Introduction
Forests dominated by the genus Polylepis characterize the higher mountain regions of South America and are distributed in small patches primarily restricted to ravines and rock outcrops. Recent studies suggest that the peculiar distribution of Polylepis trees may be the result of thousands of years of human activities (Kessler Reference Kessler2002, Renison et al. Reference Renison, Hensen, Suarez and Cingolani2006). A long history of logging (Fjeldså and Kessler Reference Fjeldså and Kessler1996, Renison et al. Reference Renison, Hensen, Suarez and Cingolani2006), fire management (e.g. Renison et al. Reference Renison, Cingolani and Schinner2002, Cingolani et al. Reference Cingolani, Renison, Tecco, Gurvich and Cabido2008), overgrazing (e.g. Teich et al. Reference Teich, Cingolani, Renison, Hensen and Giorgis2005, Cingolani et al. Reference Cingolani, Renison, Tecco, Gurvich and Cabido2008), and soil degradation (Renison et al. Reference Renison, Hensen, Suarez and Cingolani2006, Torres et al. Reference Torres, Renison, Hensen, Suárez and Enrico2008) has reduced these forests to small isolated patches restricted to rocky outcrops where the impact of livestock and burning is low (Fjeldså and Kessler Reference Fjeldså and Kessler1996; Renison et al. Reference Renison, Hensen, Suarez and Cingolani2006, Cingolani et al. Reference Cingolani, Renison, Tecco, Gurvich and Cabido2008; Coblentz and Keating Reference Coblentz and Keating2008). Consequently, Polylepis forests are now considered one of the most threatened Neotropical vegetation types (Jameson and Ramsay Reference Jameson and Ramsay2007). Despite the limited extent and patchy distribution of Polylepis forest, they still harbour a high bird species richness including many endemics that have been relatively well studied in the tropical Andes (Fjeldså Reference Fjeldså1993, Kessler et al. Reference Kessler, Herzog, Fjeldså and Bach2001, Herzog et al. Reference Herzog, Soria and Matthysen2003, Fjeldså and Kessler Reference Fjeldså and Kessler1996, Cahill and Matthysen Reference Cahill and Matthysen2007, Lloyd and Marsden Reference Lloyd and Marsden2008, Lloyd Reference Lloyd2008a,Reference Lloydb). However, in Argentina quantitative data for Polylepis forest bird communities are scarce.
In Argentina, several Important Bird Areas (IBAs) include Polylepis forests with globally threatened bird species (Di Giácomo Reference Di Giácomo2005) and restricted-range bird species belonging to three Endemic Bird Areas (EBA): EBA 056 ‘High Andes of Bolivia and Argentina’, EBA 057 ‘Argentina and south Bolivian yungas’, and EBA 058 ‘Mountains of central Argentina’ (Stattersfield et al. Reference Stattersfield, Crosby, Long and Wegge1998). However, very few attempts have been made to study avian communities of Polylepis forests. To our knowledge, there is only one study that compares avian composition among forests of P. australis, P. tomentella and P. hieronymi (Renison et al. Reference Renison, Bellis, Guzmán, Grau, Pacheco, Rivera, Politi, Martin, Cuyckens, Marcora, Robledo, Cingolani, Perasso, Cornell, Dominguez, Landi, Hensen and Arnalin press) and two non-specific studies that included, but were not exclusive to, Polylepis forests in the mountains of central Argentina (Heil et al. Reference Heil, Fernández-Juricic, Renison, Cingolani and Blumstein2007, García et al. Reference García, Renison, Cingolani and Fernández-Juricic2008).
To initiate the study of Polylepis forest birds in Argentina, our goal is to perform a first characterization of Polylepis australis bird assemblages along their latitudinal gradient of distribution. Specifically we determined: a) richness, equitability, diversity and abundance of avian communities; b) bird species composition turnover; and c) bird conservation status. These data may be useful to direct future studies and conservation priorities and serve as a baseline for monitoring avifaunal changes over time.
