The Condor98:312-321 0 The CooperOrnithological Society 1996

ASPECTS OF THE LIFE HISTORY AND FORAGING ECOLOGY OF THE ENDANGERED AKIAPOLAAU ’

C. JOHN RALPH AND STEVEN G. FANCY U.S.D.A. Service, Redwood Sciences Laboratory, 1700 BayviewDr., Arcata, CA 95521 and National Biological Service,Pacific Islands ScienceCenter, P.O. Box 44, National Park, HI 96718

Abstract. Relativeabundance, breeding ecology, annual survival,home range, and for- agingecology of Akiapolaau(Hemignathus munroi), an endangeredHawaiian honeycreeper, werestudied on theisland of Hawaii. The speciesis a specialist:Akiapolaau used koa ( koa)for foragingmuch more than expectedbased on koaavailability, and mostAkiapolaau occurredin old-growthkoa and ohia (Metrosiderospolymorpha) .Male Akiapolaau most often foragedon the trunks and largebranches of koa, whereasfemales used small branchesand twigs.The longer bill of males is apparentlyadapted to the greaterbark thicknessof largerbranches. Lichen-covered and deadbranches were preferred feeding sites. Akiapolaaushowed serial monogamy and had a relatively low reproductiverate of 0.86 youngpair-l year-‘, with a long parentaldependency period. Home rangesizes averaged 10.7ha and did not differbetween males and females.Annual survivalfor adultswas 0.71. Avian diseasesappear to restrictAkiapolaau to higherelevation forests where mosquitos are rare.Protection of remainingold-growth koa and ohia forestsabove the mosquitozone are criticalto the survivalof the species. Key words: Akiapolaau; Hemignathusmunroi; demography;survival; home range;for- aging ecology;Hawaii.

INTRODUCTION and (Perkins 1903, this study). The Ak- Hawaiian honeycreepers (Fringillidae: Drepa- iapolaau is the only surviving specieson the is- nidinae) are a showcaseexample of adaptive ra- land of Hawaii to exploit this woodpeckerniche, diation and speciation on a remote island chain. and yet it is rare and restrictedin its range despite This endemic subfamily, thought to be derived the apparent lack of competition. from a single colonization by a cardueline finch Akiapolaau are found only on the island of from North America (Johnson et al. 1989), has Hawaii, where they occur in four disjunct pop- evolved into a diverse group of speciesthat in- ulations totaling fewer than 1,500 birds (Scott et cludes the sickle-billed, nectarivorous Iiwi (Ves- al. 1986, Fancy et al. 1996). Their distribution tiaria coccinea), the cross-billed, insectivorous rangesfrom 1,300 to 2,100 m elevation in mesic Akepa ( coccineus),and the stout-billed, and wet ohia (Metrosiderospolymorpha)and koa -eating Palila (Loxioides bail&i) (Scott et (Acacia koa) forests in the Kau, Hamakua, and al. 1986, 1988; Freed et al. 1987). Even among Kona districts of Hawaii (Fig. 1) and from 1,900 such an assemblage,the bill and feeding motions to 2,900 m elevation in dry mamane (Sophora of the Akiapolaau (Hemignathus munroi) are bi- chrysophylla) and naio (Myoporum sandwich- zarre. The maxilla is long (2 l-29 mm; Amadon ense) forests on . Total population 1950, Pratt et al. 1994) and decurved. By con- size in the dry, high-elevation forest is probably trast, the mandible is straight and robust, and now less than 15 birds, and the dry forest pop- only half the length of the maxilla (Pratt et al. ulation may become extinct within the next five 1994). The utility of the beak is realized in the years (Fancy et al. 1996). species’ woodpecker-like foraging motions and The species’ foraging methods are well de- its ability to extract insectsfrom beneath the bark scribed (Perkins 1903, Berger 198 l), as are its of as well as to probe and pry insects from eggsand nest (Bank0 and Williams 1993). Meth- surface deformities and glean them from twigs ods to sex and age Akiapolaau were developed by Pratt et al. (I 994). However, the species’rarity and the inaccessibility of its habitat have pre- I Received13 June 1995. Accepted26 February1996. vented any previous detailed study of its foraging

13121 AKIAPOLAAU LIFE HISTORY AND FORAGING ECOLOGY 313 ecology or life history. We report here on the relative abundance, breeding ecology, annual survival, home range, and foraging ecology of Akiapolaau. This information was obtained dur- ing a larger study of the demography and foraging ecology of Hawaiian forest birds at five study sites on Hawaii.

