Distribution, Relative Abundance, and Habitat Relationships of Landbirds in the Vavau’ Group, Kingdom of Tonga ’

Home , Fiji shrikebill, Morinda, Polynesian ground dove, Polynesian starling, Polynesian triller

The Condor 100:60!-628 0 The Cooper Ornithological Society 1998

DISTRIBUTION, RELATIVE ABUNDANCE, AND HABITAT RELATIONSHIPS OF LANDBIRDS IN THE VAVAU’ GROUP, KINGDOM OF TONGA ’

DAVID W. STEADMAN Florida Museum of Natural History, University of Florida, P.O. Box II 7800, Gainesville, FL 32611, e-mail: steadman@~mnh.uf!.edu

HOLLY B. FREIFELD* Department of Geography, University of Oregon, Eugene, OR 97403

Abstract. We assessedthe distribution, relative abundance,and habitat preferences of the 12 indigenous,resident speciesof landbirdsthat survive in the Vava‘u Group, Kingdom of Tonga. We surveyed 16 islands, 10 of which are previouslyunmentioned in ornithological literature. The islands vary in land area (0.02-96 kn?), habitat composition,elevation (20- 215 m), and distance (O-lo.1 km) from the largest island of ‘Uta Vavdu. We conducted point counts along transectson 14 of the islands, and placed each count location into one of five habitat categories.Of the 11 species of landbirds that are widespread and at least locally common, 7 (Purple-cappedFruit-Dove Ptilinopus porphyraceus, Pacific Pigeon Du- cula pacijca, Common Barn-Owl Tyto alba, White-rumped Swiftlet Collocalia spodiopygia, Collared Kingfisher Halcyon chloris, Polynesian Tiiller Lalage maculosa, Polynesian Star- ling Aplonis tabuensis) certainly or probably occur on each of the 16 islands. One species (West PolynesianGround-Dove Gallicolumba stairii) is extremely rare (one small population on one island). Three speciesthat we did not record (Many-colored Fruit-Dove Ptilinopus perousii, Blue-crowned Lorikeet Vini australis, Fiji Shrikebill Clytorhynchus vitiensis) probably have been extirpated from Vava‘u. The species richness and relative abundanceof landbirdson islands in Vava‘u are affected more by deforestationand other human activities than by island area, elevation, or isolation. Key words: endangered species, forest birds, island biogeography, Tonga.

INTRODUCTION Ha’apai Group in 1995 and 1996. Elsewhere in The Kingdom of Tonga is part of the West Poly- West Polynesia, transect surveys of landbirds nesian avifaunal region that also comprises have been conducted in American Samoa Niue, Samoa, Wallis and Futuna, Fiji, and Ro- (Amerson et al. 1982, Engbring and Ramsey tuma (Watling 1982). The extant, indigenous 1989, Freifeld 1998) and Western Samoa (Bel- landbirds of Tonga represent 17 genera (none lingham and Davis 1988, Lovegrove et al. 1992, endemic) and 18 species (two endemic). Indi- Park et al. 1992). Another survey in Western vidual Tongan islands > 10 km2 sustain lo-13 Samoa by Evans et al. (1992) was based on speciesof landbirds today, whereas such islands sightings only and thus is less useful than sight/ each supported at least 27 species of landbirds sound surveys in quantifying relative abun- when humans arrived about 3,000 years ago dance. Mist-net surveys have been done on the (Steadman 1993, 1995, 1998). Tongan island of Late by Rinke (1991), although Based primarily upon transect surveys, we as- capture data from mist-nets may not be a reliable sess the distribution, relative abundance, and indicator of the relative abundance of birds habitat relationships of native landbirds in the (Remsen and Good 1996). Surveys such as ours Vava’u Group, Kingdom of Tonga. The only provide a first look at the incidence and abun- other published transect surveys of Tongan birds dance of landbirds in various habitats. Using are those conducted by Steadman (1998) in the similar methods on a regular basis and ideally throughout the year, these surveys are an effi- cient way to monitor bird populations among ’ ’ Received 3 November 1997. Accepted9 June habitats that differ in degree of human distur- 1998. bance. z Currentaddress: Florida Museumof NaturalHis- tory, Universityof Florida,PO. Box 117800,Gaines- Ornithological exploration of Vava’u by non- ville, PL 32611, e-mail: [email protected] Tongans began with four pencil drawings of

