The Condor 87:471483 0 The Cooper Ornithological Society 1985
HABITAT ASSOCIATION PATTERNS OF FOREST AND STEPPE BIRDS OF NORTHERN PATAGONIA, ARGENTINA
C. JOHN RALPH
ABSTRACT. -1 censusedbirds acrossa moisture gradient in northern Patagonia, Argentina, in the vicinity of Bariloche. Over a 60-km distance, the 12 sitesranged from grasslandat lower elevations to upland climax Nothofagusforests of the eastern Andes. Here, I correlated bird abundance and diversities with various vegetation measures.Using all sites, bird diversities and abundanceswere posi- tively correlated with various foliage measures.When grasslandswere excluded, however, an inverse relationship was found: birds were more diverse and abundant in the lower stature shrub communities than in complex forests. Multiple regres- sion analysesof this apparently paradoxical situation indicated that certain species of plants probably had important effects on community structure.
As a habitat in a region becomes more com- nearby, dense beech forests. Vuilleumier also plex, the bird community in that habitat usu- thought that bird speciesdiversity was higher ally becomes more complex as well (e.g., in the scrub-steppe habitat. A re-calculation MacArthur and MacArthur 196 1, Willson (Church 1974) of his data, however, showed 1974). In southern Argentina, however, Vuil- that Vuilleumier’s diversity values (H’) in the leumier (1972) concluded that the southern two habitats were essentially identical (dense beech (Nothofagus)forests had a less complex forest- 1.993; scrub-steppe forest- 1.837). avifauna than nearby, simpler, scrub areas. In One other study has suggestedthat bird species order to describe the avifauna, its habitat as- diversity was unrelated to vegetation com- sociations,and to test this seemingly paradox- plexity (Howell 197 1). This study, although ical situation, I sampledthe avifauna and vege- intensive, had samples drawn from very few tation associationsof the region in the spring study sites. of 1980. In this paper, I describethe vegetation The avifauna of Patagonia is relatively little types, the bird species associated with each described,at least in terms of population num- type, and their interrelationships. bers. My study is the first detailed quantifi- After the pioneering work of MacArthur and cation of these birds and their habitat associ- MacArthur (196 l), many studies have con- ations. Such data are important becausemost firmed their initial finding that, for terrestrial ecological generalizations are based on north- communities, the number of bird species,as ern hemisphere temperate communities. well as their diversity, are strongly positively Comparative data from other areas can pro- correlated with aspectsof the structural com- vide information on the applicability of these plexity of vegetation (e.g., MacArthur 1964, generalizations. Recher 1969, Karr and Roth 197 1, Pearson and Ralph 1978). That is, the more complex METHODS the structure or composition of the vegetation, the more likely it is that the habitat will contain STUDY SITES more bird species (higher richness) or more In the region of my study (eastward from the even abundancesof the species(evenness). This crest of the Andes in southern Argentina), an- relationship is usually interpreted as evidence nual rainfall diminished from more than 2,000 for the underlying dependence of birds upon mm to 200 mm over about a 50-km area (De their habitat, and thus, lends some degree of Fina 1972). The vegetative growth pattern re- predictability. flected this moisture gradient. Although there Exceptions to this pattern should be of in- was some effect of elevation and temperature terest becauseof what they may reveal about on the vegetation, De Fina considered it com- the pattern itself or the systemsit attempts to paratively small. Owing to a varied topogra- summarize. Thus, it is of interest that Vuil- phy, the habitat types were somewhat patchy leumier (1972) regarded the northern Pata- in this region, with each “patch” occupying gonian forest bird communities of southern severalhundred or more hectares.Habitats be- Argentina as such an exception. Although his came uniform only where the more or lesscon- data were few, they indicated that more species tinuous grasslandsteppe was reached at about of birds inhabited the structurally simpler 200 mm annual rainfall. scrub-steppehabitats than the more complex, In the region of this moisture gradient, I
14711 472 C. JOHN RALPH
I -_ \ BARILOCHE \ E ;
,i CREST OF (-ANDES \ \ )
I o,,,y,,,;o‘ \ \ Km \
FIGURE 1. Map of study locations in northern Patagonia,Argentina. The letters refer to the locations described in Methods.
selected 12 sites (Fig. 1) at which to census and lower in the altitudinal gradients. The lo- birds and measure vegetation. I made no at- cation of the sites on Figures 1 to 4 are indi- tempt to selectany certain habitat type; rather, cated by the letter following their namesbelow. at any given site, I used the largest area of Grasslandsites. Bunchgrass Grassland (A) - reasonably contiguous, similar vegetation 42 km north-northeast of Bariloche, on open, (summarized in Table 2), with a minimum size windswept hills, 2 km south of Arroyo Corral. of approximately 25 ha. Censusstations at each The vegetation was less than 0.5 m tall, and site were at least 100 m from what I judged to consistedof primarily bunchgrass(Festuca sp.) be the edge of another habitat type. The only and a small prickly shrub, Mulinum spinosum. exception to theselast two rules was the Sedge- I estimate that annual rainfall was probably scrub Site (see below), which was only about 300-500 mm. 5 ha, and several of its stationswere within 25 Cactus Grassland (B)-a mixed grass-scrub m of the edge. The vegetation data that I gath- area within 1 km south of Laguna Blanca, an ered, basedon an index to foliage density, were alkaline lake 20 km southeastof Zapala, Neu- subjectedto a cluster analysis procedure (SAS q&n. The vegetation was mostly lessthan 1.5 1979), which confirmed the vegetation (struc- m tall, and consistedof primarily bunchgrass, tural) similarity of the stations at each site. a small cactus,and two small shrubs(less than That is, the great majority of the censussta- 0.5 m tall), one being Acaena ovulifoliu.Rain- tions at a site were more similar to each other fall was probably 200-400 mm annually. than to stations at other sites. The rainfall at Herbaceous Grassland (C)-73 km north of the twelve siteswas extrapolated from the data Bariloche and 4 km east-northeastof Caleufti, in De Fina (1972) from measurementshigher NeuquCn, 0.5 km to the south of Rio Caleufu. PATAGONIAN BIRD HABITATS 413
