bank.

Nest placement. All nests were found along well-shaded streams (Figure 12), the majority within primary forest. Of 316 nests, only five were not located directly above flowing water. The streams below these nests, however, appeared to have recently shifted direction, and it is likely that the nests were over water at the time of construction. One such nest was never occupied during my study, perhaps because shifting water caused abandonment. Two others were not occupied after initially being found active, and I do not know the fate of the final two. Mean nest height above the water was 1.7 ± 0.9 m (range = 0.25 - 5 m, n = 304). Nests showed a remarkable variety of substrates and means of attachment, a few of which are illustrated in Figures 11-15. Of 305 nests, 74% of nests were considered “fixed” and attached directly to an immobile primary substrate. The other 26% were attached to a mobile secondary substrate (such as a hanging clump of epiphytes), where I could easily alter the orientation of the opening, and thus were considered “hanging.” As primary substrates I recorded horizontal tree trunks (36%), tree trunks angled at ≈45° (30%), soil banks (17%), rock faces (13%), and vertical tree trunks (4%), whereas secondary substrates included moss epiphyte clumps (85%), single thin vines (8%), multiple thin vines (3.5%), and multiple small branches or tangled vines (3.5%). Of 291

33

nests, 202 were located directly on the primary substrate (Figures 11a-c, 12c-d, 13, 15), with the remaining 89 nests attached to secondary substrates (Figures 11d. 12a-b, 14). Although many of the nests fixed into hanging epiphyte clumps were likely built into some sort of pre-existing vegetative niche, 20 of 291 nests were built into actual cavities (Figures 11a, 15) such as crevices in rotting wood (45%) or soil banks (55%). One of these, as described previously (see Nest architecture) was built into the entrance of an abandoned Highland Motmot nest tunnel. In six cases, nests were attached, to some degree, and usually sharing at least one wall with, a previously constructed nest (Figure 13b).

FIGURE 13. Nests of Spotted Barbtail at Yanayacu Biological Station, Napo, Ecuador, 1950-2200 m, arrows indicate nest entrances: a, fixed nest attached by the back and top to a mossy earthen bank; b, two fixed nests built adjacently, sharing a wall between them and attached by the back to an clay overhang; c, fixed nest attached by the back to a large clump of earth broken away from a dirt overhang; d, fixed nest attached by the side and the bottom to a large clump of mosses and epiphytes.

34

FIGURE 14. Nests of Spotted Barbtail at Yanayacu Biological Station, Napo, Ecuador, 1950-2200 m, arrows indicate nest entrances: a, hanging nest attached by the top to the end of a large vine, the epiphytic bromeliad is growing on the outer wall of the nest; b, hanging nest attached narrowly at the top to the broken end of a thin, horizontal branch; c, hanging nest attached to a single thin vine; d, hanging nest attached by a thin sheet of moss to a U shaped vine which has become dislodged from its original large mossy tangle of vines and epiphytes.

35

FIGURE 15. Nests of Spotted Barbtail built into natural cavities at Yanayacu Biological Station, Napo, Ecuador, 1950-2200 m, arrows indicate nest entrances: a, fixed nest wedged into a vertical crevice in a horizontal tree trunk; b, fixed nest built into the entrance of an irregular earthen cavity in a vertical bank; c, fixed nest built into a small niche in a rock wall; d, fixed nest built into a downward opening cavity in the ceiling of a dirt overhang (photo taken from below looking straight up).

36

Nest construction behavior. Both sexes participate in nest construction, and from the first adult arrivals with beaks full of moss, the entire process can take over one year to complete. At one nest, that consisted of only a few scraps of fresh moss in early October 2006, the first clutch of eggs had not been laid as of January 2008. I am still observing several nests which were started before October 2006 and were still not finished as of January 2008. In three cases, however, I found nests that were started and finished within 4 months. Six additional nests were completed within seven months from when first found, but in all cases there was already a substantial amount of moss present, forming most of the basic outer nest structure. Most nests I observed, probably long after their completion, were composed of living mosses and liverworts that had become so thickly interwoven over time they appeared to be part of the surrounding vegetative material and had become host to epiphytic orchids and bromeliads. Often nests appeared to have been there so long that inner portions of the mossy walls had nearly decomposed into soil, mixed with a thick network of living rootlets from surrounding plants. Nest construction can be divided into three phases. The first involves placement of live mosses and liverworts into holes and cracks surrounding the area that will eventually hold the nest. Depending on the nest site, there may be cracks in the surrounding clay, holes in the wood or, in the case of nests completely outside of cavities, irregularities in the surrounding epiphytic vegetation. The mosses and liverworts used to fill irregularities surrounding the nests’ eventual attachment point create a partial cavity or a more regular mossy surface to which the nest will be anchored. There is frequently a 1 week to twomonth delay before the second phase of construction begins. The second phase involves formation of the outer nest structure, built of mosses, liverworts, and rootlets. This phase is often protracted, lasting 2-8 months. During this phase the birds complete the globular outer structure of the nest. This is usually accomplished by first building a simple ring of moss. This two-dimensional ring is then added to, both forwards and backwards, until the three-dimensional, spherical structure of the outer nest is formed. Once the globular structure is complete, adults add material to the front, upper portion, and work forward until the material forms the downward-facing entrance tube. At the single nest where I observed this entire process, the outer portion of the nest was completed in ca. eight months. After these eight months of observation, I blocked the entrance with a leaf, which remained untouched for three months.

