Distribution, natural history and conservation status of two endemics of the Bolivian Yungas, Bolivian Recurvebill Simoxenops striatus and Yungas Antwren Myrmotherula grisea

Summary

Since their description in the first half of the 20^th century by M. A. Carriker, Bolivian Recurvebill Simoxenops striatus and Yungas Antwren Myrmotherula grisea have been regarded as extremely poorly known endemics of the Bolivian Yungas and adjacent humid foothill forests. They are considered ‘Vulnerable’ under the IUCN criteria of small population, predicted population decline (criterion C2a) and, in the case of Bolivian Recurvebill, small extent of occurrence (criteria B1a+b). Here we summarise the information published to date and present extensive new data on the distribution (including the first records for extreme southeast Peru), natural history, population size and conservation status of both species based on field work in the Bolivian Andes over the past 12 years. Both species primarily inhabit the understorey of primary and mid-aged to older regenerating forest and regularly join mixed-species foraging flocks of insectivorous birds. Bolivian Recurvebill has a strong preference for Guadua bamboo, but it is not an obligate bamboo specialist and persists at often much lower densities in forests without Guadua. Yungas Antwren seems to have a preference for dense, structurally complex understorey, often with Chusquea bamboo. Both species are distributed much more continuously at altitudes of mostly 600–1,500 m, occupy a greater variety of forest types (wet, humid, semi-deciduous forest) and have a much greater population size than previously thought. Consequently, none of the IUCN criteria currently used to assign ‘Vulnerable’ status to both species actually apply, and we recommend them to be downlisted to ‘Near Threatened’.

Resumen

Desde su descripción en la primera mitad del siglo 20 por M. A. Carriker, el Ticotico Boliviano Simoxenops striatus y el Hormiguerito Boliviano Myrmotherula grisea han sido considerados como especies muy pobremente conocidas y endémicas de los Yungas de Bolivia y bosques húmedos del pie de monte colindante. Son considerados como Vulnerable a la extinción bajo los criterios de la UICN de una población pequeña, un declive poblacional predicho (criterio C2a) y, en el caso del Ticotico Boliviano, una extensión de presencia (área de distribución) pequeña (criterios B1a+b). En el presente trabajo resumimos la información publicada hasta la fecha, y presentamos datos nuevos amplios sobre la distribución (incluyendo los primeros registros para el extremo sudeste del Perú), historia natural, el tamaño poblacional y estado de conservación de ambas especies, basándonos en nuestro trabajo de campo en los Andes de Bolivia durante los últimos 12 años. Ambas especies principalmente habitan el sotobosque de bosque primario y de bosque en regeneración de edad media y maduro, y a menudo se unen a bandadas mixtas de forrajeo de aves insectívoras. El Ticotico Boliviano tiene una preferencia marcada para bambú del género Guadua (tacuara), pero no es un especialista obligado de bambú y persiste en bosques sin tacuara, aunque a menudo a densidades mucho mas bajas. El Hormiguerito Boliviano parece tener una preferencia para un sotobosque denso, con una estructura de la vegetación compleja y a menudo con bambú del género Chusquea. A diferencia del estado del conocimiento anterior, ambas especies tienen una distribución mucho más continua, en su mayor parte a altitudes de 600–1500 m, habitan una mayor variedad de tipos de bosque (muy húmedo, húmedo, semi deciduo) y tienen un tamaño poblacional mucho más grande. Como consecuencia, ninguno de los criterios de la UICN actualmente usados para asignar la categoría Vulnerable a ambas especies aplican, y recomendamos que sean cambiadas de categoría a Casi amenazado.

Introduction

The Bolivian Yungas are situated in Endemic Bird Areas 054 and 055 (Bolivian and Peruvian lower and upper Yungas, respectively; Stattersfield et al. 1998, www.birdlife.org/action/science/endemic_bird_areas/index.html), both of which are considered of urgent conservation priority, and in the southern part of the Tropical Andes biodiversity and conservation hotspot of Myers et al. (2000). They extend from about 13°30’ S in depto. La Paz to 18°00’ S in depto. Santa Cruz and reach from Andean foothills (c. 500 m altitude) up to isolated Polylepis forest patches (to c. 4,200 m) above the closed timberline. The Yungas ecoregion is Bolivia's top priority for bird conservation (Herzog et al. 2005b) and contains about 600 (43%) of Bolivia's c. 1,420 bird species (Hennessey et al. 2003b, authors’ unpubl. data), including 14 Bolivian endemics and near-endemics, 43 restricted-range species (see Stattersfield et al. 1998) and 12 threatened species (see Hennessey et al. 2003b).

