Ecology, Behavior, and Conservation of the Maui Parrotbill’

The Condor 89:24-39 0 The Cooper Ornithological Society 1987

ECOLOGY, BEHAVIOR, AND CONSERVATION OF THE MAUI PARROTBILL’

STEPHEN MOUNTAINSPRING~ U.S. Fish and Wildlife Service,Patuxent Wildlife ResearchCenter, Mama Loa Field Station, P.O. Box 44, Hawaii National Park, HI 96718

Abstract. The distribution,habitat response, sexual dimorphism, foraging, breeding, and flockingbehavior of Maui Parrotbills(Pseudonestor xanthophrys) were studiedover a five year period.The speciespresent’ range is confinedto montanerainforest on easternMaui, but dry lowlandhabitats on Maui and Molokai wereoccupied before Polynesian contact. Birdsoccurred from 1,250 to 2,150 m elevation,becoming most abundantat 1,750 to 2,000 m. In Kipahulu Valley, birds moved to lower elevationsin some seasons.Maui Parrotbillsassociated with areaswhere vegetation was less disturbed from feralpig activity. Activity areashad more developedunderstories and more opencanopies than nonactivity areas. Birdstended to foragein the subcanopyand understory,with 66% of the prey captured 1 to 5 m aboveground. Plant speciesuse deviated from expectationsbased on availability. The mostfrequent means of preycapture was excavation for timber-boringinsects in dead brancheson live plants.Foraging accounted for 39% of the diurnal time budget;an average prey item appearedto accountfor 1% of the daily energyintake. Male birds had longer, more deeplyhooked bills than females,and tendedto excavateto greaterdepths. Sexes differedin tree speciesuse and foragingmaneuvers, but not in substrate use or foraging height. The principal limiting factors appeared to be habitat loss, avian disease, habitat degradation, predation, and competition from exotic species. Control of pig populations is a needed managementaction. Key words: Activity budget;endangered species; jlocking behavior;foraging behavior; habitat selection;Hawaiian Islands;Maui Parrotbill.

INTRODUCTION natural history accounts (Perkins 1895, 1903; The Hawaiian honeycreepers(Drepanidinae) of- Henshaw 1902) were followed by a half-century fer striking examples of adaptive radiation in bill ofneglect, and the specieswas feared extinct until morphology from a single colonizing ancestor. rediscovered in 1950 (Richards and Baldwin Specialized morphological and behavioral ad- 1953). Other sightings have since been made in aptations for feeding on seeds, nectar, insects, the remote montane rainforests of East Maui fruit, bird eggs,and molluscs span the range of (Bank0 1968; Casey and Jacobi 1974; Shallen- passerine variation. The Maui Parrotbill (Pseu- berger 1974, 198 1; Scott and Sincock 1977; Co- donestorxunthophrys), a fascinating but poorly nant 1981; Carothers et al. 1983; Scott et al. known honeycreeper,is a fairly small (length 140 1986), but little more has been learned of its mm) bird distinguished by a disproportionately natural history. A major survey conducted in large parrot-like bill used to excavateborers from 1980 on the distribution and abundance of forest timber. The sexes exhibit an extreme example birds on Maui, the Hawaii Forest Bird Survey of bill dimorphism among the honeycreepers (HFBS), estimated the population of the Maui (Amadon 1950). Parrotbill to be 500 birds (95% confidence in- The Maui Parrotbill has apparently retreated terval, 270 to 730), with a geographic range of from over 95% of its original range (Scott et al. 50 km2 (Scott et al. 1986). The specieshas federal 1986). Despite its striking appearance and ap- status as an endangered species (U.S.F.W.S. proachability, the specieswas not recognized by 1983) and studies of its natural history are es- 19th century Hawaiians (Munro 1944). Early sential in determining appropriate management actions and priorities to ensure its survival (Kep- ’ ’ Received6 August1985. Final acceptance29 Sep- ler et al. 1984). tember 1986. The objectives of this report are to quantita- 2 Previouslypublished as StephenR. Sabo. tively describe aspectsof the natural history of

1241 MAUI PARROTBILL BIOLOGY 25 the Maui Parrotbill, particularly its niche, hab- extending from 1,825 to 2,175 m elevation be- itat, and behavior, and to discuss its limiting tween the forks of Hanawi Stream in the Koolau factors,and management needs.This study com- Forest Reserve, and located in an area of high bines intensive field work during 1980 to 1985 Maui Parrotbill density. Topography in the area with analyses of museum specimens, literature is ruggedand steep,with an average slope of 40% reports, and unpublished HFBS data. on ridges.During May 198 1, the daily mean temperature at dawn was 7°C the mean afternoon STUDY AREAS high was 18°C and the overall 24 hr mean was 11.2”C. Temperatures of -4°C occurred in Jan- THE HFBS STUDY AREA uary 1985. Annual rainfall at 2,000 m elevation Native bird populations and general habitat in the area averages 500 cm, well distributed structure were surveyed in all areas of native throughout the year, except for irregular 6 to 10 upland forest on Maui during the HFBS. Four week periods of low rainfall (Blumenstock and major habitat types occurred within the range of Price 1967). the Maui Parrotbill. A mesic to wet forest type, At the upper limit of the study area, small dominated by koa (Acacia koa) and ohia (Metro- patches of alpine grassland dominated by Des- siderospolymorpha), occurred along the lateral champsia australis and Holcus lanatus lie margins of the main ohia rainforest of northeast imbedded in mesic subalpine scrub composedof Haleakala. This wet ohia forest had an upper Vaccinium reticulatum, V. berbertfolium, pu- strip adjacent to the treeline of mesic ohia forest kiawe (Styphelia tameiameiae), kukae-nene (Co- that was shorter in stature, somewhat drier, and prosma ernodeoidesvar. mauiensis), pilo (C. poorer in tree speciesthan areas below. Above montana), and the fern Sadleria cyatheoides.The the treeline a mesic subalpine scrub community subalpine scrubgradually gives way to mesic ohia marked the lower boundary of a thermal inver- forest at 2,070 m elevation, which grades into sion layer. wet ohia forest at 2,000 m elevation. The forest For this paper, the range of the species was canopy is dominated by ohia; the major subcan- divided into four contiguous units (Fig. 1). Ele- opy trees are olapa (Cheirodendron trigynum), vational boundaries of each unit were deter- kolea (Myrsine lessertiana), pilo (Coprosma mined by the highestand lowest Maui Parrotbill ochracea),alani (Pelea clusiaefoliaand Pelea un- recordsin the area. The Waikamoi unit extended describedspecies), kawau (Rex anomala), oheohe from the pasture-forestboundary north of Hos- (Tetraplasandra kavaiensis)at lower elevations, mer Grove to the east edge of Koolau Gap on and hoawa (Pittosporumconfertiforum) near the the north slopesof Haleakala, and contained me- treeline. The major shrubsin the forest are ohelo sic subalpine scrub and mesic ohia, wet ohia, and (Vaccinium calycinum), naenae (Dubautia plan- koa-ohia forest. The Wailua unit which extended tagineaand Dubautia undescribed species),akala eastfrom Koolau Gap to the west fork of Hanawi (Rubus hawaiiensis), kanawao (Broussaisia ar- Stream, and the Hana unit which extended from guta), and pukiawe. Henrickson (197 1) de- Hanawi Stream to the north headwall of Kipa- scribed the flora of the area in detail. Feral pigs hulu Valley contained subalpine scrub and mesic (Sus scrofa) have caused moderate to severe and wet ohia forest. The Kipahulu unit was de- damage to the dense understory. fined by the headwalls of Kipahulu Valley and contained mesic ohia, wet ohia, and at lower METHODS elevations koa-ohia forest. Additional areas out- LARGE-SCALE SURVEY side the Maui Parrotbill range were surveyed on windward East Maui at elevations below the The sampling design, field methodology, statis- known range, and over a wide variety of habitats tical analysis, and general results of the HFBS at Waihoi Valley, Kaumakani, Manawainui Pla- were described in Scott et al. (1986). Transects neze, Haleakala Crater, Kaupo Gap, Kahikinui, were systematically positioned at 1.6 to 3.2 km Auwahi, Kula, and the West Maui Mountains. intervals on Maui in native forest bird habitat. Stations were placed 135 m apart along the tran- THE HANAWI STUDY AREA sectsand sampled once in May to August 1980. Studieswere conductedintermittently from 1980 At each station observers conducted 8-min to 1985 on the 50-ha Hanawi study area (Fig. l), counts, recording distancesto all birds detected, 26 STEPHEN MOUNTAINSPRING

