Andean bird responses to human disturbances along an elevational gradient

lable at ScienceDirect

Acta Oecologica 65-66 (2015) 51e60

Contents lists avai

Acta Oecologica

journal homepage: www.elsevier .com/locate/actoec

Original article

Andean bird responses to human disturbances along an elevationalgradient

Flavia A. Monta~no-Centellas a, *, �Alvaro Garitano-Zavala b

a Instituto de Ecología, Universidad Mayor de San Andr�es, Cota Cota c. 27 s/n, Casilla, 10077 La Paz, Boliviab Unidad de Manejo y Conservaci�on de Fauna, Instituto de Ecología, Universidad Mayor de San Andr�es, Casilla, 10077 La Paz, Bolivia

a r t i c l e i n f o

Article history:Received 31 March 2014Received in revised form6 May 2015Accepted 12 May 2015

Keywords:BoliviaDisturbanceEnvironmental gradientSpecies compositionSpecies richness

* Corresponding author. Present address: DepartmConservation, University of Florida, Gainesville, Florid

E-mail addresses: [email protected]@umsa.bo (�A. Garitano-Zavala).

http://dx.doi.org/10.1016/j.actao.2015.05.0031146-609X/© 2015 Elsevier Masson SAS. All rights res

a b s t r a c t

Understanding patterns of species diversity along environmental gradients is essential to the study ofbiodiversity. Numerous studies have found variation in species richness and composition of bird com-munities along elevational gradients, and several others described bird diversity changes followinganthropogenic disturbances. Surprisingly, few studies have attempted to disentangle their separate ef-fects on bird assemblages. Here, we explored variation in bird species richness and composition atdifferent levels and types of disturbance along a 4000-m elevational range in the tropical Andes. Birdcounts and disturbance measurements were conducted at 85 points distributed along the gradientwithin Cotapata National Park, Bolivia. Disturbances accounted for in our study correspond to the oftenoverlooked ‘moderate’ levels of disturbance that occur in the Tropical Andes. Diversity patterns weredescribed and compared with GLM models (for species richness) and CCA models (for species compo-sition). We found that bird communities were structured by elevation and disturbance. Species richnessdecreased with both elevation and habitat openness. Anthropogenic disturbances also modified com-munity composition within the same elevational ranges. We conclude that, whereas elevation remainedthe most important variable explaining bird species composition, disturbance explained species richnesspatterns to a higher extent.

© 2015 Elsevier Masson SAS. All rights reserved.

1. Introduction

The tropical Andes are naturally complex and their topography,biogeographic heterogeneity and elevational gradient have shapedthem into one of the most diverse ecosystems worldwide. Yet it isalso one of themost endangered (Myers et al., 2000; Swenson et al.,2012), with habitat modification being the major threat to this re-gion (Brooks et al., 2002). In the Bolivian Andes, natural hetero-geneity has increased since pre-Columbian times due toanthropogenic disturbances that have produced new environ-mental gradients for wildlife to interact with. Understanding hownatural and anthropogenic gradients interact to affect wildlife iscentral to conservation strategies for these complex landscapes(Cleary et al., 2005; Colorado, 2011; Swenson et al., 2012).

Wildlifeehabitat relationships and community responses to

ent of Wildlife Ecology anda, USA.(F.A. Monta~no-Centellas),

erved.

environmental gradients have been studied extensively, and birdshave been recognized as an ideal group to test hypotheses relatedto these topics (McCain, 2009). Birds are used in these type ofstudies because: (1) their taxonomy is well known, (2) surveymethods are standardized, facilitating comparisons among studies,and (3) they are diverse and have evolved a broad range of strate-gies (ecological niches). Empirical studies have shown that birdspecies richness may decrease linearly with increasing elevation, ormay show a mid-elevation peak (McCain, 2009; Wu et al., 2013).However, compositional changes in bird communities along ele-vations are still poorly understood (Acharya et al., 2011; Blake andLoiselle, 2000; Jankowski et al., 2009; Terborgh, 1971, 1977). Birdresponses to anthropogenic disturbance seem even more difficultto predict (Borges and Stouffer, 1999; Rodewald and Yahner, 2001)as bird responses to different types and intensities of human-caused disturbances are species-specific (Forsman et al., 2010;Lee et al., 2005; Petit et al., 1999; Verhulst et al., 2004; Walters,1998). Thus, any given combination of elevation and anthropo-genic disturbance is likely to shape bird assemblages in differentways (Cleary et al., 2005; Nogues-Bravo et al., 2008). Surprisingly,

F.A. Monta~no-Centellas, �A. Garitano-Zavala / Acta Oecologica 65-66 (2015) 51e6052

to our knowledge, no studies have attempted to disentangle theseparate effects of elevation and human disturbances on birdcommunities in the hyper-diverse tropical Andes.

In this study, we examine avian responses to anthropogenicdisturbances along an elevational gradient within a protectedarea in the Andes of Bolivia: Cotapata National Park and Inte-grated Management Natural Area (hereafter Cotapata NP). Cota-pata NP occurs within a region that is recognized for its highlevels of endemism (Swenson et al., 2012) and is the most diversenational park in Bolivia, in terms or bird species per area(Balderrama, 2009). Due to the complex topography large humansettlements in this landscape are uncommon, and most of thecurrent human activities are localized. Here, we consider ‘humandisturbance’ as temporal change in environmental conditions dueto human activities, and we use four proxies to measure it: size ofhuman settlements, amount of domestic animals (cattle), amountof waste, and presence of burn areas. Additionally, we measuredtwo environmental variables that could result both from theelevational gradient and from the human activities in the area:vegetation cover and structural complexity (see Environmentalvariables for details). Disturbances accounted for in our studyoccur across the 4000-m elevational gradient of Cotapata NP andrepresent the most characteristic human-caused disturbances inthe region.

Due to the strong temperature and humidity shifts along theelevational gradient of the park (Bach et al., 2007; Jones et al., 2011),we expected elevation to be the most important variable definingbird communities (McCain, 2009). Nevertheless, we also expecteddisturbances to modify communities, resulting in different as-semblages at similar elevational ranges (Nogues-Bravo et al., 2008).Moreover, because natural disturbances such as landslides andtreefall gaps are common in mountain forests, we expected humandisturbances that resemble natural disturbances (i.e. small clear-ings) to be tolerated by most species whereas disturbances that donot resemble natural clearings (i.e. fires) might have a strongereffect on less resilient birds.

2. Materials and methods

2.1. Study area

We conducted our study in Cotapata NP (67�430e68�020W;16�100�16�200S; Fig.1), La Paz, Bolivia. Cotapata NP protects distinctvegetation types that develop between the high Andean plateauand the Andean foothills (5600e1000m asl). The park is dominatedby evergreen humid montane and cloud forests, with steep slopesand deep valleys, while natural grasslands occur at higher eleva-tions (>3900 m asl). Small natural disturbances, such as landslidesand treefall gaps, are common, as are scattered active and aban-doned small agricultural clearings (~0.5e3 ha) within a forestedmatrix (Ribera, 1995; Sevilla Callejo, 2010). Climate is seasonal,with a long wet season that peaks in January and February and atwo or three month dry season that peaks in June or July (Joneset al., 2011; Molina-Carpio, 2005).