It resembled the Cactus Grassland Site, with massof bamboo. Rainfall was probably 1,500- similar rainfall, but had less bunchgrassand 2,000 mm annually. more Mulinum and herbs. Beech (antarctica-dombeyi-bamboo)For- Scrub-cedarsites. Scrub-sedge (D)-an oa- est (IS)-at Ruca MalCn, near the northwest sis of shrubs in a swale surrounded by grass- end of Lago Correntos, and 19 km north- land, on low hills near the BunchgrassGrass- northwest of Villa la Angostura. It was a mixed land Site at Arroyo Corral and with similar forest, lo-30 m high, and consisted of pri- rainfall. The area of this patch was about 5 ha. marily Nothofagusdombeyi and some N. ant- The vegetationwas a mixture of grassland,with arctica,with a Berberisand Chusqueaunder- barberry (Berberissp.) l-2 m high and a small story. Except for the addition of N. antarctica, (2-3 m high) tree, probably Maytenusboaria. the forest resembled that of the previous site, Scrub-beech (E)-4 km east-southeast of with rainfall probably 1,500-2,000 mm. Bariloche, and 0.5 to 1.5 km south of the Beech (pumilio-dombeyi-bamboo) Forest southern base of Cerro Otto. The vegetation (L)-a beech forest 8 km east-southeastof the consistedof several shrub species,mostly 2-3 summit of Mt. Tronador, within 1 km of Rio m in height. Nothofagusantarcticus was the Castafio Overo. It was similar to the Llao Llao most common shrub, occasionally reaching 5 Peninsula site, but had an almost equal mix- m in height. Diosteajuncea, Berberis spp., and ture of Nothofagusdombeyi and N. pumilio. Lomatia hirsutawere less common. This was Rainfall was probably about 2,000-2,500 mm. the “Nothofagussteppe ” of Vuilleumier (1972). Rainfall was probably 500-700 mm. CENSUS PROCEDURES Sparse Cedar Forest (F)-3.5 km southeast I censusedbirds in the three grasslandsites by of the junction of the Rio Cuyin Manzano and using variable distance strip censuses(Emlen the Rio Traful, and 6 km east-southeast of 197 l), which I believe are the most appropri- Confluencia, Neuquen. The cedar (Austroced- ate for habitat of such low stature. Each seg- ruschilensis), often 5-8 m high, was the most ment that was 0.1 km long was tallied sepa- abundant plant, followed by shrubsof the gen- rately. I measured the vegetation at the end of era Diostea,Lomatia, and Berberis.Rainfall every fifth (0.5-km) segment. At sites of taller was probably 400-600 mm. stature, I censusedthe birds by usinga variable Dense Cedar Forest (G)-2.5 km northeast distance station count (Reynolds et al. 1980) of Confluencia, at the junction of Rio Limay with stations 100 m apart on a line through and Rio Traful, Neuqutn. Similar to the pre- the habitat. Vegetation was measured at each ceding site in rainfall, its vegetation differed station. I believe that any differences that re- largely in having a higher density of cedar. sulted from comparing data by using the two Southernbeech forest sites. Beech (dom- different censusmethods are minimal and do beyi-no bamboo) Forest (H)- 1.5 km north- not affect speciesrichness or diversity calcu- east of the preceding site in a rather steep- lations. sided canyon, it had an overstory lo-30 m I censusedbirds on about 75% of the morn- high, of primarily Nothofagusdombeyi and ings between 7 November and 10 December Austrocedrus.The understory was Berberisand 1980. This was in the spring,when singingbird Ribesmagellanicus. The rainfall was probably activity was at its height. Before beginningcen- between 750 and 1,000 mm. susing, I spent more than a week becoming Beech (pumilio-no bamboo) Forest (I)- familiar with all the songsand calls of birds in 1.5 km east-southeastof the summit of Cerro the areas. During non-census periods, I con- Otto, and 5 km west-southwestof Bariloche. tinually observed birds and spent a great deal The trees were primarily Nothofaguspumilio of time confirming the identities of birds that 20-30 m high, with an understory of Ribes were singing and calling. On each morning, I magellanicus,Berberis pierceii, and Lomatia, began within 30 min of dawn and continued with almost no bamboo (Chusquea spp.) either for 2-3 h or until wind interfered with understory. This was the “Nothofagusdom- my hearing the birds. I usually censusedbe- beyi” (sic) forest of Builleumier (1972). Rain- tween seven and 12 stations or segmentsper fall was probably between 800-l ,200 mm an- morning, spending 10 min at each station or nually. on each segment. Each site was usually cen- Dense Beech (dombeyi- bamboo) Forest susedwithin a 3- to 5-day period. (J)-a mixture of mature Nothofagusdombeyi Using the statistics derived from Ramsey 20-30 m high and small amounts of Austro- and Scott (1979) I calculated the “Effective cedrus,within 200 m of the east shore of Lago Detection Distance” (EDD) for each bird Escondido on the Llao Llao Peninsula, and speciesfor all sites combined. Using this dis- 26 km west-northwest of Bariloche. The tance to estimate the area that I surveyed for understory was a dense, almost impenetrable, each species,I then estimated the density from 474 C. JOHN RALPH the number seen at each site. I felt justified in Ohmart, and Rice 1983). The validity of the combining data from sites becauseI found no method as a quick and accurate index to vege- significant (t-test; P < 0.05) difference be- tation density is well established (Anderson, tween the EDD for the samespecies in different Ohmart, and Hunter 1983). At each station, I habitats. I determined this from an examina- estimated asaccurately aspossible the distance tion of the six speciesthat occurred in more at which a l-m2 board would, on the average, than one of the three major non-grasslandhab- be 50% obscuredby foliage at various heights, itat types with more than 25 individuals in combining all compass directions. These each of at least two habitats. In none of these heightswere O-O.1, 0.5, 1, 2, 5, 10, 20, 30,40, species did the average EDD differ between and 50 m. If the distancewas greater than 200 habitats. This is consistent with the studies m, no data were taken at that height. At each summarized by Dawson (198 1:13). Those height, I estimated the percent that each plant speciesthat occurred so sparselythat no mean- speciesor growth form (Table 2) contributed ingful EDD could be calculated were assigned to the foliage. The distance can be translated EDDs of birds with similar habits. This as- into foliage density, by species,at each height signment had little effect upon diversity (H’) by use of the following derivation of Mac- calculations,since rare speciescontribute little Arthur and MacArthur’s (196 1) formula: to the diversity measure. Although wide-rang- k = 0.69315 ing species,such as raptors, do not meet most of the assumptionsof the censusmethods, be- D ’ cause they are relatively rare, they also con- where k is the estimate of foliage leaf area (m2/ tribute almost nothing to diversity calcula- m3), D is the distance to the imaginary board, tions. and 0.693 15 is log2. Diversity was derived from the standard H’ In the analysesbelow, I usedthe total amount method (MacArthur and MacArthur 196 1) for of foliage surface area to calculate the abun- each of my sites. The percent used in the di- dance of each plant species.This number was versity calculation was the percent that each the total leaf surface area, derived from the speciescontributed to the total density at each calculation in method (2) below in determining site. Because of the criticism that some in- Foliage SpeciesDiversity. vestigators have applied to variable distance FoliageSpeciesDiversity. This wascalculated census methods (e.g., Dawson 198 l), I also from the volume of each species or growth used the number of individuals that I actually form, either: (1) proportional to the foliage area saw or heard at each site as the basis of my (m2/m3) that each speciesor growth form oc- diversity calculations. The subsequentresults, cupies, weighted by the height of that span; however, did not differ noticeably, except