Pacific Ocean

@ Hanawi study area

l fossil record

m Maui Parrotbill range

FIGURE 1. Location of Hanawi study area, four range units used in analysis, fossil records, and upper elevational limits of mosquitoes(from Scott et al. 1986).

and botanists recorded habitat structure and pig tion, but for the elevation, habitat type, and pig damage to the vegetation. Incidental observa- damagedistributions, the recordsused were those tions of Maui Parrotbills and other endangered where the birds were within 50 m of the station speciesalong the transects, outside the system- or where the habitat of the bird was unequivocal. atic sampling periods, were also noted. Pig dam- Records of distant calling birds were excluded agewas classifiedinto five generalcategories based from the habitat analysis becausethey may have on a variety of native and exotic indicator plants occupied different habitats than at the station. that differed in sensitivity and response to pig This method of analysis differs from that in Scott damage. The categoriesconstituted a monotonic et al. (1986) which included distant birds in the scalein the order: none, slight, moderate, heavy, habitat analysis and density estimate for each and severe. A general vegetation map showing station; consequently, results in this report are the habitat types in the survey area was con- presentedas birds per station rather than per unit structedusing field notes and aerial photographs. area. In this report, the elevation and contour range limits for the geographicdistribution of the Maui HABITAT DESCRIPTION Parrotbill were inferred from HFBS records and I sampled habitat charactersin the Hanawi study the vegetation map. The census stations lying area at 28 randomly placed points. The points within the range were used to characterize the were located in the field by using a 1:24,000 U.S. expecteddistribution of the Maui Parrotbill with Geological Survey topographic map, an infrared respectto elevation, geographicarea, habitat type, aerial photograph, and landmark features. I re- and pig damage. The data for characterizing the corded crown cover, canopy height, area1 cov- observed distributions for these variables were erage, and average height for individual plant the appropriate subsets of records for station species, and the foliage height profile (vertical counts and incidental observations: all records profile of foliage distribution) in a 50-m radius were used to characterize geographic distribu- around each sample point. The foliage height MAUI PARROTBILL BIOLOGY 27 profile was obtained by optically estimating the prey items by extracting additional material at percent area1foliage coverage in l-m height in- these sites and from identical burrows on the tervals. Sample points were classifiedas “activity same plant, and by matching prey burrow char- areas” and “nonactivity areas.” A particular acteristics with published accounts. Insects ex- sample point was interpreted as being an activity tracted from boring sites were referenced to a area if Maui Parrotbills were observed making monograph on the insect faunas of Hawaiian tree at least five prey captures within 50 m of the species(Swezey 1954). point within a two-day period, other points where The sexof foraging birds was determined when feeding was not observed to this extent were in- light conditions were adequate. Males were iden- terpreted as nonactivity areas. In addition to ac- tified by their large bill, bright yellow superciliary tivity areas found among the 28 points, I also line, olive back, warm yellow underparts, and characterized the habitat at nine other sites in larger size. Females were identified by their the Hanawi study area where Maui Parrotbills smaller bill, dull yellow superciliary line, grayish were observed foraging extensively. back, olive-washed underparts, and smaller size. The habitats of activity and nonactivity areas In specimen series these differences were fairly were compared through a stepwisediscriminant obvious. Data from immature birds were not function analysis of crown cover, canopy height, included in male/female comparisons. Imma- and plant speciescoverages (see Martinka 1972), tures usually resembled small drab females and using a multiple regressionalgorithm with zero were poorly representedin collections. Although assignedto nonactivity areas and one to activity immatures were readily identifiable in family areas. groups, I may have incorrectly identified a few lone immatures as adult females becauseof their FORAGING BEHAVIOR somewhat similar appearance, and because the Most foraging observations were made in the timing and sequenceof postjuvenal molt is un- Hanawi area, but 20% of the observations came known. from other sites. Data on foraging were gathered General foraging patterns and sexual differ- for as long as a bird was in view, with 1 to 15 ences were characterized from data collected maneuvers recorded per bird. Concealing habi- throughout the range, but only data collected in tats such as the upper canopy, dense brush, and the Hanawi study area were used in comparing ravines were carefully searched to reduce bias foraging substrates with available habitat. The toward conspicuousbirds. Associatedbirds were relative availability of different plant speciesas noted. Time intervals between perch changesfor foraging sites was calculated as follows: (1) an Maui Parrotbills and associatedMaui Creepers index of the biomass and foliage volume of each (Paroreomyza montana) were recorded on ran- tree speciesat each sample point was constructed domly selectedfocal individuals (Altmann 1974) by multiplying the area1coverage by the average using a watch with a sweep second hand. Each height for that species;(2) the biomass indices at maneuver that appeared to result in a prey cap- each sample point were proportioned to add to ture (or food intake in the caseof frugivory) was 100%; and (3) the mean relative availability was characterized by the height of the bird above calculated acrossall points classified as activity ground, plant species,prey substrate,living/dead areas. condition of both plant and substrate, and maneuver type. Maneuver types were classified as TIME BUDGETS AND ENERGETICS excavating(stripping bark and digginginto woody Diurnal time budgets were estimated using substrates),twig-splitting (cracking open a twig watches with a sweep second hand or a digital by forcibly closing the mandibles on it), gleaning second readout to record the duration of each (picking from a surface), probing (inserting bill activity of continuously observed birds on the into a floret or leaf cluster), or plucking (picking Hanawi study area. Activities were classifiedinto and eating a fruit). Fruits cut or crushed open to these mutually exclusive categories:quiet perch- obtain insectswere classifiedas excavated.Where ing, pedal locomotion (hopping, climbing), flight, possible, the foraging site was examined, and foraging, self-maintenance (bill wiping, scratch- depth of the excavation measured with a dull ing, defecating),and vocalization. I attempted to probe. I attempted to determine the identity of differentiate foraging activity into active visual 28 STEPHEN MOUNTAINSPRING

the tribes Psittirostrini and Hemignathini that ranged from 10 to 40 g, and from the wing arc lengths given in Amadon (1950). Regressionsto other body dimensions (culmen, tarsus, tail) and nonlinear effectsgave equivalent or poorer fits. Wing length was used becauseof its high repro- ducibility. This proceduregave a predicted mean body mass of 24 g for male birds and 17 g for females, with a standard error of estimate of 1 g. The mass of a male bird mist-netted in March 1984 in Kipahulu Valley was 20 g (M. A. Stem- mermann Kjargaard, pers. comm.).

