Herzog Et Al 2003. Seasonal Variation in Avian Community Composition in a High-Andean Polylepis (Rosaceae) Forest Fragment

438

Wilson Bull., 115(4), 2003, pp. 438–447

SEASONAL VARIATION IN AVIAN COMMUNITY COMPOSITIONIN A HIGH-ANDEAN POLYLEPIS (ROSACEAE)

FOREST FRAGMENT

SEBASTIAN K. HERZOG,1,3,4,5 RODRIGO SORIA A.,1,3 AND ERIK MATTHYSEN2

ABSTRACT.—We analyzed seasonal variation in avian species richness and relative abundance at the com-munity and guild level during a 13-month period in central Bolivia in an 11-ha patch of Polylepis (Rosaceae)forest, a high-Andean ecosystem that has suffered from extreme anthropogenic fragmentation. Birds were sur-veyed audio-visually (supplemented with mist net data) and assigned to five relative abundance categories for2-month survey periods. We recorded 35 core species, including 16 insectivores (46%), 11 frugi-granivores(31%), seven nectarivores (20%), and one carnivore (3%). Core species richness varied from 25 (June–July) to33 (October–November). Insectivores had a significantly higher proportion of year-round residents (81%) thanfrugi-granivores (45%) or nectarivores (14%); the same trend was evident with respect to seasonal variation inspecies richness of each guild. Although most species varied considerably in their relative abundance, no guildshowed significant variation in relative abundance scores among survey periods. However, frugi-granivores andnectarivores combined, both of which depend upon plants as food resources, reached a significant minimum inmid-winter (June–July), and the same result was found at the community level. The insectivore guild thus wasthe most temporally stable both in terms of species richness and abundance. Qualitative observations indicatedthat the fluctuations in frugi-granivores and nectarivores were related to the availability of food resources.Received 22 April 2003, accepted 1 August 2003.

Seasonal variation in the composition ofNeotropical bird communities has been inves-tigated in only a small number of case studies(Karr 1976, Brooks 1997, Hilty 1997, Schmittet al. 1997). Such research often is hamperedby high levels of Neotropical community di-versity, a fairly large proportion of hard-to-detect cryptic or naturally rare species, andlogistic constraints. These factors frequentlyimpede the use of standard survey techniquesand make it difficult to obtain sufficientlycomplete, comparable species inventories(Herzog et al. 2002a). Most studies on sea-sonal variation therefore have limited theirscope to subsets of communities, such ashummingbirds (Stiles 1980, 1985a), understo-ry frugivores (Loiselle and Blake 1991),mixed species flocks (Develey and Peres2000), or canopy birds (Loiselle 1988).

1 Centro de Biodiversidad y Genetica, Univ. Mayorde San Simon, Casilla 538, Cochabamba, Bolivia.

2 Lab. of Animal Ecology, Dept. of Biology, Univ.of Antwerp, B-2610 Antwerp, Belgium.

3 Current address: Asociacion Armonıa BirdLife In-ternational, Ave. Lomas de Arena 400, Casilla 3566,Santa Cruz de la Sierra, Bolivia.

4 Inst. fur Vogelforschung ‘‘Vogelwarte Helgoland,’’An der Vogelwarte 21, 26386 Wilhelmshaven, Ger-many.

5 Corresponding author;e-mail: [email protected]

Temporal variability in community struc-ture is caused by local movements (Karr andFreemark 1983, Fjeldsa 1991, Davis 1993), al-titudinal migration (Stiles 1988, Loiselle andBlake 1991), or long distance latitudinal mi-gration (e.g., Marantz and Remsen 1991, Da-vis 1993) of a considerable number of species.Movement and migration patterns have beenhypothesized or documented to reflect foodresource availability (Karr 1976, Stiles 1980,Loiselle 1988, Blake and Loiselle 1991, Loi-selle and Blake 1991, Levey and Stiles 1992,Brooks 1997). Different food resources tendto show asynchronous abundance peaks (Karr1976, Develey and Peres 2000), and differentguilds therefore may be expected to show dis-similar patterns, but little data for within-com-munity comparisons are available to test thisprediction (but see Blake and Loiselle 1991).

Based on studies in lowland Panama, Karr(1976) hypothesized that in the tropics sea-sonal variability in diversity should increasefrom omnivores to frugivores to insectivores.This was supported by Brooks (1997), whofound the highest seasonality for insectivoresin the Paraguayan chaco. In Neotropical mon-tane forests, however, most altitudinal mi-grants, which often comprise .30% of localavifaunas (Loiselle and Blake 1991), are fru-givores or nectarivores (Stiles 1985b, Blake et

439Herzog et al. • HIGH-ANDEAN POLYLEPIS COMMUNITY

al. 1990, Hilty 1997), and insectivore fluctu-ations therefore may have little impact oncommunity composition in this habitat. Thus,studies in other habitats and biogeographic re-gions are required to determine general pat-terns and key factors determining the com-position of tropical bird communities.

We analyzed seasonal variation in speciesrichness and relative abundance at the com-munity and guild level in a high-Andean Po-lylepis (Rosaceae) forest fragment during a13-month period in central Bolivia. Specifi-cally, we investigated whether seasonalchanges in community richness and overallabundance occurred, and if insectivoresshowed more pronounced seasonality thanother guilds in this marginally tropical, high-elevation forest ecosystem as proposed byKarr (1976) for the tropical lowlands. SincePolylepis forests have suffered from stronganthropogenic fragmentation (Fjeldsa andKessler 1996, Kessler 2002), we paid partic-ular attention to abundance patterns of habitatspecialists to assess implications for their con-servation.