that or (2) a span measurement, with each height correlationswith habitat variables were almost measurement contributing equally, regardless invariably slightly higher using the density fig- of the area that the estimate sampled. Al- ures. This indicates that the somewhat arbi- though (2) has been used by other investi- trary assignmentof EDDs probably had little gators, I wished to see if (1) might be a better effect upon the data. Although the number of predictor, as it estimates the total leaf surface stationsthat I censusedat each site varied, and area of vegetation. therefore resulted in potentially more species At each site, I discriminated between plant at siteswith more effort, theserare speciescon- species whenever possible, but many of the tributed little to the H’ calculations.I therefore plants were recorded as “shrub A,” “shrub B,” combined all observations at a site for the H’ and so forth. Speciesdiversity was calculated calculation. I calculated bird speciesrichness on this basis. For Table 2, however, and for (BSR) by calculating the mean of the number comparisons between sites, these were com- of speciesthat I saw at each station in each bined into such categoriesas “grass, sp.” and site. “shrub, sp.” Englishand scientificnames of birds are tak- FoliageHeight Diversity. This measurement en from Humphrey et al. (1970) Johnson was calculated from the percent of the total (1967), and Olrog (1959) in that order of foliage area of all speciesof plants contributed priority. Scientific names not contained in the by each level in a similar fashion to the species text are given in Appendix 1. diversity measurewith either: (1) proportional diversity, with each estimate at each level con- tributing proportionally to the area it samples; VEGETATION MEASURES or (2) interval diversity, with each level con- Vegetation was quantified by a method origi- tributing equally. The latter is the method that nated by MacArthur and MacArthur (196 1) MacArthur and MacArthur (1961) used. and used extensively by Ohmart and his col- Comparisonof bird andplant communities. leagues (e.g., Rice et al. 1983; Anderson, I compared bird communities as a whole with PATAGONIAN BIRD HABITATS 415 various aspects of the plant community, by measures)increased at each site (all P < 0.0 1; using the average of each variable at each of Table 3). FSD, total leaf area, and average can- the 12 sites. For these and other analyses, I opy heightwere not significantlycorrelated with used a simple correlation analysis, as well as any of the bird measures, and BSD was not stepwisemultiple regressionanalyses. This lat- correlated with any of the plant measures. In ter technique adds independent variables to a stepwisemultiple regressionanalysis, the best the one that was most correlated with the de- predictor of the bird variables was always pro- pendent variable. As it adds a variable, it de- portional FHD. Adding a second variable to termines the “maximum r2 improvement” these models markedly improved the percent (Hocking 1976, SAS 1979) or the highest per- of the variation explained. Adding total leaf cent of the variation of the dependent variable area to FHD improved both prediction of bird explained by two or more independent vari- speciesrichness (r2 = 0.88; P < 0.0001) and ables. Thus, a model with an r2 of 0.95 ex- total density (r2 = 0.74; P < 0.01). BSD was plained 95% of the variation. Surprisingly, no bestexplained by addingaverage canopy height vegetation variables were highly intercorrelat- to proportional FHD (r2 = 0.62; P < 0.05). ed with any others;therefore, all were included All of these relationships were positive, as in the analyses. many other workers have found in many com- munities. Inspection of the correlations de- rived (e.g., Fig. 2), however, shows that these RESULTS results were largely due to the low diversity BIRD CENSUSES and abundance of both birds and plants at the My resultsrepresent data taken at 275 stations, grasslandsites. and within the limitations of the method (see Non-grasslandsites. Analyses that excluded papersreferenced in Ralph and Scott 198 1:x), grasslandsites, conversely, showed that BSD I feel that they fairly approximate the actual decreased as vegetation measures increased populations present. The data (Table l), ex- (Table 3). This is the paradoxical situation that pressed in individuals per 10 ha, should not Vuilleumier (1972) inferred from his obser- be construed to be as precise as this figure. vations, which did not include data from grass- Rather, these figuresare for comparative pur- land habitats. posesonly. A predictive model using stepwise regres- I recorded 50 species at all the sites com- sion analysis,even with three vegetation vari- bined. The most common wasthe White-crest- ables, failed to predict either bird speciesrich- ed Elaenia (Elaenia albiceps),a small tyrannid ness(BSR; r* = 0.45; P = 0.36) or estimated that gleans foliage as well as catches flying total density (r2 = 0.12; P = 0.68). BSD was insects.It was absentonly in the grasslandsites. best predicted, however, by a combination of The House Wren (Troglodytesaedon) was next average leaf area and interval FSD (r2 = 0.9 1; in abundance, followed by the Thorn-tailed P < 0.001). Rayadito (Aphrastura spinicauda) and the The cause of these apparently paradoxical Green-backed Firecrown (Sephanoides se- correlations can best be seen in an example phanoides), a small hummingbird. (Fig. 2). If grasslandsites are excluded from a comparison of BSD and proportional FHD, VEGETATION MEASURES the relationship is inverse: BSD declines as FHD increases. The five beech forest sites had much higher Thus, birds reach their greatest abundances total leaf area (57-l 20 m2/100 m3) than the and diversities at intermediate densities and other sites(Table 2). The cedar and scrubareas diversities of foliage and plant species.A fur- rangedfrom 16 to 37 m2/1 00 m3, and the grass- ther example of this pattern is bird species lands between 3 and 7 m2/100 m3. diversity as compared to total leaf area (Fig. No consistent pattern was found in Foliage 3). SpeciesDiversity (FSD; Table 2), either with An interesting pattern emerges when com- the proportional or the interval method. Fo- paring the bird speciesrichness with either av- liage Height Diversity (FHD) was more pre- eragecanopy height (not shown)or averageleaf dictable, however, with the taller, more com- area (Fig. 4). A regression line, drawn sepa- plex forests generally having, as expected, rately on either the beech forests or the com- higher values than the scrub or grasslandhab- bined scrub-cedar sites, shows a significant itats. negative correlation. That is, in the more com- plex habitats, the number of speciesdeclined. COMPARISON OF COMMUNITY DIVERSITIES When these very different habitats were com- AND ABUNDANCES bined, however, this relationship disappeared. All sites. The number of bird speciesand the Becauseof this persistentdifference between density of birds increased as FHD (by two grasslandand non-grasslandsites, many of the 476 C. JOHN RALPH
TABLE 1. Resultsof bird censuswith calculationsof density (Den.) in birds per 10 ha, numbers of individuals (Ind.) of each speciesthat were seen at each location, and summary statistics.A plus (+) indicates less than 0.5 birds per 10 ha.