STUDY AREAS AND METHODS We studied Akiapolaau between November 1976 and January 1982 at five study areas on Hawaii: Keauhou Ranch (193 1‘ N, 155”2O’W, 1,740 m elevation), Kilauea Forest (19”3 1‘ N, 155”19’W, 1,630 m), Hamakua (19”47’N, 155”2O’W; 1,770 m), Kau Forest (19”13’N, 155”39’W; 1,750 m), and a Dry Forest site (19”49’N, 155”33’W, 1,865- 2,800 m). The vegetation and level of distur- bance to each of the first four, wet forest, study areas is described in Ralph and Fancy (1994a). The Kau Forest study area is dominated by ohia and lacked koa and naio, and we never found FIGURE 1. Location of study areas (X) and distri- any Akiapolaau at that site. The Dry Forest site bution of Akiapolaau on the island of Hawaii during is a 4.3 km belt transect on the southwest slope the 1970s. of Mauna Kea dominated by mamane and naio and rangesfrom 1,865 m to line at 2,800 m. (1985) to determine food habits. Sex was deter- We surveyed 25-162 stations each month at mined by presence of a brood patch or cloaca1 each study site using the variable circular-plot protuberance (Ralph et al. 1993) by plumage method with 8-min counts (Scott et al. 1986) to characteristics(Pratt et al. 1994) in the case of a obtain indices ofAkiapolaau abundance.We were few adult males, or by observations of nesting unable to obtain accurate densities by this meth- birds. We identified juvenile birds on the basis od because of the species’ rarity and large dif- of plumage characteristics,skull ossification(Pyle ferences in detectability among months; there- et al. 1987), calls, and behavior. fore, we used the mean number of Akiapolaau We conducted a discriminant analysis of cul- detected per station as an index of abundance. men length, wing length, and body mass of We excluded birds detected > 100 m from the known-sex, adult (AHY) Akiapolaau to develop station to reduce the effects on the abundance additional methods for sexing Akiapolaau from index of using different observers and counting measurements. Wing length (chord) was mea- the same bird from multiple stations. Relative sured to the nearest millimeter with a metal rule, abundance among study sites and months was from the bend to the tip of the longest primary compared by two-way ANOVA and Tukey’s test of the unflattened, folded wing. Exposedculmen using survey data for 1979-1980, when all three was measuredwith a flexible rule along the upper study sites were sampled. We used Type-IV sum surfaceof the maxilla from the tip to the start of of squares to compare differences because of the feathered area at the base of the maxilla. We missing data for some months at the Hamakua classifiedeach specimen by sex using linear dis- site. criminant functions, and we produced unbiased We captured birds in mist nets at the Keauhou error rates by classifying individuals by a jack- Ranch (n = 62,006 net hours, November 1976 knife procedure. to January 1982) and Kilauea Forest (n = 16,958 We estimated annual survival of adult Akia- net hours, April 1978 to November 1979) study polaau at the Keauhou Ranch site with a Jolly- areas,and processedbirds as describedby Ralph Seber model (Pollock et al. 1990) from capture- and Fancy (1994b). Fecal samples from cloth recapture and resighting records for 20 color- bags in which the birds were placed were pre- banded birds. We excluded juvenile (hatching served and analyzed as described in Ralph et al. year) birds from this analysis because of small 314 C. J. RALPH AND STEVEN G. FANCY sample sizes.We selecteda four-month sampling availability of ohia, koa, and naio with their use period each year from September through De- by Akiapolaau for foraging. Foliage volume was cember, 1976-l 980, based on goodness-of-fit measured at each of the bird censusstations on tests from preliminary runs. Birds captured or each study area following methods of MacArthur resighted during the eight-month period from and MacArthur (196 1) and Anderson et al. (1983). January through August were coded as resight- Bark surface area was derived from the point ings and used to