Population Status of the Endangered Hawaiian Hawk

j. RaptorRes. 31(1):11-15 ¸ 1997 The Raptor ResearchFoundation, Inc.

POPULATION STATUS OF THE ENDANGERED HAWAIIAN HAWK

LINNEA S. HALL AND MICHAEL L. MORRISON Departmentof BiologicalSdences, California State University, Sacramento, CA 95819 U.S.A.

PETER H. BLOOM WesternFoundation of VertebrateZoology, 439 CalleSan Pablo,Camarillo, CA 93010 U.S.A.

ABSTRACT.--Weassessed the current abundance and distribution of Hawaiian Hawks ('io; Buteosolitarius) on the island of Hawaii to determine if this federally endangered bird should be downlistedto threatened status.We found a densityof 0.004 hawks/ha on the island. Using an estimateof 400 000 ha of suitable'io habitat on Hawaii, we estimateda total of 1600 hawks (1120 adults;560 pairs) on Hawaii. Basedon the wide distribution of 'io among vegetationtypes on the island and little apparent change in numbers during the past decade, we agreed with the recommendation for downlistingthe hawk but suggestedthat researcherscollect long-term demographic data to better understand the statusof this species. KEYWORDS: Buteo solitarius,Hawaiian hawk, 'io,population status.

El estadode poblaci6ndel Buteosolitaruis en peligro de extinci6n. RESUMF.N.---Nosotros fijamos la cantidad corriente y distribuci6n de Buteosolitarius en la isla de Hawaii para determinar si el p•tjaro en peligro de extinci6n por leyesfederales debe ser reducido a estado amenazado. Nosotros los encontramos una densidad de 0.004 halc6n/ha en la isla. Usando la estimaci6n de 400 000 ha de h•tbitat conveniente en Hawaii, nosotros estimamos un total de 1600 halcones (1120 adultos;560 parejas) en Hawaii. En base de la distribuci6n amplia de B. solitariusentre clasesde vege- taci6n en la isla y poco cambio aparente en la cantidad durante la dgcadapasada, nosotros estamos de acuerdo con la recomendaci6npara reducir el halc6n pero sugerimosque los investigadoresjunten datosdemogrfificos de larga duraci6n para poder entender el estadode este especiemejor. [Traducci6n de Rafil De La Garza,Jr.]

The Hawaiian Hawk (Buteo solitarius), or 'io, was fin (1985) found that the speciesmight be rela- federallylisted as an EndangeredSpecies in 1967 tively unaffected by habitat modifications com- (37 FR 4001, 11 March 1967) based on its restrict- pared to many other native bird speciesafter find- ed range on the island of Hawaii (hereafter Ha- ing that foraging and nesting occurred in agricul- waii), its low numbers at the time of listing (Berger tural areas and in stands of exotic vegetation 1981), and the perceived threats to its preferred (Baskett and Griffin 1985). Griffin (1985, 1989) es- habitat from agricultural and commercial devel- timated the population at 900 breeding pairs and opments (U.S. Fish and Wildlife Service [USFWS] a total of 2700 individuals in 1983. Because of this, 1984). At the time of listing, no intensivestudy of the USFWS proposed downlistingthe 'io from en- the ecology of the 'io had ever been conducted, dangered to threatened status (58 FR 41684, 4 Au- and anecdotal accountsgave differing reports on gust 1993). Because of questions over the validity its abundance across the island (Munro 1944, of basing such a reclassificationon 10-year-old USFWS 1984). data, the USFWS requestedthat an island-widesur- Uncertainty over 'io abundance continued vey be conductedof the 'io population to obtain a through the next decade. An intensive surveyini- more current estimate of the population size. tiated by the USFWS in 1976 on Hawaiian forest Herein, we present our survey results and sam- birds wasunable to estimatethe 'io population size pling design to provide a baseline for future sur- (Scott et al. 1986). After a detailed study of 'io veysdesigned to monitor the size of the 'io popu- breedingbiology in < 1% of the island'sarea, Grif- lation on Hawaii.