To conduct our studies, we used all the existing hiking trails inthe park as our sample routes because the complex topographymakes it virtually impossible to access most areas within CotapataNP (Fig. 1). Taken together, the trails extend across most of thepark's gradient (~1000e5000m asl), and traverse different levels ofdisturbance. Most of the trail network is narrow (between < 1 and1.5 mwide), although some sections can be up to 3.5 m wide. Also,some sections of the trail network are pre-Inca: half are regularlyused by local people and pilgrims, whereas the other half was re-opened for this study. Thus, reopened trails constituted a longtransect that crossed well-conserved forest across elevations

whereas more actively use trails crossed disturbed areas acrosselevations. As described below, we measured human disturbancesat determined points across elevations. For comparison purposesonly, our higher level of disturbance would correspond to the‘moderate’ level of disturbance in most studies (Gray et al., 2007;Lefevre et al., 2012). Therefore, human disturbance levels in ourstudy range from relatively ‘undisturbed’ (the only impact being usopening a transect) to ‘moderately disturbed’ habitats (hereafter‘disturbed’).

2.2. Bird surveys

Between April 2006 and April 2007, we recorded birds at 85points (Fig. 1). We used a GPS unit (E-trex, Garmin®) to recordUTM coordinates and an altimeter (KONUS®) to record elevationat each point. Because of the tortuous nature of the trails andcomplex topography, and based on our previous experience inthis forest, walking was selected as a proxy for distance in thefield; points were separated by 50 min walk from each other. Theminimum linear distance (calculated using ArcView 3.2) betweenconsecutive points was ~340 m (mean ¼ 843.1 ± 507.5 SD), anappropriate distance to consider points independent and todetect the often overlooked small-scale (<1 km) shifts in envi-ronmental conditions that characterize tropical montane forests(Jones et al., 2011; McCain, 2009). We used non-fixed-radius 10-min point counts to sample birds (Blondel et al., 1981); countswere carried out by a single experienced observer accompaniedby a field assistant. Each of the 85 points (replicates) was sampledsix times throughout the year (approximately every two months)to account for seasonal variation. Similarly, each point wassampled at different times of the day (between sunrise and 11:30and between 13:30 and 17:30) to help account for diurnalchanges in bird activity. Because we wanted to make sure thatregistered birds were actually present in the small areas wheredisturbances were measured (see Environmental variables), weregistered only birds detected visually; we noted the number ofspecies and individuals at each point. Thus, when we heard a birdsinging, we looked for the bird and registered it only if it wasseen. Although we are aware that these types of surveys mayundersample some groups, it has been suggested that employingidentical methods using the same observer would producecomparable results (Woltmann, 2003). Moreover, we assessed thecompleteness of our counts at each point by calculating the ex-pected species richness using the Jack1 richness estimator withthe software EstimateS (Colwell, 2009), and ran all our analyseswith both observed and estimated richness.

2.3. Environmental variables

We measured seven environmental variables in a 25-m radiuscircular area surrounding each point: elevation, vegetation cover,structural complexity, house density, waste, burn and cattle.Although vegetation cover and structural complexity can not besolely attributable to anthropogenic disturbances, they do decreasewith disturbance. Thus, in combination with disturbance mea-surements (burn, cattle, human settlements and waste) they betterdescribe the environmental changes caused by human activities inthe evaluated points. Environmental variables were measured asfollow: (1) elevation (to the nearest meter); (2) vegetation cover,estimated visually and expressed as percentage of surface coveredby any type of vegetation and then categorized as follows: <20%,20e40%, 40e60%, 60e80% and 80e100%; (3) structural complexity,based on the ‘habitat heterogeneity hypothesis’(MacArthur andWilson, 1967) we visually examined each point and categorized itas belonging to one of seven increasingly complex habitat

Fig. 1. Map of Cotapata NP (Department of La Paz, Bolivia) showing the location and distribution of study points across the park. Isometric lines were drawn every 500 m between1000 and 5500.

F.A. Monta~no-Centellas, �A. Garitano-Zavala / Acta Oecologica 65-66 (2015) 51e60 53

categories (i.e. each with more microhabitats available for birds)(Hewitt et al., 2005; Tews et al., 2004): high Andean grassland, highAndean wetlands, traditional plantations, 5e10-year abandonedplantations in early recovery, young secondary forest, young sec-ondary forest with bamboo, and old-growth forest; (4) house,expressed as the number of houses visible from the center of thepoint and categorized as: 0, 1, 2e10, and >10 houses; (5) waste,visually estimated as the number of visible inorganic items withinthe 25-m radious point and categorized as: none,1 to 4, 5 to 10, and>10 items; (6) burn, a visually estimated dummy variable thataccounted for evidence of recent fire (no burn, burn) where burnwas noted only if all or most of the 25-m radius sample wasaffected; (7) cattle, recorded as the number of cows and other do-mestic animals (i.e. horses, llamas and sheeps) visible from thecenter of the circle: 0, 1e4, 5e15, and >15 animals, grouping alldomestic animals together. Waste and cattle weremeasured duringeach of the six surveys and averaged over time for each point,whereas all other variables were measured once during the firstsurvey. No fires occurred during the sampling period. A summary ofthese variables is presented in Table S1.

2.4. Analyses

No birds were recorded at two census points so these pointswere dropped from all analyses. Thus, we used the same data (83points) for both GLMs and ordinations to better contrast their re-sults. All birds species were included in the analysis except thosethat only flew over the point (Bibby et al., 1992). The overall birdcommunity was described with species richness and rankeabun-dance curves (Wilson, 1991). We described the bird assemblage ateach point on the basis of (1) species richness, (2) species compo-sition and (3) maximum abundance of each species (i.e. highestnumber of individuals registered during a single count).

Our exploratory analyses showed significant associations amongsome of the environmental variables (Table S2). However, no strongassociations between elevation and any categorical variable wasfound, thus we used a principal coordinate analysis (PCoA, withGower distance) to reduce dimensionality of the categorical vari-ables (Gower, 1966). We decided to keep the first two PCoA axes forfurther analyses because they explained 80% of the variation in thedata and were highly correlated with determined environmental

F.A. Monta~no-Centellas, �A. Garitano-Zavala / Acta Oecologica 65-66 (2015) 51e6054

variables (Table 1): the first axis (PCoA1) was positively correlatedwith vegetation cover and structural complexity, and negativelycorrelated with waste and house e two indicators of constant hu-man activity. Thus, this new variable describes a gradient of“habitat openness”. The second axis (PCoA2) was negativelycorrelated with burn, and positively correlated with cattle.Although more difficult to interpret, we considered this as agradient of burn. Neither of the PCoA axes was strongly correlatedwith elevation (Table 1).

