j. RaptorRes. 37(1):19-30 ¸ 2003 The Raptor ResearchFoundation, Inc.

ABUNDANCE OF SOARING RAPTORS IN THE BRAZILIAN ATLANTIC RAINFOREST

SANTI MAlqOSA,1 EDUARDO MATEOS, AND VITTORIO PEDROCCHI Departamentde Biolo•a , Universitatde Barcelona, Facultat de Biologia, Aving•tda Diagonal, 645, 08028 Barcelona,Catalonia, Spain

ABSTRACT.--I8 August-4 September 1998, we conducted 23 3-4 hr point-countsin an Atlantic rainlbrest area of southeastern to evaluate the richness and relative abundance of raptors in two adjacent protected areas,Parque EstadualIntervales and the Parque EstadualTuristico do Alto Ribeira. During 88.2 hr, we recorded 334 contactswith raptors,involving 734 individualsof nine .Gontacts per hour and the number of speciestallied showed that the counts were higher between 0900-1200 H (Local Standard Time), and that counts of 3 hr were the most cost effkctive. Reasonableprecision for abundance indices was achieved with samples sizes of 20-30 points, but samples of 12 should give satisfactoryresults for the more common species,as long as counting points are distributedsufficiently in space.We derived abundanceindices for speciesof raptorsmost commonlyseen in the area. In 14 3-hr counts,Black Vultures (Cora•ps atratus)were observedin 100% of them, Mantled Hawks (Leucop- ternispolionota) in 71%, BlackHawk- (Spizaetus tyrannu•') in 50%, Turkey Vultures (Cathartesaura) in 29%, Ornate Hawk-Eagles(Spizaetus ornatus) in 21%, RoadsideHawks (Buteomagnirostris) in 14%, Short-lailedHawks (Buteobrachyurus) in 14%, Crested Carataras (Polyborusplancus) in 14%, and Tiny Hawks (Acdpitersuperciliosus) in 7%. Bat Falcons(Falco rufigularis) and White-tailedHawks (Buteoalbi- caudatus)also were reported in the area, but outside the counting periods.

KEYWORDS: Brazilianrainforest;, point count;richness; Sdo Paulo State.

ABUNDANCIA DE RAPACESPLANEADORAS EN LA SELVA ATL•NTICA BRASILElqA R•SUM•N.--Entre el 18 de agosto-4 de septiembre de 1998, realizamos23 tensos puntales de 3-4 hr de duraci6n en la selvaAtl•tntica del Brasil, para estimar la riqueza y abundancia de avesde presa en dos zonasprotegidas adyacentes, el Parque EstadualIntervales y el Parque EstadualTuristico do Alto Ribeira. Tras 88.2 hr de tenso se obtuvieron 334 observacionesde rapacescorrespondientes a 734 individuosde 9 especies.E1 nfimero de contactosy de especiesobservadas indican que la mejor hora para realizar los tensos se sitfia entre las 0900-1200 H (Hora Local Time) y que su duraci6n 6ptima es de 3 hr. Una buena precisi6n en las estimasde abundanciase obtiene a partit de ta•nafiosmuestrales de 20-30 puntos,pero muestrasde 12 puntospueden ofrecer resultadossatisfactorios para las especies m•tslkecuentes, siempre y cuandolos puntosse encuentrenbien repartidospot el area de estudio.En catoroe tensos de trcs horas 11evadosa cabo entre las 0900-1200 H, el Zopilote Negro (Coragypsatratus) se observ6 en el 100% de los tensos,el BusardoBlanquinegro (Leucopternispolionora) en el 71%, el Aguila-AzorNegra (Spizaetustyrannus) en el 50%, el Aura Gallipavo(Cathartes aura) en el 29%, el Aguila- Azor Galana (Spizaetusornatus) en el 21%, el BusardoCaminero (Buteomagnirostris), el Busardo Colicorto (Buteobrachyurus), el CaracaraCarancho (Polyborusplancus) en el 14%, y el GavilancitoAmericano (Accipitersuperciliosus) en el 7% los casos.El Halc6n Murcielaguero (Falcorufigularis) y el BusardoCol- iblanco (Buteoalbicaudatus) fueron observadosfuera de los periodos de tenso. [Traducci6n de los Autores]

The Atlantic rainfbrest of Brazil is one of the 1985, Albuquerque 1995, Myers et al. 2000). One most threatened biomes in the world. Only 2-8% of the best-preservedareas of Atlantic raintbrest •s of the 106 km 2 offbrest that once covered a narrow the Paranapiacaba tbrest fragment (140000 ha) stretch of land along the southeasterncoast of Bra- (Mateos et al. 2002). This fragment includes some zil remains in scattered,small tkagments (Fortseca areas of unprotected tbrest and tbur protected ar- eas (Parque Estadual Turfstico do Alto Ribeira, E-mail address: [email protected] Parque EstadualIntervales, Parque Estadual Garlos