calculate survival probabilities quarter method (Mueller-Dombois and Ellen- (Model AX of JOLLY; Pollock et al. 1990:85). berg 1974: 110) and measurements of the total The complement of survival probability that we length of the trunk and of small, medium, and report here includes both mortality and perma- large branchesof sampled trees. Surfacearea was nent emigration. calculated from length measurements and the Home ranges of 17 Akiapolaau were deter- formula for the surface area of a cylinder. mined from locations of birds captured in nets Values presented are means -+ SE. or resightedduring weekly surveys throughout a 600 x 600 m site at Keauhou Ranch, following RESULTS methods described in Ralph and Fancy (1994b). RELATIVE ABUNDANCE Individuals with fewer than ten locations were The number of Akiapolaau detected per station excluded. Home range size was correlated with each month was highly variable, with peak num- the number of locations; therefore, we compared bers usually detected during August-November minimum convex polygon size between males each year (Fig. 2). Abundance indices of Akia- and females by analysis of covariance using the polaau at Keauhou Ranch and Kilauea Forest number of locations as a covariate. were highly correlated (r = 0.52, n = 27, P = We quantified foraging activities and use of 0.005). During 1979 and 1980, when we con- different foraging substratesduring weekly sur- ducted variable circular-plot surveysat Keauhou veys at the Keauhou Ranch site. During each Ranch, Kilauea Forest, and Hamakua, we found survey, we walked throughout the entire grid and significant variation in Akiapolaau abundance recorded activities, locations, and band combi- among study sites (F,,,, = 4.21, P = 0.03) and nations of Akiapolaau; we attempted to obtain among months (F ,,,, 9 = 3.13, P = O.Ol), with at least 35 sampling bouts during each survey. abundance (mean birds/station) at Keauhou Each bird was observed for a mean of 24.8 f Ranch (0.137 + 0.0 16) being higher than at Ki- 0.32 SE seconds(range 3-16 1 seconds),and usu- lauea Forest(0.074 & 0.0 16) and Hamakua (0.092 ally fewer than three activity budgetswere taken f 0.047); Tukey’s test, df = 19, P < 0.05). At on an individual during each survey. Each time Keauhou Ranch, where we did the majority of that a bird was observed foraging, we recorded our mist netting, monthly capture rates of Akia- the foraging activity (probe, glean, peck, or pry), polaau were highly correlated with the number height of the bird above the ground, height of ofAkiapolaau detectedper station (r = 0.35, n = the tree, speciesof tree, size (diameter) of branch 52, P = 0.01) and with the percent of stations at that the bird was on (twig = < 1 cm; small branch which Akiapolaau were detected (r = 0.49, n = = l-5 cm; medium branch = 5-15 cm; large 52, P = 0.002). Mean number of Akiapolaau branch = > 15 cm; and trunk), whether the bird detectedper station at Keauhou Ranch decreased was foraging on live or dead wood, and whether during the study (regression,n = 55, P = 0.000 1). or not the bird was foraging on bark with > 10% coverageby the epiphytic, fruticose lichen Usnea BODY MEASUREMENTS spp.We comparedforaging heights between males Discriminant analyses with exposed culmen, and females using analysis of covariance with wing, and body mass as variables indicated that tree height as a covariate, and we compared the discriminant function that best identified heights of trees used by each sex using a t-test. males from females by measurements included Differences in the percentage of time spent by only exposedculmen (EXPCUL, mm) and wing males versus females on different tree species length (WING, mm): (koa, ohia, and naio) and branch sizeswere com- D = 0.599*EXPCUL + 1.566*WING - 140.0 pared by Tukey’s test to control for Type-I errors. We developed indices of foliage volume and where D is the discriminant score for an indi- bark surface area at each site to compare the vidual. Akiapolaau with positive scoresshould AKIAPOLAAU LIFE HISTORY AND FORAGING ECOLOGY 315