RESULTS DISTRIBUTION FIGURE 2. Measurementof upper mandible cur- In the present study, Maui Parrotbills were found vature(UMC). to be restricted to montane rainforest on the northeast slopes of Haleakala, with the distri- butional range centered on the eastern Wailua searching,actual prey excavation (including twig- and Hana units (Fig. 3). The HFBS sampled 203 splitting and gleaning), and processingafter cap- stations in range and recorded 30 systematicand ture. 11 incidental detections of individual birds. The Energy expenditure in kcal/day was estimated distribution of records in each unit will be dis- by using the regression equation of Kendeigh cussedbelow. (1970) for body mass (24-g male, 17-g female) In the Waikamoi unit, eight sightings were and mean ambient temperature (I 1.Z°C). The net made during the 1980 to 1984 period in mesic energy intake for prey items captured by exca- ohia, wet ohia, and koa-ohia forest. The 1980 vation was calculated from borers extractedfrom HFBS recorded only two birds on the 43 stations feeding sites and nearby sites of similar appear- surveyed in this portion of the range, suggesting ance. that densitiesin this unit were lower than in areas east of Koolau Gap. On 3 May 1984 at 1,585 m MORPHOLOGY elevation I observed a female bird in koa-ohia Morphological dimensions were measured on forest only 400 m from the open range boundary specimens in the Bemice Pauahi Bishop Mu- to the west. No family groupshave been recently seum, Honolulu, the American Museum of Nat- observed in the Waikamoi unit, suggestingthat ural History, New York, and the Academy of breeding may be sporadic west of Koolau Gap. Natural Sciences, Philadelphia, following the Around 1900, Maui Parrotbills were observed procedures of Baldwin et al. (193 1). Measure- primarily in koa-ohia forest in the Waikamoi ments were made of the culmen (length to feather unit (H. Palmer in Rothschild 1893-1900; Per- bases), length of the lower mandible along the kins 1895, 1903; Henshaw 1902), probably in- midline, width and depth of the upper mandible cluding areas that have since been deforested. at the nares, upper mandible curvature (maxi- These early observers considered the speciesto mum perpendicular distance from the cutting be rare and local, and were apparently unaware edge to a plane positioned across the gape and that the speciesalso occurred in ohia forests to intersecting the bill tip; see Fig. 2), wing length the east (Bank0 1968). along the chord, length of middle tail feathers, Owing to its remoteness, the Wailua unit has tarsal length, length of middle toe without claw, seldom been visited by ornithologists. Maui Par- and length of hind toe claw. I estimated the body rotbills occurred in mesic and wet ohia forest in mass of Maui Parrotbills by regression. The this area. The HFBS recorded 16 birds on the 47 equation, Body Mass (g) = -53.9 + 1.026 Wing stations in range, with densities in the eastern (mm), r = 0.95, was calculated from the live portion of the unit (13 birds on 30 stations) ap- massesof five specieson the island of Hawaii in parently being the highest in the range of the MAUI PARROTBILL BIOLOGY 29

- range limit Hanowl study area ---cliff edge mesic subalpine scrub n I

meet ohia forest x other 1967-65 record

, unverified literature koa-ohia forest m record

exotic forest

FIGURE 3. Rangeof the Maui Parrotbill,showing recent records and habitattypes.

species.The only previous published record from mesic subalpine scrub, mesic ohia forest, and wet this area was by Richards and Baldwin (1953). ohia forest. Family groups were found 0.8 to 1.O km north Maui Parrotbills were first noted in the Ki- and northeast of Puu Alaea in April 1979 (T. A. pahulu unit by Palmer in 1892, presumably Burr, pers. comm.) and June to July 1980. around 1,000 to 1,300 m elevation in koa-ohia In the Hana unit, the HFBS recorded 11 birds forest (Rothschild 1893-1900). Only one bird on the 57 stations in range, and in the Hanawi was sighted during the 1967 Kipahulu Valley study area I recorded 6 birds on 28 stations in expedition, at 2,000 m elevation in mesic ohia May 198 1. Frequent visits by trained observers forest (Bank0 1968), and the HFBS recorded one to the Hanawi and Wai Anapanapa areas have bird on 56 stations in range, at 1,825 m elevation resulted in numerous sightings(Casey and Jacobi in mesic ohia forest. Conant (198 1) failed to find 1974, Shallenberger 1974, Scott and Sincock the speciesbelow 1,980 m elevation in the valley, 1977, Conant 1981, Carothers et al. 1983, Co- except for an unverified aural detection of a bird nant and Stemmermann Kjargaard 1984, Scott at 1,700 m elevation near the south wall (Conant et al. 1986). Family groups were found above and Stemmermann 1980). Despite this apparent Kuhiwa Valley in May 1980 and in the upper rarity and restriction to higher elevations in Ki- Hanawi area in June 1974 (T.L.C. Casey, pers. pahulu Valley, however, expeditions in March comm.), June to July 1983, July 1984, May 1985, 1983, July 1983, and March 1984 by the Na- and November 1985. Maui Parrotbills appeared tional Park Service reported 13 sightings of in- to have their greatest elevational range in the dividual birds and family groups, from 1,250 to Hana unit, occurring from 1,300 to 2,150 m, in 2,070 m elevation in mesic ohia, wet ohia, and 30 STEPHEN MOUNTAINSPRING

TABLE 1. Distributionof Maui Parrotbillrecords with respectto pig damagecategories.

Maui Parrotbill Damage category Stations in range records

None 22 3 Slight 48 12 Moderate 61 6 Heavy 58 4 Severe 14 0

Elevataon (m) FIGURE 4. Elevationaldistribution of theMaui Par- rotbill (1980 HFBS data). of the Maui Parrotbill probably included extensive areas of dry forest on Maui, Molokai, and koa-ohia forest (C. P. Stone and P. C. Banko, possibly Lanai and Kahoolawe. Pleistocenecon- unpubl. data). These sightingsindicate that birds nection of these islands (Steams 1966) would move down into Kipahulu Valley in some years, have facilitated such a distribution. probably from the Hana unit. A similar movement into the area in 1978 to 1979 was noted LARGE-SCALE HABITAT RESPONSE for Crested Honeycreepers (Palmeria dolei) Recent records of Maui Parrotbills were distrib- (Conant 198 1). uted from 1,250 to 2,150 m elevation. The HFBS The only historical recordsof Maui Parrotbills data showed that the birds were three times as outside the range described above were a sec- abundant at 1,750 to 2,000 m elevation as at ondhand report of a single bird in Kaupo Gap 1,400 to 1,750 m (Fig. 4). In Kipahulu Valley in 1928 (Munro 1944), and an unverified aural and perhaps elsewhere, substantial year-to-year detection at 2,000 m elevation near Kuiki Peak differences occurred in elevational distribution (Conant and Stemmermann 1980). Maui Par- that may have representedsome seasonalmove- rotbills probably occur at least occasionally in ment. the ohia and koa-ohia forests between Kaupo About 80% of the range was wet ohia forest, Gap and Kipahulu Valley on the Manawainui 14% mesic ohia forest, 5% mesic koa-ohia forest, Planeze and Kuiki Peak, although recent surveys and 1% mesic subalpine scrub. Habitat type was failed to find them there (Stemmermann 1976, determined for 35 Maui Parrotbill records made Scott et al. 1986). during the HFBS. The habitat types of these rec- Based on vegetation remnants and historical ords were compared with the habitat of all sta- accounts(Rock 19 13:2 l-22), the leeward south- tions in the geographicrange; no significant dif- ern and northwestern slopes of Haleakala (Ka- ference(Kruskal-Wallis test, P > 0.75) was found hikinui, Auwahi, and Kula areas) formerly sup- in the frequency of occurrencein the three prin- ported a variety of dry to mesic foreststhat were cipal habitat types that constitute 99% of the probably inhabited by Maui Parrotbills before range. Maui Parrotbills occurred infrequently in human contact in ca. 400 A.D. Subfossilremains the fourth habitat type, the subalpine scrub im- have been found at 200 m elevation in the Ka- mediately above the treeline. Throughout the hikinui area and at 300 m elevation in Puu Naio range, both overstory and understory were dom- (H. F. James, pers. comm., Fig. 1). Even more inated by native plant species.Maui Parrotbills significant were a fossil quadrate and mandibular were absent from areas adjacent to the range in ramus with articulation dated about 3,000 B.C. the Waikamoi unit that were dominated by ex- found in dune deposits near Ilio Point on west otic trees. Molokai (Olson and James 1982b). Drier low Pig damage to the vegetation structure in the elevation forestsoriginally supported a rich flora range of the Maui Parrotbill varied from vir- and avifauna, but Hawaiians burned and cleared tually none to severe, with 82% of the HFBS extensive areas of these forests, causing large- stations falling in the three intermediate cate- scale extinctions and range truncations among gories (Table 1). Comparing this spectrum with the native biota (Zimmerman 1948:172, Kirch actual occurrencesof Maui Parrotbills in the vi- 1982, Olson and James 1982a). The original range cinity of rated stations showed that the species MAUI PARROTBILL BIOLOGY 3 1

