The Condor 99:719-727 0 The Cooper Ornithological Society 1997

BIRD-HABITAT RELATIONSHIPS IN A VEGETATIONAL GRADIENT IN THE ANDES OF CENTRAL CHILE1

CRISTIAN E ESTADES~ Departamentode Manejo de RecursosForestales, Universidadde Chile, Casilla 9206, Santiago, Chile

Abstract. An unusualpattern has been describedin Patagonianforests, where spe- cies diversity seems to be inversely correlated with vegetation complexity. To test the oc- currence of this pattern in the northern Nothofagusforests of central Chile, a study was conducted at the Ruble National Reserve during the austral summers of 1994 and 1995. Twenty-nine plots were set along a vegetational gradient. In each plot, the density of all diurnal bird specieswas estimated, and the vegetationcomposition and structurewere mea- sured. The highest and the lowest bird speciesdiversity were found in semideciduousscrub habitat and Poaceae steppes,respectively. In nongrasslandsites, no vegetation variable ex- plained bird speciesrichness or diversity. Total foliage volume and foliage height diversity were significantly correlatedwith bird speciesdiversity in the nonforest sites but not in the forested areas.However, foliage height diversity significantly explained the variation of bird density and bird richness in the forest sites. The presence of the tree obliqua was positively correlated with bird diversity and insectivore density. The patterns observed at Nuble lie between the traditional positive relationship between bird diversity and foliage diversity and the inverse relationshipfound in Patagonia.I proposethat particular floristic features of the vegetation rather than structureitself might explain better the diver- sity of bird speciesin Patagonianforests. Key words: bird communities,Chile, Patagonianforests, diversity,vegetation structure, Nothofagus.

INTRODUCTION METHODS Temperate forest have been poorly studied The study was conducted at the Ruble National in Chile, and few investigations have described Reserve in central Chile (Fig. 1). This area in- the composition and structure of entire bird cludes two ecological subregions: “Bosque Cad- communities (Jaksic and Feinsinger 1991). ucifolio Montano” and “Cordilleras de la Ar- Moreover, all these studies have been conducted aucanfa” (Gajardo 1994). Most of the 720 km* in evergreen or semideciduous lowland forests, Reserve is covered with upland steppesof Fes- leaving the bird assemblages of the deciduous tuca and , deciduous scrub of Notho- beech forests in the Andes cordillera of central fagus antarctica or N. pumilio, and deciduous and southern Chile virtually unknown. Vuilleu- forests of N, obliqua and N. pumilio (Faiindez mier (1972) described an unexpected pattern in et al. 1994). Mean annual precipitation within bird communities in northern Patagonia in Ar- the Reserve ranges from 2.3 to 3.3 m; mean gentina. He found fewer speciesin beech forests minimum temperature in July ranges from than in the structurally simpler scrub-steppehab- -1.2”C to 0.9”C; and mean maximum tempera- itats. In the same area, Ralph (1985) found a ture in January ranges from 14.8”C to 22.1”C significant inverse relationship between foliage (Santibtiez and Uribe 1993). complexity and bird diversity in nongrassland Bird censuses were carried out during Feb- sites. ruary 1994 and January 1995, corresponding to The aim of the present study is to describe the late breeding season. Twenty-nine census habitat associations of bird communities in a sites were established along a vegetational gra- forest-scrub-steppegradient in the Andes of cen- dient from steppe to forest. When possible (20 tral Chile and to determine whether they exhibit sites), variable line transects (Bumham et al. the patterns found by Ralph (1985). 1980) were used to estimate bird density. At sites where walking was difficult (nine sites) variable circular plots were used (Buckland 1Received 5 September 1996. Accepted 22 April 1987). I assumed that these two methods pro- 1997. 2 Current address:Department of Wildlife Ecology, duced similar estimatesof a population’s density University of Wisconsin, Madison, WI 53706. (Ralph 1985). Observations were carried out be-

17191 720 CRISTIAN F. ESTADES

+ censusplots

National

5 10 km

FIGURE? 1. Geographic location of the study area, Ruble National Reserve, in central Chile.