Up to the early 1990s the majority of ornithological field work in the Yungas and other regions in Bolivia was carried out by collecting expeditions primarily from North American museums, and specimen localities were far and few between. Not surprisingly, many bird species were considered poorly known (e.g., Remsen et al. 1982, 1987, 1988; Remsen and Traylor 1983; Remsen 1984; Parker et al. 1992). Over the past 10 to 15 years, however, our knowledge of bird distributions in the Bolivian Yungas has improved considerably (see, e.g., Perry et al. 1997, Herzog et al. 1999, Hennessey 2003a,b, Hennessey et al. 2003a, Hennessey and Gomez 2003, Herzog et al. 2005a, MacLeod et al. 2005), allowing for a reassessment of the natural history and conservation status of numerous species once considered poorly known (e.g., Bryce et al. 2005), including the threatened (BirdLife International 2000, 2004) Yungas endemics Bolivian Recurvebill Simoxenops striatus (Furnariidae) and Yungas Antwren Myrmotherula grisea (Thamnophilidae). Both are currently considered ‘Vulnerable’ under IUCN criteria of small population, predicted population decline and, in the case of Bolivian Recurvebill small extent of occurrence and area of occupancy (BirdLife International 2007a,b).

Bolivian Recurvebill is among the country's most poorly understood bird species. For over 50 years, this species was known from only five specimens, four of which were collected in the 1930s by M. A. Carriker in humid Andean foothill forest (Vaurie 1980) at Santa Ana on the Río Coroico (670 m), depto. La Paz (Carriker 1935), and at El Palmar (800 m), depto. Cochabamba (Bond and Meyer de Schauensee 1941). The fifth specimen (housed at the Louisiana State University Museum of Natural Science) was collected in 1954 by F. Steinbach at Alto Palmar (1,100 m) in lower Yungas forest above El Palmar. In 1989, Parker et al. (1992) rediscovered the Bolivian Recurvebill at a new locality in the upper Río Saguayo valley (700–800 m) of Amboró National Park, depto. Santa Cruz, providing the first data on the species’ natural history, and in 1992 Whitney et al. (1994) observed two birds 48 km by road southwest of Villa Tunari (1,080–1,090 m), depto. Cochabamba. Until recently, those were the only known localities for the Bolivian Recurvebill.

Yungas Antwren also was described by Carriker (1935) based on six specimens from Santa Ana (670 m), depto. La Paz. Two specimens at the Carnegie Museum, collected as early as 1917 at Cerro Hosane, Río Blanco (1,200 m), depto. Santa Cruz, had apparently been overlooked by Carriker (Remsen et al. 1982). A further specimen was collected in 1937 at El Palmar (800 m), depto. Cochabamba (Collar et al. 1992). From 1960 to 1981 an additional six specimens were collected (housed at the Louisiana State University Museum of Natural Science) at four localities in the departments of Santa Cruz (Santa Rita, 500 m), Cochabamba (Alto Palmar, 1,100 m) and La Paz (Serranía Bellavista, 1,550 m; Río Beni 20 km N Puerto Linares, 600 m) (Remsen et al. 1982), and the species was observed by R. S. Ridgely (Remsen et al. 1982) north of Caranavi (1,400 m), depto. La Paz, by Parker et al. (1992) in the upper Río Saguayo valley (650–1,000 m), depto. Santa Cruz, and by T. A. Parker (Collar et al. 1992) in the Serranía Raigones (800 m), depto. Santa Cruz. The only primary literature sources on the natural history of Yungas Antwren are Remsen et al. (1982), Parker et al. (1992) and Perry et al. (1997), who stressed the participation of the species in mixed-species foraging flocks from the understorey (Remsen et al. 1982, Perry et al. 1997) to the lower canopy (Parker et al. 1992, Perry et al. 1997). Parker et al. (1992) stated that “this seemingly rare species is almost unknown in life.” In fact, its loud song was unknown until 1996 when recorded for the first time by SKH (see Mayer 2000, Isler and Whitney 2002).

Over the past 12 years we obtained new data on the natural history of Bolivian Recurvebill and Yungas Antwren at a number of new localities in the Bolivian Yungas that notably increase their known distributional ranges. We also report on the first records of each species for extreme southeast Peru. Additionally, for the first time we present population density and size estimates for both species in the central portion of their range. In conjunction these data indicate that both species no longer merit threatened status, and that they should be downlisted to ‘Near Threatened’. Our new locality data up to 1999 were included as personal communications in Threatened birds of the world (BirdLife International 2000, 2004), but they are presented formally and in detail herein. Data collected after 1999 are presented here for the first time. Because both species have very similar geographical distributions and often occur syntopically they are treated together rather than in separate publications.