STUDY AREA AND METHODS

Study area.—We studied the bird commu-nity of an 11-ha fragment of Polylepis besseriforest located just north of the small com-munity of Sacha Loma in the department ofCochabamba, Bolivia (178 449 S, 658 349 W,3,710–3,880 m elevation) from November1999 to November 2000. The fragment con-sisted of 9.3 ha of forest enclosing a 1.7-haboulder field largely devoid of vegetation.Tree height varied from 4–10 m. Shrubs (Gy-noxis, Berberis, Ribes, Baccharis, Brachy-otum) were distributed patchily, and theground largely was covered by bunch grass(Festuca, Stipa, Calamagrostis). Except forsome densely vegetated spots, most of thevegetation was open enough to allow unob-structed passage on foot. Human impact onthe forest (cattle grazing, fire wood cutting)was low, but the surrounding grassland wasused intensively for sheep grazing. For a de-tailed description of the vegetation see Fer-nandez et al. (2001).

The study site is embedded in a landscapemosaic of high-Andean puna grassland andmore or less isolated Polylepis besseri frag-ments (varying in size from 0.01–100 ha) in

the transition zone from humid montane for-ests on the eastern Andean slope to dry inter-montane valleys of southern Bolivia. Thestudy fragment is 10 m from an about 40-haforest patch with a considerably lower treedensity and about 600 m from a 100-ha frag-ment with a similar vegetation structure. Cli-matic differences between seasons in thestudy area are pronounced. The rainy season(austral summer) extends from November toMarch, whereas dry conditions prevail fromMay to September during the austral winter.Brief transition periods occur in April and Oc-tober. No climatic data are available for thestudy area, but Fernandez et al. (2001) presenttemperature and precipitation values from twonearby climatic stations (Alalay, Toralapa) atelevations similar to that of the study frag-ment, however without stating the number ofyears of data. Mean annual precipitation at Al-alay is 72 cm with a maximum of 17 cm inJanuary. The five rainy season months com-bined receive over 60 cm of precipitation,whereas the five dry season months combinedreceive slightly less than 5 cm. Mean annualtemperature at Toralapa is 8.78C with monthlymaxima of about 10–118 C in the rainy sea-son, which drop to a minimum of about 78 Cin June. Mean monthly minima drop to belowfreezing from April to October (with a mini-mum of about 258 C in June and July) andremain slightly above 08C during the rainyseason.

Field surveys.—We surveyed the Polylepisbird community for several days each monthexcept in February and March 2000, when so-cial unrest did not allow field work in SachaLoma. Survey dates were 10–12 Novemberand 7–12 December 1999, 27–31 January, 13–18 April, 10–13 and 24–29 May, 11–13 June,2–4 and 9–11 July, 3–7 August, 15–17 Sep-tember, 14–16 and 22–24 October, and 13–17November 2000. SKH and RSA determinedthe presence and relative abundance of birdspecies by daily walks through the fragmentduring most of the daylight period, and wemonitored vocal activity of nocturnal speciesfrom our camp site near the forest edge. Weidentified birds both visually and acoustically,and SKH was familiar with sight and soundidentification of all species before beginningdata collection. Also, we operated 10–15 mistnets daily during the same visits starting in

440 THE WILSON BULLETIN • Vol. 115, No. 4, December 2003

December 1999 (see Herzog et al. 2002b). Weused intensive observations of foraging be-havior (e.g., mixed species foraging flocks;Herzog et al. 2002b) and opportunistic ex-amination of fecal material (mainly duringhandling of netted birds, rarely upon encoun-ter in the forest; fecal matter was screened forpresence of fruit pulp, seed remnants, or chi-tinous insect parts) to determine the primarydiet of each species in the study area. Forsome species we also consulted standard ref-erences (e.g., Fjeldsa and Krabbe 1990).

As some species and many individuals re-mained unmarked even after several monthsof banding operations, and because thoroughspot map surveys of all species were unfea-sible due to time constraints, we estimated rel-ative rather than absolute abundances of spe-cies for each month based on the frequencyof encounter of a given species. We distin-guished five relative abundance categories: (a)rare, no more than three observations of singleindividuals or one observation each of a pairand a single bird, suggesting that no more thantwo to three individuals may have been pre-sent; (b) uncommon, single individuals orpairs observed occasionally but not daily,probably four to six individuals present; (c)fairly common, small numbers observed dailyor almost daily, probably 7–12 individualspresent; (d) common, observed repeatedly ev-ery day, probably 13–20 individuals present;and (e) abundant, observed daily in largenumbers, .20 individuals present. Due to thebiases of mist net capture data (e.g., Remsenand Good 1996), relative abundance estimateswere based almost exclusively on visual andacoustical observations. For rare and uncom-mon species, we consulted data from mist netcaptures, and in a few cases we increased aspecies’ relative abundance by one rank whenmist net data indicated a slightly higher abun-dance than observational data.