19 20 MAF4OS^ET At,. VOL. 37, No. 1

Botelho, and Estacao Eco16gicaXitu•) known as accuracyof the results and efficiency of the count- the ParanapiacabaEcological Continuum (Pisciot- ing eftbrt, the objectivesof our researchwere (1) ta 2002), which, all together,form one of the larg- to evaluate the effect of several variables (time of est remaining patches of uninterrupted Atlantic day, duration of counts, number of counts, selec- rainforest (Fig. 1). tion of counting points) on the efficiency of the In spite of its fragmentation, the Atlantic rain- point-count method in estimating speciesrichness remains among the richestof all ornitholog- and abundance indices of raptor assemblagesin ical areas in the world (Cracraft 1985, Wege and the Atlantic rainforest, and (2) to derive the best Long 1995, Startersfieldet al. 1998) and is consid- estimateof abundance indices of raptors in a well- ered a hotspot for biodiversityconservation (Myers preserved area of Brazilian Atlantic rainforest. 1988, Bibby et al. 1992b, Myers et al. 2000). The MATERIAI, AND METHODS area givesrefuge to many endemic (Myers et al. 2000), among which are severalspecies and sub- Study Areas. Counts were conducted in two adjacent areas of the Parque Estadual Turlstico do Alto Ribeira speciesof raptors such as Grey-headedKites (Lep- (Petar) and the Parque Estadual Intervales (Intervales), todoncayanensis), White-necked Hawks ( within the Paranapiacaba forest fragment in Serra do lacernulata),Mantled Hawks (Leucopternispoliono- Mar mountain range, in the state of S5o Paulo, Brazil ra), and Black Hawk-Eagles( Spizaetust. tyrannus) (Fig. 1). The Serra do Mar extends parallel to the Atlan- (Collar et al. 1992, del Hoyo et al. 1994, Bierre- tic coastfor over 900 kin. Its slopesrise to abrupt moun- tain peaks between 800-1100 m in elevation. The Par- gaard 1998, Bildstein et al. 1998). This justifies the anapiacaba forest fragment (Pisciotta 2002) consistsof need for scientific and conservation initiatives to four legally-protected,neighboring reserves,known as protect this important element of the global bio- the ParanapiacabaEcological Continuum (1258 km2), diversity (Burnham et al. 1994). plus some adjacent private forest areas. Together; they comprisesome 1400 km2 of uninterruptedforest in sev- Raptors usually live at low densitiesand are dif- eral ecologicalsuccessional stages (Fig. 1). Most of the ficult to detect, so that the methods usually em- Paranapiacaba forest fragment is covered by Low Eleva- ployed to evaluategeneral populations are not tion South Hillside Atlantic rainforest (Guix 2002), typi- adequate (Forsman and Solohen 1984, Thiollay cal of the altitudinal range between 50-100 m and 1200- 1989, Bibby et al. 1992a). Accurate speciesrichness 1600 masl, and tree height ranging from 20-30 m. Some surrounding and marginal areas (near 10% of the total and abundance estimatesrequire surveyinglarge area) is planted Araucariaangustifolia, Pinus sp. and Eu- areas and conducting a large number of counts. calyptussp. forest, banana plantations, and pastures.Four The dense structure of the rainforest adds even main vegetation types can be distinguishedin the Par- more difficulty to the studyof theseraptors (Thiol- anapiacaba forest fragment: mature ,which have been subject to no episode of deforestation or intensive lay 1989), and a great deal of time and effort are selectiveextraction, or only a single one more than 80- often required to prepare and to conduct the 100 yr ago; late-secondaryforests, which suffered the last counts (Whiracre and Turley 1990, Mafiosa and episode of deforestation or intensive selectiveextraction Pedrocchi 1997). 50-80 yr ago;young-secondary forests, which sufferedthe For these reasons, the conservation status of lastdeforestation or selectiveextraction episode 20-40 yr ago; and "capoeiras"or shrublands,which are forestar- many woodland raptors in the neotropics is poorly eas that have undergone the last detbrestationepisodes known (Albuquerque 1986, Thiollay 1994), in spite just 5-15 yr ago. The studyarea is coveredby mature and of the fact that more than half of the speciesof secondaryforest types. In some zones of shallowcalcar- neotropical raptors are endangered (Thiollay eous soil in the Petar site, mature or late-secondaryfo•= ests are lower in stature than those of the same succes- 1985, Bildstein et al. 1998). Several research and sional stagesin Intervales. monitoring programs are currently providing new The studyarea in Intervales was the main valley of the information about raptors in tropical •brests Formoso-Pi16esRiver, ti'om just above Basedo Carmo to (Thiollay 1989, Vannini 1989, Whitacre and Thor- a ikw kilometers beyond Base do Alecrim (Fig. 1). The strom 1992), but only preliminary research has Atlantic rainforest in this valley is mature or late-second- ary, with a mean height of 18 m and emergent trees been conducted in the Atlantic rain,brest (Willis reaching 24-30 m. Pahnito (Euterpeedulis) is abundant and Oniki 1981, Guix et al. 1992, Mateos and Marl- but becomes scarcer at the bottom of the valley,where osa 1996, Mafiosa and Pedrocchi 1997, Veilliard taquara (Merostachyssp.) is widespread. On mountain and Silva 2001, Mafiosa et al. 2002). sidesand summitsand facing north northwest,the Atlan- tic rainforestis most intact and there is hardly any ta- Considering the need of improving our knowl- quaral present.On the highest areasof the valley,near edge on raptor assemblagesin Atlantic rainforest Base do Carmo, human influence is minimal or nonex- areas, as well as the need of achieving maximum istent: mature or old-secondaryibrests cover the region. MARCH 2003 SOARiNG RAPTORS IN ATIANTIC RAINFOREST 21

! • '•'-( if----• L... Paranapiacaba

24ø37'W 48 ø 43' S i--Main AccessRøad

N 0 2 4 6 8 10km Figure 1. Location of the studyarea in Brazil. The limits of the four protected areas conforming the Ecological Continuum of the Paranapiacabaforest fragment within the S•o Paulo State are indicated. (A: Parque Estadual Turistico do Alto Ribeira; B: Parque Estadual,Intervales; C: Parque Estadual,Carlos Botelho;D: Estag•oEco16g•ca Xitu6). The study area is outlined and enlarged, showingthe location of the 23 counting points indicated by the numbers. 22 MA•osA ET AL. VOL. 37, NO. 1