Study areas

Most of the new distributional data presented here were obtained by the authors during rapid assessment surveys at a number of localities throughout the Bolivian Yungas. Coordinates, altitudes and brief notes on habitat for most localities are presented in Table 1 and in the Supplementary materials. Two sites where we obtained important additional data are described in more detail here.

Table 1. Summary of Bolivian Recurvebill Simoxenops striatus (SIST) and Yungas Antwren Myrmotherula grisea (MYGR) locality data. See text and Supplementary materials for details on locality records. Coordinates and altitudes of literature and specimen localities were verified based on field knowledge of the authors, topographic maps and Google Earth.

¹ The coordinates for Río Saguayo given by Parker et al. (1992) (17°50’S, 63°39’W) were an error.

² Paynter (1992) stated that his identification of this locality was uncertain and gave coordinates for a site in the lowlands c. 15 km north of Buena Vista, which is at an altitude of c. 300 m (not 500 m as given by Remsen et al. 1982) and falls outside the distributional range of Yungas Antwren. The coordinates given here correspond to a semi-deciduous forest site c. 26 km southwest of Santa Cruz de la Sierra, which matches the altitude given by Remsen et al. (1982) and is much more likely to be the actual collecting locality.

Cordillera Mosetenes, a remote mountain range located in the Andean foothill region of depto. Cochabamba, is separated from the main Andes by the Altamachi River. It is about 130 km long, 20–25 km wide (covering c. 3,000 km²), up to 2,050 m high and probably contains the largest completely uninhabited forest tract in Bolivia. It is one of the country's most humid regions and, due to its steep topography, is of extremely difficult access. Parts of the area are formally protected in Isiboro-Sécure Indigenous Territory and National Park. Bird surveys on Cordillera Mosetenes were conducted by SKH in a radius of c. 1 km around a small lagoon (Laguna Carachupa; Table 1) from 28 August to 21 September 2003, with several trails reaching from stream valleys at about 1,200 m altitude to a mountain ridge at 1,600 m. Rather than habitat characteristics, the sole criterion for choosing this site was the availability of an open area where a helicopter could land.

The vegetation at our study site on Cordillera Mosetenes (MK, unpubl. data) can be divided into five main categories. As determined from several over-flights by small airplane and helicopter by the authors, this habitat mosaic stretches across 80–90% of the area of Cordillera Mosetenes. (1) Mature forest was 20–25 m tall, had a rather open understorey and was restricted to scattered patches about 0.1–0.5 ha in extent. Epiphytic plants were present in moderate density and abundant only on scattered, up to 35 m tall emergent trees (Ficus spp.). (2) Most of the area was covered by regenerating forest of variable age, caused by natural disturbances due to landslides or tree falls. Older regenerating forest differed from mature forest in its lower canopy height (10–20 m), lack of thick-trunked trees, denser understorey, and fewer epiphytes. (3) Less widespread young successional stages were dominated by dense scrub (Panicum sp., Solanum spp., Acalypha spp., Hedyosmum spp.), with emergent trees (Cecropia spp.) to 12 m tall. (4) Large tracts were formed by up to 15 m tall Guadua bamboo forest. Within dense bamboo stands very little tree regeneration was observed. (5) Stunted ridge forest occurred at 1,500–1,600 m, with trees 3–12 m tall (exceptionally to 15 m) and thickly covered in liverworts, mosses and vascular epiphytes. Small Chusquea bamboo was frequent, but did not form dense thickets.

Los Volcanes (Table 1) is a private 297 ha reserve immediately adjacent to Amboró National Park, depto. Santa Cruz. It is situated on the transition from the tropical humid Yungas to subtropical inter-Andean dry valleys of southern Bolivia and consists of steep sandstone hills intersected by narrow valleys at altitudes of 1,000–1,350 m. Mean annual precipitation is c. 1,300 mm (Herzog and Kessler 2002) and characterized by strong seasonality, with about 80% of the rainfall occurring from October to April (SKH, unpubl. data). Topographical factors lead to a marked differentiation of vegetation types, with drought-deciduous (dry) forest on sunny north-facing slopes, semi-deciduous forest on partly shaded slopes and evergreen forest on valley bottoms and adjacent, mostly shaded slopes. Chusquea bamboo was abundant in all forest types with a flowering and fruiting climax in the rainy season (November to March) 2000/2001, followed immediately afterwards by a complete, area-wide bamboo die-off. In 2003 bamboo began to recover slowly, but by late 2005 it was still patchy and not nearly as abundant as during its previous climax.