+ Keauhou

+ Kilauea 0.8 l-----lX- Hamakua .-5 g 0.6

z a. ul 0.4 .--F a 0.2

o.ohJASONDJFMAMJJASONDJFMAMJJASONDJFMAMJJASONDJFMAMJJASONDJ 1977 1978 1979 1980 1981 FIGURE 2. Mean monthlydensity of Akiapolaauat KeauhouRanch, Kilauea Forest, and Hamakua,Hawaii, 1977-1982. be classified as males, and those with negative never observed a pair nesting together for two scoresshould be classified as females. We cor- seasonsand we observed four caseswhere Ak- rectly classified 21 of 22 Akiapolaau of known iapolaau definitely switched mates between one sex by this method. breeding season and the next. In two casesin- In separatef-tests, we found that exposedcul- volving a banded male and an unbanded female, men lengths of 12 adult males (29.97 t- 0.49) it is possible that the pair remained together for were longer than those of ten females (25.15 f at least two breeding seasons,but we could not 0.88; t = 5.03, df = 20, P = 0.001). Males had confirm this. In three caseswhere an individual mean wing lengths of 80.87 k 0.34 mm, com- changedmates, the other member of the pair left pared with 77.38 k 0.48 for females (t = 6.10, the study area. However, one female who had df = 20, P -c 0.0001). Males were also heavier, mated and produced a young in 1978 left that with body massesof 28.86 k 0.40 g, as compared male for another in early January 1979. She con- with 25.93 k 0.53 g for females (t = 4.51, df = sorted with the new male until mid-February, 20, P = 0.0002). and then was seenin the company of still another male in April 1979. The first two males remained BREEDING AND NESTING in the area and were seen infrequently until De- Akiapolaau demonstrated serial monogamy in cember 1979 and December 1980, respectively. our study:of 14 pair-yearsof observation in which Only one young per year was produced by the at least one member of the pair was banded we pairs with at least one banded parent. Twelve 316 C. J. RALPH AND STEVEN G. FANCY fledged young were produced from the 14 pair- defend part of their large home ranges as terri- years of observation. During regular surveys to tories. quantify activities and search for banded birds, ANNUAL SURVIVAL we observed an Akiapolaau with more than one young on only three of 2 17 occasionswhere an We found a high probability (0.80 f 0.21) of adult was seen with one or more young. The resighting an individual during the September- duration of the parental dependency period ap- December sampling period if it was present on peared to be very long: four young remained on the study area. This allowed us to fit a Jolly-Seber the study area with their parents for at least four model to our data (x2 = 0.12, df = 1, P = 0.72), months, and one young was still observed with even though we had data for only 20 adult birds. its parents five months after fledging. Annual survival for adults (the complement of We observed two nests, one for 592 minutes which includes permanent emigration) was 0.7 1 of observation during nest construction and in- f 0.03. We were unable to estimate survival for cubation at Keauhou Ranch during 16 July-l 5 juvenile Akiapolaau since most of the birds ei- August 1979, and the second, at Kilauea Forest ther left the study area or died before they could during 21 February-6 March 1980, for 1,9 10 be banded. However, ten young remained on the minutes. Observations began five days before the grid with one or more of their banded parents eggshatched and ended after the young fledged. for a mean of 105 days (range 38-157) before The nest at Keauhou Ranch was a cup nest in they disappeared. the small branches near the top of a 20-m high FORAGING BEHAVIOR koa tree, and the Kilauea Forest nest was 11.3 m off the ground in a cavity where a limb had The use of koa by Akiapolaau for foraging at broken off of a 20-m high ohia tree. Keauhou Ranch, Kilauea Forest, and Hamakua Only the female incubated and brooded the was much greater than expected based on the young. From observations of a single nest, we availability of koa (Fig. 3). For example, 63% of observedthat on days l-3 of the brooding period foraging observations at Keauhou Ranch were the female fed the young and removed fecal sacs, in koa, but koa comprised only 13% of the bark and the male fed the female off the nest in a surface area and 12% of foliage volume of trees fashion typical of drepanidines (Perkins 1903, at that site. Selection for koa was consistent Berger 198 1). On days 4-5 the male began com- throughout the year (Fig. 4), with 45-65% of for- ing to the nest, feeding the female there, and aging observations in koa during each month. At possibly also the young. The female continued Keauhou Ranch and Kilauea Forest, naio was to feed the young and remove fecal sacs.On days also used much more than expected, based on 6-9, the male played an increasing role, feeding its availability (Fig. 3). At the Dry Forest site, the young directly, removing fecal sacs,and con- 61% of 93 foraging observations were on ma- tinuing to feed the female. The female continued mane and 38% were on naio, compared to the to make more trips to the nest than the male. relative availability of mamane of 26% and naio The single young fledged on day 2 1. of 74%, by both the bark surfacearea and foliage volume methods. This indicated that Akiapo- HOME RANGE laau selectedmamane over naio for foraging. Ko- We found no difference in home range size be- lea (Myrsine lessertiuna)was rare at all study sites tween nine males and eight females after adjust- but was used more than expected for foraging ing for the number of locations at which each (Fig. 3). At the Hamakua study site, 16% of for- bird was observed (analysis of covariance, F,,+, aging observations were in kolea although the = 1.45, P = 0.25). Mean home range size for availability ofkolea was only 2% basedon foliage both sexescombined (n = 17) was 10.73 + 1.01 volume. ha (range 4-17 ha). Home range size for some Observations of Akiapolaau foraging on koa birds approached the size of the study area and indicated that they selected for lichen-covered was probably underestimated. We observedcon- and dead branches, presumably because these siderable overlap among home ranges, but ob- substratesharbored more insects. For example, servations of Akiapolaau reacting aggressivelyto Akiapolaau spent an averageof 69% of their time playbacks (M. Moore, unpubl.) and chasing in- on lichen-covered branches at Keauhou Ranch dividuals of the same sex suggestthat they may (n = 1,954 observation bouts), 75% of time at AKIAPOLAAU LIFE HISTORY AND FORAGING ECOLOGY 317

KEAUHOU RANCH KllAUEA FOREST HAMAKUA

100 l- N X L i- f -r 80 K 60 N ; K

k K z T 60 K r 5 a l- 40 0 40

0 0

20 20

0 e 60 I - ; - - - ! 0 EAF JSED 0 ILLAL- LkiF BiitK %USED FIGURE 3. Availability of treespecies as determined by foliagevolume () and barksurface area (BARK), and percentof foragingobservations of Akiapolaau(% USED) on ohia (0), koa (K), naio (N), kolea(L, Myrsine spp.),and all othertree speciescombined (X), at KeauhouRanch, Kilauea Forest,and Hamakua,Hawaii.