To test for species richness responses, we first used Moran's Icoefficient implemented in the ape package (Paradis et al., 2008) toexamine if species richness was explained by geographic distancesamong points (i.e. spatial autocorrelation). Moran's I coefficient hasa value near zero when there is no spatial autocorrelation in thedata; positive or negative values of Moran's I indicate positive ornegative autocorrelation, respectively (Diniz-Filho et al., 2003;Legendre, 1993). Because we did not find evidence of significantautocorrelation among points for observed richness data(I ¼ �0.03, P ¼ 0.07) nor for estimated richness (I ¼ �0.02,P ¼ 0.08), we did not include spatial predictors in our analyses ofspecies richness. The relationship between avian species richnessand environmental variables was analyzed with a generalizedlinear model (GLM), assuming Poisson distribution errors, withelevation, PCoA1 and PCoA2 as explanatory variables. Interactionsamong explanatory variables were also considered in themodel. Notransformation was applied to species richness data.

Species composition responses were analyzed with canonicalcorrespondence analyses (CCA) (ter Braak, 1986) performed inVegan and ade4 packages (Dray and Dufour, 2007; Oksanen et al.,2013). Species abundances were log-transformed [log(xþ 1)](Legendre and Gallagher, 2001) and environmental variables wereusedwithout transformation.We examined linear dependencies bycomputing each environmental variable's variance inflation factor(VIF), which measures the proportion by which the variance of aregression coefficient is inflated in the presence of other explana-tory variables. Values of VIF above 10 suggest potential collinearityand should be avoided (Borcard et al., 2011). All variable VIFs werebelow 4 (2.0, 3.6, 3.5, 2.8, 1.9, 2.1 and 3.7 for elevation, cover,complexity, burn, cattle, house and waste, respectively), thus, allseven variables were initially included in the analyses (McCuneet al., 2002).

We examined if similarity in bird communities among points(BrayeCurtis distances) could be partially explained by geographic(Euclidean) distances with a Mantel test (1000 Monte Carlo per-mutations) performed in the Vegan package (Oksanen et al., 2013).We found a weak yet significant lack of geographic independence

Table 1The two first axes of the principal coordinate analyses (PCoA) of categorical envi-ronmental variables (PCoA1 and PCoA2) and their correlation with explanatoryvariables. Pearson's correlations between explanatory variables and these two PCoAaxes are presented as r (P). Correlations are considered significant if they are above0.70 and with P < 0.05.

PCoA 1 PCoA 2

Eigenvalue 2.79 1.01Proportion of explained variation 0.59 0.21Cummulative explained variation 0.59 0.80Correlations with explanatory variablesElevation 0.07 (0.51) 0.53 (0.001)Vegetation cover 0.79 (<0.001) 0.19 (0.09)Structural complexity 0.76 (<0.001) �0.37 (<0.001)House �0.72 (<0.001) 0.07 (0.54)Waste �0.92 (<0.001) 0.08 (0.49)Burn �0.56 (<0.001) �0.74 (<0.001)Cattle �0.46 (<0.001) 0.72 (<0.001)

among points (Mantel test: r ¼ 0.137, P ¼ 0.01), which could resultfrom the natural east-to-west elevational gradient in Cotapata NP(see Fig. 1). Therefore, in addition to the full CCAmodel as describedabove, we run a partial CCA model to examine composition re-sponses to elevation and environmental variables while controllingfor the spatial autocorrelation (Borcard et al., 2011). For this, wefollowed Dray et al. (2006) and used Moran's eigenvector maps(MEM) to extract the eigenvalues that maximize Moran's index ofautocorrelation among points. MEMs were calculated with thepackage PCNM (Legendre et al., 2012). Then, we followed Blanchetet al. (2008) and used a forward-selection procedure to the MEMspatial predictors (eigenvectors). Eleven MEMs, explaining 37% ofthe total variation of species composition, were selected. Finally, weincluded these eleven eigenvectors as spatial explanatory variablesand controlled for their potential effect with a partial CCA.

Significances of the model, axes and explanatory variables ofeach canonical model were tested by contrasting them with nullmodels generated by 1000 Monte Carlo permutations (Legendreand Legendre, 2012). Finally, we used a forward-selectionapproach to select for the most parsimonious CCA modelincluding only the significant predictors. Because both modelsprovided the same conclusions, we report and base our discussionon the full CCA model (partial model can be reviewed in theSupplementary Material). Analyses were done within the R envi-ronment (R Development Core Team, 2009).

3. Results

3.1. Species richness responses

We registered a total of 1434 individuals from 178 bird species(Appendix). A relatively low number of species dominated thecommunity due to their abundance or frequency across points (e.g.Zonotrichia capensis, Psarocolius atrovirens, Psarocolius decumanus,Ramphocelus carbo, Cinclodes albiventris, Atlapetes rufinucha, Myio-borus miniatus; Fig. S1). Observed species richness per point rangedfrom 1 to 19 (6.59 ± 3.7SD) and estimated species richness (Jack1)ranged from 1 to 27 species per point (11.0 ± 6.2SD). Two envi-ronmental predictors were determinants for observed and ex-pected species richness: elevation and habitat openness (PCoA1).Both observed and estimated species richness decreased withelevation and habitat openness (Fig. 2), with considerable variationamong the data. Contrary to expectations, elevation explained lessof the variation in species richness than did habitat openness(Table 2).

3.2. Species composition responses

The full CCA model (F ¼ 1.69, P ¼ 0.001) had three significantaxes that explained 73.4% of the variation in bird species compo-sition (Table 3). Overall, the partial CCA model showed the samepatterns (Table S3). Axis 1 of the full CCA model (l ¼ 0.78) washighly correlated with elevation (r ¼ 0.91, P < 0.001); speciesknown to inhabit Andean highlands, such as Phrygilus spp., C.albiventris and Muscisaxicola cinereus were separated along Axis 1from species associated with lower elevations (Fig. 3). Species inthe middle of the plot might be either altitudinally ubiquitous orassociated with middle elevations. Axis 2 (l ¼ 0.61) represented adisturbance gradient as it was positively correlated with vegetationcover and structural complexity (r ¼ 0.79, P < 0.001 and r ¼ 0.65,P < 0.001, respectively) and negatively related with waste(r¼�0.72, P < 0.01) and house density (r¼�0.53, P < 0.001). Thus,this axis separated species based on their tolerance to anthropo-genic disturbances. Axis 3 (l ¼ 0.40; not shown in the figure) wasweakly correlated with presence of cattle, occurrence of fire, and

Fig. 2. Relationship between bird species richness and elevation (a and c for observed and expected species richness, respectively) and habitat openness (b and d for observed andexpected species richness, respectively) along the altitudinal gradient of Cotapata NP.

Table 2Generalized linear models explaining the relationship between observed avianspecies richness and environmental variables (Model 1) and estimated speciesrichness and environmental variables (Model 2). Final models include only signifi-cant terms, and were selected by a backward selection procedure.

Model Coefficients

Coefficient SE Z value P

Model 1Intercept 2.07 0.11 18.31 <0.001Elevation (m � 103) �0.09 0.04 �2.03 0.042PCoA 1 �1.25 0.22 �5.72 <0.001Model 2Intercept 14.18 1.76 8.07 <0.001Elevation (m � 103) �1.38 0.04 �2.00 0.049PCoA 1 �11.47 3.44 �3.33 0.001

Table 3The four significant axes in the ordination of species and point counts, as a functionof the seven environmental variables tested (elevation, vegetation cover, structuralcomplexity, house density, cattle, waste and burn), obtained by the full CCA model.