tween Ok00 and 12:O0. Detections were record- standardized as the number of specieson a 2-ha ed at intervals of 10 m, with a maximum detec- area. tion distance of 80 m. Flying birds were not re- At every site, foliage structure and spe- corded unless they showed clear evidence of us- cies composition were described in four hori- ing the plot (e.g., swallows hunting insects). zontal layers (o-0.2, 0.21-2, 2.1-8, and 8.1-16 Censuses were conducted by two observers, m). In each layer, total foliage volume (TPV,,,,) each contributing roughly 50% of the observa- was estimated as the proportion of the layer oc- tions within each plot. Transects or points were cupied by plant leaves, twigs, and branches.This surveyed 4-10 times in 1994 and 6 times in proportion was multiplied by the layer’s height 1995. Data for each bird species at each census to obtain its TFV,,,, (m3 per 0.01 ha). The rel- site were pooled to estimate population density. ative species frequency at each layer was mul- Densities from 1994 and 1995 were averaged tiplied by the TFV,,,, to obtain foliage volume and their variances were calculated. for individual species. The site’s total foliage Bird species diversity (BSD) at each site was volume (TFV) and individual species’ foliage calculated using the Shannon-Wiener (H’) di- volume were calculated by the addition of the versity index (MacArthur and MacArthur 1961). four TFV,aye,values. Vegetation measures were Bird speciesrichness (BSR) at each site was es- estimated in both 1994 and 1995 and then av- timated using the jackknife procedure (Heltshe eraged. Plant species diversity (PSD) was esti- and Forrester 1983). To eliminate the effect of mated for each layer and for the entire site using sample size, richness estimates were corrected the proportion of the TFVlaye,and the proportion using a logarithmic transformation and were of the TFV represented by all the species, re- BIRD-HABITAT RELATIONSHIPS IN CENTRAL CHILE 721 spectively. With these values, an H’ index of di- vation (1,200-1,400 m) deciduous scrub. In versity was calculated. In order to determine an open valley bottoms. Sparse scrub with dense index of foliage height diversity (FHD), the grass (mainly Festuca) cover. Scrub 2-5 m tall three upper layers were divided into two equal (five sites). sub-layers whose TFV,,,, were calculated. This 8. Chusquea argentina steppe (CaS). Mid-to subdivision was done because the four original high-elevation (1,300-1,900 m) steppe. Occurs layers were considered inappropriate to describe in many topographic situations. Composed the foliage profile precisely. Finally, the TFV,,,,j mainly of C. argentina and Festuca scabrius- TFV ratios for the seven resulting layers were culu. Lower than 0.5 m tall (three sites). used to estimate the H’ index. 9. Poaceae steppe(PS). Mid to high-elevation As a way of identifying different kinds of (1,300-2,000 m) steppe. Located in windy habitats for birds, the census sites were grouped zones. Main species are grasses of the genus initially in terms of their vegetation structureand Agrostis, Festuca, Poa, and Hordeurn. In some composition using hierarchical clustering (Digby zones there is a sparse cover of mainly and Kempton 1987). However, to make the re- Chiliotrichum rosmarinifolia (two sites). sulting classification more consistent with field The use of these habitats by bird specieswas observations of bird assemblages,two sites were described in terms of their mean absolute den- arbitrarily reassigned to another type of habitat. sities. A measure of habitat breadth (B) was cal- Finally, nine vegetation types (bird habitats) culated for all speciesusing the relation B(x) = were recognized at the studied area. These are: eZpiLOg@used‘), by MacArthur (1964), where pi is 1. Nothofugus pumilio forest (NpF). High-el- the density of speciesx in habitat i divided by evation (1,300-1,700 m) deciduousbeech forest, the sum of densities for the species at all the on wind-protected slopes. Trees lo-14 m tall. habitats studied. Birds were assigned to trophic Sparse understory (two sites). guilds following Johnson (1967) and Jaksic and 2. N. obliqua forest (NoF). Low-elevation Feinsinger (1991) and using the author’s field ( 1,000-l ,200 m) deciduousbeech forest, located experience. Bird species names follow Araya in protected valleys and slopes. Trees 8-12 m and Chester (1993), and plant names follow tall. Sparse to dense understory (six sites). Marticorena and Quezada (1985). 3. N. obliqua-pumilio forest (NopF). Mid-el- Linear regression analysis and Pearson cor- evation (1,200-l ,400 m) highly stratified decid- relations were performed in order to relate birds uous beech forest. Located mainly on south-fac- with vegetation variables and individual plant ing slopes in protected valleys. Dense understo- species. For this last analysis, only species ry of Chusquea argentina. Trees 14-16 m tall ( and birds) recorded at five or more sites (five sites). were considered. 4. Austrocedrus chilensis-N. obliqua forest RESULTS (AcNoF). Low-elevation (l,OOO-1,200 m) mixed cedar-beech forest. Occurs on rocky BIRDS slopes. Very sparse overstory and understory. Forty-six bird species were recorded during the Trees 8-12 m tall (one site). transect and point surveys (Table 1). The most 5. Semideciduous scrub (SS). Low-elevation abundant specieswere the White-crested Elaenia (l,OOO-1,200 m) scrub. Located on dry slopes. (1.37 ha-l), Black-chinned Siskin (1.30 ha-‘), Composed mainly of Colletia hystrix, C. argen- Thorn-tailed Rayadito (0.86 ha-‘), Common tina, Diostea juncea, and imbricata, Diuca-Finch (0.78 ha-i), and House Wren (0.77 with a few scatteredbeech trees. Scrub 0.5-4 m ha-i). Another 22 species were recorded on the tall (three sites). sites but not on official counts (Table 1). The 6. N. antarctica low scrub (NalS). High-ele- latter species were very uncommon in the re- vation (1,600-2,000 m) deciduous scrub. Locat- serve and were normally observed only once or ed in very windy and cold zones. On some sites twice during the entire study. N. antarctica is replaced by N. pumilio but this Bird communities showed important differ- does not alter the overall structureof the habitat. ences among habitats (Table 1). Estimated mean Scrub 0.5-1.5 m tall (two sites). density of all birds ranged from 3.36 ha-’ in Po- 7. N. antarctica tall scrub (NatS). Mid-ele- aceae steppeto 20.64 ha-i in N. obliqua-pumilio