"Capoeiras" occur only along the main road and trails, Evaluation of Biases. The effect of the time of day due to the extraction of palmfro palms and wood. The (morning only) on the results of the counts was evalu- valley bottom near Alecrim has been inhabited and mod- ated by analyzing the number of speciesreported at sev- ified for centuries. Late-secondaryforest dominates this eral hourly intervals. For speciesin which the number o1 region, with a scatteringof small banana plantations.Ta- observationswas large (Manded Hawk, Black Vulture, quaral areas around Alecrim are the remains of opened Turkey Vulture, and Black Hawk-), an hourly re- areas used for subsistencefarming 15-25 yr ago. porting pattern was also obtained. An index of activity The study area in Parque Estadual Turfsrico do Alto was computed for each hourly interval and species.Each Rabeirawas adjacent to the village of Nilcleo Caboclos hourly period of observationwas divided into twelve 5- (F•g. 1). The Nilcleo Caboclosis 600 masl and is covered rain intervals. During each interval, we recorded the by typical Hillside Atlantic rainforestvegetation (Carval- number of individuals in view. The activity index for a hoet al. 2002) with a high botanical diversity.This high given species during a specific hourly interval was the diversity is caused by the existence of calcareoussedi- product of the proportion of 5-min intervals in which the ments that enhance the growth of fbrestswith a partic- specieswas in view during that period (from 0/12-12/ ular fioristic composition. Secondary vegetation growsin 12) by the addition of all individualsrecorded during the the marginsof dirt tracksand trails;it is characterizedby twelve 5-rain intervals. taquaral and shrublands. To assessthe optimal duration of counts, we analyzed Field Procedures. The point-count method (Whiracre how the number of detected speciesand abundance in- and Turley 1990, Whiracre et al. 1992a) wasused because dices estimateschanged in relation to count duration. We it was the most effective method in a preliminary survey limited this analysisto the counts initiated at 0900 H and in 1994 (Mafiosa and Pedrocchi 1997). Counts were con- that were extended to 1300 H (N = 9), so that the effect ducted during the dry season,between 18 August and 4 of sampling at different times of day was removed. September 1998, when some raptors (i.e., Manfled Hawk, The relationship between precision of estimates and Black Hawk-Eagle) are expected to be displayingin the sample sizewas modelled using bootstrappingtechniques area (Vielliard and Silva 2001). One or two observers (see above) by taking samplesof progressivelyincreasing stayedon high points of the landscapeor perched on the size (1-50). To evaluate how the precision of the abun- topsof trees,having a viewangle of 80-294ø and a range dance indices improved with sample size, the -+95% GI of view above the canopy of 1000-4000 m. We selected of these estimates were plotted against the number of 23 points along main tracksor footpaths (Fig. 1), which counts. To make comparisonsbetween parameters and we judged to be representativeof the dominant habitat speciespossible, precision was expressedas (upper 95% m the area. Seventeenof thesepoints were along 27 km GI - lower 95% GI)/mean. of the Formoso-Pi16esValley and six around the Caboclos To test how the selectionof lhe counting points could area. Each point was selected in order to give an inde- affect the results, we divided our initial set of 23 points pendent view of a part of the study area. The mean dis- into two sets of spatially alternating points, a set of 12 tancebetween neighboring points was 997 -+ 441 m (SD) (set A) and a set of 11 (set B) points. These two data in Intervales(range = 267-1493) and 1900 .+ 1665m in subsetswere used to derive two separate estimatesof the Petar (range = 600-5161). Even the closestpoints of- fered non-overlapping views of the study area. The relative-abundance parameters. Comparison between counts were initiated between 0800-1200 H (Local Stan- these two setsshould give some indication of the repro- dard Time, sunrise 0519 H), and lasted for 2.5-4 hr. The ducibility and accuracyof our results. mean length of the counts was 3.8 -+ 0.4 hr (SD). Each Computing Standard Abundance Estimates. Standard- point was sampled only once. The 23 counts totalled 88.2 ized abundance indices estimatesof raptors t¾r the study area as a whole and IBr each of the two Preserve Areas hr of observation, concentrated fi'om 0900-1300 H (72.1 hr). Counts were conducted in clear weather, except for were derived, considering ouly the 14 counts in the one that was conducted in light rain but yielded com- 0900-1200 H period. The rcmainiug nine counts were parable results.During each count interval, we recorded omitled becausethey lasted for lesstban three hours or every raptor in view to an unbounded distance. In this because they did not include the above-mentioned peri- manner, we obtained a list of the minimum number of od. •ndividualsand groups observed during every count. Computation of Relative Abundance Parameters. We RESUI•TS computed the following abundance indices lbr each spe- cies and area: percentage of counting points in which a We recorded 11 raptor speciesin the Paranapia- specieswas detected (percent presence .+95% Confi- caba forest fragment, two of which, the Bat Falcon dence Interval [GI]), mean group size (individuals +- (Falcorufi•ffularis) and the White-tailed Hawk ( SD), and contacts/hr (-+95% GI). Gonfidence intervals albicaudatus), were observed lbr the first time in were computed by bootstraptechniques, using the Resam- phng Statssuite (Bruce et al. 1995): a parameter value the area (Table 1). The 23 counts yielded 334 ob- wascomputed for each original counting point. This gave servations of raptors (3.8 contacts/hr) involving an initial sample of values, which was randomly resam- 734 individuals,belonging to nine speciesof rap- pled with replacement to obtain a sample size identical tors (Table 2). Only one of the 23 counts (4%) to the original one. This processwas repeated 10000 times. Then, the mean and confidence intervals of the resulted in no raptor observations, four yielded 10 000 parameter estimateswere obtained. one species,five yielded two species,four yielded MARCH 2003 SOARING RAPTORS IN ATLANTIC RAINFOREST 23

Table 1. Composition of the raptor assemblagein the Paranapiacabaforest fragment, according to the species recorded in severalsurveys to the area. "x" indicatesthe specieswas recorded outside the primary sampleperiods. Vagrant species,as well as migratoryspecies that are absentfrom the area in August,are excluded.