Grasslands Scrub Bunchgrass cactus Herbaceous Sedge Beech Den. Ind. Den. Ind. Dell. Ind. Den. Ind. Den. Ind.
Buff-neckedIbis _ _ - _ _ 1 6 - - Ashy-headed Goose _ _ - + 4 _ _ + 2 Andean Condor - - _ _ _ - - + 1 Black Vulture - _ _ _ _ + 1 _ _ Turkey Vulture _ - ______Red-backed Hawk - - 1 _ _ + 1 _ _ CinereousHarrier - - _ _ _ + _ _ Chimango Caracara - - _ + 1 + ; + 1 American Kestrel _ _ _ _ + _ _ California Quail _ _ _ 2 8 9 3: 4 19 Southern Lapwing _ _ _ _ 1 2 2 16 - - Eared Dove + 1 _ + 1 + 3 + 2 Austral Parakeet _ _ _ _ - - _ _ - _ Band-winged Nightjar _ _ + 1 _ _ _ - _ _ Green-backed Firecrown _ _ _ - _ _ 1 1 1 1 Magellanic Woodpecker - - ______Austral Flicker - - _ _ _ _ - _ 2 16 Striped Woodpecker ______- - - _ Short-billed Miner _ _ + 1 _ _ _ _ - _ Rufous-banded Miner ______1 2 - _ Scale-throatedEarthcreeper - _ + 3 _ _ _ _ + 2 Thorn-tailed Rayadito ______- - _ Lesser Canastero _ _ 1 3 _ _ 36 46 9 13 Austral Canastero _ _ - _ 1 1 16 11 8 6 White-throated Treerunner _ _ - - _ _ - _ + 1 Huet-huet _ _ - _ - - _ _ - Chucao _ _ - - _ _ _ - + 5 Magellanic Babbler _ _ - _ _ _ - _ - _ Great Shrike Tyrant _ _ - _ _ - 4 6 - _ Grey-bellied Shrike Tyrant _ _ 1 _ _ _ - - _ Chorlote _ _ + ; _ _ _ _ - _ Fire-eyed Diucon _ _ - + 1 _ _ 3 9 Spot-billed Ground-Tyrant _ _ 1 9 5 10 - _ _ _ Cinnamon-bellied Ground-Tyrant _ _ + 1 _ _ - _ _ _ Tufted Tit Tyrant - - - _ - - 7 7 10 11 White-crested Elaenia _ _ - - _ _ 3 4 66 102 Rufous-tailed Plantcutter _ _ - _ _ _ 2 16 + 1 Blue and White Swallow _ _ _ - + 1 _ - 1 House Wren _ _ _ - _ _ 30 38 19 2: Austral Thrush - - _ - 1 2 9 28 1 4 Hellmayr’s Pipit 1 2 _ - _ _ 1 4 _ _ Austral Blackbird _ _ - - _ _ 1 3 _ _ Greater Red-breastedMeadowlark _ _ - - _ _ 5 41 _ _ PatagonianSierra Finch _ _ _ - _ - _ _ 11 33 Grey-headed Sierra Finch _ _ _ - - - 2 _ _ Mourning Sierra Finch _ _ _ - _ - : 5 _ _ Plumbeous Sierra Finch _ _ - _ _ _ _ _ Rufous-collared Sparrow _ _ 6 41 1 3 _ - 19 51 Black-chinned Siskin _ _ _ - _ _ 1 12 36 Common Diuca Finch _ _ - _ 9 18 24 3: 1 2 Total density 1 3 10 70 21 52 158 323 167 346 Number of species 2 10 12 26 23 Number of stations censused 14 60 21 22 25 Diversity (density) 0.6365 1.4632 1.7522 2.3556 2.0726 Diversity (individual) 1.0986 1.4393 1.9647 2.6447 2.3469 Average no. species/stations 0.2 0.1 1.0 1.7 6.1 PATAGONIAN BIRD HABITATS 471
TABLE 1. Extended.