724 CRISTIAN F. ESTADES

. l .** ,.. 0 ..d 0 ,/ ’

. q.0.” / /.” ,.,c”

.,.. / .I., 1 ’ l o - l o 0 . .. ’ / ’ .* 0 . B D/0 r* = 0.49***: 0 0.50 **: . 0.31 *

/ 0 *-0 0 0 I+ = 0.08; 0 0.36 *; 0 0.01 rZ = 0.23 **; 0 0.23; . 0.31 *

0 O F 0 & 2.6 O M” 0 . l . . . "...... *...... ~...... ~ . l e .

0 -j 1.2 * - 0.01; 0 0.50 **; l 0.07 0 r* - 0.16 *; 0 0.34 *; 0 0.14 ual,O, " " 0 100 200 300 400 500 600 700 800 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Total foliagevolume (TFV) (m30.01. ’ ha’ ) Foliageheight diversity (FHD) (H’) FIGURE 2. Relationshipsbetween vegetation variables and bird community variables. The open circles rep- resent the nonforest sites and the black circles, the forest sites. Solid lines representthe least squaresregression line for the entire data set; dotted lines represent the least squaresregression line for the nonforestand forest sitesanalyzed separately. The asterisksindicate the statisticalsignificance of correlations(*P < 0.05; **P < 0.01; ***p < 0.001). forests (Table 1). Highest BSD was found in the Fig. 2E and 2F). No significant linear association semideciduousscrub with a mean value of 2.41, was found between PSD and any of the bird whereas the lowest BSD was found at the Po- community parameters. Removing the grassland aceae steppe, with only 1.49. sites from the model (following Ralph 1985) re- Species with the greatest habitat breadth (B) duced the amount of variance explained by FHD were the House Wren (B = 7.51) and the Black- and TFV for all bird variables. In this case, only chinned Siskin (B = 7.31), recorded at all hab- the density of all birds was significantly related itats, followed by the Rufous-collared Sparrow (P < 0.01) with vegetation variables. (B = 6.50) and the White-crested Elaenia (B = Correlations changed substantially when non- 6.47). forest and forest sites were analyzed separately (Fig. 2, dotted lines). At nonforest sites, all bird VEGETATION-BIRD COMMUNITY community variables were significantly correlat- RELATIONSHIPS ed (P < 0.05-0.01) with TFV (Fig. 2A-2C). Variance in density of all birds per site was high- FHD significantly explained the variance in den- ly correlated (P < 0.001) with both TFV and sity of all birds (P < 0.01) and BSD (P < 0.05; FHD (Fig. 2A and 2D, solid lines). Variation of Fig. 2D and 2F), but failed to explain the vari- BSR and BSD was only explained significantly ation of BSR (Fig. 2E). At forest sites, no bird by FHD (P < 0.01 and P < 0.05, respectively; community variable was correlated with TFV BIRD-HABITAT RELATIONSHIPS IN CENTRAL CHILE 725