11 12 HALL ET AL. VOL. 31, NO. 1

STUDY AREA AND METHODS Table 1. Vegetation descriptionsand codes for survey The most efficient way to sampledominant vegetation transects used in analysesof Hawaiian Hawk numbers types acrossHawaii for the occurrence of 'io was to con- across Hawaii in December 1993. a duct unlimited distance point counts (Blondel et al. 1981) along paved and dirt roadsacross the island.Point counts were selected to make our methods generally comparableto those of Scott et al. (1986), who conduct- CODE QUENCYb DESCRIPTION ed the most complete previouscensus of 'io on Hawaii 1 63 Sugar cane fields with exotic and/or as part of the USFWS's Hawaiian Forest Bird Survey native trees or shrubsat edges, as (1976-79). We also needed a method that could sample windrows. the 'io's use of vegetationranging from lowland agricultural areas to subalpine woodlands (Scott et al. 1986, 2 41 Short or tall exotic trees with exotic Griffin 1989), and would be applicable to birds with shrubs,and sometimesexotic grass- home range sizesvarying from 48 ha in agriculturalareas es. to 490 ha in forestsand mid-elevationpasturelands (Bas- 3 22 Macadamianut or papayaorchard kett and Griffin 1985). Use of roadwayswas the only fea- with native and/or exotic trees or sible means of satisfyingthese objectives.Some studies have indicated that roadsidecounts can give biasedesti- shrubsat edges. mates of bird densitiesand vegetation associations,but 4 38 Grassland with scattered exotic other studies have indicated that road counts can be use- and/or native trees (especially ful and appropriatewhen large areasneed to be sampled o'hia); scattered homes. and monitored long term (Fuller and Mosher 1981). 5 5 Pioneer exotic vegetation growing on When possible,we used roads that crisscrossedan area lava. to more thoroughly samplefor 'io. We conductedpoint counts 0.1-16 km off main roads to ensure that traffic 6 68 Native trees and native shrubs occa- noise did not interfere with the counts and that we more sionallywith scatteredorchard adequately sampled vegetation that could contain 'io. trees, or exotic understoryand Count stationswere located disproportionatelyamong homes. vegetationtypes (Table 1), basedon information that 'io 7 99 Mixed exotic and native trees, some- were unlikely (or very uncommon) in shrublands(vegetation type 10), upper-elevationmamane-naio (Sophora times with mixed exotic and native chrysophylla-Myoporumsandwicense) woodlands (vegetation shrubsor grass. type 12), and exotic pioneering lava vegetation (vegeta- 8 24 Residential area with scattered exotic non type 5) (J.Jeffrey and J. Giffin pers. comm.). and native vegetation. All count stationswere 0.8-3.2 km apart, and counts 9 40 Native tree and mixed exotic and na- were conductedby 1-2 observersbetween 0900-1700 H. Each point count lastedfor exactly10 rain, which wasthe tive shrub vegetation on lava, some- same count length used by Scott et al. (1986), and in- times with scattered homes; a cluded 8 rain of listeningand watchingfor hawks,plus 2 pioneer community. m•n of playbackof taped adult territorial and fledgling 10 2 Mixed exotic and native shrubs with callsof 'io. After the first minute of the tape elapsed,we scattered native and exotic trees. turned it off and observed the area for any hawksfor 7 11 16 Native trees and grassland;non-pio- min. Anytime an 'io responded to the tape, either by calling or flying to the point, we immediatelystopped the neer community. tape, but continued the count to determine if any addi- 12 4 Mamane-naio vegetation,with grass tional hawks were observed.We then played the tape and/or exotic shrub understory, or again for 1 rain, and watched for the last minute of the sometimes with scattered exotic count. trees. Although surveyshave not previouslyused broadcast a Scientific names of plants listed: sugar cane (Saccharumoffici- calls, Banko (1980) and Baskett and Griffin (1985) re- narum), macadamianut ( Macadamiaternifolia), papaya ( Caricapa- ported that 'io call and defend their territories in the paya), o'hia (Metrosiderospolymo•pha), mamane (Sophorachryso- w•nter. On 12-13 December 1993, we tested if broad- phylla), and naio ( Myoporumsandwicense). castedterritorial callselicited responsesfrom 'io by going b Frequency= total number of times the describedvegetation to areas known to have 'io present (J. Jeffrey pers. wasrecorded along surveyroutes. comm.). We watched 'io that were >200 m awaywhile we played the taped calls. Eighty percent of the hawks respondedby taking flight, calling or coming to the tape. No countswere conducted when precipitation exceed- we tried to determine age and sex of hawksin the field, ed a light rain, or when wind exceeded 24 km/h. We it was often difficult to make a positiveidentification, so recorded the distancefrom the point of initial detection in our analyseswe combined all sightings. of all observed hawks, the detection mode (visual, aural We used program DISTANCE (Buckland et al. 1993, or both), the morph (light, dark, unknown), and the veg- Laake et al. 1993) to estimate the densities of 'io in the etation where the bird wasobserved (Table 1). Although 12 major vegetationtypes recorded during our surveys M_ARCH 1997 STATUS OF HAWAIIAN HAWK 13