Beech forests Cedar forests No bamboo Bambooldombeyi SiXUSe Dell= pumilio dombeyi Dense UtlfarCtiCQ pumdio Den. Ind. Den. Ind. Den. Ind.- Den. Ind. Den. Ind. Den. Ind. Dell. Ind. Total
- - ______- _ 6 _ _ - _ _ _ _ _ + 1 _ _ - _ I _ _ + 1 _ _ - _ _ _ _ _ - _ 2 ______- _ - 1 _ _ + 6 - _ _ _ - - _ _ _ _ 6 + 1 + 2 1 4 _ - _ _ + 1 - - 10 _ - ______2 ______- _ _ _ _ - _ 4 _ _ _ _ - - _ _ - - _ _ _ _ 1 1 15 5 16 ______96 _ _ _ _ - ______- _ - 18 _ - ______I _ _ - _ _ _ - _ + 3 + 3 1 4 10 ______- - _ - 1 _ _ 4 3 8 8 _ _ 41 33 48 70 35 20 136 _ _ _ _ 1 4 _ _ + 1 _ - _ - 5 _ _ _ - + 1 _ _ + 2 + 2 _ - 21 _ _ _ _ 1 I _ _ _ - _ _ + 1 8 ______- - _ - 1 _ _ _ _ - _ _ _ - _ _ _ _ _ 2 _ _ _ - _ _ _ - _ _ - - _ - 5 _ _ 3 4 30 52 15 I 35 48 32 80 29 28 219 17 13 11 11 _ _ - _ _ _ - _ - _ 86 7 3 9 5 _ _ _ _ - _ _ _ _ _ 26 _ _ _ _ 3 10 1 1 3 9 4 17 4 I 45 _ _ _ _ 2 9 1 2 6 3 1 3 4 14 31 + 3 _ _ 1 19 _ - 2 22 3 66 2 13 128 + 2 _ - + 3 _ _ + 1 1 11 + 3 20 _ _ _ _ _ - ______6 _ _ - _ _ _ - _ _ _ _ _ - _ 8 ______- _ - 2 _ _ + 1 1 4 _ _ _ _ 1 4 7 13 32 _ - _ - _ _ _ _ _ - _ _ _ _ 19 ______- _ 1 21 15 13 10 _ _ _ - _ _ 3 5 _ - 48 31 25 13 14 84 130 41 18 70 87 41 106 12 63 549 _ _ + 1 - ______18 - - _ _ 2 7 _ _ _ _ 3 14 2 3 26 21 20 36 38 49 71 20 8 26 30 11 23 50 41 296 1 2 4 9 3 9 2 2 + 1 1 4 _ - 61 ______- _ _ - 6 1 1 ______3 14 2 3 21 ______41 2 3 5 11 15 43 I 6 10 23 4 18 12 19 156 ______- _ - 2 _ - _ _ _ - ______- _ _ _ 1 1 _ _ 2 2 _ _ _ _ - _ : ______- - _ - 1 2 _ - 91 10 16 9 19 11 33 1 1 + 1 1 6 1 1 115 _ _ _ _ - _ _ _ - - _ _ - _ 56 129 119 112 152 210 414 91 41 189 265 163 449 220 233 2,413 13 17 17 9 15 19 15 13 18 25 7 20 36 14 215 1.9360 2.1750 1.7680 1.5328 1.6316 1.8595 1.8325 2.1573 2.3919 2.1603 1.7822 1.9584 2.2526 2.2102 5.7 4.1 7.0 4.4 6.0 6.2 7.8
PATAGONIAN BIRD HABITATS 419
TABLE 3. Comparison of community diversities and abundances.
Foliage species diversity Foliage height diversity Foliage surface Bird variables Canopy height area Proportional Interval Proportional Interval
All locations (n = 12) Average no. species +0.48 +0.45 +0.29 +0.36 +0.85*** +0.79** Bird speciesdiversity +0.03 +0.03 +0.47 +0.39 +0.47 +0.37 Average density +0.58* +0.55 +0.21 +0.29 +0.86*** +0.78** No grasslands (n = 9) Average no. species -0.22 -0.34 +0.08 -0.22 -0.10 -0.22 Bird speciesdiversity -0.86** -0.86** +0.34 -0.17 -0.76* -0.87** Average density +0.16* +0.08 -0.09 -0.27 +0.24 +0.03
P <: * = 0.05; ** = 0.01; *** = 0.001. results below are divided into analyses of all -0.78; P -C 0.01). In a stepwiseregression, fo- sites, and all sites except the grasslands. liage of Mulinum and all Nothofagus com- bined explained 86% (P < 0.001) of the vari- BIRD SPECIES RICHNESS (BSR) VS. INDIVIDUAL PLANT SPECIES ation in BSD. Additional plants explained only about 4.5% more of the variation. BSR was negatively correlated with the leaf Excluding grasslands,increased BSD was surfacearea of Mulinum, a common grassland correlated with increasing foliage of all shrub plant (r = -0.86; P < O.OOl), and positively speciescombined (r = 0.84; P < O.Ol), grasses correlated with foliage area of Berberis (r = (r = 0.78; P < 0.05) and with decreasing 0.78; P < 0.01). A stepwise regressionusing Nothofagus foliage area (r = -0.90; P < both speciesexplained 89% of the variation in 0.0001). A stepwise regression with the best BSR (P < 0.0001). Addition of other plants two variables chose Nothofagu and all shrubs added less than 3% to the model. (r* = 0.95; P < 0.0001). Addition of another Excluding the grasslandsites, foliage of Ber- species or group increased the variation ex- beris was a good correlate of BSR (r = 0.75; plained by less than 3%. P < 0.05). Increasing BSR in the non-grass- land sites, however, was also correlated with ESTIMATED TOTAL BIRD DENSITY VS. decreasing foliage of the cedar, Austrocedrus INDIVIDUAL PLANT SPECIES (r = -0.88; P < 0.01). The “best” stepwise Greater leaf surface area of a grasslandplant, regressionof three variables included Austro- Mulinum, was correlated with decreasingbird cedrus, Nothofagus antarcticus, and a small density of all species combined (r = -0.73; herb (Acaena; r* = 0.93; P -C 0.01). Adding P < 0.0 l), reflecting the depauperatenature of further variables increased the percent varia- the grassland bird community. A stepwise tion explained by less than 4%. regressionadded another negative vegetation component to Mulinum, that of cedar, result- BIRD SPECIES DIVERSITY (BSD) VS. INDIVIDUAL PLANT SPECIES ing in a significant model (r2 = 0.80; P < 0.001). In addition, increasing Berberis was The only plant speciesthat had a significant correlated (r = 0.73; P < 0.01) with increas- correlation with BSD was Mulinum (r = ing bird density. Among the non-grassland sites, increasing c: bird density was correlated with decreasing : I-----’ \‘ Austrocedrus(r = -0.84; P -=c0.01). This was : , I4 \‘ I \\ the only significantcorrelation and, sincethere E : t : - SOUTHERN BEECH,’ : were 17 plant variables, could be expected by : \J‘ Li chancealone. The best three-variable, stepwise ITJ .\_‘ K :_----_.> ___
/---- T ,: J \\‘ I \ z \ H \ = loo- \ \ SOUTHERN BEECH : \ “E \ \ \‘ 1 \ K I : ;‘ I 5 \‘ L : LL 5o- \.__/‘ ’ CEDAR-SCRUB __--__ 2 ,,-_ G -., .d . ’ \\‘ -I - 3‘ GRASSLAND 1 E 2 ~______~ \F‘ D < ’ B o (A‘ 1‘ .-______/‘ : _ C! *. -______r!o______-_--__,‘ ’ 2!. I
BIRD SPECIES DIVERSITY (H)’ FIGURE 4. Average number of speciesper station for each location in three habitat types in northern Patagonia, FIGURE 3. Bird species diversity compared with the Argentina, as compared to average leaf areas of all plant total leaf area from three habitats in northern Patagonia, speciescombined. Letters refer to locations described in Argentina. Lettersrefer to locationsdescribed in the Meth- Methods. ods.