16091 610 DAVID W. STEADMAN AND HOLLY B. FREIFELD birds from the expedition of Alejandro Malaspi- 26.6”C, respectively, in August. Sixty-seven per- na in 1793 (Sotos Serrano 1982), currently under cent of the 2,312 mm mean annual rainfall oc- study by S. L. Olson. E. H. Layard visited ‘Uta curs from November through April. The wettest Vava‘u in the early 1870s and collected speci- months are March (364 mm, range 126-799) mens now in the British Museum (Natural His- and January (285 mm, range 50-1,009), and the tory) (Layard 1876). C. H. Townsend collected driest are June (105 mm, range 9-393) and Au- birds on ‘Uta Vava‘u on 4-5 December 1899 gust (116 mm, range 12-273). Vava’u lies near (Townsend and Wetmore 1919). The Whitney the northern limit of the South Pacific trade wind South Seas Expedition (WSSE) visited Vava’u zone, with northeast to southeastwinds prevail- in August 1925, yielding the first records of ing about 60-70% of the time, regardlessof seabirds from several islands other than ‘Uta Va- son. va’u. No comprehensive list of species and is- The human population of Vava’u was about lands was ever published based on the WSSE 16,000 in 1993 (Christopher 1994). Archeolog- data. ical sites on ‘Eua and in the Ha‘apai Group of Studies of birds in Vava‘u since 1925 have Tonga indicate continuous human occupation for been based primarily on sight/sound records. the past 2,800-3,000 years (Burley 1994, Dick- The birds of Vava’u are treated in a general inson et al. 1994, Burley et al. 1995). One con- manner in regional books such as duPont (1976), sequenceof this human occupation is the loss of Watling (1982), and Pratt et al. (1987). In spite most indigenous species of birds (Steadman of accountsof selectedspecies on certain islands 1993, 1995, 1997a). A similar situation is likely (Gill 1988, 1990, Rinke et al. 1992), meaningful to have taken place in Vava’u, although no ar- lists of species (sensu Remsen 1994) were not cheological excavations have taken place there available before our work for any individual is- (Davidson 197 1). land in Vava‘u, nor had the relative abundances Volcanic eruptions of Late and Fonualei dur- of any species been estimated. Ten of the 16 ing the Holocene (last 10,000 years) covered the islands we visited have not been reported pre- islands of Vava‘u with tephra depositsthat have viously in the ornithological literature. weathered into rich soils suitable for cultivation of food plants. The vegetation on most islands METHODS today is thus a mosaic of mature forests (typically covering only the steep land between the STUDY AREA limestone terraces), successional or disturbed The Vava’u Group is the northernmost of three forests, and active (open) or abandoned (wood- main clusters of islands in the Kingdom of Ton- ed) agricultural plantations. In the absence of ga (Fig. l), which is located east of Fiji and human influence, mature forests would be the southwest of Samoa. The volcanic islands of dominant vegetation on each island we sur- Late and Fonualei lie to the west and north and veyed. are separated from Vava’u by deep water. Va- va‘u consists of 58 raised limestone islands on HABITAT CATEGORIES a single submarine platform sloping gently to We divided the terrestrial habitats of Vava’u into the south, so that most emergent land is on the five categories:village, open plantation, wooded north side of the platform (Fig. 2). Lowered sea plantation/early successional forest, submature/ levels during the late Pleistocene glacial interval disturbed mature forest, and mature forest (Table probably united all or nearly all of these islands 2). into one island > 300 km2 in land area. By far 1. Village refers to any human habitation site. the largest island in Vava’u is ‘Uta Vava‘u (Ta- These areas are devoid or nearly so of native ble 1). Eleven other islands in Vava’u have land trees or shrubsexcept coconuts (Cocos nuciferu) areas > 1 km2, five of which we surveyed for but have scatterednon-native trees. There is no birds. regeneration of native vegetation. An irregular The only long-term climate and weather data grid of dirt roads and paths lies among the hous- for Vava’u are from ‘Uta Vava‘u (Thompson es and other buildings. Non-native vertebrates 1986), where the mean daily minimum and max- including dogs, pigs, rats, and chickens are imum temperatures range from 23.5”C and abundant. 30.2”C, respectively, in February to 2O.O”C and 2. Open Plantation refers to active agricultural LANDBIRDS OF VAVAU,‘ TONGA 611