Axis 1 Axis 2 Axis 3

Eigenvalue (l) 0.782 0.607 0.401% Explained inertiaa 25.07 18.28 13.92F (P)b 2.67 (0.001) 2.08 (0.001) 1.39 (0.001)

a Percentage of the restrained to environmental variables inertia, explained byeach axis.

b Fischer's F value for each axis, obtained with an ANOVA, contrasting a nullmodel generated by 1000 Monte Carlo permutations. Statistical significance of P-value is presented in parentheses.

F.A. Monta~no-Centellas, �A. Garitano-Zavala / Acta Oecologica 65-66 (2015) 51e60 55

presence of waste (r ¼ 0.57, 0.43, and 0.28, respectively). Speciessegregated along both Axis 1 and 2, suggesting that disturbancesaffected species composition along the elevational gradient (Fig. 3).Significance test of explanatory variables in the CCA model showsthat elevation, vegetation cover, structural complexity, cattle andwaste were significantly contributing to the model (Table S4).

Species-specific responses to disturbance and elevation wererelatively evident from multivariate analyses. Species such asZcapensis, Turdus spp., Pitangus sulphuratus, Thraupis palmarum,Thraupis sayaca, Troglodytes aedon and Myiarchus ferox werepositively related to disturbance. In contrast, several species thatwere relatively frequent in our study were strongly associatedwith lower disturbance levels, such as Pipreola intermedia, Pipreolaarcuata, Entomodestes leucotis, Catharus dryas, Margarornis squa-miger, Tangara vassorii, Conirostrum albifrons, Heliangelus ame-thysticollis, Hemispingus trifasciatus and Hemispingusxanthophthalmus (Fig. 3). Finally some of the most frequentlyobserved species in montane forest were apparently associatedwith intermediate levels of disturbance, such as Ampelionrubrocristata, Anisognathus igniventris, Buthraupis montana,Myiothlypis spp., Mionectes striaticollis, Ochthoeca spp. and Chi-roxiphia boliviana.

4. Discussion

Bird communities along our elevational gradient in CotapataNP are structured by both elevation and disturbance. Speciesrichness responses were not strong: although an overall pattern ofrichness decline with elevation and habitat openness (i.e. acombination of decreasing vegetation cover and increasing set-tlements and waste) was found, the rate of decrease was low. On

Fig. 3. Ordination of bird species plotted as a function of the first two axis of CCAwhenconsidering all environmental variables as response variables. Axis 1 represents anelevational gradient (l ¼ 0.78, highly correlated with elevation r ¼ 0.91) with higherelevations to the right. Axis 2 represents a disturbance gradient [l ¼ 0.61; positivelycorrelated with vegetation cover and structural complexity (r ¼ 0.79 and r ¼ 0.65,respectively) and negatively related with waste (r ¼ �0.72, P < 0.01) and house density(r ¼ �0.53, P < 0.001), with less disturbed areas in the upper half of the plot and moredisturbed areas in the lower part. Thus, Axis 2 separates species based on theirtolerance to anthropogenic disturbances. Bird codes are described in the Appendix.

F.A. Monta~no-Centellas, �A. Garitano-Zavala / Acta Oecologica 65-66 (2015) 51e6056

the contrary, compositional responses were clearer. We found thatelevation remained the most important variable explaining birdcomposition, yet anthropogenic disturbances modified commu-nity composition, even within the same elevational ranges.Furthermore, these changes in composition suggest that avianresponses may not only be species-specific but may depend on thenature of the disturbance: disturbances that mimic naturalclearings such as small clearings and pastures had smaller effectson composition than disturbances that do not resemble naturalconditions in montane habitats (burn, human settlements).Interestingly, despite these differences in the strength of the re-sponses in species richness and composition, we found that thesame combination of variables were determinant for both rich-ness and composition. The most important PCoA axis and themost important CCA axis (one used as an explanatory variable forspecies richness and the other explaining composition) werecompose by the same environmental variables: decreasing vege-tation cover and structural complexity, and increasing humansettlements and waste.

Bird species richness is known to decrease with elevation inmontane habitats (Blake and Loiselle, 2000; Herzog et al., 2005;McCain, 2009; Terborgh, 1977), and these patterns have beenrelated to changes in productivity along elevational gradients thatare associated with changes in temperature and humidity(Rensburg et al., 2002). However, these patters could be modifiedby processses that occur more locally, such as human disturbances(Nogues-Bravo et al., 2008). Our results are consistent with thisstatement. We found a regular but slow rate of decline in bothobserved and expected species richness along the elevational

gradient at Cotapata NP. Yet, in addition to the effects of elevation,species richness per point decreased with habitat openness andincreasing human settlements and waste (Table 2). Contrary toour expectations e for such a broad gradientethe coefficient forelevation in our GLM model was lower and less significant thanthat of habitat openness (i.e. a combination of decreasing vege-tation cover and increasing settlements and waste), suggestingthat human disturbances are as important as elevation in deter-mining species-richness patterns along our gradient. This resultcould be explained by human disturbances causing substantialvariation among point counts at similar elevations. As most dis-turbances in Cotapata NP are moderate (Gray et al., 2007; Lefevreet al., 2012), we expected some species to be able to take advan-tage of moderately disturbed habitats, thereby enriching theassemblage as a whole (Terborgh and Weske, 1975) and, at thesame time, creating great variation among points at the sameelevation. For example, we found that assemblages of highlydisturbed areas were increased because of the addition of tolerantspecies such Z. capensis, R. carbo, Pygochelidon cyanoleuca and P.decumanus, that were broadly distributed (more ubiquitous) andassociated with intermediate and high disturbance levels. Besidesreducing vegetation cover and modifying natural heterogeneity,small human settlements in Cotapata NP (up to 40 families)introduce exotic resources such as flowers and fruits in gardensand orchards that could attract generalists and increase speciesrichness (Soh et al., 2006). Similarly, localized fires in tropicalforests may provide more resources for generalists (particularlyinsectivorous and frugivorous species) and offer safe perches forthem to watch for predators (Borges and Stouffer, 1999;Woltmann, 2003). In Cotapata NP, localized burns are used as away to open new agricultural sites; however, most of the sites arenot used and are left for natural regeneration to take over. Mostburned areas (however recent) contain several dead or recoveringtrees, and birds were often seen perching and gliding close tothem. The fact that opportunistic species use these disturbedareas might partially explain why burn and cattle did not explainmuch of the variation in species richness.

Species resilience to disturbances will depend on their naturalhistory attributes as well as the type and intensity of disturbance(Devictor et al., 2008; Karp et al., 2011; Soh et al., 2006). Somespecies appeared to be highly resilient in Cotapata NP. For example,R. carbo, Tersina viridis, T. sayaca and Tyrannus melancholicus wheremore associated with burnt areas, whereas species of the generaPhrygilus and Sicalis, as well as the nectarivore Amazilia chia-nogaster were frequent in areas with low complexity, low vegeta-tion cover and cattle, where human disturbances resemble naturalopen areas that these species are known to use. As expected, mostspecies inhabiting the inhospitable high Andean plateau werefound in the lower half of the plot, whereas species in middle andlower elevations segregated along the disturbance gradient (Axis2). This result could reflect a more resilient nature of these speciesadapted to cope with harsh conditions. Alternatively, this patterncould be an artifct due to the natural depauperate vegetationstructure in the highlands. However, because the elevation vector isdirected towards the upper half of the plot and most of thesespecies are located nearby the cattle vector, we think our resultsmight provide more support for the first hypothesis.