Data analysis.—We included only core spe-cies (those that regularly breed, winter, or mi-grate through a given habitat; Remsen 1994)of the Polylepis fragment in the present anal-ysis. Species considered rare visitors fromnearby habitats, dispersing individuals, or va-grants were not analyzed. Valid, meaningfulcomparisons of community composition with-in or between sites require similar levels ofsampling effort and survey completeness

(Remsen 1994). Because survey effort variedamong months, we combined monthly datainto five 2-month periods based on climaticand biological characteristics: December–Jan-uary (11 survey days), April–May (16), June–July (9), August–September (8), and October–November (14). December and January arethe two months of greatest precipitation (Fer-nandez et al. 2001) and correspond to the latebreeding season for most Polylepis species(authors pers. obs.). April and May representthe transition period from the rainy to the dryseason, whereas June and July are character-ized by the lowest temperature and precipita-tion values (Fernandez et al. 2001). The onsetof territorial behavior in several core speciesoccurs in August and September (Herzog etal. 2002b), followed by the peak of the breed-ing season for most Polylepis species (seeFjeldsa and Krabbe 1990; authors pers. obs.)and the transition period from the dry to therainy season in October and November.

Taking into account the small size of theforest fragment, its open vegetation structure,and relatively low species richness, as well asSKH’s familiarity with the avifauna, we con-sider that despite the differences in survey ef-fort .95% of all core species were recordedin each 2-month period (see also Fjeldsa 1993for an assessment of minimum survey effortin Polylepis woodlands). Although the April–May period received the highest survey effort,six days of field work during late May addedonly a single core species that had not beenrecorded during April or early May. This in-dicates that our assumption of .95% com-pleteness in all periods is warranted.

We assigned all core species to one of fourprimary diet categories: insectivores (speciesthat primarily consume insects and other ar-thropods), frugi-granivores (species that pri-marily consume fruit, seeds, and flowers),nectarivores (species that primarily consumenectar), and carnivores (species that primarilyconsume vertebrate prey). We ranked relativeabundance categories on an ordinal scale ofincreasing abundance from 0 (not recorded) to5 (abundant). If a species’ abundance variedbetween months within the same 2-month pe-riod, the mean was used (e.g., 2.5 for a speciesuncommon in June and fairly common inJuly). However, a species that was rare in onemonth and not recorded in the other month of


a given 2-month period was still consideredrare and assigned an abundance rank of 1.

We performed sign tests to analyze changesin relative abundance within guilds and for theentire bird community between consecutive 2-month periods using STATISTICA for Win-dows (StatSoft, Inc. 1997). We examined dif-ferences among guilds in the proportion ofyear-round residents with a Fisher exact testusing StatXact (Cytel Software Corp. 1995).

RESULTS

Species richness.—We recorded a total of35 core species, including 16 insectivores(46%), 11 frugi-granivores (31%), 7 nectari-vores (20%), and 1 carnivore (3%; Table 1).An additional 25 vagrant or visiting specieswere observed in the study site (Appendix).Nineteen (54%) core species were year-roundresidents (i.e., they were recorded during all2-month periods), whereas three core species(9%) each were observed during only one pe-riod. Among insectivores, 13 species (81%)were year-round residents, whereas only 5 fru-gi-granivores (45%) and 1 nectarivore (14%)were recorded during all periods. These dif-ferences in the proportion of year-round resi-dents among guilds were significant (Fisherexact test, P 5 0.008). The single carnivore,the Yungas Pygmy-Owl (Glaucidium boli-vianum), was observed during all periods ex-cept December–January, but it may have beenoverlooked then due to its crepuscular andnocturnal activity.

Core species richness of the Polylepis birdcommunity was lowest in winter, with 25 spe-cies during June–July and 26 species duringAugust–September, and it was highest duringthe two transition periods, with 29 speciesduring April–May and 33 species during Oc-tober–November (Fig. 1A). The latter twomonths correspond to the height of the breed-ing season of most Polylepis birds (authorspers. obs.; see Fjeldsa and Krabbe 1990). Dur-ing December–January, species richness wasslightly higher (27 species) than during winter.Insectivore richness showed low variabilityamong periods, varying from 13 species dur-ing December–January to 16 species duringOctober–November. In contrast, nectarivorerichness was twice as high during April–May(six species) and more than twice as high dur-ing October–November (seven species) than

during the periods from June through Septem-ber (three species). Variability in frugi-grani-vore richness was intermediate with a mini-mum of six species during June–July andmaxima of nine species during April–May andOctober–November (Fig. 1A).

Relative abundance.—Most species variedconsiderably in their relative abundance, theextreme case being the Sparkling Violetear(Colibri coruscans), which was common dur-ing December but absent from June throughSeptember (Table 1). Of the 19 year-roundresidents, 11 varied in their abundance bymore than one category (Table 1). Thus, onlyeight species (23% of the entire core com-munity) were fairly constant year round.

Overall abundance at the community levelpeaked during October–November (sum ofabundance scores 75.0, mean 2.3) and waslowest during June–July (47.0, 1.9), and wefound similar patterns at the guild level (Fig.1B). When comparing species abundancescores between consecutive periods, none ofthe three guilds analyzed alone showed anysignificant variation (Table 2, results shownfor insectivores only). Because frugi-grani-vores and nectarivores both depend uponplants as food resources, we pooled the datafor the two guilds, which revealed a signifi-cant decrease from April–May to June–July (Z5 2.41, df 5 15, P 5 0.016) and a significantincrease from June–July to August–September(Z 5 2.02, df 5 11, P 5 0.043) in abundance(Table 2). We obtained similar results for theentire bird community, except that the de-crease in abundance from October–Novemberto December–January also was significant(Table 2). Note that we compared December–January to October–November of the nextyear, assuming between-year continuity.