Methods
Study areas
Fieldwork was carried out in P. australis forest fragments located in the eastern slopes of the Andes in northern Argentina and in Sierras Grandes of Argentina. We selected three sampling sites along the P. australis latitudinal gradient (Figure 1). The North site corresponds to Alto Calilegua, Jujuy province (23° 37′ S, 64° 54′ W; 2,714 m) and Centre site corresponds to La Ovejeria, Tucumán province (26° 50′ S, 65° 45′ W; 2,400 m). These two sites belong to the upper cloud forest of the southern Yungas (or subtropical montane forests) where P. australis and Alnus acuminata forests are followed by alpine Andean grasslands and meadows (Cabrera Reference Cabrera1976). Annual vertical precipitation reaches 1,500 mm and is mainly concentrated in the austral summer (December–March). Fog and horizontal precipitation can be equivalent to the vertical precipitation or even exceed this volume (Hunzinger Reference Hunzinger, Brown and Grau1995). The South site corresponds to Sierras Grandes, Córdoba province (31° 58’ S, 64° 56’ W; 1,950 m) which is located in a small chain of mountains 400 km east of the Andean Mountains. In the South site, P. australis dominates the upper strata of woodlands and shrublands with other less abundant woody species, such as: Maytenus boaria, Escallonia cordobensis, Berberis hieronimii, Satureja spp., and Gaultheria poepigii (Cabido and Acosta Reference Cabido and Acosta1985, Cingolani et al. Reference Cingolani, Renison, Zak and Cabido2004). Mean annual precipitation is 840 mm, with 83% of rainfall concentrated in the warmer months (October to April, Renison et al. Reference Renison, Cingolani and Schinner2002).
Figure 1. Distribution range of Polylepis australis forest to Argentina. Filled circles represent the different study sites. North site: Alto Calilegua (Jujuy province) and Centre site: La Ovejeria (Tucuman province) are located on the Andean slopes of the Yungas, and South site: Sierras Grandes in the mountains of central Argentina.
Bird survey
Bird data were collected during the summer season (January to March 2006) when bird richness is highest due to the arrival of summer resident species that use this area for breeding. Additionally, during the summer, differences in bird communities are easier to detect due to their territorial behaviour, compared to their aggregated distribution in the winter when large mixed flocks can be seen (Ordano Reference Ordano1996, García et al. Reference García, Renison, Cingolani and Fernández-Juricic2008). Bird species and abundance were quantified using 10-minute point counts randomly distributed in each P. australis study site. This count period maximized probability of bird detection in cryptic species (Bibby et al. Reference Bibby, Burgess and Hill1992, Ralph et al. Reference Ralph, Sauer and Droege1995, Lee and Marsden Reference Lee and Marsden2008). We located an average of 30 survey points separated by at least 150 m (or 10 min travelling distance) to avoid double-counts between neighbouring points. At each point, the surveyor waited 5 min as a settling down period before starting the counts (Bibby et al. Reference Bibby, Burgess and Hill1992, Ralph et al. Reference Ralph, Sauer and Droege1995). Birds occurring within 50-m fixed-radius of each point were recorded visually or acoustically. Point counts were surveyed twice in favourable weather conditions. Daily counts included 3-hour periods after sunrise. To reduce biases, experienced observers conducted all bird surveys (LR and EM with 10 and four years of experience, respectively). Nocturnal species and species that only overflew the forest (e.g. Andean Condor Vultur gryphus) were not considered. Taxonomic identification of birds was determined following Mazar-Barnett and Pearman (Reference Mazar Barnett and Pearman2001).