(Fig. 2A-2C). FHD was positively correlated ranged from 0.59 to 0.71; P < 0.001) with both with density of all birds (P < 0.05; Fig. 2D) and primary and all foliage-gleaners. BSR (P < 0.05; Fig. 2E) but did not explain Foliage volume of the tree N. obliqua was BSD’s variation (Fig. 2F). Both at nonforest and positively correlated (P < 0.001) with the den- forest sites, PSD showed no significant correla- sity of all frugivores, primary and all insecti- tion with bird community variables. vores, primary and all foliage-gleaners, and all air-hunters (r ranged from 0.58 to 0.70). Signif- VEGETATION-BIRD SPECIES RELATIONSHIPS icant negative correlations (P < 0.05) were Over the entire vegetation gradient, several sig- found between Baccharis magellanica and all nificant correlations were found between vege- granivores, all herbivores, and all foliage-glean- tation variables and individual bird species.Data ers (r ranged from -0.38 to -0.43). concerning these and the following analyses are Correlations of bird guilds and vegetation available from the author on request. TFV and variables within non-forest sites differed notably FHD were significantly positively correlated from the correlations in forest sites. Few guilds were positively related with FHD in forest sites, with densities of typical forest bird species (r but many were at the nonforest sites. ranged from 0.43 to 0.68), including the Thorn-tailed Rayadito, the Magellanic Wood- DISCUSSION pecker, the Patagonian Tyrant, the White-crested Even though FHD and TFV were significantly Elaenia, the White-throated Treerunner, and the correlated (Pearson r = 0.68; P < 0.001) among Chucao Tapaculo. Densities of grassland and the 29 sites assessed,they differed in relation to scrub species,such as the Common Diuca-Finch some bird variables (Fig. 2), indicating that these and the Long-tailed Meadowlark, were signifi- two variables might effectively describe differ- cantly negatively correlated with these vegeta- ent features of the habitat for certain bird spe- tion parameters (r ranged from -0.39 to -0.51, cies. For example, species like the Des Mur’s P < 0.05 and 0.01, respectively). Wiretail, the Andean , and the Chest- Plant species such as Chusquea argentina, nut-throated Huet-huet, that were influenced by Hypericum perforatum, Maytenus disticha, TFV but not by FHD, forage within the densest Nothofagus obliqua, and Ribes punctatum ex- foliage of the understory. On the other hand, plained an important part of the variation of for- speciesinfluenced strongly by FHD, such as the est-bird species densities. On the other hand, Green-backed Firecrown or the Fire-eyed Diu- plant speciessuch as Acaena pinnati$da, Mulin- con, usually hunt for insects flying between urn spinosum, and N. antarctica were mainly trees. Similarly, the Austral Thrush looks for correlated with scrub and openland bird species. berries in the openings of forests &though it also The only two significant negative correlations forages in prairies (Table 1). Assemblages of between plant and bird species were found be- species with such different traits could be re- tween Chusquea argentina and the Rufous-col- sponsible for the different patterns observed be- lared Sparrow (r = -0.40), and between N. ob- tween the bird communities in forested and non- liqua and the Common Diuca-Finch (r = forested sites. -0.50). Results indicate that the relationship of the Many correlations between birds and vegeta- studied bird communities with the vegetation tion variables changed when nonforest and for- variables changes with the scale of analysis. est sites were analyzed separately,producing, on Considering the whole vegetation gradient, bird average, more bird-plant species relationships assemblagesbehaved as predicted by MacArthur and fewer bird-vegetation relationships. and MacArthur (1961), increasing their richness and diversity as FHD increased. However, when VEGETATION-BIRD GUILDS RELATIONSHIPS grassland sites were deleted from the analysis, TFV and FHD showed similar overall relation- no such relationships were found. FHD in non- ships with the different bird trophic guilds. Den- forest sites was positively related with BSD, but sities of primary insectivores, all insectivores, when analyzing the forest plots alone, no signif- and all frugivores increased significantly as TFV icant linear relationship was found between and FHD increased (P < 0.001). The latter vari- these two variables. These results agree with the ables also were highly positively correlated (r patterns observed by Willson (1974). 726 CRISTIAN F. ESTADES