I I I

Niuafoou’ @

dTafahi

v 6 ’ Niuatoputapu

KINGDOM OF

TONGA

~Fonualei

’ ’ Toku

VAVAU’ GROUP Vavau’ Hunga @ ,%U Pangaimotu o Noa~a~u , ;\ Late Ovaka +,,“R,“,”

HAAPAI’ GROUP MO’UtlQtl’One i, Haano’ DKao . >Foa 8 Tofua 1. _,JLifuka .i Uiha.0 Uoleva 20” ..‘

Nomuka FZ3lCOIl P M.WlQO~’ ‘. :-~&,“a

. . Hunga Hunga Haapai’ Tonga

TONGATAPU GROUP wJEuaiki’ Tongatapu Eua‘ a KZllfiU ’

22”

_Ata‘

0 20 40 R I ” t/“N f 0 20 40 $0 80 k

176 ’ 174O I I I

FIGURE 1. The Kingdom of Tonga. 612 DAVID W. STEADMAN AND HOLLY B. FREIFELD

Vava’u Island Group

3ELIFUKA VAKkEITU LP,NGITo O-’ $ SISIA

@ AVALAU 8 ,E”AKAF*@$J -4 B TOTOKAMAKA @ EUEIKI‘

FUAAMOTU’ b 6% MUOMU’ A’ @3 0 I FANGASITO ~ONUAONE’ ONE’ Q N

TAULAb

Q MANINlTA 114”W

FIGURE 2. Vava‘u Group, Tonga. Islands surveyed are underlined.

plots. Crops in Vava‘u are usually grown be- Pandanus spp., and bananas Eumusa spp., Aus- neath an overstory of coconut trees. Most of the tralimusa spp. (sequence and nomenclature fol- cultivars are speciesintroduced or managed pre- low Kirch 1994). Modern crops also include historically in West Polynesia, such as breadfruit species introduced to Polynesia within the past Artocarpus altilis, paper mulberry Broussonetia 200 years, such as mango Mangzfera indica, cit- papyrzfera, Indian mulberry Morinda citrzfolia, rus Citrus spp., maniac Manihot esculenta, pa- kava Piper methysticum,Tahitian chestnut Zno- paya Carica papaya, pineapple Ananas como- carpusfagzfer, sweet potato Zpomoeabatatas, ti sus, and tannia taro Xanthosoma sagittifolium. Cordylinefruticosa, yams Dioscorea spp., sugar Woody but weedy exotics include Adenanthera cane Saccharum ofJicinarum,kape Alocasia ma- pavonina and Mimosa pudica. Coconut, panda- crorrhiza, taro Colocasia esculenta, pandanus nus, and Indian mulberry are indigenous to Ton- LANDBIRDS OF VAVAU,’ TONGA 613

TABLE 1. Summary of characteristicsof islands surveyed. Areas are rounded to the nearest0.01 km2, elevations rounded to the nearest 5 m, and distancesrounded to the nearest 0.1 km. * Transect surveys were not conductedon Nuku or Kulo, althoughhabitat assessmentsand countsof all birds were made. Habitat categories defined in Table 2. ** Includes Pangaimotu(connected by a causeway).