Data analysis
As a measure of relative density we calculated encounter rates per bird species (ER) recorded in each point count. ER was expressed as number of detections per point (10 min and 0.78 ha) for each species recorded. Richness and diversity of bird communities were calculated using EstimateS v.8.0 software (Colwell Reference Colwell2006). Sample species richness was estimated from the sample-based rarefaction curves (Mau Tau; Sobs; Mao et al. Reference Mao, Colwell and Chang2005). Sample was randomised 50 times for each dataset. To examine changes in species composition along the latitudinal gradient, the robust bootstrap estimator (Sboot; Colwell and Coddington Reference Colwell and Coddington1994) was used as a richness measure and Simpson's index (1/D) as a diversity measure. This index better reflects the entire species abundance distribution than other indices of the same general form (Magurran Reference Magurran1988). We also calculated the Simpson Equitability index (SE, Magurran Reference Magurran1988) to compare along the gradient the evenness with which individuals were distributed among the different species. Changes in composition of avian communities along the gradient (beta diversity) were tested with the Sorenson qualitative index (Magurran Reference Magurran1988).
We also analyzed the contribution to the Polylepis bird assemblage from birds of surrounding habitat types: (1) subtropical montane forests of the Andes (i.e., selva Tucumano-Boliviano), (2) shrublands/grasslands, or (3) not restricted to a single habitat, here defined as generalists (Fjeldså and Krabbe Reference Fjeldså and Krabbe1990, Stotz et al. Reference Stotz, Fitzpatrick, Parker and Moskovits1996, Narosky and Yruzieta Reference Narosky and Yzurieta2003). Functional groups were established according to habitat use of species as determined from the literature (Fjeldså and Kessler Reference Fjeldså and Kessler1996, Dardanelli et al. Reference Dardanelli, Nores and Nores2006). Groups were classified as: disturbance-sensitive species (DS, species that use mainly the forest: understorey, ground, medium stratum or canopy), disturbance-tolerant species (DT, species that primarily use forest edges, shrublands and open areas) and neutral species (N, habitat generalists). Here we defined habitat in the narrow sense of vegetation structure rather than as the full array of biotic and abiotic factors in the environment. We tested changes on bird communities among sites with a Kruskal-Wallis test and differences in the number of species of functional groups within and among sites with χ2 goodness of fit test (Zar Reference Zar1984).
Results
We recorded a total of 543 birds belonging to 50 species (Table 1). The most commonly recorded species were: Rufous-collared Sparrow Zonotrichia capensis with 27.6% (n = 150) of all detected individuals, followed by Chiguanco Thrush Turdus chiguanco 11.9% (n = 65), Red-tailed Comet Sappho sparganura 5.9% (n = 32) and Brown-capped Tit-spinetail Leptasthenura fuliginiceps 2.8% (n = 15). Bird richness (Sboot) decreased with latitude (Kruskall-Wallis Sboot = 14.42, P = 0.0007) with more species recorded at North and Centre sites than at the South site. Bird diversity followed a different pattern (Kruskall-Wallis 1/D = 21.88; P < 0.0001), with the highest diversity at the Centre site, intermediate at the South site and the lowest at the North site. The difference in richness - diversity patterns were due to a strong North - South increment in equitability index (Table 2).
Table 1. List of bird species and encounter rates expressed as number of detections per point (10-min. sampled period) recorded in Polylepis australis forest of Argentina. North (Jujuy province), Centre (Tucumán province) and South (Córdoba province) sites. DS: disturbance-sensitive species, DT: disturbance-tolerant species and N: neutral species. (*) Endemic subspecies. (+) Species closely associated with P. australis forests.
Table 2. Species richness, diversity and functional group composition estimated for bird communities of North, Centre and South sites of Polylepis australis forest of Argentina. ER: encounter rates expressed as the number of detections per 10-minute survey period. Sobs Mau Tau: sample species richness from sample-based rarefaction curves. Sboot: bootstrap estimator; 1/D: Simpson's index. Different letter shows significant differences (P < 0.05).
Sorenson's similarity index was low, showing a species turnover along the latitudinal gradient. North and Centre sites shared 11 species (40%), North and South shared five species (12%) and Centre and South sites shared seven species (34%). Approximately half of the species in each site occurred only in that site; in the North and Centre sites species richness was determined by species associated with subtropical montane forest, while the number of species of shrublands and grassland was similar in the three sites (Figure 2).