Kilauea Forest (n = 661), and 79% of time at was higher than that for males in koa (analysis Hamakua (n = 270). Similarly, mean time spent of covariance, F,,,,,, = 58.3, P < 0.0001) and on dead branches was 50% at Keauhou Ranch, naio (F,,438= 12.2, P = O.OOOS),but not in ohia 70% at Kilauea Forest, and 62% at Hamakua (F,,,d, = 3.4, P = 0.067). Males selectedtaller koa (sample sizes same as above). and ohia trees for foraging than did females (Ta- Branch sizes selectedfor foraging by Akiapo- ble 2). laau differed between sexes(Table 1). Males spent Percent occurrence of arthropods in 38 fecal more time foraging on the trunk and larger samples from adult birds are shown in Table 3. branches of trees, whereas females spent more Mandibles of lepidoptera larvae were found in time on twigs and small branches.After adjusting 92% of the droppings. Spiders were fairly com- for tree height, the foraging height for females mon (53%), as were cerambicid larvae (42%).

.7- Cl-KOA A-NAIO .6- 0 0 0 0- OHIA 0 4 0 9 4 4 J 4‘ 4‘ + 4 4 4 o 4 “-4Q o 0 0 JF’ M’ A’ M’ J’ J’ A’ S’ O’ N’ D’ ’ MONTH FIGURE 4. Mean monthly time spentforaging on koa, ohia, and naio by Akiapolaauat KeauhouRanch, Hawaii, 1977-1982. 318 C. J. RALPH AND STEVEN G. FANCY

TABLE 1. Mean percentof time spentforaging by male (n = 1,044) and female(n = 492) Aldapolaauon differentsizes of branchesof koa, ohia, and naio trees,Keauhou Ranch, Hawaii.

M&S Females Tree Branchsize Mean % SE Mean % SE P Koa Twigs 1.71 0.35 14.73 1.48 ** Small branches 17.48 1.10 39.78 2.08 ** Medium branches 21.94 1.22 14.10 1.49 ** Largebranches 6.50 0.72 2.07 0.59 ** Trunks 2.83 0.48 1.22 0.46 Ohia Twigs 0.68 0.23 1.63 0.55 Small branches 3.03 0.51 1.48 0.52 Medium branches 3.66 0.56 0.47 0.29 Largebranches 0.49 0.21 0.41 0.29 Trunks 1.45 0.36 0.30 0.22 Naio Twigs 3.93 0.56 5.10 0.96 Small branches 10.11 0.87 7.99 1.18 Medium branches 5.68 0.67 1.09 0.45 ** Largebranches 1.93 0.40 0.20 0.20 Trunks 4.48 0.61 0.99 0.41 ** Significantdifference found by Tukey’s test,P < 0.05.

DISCUSSION vember (Ralph, unpubl.). Akiapolaau breed throughout the year (Ralph and Fancy 1994a), Our monthly surveys suggestedthat the Akia- but we recorded the greatestnumber of sightings polaau population at Keauhou Ranch and Ki- of immature birds during the fall and the lowest lauea Forest increased each fall and then de- number in April. creasedto lowest levels during January through Akiapolaau were more abundant at Keauhou April. However, we found that marked individ- Ranch than at Kilauea Forest during the study uals remained on the study area throughout the even though Kilauea Forest had a much higher year and thus the seasonalpattern in abundance density of koa (Fig. 3). Keauhou Ranch was reg- indices was not causedby a regular migration of ularly grazed, but because the site had no per- birds into and out of the study area. A combi- manent water, it was subjected to less grazing nation of larger numbers of young and higher pressurethan Hamakua. Also, the fortuitous pil- detection rates from increased calling rates, par- ing of koa logs from a previous harvest created ticularly by juvenile birds, could account for the small areas ranging from lo-500 m2 that kept apparent increasein populations each fall. Ralph out and allowed some regeneration of na- and Fancy (1994a) found that singing rates of tive species.Akiapolaau density decreaseddur- Akiapolaau peaked in June but were otherwise ing our study as adjacent areas were being logged fairly constant throughout the year, but calling for koa, but it is unclear whether there was a rates were higher during September-December cause and effect relationship. than in other months, with peak calling in No- We found that Akiapolaau pairs usually pro-

TABLE 2. Heightsat which male and femaleAkiapolaau were observed foraging in three speciesof tree at Keauhou Ranch, Hawaii.