the ground cover (Table 2). Activity areastended to have greater ohelo cover, less ohia, pukiawe, and akala cover, and greater canopy height than nonactivity areas, as shown in a stepwise discriminant function analysis (F = 6.04; df = 5, 31; P < 0.001). Two a posterior-i divisions of plant coveragewere significant: no activity area had < 10% ohelo cover, but 10 nonactivity areas did (x2 = 9.3, df = 1, P < 0.01); no activity area had >65% ohia cover, but 10 nonactivity areas did (P < 0.01). The vertical distribution of foliage in activity areas had greater understory and less canopy fo- FIGURE 5. Mean foliage height profiles in activity liage than in nonactivity areas (Fig. 5). For l-m areasand nonactivity areas in the Hanawi study area. height increments, foliage volume in the 0- to 6-m height interval averaged 4 1% greater in activity areas compared to nonactivity areas (x2 = tended to occur more frequently in areas with -2 In P = 33.4, df = 12, P < O.OOl), and in the less pig damage (Mann-Whitney U-test, P < 6- to 12-m interval 41% less (x2 = 24.9, df = 12, 0.05),with 84% of the recordsfalling in the three P -c0.02). Maui Parrotbills thus tended to utilize lightest categories of pig damage. Occasionally areaswith more developedunderstories and more birds did occur in areas of heavy to severe dam- open canopies. age. Incidental observations suggestedthat Maui Parrotbills were relatively sedentary in their dai- SMALL-SCALE HABITAT RESPONSE ly movements. In the Hanawi study area, I tracked In the Hanawi study area, 15 activity areas and individual birds and family groupsduring breed- 22 nonactivity areas were identified from sys- ing and nonbreeding periods for as long as 2 hr tematic surveys and incidental observations of and found that birds tended to stay within areas foraging birds. In both types of areas, ohia dom- 200 to 300 m in diameter. On subsequentdays, inated the canopy; pilo, olapa, ohelo, and pu- what appeared to be the same birds or family kiawe the understory; and graminoids and ferns groupswere found in the vicinity of the last sight-

TABLE 2. Cover and height characteristicsof activity and nonactivity areas.

Plant ccner(%) Activity Nonactivity Discriminant Height (III) areas areas function Habitat component x SD R SD P SD (&statistic)

Canopy height (m) - - 9 2 10 3 2.49 Crown cover - - 57 17 57 29 - Woody plants Akala 0.4 9 8 16 19 -2.98 Alani :.i 1.6 2 3 2 3 - Kanawao 1:6 0.4 1 2 3 7 Kawau 4.3 1.7 1 1 2 1 Kolea 3.3 1.5 2 : 2 4 - Naenae 2.0 1.0 1 2 4 Ohelo 3.1 0.9 17 ; 10 6 2.01 Ohia 8.7 2.3 40 18 52 30 -3.54 Olapa 3.1 0.9 12 15 10 7 - Pi10 2.9 1.1 20 16 18 15 - Pukiawe 2.6 0.8 11 14 17 18 -3.50 Ground cover Ferns 0.6 0.4 36 17 34 18 - Graminoids 0.2 0.1 22 17 25 15 - 32 STEPHEN MOUNTAINSPRING

TABLE 3. Prey capturesof the Maui Parrotbillby substrateand maneuver.

Live Dead-on-live Dead CZlk@.?ry Branch Twig Leaf Flower Fruit Branch Twig Branch Total Plant species Ohia 28 5 5 3 - 43 3 94 Olapa 4 9 4 - 15 : 56 Ohelo 5 24 8 - 9 224 - - 50 Pi10 4 15 8 - 3 16 - - 46 Kanawao - - 25 - 1 - 26 Alani ; - - - - 18 2 - 23 Kolea 3 4 - - - 8 1 - 16 Koa 2 1 8 - - - - - 11 Oheohe 4 2 3 - - - - 11 Akala ii - - - - 2 - 4 Kawau z - - - - - 2 - 4 Naenae 1 1 1 - - 1 - - 4 Hoawa - - - - - 2 - - 2 Manono ------2 - 2 Pukiawe - - - - - 1 - - 1 Maneuvers Excavating 30 - - - 28 106 - 9 173 Gleaning 26 44 28 3 - 9 - - 110 Twig-splitting - 19 - - - - 28 - 47 Probing - - 8 3 - - - - 11 Plucking - - - - 9 - - - 9 Total 56 63 36 6 37 115 28 9 350

ing. Perkins (1895) too noted that the birds did major woody speciespresent (Table 3). The most not make long-distance flights. If the interpre- frequently visited specieswas the dominant tree, tation of relative sedentarinessis accurate, then ohia (27%) but 70% of the prey captures were it seems reasonable to conclude that Maui Par- made on subcanopy treesand understory shrubs, rotbills actively select the most suitable habitat especially olapa, ohelo, pilo, alani, and kanawao. sites in an area to concentrate their foraging ef- In addition to the speciesin Table 3, Maui Par- forts. They may thus be more sensitive to adverse rotbills have also been reported to feed on olo- local habitat modification than speciesthat reg- mea (Perrottetia sandwicensis) (Berger 198 1). ularly move several kilometers on a daily basis, Henshaw (1902) and Perkins (1903) found that such as the Iiwi (Vestiaria coccinea), Crested birds fed extensively on koa trees, probably re- Honeycreeper, or Apapane (Himatione sangui- flecting the many borers associated with koa nea). (Swezey 1954) and the location of their observations in koa-ohia forest in the Waikamoi unit. FORAGING SUBSTRATES Compared to available plant biomass in the Maui Parrotbills foraged mostly in the subcan- Hanawi study area, Maui Parrotbills fed on ohia opy and understory at a mean height of 3.7 + 50% and on pukiawe 93% less often than ex- 2.4 SD m (n = 289) above ground; 4% of the pected, but they fed on olapa 89%, ohelo 44%, prey captures were at 0 to 1 m height, 54% at 1 and pilo 36% more often than expected (Chi- to3m,16%at3toSm,22%at5to8m,and square tests,P -C 0.05, Table 4). The most strik- 3% above 8 m. For heights > 1 m in the Hanawi ing preferences were for the relatively uncom- study area, the height distribution of foraging did mon alani and kolea, where feeding was 10 and not differ significantly from the height distribu- 5 times more frequent than expected, respec- tion of foliage in activity areas (Kolmogorov- tively. Perkins (1903) also noted frequent use of Smimov test, P > 0.05), suggestingthat foraging alani. The apparent feeding preferencesprobably was random with respectto available foliage vol- reflectedthe abundance of timber-boring insects. ume at different heights. For example, several timber-borers that infested Maui Parrotbills captured prey on virtually all pilo left characteristic holes in dead branches of MAUI PARROTBILL BIOLOGY 33

TABLE 4. Distribution of prey captures by plant per mandible into a bark deformity, such as a specieson the Hanawi studyarea comparedto biomass small crack, crevice, or exposed borer tunnel. indices in activity areas;* indicates P -C 0.05 ** P < 10-S. They quickly dug in the hook to enlargethe open- ing, graspedbark laterally with both mandibles,

Prey cap Biomass pried on the bark, and chippedor peeledoff pieces Plant species t”reS (%) W) x2 ddkenca of bark and wood 1 to 3 (rarely to 5) cm long. Ohia 21 54 34.1** When birds uncovered a burrow, they tracked it Olapa 19 8 20.1* by peeling up the bark until coming to a frass Pi10 18 13 4.3* deposit, dug in with their upper mandible, and Ohelo 16 11 extractedthe larva or pupa using their mandibles Alani 7 1 1,;.;I* Kolea i.4 1 34:4* and tongue. The upper mandible was often used Pukiawe 5 12.6* to plough up and scrapeout old wood or to peck All others 8 4 7.5* like a small woodpecker. The lower mandible