However, after three months (11 months after I first

37

discovered the beginnings of this nest), adults began to visit the nest regularly, as evidenced by the continual removal of the leaf placed in the entrance. On several occasions I found an adult roosting in the nest, and I suspect that the nest was used as a nocturnal roost during most of this period. Finally, one year after construction commenced, the adults began the final phase, lining the nest with pale fibers over the course of two weeks. After this, adults occasionally roosted in the nest at night, but added little or no material to the nest. Five months after lining was added the first clutch had still not been laid, giving a building period of over 17 months for this nest. In some instances, it appears that dormitory nests (Skutch 1967, see General Natural History) were augmented, eventually becoming breeding nests. Although additional observations on nest construction are needed, I suggest that most nests are constructed in this manner. Those nests apparently hanging by few, or even a single, thin attachment point (Figures 11d, 12a-b, 14) were likely built in the manner described above, but have subsequently become dislodged from their original attachment due to the dislodgement of the soil, vegetation, or wood around them. Similarly, I suspect that the few nests found in situations such as small vine or branch tangles, where they appeared to have been attached by simply wrapping moss around small branches, were originally constructed into larger, denser clumps of vegetation that had subsequently fallen away. Because of the extremely slow rate at which nests are built, I have little information on adult behavior during construction. From videotaping behaviors at one nest during the lining phase, I found that adults may build at a rate of over 20 visits per adult per hour, and then not return to the nest for several weeks or months. Most of my direct observations of nest construction were during the lining phase.

While lining the nest, both adults

frequently make their characteristic, loud, trilling call (Hilty & Brown 1986, Ridgely & Greenfield 2001, Areta 2007). Before reusing old nests, the previous lining is removed and a new lining is added.

38

NEST ORIENTATION As barbtail nests may be built partially or entirely into various natural cavities, completely pendant, or attached in various manners to logs, banks, and epiphyte tangles, they may potentially be oriented in any direction. Here I describe the unique, non-random pattern of nest orientation shown by this species.

FIGURE 16. Nest entrance orientation of barbtails in relation to the direction of water flowing directly below the nest (left). Nests oriented right (90°) are those opening towards the far side of the stream and nests oriented left (270°) are those opening towards the closest bank. Orientation of barbtail nests in relation to the four cardinal directions (right)

Barbtail nest orientation was strongly correlated with stream flow, such that nests were oriented in one of three ways in relation to the direction of water movement (Figure 21, r = 0.16, n = 196; Raleigh’s test; Z = 5.401, p = 0.005: Rao’s test; U = 303.43, p < 0.01: Watson’s test; U2 = 1.121, p < 0.005). Of the three categories (upstream ±20°, downstream ±20°, across-stream ±30°), most were oriented either with, or against, the flow of water

39

(Figure 17), but showed no significant preference for parallel or perpendicular (X2 = 1.1, p >> 0.05, df = 1). Nests oriented perpendicular to the stream showed no relationship with regards to the closest bank, sometimes opening across the stream and sometimes opening towards the nearest bank (X2 = 0.2, p >> 0.05, df = 1). Nest orientation showed only a weak relationship to compass bearing (Figure 16, r = 0.113, n = 196: Raleigh’s test; Z = 2.503, p = 0.082: Rao’s test; U = 240.61, p < 0.01: Watson’s test; U2 = 0.187, p > 0.05). As Rao’s test is more sensitive to patterns with high levels of dispersion (indicated by low r value; Bergin 1991), it is understandable that it was the only of the tree tests to yield a significant result. This propensity to be oriented east is most likely a reflection of the study areas streams’ significantly ESE direction (r = 0.437, n = 196: Raleigh’s Z = 37.423, p