Whereas along the whole vegetation gradient use all the habitat complexity available to them, most bird species tended to be related to struc- and Ralph (1985) suggestedthat this effect was tural variables (FHD and/or TFV), within small- caused by the reduced area and isolation of for- er subgroups of sites some bird species were ests which diminished the number of potential more associated with particular plant species. bird species present. But in the present study, This pattern has been described before by Ro- the birds did seem to be making some use of the tenberry (1985) in grasslandsand by Bersier and habitat complexity (the estimate of speciesrich- Meyer (1994) in forests, and is probably a mat- ness did increase with forest FHD). This local ter of larger scales masking relationships that pattern is probably due to the relatively higher exist between birds and individual plant species. insect diversity found in N. obliqua forests Thus, a certain bird species may forage on the (Welch 1988) than in other Patagonian and Tas- insects that live on a certain plant in a shrubland, manian Nothofagus forests, where the depauper- but in a forest will probably have to select other ate insect fauna has been suggestedto be a fac- species that can fulfill its requirements. In both tor in determining the low number of insectiv- habitats the bird’s presence would be associated orous birds found in these ecosystems (Mc- with one plant species, but such a relationship Quillan 1993). Thus, even though biogeographic might not arise along the entire vegetation gra- isolation could be ultimately limiting the poten- dient. tial number of bird speciespresent in Patagonian Nonetheless, considering the natural history forests, the proximate constraint on most of of the studied bird species, many significant as- these forests could be the diversity and abun- sociations found between them and individual dance of insect prey available to birds, which plant species lack a known direct cause-effect might be more strongly associatedwith floristic relationship. However, the relative importance of rather than structural features of the vegetation. N. obliqua forest for insectivorous birds agrees Testing this hypothesis remains one of the sev- with the observations of Welch (1988), who eral tasks that are needed to understand the na- concluded that this tree supports higher insect ture of the forest bird communities of Patagonia. diversity than any other Nothofagus in the area. The patterns described in the fluble’s bird ACKNOWLEDGMENTS communities seem to lie between the traditional This study was partially funded by the Chilean Forest belief that as foliage diversity increases so does Service (Conaf) and the Chile’s National Petroleum Company (ENAP). J. F? Gabella helped with bird bird community diversity, and the inverse rela- counts and M. M. S&chez helped with vegetation de- tionship of these two parameters found by Vuil- scription. L. Falindez kindly identified some plant spe- leumier (1972) and Ralph (1985) in Patagonian cies. H. J. HemBndez provided the cartographicinfor- forests. mation. D. R. Curson, L. X. Payne, S. A. Temple, J. Because of their regional isolation, Patago- V. Remsen, W. D. Koenig, and an anonymousreviewer made very helpful comments on different versions of nian forest-bird communities show some char- the manuscript. acteristicsof an island biota, such as a relatively low diversity, and species with wide habitat- LITERATURE CITED niches (Vuilleumier 1985). On the other hand, ARAYA,B., AND S. CHESTER.1993. The birds of Chile: the steppe and scrubland ecosystems are wide- a field guide. Latour, Santiago, Chile. spread along the Andes connecting the bird BERSIER,L. E, AND D. R. MEYER. 1994. Bird assem- blages in mosaic forests: the relative importance communities with others in South America, thus of vegetation structure and floristic composition reducing isolation. However, within the entire along the successionalgradient. Acta Oecologica range of Patagonian forests, tree species com- 15561-576. position varies in important ways. The commu- BUCKLAND,S. T. 1987. On the variable circular plot nities describedin Ruble are at the northern lim- method of estimating density. Biometrics 43:363-384. it of Patagonian forests, and therefore they might BLJRNHAM,K. F!, D. R. ANDERSON,AND J. L. LAAKE. show some extreme characteristics.The forests 1980. Estimation of density from line transect studied by Vuilleumier (1972) and Ralph (1985) sampling of biological populations. Wildl. Mon- were well to the south and lacked N. obliqua, a ogr. 72: l-202. DIGBY, P G. N., AND R. A. KEMPTON.1987. Multi- common tree at Ruble. variate analysis of ecological communities.Chap- Rosenzweig (1995) statedthat the birds in the man and Hall, London. Patagonian Nothofagus forests did not seem to FA~NDEZ, L., M. T. SERRA, AND H. J. HERNANDEZ. BIRD-HABITAT RELATIONSHIPS IN CENTRAL CHILE 727