Foragingheight Tree height F’roportionalheight Tree specks ” MGUI SE Meall SE MCUI SE

Koa Male 662 16.46 2.12 22.04 2.11** 0.699 0.006 Female 435 12.20 0.27 16.26 0.29 0.740 0.007 Ohia Male 116 10.73 0.43 16.19 0.50** 0.668 0.017 Female 26 8.00 0.92 10.31 1.07 0.794 0.036 Naio Male 350 6.53 0.16 10.12 0.18 0.635 0.009 Female 89 6.94 0.30 9.75 0.31 0.706 0.017

**t-test,P < 0.05. AKIAPOLAAU LIFE HISTORY AND FORAGING ECOLOGY 319 duce only one young per year and the relatively TABLE 3. Percentoccurrence of variousinsect and long parental investment period, which in most arachnidtaxa in 38 Akiapolaaudroppings. speciesis correlated with a low reproductive rate Percent (Skutch 1985) apparently causessome pairs to TW.Oll Age of insect OCC”IIfXlCe forego breeding in some years. The net produc- tivity of 0.86 young pair’ year-’ is low com- Coleoptera Adult 15.8 Larvae 5.3 pared to most passerines.For example, van Rip- Carabidae Adult 2.6 er (1987) found that Common Amakihi (Hem- Cerambycidae Larvae 42.1 ignathusvirens) on Mauna Kea produced 2.5 Curculonidae Adult 2.6 young pair-’ year-‘, whereas Palila (van Riper Lepidoptera Larvae 92.1 Hymenoptera Adult 7.9 1980) another endangered honeycreeper, pro- Homoptera duced 1.8 young pair-’ year-‘. In a compilation Cicadelidae Adult 2.6 of several mainland studies, the percent of eggs Delphacidae All 5.3 (n = 21,951) that subsequently fledged a young Psyllidae All 5.2 averaged46% (Nice 1957), and when largerclutch Heteroptera Lygaeidae Adult 2.6 sizes and multiple clutches per season for most Diptera Adult 2.6 speciesare taken into account (Lack 1954, Skutch Neuroptera Larvae 23.7 1985) the number of young produced annually Psocoptera by Akiapolaau is low. Psocidae All 5.3 Spiders All 52.6 Adult survival seems to be average, with an annual survival rate of 0.7 1. Annual survival of other Hawaiian passerines, estimated with the same methods, was 0.73 for Hawaii Creeper (Or- polaau seem to be (1) habitat modification, par- eomystismana) and 0.70 for Akepa (Ralph and ticularly from logging and grazing, (2) predation Fancy 1994b), 0.72 for Apapane (Himatione by introduced rats and cats (Feliscatus) and na- sanguinea;Ralph and Fancy 1995) 0.66 for Pa- tive raptors, and (3) avian disease. Koa is a le- lila (Lindsey et al. 1995) and Omao (Myadestes gume, and its young leaves are highly palatable obscurus;Ralph and Fancy 1994c), and 0.55 for to cattle. Koa wood is also highly prized for Iiwi (Ralph and Fancy 1995). Mean survival for woodworking, and many areasof old-growth koa 35 speciesof temperate and tropical forest birds forest have been logged or grazed. Sakai (1988) was 0.56 (Karr et al. 1990). found that Akiapolaau were absent from almost We found evidence for divergence of male and pure stands of regenerating young koa following female Akiapolaau within their narrow foraging mechanical clearing of ohia-koa forest. Several niche. Mean measurements between sexes dif- Akiapolaau were detectedin the regeneratingar- fered by 19% for exposedculmen, 11.3% for body eas in 1993 and 1994, ca. 14 years after logging mass, and only 4.5% for wing length, indicating and bulldozing, but it is not known whether they that selection has been highest for dimorphism had recolonized the areas or were only foraging of the feeding apparatus. Males selected trunks there. Akiapolaau rely heavily on koa, and to a and large and medium branches for foraging, lesser extent mamane (also a ) and naio, whereasfemales were most often observed high- and they are the only forest passerine that did er in the trees on small branches and twigs. The not use from the extremely abundant ohia longer bills of males have probably evolved to (Ralph, unpubl.). We never found any exploit the thicker bark on the larger branches Akiapolaau at our Kau Forest site, where koa of the tree. was absent. Fancy et al. (1996) found few Ak- Other than the narrow foraging niche, the re- iapolaau outside of koa-dominated forests dur- quirements of the species appear to be broad. ing surveys in 1990-1995. Nests have been found in ohia, koa, and mamane The diversity of insectsfound in fecal samples trees as either open statant cups or in cavities from Akiapolaau was low; only the Iiwi, which (Bank0 and Williams 1993; this study). Breeding is primarily nectarivorous (Ralph, unpubl.), had and molting can occurthroughout the year (Ralph a less diverse diet. Akiapolaau were most often and Fancy 1994a) and the speciesoccurs in both observed foraging on lichen-covered surfacesof wet and dry forests (Scott et al. 1986). koa branches. Lichen is characteristic of larger The primary threats to the survival of Akia- trees, and Akiapolaau are rarely found in signif- 320 C. J. RALPH AND STEVEN G. FANCY icant numbers outside of old growth forests.The LITERATURE CITED need for a relatively large home range for spe- AMAZON, D. 1950. The Hawaiian honeycreepers cialized foraging within a limited distribution of (Aves, Drepaniidae). Bull. Am. Mus. Nat. Hist. suitable old-growth koa forest undoubtedly con- 95: 157-262. tributes to the species’ rarity. ANDERSON,B. W., R. D. OHMART,AND J. RICE. 1983. Avian and vegetation community structure and Predation by black rats (Rattusrat&s), feral their seasonalrelationships in the lower Colorado cats, and the native Hawaiian Hawk (Buteo so- River valley. Condor 85392-405. litarius)and Short-eared Owl or Pueo (Asioflam- BANKO,P. C., AND J. WILLIAMS. 