Habitat specialists.—Three core species,the Tawny Tit-Spinetail (Leptasthenura yan-acensis), the Giant Conebill (Oreomanes fras-eri), and the Thick-billed Siskin (Cardueliscrassirostris) are strongly specialized on Po-lylepis forest, and five species (Rusty-ventedCanastero, Asthenes dorbignyi; Streak-frontedThornbird, Phacellodomus striaticeps; Ru-fous-webbed Bush-Tyrant, Polioxolmis rufi-pennis; D’Orbigny’s Chat-Tyrant, Ochthoecaoenanthoides; Rufous-bellied Saltator, Salta-tor rufiventris) show a marked affinity for thishabitat (see Fjeldsa and Kessler 1996). The


TABLE 1. Most of the 35 core species recorded in a Polylepis woodland at Sacha Loma, central Bolivia,varied in relative abundance (n.r. 5 not recorded; R 5 rare; U 5 uncommon; F 5 fairly common; C 5 common;A 5 abundant; see text for details) between 2-month survey periods, November 1999 to November 2000. Atthe guild level (ins 5 insectivore; fru 5 frugi-granivore; nec 5 nectarivore; car 5 carnivore), most insectivoreswere year-round residents, whereas the reverse was true for frugi-granivores and nectarivores. No data werecollected in February and March. Taxonomy follows South American Checklist Committee (2002).

Species

Survey period

Dec–Jan

Apr–May

June–July

Aug–Sept

Oct–Nov Diet

Ornate Tinamou (Nothoprocta ornata)Spot-winged Pigeon (Columba maculosa)Gray-hooded Parakeet (Psilopsiagon aymara)Yungas Pygmy-Owl (Glaucidium bolivianum)Sparkling Violetear (Colibri coruscans)

RRRn.r.C-R

RUn.r.RR

U-RU-Rn.r.Rn.r.

UUn.r.Un.r.

RRn.r.-URR

frufrufrucarnec

Black-hooded Sunbeam (Aglaeactis pamela)Andean Hillstar (Oreotrochilus estella)White-sided Hillstar (O. leucopleurus)Giant Hummingbird (Patagona gigas)Red-tailed Comet (Sappho sparganura)

Rn.r.n.r.RR

RR-Un.r.RR

n.r.R-Un.r.Rn.r.

n.r.FR-Un.r.n.r.

RU-RU-n.r.RU

necnecnecnecnec

Plain-breasted Earthcreeper (Upucerthia jelskii)Bar-winged Cinclodes (Cinclodes fuscus)Brown-capped Tit-Spinetail (Leptasthenura fuliginiceps)Tawny Tit-Spinetail (L. yanacensis)Cordilleran Canastero (Asthenes modesta)

RCn.r.CR

R-UC-Rn.r.Cn.r.

RRn.r.CR

RRn.r.CR

RF-CR-UCR-n.r.

insinsinsinsins

Rusty-vented Canastero (A. dorbignyi)Puna Canastero (A. sclateri)Streak-fronted Thornbird (Phacellodomus striaticeps)Puna Tapaculo (Scytalopus simonsi)White-throated Tyrannulet (Mecocerculus leucophrysTufted Tit-Tyrant (Anairetes parulus)Rufous-webbed Bush-Tyrant (Polioxolmis rufipennis)D’Orbigny’s Chat-Tyrant (Ochthoeca oenanthoides)White-browed Chat-Tyrant (O. leucophrys)Red-crested Cotinga (Ampelion rubrocristatus)

CFFCURF-URFn.r.

C-FUC-FFR-FUC-UFRU-n.r.

FRUUFU-FU-RFRn.r.

FFUUFUUFUn.r.

CFFF-CFRFUF-Cn.r.

insinsinsinsinsinsinsinsinsfru

Andean Swallow (Haplochelidon andecola)Chiguanco Thrush (Turdus chiguanco)Giant Conebill (Oreomanes fraseri)Gray-bellied Flowerpiercer (Diglossa carbonaria)

n.r.UFF

n.r.C-FFn.r.-U

R-UU-FFR

RFFU

R-UF-CFF-C

insfruinsnec

Common Bush-Tanager (Chlorospingus ophthalmicus)Rufous-collared Sparrow (Zonotrichia capensis)Ash-breasted Sierra-Finch (Phrygilus plebejus)Rufous-bellied Saltator (Saltator rufiventris)Thick-billed Siskin (Carduelis crassirostris)Hooded Siskin (C. magellanica)

n.r.A-CA-Cn.r.U-FC

n.r.C-FCRRU-n.r.

n.r.U-RCRn.r.n.r.

n.r.FAn.r.FR

RAAn.r.U-Fn.r.-F

frufrufrufrufrufru

Tawny Tit-Spinetail and the Giant Conebillhad constant abundance year round (Table 1).The Rusty-vented Canastero and the Streak-fronted Thornbird showed slight variation, be-ing least abundant during the austral winter,whereas the D’Orbigny’s Chat-Tyrant had itslowest abundance during December–January,and the Rufous-webbed Bush-Tyrant lacked aclear pattern. The Thick-billed Siskin was notrecorded during June–July and peaked inabundance from August through January. The

Rufous-bellied Saltator was recorded onlyduring April–May and June–July when it wasrare.

Noteworthy distributional record.—The oc-currence of the White-sided Hillstar (Oreotro-chilus leucopleurus) at Sacha Loma during theaustral winter (Table 1) represents a northwardrange extension by about 400 km for this An-dean species, which breeds mainly in Argen-tina and Chile and previously was known onlyfrom as far north as the department of Tarija


FIG. 1. Insectivores were the most temporally sta-ble guild in the Polylepis bird community at SachaLoma, central Bolivia, from November 1999 to No-vember 2000, both in terms of (A) species richnessand (B) relative abundance. No data were collected inFebruary and March.