WILLIS AND VIELI.IARD MAI•OSA ONIKI AND SILVA ET ^I,. This (1981)a (2001)b (1997)c studya Black Vulture (Cora•ps atratus) 49 100% 50% 100% Turkey Vulture (Cathartesaura) 59% x 29% Grey-headedKite (Leptodon cayanensis) 9% >e Rufous-thighedKite ( Harpagusdiodon) 18% Grey-belliedGoshawk (Accipit• poliogast•) 14% Tiny Hawk (Accipit• sup•ciliosus) 25% 7% Sharp-shinnedHawk (Accipit• striatus) x Crane Hawk ( G•anospizaca•ulescens) x Mantied Hawk ( Leucopt•nispolionota) 2 86% 50% 71% Great Black-Hawk ( Buteogallusurubitinga) 4% Black-chestedBuzzard-Eagle (G•anoaetus melanoleucus) x RoadsideHawk ( Buteomagnirostris) 34 95% x 14% White-rumped Hawk ( Buteoleucorrhous) 2 x Short-tailed Hawk ( Buteot,'achyurus) 7 54% x 14% White-tailed Hawk ( Buteo albicaudatus) x Guiana Crested Eagle (Morphnusguianensis) x Black Hawk-Eagle ( Spizaetustyrannus) 68% x 50% Ornate Hawk-Eagle( Spizaetusornatus) 25% 21% Crested Caracara ( Polyborusplancus) 2 18% 14% Chimango Caracara (Milvago chimachima) 17 54% x Laughing Falcon ( Herpetoth•escachinnans) 5 x Barred Forest-Falcon( Micrasturruficollis) 10 77% CollaredForest-Falcon (Micrastur .semitorquatus) 4% x Bat Falcon (Falcorufigttlaris) a Individuals/100hr observation(N = 41.2 hr) reported in the CarlosBotelho area between24-28 Februaryand 5-10 July 1979 (Swallow-tailedKite [Elanoidesforficatus excluded] ). b Percentof 22 visitsto the Intervalesarea betweenAugust 1988 and December 1992 in which the specieswere reported. "x" indicates presenceduring other occasional visits to the area (Osprey[Pandion haliaetus], Swallow-tailed Kite, and PlumbeousKite [Ictiniaplumbea excluded] ). •'Percent of point counts(N = 4) in which the specieswas detected during a visit to Intervalesbetween 1 and 12 August1994. "x" indicates detection outside the counts. d Percentof point counts(N = 14) in which the specieswas detected during a visitto Intervalesand the Petar in August-September 1998. "x" indicates detection outside the counts. e Identification was uncertain. three species,six yielded four species,and three afternoon, whereas the activity of Black Hawk-Ea- yielded five species. gles tended to be maximum between 0900 and Effect of Time of Day. Only Mantled Hawksand 1200 H. Overall, the period 1000-1200 H em- RoadsideHawks (Buteomagnirostris) were observed braced the peak activityperiods of the three most before 0900 H. Although observation effort was common species,as well as all the observationsof minimal, only Black Vultures were consistentlyob- Short-tailed Hawks (Buteobrachyurus). A high per- served during the afternoon (1300-1600 H). The centage of the records of the lessdetectable spe- activityof this speciesremained more or lesscon- cies, such as Tiny Hawks (Accipitersuperciliosus, stantfrom 1000 H onwards(Fig. 2). Comparatively, 100%, N = 1), CrestedCaracaras (Polyborusplancus, the activity of Turkey Vultures was more concen- 50%, N = 2), and Ornate Hawk-Eagles(Spizaetus trated around mid-day.Activity of Mantled Hawks ornatus, 50%, N = 4) were concentrated between remained constantduring the morning and early 0900 and 1000 H. 24 lVI•OSA ET AL. VOL. 37, NO. 1

Table 2. Raptor abundance indices in Parque EstadualIntervales, Parque EstudualTur•stico do Alto Ribeira and pooled estimate.Means and _+95%Confidence Intervals are givenin parenthesis.Nis the number of counts.

INTERVALES PETAR POOLED (N = 9) (N = 5) (N = 14) Hours of observation 27 15 42 Number of species 8 8 9 Black Vulture Contacts/hr 2.5 (1.5-3.7) 3.3 (2.3-4.2) 2.8 (2.0-3.6) Percent presence 100 (100-100) 100 (100-100) 100 (100-100) Man fled Hawk Contacts/hr 0.7 (0.3-1.3) 0.8 (0.4-1.2) 0.8 (0.4-1.2) Percent presence 67 (33-100) 80 (40-100) 71 (50-93) Black Hawk-Eagle Contacts/hr 0.6 (0.1-1.2) 0.7 (0.1-1.4) 0.6 (0.2-1.1) Percent presence 44 (11-78) 60 (20-100) 50 (21-79) Turkey Vulture Contacts/hr a <0.1 (0.0-0.1) 0.5 (0.1-0.8) 0.2 (0.05-0.4) Percent presence 11 (0.00-33) 60 (20-100) 29 (7-57) Contacts/hr 0.2 (0.0-0.6) 0.1 (0.0-0.2) 0.1 (0.0-0.4) Percent presence 11 (0-33) 20 (0-60) 14 (0-36) Ornate Hawk-Eagle Contacts/hr <0.1 (0.0-0.1) 0.1 (0.0-0.3) 0.1 (0.0-0.1) Percent presence 11 (0-33) 40 (0-80) 21 (0-43) Short-tailed Hawk Contacts/hr 0.0 0.2 (0.0-0..5) 0.1 (0.0-0.2) Percent presence 0 40 (0-80) 14 (0-36) Crested Caracara Contacts/hr <0.1 (0.0-0.1) 0.1 (0.0-0.2) 0.05 (0.0-0.1) Percent presence 11 (0-33) 20 (0-60) 14 (0-36) Troy Hawk Contacts/hr <0.1 (0.0-0.1) 0.0 0.02 (0.0-0.1) Percent presence 11 (0-33) 0 7 (0-21) '• S•gnificant differences between areas at P < 0.0.5; Kruskal-Wallis test.

Effect of Count Duration. The number of spe- Effect of the Number of Counts. The number cies observed (+95% CI) tended to increase as the of speciesdetected increased quickly with the num- duration of the counts increased from 1 hr (1.4 + ber of counts. Some 20 counts were needed to de- 0.5 species)to 2 hr (2.7 -+ 0.9), but leveled off tect 90% of the total speciesdocumented, but the progressivelyfor 3 hr (3.0 _+0.9) and 4 hr counts rate of addition of new specieswas low with the (3.2 -+ 1.0). Among the four more frequent spe- addition of more counts (Fig. 4). Precision of cies, the cumulative percentage of counts levelled abundanceindices estimates quickly increased with off after 2-3 hr of sampling for the Black Vulture samplesize (Fig. 5). For the more frequent species and the Black Hawk-Eagle. For the Turkey Vulture (i.e., percent presence70-100%: Black Vultures, and the Mantied Hawk, this parameter increased Mantied Hawks), the precision of the abundance slightly, although not significantly, during the estimates falls below one +95% CI of the mean fourth hour of the count (Fig. 3). after 12 counts.For specieswith <30% presence M•CH 2003 SOARING RAPTORS IN ATLANTIC RAINFOREST 25

Black Vulture Turkey Vulture 14 Black Vulture 100 12 80 80 10 100 8 B 60 60 • 6 == 40 4 20 2 • 20 0 • 0 t i i T 4ø0i 8-9 9-10 10-11 11-12 12-13 13-14 1 2 3 4 1 2 ß-• 02 Turkey Vulture Mantied Hawk Black Hawk-eagle '• 100 100 o.1 80 g• 60 60 40 40 oo 20 20 8-9 9-10 10-11 11-12 12-13 13-14 0 0 i i i 1 6 1 2 3 4 1 2 3 4 Mantled Hawk 5 Durationof the Counts(hr)