Table 2. Summaryof 'io densityestimates (all agestogether) by vegetationtype, calculatedby program DISTANCE (Laake et al. 1993) from surveydata acrossHawaii, December 19937

ESTI-

MA- DENSITY ESTIMATIONS e To- No. TOR T•tL No. 'Io MoI)- DE- DENSITY BOOTSTRAP f VEGETATION EF- Pox- OBS- EL ESTI- TECT. ENC. CODEb FORT• NTS YD. No. d MATE SE %CV 95%CI df PaOB. RATE EST. %CV RUNS

All veg types 399 399 98 5 0.004 0.0007 15.9 0.003-0.006 345 0.02 0.24 0.004 23.7 400 Veg I 63 63 26 1 0.002 0.0006 29.1 0.001-0.003 86 0.08 0.41 0.002 56.2 100 Veg 2 41 41 5 1 0.0004 0.0003 77.1 0.0001-0.002 12 0.11 0.12 -•g Veg 3 22 22 5 1 0.003 0.0016 58.8 0.0009-0.008 22 0.03 0.23 -- Veg 4 38 38 16 1 0.004 0.0011 29.1 0.002-0.006 52 0.04 0.42 -- Veg 5 5 5 0 __h Veg 6 68 68 10 1 0.003 0.0014 41.3 0.002-0.007 76 0.02 0.15 -- Veg 7 99 99 22 1 0.005 0.0013 26.6 0.003-0.008 118 0.02 0.22 0.005 67.6 100 Veg 8 24 24 7 1 0.009 0.0046 54.0 0.003-0.024 26 0.01 0.29 -- Veg 9 40 40 I -- Veg 10 2 2 0 -- Veg 11 16 16 7 1 0.005 0.0024 50.5 0.002-0.013 17 0.03 0.44 Veg 12 4 4 0 -- a For explanationof DISTANCE program estimations,see text. b For vegetationcode explanation,see Table 1. cTotal effort = the sum of the number of times each point and its correspondingvegetation was sampled. d EstimatorModel No. = the mathematicalestimator model selectedby program DISTANCE to analyzethe point data, where the chosen model was the one that had the smallest Akaike's Information Criterion value (AIC). e Densityestimations: Estimate = densityin number of 'io/hectare; SE = standarderror of the estimation;%CV = percentcoefficient of variationof the estimate;95% CI = 95% confidenceinterval for the estimate;df = degreesof freedomused in the analysis;Detect Prob. = the estimateof averageprobability of detectingan 'io; Enc. rate = the number of animalsexpected to be observedper point. f DensityBootstrap values: Est. = bootstrappeddensity estimate; %CV = percent coefficientof variation of this estimate;Runs = total number of bootstrapruns conducted. g "--" = tOOfew degreesof freedom to conduct bootstrap analyses. h ,,__,, = no or too few 'io observedalong this surveyroute, so no densityanalysis could be performed.