Seven birds were positively correlated with bined (Table 4). I found 24 (89%) of these grasses,and many of these were open-grass- significantlycorrelated with foliage surfacearea land foragers. Some, however, reached their of at least one plant species.Surprisingly, an peak densitiesin the scrub-sedgehabitat. Two increasein the surface area of most plants was other specieswere negatively correlated. Nine significantly correlated with an increase in the birds were positively correlated with Notho- bird count. If chance had been operating to fag-us, either with one of the species, or by any significant degree, I would have expected combiningall Nothofagusspecies. Most ofthese many more negative relationships. speciesforage commonly in Nothofagus.Bam-
TABLE 4. Summary of correlationsbetween speciesor growth forms ofplants and speciesof birds and the best plant variables used in constructinga two-variable stepwiseregression model. Numbers in the table are simple correlation coefficients.’ r2 is the percent of the variation explained by the model.
AUSWO- Species n Berberis Dim-tea Churquea cedrus DeSfontainea Grass sp. Herbs sp.
Red-backed Hawk - - - - - California Quail :: - - - - - 0%’ Southern Lapwing 18 0.40 - - - - 0.89”” Austral Parakeet 10 - - - - Green-backed Firecrown 136 0.64+ - 0.91++++ - 028’ 1 Austral Flicker - - - -0.04 - Thorn-tailed Rayadito 2;; - - 0.73’ - - - Lesser Canastero 86 - - - - - 0.76” Austral Canastero 26 - - - - - 0.71 White-throated Treerunner 45 - - 0.72” - - - Huet-huet 31 ------Chucao 128 0.72” - 0.87”’ - 0.78” - Magellanic Babbler 20 0.63’ - 0.58+ - 0.77” - Fire-eyed Diucon 32 ------Spot-billed Ground-Tyrant 19 - - - - - Tufted Tit Tyrant 48 - -(0.57) - - - White-crested Elaenia 549 ------0.62’ Rufous-tailed Plantcutter 18 ------0.91”” Greater Red-breastedMeadowlark 4 1 - - - - - 0. go++++ House Wren 296 ------Austral Thrush 61 - - - - - 0.81++ Austral Blackbird 21 0.83”’ - - - 0.88+++ - Blue and White Swallow 26 0.67’ - - - 0.76” - PatagonianSierra Finch 156 - - - - -0.02 - Rufous-collared Sparrow 97 - - - - - Black-chinned Siskin 115 - 026’ - - - - Common Diuca Finch 56 - - - - - 0.86+++
a Their significance level is indicated to the right of each number (P C: + = 0.05; ++ = 0.01; +++ = 0.001.; ***+ = 0.0001). Italic numbers are those variables that contributed to a two-variable model (correlation values are given for those variables that were not slgniticantly correlated, but contributed to a model). b Although all Nothofagus species combined were the best correlated, the two species separately contributed best to a two-variable model. EAlthough positively correlated (P = 0.0539, the information it contributed to the predictive model was negative. d Mu/inum alone was the only combination of variables that would significantly predict the species’ abundance. PATAGONIAN BIRD HABITATS 481 boo (Chusqueu)was positively correlated only with bird abundance. Specifically, in the non- with a rhinocryptid, the Chucao (Scelorchilus grasslandsites, as foliage height diversity, av- rubecula),and the Green-backed Firecrown. erage canopy height, and average leaf surface Abundances of three other birds were corre- area all increased, bird species diversity de- lated with bamboo, but it was not selectedin creased(Table 3). stepwise models, possibly reflecting only an The explanation of this paradoxical situa- ancillary role in organization of the bird com- tion probably lies in the specific habitat rela- munity. Some shrub species were also in- tionships of each of the bird speciesto each volved in models or correlations. Their roles plant species.My analysesshow that the prin- as a substratefor feeding or as food themselves cipal correlates and predictors of the abun- are unknown. Most important were Mulinum, dances of individual bird species were the with mostly negativeassociations, Lomatia and abundancesof certain plant taxa, namely: pos- Berberiswith all positive associations, and itive correlations of Berberis,grass, Chusquea, Desfontaineawith mostly positive associa- Nothofagusdombeyi, N. pumilio, Desfontai- tions. nea, and Lomatia; and a negative correlation A few other plants were fairly common and with Mulinum.All of these,with the exception in more than one site, but were not well cor- of Lomatia, are plants of either the grassland related with any bird species.These were the or the beech forests. Those plants most com- cedar (Austrocedrus),a currant (Ribesmagel- mon (Table 2) in the intermediate scrub and lanicus),and a shrub(Diosteu junceu). Of these, cedar forest sites, such as Austrocedrus,Em- the lack of associationwith Austrocedruswas bothrium,and Diostea,were all rather unim- most surprising,as it is a major component of portant as predictors or correlates of bird some ecosystems. speciesabundances (Table 4). I found no real indication that bird species DISCUSSION diversity was related in any simple fashion to My major finding is the first well-documented plant speciesdiversity. The sites with higher negative correlation of vegetation abundance plant diversity values were distributed irreg-
TABLE 4. Extended.