Sensitive species (sensu Stotz et al., 1996) were expected to benegatively affected by disturbance (Chapman and Reich, 2007),especially those with naturally restricted distributions. Forinstance, Cranioleuca albiceps and Pseudocolaptes boissonneautii,two distribution-restricted species that inhabit Cotapata NP, wereabsent from disturbed areas and strongly associated with undis-turbed sites in our study. However, other species considered assensitive, such as E. leucotis, C. dryas, M. squamiger and H.

F.A. Monta~no-Centellas, �A. Garitano-Zavala / Acta Oecologica 65-66 (2015) 51e60 57

trifasciatus, were relatively more resilient to disturbances in Cota-pata NP. Although they were strongly associated with undisturbedareas, they were also present in areas with intermediate levels ofdisturbance.

Species inmixed-species flocks (e.g. Chlorospingus flavopectus, A.igniventris, Conirostrum ferrugineiventre, C. albifrons, Myiothlypisbivittatus, Myiothlypis signatus and Myiothlypis luteoviridis) per-sisted but were less abundant at higher disturbance levels and,thus, were associated with medium levels of disturbance. As ex-pected, some of them (C. flavopectus, A. igniventris,C. ferrugineiventre) along with other species associated with me-dium levels of disturbance (C. boliviana, T. viridis, Tangara xantho-cephala, among the most frequent), are understory birds that feedon successional species (Gomes et al., 2008). Fruit and flowerabundance might be higher in moderately disturbed habitatsbecause of improved light conditions, and fruits in open habitatsmay be especially sugary and attract more birds than fruits in themore shaded forest (Lumpkin and Boyle, 2009). Irrespective of themechanism, even low levels of disturbance in tropical forests cansignificantly change community composition and affect speciesinteractions (Lefevre and Rodd, 2009; Lefevre et al., 2012). Thepotential effects of habitat disturbances on birds that engage inmixed flocks in Cotapata NP is still unknown, and could representan important predictor of forest health and ability to harbor func-tional biodiversity. Although most Trochilidae and other nectar-feeding species were associated with undisturbed habitats, a fewspecies were strongly related with human settlements (e.g. Colibricoruscans and Setophaga pitiayumi), where ornamental and culti-vated plants provide an abundance of resources.

Despite its great potential for bird conservation, no prior studieshave analyzed birdehabitat relationships in Cotapata NP, evenwhen this information is key to determining the park's conserva-tion potential (Colorado, 2011). In this study, we present the firstattempt to explore how avifaunas in such a complex landscaperespond to disturbances along the elevational gradient. Because weonly included birds that were seen, our results may be biased

Birds observed along an elevational gradient at Cotapata National Park. Scientific names

Family Species

Cracidae Nothocercus nigrocapillusChamaepetes goudotiiPenelope montagniiOrtalis guttata

Threskiornithidae Plegadis ridgwayiTheristicus melanopis

Cathartidae Cathartes auraScolopacidae Gallinago andinaLaridae Chroicocephalus serranusColumbidae Patagioenas fasciata

Leptotila verreauxiLeptotila megaluraGeotrygon frenata

Cuculidae Piaya cayanaTrochilidae Phaethornis malaris

Doryfera ludovicaeColibri delphinaeColibri thalassinusColibri coruscansHeliangelus amethysticollisAdelomyia melanogenysAglaiocercus kingiiOreotrochilus estellaChalcostigma ruficepsMetallura tyrianthinaMetallura aeneocaudaEriocnemis glaucopoides

towards species that occur in more open areas or are more abun-dant or vocal. Also, a common caveat of point counts as a surveymethod to compare different habitats is the higher detectability ofspecies in more open vegetation, thus we likely missed severalquiet or secretive species, especially those in denser vegetationcover. Furthermore, despite our efforts to sample representativeareas of the park by re-opening abandoned trails, as in manytransect-based avian surveys, we cannot extrapolate our results tothe whole park because our sampling was restricted to the fewaccesible trails within it. However, evenwhen limited to this subsetof the potential species inhabiting the park, our results suggest thatdisturbances along the trails in Cotapata NP have created a richmosaic landscape, where different bird assemblages can be seenalong the elevational gradient.

Author contributions

FMC originally formulated the idea, conducted fieldwork, dataanalyses and wrote the manuscript. AGZ collaborated developingthe methodology and with conceptual and editorial advice.

Acknoweldgements

We thank M. Villegas, J. M. Molina, F. Saavedra, P. Duchen andlocal guides for their help in the field work; L. Pacheco providedvaluable advice that helped improve the project. Comments from J.G. Blake, B. A. Loiselle, B. Baiser and four anonymous reviewersgreatly improved the manuscript. Permissions to enter Cotapata NPwere provided by the Ministerio de Desarrollo Sostenible yPlanificaci�on (MDSP) and the Servicio Nacional de �Areas Protegidas(SERNAP). The study was funded with a WWF Russell E. Trainscholarship granted to FMC.

All authors have approved the final version of the article.

Appendix

follow Remsen et al. (2013).

English name Code

Hooded Tinamou NONISickle-winged Guan CHGOAndean Guan PEMOSpeckled Chachalaca ORGUPuna Ibis PLRIBlack-faced Ibis THMETurkey Vulture CAAUPuna Snipe GAANAndean Gull LASEBand-tailed Pigeon COFAWhite-tipped Dove LEVELarge-tailed Dove LEMEWhite-throated Quail-Dove GEFRSquirrel Cuckoo PICAGreat-billed Hermit PHMAGreen-fronted Lancebill DOLUBrown Violetear CODEGreen Violetear COTHSparkling Violetear COCOAmethyst-throated Sunangel HEAMSpeckled Hummingbird ADMELong-tailed Sylph AGKIAndean Hillstar ORESRufous-capped Thornbill CHRUTyrian Metaltail METYScaled Metaltail MEAEBlue-capped Puffleg ERGL

(continued on next page)

(continued )

Family Species English name Code

Coeligena torquata Collared Inca COTOCoeligena violifer Violet-throated Starfrontlet COVIPterophanes cyanopterus Great Sapphirewing PTCYOcreatus underwoodii Booted Racket-tail OCUNChlorostilbon mellisugus Blue-tailed Emerald CHMLThalurania furcata Fork-tailed Woodnymph THFUAmazilia chionogaster White-bellied Hummingbird AMCH

Trogonidae Trogon personatus Masked Trogon TRPEBucconidae Malacoptila fulvogularis Black-streaked Puffbird MAFUCapitonidae Eubucco versicolor Versicolored Barbet EUVERamphastidae Aulacorhynchus prasinus Emerald Toucanet AUPR

Aulacorhynchus derbianus Chestnut-tipped Toucanet AUDEAulacorhynchus coeruleicinctis Blue-banded Toucanet AUCO