TABLE 2. Frugi-granivores and nectarivores combined had a significantly low abundance in mid-winter(June–July) in a Polylepis woodland at Sacha Loma, central Bolivia, November 1999 to November 2000, asrevealed by binomial sign tests comparing relative abundance scores between 2-month periods. This is reflectedat the community level, whereas insectivores lacked significant variation. No data were collected in Februaryand March.

Comparison

Insectivores (n 5 16)

Z P

Frugi-granivores andnectarivores (n 5 18)

Z P

All species (n 5 35)

Z P

Dec–Jan to April–MayApril–May to June–JulyJune–July to Aug–SeptAug–Sept to Oct–NovOct–Nov to Dec–Jan

0.00.600.01.581.06

1.0000.5461.0000.1140.289

0.292.412.021.031.58

0.7730.0160.0430.3020.114

0.202.352.121.542.01

0.8380.0190.0340.1240.044

in southern Bolivia (Fjeldsa and Krabbe 1990,Arribas et al. 1995, Fjeldsa 1999). We firstmist netted a male on 9 June 2000 in a nearbyPolylepis patch, followed by a capture of amale on 7 August 2000 and subsequent ob-servations until 16 October 2000 in the studyfragment. Males of the White-sided Hillstardiffer from those of the sympatric central Bo-livian subspecies of the Andean Hillstar (O.estella boliviana) primarily by their blue-black rather than chestnut median stripe on theunderparts (Fjeldsa and Krabbe 1990).

These observations represent the first evi-dence that the White-sided Hillstar is an aus-tral migrant. Although Marın et al. (1989)suggested that Chilean birds might migratebased on a single record in southern Antofa-gasta, de la Pena (1994) and Fjeldsa (1999)mentioned only altitudinal movements for thespecies, and Parker et al. (1996) did not listthe White-sided Hillstar as an austral migrant.

DISCUSSION

The insectivore guild was the most tempo-rally stable at Sacha Loma both in terms ofspecies richness and abundance, and we foundno significant changes in abundance betweenseasons. In contrast, frugi-granivores and nec-tarivores combined showed considerable sea-sonal variation with a significant abundancelow during the austral winter (June–July).Thus, our results indicate that Karr’s (1976)hypothesis of highest seasonality in insecti-vores, which was derived from studies in trop-ical lowlands, does not apply to marginallytropical, high-elevation Polylepis forests. Sea-sonal variation at the community level largelyreflected that of frugi-granivores and nectari-


vores, i.e., the two guilds dependent uponplants as food resources. The Sacha Lomabird community may be characterized as high-ly seasonal since only about one quarter of itscore species were year-round residents withfairly constant abundances.

Relative abundance values do not necessar-ily reflect true abundances of species, but maybe influenced by interspecific differences andseasonal changes in detectability. In the Bo-livian Andes, vocal activity (and thus detect-ability) levels of forest birds tend to decreaseconsiderably during the late rainy and firsthalf of the dry season (SKH pers. obs.).Therefore, the question arises to what degreeour data are biased by varying detectabilitiesof species. The relatively open vegetationstructure of the Polylepis forest certainly re-duced such biases, but even so, visually in-conspicuous species that prefer to stay indense vegetation or close to the ground, i.e.,Ornate Tinamou (Nothoprocta ornata), Cor-dilleran (Asthenes modesta) and Puna (A. scla-teri) canasteros, Puna Tapaculo (Scytalopus si-monsi), and White-browed Chat-Tyrant (O.leucophrys), probably became less detectableduring the late rainy season. The apparentlylow abundance of the Puna Tapaculo, a highlysedentary and almost flightless species (Fjeld-sa and Krabbe 1990), during the austral winteralmost certainly is a result of reduced vocalactivity during that season. However, exceptfor the Ornate Tinamou, these visually crypticspecies are insectivores. Thus, if seasonalchanges in detectability biased our data, onewould expect to find pronounced seasonal var-iation in abundance for insectivores ratherthan for frugi-granivores and nectarivores,which was not the case in Sacha Loma. Infact, the true seasonality in insectivores com-pared to frugi-granivores and nectarivorestherefore is likely to be even less than whatFig. 1 indicates.

Several studies have shown that variationsin abundance of birds over time are related tothe availability of food resources (Karr 1976,Blake and Loiselle 1991, Loiselle and Blake1991, Levey and Stiles 1992, Strong andJohnson 2001). Although we lack quantitativedata on food supply, some comparisons canbe made based on qualitative observations.Flowering and seeding of Polylepis trees, forexample, is strictly seasonal. The first Poly-

lepis trees had open flowers during July, andmost trees were flowering during early Au-gust. Seeds still were present during late Jan-uary and extended into February, but nonewere observed during mid-April. This patternlargely coincided with the abundance fluctu-ations of the Thick-billed Siskin, a habitatspecialist that mainly feeds on flowers andseeds of Polylepis (Fjeldsa and Krabbe 1990,Herzog et al. 2001). Other common woodyplants such as Berberis (Berberidaceae) alsostarted flowering during August, producingfruits and seeds during the rainy season.