(Table 1). Observations of 'io were entered as the radial served. Thirty-three hawks were identified as distance to the hawk from the point. We truncated the adults, 7 as iramatures, and 58 as unknown-aged. distances at 3000 m (the maximum distance at which most 'io were observed) to allow all hawk observations to Forty-five hawks were light morph birds and 14 be entered into the analyses.We instructedthe program were dark morphs. to select the most appropriate density estimation model Densities ranged from a low of 0 in vegetation for each analysis,based on maximum likelihood ratio types5, 9, 10, and 12 to a high of 0.009 hawks/ha tests of the models vs. each other. We also instructed the in vegetationtype 8 (Table 2). Most densitieswere program to conduct400 bootstrapsamples for the island- wide data, to obtain reliable estimates of the variances between 0.003 and 0.005 hawks/ha, with an overall around the densityestimates, and 100 bootstrapsamples mean of 0.004 hawks/ha. Vegetation types5 and 9 for each of the analysesof densityby vegetationtype. had lava as a major ground component, and thus We estimatedthe current population sizeof the 'io on had poorly-developedtree cover. Type 10 vegeta- Hawaii basedon the densityof hawksper vegetationtype, and the estimatedpercent coverby each vegetationtype tion was dominated by shrubs,and type 12 wasin on the island of Hawaii (Jacobiand Scott 1985, Cuddihy mamane-naiowoodland. Type 2 vegetationwas typ- and Stone 1990). ified by exotic trees of various sizesand had very few hawks.Type 8 vegetationconsisted of residen- RESULTS AND DISCUSSION tial areaswith both native and exotic tree compo- We sampled40 transectsacross Hawaii, with 399 nents and showed the highest hawk densities. points covering approximately 500 km of roads. Grasslands with scattered exotic and native trees Among these points, 98 different 'io were ob- (vegetationtype 4) also had moderately-highden- 14 H•I• ET AL. VOL. 31, NO. 1 sities of birds. Bootstrapped density estimates 1600 birds (range = 1200-2400), with 1120 adults matched the model estimates in all cases where ad- or 560 pairs. equate degreesof freedom existed. Our estimated density of adults (1120) is about Surveysfound 'it most commonly in areaswith 25% below the lower end of the target range nec- native and/or exotic tree cover, usually with un- essaryfor a stable 'it population, accordingto the derstoriesof exotic grass,and sometimeswith na- Recovery Plan (target = 1500-2500 adult birds). tive and/or exotic shrub understories.Although Assuming that all birds alive during our surveys 'it were not found frequently in small patchesof survived to breed, the total number of birds we mixed nativeand exoticforest surrounded by open estimated(1600) isjust abovethe lower end of the fields or orchards,they were commonly observed target range, but is still below the mean target val- over the open areas, or in open places,with scat- ue of 2000. The target value of adults is not entered native and/or exotic trees. For example, in closed in the confidence interval around the sugar cane fields with ribbons of native or exotic 0.004/ha value (95% C.I. = 0.003-0.006), but is trees between fields, or with trees extending down enclosed if we assume that all hawks alive breed from higher elevationforests; in open pastureland (400 000 ha x 0.006 = 2400 hawks). with scatterednative trees; in orchards (especially We found a relativelyhigh number of birds that macadamia nut) with taller native and/or exotic were widely distributedamong vegetation types on trees at the perimeters. This indicated that 'it are Hawaii, including heavily-disturbedareas. In addi- now using areas that are not pure native forest. tion, our results were similar to those found 10 Based on these data and anecdotal breeding rec- yearsearlier by Griffin (1985, 1989), indicating the ords from these more open areas, it appears that likelihood of a relatively stable population during they are also able to successfullybreed there (J. the past decade. Thus, we concluded that down- Jeffrey and J. Giffin unpubl. data). listing to threatened statuswas supported. Griffin (1985, 1989) estimated that the popula- As other biologists have suggestedfor the 'it, tion of 'it on Hawaii was about 2700 hawks in long-term demographic studiesare necessaryto ac- 1983. Of this, 1800 were adults. This estimate curately assessthe overall statusof the population served as the basis for the Hawaiian Hawk Recov- (USFWS 1984, Griffin 1989). Our fieldwork did ery Plan developedby the USFWS (1984). It used not assesspopulation trends, reproductive fecun- an abundance of 2,000 hawks (the midpoint be- dity and success,dispersal or mortality,all of which tween the 1500 and 2500 adult birds thought to be have been shown to be problematic for other for- needed for a self-sustainingpopulation) as the tar- est birds on Hawaii (Scott et al. 1986). Thus, we get to downlist the speciesto threatened status. think that the USFWS should initiate a long-term The island-wideestimate of 'it densitywas based demographic study so future density estimations on a total forested area of 343 000 ha (J.M. Scott can be evaluatedin light of other population data. pers. comm., Griffin 1989). This value correspond- Such a study is necessarybefore delisting from ed roughly to the potential 'it habitat contained threatened status is considered. within the Hawaiian Forest Bird Survey area (Ja- cobi and Scott 1985). Using this area, and the over- ACKNOWLEDGMENTS all estimate of 'it densityfrom our surveys(0.004 We thank S. Johnston at the USFWS Pacific Islandsof- birds/ha, 95% Confidence Interval [C.I.] = 0.003- rice in Honolulu, Hawaii; J. Jeffrey at the USFWS office 0.006), we .obtaineda total density of 1372 'it in Hilt; L. Katahira at the NPS office at Hawaii Volcanoes (range = 1029-2058) on Hawaii. Much of the low- National Park; M. Reynolds,J. Jacobi, and T. Pratt at the land forested areas of Hawaii, including the sugar National BiologicalService office at HawaiiVolcanoes Na- tional Park; J. Giftin at the Forestry and Wildlife office cane, macadamia nut and other disturbed areas oc- in Kamuela; and all of the associated field assistants in cupied by 'it, were excluded from Griffin's area Hawaii for their assistancein establishingtransects and estimate. We therefore modified the Griffin esti- conducting surveys.We also thank reviewersof earlier mate by adding 60 000 ha of mixed sugar cane- reports and manuscript drafts, including C.R. Griffin, lowland forested area and various other minor veg- D.E. Anderson, J.M. Scott, E Duvall II, S. Fancy, and B. Harper for significantlyimproving our analysesand con- etation types (Cuddihy and Stone 1990), bringing clusions.This researchwas funded by CooperativeAgree- the total potential 'it habitat to 400 000 ha. This ment No.14-48-0001-93676 between the USFWS and the raised our estimate of 'it on the island to about Western Foundation of Vertebrate Zoology. •VIARCH 1997 STATUS OF HAwAnAN HAWS: 15