1993. Eggs, nests, meussandwichensis), may affectAkiapolaau more and nesting behavior of Akiapolaau (Drepanidi- than other native birds (Snetsinger et al. 1994). nae). Wilson Bull. 105427-435. BERGER,A. J. 1981. Hawaiian birdlife. 2nd ed. Univ. The vocalizations of adult Akiapolaau are loud Hawaii Press, Honolulu. and frequent, and Akiapolaau forage more often FANCY,S. G., S. A. SANDIN,M. H. REYNOLDS,AND J. on the trunk and larger branches of trees where D. JACOBI. 1996. Distribution and population they are more accessibleto predators. Hawaiian statusof the endangeredAkiapolaau. Pac. Sci. 50: Hawks respond to playbacks of juvenile Akia- in press. FREED,L. A., S. CONANT,AND R. C. FLEISCHER.1987. polaau calls that are used to lure adults into mist Evolutionary ecology and radiation of Hawaiian nets. In the dry mamane forest, where trees are passerinebirds. Trends Ecol. Evol. 2: 196-203. of lower stature, we suspectthat predation is the JACOBI,J. D. 1990. Distribution maps, ecologicalre- primary factor that will soon lead to the local lationships, and status of native communi- ties on the Island of Hawaii. Ph.D.diss., Univ. of extinction ofAkiapolaau (Fancy et al. 1996). Only Hawaii, Honolulu. one juvenile Akiapolaau has been observed in JOHNSON, N. K., J. A. MARTIN,AND C. J. RALPH. 1989. the dry forestpopulation since 1990, despiteclose Genetic evidence for the origin and relationships monitoring of several banded pairs (T. Pratt, un- of Hawaiian honeycreepers(Aves: Fringillidae). Condor 91:379-396. publ.), and the remaining population consistsonly KARR, J. R., J. D. NICHOLS,M. K. KLIMKIEWICZ, AND of a few males. J. D. BRAWN. 1990. Survival rates of birds of Avian malaria and pox have had a profound tropical and temperate forests:will the dogma sur- effect on native Hawaiian forest birds (Warner vive? Am. Nat. 136:277-291. 1968, van Riper et al. 1986), and many areas of LAAKE,J. L., S. T. BUCKLAND,D. R. ANDaRsON,AND K. P. B~~NHAM. 1994. DISTANCE User’s auide. apparently suitable old growth koa-ohia forest 2.1. Colorado Coop. Fish & Wildlife Res. -Unit: where Akiapolaau are absent can be found on Colorado State Univ., Fort Collins CO. Hawaii below approximately 1,500 m (Scott et LACK, D. 1954. The natural reaulation of animal al. 1986; Jacobi 1990; Fancy et al. 1996). There numbers. Clarendon Press, Oiford. are no data to determine whether Akiapolaau are LINDSEY, G. D., S. G. FANCY,M. H. REYNOLDS,T. K. PRATT,K. A. WILSON, P. C. BANKO, AND J. D. more susceptibleto avian diseasesthan are other JACOBI. 1995. Population structureand survival Hawaiian honeycreepers, but the fact that Ak- of Palila. Condor 97:528-535. iapolaau are currently found only where mos- MACARTHUR,R. H., AND J. W. MACARTHUR. 1961. quitos are absent or rare suggeststhat avian dis- On bird speciesdiversity. Ecology 42:594-598. ease is a major cause of their range contraction. MUELLER-D• MBOIS,D., AND H. ELLENBERG.1974. Aims and methods of vegetation ecology. John The protection and enlargement of existing hab- Wiley and Sons, New York. itat among Hawaii’s remaining old-growth ohia NICE, M. M. 1957. Nesting successin altricial birds. and koa forests above the zone of mosquito oc- Auk 74:305-321. currence is thus critical for the continued sur- PERKINS,R.C.L. 1903. Vertebrata (Aves), p. 368- vival of Akiapolaau. If adequate habitat is avail- 465. In D. Sharp [ed.], Fauna Hawaiiensis. Vol. 1, University Press, Cambridge. able, the speciesappears capable of maintaining POLLXK, K. H., J. D. NICHOLS,C. BROWNIE,AND J. viable populations. E. HINES. 1990. Statisticalinference for capture- recaptureexperiments. Wildl. Monogr. 107:1-97. ACKNOWLEDGMENTS PRATT,T. K., S. G. FANCY, C. K. HARADA, G. D. We thank Dawn Breese, Marc Collins, Tim Ohashi, LINDSEY, AND J. D. JACOBI. 1994. Identifying sex Peter Paton, Howard Sakai, and Claire Wolfe for their and ageof Akiapolaau. Wilson Bull. 106:42l-430. assistance with field work. Kamehameha Schools/ PyLE, P., S.N.G. Howau, R. P. YUNICK, AND D. F. Bishop Estatekindly provided accessto their lands for DESANTIZ. 1987. Identification guide to North this study. Sheila Conant, Lenny Freed, Jeff Hatfield, American passerines.Slate Creek Press, Bolinas, Thane Pratt, and Charlesvan Riper III provided many CA. helpful comments on an earlier draft ofthe manuscript. RALPH,C. J., AND S. G. FANCY. 1994a. Timing of AKIAPOLAAU LIFE HISTORY AND FORAGING ECOLOGY 321