1994. Flora y vegetaci6n de la Reserva National MCQUILLAN,P R. 1993. Nothofagus (Fagaceae) and Nuble. Caracterizaci6n y Monitoreo. Project Re- its invertebrate fauna-an overview and prelimi- port. Universidad de Chile, Departamentode Ma- nary synthesis.Biol. J. Linn. Sot. 49:317-354. nejo de RecursosForestales and Corporaci6n Na- RALPH,C. J. 1985. Habitat associationpatterns of for- cional Forestal, Santiago, Chile. est and steppebirds of northern Patagonia,Argen- GAJARDO,R. 1994. La vegetaci6n natural de Chile. tina. Condor 87:471-483. Clasificacibn y distribucidn geogrifica. Editorial ROSENZWEIG,M. L. 1995. Species diversity in space Universitaria, Santiago, Chile. and time. Cambridge Univ. Press, Cambridge. HELTSHE,J. E, AND N. E. FORRESTER.1983. Estimating ROTENBERRY,J. T. 1985. The role of habitat in avian species richness using the jackknife procedure. community composition: physiognomy or florist- Biometrics 39: l-l 1. its? Oecologia 67:213-217. JAKSIC,E M., AND l? FEINSINGER.1991. Bird assem- SANTIBANEZ,F., AND J. M. URIBE. 1993. Atlas agro- blages in temperate forests of North and South climkico de Chile. Regiones Sexta, SCptima,Oc- America: a comparison of diversity, dynamics, tava y Novena. Universidad de Chile-Ministerio guild structure,and resourceuse. Revista Chilena de Agricultura-Corporaci6n de Foment0 de la Produccibn, Santiago, Chile. de Historia Natural 64:491-510. VUILLEUMIER,E 1972. Bird species diversity in Pata- JOHNSON,A. W. 1967. The birds of Chile and adjacent gonia (temperate South America). Am. Nat. 106: regions of Argentina, Bolivia, and Peni. Platt Es- 266-271. tablecimientosGrBficos, Buenos Aires. VUILLEUMIER,E 1985. Forest birds of Patagonia:eco- MACARTHUR,R. H. 1964. Environmental factors af- logical, geography, speciation, endemism, and fecting bird species diversity. Am. Nat. 98:387- fauna1history. Ornithol. Monogr. 36:255-304. 397. WELCH,R. C. 1988. Phytophagousinsects on decid- MACARTHUR,R. H., AND J. W. MACARTHUR.1961. On uous Nothofagus in Chile and Argentina. Mon. bird speciesdiversity. Ecology 42:594-598. Acad. Cs. Exactas,Fisicas y Naturales(Argentina) MARTICORENA,C., AND M. QUEZADA.1985. Catalog0 4:107-114. de la flora vascularde Chile. Gayana Botanica 42: WILLSON,M. E 1974. Avian community organization l-155. and habitat structure.Ecology 55: 1017-1029.