Picidae Colaptes rivolii Crimson-mantled Woodpecker CORIFalconidae Phalcoboenus megalopterus Mountain Caracara PHME

Falco sparverius American Kestrel FASPFalco peregrinus Peregrine Falcon FAPE

Psittacidae Aratinga mitrata Mitred Parakeet ARMIPyrrhura molinae Green-cheeked Parakeet PYMOHapalopsittaca melanotis Black-winged Parrot HAME

Thamnophilidae Thamnophilus ruficapillus Rufous-capped Antshrike THRUDysithamnus mentalis Plain Antvireo DYMEDrymophila striaticeps Streak-headed Antbird DRCA

Conopophagidae Conopophaga ardesiaca Slaty Gnateater COARGrallariidae Grallaria squamigera Undulated Antpitta GRSQ

Grallaria rufula Rufous Antpitta GRRURhinocryptidae Scytalopus bolivianus Bolivian Tapaculo SCBOFurnariidae Geositta cunicularia Common Miner GECU

Sittasomus griseicapillus Olivaceous Woodcreeper SIGRDeconychura longicauda Long-tailed Woodcreeper DELODendrocolaptes picumnus Black-banded Woodcreeper DEPIXiphorhynchus ocellatus Ocellated Woodcreeper XIOCXiphorhynchus triangularis Olive-backed Woodcreeper XITRLepidocolaptes lacrymiger Montane Woodcreeper LELAPseudocolaptes boissonneautii Streaked Tuftedcheek PSBOUpucerthia validirostris Buff-breasted Earthcreeper UPJECinclodes albiventris Cream-winged Cinclodes CIFUAnabacerthia striaticollis Montane Foliage-gleaner ANSTMargarornis squamiger Pearled Treerunner MASQAsthenes humilis Streak-throated Canastero ASHUAsthenes harterti Black-throated Thistletail ASHACranioleuca albiceps Light-crowned Spinetail CRALSynallaxis azarae Azara's Spinetail SYAZ

Tyrannidae Zimmerius bolivianus Bolivian Tyrannulet ZIBOTyrannus melancholicus Tropical Kingbird TYMETolmomyias sulphurescens Yellow-olive Flycatcher TOSUSayornis nigricans Black Phoebe SANIPyrrhomyias cinnamomeus Cinnamon Flycatcher PYCIPseudotriccus ruficeps Rufous-headed Pygmy-Tyrant PSRUPoecilotriccus plumbeiceps Ochre-faced Tody-Flycatcher POPLPitangus sulphuratus Great Kiskadee PISUPhylloscartes ventralis Mottle-cheeked Tyrannulet PHVEOchthoeca rufipectoralis Rufous-breasted Chat-Tyrant OCRUOchthoeca pulchella Golden-browed Chat-Tyrant OCPUOchthoeca oenanthoides d'Orbigny's Chat-Tyrant OCOEOchthoeca leucophrys White-browed Chat-Tyrant OCLEOchthoeca fumicolor Brown-backed Chat-Tyrant OCFUOchthoeca cinnamomeiventris Slaty-backed Chat-Tyrant OCCIMyiodynastes maculatus Streaked Flycatcher MYMAMyiodynastes chrysocephalus Golden-crowned Flycatcher MYCHMyiarchus ferox Short-crested Flycatcher MYFEMyiarchus cephalotes Pale-edged Flycatcher MYCEMuscisaxicola rufivertex Rufous-naped Ground-Tyrant MURUMuscisaxicola cinereus Cinereous Ground-Tyrant MUCIMionectes striaticollis Streak-necked Flycatcher MISTMionectes oleagineus Ochre-bellied Flycatcher MIOLMecocerculus leucophrys White-throated Tyrannulet MELELessonia oreas Andean Negrito LEORHemitriccus spodiops Yungas Tody-Tyrant HESPElaenia pallatangae Sierran Elaenia ELPAElaenia obscura Highland Elaenia ELOBElaenia albiceps White-crested Elaenia ELALAgriornis montanus Black-billed Shrike-Tyrant AGMO

Cotingidae Pipreola intermedia Band-tailed Fruiteater PIINPipreola arcuata Barred Fruiteater PIARPipreola frontalis Scarlet-breasted Fruiteater PIFR

F.A. Monta~no-Centellas, �A. Garitano-Zavala / Acta Oecologica 65-66 (2015) 51e6058

(continued )

Family Species English name Code

Ampelion rubrocristatus Red-crested Cotinga AMRURupicola peruvianus Andean Cock-of-the-rock RUPE

Pipridae Chiroxiphia boliviana Yungas Manakin CHBOVireonidae Vireo leucophrys Brown-capped Vireo VILECorvidae Cyanolyca viridicyanus White-collared Jay CYVI

Cyanocorax cyanomelas Purplish Jay CYCYHirundinidae Pygochelidon cyanoleuca Blue-and-white Swallow PYCY

Orochelidon murina Brown-bellied Swallow ORMUOrochelidon andecola Andean Swallow ORAN

Troglodytidae Troglodytes aedon House Wren TRAETroglodytes solstitialis Mountain Wren TRSOHenicorhina leucophrys Gray-breasted Wood-Wren HELE

Turdidae Myadestes ralloides Andean Solitaire MYRACatharus dryas Spotted Nightingale-Thrush CADREntomodestes leucotis White-eared Solitaire ENLETurdus amaurochalinus Creamy-bellied Thrush TUAMTurdus fuscater Great Thrush TUFUTurdus chiguanco Chiguanco Thrush TUCH

Mimidae Mimus dorsalis Brown-backed Mockingbird MIDOThraupidae Hemispingus superciliaris Superciliaried Hemispingus HESU

Hemispingus melanotis Black-eared Hemispingus HEMEHemispingus xanthophthalmus Drab Hemispingus HEXAHemispingus trifasciatus Three-striped Hemispingus HETRThlypopsis ruficeps Rust-and-yellow Tanager THRFRamphocelus carbo Silver-beaked Tanager RACAButhraupis montana Hooded Mountain-Tanager BUMOChlorornis riefferii Grass-green Tanager CHRIAnisognathus igniventris Scarlet-bellied Mountain-Tanager ANIGAnisognathus somptuosus Blue-winged Mountain-Tanager ANSOThraupis sayaca Sayaca Tanager THSAThraupis palmarum Palm Tanager THPLThraupis cyanocephala Blue-capped Tanager THCYTangara cyanicollis Blue-necked Tanager TACYTangara vassorii Blue-and-black Tanager TAVATangara cyanotis Blue-browed Tanager TACNTangara xanthocephala Saffron-crowned Tanager TAXATersina viridis Swallow Tanager TEVIDacnis cayana Blue Dacnis DACACyanerpes caeruleus Purple Honeycreeper CYAEConirostrum cinereum Cinereous Conebill COCIConirostrum albifrons Capped Conebill COALConirostrum ferrugineiventre White-browed Conebill COFEDiglossa mystacalis Moustached Flowerpiercer DIMYDiglossa brunneiventris Black-throated Flowerpiercer DIBRDiglossa sittoides Rusty Flowerpiercer DISIDiglossa cyanea Masked Flowerpiercer DICYPhrygilus punensis Peruvian Sierra-Finch PHPUPhrygilus unicolor Plumbeous Sierra-Finch PHUNPhrygilus plebejus Ash-breasted Sierra-Finch PHPLDiuca speculifera White-winged Diuca-Finch DISPSicalis uropygialis Bright-rumped Yellow-Finch SIURSicalis olivascens Greenish Yellow-Finch SIOLCoryphospingus cucullatus Red-crested Finch COCUCoereba flaveola Bananaquit COFL