breeding and molting in six speciesof Hawaiian Scorr, J. M., S. MOUNTAINSPRING,F. L. RAMSEY, AND honeycreepers.Condor 96: 15l-l 6 1. C. B. KEPLER. 1986. Forest bird communities of RALPH,C. J., ANDS. G. FANCY. 1994b. Demography the Hawaiian islands:their dynamics,ecology and and movements ofthe endangeredHawaii Creeper conservation. Stud. Avian Biol. 9: l-43 1. and Hawaii Akena. Wilson Bull. 106:615-628. SKUTCH,A. F. 1985. Clutch size, nestingsuccess, and RALPH,C. J., ANDS. G. FANCY. 1994~. Demography predation on nestsof neotropical birds, reviewed. and movements of the Omao (Myadestes obscu- Omithol. Monogr. 36:575-594. rus). Condor 96503-5 11. SNETSINGER, T. J., S. G. FANCY,J. C. SIMON, AND J. RALPH, C. J., AND S. G. FANCY. 1995. Demography D. JACOBI. 1994. Diets of owls and feral cats in and movements of Apapane and Iiwi in Hawaii. Hawaii. Elepaio 54:47-5 1. Condor 91:729-742. VANRIPER, C., III. 1980. Observationson the breed- RALPH, C. J., G. R. GEUPEL, P. PYLE,T. E. MARTIN, ing of the Palila Psittirostrabailleui of Hawaii. Ibis ANDD. F. DESANTIZ.1993. Manual of field meth- 122~462-475. ods for monitoring landbirds. U.S. Forest Service, VAN RIPER, C., III. 1987. Breeding ecology of the Gen. Tech. Rep. PSW-GTR-114. Albany, CA. Hawaii Common Amakihi. Condor 89:85-102. RALPH,C. P., S. E. NAGATA,ANLI C. J. RALPH. 1985. VANRIPER, C., III., S. G. VAN RIPER,M. L. GOFF,AND Analysis of droppings to describe diets of small M. LAIRD. 1986. The epizootiology and ecolog- birds. J. Field Omithol. 56: 165-l 74. ical significanceof malaria in Hawaiian land birds. SAKAI, H. F. 1988. Avian response to mechanical Ecol. Monogr. 56:327-344. clearing of a native rainforest in Hawaii. Condor WARNER,R. E. 1968. The role of introduced diseases 90:339-348. in the extinction of the endemic Hawaiian avifau- Scorr, J. M., C. B. KEPLER, C. VAN RIPERIII, AND S. na. Condor 70:101-120. I. FEFER. 1988. Conservation of Hawaii’s van- ishing avifauna. Bioscience 38:238-253.