sCalculated from a sampleof 253 PRY captures was used mainly in conjunction with the upper to grasp bark. Branches were often stripped of their moss or lichen cover in the searchfor prey. live trees, and borer sites were found on 29 of On one occasion a bird fully opened its bill and 50 mature pilo plants located within 10 m of a then hammered with the lower mandible, fol- randomly placed transect at 2,100 m elevation. lowed by probing with the upper; this pattern is Similar high infestation levels appeared on alani virtually identical to the distinctive common trees. On the other hand, ohia trees have been feeding behavior of the Akiapolaau (Hemigna- reported to have relatively low densities of tim- thusmunroi) (Perkins 1903). Occasionally a bird ber-boring insects(Gressitt and Samuelson 198 l), dug into a site without peeling away the bark, as and I seldom noted infestations on ohia in the on dead trees with little bark, and other times Hanawi study area. peeling alone uncovered the prey. Birds some- About 57% of the observed prey captureswere times assumed inverted postures while peeling made on live substrates,and 4 1% were made on or digging on the side of branches, probably for dead substrates on live plants (Table 3). Exca- a better attack angle or foothold. Usually 30 to vating and twig-splitting presumably took more 60 set was spent at an excavation site. One spec- energy than other maneuvers, and 6 1% of these imen in the Bishop Museum (#4095) has wear were on dead-on-live substrates. Maui Parrot- marks on the upper mandible that appear to have bills made 5 1% of their prey captureson branch- resulted from excavating or twig-splitting. es, 26% on twigs, and 10% on leaves. A similar In twig-splitting (14% of observations), birds reliance on branchesand twigs was noted by Per- typically alit on a small branch and after l- to kins (1903). About 54% of the prey captures on 5-set inspection graspedthe twig perpendicular brancheswere made at siteswith a lichen or moss to the bill symmetry plane and rather quickly cover. The prey captures made by excavating split it open by clamping the bill shut. Sometimes larvae from fruit (8%) were mostly observed in they quickly picked at the twig stub with their November 198 5, when birds were exploiting in- upper mandible or crackedthe twig again farther festationsin kanawao fruit-on seven randomly- toward the base. When splitting stubborn twigs, chosenplants at 1,650 m elevation, 50% (f27% they frequently used strong back-and-forth SD) of the ripe berries had been excavated (C. movements of the head (“wrenching”) for ad- B. Kepler, pers. comm.). Infrequently (2% of all ditional shearing force. Twigs were also grasped observations) birds probed ohia flowers or parallel to the mandibular edge and split longi- gleaned in oheohe flowers for nectar or insects. tudinally. Twig-splitting and wrenching struck Ripe and green fruits of ohelo and pilo were oc- Henshaw (1902) and Perkins (1903) as the most casionally eaten (3% of observations). impressive behavior of the Maui Parrotbill. Gleaning (32% of all maneuvers) appeared to FORAGING MANEUVERS be an opportunistic maneuver. Perkins (1903) Excavating constituted 49% of all prey captures reported that Maui Parrotbills fed caterpillars to (Table 3) and was also noted by Perkins (1903) their young during the breeding season,suggest- as being a common maneuver. Typically, exca- ing that gleaning may be important then. Glean- vating birds began by inserting their hooked up- ing constituted 63% of all maneuvers in July 1983 34 STEPHEN MOUNTAINSPRING when most birds were in family groups, but only TABLE 5. Time budget of the Maui Parrotbill. 18% in May 198 1 when no signsof breeding were noted. Behavior Seconds % Probing (3% of all maneuvers) was used to Quiet perching 1,642 27.3 explore ohia flowers and deformed, rust-infested Pedal locomotion 742 12.3 koa leaf clusters. Plucking (3%) was used to de- Flight 110 1.8 scribe fruit taking, but otherwise resembled Foraging gleaning. Visual searching 587 9.7 Recurring patterns in search behavior sug- Excavatinga 2,357 39.1 gestedthat Maui Parrotbills cued in on recently Processing 106 1.8 dead branches with loose bark and on bark de- Self-maintenanceb 32 0.5 Vocalization 449 7.5 formities that indicated probable borer activity. Total 6,025 100.0 Perkins (1903) noted that both this speciesand aIncludes a small proportionof twig-splitting and gleaning the Nukupuu (Hemignathus lucidus) had sys- bBill wiping, defecating,preening, scratching. tematic, methodical patternsof searchingfor prey, suggestingthe use of some sort of searchimage. substantial individual and seasonal variation doubtless occurs, one general basis of the slow TIME, RESOURCES, AND ENERGETICS foraging rate and long rest periods is that an av- The largestexpenditure in the diurnal time bud- erage prey capture appears to yield about 1% of get (Table 5) was the time spent capturing prey the daily energyrequirements. The metabolic rate by excavation and twig-splitting (39%). Pedal lo- (and hence energy requirements) may moreover comotive activity was often associatedwith ac- be lower than expectedfrom the regressionequa- tive feeding. During active foraging Maui Par- tion, similar to the physiologicaladaptation found rotbills averaged 20 ? 12 SD set between prey among other Hawaiian honeycreepersto cold wet captures (n = 1 lo), but between foraging bouts environments (MacMillen 198 1). they often restedfor 3 to 10 min, and quiet perching accountedfor 27% of the diurnal time budget. Maui Parrotbills often fed from dead sub- MORPHOLOGY AND SEXUAL DIMORPHISM strateson live plants where borers were frequent. Morphological adaptations that probably facili- Prey items included the larvae and pupae of bee- tate excavating and twig-splitting included the tles and moths: Plagithmysussp. (Cerambycidae) extremely stout, sharply decurved bill (Fig. 2) on ohia and alani, Clytarlus spp. (Cerambycidae) and relatively strong and well-developed legs on koa, Proterhinussp. (Aglycyderidae) on ohia, compared to other honeycreepers (S. Moun- Nesotocusnewelli (Curculionidae) on olapa, and tainspring, unpubl. data). Each mandible ends in microlepidopterans, probably Hyposmocomasp. a sharp point that was frequently inserted into (Hyponomeutidae), on pilo, and Heterocrossasp. borer tunnels. The slight downward flaring of the (Carposinidae) on kanawao. The insects extract- cutting edge at the middle of the lower mandible ed were moderately large, generally 8 to 20 mm appeared to bring greater shearing force to bear long. A sample of 20 borer larvae and pupae had during twig-splitting and to improve the grasp an average mass of 0.2 g (-to.1 SD) per borer. during bark-peeling (S. Mountainspring, pers.ob- Assuming that borers averaged 1.2 kcal/g wet serv.). The internal morphology is poorly known, weight of metabolizable energy like other insects, and a few intact fluid, skeletal, and frozen tissue and that Maui Parrotbills averaged 66% assim- specimens should be collected, examined, and ilation efficiency like other passerines (Ricklefs preserved for further study. S. L. Olson and H. 1974), an averageprey item contributed approx- F. James (pers. comm.) found that the cranial imately 0.15 kcal (+- 0.10 SD) to the energy bud- morphology of this species “is unique among get. passerinesin being highly convergent with that Based on regressionpredictions, the energetic of parrots and that examination of the dried jaw expenditure of male birds was about 18 kcal/day musculature remaining on the skull indicates that and females 15 kcal/day, and hence 100 to 120 it too is peculiar.” average-sizeprey items would be needed per day, Amadon (1950) reported that Maui Parrotbills or an average of one prey excavation every 5 to had the most extreme sexual dimorphism in bill 8 min, for a balanced energy budget. Although length of the Hawaiian honeycreepers.Males were MAUI PARROTBILL BIOLOGY 35

TABLE 6. Morphological dimensions of the Maui TABLE 7. Sexualdifferences in the foragingbehavior Parrotbill;* indicatessignificant different between sexes, of the Maui Parrotbill; * indicates P < 0.05, ** P i Mann-Whitney U-test, P < 0.05. 0.001.