3 2 Figure 3. Simulation of cumulativepercent presenceof 1 the most frequently-detected speciesin relation to the 0 duration of counts (hr). All counts began at 0900 H 8-9 9-10 10-11 11-12 12-13 13-14 Computations are based on the same nine point counts Black Hawk-eagle

timates from the overall sample, but were not s•g- nificantly different from one another, nor from the pooled estimate. 8-9 9-10 10-11 11-12 12-13 13-14 Composition of the Raptor Assemblage.During Hour Interval the fourteen standardized counts conducted be- Figure 2. Hourly variation (Local Standard Time) of tween 0900 H and 1200 H, nine raptor species the activity index of the most common speciesin the were detected (Table 2) of the 24 that have been studyarea (both sitescombined). For a given hour pe- reported in the area in this and in previoussurveys riod, the activityindex for each specieswas computed as (Willis and Oniki 1981, Mafiosa and Pedrocchi the product of the proportion of 5-min intervalsin which 1997, Vielliard and Silva2001; Table 1). Four spe- the specieswas in view by the total number of individuals observed. Sunrise = 0519 Local Standard Time. cies are reported in the four lists, four speciesin three, six speciesin two, and 10 speciesin only one. Of the 15 potential specieswhich were missed in our point counts, only five had been reported (i.e., Turkey Vulture), more than 50 samplesare needed to reach such a precisionvalue. However, the rate of increase in precision of parameter es- timates is very low after 10-20 counts.At any given • 9 sample size, percent presence was slightly more precisethan contacts/hr. 7 Reproducibility of Restfits. For the data set in- • 6 cluding all the counting points the mean distance between neighboring points (SD) was 1.2 --- 1.0 km (range = 0.3-5.2 km). For the two subsetsof alter- nating points, these distancesincreased to 2.4 + 1.3 km (range = 0.8-6.0 kin) for subsetA and 2.4 +1 + 1.6 km (range = 0.5-6.45 km) for subsetB. Only the three most common specieswere detected in co 0 5 10 15 20 25 30 35 40 both subsetsthat, because of small sample size, yieldedonly five and six species,respectively. Abun- Number of Counts dance indices estimatesbased on these separate Figure 4. Simulation of the cmnulative number of spe- data sets had larger confidence intervals than es- cies detected in relation to the number of counts. 26 MA•qOSAET AL. VOL. 37, NO. 1

, I Contacts/hour BlackVulture though the difference was only significantfor the \ - MantledHawk Turkey Vulture. • 3 \ BlackHawk-eagle DISCUSSION \\ %x.--_ -•• TurkeyVulture The timing, duration, efficiencyof, and number of point counts is important in obtaining an opti- • 0 mal balance between sampling effort and an ade- quate descriptionof raptor assemblages.If extend- -1 ed count periods do not result in the detection of 0 5 10 15 20 25 30 35 40 45 50 more species,then additional time would be best investedin samplinga larger area or searchingand installing additional points. Within a counting -/ %presence atsome time during counts point, time should be limited to that needed to //\ detect all speciesactive on a given morning. In our 8 2 counts, every few species were detected before

t= 1 0900 H, whereas, after 1200 H detection frequency ./- .--- ..• • of most raptors, including the Black Hawk-Eagle, • 0 declined. Counts of less than 3 hr duration had a high probability of missingsome species, but after -1 three counting hours, no new specieswere detect- 0 5 10 15 20 25 30 35 40 45 50 ed. Also, for the common species,abundance in- Number of Samples dices stabilize after the third hour of the count. This indicates that the best cost-effective alterna- Figure 5. Simulation of the increase in precision of es- umates around the means (mean = 0) of contacts/hour tives may be conducting 3-hr counts from 0900- 1200 H (Local Standard Time), that is between 3 and percent presenceestimates in relation to number of counts for the four most common speciesin the area. hr 41 min and 6 hr 41 rain after sunrise, a bit later The _+95%Confidence Interval of the mean is expressed than what has been found in other rainforest areas as multiplesof the mean (Upper 95% ConfidenceInter- (Whiracre and Turley 1990, Whitacre and Thor- val - Lower 95% ConfidenceInterval/Mean) and plot- strom 1992). ted againstsample size. The number of counts is related to the number of speciesthat would be detected (Whitacre and Turley 1990) and how well the raptor assemblage is characterized. it is also important in determining in more than one previous survey, and only one the precision of the abundance estimates(Bibby et speciesin more than two studies (Table 1). al. 1992a). Accuracyand precision in monitoring The results of the fourteen standardized counts programsis needed to detect small differencesbe- (Table 2) indicated that the Black Vulture was the tween years,habitats, or areas.However, after a cer- most frequently detected species, involving 118 tain level of precision is reached, further improve- (59%) contacts.The specieswas usually observed ment can only be achieved at a high cost. Our flying in groups of 2-12 individuals (2.6 -+ 1.7 results indicate that the more detectable species birds/group). Only 26% of the observationsof this are fbund with few counts, but that sample sizesof speciesinvolved single birds. Mantied Hawks ac- 11-12 points may still miss some important, counted for 31 (16%) contactswith groupsof 1 or though less detectable, species.With 20 points, 2 birds (1.2 + 0.4 birds/group). BlackHawk-Eagles 90% of the more-detectablespecies are identified, involved26 (13%) contactsof singlebirds or pairs but more than half the potential speciespresent in (1.2 + 0.4 birds/group). The remaining species the area are still missed (Table 1). On the other each accounted for less than 5% of the records. hand, 12 counts may provide abundance indices Both preserve areasshowed similar speciescom- within one +95% CI of the mean, but only for the position, but among the specieswith more than two most common species(Black Vulture, Manfled one record, the Short-tailed Hawk was reported Hawk). A minimum of 20-30 counts would be only in Petar. All speciestended to be more abun- needed to obtain comparable precision for the dant in Petar than in Intervales (Table 2), al- next most frequently-detected species (Black MARCH 200:3 SOARING RAPTORS IN ATLANTI(; RAINFOREST 227