Nectarivores primarily were observed at theflowers of two species, the annual vine Sal-pichroa glandulosa (Solanaceae) and the mis-tletoe Tristerix penduliflorus (Loranthaceae), ahemiparasite on Polylepis trees, although theyalso used other, less conspicuous flowers.Whereas S. glandulosa flowering was restrict-ed to the rainy season, mistletoe flowers (andfruit) were available year round, but had theirlowest abundance during the middle of theaustral winter when species richness andabundance of nectarivores also was lowest. Itis noteworthy that the Gray-bellied Flower-piercer (Diglossa carbonaria) was the onlynectarivore with year-round presence at SachaLoma. This possibly was due to the feedingmethod of flowerpiercers, resulting in lessspecialization and allowing them to use agreater variety of flowers than any of the hum-mingbird species (Naoki 1998). Alternatively,D. carbonaria may have shifted to a more in-sectivorous diet during the dry season, giventhat Isler and Isler (1999) reported insect partsas stomach contents for this species.

Although the insectivore guild did not showsignificant abundance fluctuations, some in-sectivorous species nonetheless changed inabundance possibly in response to a variablefood supply. The Bar-winged Cinclodes (Cin-clodes fuscus) mostly forages along streams,on wet bogs or grassy glades, often in Poly-lepis woodlands (Fjeldsa and Krabbe 1990,Fjeldsa and Kessler 1996). This species wascommon during the rainy season whenephemeral aquatic microhabitats formed fre-quently, both in the Polylepis woodland andin the surrounding puna grassland, whereas itwas rare during the dry season when evensome springs near Sacha Loma dried up. Thispattern coincides with anecdotal observations


of drastic body mass decrease in two bandedC. fuscus individuals. When banded at the be-ginning of the dry season on 10 May 2000,the first bird had a body mass of 27.0 g andshowed no traces of subcutaneous fat. Thesame bird was recaptured during the height ofthe dry season on 5 August 2000 weighing20.5 g, 74% of its body mass in May. A sec-ond bird decreased in body mass from 31.5 g(again without signs of subcutaneous fat de-posits) on 28 January 2000 to 27.0 g on 13October 2000, suggesting that it reached aminimum during the dry season. Although wecan not dismiss the possibility that those bodymass losses were caused by other factors (e.g.,a heavy parasite load), the most plausible ex-planation is a strong reduction in food supplyand optimal foraging habitat during the dryseason.

Of the eight Polylepis specialists recordedat Sacha Loma, six species showed variationsin abundance, two of which left the fragmentduring part of the year. Those six speciesprobably depend upon a certain degree of con-nectivity between suitable habitat patches, es-pecially when inhabiting small fragments withmore homogeneous environmental conditionsthan larger patches that cover greater eleva-tional gradients. The need for habitat connec-tivity also might apply to some extent to theGiant Conebill, which, despite constant abun-dance during the study period, exhibitedmovements between fragments at Sacha Loma(Herzog et al. 2002b). Because anthropogenicfragmentation and deforestation of Polylepiswoodlands due to excessive burning and graz-ing are ongoing (Fjeldsa and Kessler 1996,Fjeldsa 2002, Hensen 2002), studies on themetapopulation dynamics of Polylepis spe-cialists are urgently needed to determine howthey respond to such disturbance.

ACKNOWLEDGMENTS

We thank the people of Sacha Loma for allowing usto work on their land, and J. Balderrama, J. Cahill, G.Duran, L. Lens, D. Mendez, O. Ruiz, A. Troncoso, andR. Vargas for assistance during field work. Three anon-ymous reviewers and J. A. Smallwood made valuablecomments on the manuscript. This study was support-ed by the Institutional University Cooperation (IUC)program between the Flemish Universities and theUniv. Mayor de San Simon.

LITERATURE CITED

ARRIBAS, M. A., L. JAMMES, AND F. SAGOT. 1995. Listade las aves de Bolivia. Armonıa, Santa Cruz de laSierra, Bolivia.

BLAKE, J. G. AND B. A. LOISELLE. 1991. Variation inresource abundance affects capture rates of birdsin three lowland habitats in Costa Rica. Auk 108:114–130.

BLAKE, J. G., F. G. STILES, AND B. A. LOISELLE. 1990.Birds of La Selva: habitat use, trophic composi-tion, and migrants. Pp. 161–182 in Four Neotrop-ical rain forests (A. Gentry, Ed.). Yale Univ. Press,New Haven, Connecticut.

BROOKS, D. M. 1997. Avian seasonality at a localityin the central Paraguayan Chaco. Hornero 14:193–203.

CYTEL SOFTWARE CORP. 1995. StatXact, ver. 3.0. CytelSoftware Corp., Cambridge, Massachusetts.

DAVIS, S. E. 1993. Seasonal status, relative abundance,and behavior of the birds of Concepcion, Depar-tamento Santa Cruz, Bolivia. Fieldiana Zool. 71:1–33.

DE LA PENA, M. R. 1994. Guıa de aves argentinas, vol.3: Columbiformes a Piciformes. Literature of Lat-in America, Buenos Aires, Argentina.

DEVELEY, P. F. AND C. A. PERES. 2000. Resource sea-sonality and the structure of mixed species birdflocks in a coastal Atlantic forest of southeasternBrazil. J. Trop. Ecol. 16:33–53.

FERNANDEZ, M., M. MERCADO, S. ARRAZOLA, AND E.MARTıNEZ. 2001. Estructura y composicion flor-ıstica de un fragmento boscoso de Polylepis bes-seri Hieron. subsp. besseri in Sacha Loma (Co-chabamba). Rev. Bol. Ecol. 9:15–27.