N. Nothofagus LOT7llltia MldiflUPtl antarctica N. dombeyi N. pumilio (W Embothrium Acaena Ribes Shrubs&x r’ P
- - - 0.21 - - - - 0.88 ++++ 0.51 7 ------0.79 ++ ------0.84 +++ ------0.34 n.s. - - - 0.82++ - 0.71 ’ - - - - 0.86 ++++ - 0.64 ’ ------0.96 ++++ - 0.75” +0.43 0 92++++b - - - - 0.89 ++++ - -0.24 ------0.78 ++ 0.51 ------0.88 ++++ - 0.70 ’ - 0.87”” - - 0>3 - 0.81 +++ ------0.24 n.s. - - 0.74++ +0.22 0.83”’ - - - ++++ ------025 % ++ - - - - - 0.18 n.s. 0.59 ’ - - - - - 0.37 n.s.* 0.79” - - - - 0.63 ’ 1 - - 0.69 ++ - - 0.69 ’ - - 0.77++ - - - - 0.73 ++ - -0.10 ------0.94 ++++ -0.08 ------0.91 ++++ - - 0.74” -0.21 ------0.82 ++ - -0.28 - - - 0.90 ++++ - - - - - 0.87 +++ - 0.51 0.71” - - 0.92 ++++ -0.65 ’ - 0.59 ’ 0.67 ’ - - 0.61 ’ - 0.61 + 057 - - - - 0.65 ’ - - 0.68 ++ 0.79” - 0.42 - - - 0.98 ++++ - - - -0.17 1 - - 0.76 + ’ 482 C. JOHN RALPH ularly through the habitats (Table 2). Although stability and presumably their diversity. The some of these higher plant values were in the most important is the isolation of these forests scrub-cedar habitats, this was, by no means, from others. The Nothofagusand Araucaria always the case. The highest bird speciesdi- forests of the Andes of Argentina and Chile versity and (generally) richness occurred in are separatedfrom other forestsby about 2,000 habitats that were intermediate in vegetation km of steppe and desert. This might lead to measuressuch as total foliage area and foliage relatively small populations which are more height diversity. This indicates that these hab- prone to local extinction, with resultant low itats are either unique in themselves, or they diversity. The second factor is the relative possesssome attribute(s) that causesthem to harshnessof the climate. These forests have a be rich in bird species.Their intermediate stat- more persistent snow cover in the winter than ure and foliage height diversities argue against do the steppehabitats at lower elevations. This any uniqueness, at least in their vegetation climatic factor may add to the risk of local structure. extinction and unstablebird populations,when Rather, it seemspossible that their high bird coupled with the forests’ isolation, the possible speciesdiversity (as well as possibly richness) “island” effect, and low colonization rates. Fi- can be attributed to their intermediate position nally, the abundance of bamboo (Chusquea in the moisture gradient between the wide- spp.) in the understory of many of the beech spread steppe habitats of southern South forests(Table 2) may play a role. The large leaf America, and the unique and isolated Noth- surface area of bamboo, coupled with what I ofagusforest of the Andes, as suggestedby observed to be its apparent low use by insects Rabinovich and Rapoport (1975: 146). The (possiblydue to its relative unpalatability), may scrub-cedarhabitat has elementsof both steppe produce lower food densitiesfor foraging birds and beech forests.Twelve of the 27 bird species than is normal in taller forests such as these. that were common in my census(Table 4) were An interestingtest of the role of bamboo will positively related to plant speciesfound in the come with its blooming (it blooms only once beech forests, such as Nothofugusitself, Chus- about every 40 + years) and subsequentsuper- quea,and Ribes,as well as others. Many of abundance of seeds.This may happen in the these bird speciesare probably adapted to the next few years (0. P. Pearson, pers. comm.). I beech forests,as many do not commonly occur predict that, during the year of extensivebloom outside of them (Vuilleumier 1967), although and seed set of this important component of they can be found in some of the scrub-cedar many of the beech forests, the number of bird habitats. Similarly, the birds adapted to the speciesusing the forest will increase.This may widespreadsteppe habitats can live in the tran- also result in an increase of bird speciesdi- sitional cedar forest habitats. These interme- versities. diate habitats are not common in this region; they occur only where the elevational and rain- ACKNOWLEDGMENTS fall gradientsintersect appropriately. One might I am grateful to 0. P. and A. Pearson, who provided the be tempted to ascribethis richnessto the “edge” incentive for this work, much help on the identification effect, the mixture of two diverse habitats. It of birds and plants of the area, and their hospitality. M. is more than this, however; some plant species Christie provided a great deal of active help and advice at several stagesof the research.They, as well as B. An- are generally restrictedto this habitat type (e.g., derson, A. Doyle, J. T. Emlen, D. James, J. Kikkawa, P. Lomatia and Diostea-Table 2). Therefore, Hall, B. Noon; R. Ohmart, C..P. Ralph, M. G. Raphael, floristically and faunistically, the scrub-cedar J. Vemer, and F. Vuilleumier read drafts of this manu- habitat is a mixture of steppeand beech forest, script and made many extremely helpful suggestions. containing not only substantial elements of LITERATURE CITED each, but also unique elements of its own. This combination might assurespecies richness and ANDERSON,B. W., R. D. OHMART, AND J. RICE. 1983. Avian and vegetation community structureand their diversity. seasonalrelationships in the lower Colorado River This explanation, however, contradicts the Valley. Condor 85392-405. prediction that bird species diversity, espe- ANDERSON,B. W., R. D. OHMART,AND W. C. HUNTER. cially the evennesscomponent, would be higher 1983. Quantifying variables for classifyingdesert ri- in the structurally complex beech forests, even parian vegetation, p. 32-44. In W. H. Moir and L. Hendzel [tech. coords.],Proceedings of the workshop if they are somewhat poor in bird species.One on southwesternhabitat types. USDA Forest Service, might expect that those fewer speciesshould SouthwesternRegion, Albuquerque, NM 87 102. be about equal in abundance (the evenness CHURCH,V. 1974. Bird speciesdiversity in Patagonia: component of diversity), compensatingfor the a critique. Am. Nat. 108:235-236. DAWSON,D. F. 1981. Counting birds for a relative mea- lower richness. sure (index) of densitv, D. 12-16. In C. J. Ralnh and The beech forestsof the Andes are, however, J. M.‘Scott [eds.], Es&matingthe number ofterrestrial under three constraintsthat might lower their birds. Stud. Avian Biol. 6. PATAGONIAN BIRD HABITATS 483