Emberizidae Zonotrichia capensis Rufous-collared Sparrow ZOCAArremon torquatus White-browed Brush-Finch ARTOAtlapetes rufinucha Bolivian Brush-Finch ATRUChlorospingus flavopectus Common Chlorospingus CHFL

Cardinalidae Piranga flava Hepatic Tanager PIFLParulidae Setophaga pitiayumi Tropical Parula PAPI

Myiothlypis luteoviridis Citrine Warbler BALUMyiothlypis signata Pale-legged Warbler BASIMyiothlypis bivittata Two-banded Warbler BABIMyiothlypis coronata Russet-crowned Warbler BACOBasileuterus tristriatus Three-striped Warbler BATRMyioborus miniatus Slate-throated Redstart MYMIMyioborus melanocephalus Spectacled Redstart MYML

Icteridae Psarocolius atrovirens Dusky-green Oropendola PSATPsarocolius decumanus Crested Oropendola PSDE

Fringillidae Sporagra xanthogastra Yellow-bellied Siskin CAXASporagra atrata Black Siskin CAATEuphonia laniirostris Thick-billed Euphonia EULAEuphonia cyanocephala Golden-rumped Euphonia EUCYEuphonia xanthogaster Orange-bellied Euphonia EUXAChlorophonia cyanea Blue-naped Chlorophonia CHCY

F.A. Monta~no-Centellas, �A. Garitano-Zavala / Acta Oecologica 65-66 (2015) 51e60 59

F.A. Monta~no-Centellas, �A. Garitano-Zavala / Acta Oecologica 65-66 (2015) 51e6060

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.actao.2015.05.003.

References

Acharya, B.K., Sanders, N.J., Vijayan, L., Chettri, B., 2011. Elevational gradients in birddiversity in the eastern Himalaya: an evaluation of distribution patterns andtheir underlying mechanisms. PLoS One 6, e29097.

Bach, K., Kessler, M., Gradstein, S., 2007. A simulation approach to determine sta-tistical significance of species turnover peaks in a species-rich tropical cloudforest. Divers. Distributions 13, 863e870.

Balderrama, J., 2009. Aves. Libro rojo de la fauna silvestre de vertebrados de Bolivia.Ministerio de Medio Ambiente y Agua, La Paz, Bolivia, pp. 305e409.

Bibby, C.J., Burgess, N.D., Hill, D.A., 1992. Bird Census Techniques. Academic Press,London.

Blake, J.G., Loiselle, B.A., 2000. Diversity of birds along an elevational gradient in theCordillera Central, Costa Rica. Auk 117, 663e686.

Blanchet, F.G., Legendre, P., Borcard, D., 2008. Forward selection of explanatoryvariables. Ecology 89, 2623e2632.

Blondel, J., Ferry, C., Frochot, B., 1981. Point counts with unlimited distance. Stud.Avian Biol. 6, 414e420.

Borcard, D., Gillet, F., Legendre, P., 2011. Numerical Ecology with R. Springer.Borges, S.H., Stouffer, P.C., 1999. Bird communities in two types of anthropogenic

successional vegetation in central Amazonia. Condor 529e536.Brooks, T.M., Mittermeier, R.A., Mittermeier, C.G., Da Fonseca, G.A., Rylands, A.B.,

Konstant, W.R., Flick, P., Pilgrim, J., Oldfield, S., Magin, G., 2002. Habitat loss andextinction in the hotspots of biodiversity. Conserv. Biol. 16, 909e923.

Chapman, K.A., Reich, P.B., 2007. Land use and habitat gradients determine birdcommunity diversity and abundance in suburban, rural and reserve landscapesof Minnesota, USA. Biol. Conserv. 135, 527e541.

Cleary, D.F., Genner, M.J., Boyle, T.J., Setyawati, T., Angraeti, C.D., Menken, S.B., 2005.Associations of bird species richness and community composition with localand landscape-scale environmental factors in Borneo. Landsc. Ecol. 20,989e1001.

Colorado, G.J., 2011. Ecology and Conservation of Neotropical-nearctic MigratoryBirds and Mixed-species Flocks in the Andes. The Ohio State University.

Colwell, R., 2009. Estimates: Statistical Estimation of Species Richness and SharedSpecies from Samples. Version 8.2. http://viceroy.eeb.uconn.edu/estimates.

Devictor, V., Julliard, R., Jiguet, F., 2008. Distribution of specialist and generalistspecies along spatial gradients of habitat disturbance and fragmentation. Oikos117, 507e514.

Diniz-Filho, J.A.F., Bini, L.M., Hawkins, B.A., 2003. Spatial autocorrelation and redherrings in geographical ecology. Glob. Ecol. Biogeogr. 12, 53e64.

Dray, S., Dufour, A.-B., 2007. The ade4 package: implementing the duality diagramfor ecologists. J. Stat. Softw. 22, 1e20.

Dray, S., Legendre, P., Peres-Neto, P.R., 2006. Spatial modelling: a comprehensiveframework for principal coordinate analysis of neighbour matrices (PCNM).Ecol. Model. 196, 483e493.

Forsman, J.T., Reunanen, P., Jokim€aki, J., M€onkk€onen, M., 2010. The effects of small-scale disturbance on forest birds: a meta-analysis. Can. J. For. Res. 40,1833e1842.

Gomes, L.G., Oostra, V., Nijman, V., Cleef, A.M., Kappelle, M., 2008. Tolerance offrugivorous birds to habitat disturbance in a tropical cloud forest. Biol. Conserv.141, 860e871.

Gower, J.C., 1966. Some distance properties of latent root and vector methods usedin multivariate analysis. Biometrika 53, 325e338.

Gray, M.A., Baldauf, S.L., Mayhew, P.J., Hill, J.K., 2007. The response of avian feedingguilds to tropical forest disturbance. Conserv. Biol. 21, 133e141.

Herzog, S.K., Kessler, M., Bach, K., 2005. The elevational gradient in Andean birdspecies richness at the local scale: a foothill peak and a high-elevation plateau.Ecography 28, 209e222.

Hewitt, J.E., Thrush, S.F., Halliday, J., Duffy, C., 2005. The importance of small-scalehabitat structure for maintaining beta diversity. Ecology 86, 1619e1626.

Jankowski, J.E., Ciecka, A.L., Meyer, N.Y., Rabenold, K.N., 2009. Beta diversity alongenvironmental gradients: implications of habitat specialization in tropicalmontane landscapes. J. Animal Ecol. 78, 315e327.

Jones, M.M., Szyska, B., Kessler, M., 2011. Microhabitat partitioning promotes plantdiversity in a tropical montane forest. Glob. Ecol. Biogeogr. 20, 558e569.