FJELDSA, J. 1991. The activity of birds during snow-storms in high-elevation woodlands in Peru. Bull.Brit. Orn. Club 111:4–11.

FJELDSA, J. 1993. The avifauna of the Polylepis wood-lands of the Andean highlands: the efficiency ofbasing conservation priorities on patterns of en-demism. Bird Conserv. Int. 3:37–55.

FJELDSA, J. 1999. White-sided Hillstar. Pp. 624 inHandbook of the birds of the world, vol. 5: barn-owls to hummingbirds (J. del Hoyo, A. Elliot, andJ. Sargatal, Eds.). Lynx Edicions, Barcelona,Spain.

FJELDSA, J. 2002. Polylepis forests–vestiges of a van-ishing ecosystem in the Andes. Ecotropica 8:111–123.

FJELDSA, J. AND M. KESSLER. 1996. Conserving the bi-ological diversity of Polylepis woodlands of thehighlands of Peru and Bolivia: a contribution tosustainable natural resource management. NOR-DECO, Copenhagen, Denmark.

FJELDSA, J. AND N. KRABBE. 1990. Birds of the highAndes. Apollo Books, Svendborg, Denmark.

HENSEN, I. 2002. Impacts of anthropogenic activity onthe vegetation of Polylepis woodlands in the re-gion of Cochabamba, Bolivia. Ecotropica 8:183–203.


HERZOG, S. K., M. KESSLER, AND T. M. CAHILL. 2002a.Estimating species richness of Neotropical birdcommunities from rapid assessment data. Auk119:749–769.

HERZOG, S. K., R. SORIA A., A. TRONCOSO J., AND J.CAHILL M. 2001. A Bolivian breeding range ex-tension and field identification aids for Thick-billed Siskin Carduelis crassirostris. Cotinga 16:101–102.

HERZOG, S. K., R. SORIA A., A. TRONCOSO J., AND E.MATTHYSEN. 2002b. Composition and spatialstructure of avian mixed-species flocks in a high-Andean Polylepis forest in Bolivia. Ecotropica 8:133–143.

HILTY, S. L. 1997. Seasonal distribution of birds at acloud-forest locality, the Anchicaya Valley, inwestern Colombia. Ornithol. Monogr. 48:321–343.

ISLER, M. L. AND P. R. ISLER. 1999. Tanagers. A & CBlack, London, United Kingdom.

KARR, J. R. 1976. Seasonality, resource availability,and community diversity in tropical bird com-munities. Am. Nat. 110:973–994.

KARR, J. R. AND K. E. FREEMARK. 1983. Habitat selec-tion and environmental gradients: dynamics in the‘‘stable’’ tropics. Ecology 64:1481–1494.

KESSLER, M. 2002. The ‘‘Polylepis problem’’: wheredo we stand? Ecotropica 8:97–110.

LEVEY, D. J. AND F. G. STILES. 1992. Evolutionary pre-cursors of long-distance migration: resource avail-ability and movement patterns in Neotropicallandbirds. Am. Nat. 140:447–476.

LOISELLE, B. A. 1988. Bird abundance and seasonalityin a Costa Rican lowland forest canopy. Condor90:761–772.

LOISELLE, B. A. AND J. G. BLAKE. 1991. Resourceabundance and temporal variation in fruit eatingbirds along a wet forest elevational gradient inCosta Rica. Ecology 72:180–193.

MARANTZ, C. A. AND J. V. REMSEN, JR. 1991. Seasonaldistribution of the Slaty Elaenia, a little-knownaustral migrant of South America. J. Field Orni-thol. 62:162–172.

MARıN, M., L. F. KIFF, AND L. PENA G. 1989. Noteson Chilean birds with descriptions of two newsubspecies. Bull. Brit. Orn. Club 109:66–82.

NAOKI, K. 1998. Seasonal change of flower use by theSlaty Flowerpiercer (Diglossa plumbea). WilsonBull. 110:393–397.

PARKER, T. A., III, D. F. STOTZ, AND J. W. FITZPATRICK.1996. Ecological and distributional databases. Pp.131–436 in Neotropical birds: ecology and con-servation (D. F. Stotz, J. W. Fitzpatrick, T. A.Parker, III, and D. K. Moskovits, Eds.). Univ. ofChicago Press, Chicago, Illinois.

REMSEN, J. V., JR. 1994. Use and misuse of bird listsin community ecology and conservation. Auk111:225–227.

REMSEN, J. V., JR. AND D. A. GOOD. 1996. Misuse ofdata from mist-net captures to assess relativeabundance in bird populations. Auk 113:381–398.

SCHMITT, C. G., D. C. SCHMITT, AND J. V. REMSEN, JR.1997. Birds of the Tambo area, an arid valley inthe Bolivian Andes. Ornithol. Monogr. 48:701–716.

SOUTH AMERICAN CHECKLIST COMMITTEE. 2002. A clas-sification of bird species of South America, ver.9/13/02. http://www.museum.lsu.edu/;Remsen/SACCBaseline.html.

STATSOFT, INC. 1997. STATISTICA for Windows, ver.5.1. StatSoft Inc., Tulsa, Oklahoma.

STILES, G. F. 1980. The annual cycle in a tropical wetforest hummingbird community. Ibis 122:322–343.

STILES, G. F. 1985a. Seasonal patterns and coevolutionin the hummingbird-flower community of a CostaRican subtropical forest. Ornithol. Monogr. 36:757–787.