Methods

Survey methods

Rapid assessment surveys were audio-visual and relied heavily on the use of sound recording (Parker 1991) for species documentation and on playback for the identification of unfamiliar bird voices. Field work at most sites aimed at maximizing survey area, covering a new area on most mornings starting with or before the dawn chorus. We kept track of the number of individuals per species observed either consecutively throughout the entire survey (Herzog et al. 2002) or in summary form at the end of each survey day (Hennessey and Gomez 2003). Sound recordings of Bolivian Recurvebill made by ABH in 1998 were deposited at the Macaulay Library of Natural Sounds (MLNS), Cornell Laboratory of Ornithology, U.S.A., and are included in Mayer (2000). Recordings of Yungas Antwren made by SKH in 1996 are included in Mayer (2000), Schulenberg (2000) and Isler and Whitney (2002).

Additional methods were used at two sites. During a rapid assessment expedition on Cordillera Mosetenes, survey duration and relative abundance of both species allowed for the use of time consuming territory mapping with the help of playback of song sound recordings. SKH also conducted 16-minute point counts during the first three morning hours (beginning at dawn) in an attempt to use distance sampling for population density estimation. However, only eight spatially independent (minimum distance 250 m) point count stations could be established along the available survey trails (see below), and a total of 14 counts at those stations resulted in only a single detection each of Bolivian Recurvebill and Yungas Antwren, so point count surveys were not continued. This scarcity of detections probably is at least partly a result of the vocal activity pattern of both species, which vocalize predominantly during the dawn chorus and only very sporadically during the remainder of the morning.

Due to the potential problems of territory mapping with unmarked individuals, the objective was to determine the total number of territories falling into the study area, regardless of whether they were completely contained in the area, rather than estimating the absolute size of each territory. An area of about 20 ha at altitudes of 1,200–1,600 m was surveyed intensively along newly cut trails over a 24-day period in August and September 2003. Dense vegetation and steep topography limited field work to the trail system in most of the survey area. A hand-held GPS unit was used to determine the linear horizontal distance between the origin and terminus (or a major shift in direction) of each trail (transect), ignoring minor curves and switchbacks and topographic complexity of the area. Assuming that vocalizations of both species can be detected up to a distance of 50 m on both sides of a transect (see, e.g., Schieck 1997; the song of Bolivian Recurvebill can certainly be detected at distances > 50 m; authors’ pers. obs.), we consider that 100 m of survey transect covered an effective horizontal survey area of 1 ha. Thus, total transect length in metres was divided by 100 to obtain the size of the entire survey area in hectares.

All observations of both species on Cordillera Mosetenes were marked on a detailed map drawn with the aid of aerial photographs and a GPS. After two survey weeks, SKH targeted gaps between observation locations of each species with song playback to detect additional territories and/or determine territory limits along the trail system. Special caution was exerted to minimize possible errors in territory limits resulting from the observation of unmarked individuals. Once a pair of either species was located on a given day, SKH walked to the nearest observation point or cluster of points considered to be assignable to a neighbouring territory of that species, followed by the use of song playback broadcast in all directions to elicit a response of the presumed territory holders. If a response to playback came from the direction of the previous encounter the experiment was considered inconclusive. This procedure was repeated daily for most territories during the last 10 days of the survey.

Los Volcanes was initially surveyed by SKH in 1997 as outlined above. From June 2003 to November 2005 birds were monitored monthly by SKH and VHGS using constant effort mist netting, point counts and territory mapping on and in the vicinity of three permanent 1-ha plots at 1,050–1,200 m. Natural history data from this study on Bolivian Recurvebill are presented in the Supplementary materials.