Hawk-Eagle). Precise abundance estimatesfor the These difikrences may also explain the surprising still lessdetectable raptors would only be obtained absenceof the Ornate Hawk-Eaglefrom the Willis at the expenseof much more samplingeffort (>50 and Oniki (1981) and Vielliard and Silva (2001) counts), which also would be needed to detect ad- surveys.On the other hand, we were unable to de- ditional speciesor io slightlyimprove the precision tect sonhespecies that may be relatively common of the estimatesof more common species. in the area, such as forest-falcons and the Ghiman- Biasin the selectionof counting points may also go Caracara (Milvago chimachima).Other raptor have an eftbct on the estimationof speciesrichness specieswe may have missedthat have been report- and on lhe accuracyof abundance estimates.This ed fkom other Atlantic rainforest areas are the Bi- bias can arise from selecting points not represen- colored Hawk (Accipiterbicolor) (Willis and Onik• tative of the area being surveyed,or fi'om excessive 1981), the White-necked Hawk and the Harpy Ea- proximity of sampling points, which may result in gle (Harpia harfiyja)(Albuquerque 1995). double counting of territorial raptors.Comparison The point count method provides only indices of two subsetsof our data points suggeststhat fbr of speciesabundance rather than information on the most detectable species,species richness was population densities.Theoretically, these measures little affected by the reduction of number and should be related, but detectability may differ be- placement of counting points, and that relative tween species,so comparisonsof detection rates abundance estimates were also not altered. are most reasonablebetween speciesof similar size Counts were conducted at the start of the nest- and behavior (Thiollay 1989). Different soaring ing season,when raptors were most active and propensities,behavior, and body size of the differ- probably when the point count method was most ent specieslargely affect detectabilityduring point effective (Whitacre and Turley 1990). However, counts. This is especiallytrue becausewe used an during the counts we only detected nine of the 24 unlimited observation radius, which favors larger speciespotentially present in the area (Table 1). speciesat the expense of smaller ones. Thus, the Most of the remaining speciesseem to be relatively point-count method may overestimate the abun- uncommon, but sonhemay have been missed be- dance of larger and wide-ranging species(eagles, causeof the timing of the counls. This may be the vultures) in relation to thoseof the smallerraptors case of Grey-headed Kite (Lefitodo• caya•e•sis), (Jullien and Thiollay 1996). which generallydisplays early in the breeding sea- Both the Intervales and Petar protected areas son (October, according to Vielliard and Silva •nay support similar raptor assemblages,although [2001] and Whitacre pets. comm.). Small sizemay sonhe raptors, in particular the Turkey Vulture, account for the absence from our list of Rufous- tended to be more abundant in Petar than in In- thighed Kite (Harfiagusdiodo• 0 , which was relative- tervales. The relatively high abundance of open- ly frequent in some previous surveys.The point- country speciesreveals the effect of human settle- count method may also overlook non-soaring ment in the area and the relative proximity of species,which would explain why Tiny Hawks, Bat open habitat near the preserves.In spite of this Falcons,forest-falcons (Micra.stur s•O.), and Crested influence, the Man fled Hawk, an endemic species Eagles (Mor•h•us g•ia•e•sis) were not detected. of the Atlantic rainforest, was found to be common For these species,the use of acousticalluring or in the area. Hawk-eaglesalso were common, par- pre-dawn listening may be more effective (Whita- ticularly the Black Hawk-Eagle, a specialistof sec- cre and Turley 1990). ondary or disturbed forest patches (Jullien and Comparing all raptor surveysavailable from the Thiollay 1996, Thiollay 1999). The Ornate Hawk- area (Willis and Oniki 1981, Mafiosa et al. 1997, Eagle, which is thought to be restricted to little- Vielliard and Silva •001; Table 1), the Black Vul- disturbed forest (Jullien and Thiollay 1996), was ture appears as the most common species.The less frequently observed than the Black Hawk-Ea- Roadside Hawk emerged as the second most-fre- gle. This difference in detectionsof these two spe- quently-detectedspecies in Willis and Oniki (1981) cies'may reflect the dominanceof late-secondary and Vielliard and Silva (2001) studies, instead of forest in the area, but also the lower soaringpro- the Mantled Hawk, which ranked second in our pensity (del Hoyo et al. 1994) and detectability of survey. This dif•brence may indicate temporal Ornate Hawk-Eagle, even in areas where it is more changesin speciesabundance or, more likely, dif- abundant than the former species (D. Whitacre ferences in the methods and habitats sampled. pets. comm.). Mean percent presence of hawk-ca- 28 M_•os^ ET AL. VOL. 37, No. 1 gles was higher and the