STILES, G. F. 1985b. Conservation of forest birds inCosta Rica: problems and perspectives. Pp. 141–168 in Conservation of tropical forest birds (A.W. Diamond and T. Lovejoy, Eds.). InternationalCouncil for Bird Preservation, Cambridge, UnitedKingdom.

STILES, F. G. 1988. Altitudinal movements of birds onthe Caribbean slope of Costa Rica: implicationsfor conservation. Mem. Calif. Acad. Sci. 12:243–258.

STRONG, A. M. AND M. D. JOHNSON. 2001. Exploitationof a seasonal resource by nonbreeding Plain andWhite-crowned pigeons: implications for conser-vation of tropical forests. Wilson Bull. 113:73–77.

APPENDIX

Differentiating between core species andvisitors or vagrants is not always a straight-forward process and may be disputable insome cases. Thus, we present a commentedlist of all species observed in the forest, di-rectly at its edge, or flying over that were notconsidered core species of the Polylepis frag-ment at Sacha Loma, central Bolivia. Speciesexclusively observed in the surrounding punagrassland are not included.

Darwin’s Nothura (Nothura darwinii). Ob-served only once during October 2000, whenseveral birds were heard at forest edge.

Turkey Vulture (Cathartes aura). Singlebirds flying over the forest and puna duringApril, May, and August 2000.

Andean Condor (Vultur gryphus). Raresightings of soaring birds during April, May,and July 2000.

Black-chested Buzzard-Eagle (Geranoaetusmelanoleucus). Observed only during August2000 in flight.

Red-backed Hawk (Buteo polyosoma). Seenin small numbers during most months in flight


or on ground in puna grassland. This speciespotentially may hunt in the forest fragment,but we never observed this behavior.

Mountain Caracara (Phalcoboenus megal-opterus). This common puna species was ob-served in flight and on the ground in punaduring most months.

Aplomado Falcon (Falco femoralis). Weobserved two birds hunting a passerine overpuna during December 1999, and single birdswere seen once each during May, June, July,and August 2000 either perched on top of aPolylepis tree or hovering over the fragment.This species may very occasionally prey on aPolylepis bird, but this alone does not warrantinclusion as a core species.

Blue-crowned Parakeet (Aratinga acuticau-data). We saw a flock of eight birds once dur-ing October 2000 in fast flight over the frag-ment.

Mitred Parakeet (Aratinga mitrata). We sawa flock of 8–10 birds once during December1999 in fast flight over the fragment.

Mountain Parakeet (Psilopsiagon aurif-rons). Small flocks in fast flight observed dur-ing May, July, August, and October 2000, butnever observed perched in the forest.

White-collared Swift (Streptoprocne zon-aris). A single bird circling relatively highabove the forest in a flock together with thetwo Aeronautes species below during April2000.

Andean Swift (Aeronautes andecolus). SeeWhite-tipped Swift (below).

White-tipped Swift (Aeronautes montiva-gus). A flock of about 50 birds circling to-gether with two White-collared Swifts andabout 50 Andean Swifts relatively high abovethe forest during April 2000.

White-vented Violet-ear (Colibri serriros-tris). A single bird in pale juvenal plumageseen once during November 1999.

Great Sapphirewing (Pterophanes cyanop-terus). Single birds seen twice in the forestduring October 2000. Probably a rare visitorfrom the humid treeline zone.

Green-barred Woodpecker (Colaptes me-lanochloros). Two single birds observed dur-

ing October 2000. Probably a rare visitor fromdeciduous forest at lower elevations.

Andean Flicker (Colaptes rupicola). Un-common to common almost year-round (notrecorded during June 2000) in the surroundingpuna grassland, we observed this species onlytwice in the fragment, once at the edge (a cop-ulating pair during November 1999) and oncein the boulder field.

Rufous Casiornis (Casiornis rufa). We mistnetted a single bird during August 2000. Prob-ably a vagrant from deciduous forest at lowerelevations.

Fawn-breasted Tanager (Pipraeidea melan-onota). We mist netted a single bird duringAugust 2000 and observed another during No-vember 2000. Probably a vagrant.

Cinereous Conebill (Conirostrum ciner-eum). Two single birds seen during October2000. This species regularly inhabits Polylepisforests and may form part of the core speciescommunity in some years.

Plumbeous Sierra-Finch (Phrygilus unicol-or). We mist netted a single bird during April2000. Surprisingly, we did not observe thisspecies in the puna grassland around our studysite.

Rufous-sided Warbling-Finch (Poospiza hy-pochondria). Sightings of one (dispersing?)bird in juvenal plumage during May 2000 andof two individuals in adult plumage duringOctober 2000. This species regularly inhabitsPolylepis forests and may form part of thecore species community in some years.

Plain-colored Seedeater (Catamenia inor-nata). We mist netted a single bird during De-cember 1999. This species regularly inhabitsPolylepis forests and may form part of thecore species community in some years.

Golden-billed Saltator (Saltator aurantii-rostris). A single bird heard once during No-vember 1999. Probably a rare visitor from de-ciduous scrub and gardens near houses at low-er elevations.

Black Siskin (Carduelis atrata). Four birdsmoving through scattered Polylepis bushes atthe fragment edge during April 2000. Proba-bly a rare visitor.

Documents

Herzog Et Al 2003. Seasonal Variation in Avian Community Composition in a High-Andean Polylepis (Rosaceae) Forest Fragment