Male/ Percent of prey Male FeIlIde female captures Dimension (mm) (n = 11) (n = 5) ratio Male FUllale Chi-square Category (n = 131) (n = 102) test Upper mandible length 21.5 17.4 1.24* Upper mandible width 6.4 5.7 1.11* Plant species 45.0** Upper mandible depth 8.8 7.4 1.12* Ohia 23 23 0.0 Upper mandible curvature 4.6 3.6 1.30* Ohelo 22 7 8.7* Lower mandible length 14.6 12.1 1.20* Pi10 22 5 11.7** Wing chord length 73.1 66.5 1.10* Olap 8 22 7.0* Tail length 42.6 37.6 1.13* Kanawao 3 22 17.6** Tarsus length 22.2 20.9 1.07 All others 21 23 0.0 Middle toe length 12.8 12.0 1.07 Middle toe claw length 6.1 5.5 1.11* Substratecondition 5.8 Hind toe claw length 7.3 6.6 1.12* Live 73 60 1.4 Dead-on-live 23 37 4.0* Dead 5 3 0.4 larger in bill, wing, tail, and pedal morphology Substratetype 7.2 than females (Table 6). Assuming that the male/ Branch 39 39 0.0 Twig 28 29 0.0 female ratio is proportional to functional differ- Leaf 18 8 4.6* entiation, bill length and curvature are particu- Flower 2 3 0.1 larly important, with males having longer, more Fruit 12 21 2.5 deeply hooked bills. Maneuver 20.4** Sexes differed in foraging behavior (Table 7) Excavating 45 49 0.2 as to tree species(x2 = 45.0, df = 5, P -c 0.001) Gleaning 31 ::, 0.0 and maneuvers (x2 = 20.4, df = 4, P < 0.01) Twig-splitting 9 4.6* 8 0 8.6* but not in substrate condition (e.g., live, dead) Plucking Probing 8 0 7.0* (x2 = 5.4, df = 2, P > O.OS),substrate structure (branch, twig, etc.) (x2 = 17.2, df = 4, P < 0.01) Foraging height (m) 4.2 3.5 or foraging height (t = 0.9, P > 0.3). Compared Mean SD 2.5 2.1 to females, males tended to use pilo and ohelo more, and olapa and kanawao less; females split twigs more but were not observed probing or plucking (Table 7). The depth of excavation sites by bird 1 who tried to hit bird 2 on the nape with also differed, with 24 male sites ranging 5 to 13 his upper mandible. Secondsafter the chasebroke mm and 22 female sites 3 to 10 mm (Mann- up, bird 2 landed and sang. Bird 1 flew into the Whitney U-test, P < 0.05). Foraging differences air 3 m above bird 2 and both sang; then, a that were probably correlated with the larger bills spectacularaerial chaseensued with bird 1 sing- of males included greater excavation depth and ing, wielding his bill 5 to 40 cm from the head more frequent probing and plucking. As noted of bird 2, and repeatedly diving at and attempt- in other avian taxa, sexual dimorphism in for- ing to strike and grasp the head of bird 2. This aging and morphology results in exploitation of episode ended as bird 2 flew off, but if a bird a greater variety of resources and presumably were to actually grasp the head of another bird reducesintraspecific competition (Selander 1966). death would be a likely result because of the crushing ability of the mandibles. The pro- ANNUAL CYCLE AND BREEDING BEHAVIOR nounced sexual dimorphism of bill size may Few observations exist of Maui Parrotbill breed- partly be a responseto selective pressurearising ing biology. On 19 May 198 5 during the breeding from this sort of aggressivebehavior. season, I observed two males engage in aggres- On 13 January 1985, T. W. Sutterfield found sive behavior suggestingterritorial defense. Fol- the first positively identified Maui Parrotbill nest. lowing a spirited bout of advertising songby both The nest was located at 1,800 m elevation in the birds, bird 1 dived at bird 2 and initiated a chase Hanawi study area, and was placed 11 m high sequence during which both birds continued in a fork formed by the branches and branchlets singing.Bird 2 landed, sang,and was chasedagain of the upper crown of an ohia tree on a gulch 36 STEPHEN MOUNTAINSPRING edge. The nest was a substantial structure about uary to June in various years. Nesting records 18 cm in diameter, of cup-like shape, with li- over a prolonged seasonare typical of lower lat- chens and leaves hanging on the outside, and itudes (Baker 1938), but in a given year pairs generally constructedfrom small twigs and other may tend to be synchronized by climatic events woody material. The contents could not be de- suchas the start of the wet season(Fogden 1972). termined because the nest was inaccessible. Family groups usually comprised two imma- Whether this was a cock or true nest is unknown, ture birds being cared for by two adults; however, becausethe next day a storm with winds in excess as many as four young occurred in a single fam- of 30 m/set struck and field work was terminat- ily, and occasionally one or two immatures were ed. cared for by only one adult, male or female. Sutterfield (pers. observ.) also observed nest The discoverer of a nest should note approx- building behavior at this nest site. While the pair imate age of nestlings and nest soiling, as this was in the crown of an ohia tree, the male bird would help in determining the systematic posi- picked up a piece of sedgegrowing on the tree, tion of the Maui Parrotbill. In nest sanitation, hopped around in the canopy displaying it to the the Psittirostrini do not removing fecal sacsdur- female for 10 to 15 set, and took 10 set to place ing the first 10 days after hatching, whereas the it in the nest. Then for 30 set the two birds Hemignathini do (van Riper 1980). chipped almost incessantlywhile hopping around within 15 cm of one another. The male then VOCALIZATIONS pulled up a small clump of Usneulichen, hopped Maui Parrotbills have three distinct, fairly ste- around with it for 30 set, displaying it to the reotyped vocalizations: a chip, a call, and a song. female (who frequently chipped), and deposited The chip was a sharp atonal “chick” that was it in the nest. The pair hopped about in close virtually inseparable from the chip of the Maui proximity, chipping to each other, for 3 to 4 min Creeper. The chip note was usually given when near the nest site, before flying out of the tree as Maui Parrotbills were in loose flocks with Maui two Apapanes arrived. On 9 January 1985, A. M. Ecton and P. A. Stine noted a male bird car- Creepers, and probably promoted flock cohe- rying Usneain its bill elsewherein the study area. sion. The call was a somewhat thin, upslurred This behavior may indicate nest building activ- whistle clearer than the similar notes given by ity. the Common Amakihi (Hemignathus virens), The only other report of a Maui Parrotbill nest Iiwi, and Crested Honeycreeper. This vocaliza- was a possiblerecord by R.C.L. Perkins for April tion was often given by two birds at some dis- or May 1894 (Greenway 1958). Perkins (1903: tance apart, and probably servedas contact notes. 432) reported that the nest was built in a koa It was given by apparently paired birds and by tree, “in the fork formed by a branch and the adults and young in apparent family groups. main trunk about [8 m] from the ground. The The song consisted of 4 to 10 short upslurred tree was covered with grey lichens and the nest whistled notes given consecutively without break was well-concealed, being itself covered with the and tending to descendin pitch. Only male birds same. It was of simple cup-like form, resembling have been observed singing. During this study, in this respectthe usual Drepanid nest. For sev- songswere uttered singly or continually with 5 eral days two old and two young birds, just able to 60 set breaks for as long as 5 min. Little vari- to fly, were constantly seen in this tree, and I feel ation over the course of the day was observed in sure that they must have been the former oc- the frequency of any vocalization, except that cupants of the evidently newly deserted nest.” birds appeared to be somewhat quieter at dawn, This possible nest was similar to the 1985 nest dusk, and during rainy periods and warm sunny in its fork location, cup-like form, and placement afternoons. Other vocalizations occasionally of gray lichens on the exterior. heard were solicitation notes by young birds and Records of immature birds in family groups various short burry call notes rarely given. soliciting food from adult birds occurred from March to September, with 3 to 7 recordsper field MIXED-SPECIES FLOCKING trip from May through August, and one or two Maui Parrotbills frequently formed loosely as- records/trip for March, April, September, and sociated flocks with Maui Creepers. On 20 of 39 November. This suggestsa protracted breeding encounters (5 1%) one or two (rarely as many as period, with nesting occurring perhapsfrom Jan- four) Maui Parrotbills were accompanied by two MAUI PARROTBILL BIOLOGY 37