presence of Buteohawks LITERATURE CITED lower in the Paranapiacabafragment than in sev- ALBUQUERQUE,J.L.B. 1986. Conservation and status of eral rainforest areas where raptor counts have raptors in southern Brazil. BirdsPrey Bull. 3:88-94. been conducted in Central America (Jones and 1995. Observationsof rare raptors in southern Sutter 1992, Whitacre et al. 1992b). Also, the Black Atlantic rainforest of Brazil. J. Field Ornithol.66:363- Vulture may be relatively more abundant and Tur- 369. key Vulture less abundant than in the Central BIBBY,CJ., N.D. BURGESS,AND D.A. HILL. 1992a.Bird census American study sites.These differences may indi- techniques.BTO & RSPB.Univ. Press,Cambridge, U.K cate differences in habitats among areas or reflect --, N:J. COLLAR,M 0. CROSBY,M.F. HEATH, C.H. IM- different degrees of modification among sites. BODEN,T.H. JOHNSON,A.J. LONG, AJ. SATTERSFIELD, The results of these counts confirm the impor- ANDS.J. THORGOOD. 1992b. Putting biodiversityon the tance of the ecological continuum of the Parana- map: priority areas for global conservarion.ICBP, piacaba fragment for the conservationof endan- Cambridge, U.K. gered raptors in the Brazilian Atlantic rainforest. BIERREGAARD,JR., R.O. 1998. Conservationstatus of birds of prey in the South American tropics.J. RaptorRes. The area supportsrelatively abundant populations 32:19-27. of the Mantled Hawk, a poorly known species BILDSTEIN,K.L., W. SCHELSKY,AND J. ZALI.ES.1998. Con- (Thiollay 1985, IUCN 1990), which is considered servationstatus of tropical raptors.J. RaptorRes. 32:3- endangered (Thiollay 1994) or nearly endangered 18. (Collar et al. 1992, del Hoyo et al. 1994). The rel- BRUCE,P., J.L. S•MON,AND T OSWALD.1995. Resampling atively frequent detections of two other poorly- statsfor DOS. Operation guide. ResamplingStats inc., known species,the Ornate and the Black Hawk- Arlington, VA U.S.A. Eagles (Bildstein et al. 1998), also is notable. BURNHAM,W.A., D.F. WHITACRE,AND J.P. JENNY. 1994. The Particularlyas the subspeciesof Black Hawk-Eagle, Maya project: use of raptors as tool for conservation which was documented, is an endemic of the Bra- and ecologicalmonitoring of biologicaldiversity. Pag- zilian Atlantic rainforest (Bierregaard 1998). Fur- es 257-264 in B.-U. Meyburg and R.D. Chancellor ther fragmentation of the Paranapiacaba forest [EDS.],Raptor conservationtoday. WWGBP and Pica may severely impact populations of these poorly- Press,Near Robertsbridge,U.K. CARVALHO,M.P., C.H.G. CEZARE,C.G. COSTA, R.G.T DA studied species.Although changeson the popula- CUNHA, S.S. DORNELLES,M.S. FILIIAO,J.C. GuIx, A. tions of the rare or lessdetectable speciesmay not HERNANDEZ,P.C. LAZARIN, L. 1,I•ORENS,E. MATEOS,C be adequately tracked with this general point- MIQUEL, L.M. PETRONI, B.S. PINTO, K. PISCIOTTA, C. count technique, standardization of the counting SANCHEZ,I. OLIVERAS,AND A. SERe,&.2002. Description method would allow the implementation of moni- of the study areas. Pages 27-48 in E. Mateos,J.C toring programswith the objective to detect trends Guix, A. Serra, and K. Pisciotta [EDs.], Censuses of of some of the more detectable species, such as vertebrates in a Brazilian Atlantic rainforest area: the Man fled Hawks and hawk-eagles. Importantly, Paranapiacaba forest fragment. Centre de Recursos these speciesare of conservationconcern and may de Biodiversitat Animal. Universitat de Barcelona, be good indicatorsof habitat changesin the Atlan- Barcelona,Spain. tic rainforest relevant to the entire raptor assem- COLLAR,N:J., L.P. GONZA(;A,N. KRABBLE,A. MADROiqO- blage. NIETO, L.G. NARANJO,T.A. PARKER,AND D.C. WEGE 1992. Threatened birds of the Americas. The iCBP/ ACKNOWLEDGMEN I'S iUCN Red Data Book. ICBP, Cambridge, U.K. CRACRAFT,j. 1985. Historical, biogeography,and patterns This work is dedicatedto our late friend CarlesL6pez of diftkrentiation within the South American avifau- Adzerias. We also wish to remember Roberto Bfirgi, for- mer director of the Parque Estadual Turfsrico do Alto ha: areas of endemism. Pages49-84 in P.A. Buckley, R•beira, who died in an accident during this study.We M.S. Foster, E.S. Morton, R.S. Ridgely, and EG. Buck- would like to thank Dr. Juan Carlos Guix, Dr. Antoni Ser- ley [EDs.], Neotropical . The American ra, and Katria Pisciottatbr making this expedition possi- Ornithologists' Union, Washington,DC U.S.A. ble, as well as the staff of Parque EstadualIntervales and DELHOYO, J., A. ELLIOTT,AND J. SARGATAL(EDS.). 1994. Parque EstadualTurfsrico do Alto Ribeira for all the fa- Handbook of the birds of the world. Vol. 2. New world cilitiesthey provided. Flaviade CamposMartins, Blanche Sousa Pinto, and Montse Ontafi6n assistedduring the vulturesto .Lynx Ed., Barcelona,Spain. field work. The comments of Dr. D.F. Whiracre, Dr. K.L. FONSEC^,G.A.B. 1985. The vanishing Brazilian Atlantic Bxldstein, and an anonymous referee very much im- forest. Biol. Conserr. 34:17-34. proved this manuscript. FORSMAN,D. ANDT SOLONEN.1984. Censusingbreeding MARCH 2003 SOARING RAPTORS IN ATIANTIC RAINFOREST 29