Figure 2. Bird species richness in North (N), Centre (C), and South (S) sites along the latitudinal range of Polylepis australis in Argentina. (a) Number of species recorded in the three sites (N+C+S), shared by two sites, or exclusive to one site and (b) number of species associated to subtropical montane forest, shrubland or grasslands, or not associated to a particular habitat (generalist).
In terms of habitat use groups (Table 2), there was a significantly higher number of disturbance-sensitive species (DS) detected in the North site (χ2 = 8.81, P = 0.01). The Centre and South sites did not show significant differences in the richness of sensitive, tolerant or neutral species (χ2 = 1.51 P = 0.46; χ2 = 0.86, P = 0.65, respectively). Comparing the number of species per habitat group along the gradient, only DS of the North site showed significant differences (χ2 = 6.93, P = 0.03) in respect to the other sites.
Discussion
Our results show a general pattern of decrease in bird richness and number of disturbance-sensitive species of Polylepis bird communities with latitude, coupled with an increment in the evenness of the species distribution. The lower equitability index at the North site resulted from the higher dominance of forest-dependent birds which caused a reduction of diversity index at this site. Many studies show that diversity can change with changes in evenness independently of species richness. That is, a community with evenly distributed species appears more diverse than a community that is dominated by a few species (Magurran Reference Magurran1988, Stirling and Wilsey Reference Stirling and Wilsey2001).
Besides this latitudinal avifaunal pattern, we found a high species turnover among sites. The species turnover could be influenced by vegetation types within and surrounding P. australis forest patches. North and Centre sites are structurally rich and their associations with other plant species such as vines, ferns and bromeliads makes them favourable feeding habitats for many birds (Fjeldså Reference Fjeldså1993). On the contrary, in the South site P. australis usually produces lateral basal ramifications forming shrublands and only occasionally forms dense forest stands (Enrico et al. Reference Enrico, Funes and Cabido2004, Cingolani et al. Reference Cingolani, Renison, Zak and Cabido2004, 2008, Renison et al. Reference Renison, Hensen, Suarez and Cingolani2006). These differences in the growth habit of P. australis change the forest structure (Kessler Reference Kessler2000) and availability, so could be inducing a reduction in bird richness (Bellis et al. unpubl. data). The positive response of bird species richness to habitat complexity has been well documented in tropical forests (Bennett et al. Reference Bennett, Hinsley, Bellamy, Swetnam and MacNally2004, Watson et al. Reference Watson, Whitttaker and Dawson2004, Barlow et al. Reference Barlow, Peres, Henrique, Stouffer and Wunderle2006). Specifically in Polylepis forests, habitat complexity and quality are important determinants of avian diversity (Terborgh Reference Terborgh1977, Kessler et al. Reference Kessler, Herzog, Fjeldså and Bach2001, Lloyd and Marsden Reference Lloyd and Marsden2008). Habitat features such as density of large trees, vegetation cover, and patch size are very important for many forest-dependent birds (Lloyd Reference Lloyd2008a,Reference Lloydb, Lloyd and Marsden Reference Lloyd and Marsden2008). Besides forest quality, the surrounding vegetation may also be an important driver of bird communities; we found that the birds that use Polylepis forests are mainly species of neighbouring vegetation types (i.e., subtropical montane forest; shrubland/grassland). Thus, in agreement with observations of Lloyd (Reference Lloyd2008a,Reference Lloydb) in Peruvian Polylepis forest, the matrix exerts a significant influence on avifauna composition.
In South America, the southern range distribution of avian species highly associated with Polylepis forests (e.g., Tawny Tit-spinetail Leptasthenura yanacensis, Thick-billed Siskin Carduelis crassirostris, Giant Conebill Oreomanes fraseri) only reached the Northern site, with no records at Centre or South sites. Therefore, we were unable to find bird specialists of P. australis forests, in contrast to Herzog et al. (Reference Herzog, Soria and Matthysen2003) and Lloyd (Reference Lloyd2008a,Reference Lloydb) who detected many highly associated bird species in tropical Polylepis forests of Bolivia and Peru.