or three (up to six) Maui Creepers. Flocking ap- range lies on private, state, and federal preserves peared to be facilitated by the nearly identical with some degree of legal protection. chip notes of the two species.On seven occasions At lower elevations, the range limits of the the Maui Creepers left the Maui Parrotbills be- Maui Parrotbill have a striking coincidence with hind. When foraging, Maui Parrotbills changed the upper elevation limits of mosquito distri- perches every 5 f 3 SD set (n = 79), whereas bution (Fig. 3). The Nukupuu, Maui Creeper, Maui Creepers changed perches more rapidly, Akepa (Loxops coccineus),Iiwi, Crested Hon- every 1 * 1 SD set (n = 49; Mann-Whitney eycreeper, and Poo-uli have similarly restricted U-test, P < 0.001). The faster movement rate of distributions (Scott et al. 1986). The role of mos- Maui Creepers seemed to be compensated by quitoes in serving as vectors of avian malaria longer flights of Maui Parrotbills when the two (Plasmodium relictum capistranoae) and most speciesforaged together. On rare occasionsa Nu- likely avian pox (Poxvirus avium) probably re- kupuu or Poo-uli (Melamprosops phaeosoma) stricts the ranges of speciesthat have little resis- flocked with Maui Parrotbills (Perkins 1903; tance to these diseasesto areas with zero to very Shallenberger 198 1; S. Mountainspring, unpubl. low mosquito densities (Warner 1968, van Riper data). The selective force for flocking may be et al. 1982, Scott et al. 1986). predatordetection, because Maui Parrotbills often Habitat degradation by feral pigs is the most direct most of their attention to excavating. Sev- insidious and threatening factor that limits Maui eral times when Short-eared Owls (Asio jlam- Parrotbill numbers. Maui Parrotbills tend to oc- meus) quartered over the rainforest, all passer- cur on sites with significantly lower pig damage ines including Maui Parrotbills became very quiet than the landscape norm, and the foliage profile and ducked into the vegetation. Before Polyne- they prefer to forage in is typical of areas un- sian contact, predation from now-extinct accip- damaged by pigs. Moreover, the relative seden- itrine hawks and an undescribed fossil genus of tariness of the birds would tend to make them long-leggedowl were probably important causes more sensitive to localized adverse habitat mod- of mortality of small land birds in the Hawaiian ification. Rooting by feral pigs destroys under- Islands (Olson and James 1982b). story and ground cover, most significantly the seedlingsof the shrub and subcanopy trees (Gif- DISCUSSION fin 1978, Baker 1979, Tisdell 1982) where most prey are captured. Other rare bird species ad- LIMITING FACTORS AND MANAGEMENT versely impacted in the Hanawi area by pigs are The primary limiting factors of the Maui Par- the Poo-uli and Nukupuu, whose distributions rotbill appear to be: (1) habitat loss; (2) avian correspondto areas of lower pig damage (Moun- disease; (3) habitat degradation: (4) predation; tainspring 1986) and Bishop’s 00 (Moho bish- and (5) competition from exotic species.The first opi), which prefersto feed on the nectar of fragile two factors act mainly to limit the range; the last understory lobeliads (Perkins 1903, Sabo 1982). three tend to act within the range by depressing On steep slopes in the montane rainforest, pig densities (Mountainspring 1986). damage leads to excessivesoil erosion; on more Prior to Polynesian contact around 400 A.D., level sites, habitat degradation gradually con- widespreaddryland forestssupported a rich bio- verts mature forest to mire. The urgently needed ta that included the Maui Parrotbill (Olson and management action is to control pig populations James 1982b). Burning, clearing, lumbering, in substantial areasof montane rainforest by im- ranching, and feral ungulate activity causedlarge- proving public hunting access,and in remote key scale extinctions and retreats among this biota. areas by fencing, driving, intensive hunting, and Most of the original range of the Maui Parrotbill possibly poisoning (Kepler et al. 1984). It is un- has been converted to exotic vegetation, which known to what extent predation limits Maui Par- the speciesdoes not inhabit. Although birds for- rotbill numbers. In the high elevation rainforest merly occurredon the Kahikinui Tract, that area where bird densitiesare greatest,mongoose (Her- is now too severely degraded to support them. pestesauropunctatus) and feral cats (Felis catus) Protection of the remnant koa forestsin this area are relatively uncommon, but Polynesian rats by feral goat (Capra hircus) control would en- (Rat&s exulans) and black rats (R. rattus) occur hance the potential for transplanting birds and at moderate levels of abundance (R. T. Sugihara, foster possibly spontaneousrecolonization from pers. comm.). Stomachsof both rat specieshave the main population. Virtually all of the present contained bird feathers in Waikamoi Preserve 38 STEPHEN MOUNTAINSPRING

(R. T. Sugihara, pers. comm.) and Kipahulu Val- AMERICANORNITHOLOGISTS UNION. 1983. Check- ley (C. P. Stone, pers. comm.). list of North American birds. 6th ed. American Ornithologists’ Union, Washington, DC. Indirect competition for food resources may BAKER,J. K. 1979. The feral pig inHawaii Volcanoes occur from introduced insect species that prey National Park. D. 365-367. In R. M. Linn led.1. on timber-boring insects. Yellow jackets (Ves- Proceedingsof the first conference on sciehtmc pula pennsylvanica)are common in some years researchin the national parks, New Orleans. U.S. in the Hanawi study area; Ischiogonuswasps that Natl. Park Serv. Trans. Proc. Ser., No. 5. BAKER,J. R. 1938. The relation between latitude and parasitize cerambycid beetles (Zimmerman 1948) breeding seasonsin birds. Proc. Zool. Sot. Lond., may also affect the resource base. A quantitative Ser. A. 108:557-582. study ofresource availability and probable exotic BALDWIN,S. P., H. C. OBERHOLSER,AND L. G. WORLEY. impacts would be helpful in assessing the role of 1931. Measurements of birds. Sci. Publ. Cleve- land Mus. Nat. Hist. 2:1-165. food limitation and competition. BANKO,W. E. 1968. Rediscovery of Maui Nukupuu, Hemignathuslucidus ajinis, and sightingof Maui ACKNOWLEDGMENTS Parrotbill, Pseudonestorxanthophrys, Kipahulu A grant from the Frank M. Chapman Memorial Fund Valley, Maui, Hawaii. Condor 70:265-266. of the American Museum of Natural History led to the BERGER,A. J. 1981. Hawaiian birdlife. 2nd ed. Univ. initiation of this study, and the continued support of Press of Hawaii, Honolulu. the U.S. Fish and Wildlife Servicethrough the Limiting BLUMENSTOCK,D. I., AND S. PRICE. 1967. Climates Factors Project made possible this report. The stim- of the states:Hawaii. Climatography of the United ulating discussionsand gracious hospitality of C. B. States60-5 1. U.S. Dept. Commerce, Washington, and A. K. Kepler enhanced my field experiences on DC. Maui. I especiallyextend my thanksto the investigators BOCK,W. J. 1970. Microevolutionary sequencesas of the Hawaii Forest Bird Survey who made much of a fundamental concept in macroevolutionary the presentanalysis possible, to D. Amadon, A. J. Ber- models. Evolution 24:704-722. aer. T. A. Burr. J. H. Carothers.T.L.C. Casev.R. Good- CAROTHERS,J. H., S. R. SABO,AND R. B. HANSEN. roe; H. F. James,J. D. Jacobi,C. B. Kepler, S L. Olson, 1983. Ecologicalobservations on an endangered H. D. Pratt, P. Pyle, C. J. Ralph, K. Redman, J. M. species:the Maui Parrotbill, Pseudonestorxan- Scott, and C. van Riper III for constructivecomments thophrys.Am. Birds 37:820-82 1. on earlier drafts, to A. Zeigler of the B. P. Bishop CASEY, T.L.C., ANDJ. D. JACOBI. 1974. A new genus Museum for loan of specimens,to D. Amadon and W. and speciesof bird from the island of Maui, Ha- E. Lanyon ofthe American Museum ofNatural History waii (Passeriformes:Drepanididae). Occas..Pap. for accessto specimens, to P. F. Cannel1 and R. E. Bernice P. Bishon Mus. 24:215-226. Ricklefs for specimen measurements,to C. J. Davis, CONANT,S. 1981. Recent observationsof endangered P. A. Hiaashino. and J. D. Jacobi for soecimen iden- birds in Hawaii’s national parks. ‘Elepaio 41:55- tifications, to J. D. Jacobi for mapping the vegetation 61. in the Maui Parrotbill range, to F. L. Ramsey for dis- CONANT, S., AND M. A. STEMMERMANN.1980. Birds cussionson statistical procedures,to C. P. Stone for in the Kipahulu District of Haleakala National unpublisheddata from the 1983 to 1984 National Park Park. D. 67-75. In C. W. Smith led.1.Proceedines: Serviceexpeditions into Kipahulu Valley, to EastMaui Thirdconference in natural science; Hawaii Vol- Irrigation Co.. Haleakala National Park. Haleakala canoesNational Park. Coop. Natl. Park Res. Stud. Ranch, Hawaii Department of Land and Natural Re- Unit, Univ. Hawaii, Honolulu. sources, and The Nature Conservancy for accessto CONANT,S., ANDM. STEMMERMANNKJAaGAArW. 1984. studvareas. to P. C. Banko. T. A. Burr. J. H. Carothers. Annotated checklistof birds of Haleakala Nation- T.L.C. Casey, L. Cuddihy-Anderson; S. J. Doyle, A: al Park, Maui, Hawaii. West. Birds 15:97-l 10. M. Ecton, E. GagnC,R. B. Hansen, P. A. Higashino, FOGDEN,M.P.L. 1972. The seasonalityand popula- J. D. Jacobi, A. K. Kepler, C. B. Kepler, P. H. Mc- tion dynamics of equatorial birds in Sarawak.Ibis Eldowney, P. Pyle, F. L. Ramsey, J. M. Scott, R. L. 114:307-343. Stemmermann, M. A. Stemmermann Kjargaard, P. A. FORSHAW,J. M. 1978. Parrots of the world. 2nd ed. Stine, C. P. Stone, R. T. Sugihara, T. W. Sutterfield, Lansdowne Press, Melbourne, Australia. P. W. Sykes,Jr., D. M. Taylor, F. R. Warshauer, J. E: GIF~N, J. 1978. Ecologyof the feral pig on the island Williams, and A. Y. Yoshinana for contributina their of Hawaii. Hawaii Dept. Land Nat. Res., Hono- field observations, and to S.,-O., and N. Mot&tin- lulu. spring for their encouragementand support. GREENWAY,J. C., JR. 1958. Extinct and vanishing birds ofthe world. Spec.Publ. 13, American Com- mittee for International Wild Life Protection, New LITERATURE CITED York. GRESSIT~,J. L., AND G. A. SAMUELSON.198 1. Ce- ALTMANN,J. 1974. Observational study of behavior rambycid tree borers, p. 288-290. In D. Mueller- sampling methods. Behaviour 49:227-267. Dombois, K. W. Bridges,and H. L. Carson [eds.], AMADON, D. 1950. The Hawaiian honeycreepers Island ecosystems:biological organization in se- (Aves. Drepaniidael. Bull. Am. Mus. Nat. Hist. lected Hawaiian communities. Hutchison Ross. +5:15i-26i. Stroudsburg,PA. MAUI PARROTBILL BIOLOGY 39