LITERATURE CITED GRIFFIN,C.R. 1985. Biologyof the Hawaiian Hawk (Buteo solita•ius). Ph.D. dissertation, Univ. Missouri, Colum- BANI40,W.E. 1980. Part I. Population histories-species bia, MO U.S.A. accounts. Forest birds: Hawaiian Hawk ('io). Natl. 1989. Raptors in the Hawaiian islands.Pages Park Serv. CPSU/UH Avian Rep. 6A, History of En- 155-160 in I• Steenhof, M.N. Kochert and M.N. demic Hawaiian Birds. Hawaii Volcanoes National LeFranc Jr. leds.], Proc. Western Raptor Manage. Park, HI U.S.A. Symp.Workshop. Natl. Wildl. Fed, Sci. Tech. Ser.No. BASKETT,T.S. ANDC.R. GRIFFIN.1985. Final project re- 12, Washington,DC U.S.A. port on the biology of the endangeredHawaiian JACOBI,J.D. ANDJ.M. SCOTT.1985. An assessmentof the Hawk: ecology,life history,and environmentalpollu- current statusof nativeupland habitatsand associated tion problems.Missouri Coop. Wildl. Res.Unit, Univ. endangeredspecies on the island of Hawaii. Pages3- Missouri, Columbia, MO U.S.A. 22 in C.P. Stone and J.M. Scott leds.], Hawaii's ter- BERGER,AJ. 1981. Hawaiian birdlife. Univ. Hawaii Press, restrial ecosystems:preservation and management. Honolulu, HI U.S.A. Coop. Natl. Park Resour. Stud. Unit, Univ. Hawaii, BLONDEL,J., C. FERRYAND B. FROCHOT. 1981. Point Honolulu, HI U.S.A. countswith unlimited distance.Pages 414-420 in CJ. LaaI4Z, J.L., S.T. BucIaa•D, D.R. ANDERSONAND K.P. Ralph and J.M. Scott lEds.I, Estimatingthe numbers BURNI4A•. 1993. Distance user's guide, version 2.0. of terrestrial birds. Stud. Avian Biol. No. 6. Colorado Coop. Fish and Wildl. Res. Unit, Colorado BUCKIAND, S.T., D.R. ANDERSON, K.P. ANDERSON, K.P. State Univ., Fort Collins, CO U.S.A. BUed,•HAMAND J.L. LAA•. 1993. Distancesampling: MUNRO,G.C. 1944. Birds of Hawaii. Tongg Publ. Co., estimating abundance of biological populations. Honolulu, HI U.S.A. Chapmanand Hall, London, UK. ScoTT, J.M., S. MOUNTAINSPRING,EL. RAMSEYAND C.B. CUDDIHY, L.W. AND C.P. STONE. 1990. Alteration of native KEPLER. 1986. Forest bird communities of the Ha- Hawaiianvegetation: effects of humans,their activities waiian islands:their dynamics,ecology, and conser- and introductions. Coop. Natl. Park Resour. Stud. vation. Stud. Avian Biol. No. 9. Unit, Univ. Hawaii, Honolulu, HI U.S.A. U.S. FISH AND WILDLIFE SERVICE. 1984. Hawaiian Hawk FULLER,M.R. ANDJ.A. MOSHER.1981. Methods of de- recovery plan. U.S. Fish and Wildl. Serv., Portland, tecting and countingraptors: a review.Pages 235-246 OR U.S.A. in CJ. Ralph and J.M. Scott lEds.I, Estimating the numbers of terrestrial birds. Stud. Arian Biol. No. 6. Received15 December 1995; accepted24 October 1996.