Distance No. of stations in Max. to ‘LIta each habitat type Area &V. VZWaU ‘ Island (km*) Cm) (km)** 1 2 3 4 5 Dates surveyed ‘Uta Vava‘u 95.95 215 0 18 2 41 37 41 2. 8. 9. 14. 21-25.29 ‘Juiy’1995, 7 July 1996 Pangaimotu 9.24 70 0 - 7 12 - - 7 July 1996 Kapa 6.06 100 1.2 - - 10 6 - 26 July 1995 Hunga 5.34 90 2.8 - 20 - - 11-12 July 1995 Ofll 1.32 30 2.1 - - 16 - - 1 August 1995 ‘Utungake 1.09 80 0.2 6 6 8 July 1996 Vaka ‘eitu 0.85 45 7.0 - - - 5 : 4 July 1995 ‘Euakafa 0.70 62 6.8 - - - 7 15 19 July 1995 Taunga 0.56 40 4.0 2 23 - - 20 July 1995 A‘a 0.54 45 3.4 - - 2 10 7 July 1995 Mafaua 0.46 25 1.5 - 5 8 - 31 July 1995 ‘Eueiki 0.24 35 6.6 - - - 3 - 18 July 1995 Foelifuka 0.21 40 10.1 - 14 - - 28 July 1995 Foeata 0.06 30 9.8 - - 4 - - 28 July 1995 Nuku 0.04 20 3.8 - * 30 July 1995 Ku10 0.02 30 6.8 - * 6 July 1995 Totals 26 9 148 74 72 329 Stations

ga, although they are managed to provide food, ods and nomenclature described by Franklin and matting, cordage, medicine, and other products. Merlin (1992), Bolick (1995), and Drake et al. After coconut, mango is the second most com- (1996). They typically studied nearby pairs of mon large tree in the plantations. plots in secondary and mature forest. All plant Definition of OUThabitat categories 3-5 (Table specieslisted for these habitat categoriesare in- 2) is enhanced by quantitative inventories of for- digenous unless statedotherwise. By identifying ests in Vava’u by Franklin et al. (in press) at the and recording all vascular plant species and same time that we surveyed birds. Franklin et measuring all woody stems > 5 cm, they cal- al. sampled forest vegetation in 44 plots of 600 culated the basal area for 85 species of trees m* each (20 X 30 m) on 13 islands, using meth- (11-29 species per plot). Their data classified

TABLE 2. Characteristicsof habitat types in Vava‘u. “Proportion native vegetation” refers to the approximate percentageof woody stems (> 5 cm diam.) that representsnative species.