raptors in southern Finland: methods and results. PISC1OTYA,K. 2002. The Paranapiacaba forest fragment Ann. Zool. Fenn. 21:317-320. Pages 19-25 in E. Mateos,J.C. Guix, A. Serra, and K. Gu•x, J.C. 2002. Introduction and objectives.Pages 3-18 Pisciotta [EDs.], Censuses of vertebrates in a Brazilian in E. Mateos, J.C. Guix, A. Serra, and K. Pisciotta Atlantic rainforestarea: the Paranapiacabaforest frag- [EDS.], Censuses of vertebrates in a Brazilian Atlantic ment. Centre de Recursos de Biodiversitat Animal rainforest area: the Paranapiacaba forest fragment. Universitat de Barcelona, Barcelona, Spain. Centre de Recursos de Biodiversitat Animal. Univer- STATTERSFIEI.D,AJ., M:J. CROSBY,AJ. LONG, AND D.C. sitat de Barcelona, Barcelona, Spain. WEGE.1998. Endemic bird areas of the world: priori- , A.AJ. TABANIaZ,A.N. Smv^, C. LOPEZ,C. MARTI- ties for bigdiversity conservation. BirdLife Interna- NEZ, E. MATHEU, EL. DE Souz& K.R. P•SC•OTr^, N. tional, Cambridge, U.K. BRADBURY,AND W.G. PORTILHO.1992. Viagem de re- THIOLLAY,J.-M. 1985. Falconiformsof tropical rain forest' conhecimento cientffico a algumas areas desconheci- a review.Pages 155-165 in I. Newton and R.D. Chan- das da Fazenda Intervales, Estado de Sag Paolo, dur- cellor [EDS.], Conservation studies on raptors. Vol. 5. ante o periodo de 04 a 16 outubro de 1991. Grupo ICBP Tech. Publ., Cambridge, U.K. Estud. Ecol. Sez. Doc. 4:38-94. --. 1989. Censusingof diurnal raptors in a primary IUCN. 1990. IUCN Red List of threatened . rain forest: comparative methods and speciesdetect- IUCN, Cambridge, U.K. ability.J. RaptorRes. 23:72-84. JONF•S,L.E. ANt)j. SUTTER.1992. Resultadosy compara- 1994. A world review of tropical forest raptors. ci6n de dos aftos de trabajos de censo en tres unida- Current trends, research objectives,and conservation des del complejo Reservade la BiosferaMaya/Reser- strategy.Pages 231-239 in B.-U. Meyburg and R.D. va de Calakmul. Pages69-87 in D.F. Whitacre and Chancellor lEDS.], Raptor Conservation Today. Pica R.K. Thorstrom [EDS.], Proyecto Maya: uso de aves Press,Berlin, Germany. rapacesy otra fauna como indicadoresdel medio am- 1999. Responseof an arian community to rmn biente, para el disefio y manejo de areas protegidas y forest degradation. Biodiv.Conserv. 8:513-534. para fortalecer la capacidadlocal para la conservaci6n VANNINI,J.P. 1989. Neotropical raptors and deforestation en Amdrica Latina. The Peregrine Fund, Inc., Boise, ID U.S.A. notes on diurnal raptors at finca E1 Faro, Quetzalten- ango, Guatemala.J. RaptorRes. 23:27-38. JULLIEN,M. ANDJ.-M. THIOLLAY.1996. Effects of rain for- VmLLIARD,J.M.E. ANDW.R. Smv^. 2001. Avifauna. Pages est disturbance and fragmentation: comparative 125-145 in C. Leonel [ED.], Intervales. Fundacao changes of the raptor community along natural and para a conserva½5oe a produ•to fiorestal do estado human-made gradients in French Guiana. J. Biogeogz. 23:7-25. de Sag Paulo. Secretaria de esatdo do Meio Ambiente, Governo de Estado de Sag Paulo, Brazil. MA•OSA,S. ANDV. PEDROGGHI.1997. A raptorsurvey in the BrazilianAtlantic rainforest. J. RaptorRe3. 31:203-207. WEGE,D.C. ANDA.J. LONG.1995. Key areasfor threatened , E. MATEOS, AND V. PEDROGGHI. 2002. Birds of birds in the neotropics. BirdLife Conserr. Ser. 5, prey survey (aves:cathartiformes and ) BirdLife International, Cambridge, U.K. in the Paranapiacabaforest fragment. Pages 165-179 WHITACRE,D.F. ANDC.W. TURLEY.1990. Further compar- in E. Mateos, J.C. Guix, A. Serra, and K. Pisciotta isons of tropical forest raptor censustechniques. Pag- lEDS.], Censuses of vertebrates in a Brazilian Atlantic es 71-92 in W.A. Burnham, D.F. Whitacre, and J.p. rainforest area: the Paranapiacaba forest fragment. Jenny [EDs.], ProgressReport III. Maya Project: use Centre de Recursos de Biodiversitat Animal. Univer- of raptors as environmental indices for design and sitat de Barcelona, Barcelona, Spain. management of protected areas and for building local MATEOS, E. AND S. MA•OS^. lEDS.] 1996. Memoria de re- capacityfor conservation in Latin America. The Per- suitados del segundo viaje de reconocimiento cienti- egrine Fund, Inc., Boise, ID U.S.A. rico a regiones poco conocidas de la Fazenda Inter- --AND R.K. THORSTROM.lEDS.] 1992. Proyecto vales, estado de Sag Paulo, Brasil. Publicaciones de la Maya: uso de aves rapaces y otra fauna como indica- Universidad de Barcelona, Barcelona, Spain. dores del medio ambiente, para el disefio y manejo , J.C. Gu•x, A. SERP•, AND K. PISCIOTTA.[EDS.] de areas protegidasy para fortalecer la capacidad lo- 2002. Censuses of vertebrates in a Brazilian Atlantic cal para la conservaci6n en Am6rica Latina. The Per- rainforest area: the Paranapiacaba fragment. Centre egrine Fund, Inc., Boise, ID U.S.A. de Recursos de Biodiversitat Animal. Universitat de --, L.E. JONES,ANDJ. SUTTER.1992a. Gensosde aves Barcelona, Barcelona, Spain. rapacesy de otras avesen el bosque tropical: mejoras MYERS,N. 1988. Threatened biotas: "hotspots" in tropi- hechasa la metodologia. Pages43-56 in D.F. Whiracre cal forests. Environmentalists 8:187-208. and R.K. Thorstrom [EDS.], Proyecto Maya: uso de , R.A. M•TTERME•ER,C.G. M•TrERMEmR, G.A.B. D^ avesrapaces y otra fauna como indicadoresdel med•o FONSEC&,AND J. KENT. 2000. Biodiversityhotspots for ambiente, para el disefio y manejo de areasprotegidas conservationpriorities. Nature403:853-858. y para fortalecer la capacidad local para la conserva- 30 M•OSA ET XL. VOL. 37, NO. 1

ci6n en Amfirica Latina. The Peregrine Fund, Inc., para el disefio y xnanejode fireas protegidasy para Boise, ID U.S.A. fbrtalecerla capacidadlocal para la conservaci6nen --, C. TURI,EY, A.E. HERNANDEZ, AND F. OSORIO. AmErica Latina. The Peregrine Fund, Inc., Boise,ID 1992b.Una cmnparaci6nde comunidadesde avesra- U.S.A. pacesque habitan bosquestropicales primarios y mo- Wn,LIs, E.O. •Nr) Y. ONmL 1981. Levantamento prelimi- saicosde ranchos agropecuariosque utilizan la tfic- nar de aves en treze •u'eas do estado de Sao Paulo nica de Tala y quema del bosque.An•disis de los datos Rev. Bras. Biol. 41:121-135. de 1989. Pages 89-101 in D.F. Whitacre and R.K. Thorstrom [EDs.],Proyecto Maya: uso de avesrapaces Received 14 March 2002; accepted 5 October 2002 y otra fauna como indicadoresdel medio ambiente, Associate Editor: Marco Rcstani