Range size (extent of occurrence) estimates

In a recent study determining conservation priorities on the east Andean slope of Peru and northern Bolivia (Young 2007), the extent of occurrence of Bolivian Recurvebill and Yungas Antwren were estimated quantitatively for the first time by GIS-based predictive distribution modelling. Using the inductive model Maxent (Phillips et al. 2006; see also Hernandez et al. 2006) the extent of occurrence of each species was projected based on all known locality records as well as altitude, climate and vegetation data, including current habitat availability derived from Moderate Resolution Imaging Spectroradiometer (MODIS) data (Hernandez 2007). For each species four models varying in the use of the MODIS data were run (no MODIS data, MODIS data unsummarised, summarised by 2 km, summarised by 5 km). To evaluate the models, data could not be partitioned into training and evaluation records because of the scarcity and low spatial precision of available locality data. In this situation, review by specialists familiar with the species is often the best method available to determine the modelling procedure that produces the most realistic range prediction (Hernandez et al. 2008). Thus, SKH reviewed the output, determining whether any of the Maxent models produced reasonable results, and if they did, identifying the model and threshold (from the 0–1 scale Maxent output depicting how well the variables in each pixel match those found where the species has been recorded) that produced the most reasonable map for the species according to our present understanding of its distribution and habitat availability. Predicted areas of distribution where the species is known not to occur were eliminated. The final distribution maps are available at www.natureserve.org/aboutUs/latinamerica/maps_birds.jsp.

Results

Distribution, natural history and vocalizations

Bolivian Recurvebill and Yungas Antwren are now known from a total of 21 and 25 (27 when including two vagrant records) localities, respectively, which are fairly evenly distributed along the east Andean slope and extend across a latitudinal range from about 14°S to 18°S (Table 1, Figures 1 and 2). Details on records obtained by us and other ornithologists since 1995 are given in the Supplementary materials. Perhaps the most noteworthy of our observations are the first records (sound-recordings) of both species for Peru, obtained by SKH and VHG in 2005 and 2006 at Palmerani-Urubamba (Table 1) in extreme southeast depto. Puno, less than 3 km from the Bolivian border (see Supplementary materials). Prompted by the discovery of Bolivian Recurvebill at this site, a team from the Louisiana State University Museum of Natural Science visited the general area in 2007 and obtained the first Peruvian specimens of both species (D. F. Lane and A. Cuervo, pers. comm.; Table 1). Detailed information on the natural history and vocalizations of both species are given in the Supplementary materials.

Figure 1. Locality records (dark grey dots) of Bolivian Recurvebill Simoxenops striatus in Bolivia (LP = depto. La Paz, BE = depto. Beni, CO = depto. Cochabamba, SC = depto. Santa Cruz) and extreme southeast Peru (PU = depto. Puno). Light grey shading denotes areas within the main altitudinal range of the species (640–1,500 m). Numbers correspond to the following localities or locality clusters: 1 = Curva Alegre, Palmerani-Urubamba, Río Cocos; 2 = Serranía Eslabón; 3 = San Miguel Chico; 4 = Cerro Asunta Pata; 5 = Serranía Beu; 6 = Serranía Cuchilla; 7 = Illampu; 8 = Santa Ana; 9 = Mosetén Communal Lands, Río Ipiri; 10 = Laguna Carachupa; 11 = 48 km by road southwest of Villa Tunari, El Palmar, Alto Palmar, Serranía de Callejas; 12 = Río Saguayo; 13 = Serranía La Luna; 14 = Río Salado, Los Volcanes.

Figure 2. Locality records (dark grey dots) of Yungas Antwren Myrmotherula grisea in Bolivia (LP = depto. La Paz, BE = depto. Beni, CO = depto. Cochabamba, SC = depto. Santa Cruz) and extreme southeast Peru (PU = depto. Puno). Light grey shading denotes areas within the main altitudinal range of the species (600–1,400 m). Numbers correspond to the following localities or locality clusters: 1 = Curva Alegre, Palmerani-Urubamba, Río Cocos, Putina Puncu; 2 = Serranía Eslabón; 3 = Tuichi drainage: Hacienda Ubito, Río Machariapo, Yarimita; 4 = Cerro Asunta Pata; 5 = Serranía Pilón; 6 = Río Beni N Puerto Linares; 7 = Padilla; 8 = Serranía Bellavista, Caranavi; 9 = Santa Ana; 10 = Río Ipiri; 11 = Laguna Carachupa; 12 = Río Altamachi; 13 = El Palmar, Alto Palmar; 14 = Río Saguayo; 15 = Cerro Hosane; 16 = Río Macuñucú; 17 = Santa Rita; 18 = Los Volcanes, Serranía Raigones.

Range size (extent of occurrence) estimates

GIS-based predictive distribution modelling produced extent of occurrence estimates of 37,130 km² for Bolivian Recurvebill and of 29,755 km² for Yungas Antwren (Young 2007).