Karp, D.S., Ziv, G., Zook, J., Ehrlich, P.R., Daily, G.C., 2011. Resilience and stability inbird guilds across tropical countryside. Proc. Natl. Acad. Sci. 108, 21134e21139.

Lee, T.M., Soh, M.C., Sodhi, N., Koh, L.P., Lim, S.L.-H., 2005. Effects of habitatdisturbance on mixed species bird flocks in a tropical sub-montane rainforest.Biol. Conserv. 122, 193e204.

Lefevre, K.L., Rodd, F.H., 2009. How human disturbance of tropical rainforest caninfluence avian fruit removal. Oikos 118, 1405e1415.

Lefevre, K.L., Sharma, S., Rodd, F.H., 2012. Moderate human disturbance of rainforest alters composition of fruiting plant and bird communities. Biotropica 44,427e436.

Legendre, P., 1993. Spatial autocorrelation: trouble or new paradigm? Ecology 74,1659e1673.

Legendre, P., Borcard, D., Blanchet, F., Dray, S., 2012. PCNM: MEM Spatial Eigen-function and Principal Coordinate Analyses. R Package Version 2.1e2/r106.

Legendre, P., Gallagher, E.D., 2001. Ecologically meaningful transformations forordination of species data. Oecologia 129, 271e280.

Legendre, P., Legendre, L., 2012. Numerical Ecology. Elsevier.Lumpkin, H.A., Boyle, W.A., 2009. Effects of forest age on fruit composition and

removal in tropical bird-dispersed understorey trees. J. Trop. Ecol. 25, 515e522.MacArthur, R.H., Wilson, E.O., 1967. The Theory of Island Biogeography. Princeton

University Press.McCain, C.M., 2009. Global analysis of bird elevational diversity. Glob. Ecol. Bio-

geogr. 18, 346e360.McCune, B., Grace, J.B., Urban, D.L., 2002. Analysis of Ecological Communities. MjM

Software Design Gleneden Beach. Oregon.Molina-Carpio, J., 2005. R�egimen de precipitaci�on en la cuenca de Huarinilla-

Cotapata, La Paz-Bolivia. Ecol. Boliv. 40, 43e55.Myers, N., Mittermeier, R.A., Mittermeier, C.G., Da Fonseca, G.A., Kent, J., 2000.

Biodiversity hotspots for conservation priorities. Nature 403, 853e858.Nogues-Bravo, D., Araujo, M.B., Romdal, T., Rahbek, C., 2008. Scale effects and hu-

man impact on the elevational species richness gradients. Nature 453, 216e219.Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, P.R., O'Hara, R.B.,

Simpson, G.L., Solymos, P., Stevens, M.H.H., Wagner, H., 2013. Vegan: Commu-nity Ecology Package. Version 2.0e10. http://CRAN.R-project.org/package¼vegan.

Paradis, E., Bolker, B., Claude, J., Cuong, H.S., Desper, R., Durand, B., Dutheil, J.,Gascuel, O., Jobb, G., Heibl, C., 2008. The ape package. Anal. Phylogenetics Evol.9e12.

Petit, L.J., Petit, D.R., Christian, D.G., Powell, H.D., 1999. Bird communities of naturaland modified habitats in panama. Ecography 22, 292e304.

R Development Core Team, 2009. R Version 2.9. 2. R Project for StatisticalComputing Vienna, Austria.

Remsen Jr., J.V., Cadena, C.D., Jaramillo, A., Nores, M., Pacheco, J.F., P�erez-Em�an, J.,Robbins, M.B., Stiles, F.G., Stotz, D.F., Zimmer, K.J., 2013. A Classification of theBird Species of South America [Version 31 October]. American Ornithologists'Union.

Union.Rensburg, B.J.v., Chown, S.L., Gaston, K.J., 2002. Species richness, environ-mental correlates, and spatial scale: a test using South African birds. Am. Nat.159, 566e577.

Ribera, M.O., 1995. Aspectos ecol�ogicos, del uso de la tierra y conservaci�on en elparque nacional y �area natural de manejo integrado cotapata. In: Morales, C.B.(Ed.), Caminos de Cotapata. Instituto de Ecología, FUNDECO, FONAMA, La Paz,Bolivia, pp. 1e84.

Rodewald, A.D., Yahner, R.H., 2001. Influence of landscape composition on aviancommunity structure and associated mechanisms. Ecology 82, 3493e3504.

Sevilla Callejo, M., 2010. Organizaci�on territorial y campesinado en el ParqueNacional y �area Natural de Manejo Integrado Cotapata (Bolivia). UniversidadAut�onoma de Madrid.

Soh, M.C., Sodhi, N.S., Lim, S.L., 2006. High sensitivity of montane bird communitiesto habitat disturbance in peninsular Malaysia. Biol. Conserv. 129, 149e166.

Stotz, D.F., Parker, T.A., Fitzpatrick, J.W., 1996. Neotropical Birds: Ecology and Con-servation. Cambridge Univ Press.

Swenson, J.J., Young, B.E., Beck, S., Comer, P., C�ordova, J.H., Dyson, J., Embert, D.,Encarnaci�on, F., Ferreira, W., Franke, I., 2012. Plant and animal endemism in theeastern Andean slope: challenges to conservation. BMC Ecol. 12, 1.

ter Braak, C.J.F., 1986. Canonical correspondence analysis: a new eigenvector tech-nique for multivariate direct gradient analysis. Ecology 67, 1167e1179.

Terborgh, J., 1971. Distribution on environmental gradients: theory and a pre-liminary interpretation of distributional patterns in the avifauna of theCordillera Vilcabamba, Peru. Ecology 23e40.

Terborgh, J., 1977. Bird species diversity on an Andean elevational gradient. Ecology1007e1019.

Terborgh, J., Weske, J.S., 1975. The role of competition in the distribution of Andeanbirds. Ecology 562e576.

Tews, J., Brose, U., Grimm, V., Tielb€orger, K., Wichmann, M.C., Schwager, M.,Jeltsch, F., 2004. Animal species diversity driven by habitat heterogeneity/di-versity: the importance of keystone structures. J. Biogeogr. 31, 79e92.

Verhulst, J., B�aldi, A., Kleijn, D., 2004. Relationship between land-use intensity andspecies richness and abundance of birds in Hungary. Agri. Ecosyst. Environ. 104,465e473.

Walters, J.R., 1998. The ecological basis of avian sensitivity to habitat fragmentation.In: Marzluff, J.M., Sallabanks, X. (Eds.), Avian Conservation: Research andManagement. Island Press, Washington, DC, pp. 181e192.

Wilson, J.B., 1991. Methods for fitting dominance/diversity curves. J. Veg. Sci. 2,35e46.

Woltmann, S., 2003. Bird community responses to disturbance in a forestryconcession in lowland Bolivia. Biodivers. Conserv. 12, 1921e1936.

Wu, Y., Colwell, R.K., Rahbek, C., Zhang, C., Quan, Q., Wang, C., Lei, F., 2013.Explaining the species richness of birds along a subtropical elevational gradientin the Hengduan Mountains. J. Biogeogr. 40, 2310e2323.