DE FINA, A. L. 1972. El clima de la regi6n de 10sbosques WILLSON,M. F. 1974. Avian community organization Andino-patag6nicos Argentinos, p. 35-58. In M. J. and habitat structure.Ecology 55:1017-1029. Dimitri [ed.], La regi6n de 10sbosques Andino-pa- tag6nicos.Sinopsis General, Vol. 10. Colecci6n Cien- Redwood SciencesLaboratory, Pacific SouthwestForest tlfica de1 Instituto National de Tecnologla Agrope- and RangeExperiment Station, USDA ForestService, 1700 cuaria, Buenos Aires. BayviewDrive, Arcata, California 95521. Received 2 Au- EMLEN,J. T. 1971. Population densitiesof birds derived gust 1984. Final acceptance28 May 1985. from transect counts. Auk 88:323-342. HOCKING,R. R. 1976. The analysis and selection of variables in linear regression.Biometrics 32:1-50. HOWELL,T. R. 1971. An ecologicalstudy of the birds of APPENDIX 1. List of specieswhose scientificnames do the lowland pine savanna and adjacent rain forest in not appear in the text. northeastern-Nicaragua.Living Bird 10:185-242. HUMPHREY.P. S.. D. BRIDGE.P. W. REYNOLDS.AND R. Buff-neckedIbis Theristicuscaudatus T. PETERSON.1970.’ Birds of Isla Grande (Tierra de1 Ashy-headed Goose Chloephagapoliocephala Fuego). Preliminary Smithsonian Manual. Mus. Nat. Andean Condor Vultur gryphus Hist., Univ. Kansas, Lawrence. Black Vulture Coragypsatratus JOHNSON,A. W. 1967. The birds of Chile and adjacent Turkev Vulture Cathartesaura regionsof Argentina, Bolivia and Peru. Platt Estable- Red-backed Hawk Buteopolyosoma cimientos Graficos, BuenosAires. CinereousHarrier Circuscinereus KARR, J. R., AND R. R. ROTH. 1971. Vegetation structure Chimango Caracara Milvago chimango and avian diversity in several New World areas.Am. American Kestrel Falco sparverius Nat. 105:423-435. California Quail Lophortyx californica MACARTHUR,R. H. 1964. Environmental factors af- Southern Lapwing Vanelluschilensis fecting speciesdiversity. Am. Nat. 98:387-397. Eared Dove Zenaida auriculata MACARTHUR.R. H.. ANDJ. W. MACARTHUR. 1961. On Austral Parakeet Enicognathusferrugineus bird speciesdiversity. Ecology 42:594-598. Band-winged Nightjar Caprimulguslongirostris OLROG, (?. C. 1959. -Las aves Argentinas. Instituto Magellanic Woodpecker Campephilusmagellanicus “Miauel Lillo.” Universidad National de Tucuman. Austral Flicker Colaptespitius Tucim&n, Argentina. Striped Woodpecker Dendrocopuslignarius PEARSON,0. P., AND C. P. RALPH. 1978. The diversity Short-billed Miner Geosittaantarctica and abundanceof vertebratesalong an altitudinal gra- Rufous-banded Miner Geosittarujipennis dient in Peru. Mem. de1Mus. Hist. Nat. “Javier Pra- Scale-throatedEarthcreeper Upucerthiadumetaria do,” No. 18, Lima, Peru. Lesser Canastero Asthenespyrrholeuca RABINOVICH,J. E., AND E. H. RAPOPORT.1975. Geo- Austral Canastero Asthenesanthoides graphicalvariation of diversity in Argentine passerine White-throated Treerunner Pygarrhichasalbogularis birds. J. Biogeogr.2: 141-l 57. Huet-huet Pteroptochostarnii RALPH, C. J., AND J. M. SCOTT [EDS.]. 1981. Estimating Magellanic Babbler Scytalopusmagellanicus numbers of terrestrial birds. Stud. Avian Biol. 6. Great Shrike Tyrant Agriornis livida 1979. Estimating pop- RAMSEY,F. L., ANDJ. M. SCOTT. Grey-bellied Shrike Tyrant Agriornis microptera ulation densities from variable circular plot survey, Chorlote Pseudoseisuraguttural& p. 155-l 8 1. In R. Cormack, G. Patil, and D. Robson Fire-eyed Diucon Pyropepyrope [eds.], Sampling biological populations.International Spot-billed Ground-Tyrant Muscisaxicolamaculiros- Co-operative Publishing House, Fairland, MD. tris RECHER,H. 1969. Bird speciesdiversity and habitat di- Cinnamon-bellied Ground- Muscisaxicolacapistrata versity in Australia and North America. Am. Nat. Tyrant 103:75-80. Tufted Tit Tyrant Anaeratesparulus REYNOLDS,R. T., J. M. SCOTT,AND R. A. NUSSBAUM. Rufous-tailed Plantcutter Phytotoma rara 1980. A variable circular-plot method for estimating Blue and White Swallow Notiochelidoncyanoleuca bird numbers. Condor 82:309-3 13. Austral Thrush Turdusfalcklandii RICE, J., R. D. OHMART,AND B. W. ANDERSON.1983. Hellmayr’s Pipit Anthus hellmayri Habitat selection attributes of an avian community: Austral Blackbird Curaeuscuraeus a discriminant analysis investigation. Ecol. Monogr. Greater Red-breasted Mea- Pezitesmilitaris 53:263-290. dowlark SAS (STATISTICALANALYSIS SYSTEM). 1979. SAS user’s PatagonianSierra Finch Phry.&s patagonicus guide. SAS Institute, Raleigh, NC. Gray-hooded Sierra Finch Phrygilusgayi VUILLEUMIER,F. 1967. Phyletic evolution in modem Mourning Sierra Finch Phrvnilusfnrticeti birds of the Patagonian forests. Nature (Lond.) 2 15: Plumbeons Sierra Finch Phr$gilus”unicolor 247-248. Rufous-collared Sparrow Zonotrichia capensis VUILLEUMIER,F. 1972. Bird speciesdiversity in Pata- Black-chinned Siskin Spinus barbatus gonia (temperateSouth America). Am. Nat. 106:266- Common Diuca Finch Diuca diuca 271. -