Population density and population size estimates

Along the trail system on Cordillera Mosetenes, covering an effective survey area of 20 ha, SKH determined eight Bolivian Recurvebill and six Yungas Antwren territories using territory mapping and song playback. Under the conservative assumption that only 50% of the average territory area was contained in the survey area and that a territory is occupied by two adult birds, extrapolated breeding population densities for the Laguna Carachupa area are 20 Bolivian Recurvebill and 15 Yungas Antwren pairs km⁻². When further assuming that at least two thirds of the area of Cordillera Mosetenes (c. 2,000 km², see Study area above), excluding areas above the altitudinal range of both species and marginal areas such as the northwestern tip with apparently suboptimal habitat, sustain those population densities, the overall breeding populations on Cordillera Mosetenes consist of about 80,000 Bolivian Recurvebills and 60,000 Yungas Antwrens. Taking into account a conservative error margin of 25%, those values are reduced to a minimum of 60,000 and 45,000 breeding birds, respectively. However, given the potential problems of territory mapping with unmarked individuals (primarily the potential overestimation of the number of territories by the false assignment of the same individual to more than one territory), and the extrapolation from a small study area of 0.2 km² to an entire mountain range, a cautious interpretation of these values is recommended.

Although we lack similar quantitative data from other areas, it seems safe to conclude that neither species reaches such high densities at most of its other localities. The Río Cocos site (Table 1), however, an area with moderately abundant Guadua bamboo, might be an exception for Bolivian Recurvebill. Here, SKH and VHGS evaluated the density of the species somewhat less rigorously. Due to time constraints we made no specific effort to map territories, but we recorded what appeared to be six Bolivian Recurvebill territories in an effective survey area of c. 45 ha. Although this estimate is not directly comparable to that from Cordillera Mosetenes, given the larger survey area and shorter survey duration at the Río Cocos (see Supplementary materials), similar population densities at both sites seem plausible.

Discussion

Distribution

Bolivian Recurvebill and Yungas Antwren are now known from over 20 localities each (Table 1), which are distributed much more continuously than previously thought throughout Andean foothills and lower Yungas from depto. Puno in extreme southeast Peru to western depto. Santa Cruz in Bolivia (Figures 1, 2). In addition, the majority of localities of both species are inside or on the border of protected areas (Table 1). Their respective altitudinal distributions are also rather similar. That of Bolivian Recurvebill ranges from 640 m to mostly 1,500 m, locally extending up to 1,700 m in depto. La Paz (Table 1), but perhaps only under special local ecoclimatic conditions in somewhat sheltered valleys that have higher levels of solar insolation than outer Yungas slopes under direct impact from incoming clouds (see Herzog et al. 1999). Yungas Antwren occurs mostly from 600 m to 1,400 m, but extends locally up to 1,550 m in depto. La Paz; only at the extreme southern end of its range does it occur at altitudes as low as 450-500 m (Río Macuñucú, Santa Rita; Table 1), presumably due to the absence of lowland congeners from these areas (SKH, pers. obs.). The relative scarcity of records from large parts of depto. Cochabamba (Figures 1, 2) is probably mostly due to the difficult access to appropriate altitudes.

Both species might nonetheless be distributed somewhat patchily in at least parts of their range as some studies in seemingly suitable habitat did not record either species. Such areas include Río Leche (R. MacLeod, in litt.) and Cerro Leñe (SKH, unpubl. data, but see Herzog and Kessler 1998) in eastern depto. Cochabamba, Serranías Sadiri, Tequeje and del Tigre in depto. La Paz (Madidi National Park; Hennessey 2003b) and Serranía Bala in depto. Beni (Hennessey et al. 2003a). Serranías Sadiri, Tequeje, del Tigre and Bala are the first foothill ridges in the northern Bolivian Andes that peak at about 900–1,000 m. Although they fall within the altitudinal range of Bolivian Recurvebill and Yungas Antwren, the suitable altitudinal amplitude of only about 300–350 m may be too narrow to sustain viable populations. The notion of a somewhat patchy distribution is further supported by the fact that, although both species inhabit an almost identical geographical range and a very similar altitudinal range, only 12 (33%) of the 36 localities in Table 1 were found to hold both species, which may be related to differences in (micro)habitat use between the two species.

Habitat associations

Parker et al. (1996) considered Bolivian Recurvebill as having high sensitivity to disturbance. We disagree and suggest classifying the species as having medium sensitivity to disturbance. Although not at high densities, Bolivian Recurvebill has been found in human disturbed habitats at Palmerani, 40 km southwest of Villa Tunari (S. Kreft, pers. comm.) and at the edge of logging gaps at Río Ipriri (see Supplementary materials). At Laguna Carachupa on Cordillera Mosetenes the dominant vegetation type was regenerating forest of variable age (often with stands of Guadua bamboo) caused by natural disturbances due to landslides or tree falls.