We found four species, broadly distributed in other vegetation types, which were present in the three sites surveyed: Leptasthenura fuliginiceps, Sappho sparganura, Turdus chiguanco, and Zonotrichia capensis. The Polylepis forest assemblage has a mixed origin of species associated with subtropical montane forest and species associated with shrublands and grasslands.
Regarding the habitat sensitive species group, the North site harboured significantly more disturbance-sensitive species than tolerant or neutral species. The low similarity in composition of bird communities along the gradient, beta diversity, showed a species turnover from disturbance-sensitive species to species relatively resilient to human disturbance. Generalist and dispersive species such as House Wren Troglodytes aedon, Zonotrichia capensis, and Eared Dove Zenaida auriculata, became abundant towards the South to the detriment of forest species. Two possible factors that can compound the absence of habitat disturbance-sensitive species in the South site are the structural simplification of the vegetation and the high degradation of Polylepis forest. Most Polylepis specialists are insectivorous and therefore both vegetation structure and habitat disturbance (mainly fire) may reduce food resources for them (J. Cahill in litt.). Similar patterns of beta bird diversity changes were detected in a gradient of disturbed forest of central Argentina where increasing fire pressure caused an important structural simplification of the habitat (Albanesi et al. Reference Albanesi, Dardanelli, Heredia and Bellis2008), in forests of the tropical Andes (O'Dea and Whittaker Reference O'Dea and Whittaker2007) and, in logged versus unlogged forest of Bolivia (Felton et al. Reference Felton, Wood, Felton, Hennessey and Lindenmayer2008).
Of the total species recorded, two (Olrog's Grey-flanked Cincloides Cincloides oustaleti olrogi and Plain-coloured Seedeater Catamenia inornata cordobensis) are endemic subspecies that we recorded in the South site (EBA 058) and four (Black Siskin Carduelis atratus, Plumbeous Sierra-finch Phrygilus unicolor, Streak-fronted Thornbird Phacellodomus striaticeps, and Tucuman Mountain Finch Poospiza baeri) are typical Polylepis forests birds (Fjeldså Reference Fjeldså2002) that we recorded in the North and Centre sites (EBA 057). We also recorded four species of conservation concern: Poospiza baeri, a globally threatened species categorized as ‘Vulnerable’ by Birdlife International (2007) due to its restricted range and the loss of its natural habitat as a consequence of land conversion and use of pesticides (Peris Reference Peris1997, Di Giácomo Reference Di Giácomo2005). Grey-hooded Parakeet Bolborhynchus aymara, Leptasthenura fuliginiceps, and Turdus chiguanco are declining at regional level in the South site (Miatello et al. Reference Miatello, Baldo, Ordano, Rosacher and Biancucci1999), as a consequence of habitat loss caused primarily by fires and overgrazing. Livestock overgrazing in the South site is responsible of transformation of woodlands into grasslands and eroded rock surfaces (Cingolani et al. Reference Cingolani, Renison, Tecco, Gurvich and Cabido2008). This degradation process has negatively affected bird diversity at the landscape level (García et al. Reference García, Renison, Cingolani and Fernández-Juricic2008).
Implications for conservation
This work is a first approach to understanding the assemblage and status of the avifauna in Polylepis australis forest along its distribution range in Argentina and it constitutes a starting point for more comprehensive studies. Initial management recommendations can be formulated on the basis of our study. We recommend the expansion of Calilegua National Park to include the P. australis belt in its northern ranges – a project that is being considered by the present National Parks Administration. This is especially important as these Polylepis forest sites harbour many disturbance-sensitive species, which will disappear if these areas are degraded. We also recommend managing Quebrada del Condorito National Park in the southern ranges to augment the complexity of forest structure and thus provide suitable habitat for disturbance-sensitive species.
Acknowledgments
We are very grateful to American Bird Conservancy for providing financial support for this research and the National Parks Administration for granting study permits. We also give thanks to J. Cahill and an anonymous referee who made important suggestions that greatly improved the MS. L.M.B. and D.R. are researchers from CONICET. L.R. and E. M. have doctoral fellowships from YPF Foundation and CONICET, respectively.