HENRICKSON,J. 1971. Vascular flora of the northeast RICKJ_EF~,R. E. 1974. Energeticsof reproduction in outer slopesof Haleakala Crater, East Maui, Ha- birds. Publ. Nuttall Omithol. Club 15:152-297. waii. Contrib. Nature ConservancyNo. 3, Arling- ROCK,J. F. 1913. The indigenoustrees of the Hawai- ton, VA. ian Islands. Hawaiian Gazette, Honolulu. HENSHAW,H. W. 1902. Birds of the Hawaiian Is- ROTHSCHILD,W. 1893-l 900. The avifauna of Laysan lands, being a complete list of the birds of the and the neighbouringislands. R. H. Porter, Lon- Hawaiian possessionswith notes on their habits. don. Thos. G. Thrum, Honolulu. SABO,S. R. 1982. The rediscovery of Bishop’s ‘0‘0 KENDEIGH, S. C. 1970. Energy requirements for ex- on Maui. ‘Elepaio 42:69-70. istence in relation to size of bird. Condor 72:60- SCOTT,J. M., S. MOUNTAINSPRING,F. L. RAMSEY, AND 65. C. B. KEPLER. 1986. Forest bird communities of KEPLER,C. B., T. BURR,C. B. COOPER,D. DUNATCHIK, the Hawaiian Islands:their dynamics,ecology, and J. MEDEIROS,J. M. SCOTT,M. UEOKA, AND W. conservation. Stud. Avian Biol. 9. WONG. 1984. Maui-Molokai forest bird recovery SCOTT,J. M., AND J. L. SINCOCK. 1977. Recent ob- plan. U.S. Fish Wildl. Serv., Portland, OR. servations on the birds of the Koolau Forest Re- KIRCH, P. 1982. The impact of the prehistoric Poly- serve, Maui. West. Birds 8: 113-l 16. nesianson the Hawaiian ecosystem.Pac. Sci. 36: SELANDER,R. K. 1966. Sexual dimorphism and dif- 1-14. ferential niche utilization in birds. Condor 68: 113- MACMILLEN,R. E. 1981. Nonconformance of stan- 151. dard metabolic rate with body mass in Hawaiian SHALLENBERGER,R. 1974. Field notes. ‘Elepaio 35: honeycreepers.Oecologia 49:340-343. 18-20. MARTINKA,R. P. 1972. Structural characteristicsof SHALLENBERGER,R. J. [ED.]. 1981. Hawaii’s birds. Blue Grouse territories in southwesternMontana. 3rd ed. Hawaii Audubon Society, Honolulu. J. Wildl. Manage. 36:498-5 10. STEARNS,H. T. 1966. Geology ofthe State of Hawaii. MOUNTAINSPRING,S. 1986. An ecologicalmodel of Pacific Books, Palo Alto, CA. the effects of exotic factors on limiting Hawaiian STEMMERMANN,M. A. 1976. Ornithology report, p. honeycreeperpopulations. Ohio J. Sci. 86:95-100. 209-233. In D. Petersonled.1. The scientificreoort MOUNTAINSPRING,S., ANDJ. M. SCOTT. 1985. Inter- of the Manawainui Re&ar& Project, East Maui, specificcompetition among Hawaiian forest birds. Hawai‘i. National ScienceFoundation, Washing- Ecol. Monogr. 55:2 19-239. ton, DC. MUNRO, G. C. 1944. Birds of Hawaii. Tongs Publ. SWEZEY,0. H. 1954. Forest entomology in Hawaii: Co., Honolulu. an annotated check-list of the insect faunas of the OLSON,S. L., AND H. F. JAMES. 1982a. Fossil birds various components of the Hawaiian forests. from the Hawaiian Islands:evidence for wholesale Bemice P. Bishop Mus. Spec. Publ. 44. extinctions by man before Western contact. Sci- TISDELL,C. A. 1982. Wild pigs: environmental pest ence 217:633-635. or economic resource?Pergamon Press, Sydney, OLSON,S. L., AND H. F. JAMES. 1982b. Prodromus Australia. of the fossil avifauna of the Hawaiian Islands. UNITED STATESFISH AND WILDLIFESERVICE. 1983. Smithson. Contrib. Zool. 365. Republication of the lists of endangered and PERKINS,R.C.L. 1895. Notes on someHawaiian birds. threatened species.Fed. Reg. 48:34, 182-34, 196. Ibis 1895:117-129. VANRIPER, C., III. 1980. Observationson the breed- PERKINS,R.C.L. 1903. Vertebrata, p. 365-466. InD. ing of the Palila Psittirostrubailleui of Hawaii. Ibis Sharp red.], Fauna Hawaiien& Vol. 1, Part IV. 1221462-475. The Univ. Press. Cambridee. U.K. VANRIPER, C., III, S. G. VAN RIPER,M. L. GOFF,AND PRATT,H. D. 1979. ’ A system& analysisof the en- M. LAIRD. 1982. The impact of malaria on birds demic avifaunaof the Hawaiian Islands.Ph.D.diss., in Hawaii Volcanoes National Park. Tech. Rept. Louisiana State Univ., Baton Rouge. 47, Coop. Nat. Res. Stud. Unit, Univ. Hawaii, RAIKOW,R. J. 1976. The origin and evolution of the Honolulu. Hawaiian honeycreepers(Drepanididae). Living WARNER,R. E. 1968. The role of introduced diseases Bird 15:95-l 17. in the extinction of the endemic Hawaiian avifau- RAMSEY, F. L., AND J. M. SCOTT. 1979. Estimating na. Condor 70:101-120. population densities from variable circular plot WILSON,S. B., AND A. H. EVANS 1890-1899. Aves surveys, p. 155-181. In R. M. Cormack, G. P. Hawaiienses: the birds of the Sandwich Islands. Patil, and D. S. Robson [eds.],Sampling biological R. H. Porter, London. populations. StatisticalEcology Series,Vol. 5. In- ZIMMERMAN,E. C. 1948. Insects of Hawaii. Vol. 1. tematl. Co-op. Publ. House. Fairland, MD. Introduction. Univ. Hawaii Press, Honolulu. RICHARDS,L. P., AND P. H. BALDWIN. 19$3. Recent recordsof some Hawaiian honeycreepers.Condor 55~221-222.