The statement of Ridgely and Tudor (1994) that Bolivian Recurvebill apparently shows no preference for bamboo, which was also implied by Parker et al. 1992 and based on the very limited data available at the time, needs to be revised. The fact that it reaches its highest known population densities in areas where Guadua bamboo is common to abundant (Cordillera Mosetenes, Río Cocos) combined with repeated observations of specialised foraging behaviour on Guadua bamboo (see Supplementary materials) are circumstantial evidence for a strong bamboo preference. However, the species is not an obligate bamboo specialist and persists at often much lower densities in humid and semi-deciduous forest without Guadua bamboo, where it forages in dense understorey or vine tangles, often near tree fall gaps. As observed by SKH at Los Volcanes (see Supplementary materials), it appears to employ similar foraging manoeuvres in these micro-habitats as it does in bamboo stands.

Yungas Antwren was thought to inhabit the mid-storey by Ridgely and Tudor (1994), whereas Zimmer and Isler (2003) considered it a species of mid-storey and subcanopy that descends to the understorey only at forest edge. Data presented here (see Supplementary materials) indicate that Yungas Antwren, like its closest relatives Plain-winged Antwren M. behni, Unicoloured Antwren M. unicolor and Alagoas Antwren M. snowi (the ‘plain-winged group’; Whitney 1994, Whitney and Pacheco 1997, Zimmer and Isler 2003), is an understorey species that moves up to mid-level or subcanopy only occasionally while foraging with mixed-species flocks. Its regular occurrence in semi-deciduous forest in depto. La Paz (Perry et al. 1997, this study) also is a rather recent discovery. The presence of typical humid forest species in semi-deciduous and deciduous forest of isolated rain-shadowed valleys embedded in the humid Bolivian Yungas is a general pattern that has only recently been documented to its full extent (Herzog and Kessler 2002, 2006). Nonetheless, the first record of Yungas Antwren from semi-deciduous forest originated from Santa Rita (Remsen et al. 1982; Table 1), depto. Santa Cruz, which went unnoticed due to the erroneous assignment of coordinates to this locality (see footnote 2 in Table 1). T. A. Parker's observation of several individuals on Serranía Raigones, which was mentioned only ‘in passing’ by Collar et al. (1992), originated from a deciduous forest area where semi-deciduous forest is restricted to narrow gullies (SKH, pers. obs.) at the extreme southern end of the species’ range. However, as with nearby Los Volcanes, we consider it unlikely that Serranía Raigones holds a breeding population of Yungas Antwren. Rather, these observations may be related to source-sink dynamics, where individuals from the core, evergreen areas spill into suboptimal deciduous habitat.

Range size and population size

The global extent of occurrence was previously estimated at 19,900 km² for Bolivian Recurvebill and 32,000 km² for Yungas Antwren (BirdLife International 2000, 2004, 2007a,b). The most recent estimates obtained by GIS-based predictive distribution modelling (Young 2007) projected a rather similar value of 29,755 km² for Yungas Antwren, but a notably greater extent of occurrence of 37,130 km² for Bolivian Recurvebill, which is almost twice as large as that estimated previously. These new estimates thus reconfirm the classification of both species as restricted-range species (see Stattersfield et al. 1998).

The population size of Bolivian Recurvebill and Yungas Antwren was previously thought to be between 2,500 and 10,000 mature individuals (BirdLife International 2004, 2007a,b). Our density estimates from Cordillera Mosetenes indicate that this area alone holds breeding populations of both species of well in excess of 10,000 individuals. Although total population sizes are difficult to estimate due to a lack of quantitative data from other localities, we consider that the global population of both species must number on the order of several tens of thousands of mature individuals. Cordillera Mosetenes seems to be the stronghold for both species, probably due to its high habitat heterogeneity with much natural disturbance and, for Bolivian Recurvebill, the abundance of thick-stemmed Guadua bamboo.

Conservation status

Bolivian Recurvebill and Yungas Antwren are currently considered ‘Vulnerable’ (BirdLife International 2007a,b). The latter species is listed based only on IUCN criterion C2a (BirdLife International 2007a), i.e. its population is estimated to number fewer than 10,000 mature individuals and a continuing decline in numbers of mature individuals is observed, projected, or inferred, with no subpopulation estimated to contain more than 1,000 mature individuals. Our density estimates from Cordillera Mosetenes, combined with the new locality records, indicate that the global population size of Yungas Antwren is much greater than 10,000 mature individuals, that the Cordillera Mosetenes subpopulation is much greater than 1,000 mature individuals and that the global population is much less fragmented than previously thought. Criterion C2a thus does not apply to Yungas Antwren.

Bolivian Recurvebill also is listed under criterion C2a (BirdLife International 2007b), and as with Yungas Antwren, our data strongly suggest that this criterion does not apply to the species. Additional criteria that are thought to apply to Bolivian Recurvebill are B1a+b (BirdLife International 2007b), which estimate the species’ extent of occurrence to be less than 20,000 km², combined with no more than 10 known localities and a continuing population decline being observed, inferred or projected. The numerous new localities presented above combined with the new extent of occurrence estimate of 37,130 km² (Young 2007) document that criteria B1a+b also do not apply to Bolivian Recurvebill.

In conclusion, none of the IUCN criteria currently used to assign ‘Vulnerable’ status to both species (BirdLife International 2007a,b) actually apply, and we recommend that Bolivian Recurvebill and Yungas Antwren be downlisted to ‘Near Threatened’. Although no direct evidence for a population decline exists, given their small range it remains a valid threat to both species due to high levels of habitat conversion in the Andean foothills and lower Yungas of Bolivia. Further, even though there remains no doubt that both species have a global population size of well above 10,000 mature individuals, our estimates of > 50,000 individuals are essentially based on a single locality, and confirmation of our density estimates from additional localities on Cordillera Mosetenes are desirable. Both species have been recorded in several protected areas, but most Bolivian national parks offer little real protection to their threatened species because areas such as Amboró National Park often suffer locally from illegal invasions and forest clearing by subsistence farmers (e.g., Bayá 2006). In addition, under current Bolivian legislation, the exploitation of natural resources has priority over the preservation of protected areas and there remains the risk that these areas could be seriously altered in future extractions of natural resources or road construction projects.

Other threatened Yungas species

Six bird species currently considered threatened occur in the humid forests of the Bolivian Yungas (excluding birds of treeline habitats and Polylepis woodlands such as Royal Cinclodes Cinclodes aricomae), and one might wonder if the conservation status of some of them should not also be re-evaluated in light of newly available data. Two are listed as ‘Endangered’ (Horned Curassow Pauxi unicornis, Yellow-rumped Antwren Terenura sharpei) and four as ‘Vulnerable’ (Military Macaw Ara militaris, Scimitar-winged Piha Lipaugus uropygialis, Green-capped Tanager Tangara meyerdeschauenseei, Cerulean Warbler Dendroica cerulea) (see www.birdlife.org/datazone/species/index.html).

The Nearctic-Neotropical migrant Cerulean Warbler reaches the southern terminus of its non-breeding range in northern Bolivia, where it is known from only a handful of records (see Tobias and Seddon 2007), so virtually nothing can be said about its conservation status based on such data. The situation is similar for Green-capped Tanager, a species formerly considered as endemic to Peru and reported from a single Bolivian locality (Hennessey and Gomez 2003). The status of Scimitar-winged Piha was recently reviewed by Bryce et al. (2005), who recommended continued classification as ‘Vulnerable’. The widely distributed Military Macaw, on the other hand, does not appear to be highly threatened in Bolivia, where it is common throughout most of the Yungas of depto. La Paz and northern depto. Santa Cruz (rare to uncommon elsewhere in the Bolivian Andes; authors’ unpubl. data), and it is found rather infrequently in the local pet trade (Herrera and Hennessey 2007).

To the contrary, both ‘Endangered’ species, which are also restricted-range species (Stattersfield et al. 1998), are properly listed as such. Both seem to be naturally rare and are threatened by ongoing, widespread habitat conversion in the foothills and, in the case of Horned Curassow, a high sensitivity to hunting pressure (BirdLife International 2000). Bolivian records of Yellow-rumped Antwren are restricted to only three areas in deptos. La Paz and Cochabamba (Cerro Asunta Pata, Louisiana State University Museum of Natural Science specimen; Serranía Bellavista, Remsen et al. 1982, BirdLife International 2000; the Chapare region near San Onofre, Remsen 1982, Brumfield and Maillard 2007), but the species no longer seems to occur on Serranía Bellavista (VHGS, unpubl. data) and has not been found at additional sites despite specific efforts (Hennessey 2004, SKH and VHG, unpubl. data).