Proportion native Habitat type vegetation 1. Village <20 Isolated trees; largest trees non-native; ground cover grass,bare soil, buildings, or roads; no regenerationby native species. 2. Open plantation <20 Open canopy (618 m); largest trees non-native; active cultivation of non-native crops; no regenerationby native species. 3. Wooded plantation/early 20-90 Semi-open to closed canopy (6-20 m); largest trees and understo- successionalforest ry both native and non-native; few or no trees > 40 cm diam.; all regenerationby native species. 4. Submature/disturbedma- >90 Closed or nearly closed canopy of native species(lo-22 m); un- ture forest derstory open to dense; few trees > 40 cm diam.; all regeneration by native species. 5. Mature forest >90 Closed canopy of native species (lo-24 m); rather open understory; many trees > 40 cm diam.; all regenerationby native species. 614 DAVID W. STEADMAN AND HOLLY B. FREIFELD the overstory vegetation into forest types that leeward forest has been lost to village sites or correlate with our categories3-5 of bird habitat. fishing camps. Characteristic species include 3. Wooded plantation/early successionalforest those usually associatedwith beach forest (Pan- refers to abandoned plantations. Native woody danus tectorius, Casuarina equisetzfolia, Guet- species begin to reoccupy plantations after sev- tarda speciosa,Neisospenna oppositzfolia,Och- eral years, although on inhabited islands such rosia vitiensis, Hibiscus tiliaceus, and Excoecar- land usually is cleared again within 30 years (of- ia agallocha) as well as some speciesassociated ten only 5-10 years), before or just as the native with lowland rain forest (Planchonella grayana, species begin to develop a canopied secondary Xylosma simulans, and Zanthoxylumpinnatum). forest. In these situations the indigenous trees Evidence of past cultivation may be minimal. In tend to occur as scattered individuals or small category 4, we also place tracts of forest that we groups, seldom with stem diameters > 40 cm. would have judged to be mature forest (category Franklin et al. (in press) distinguished two 5) except for evidence of selective logging dur- grades of early to mid-successional forest based ing recent decades. on species composition: (a) Rhus taitensisdl- 5. Mature forest has a closed canopy, many phitonia zyzyphoidesearly successionalrain for- large (> 40 cm dbh) trees, no evidence of recent est, found on gently sloping (O-15”) sites be- logging, and is dominated by native species. tween lo-125 m elevation on large or small is- This correspondsto the following type described lands, with cultivated speciessuch as Cocos nu- by Franklin et al. (in press): Maniltoa grandzflo- czfera, Adenanthera pavonina, Citrus sinensis, ra-Pleiogynium timoriense-Planchonella gray- Morinda citrifolia, and Artocarpus altilis pres- ana late-successionalrain forest, found on steep, ent; and (b) Cryptocazya turbinata-Elattostachys rocky slopes on both large and small islands falcata-Rhus taitensisdlphitonia zizyphoides from 50-l 80 m elevation. Co-dominants include mid-successional rain forest, with higher basal Garuga floribunda, Chionanthus vitiense, and area and higher density than the first type, and sometimes Elattostachys falcata, Xylosma si- most cultivated speciesabsent from the oversto- mulans, and Zanthoxylum pinnatum. This forest ry, which is co-dominated by Zanthoxylum pin- type is considered mature and relatively undis- natum, Pleiogynium timoriense, Elattostachys turbed based on composition (low or no cover falcata, Elaeocarpus tonganus, and Cocos nu- by early-successional species or cultivars), site czfera. characteristics,and in a few casesfrom air pho- 4. Submature/disturbedmature forest refers to tos taken in 1968 showing no clearing since at habitat with a closed or nearly closed canopy of least that year. native trees, but primarily with individuals of Banyans (Ficus obliqua, F. prolixa) are a early successional species found in the largest patchily-distributed but significant component of size classes. This habitat type corresponds to habitat categories4-5. Although seldom record- two forest types described by Franklin et al. (in ed in plots because they are usually found on press): (a) Cryptocarya turbinata-Ellatostachys steep slopesor cliffs, Ficus spp. are an important falcata-Zanthoxylum pinnatum-Maniltoa gran- food resource for tropical frugivores (Frith et al. diflora disturbed mid- or late-successional rain 1976, Snow 1981, pers. observ.). forest, found in sites with some evidence of disturbance or past cultivation, with subdominants BIRD SURVEYS Dysoxylum forsteri, Rhus taitensis, and Alphi- The scientific, English, and Tongan names for tonia zyziphoides, the last two suggesting that each species are given in the Species Accounts this forest type is late-successional rather than below, which also review previous records from mature; and (b) transition beach forest-lowland Vava’u. We surveyed birds on 16 islands in Va- rain forest, found at elevations of 7-50 m in a va’u during 2 July-l August 1995 and 7-8 July narrow (20-60 m wide) perimeter around large, 1996 (Table l), a season in which West Poly- protectedislands but covering much or all of the nesian birds are vocally conspicuous (pers. ob- islands lying near the margins of the Vava’u serv.). To estimate relative abundance and dis- Group, such as Foeata, Foelifuka, and Hunga. tribution of birds among islands and in different Influenced somewhat by salt spray, these forests forest types, we conducted 5-min, fixed-radius are taller and richer in woody specieson leeward counts along transects(Hutto et al. 1986). Time than on windward coasts, although much of the and logistical constraintsdid not allow us to cen- LANDBIRDS OF VAVAU,‘ TONGA 615

TABLE 3. Mean number of birds per station for each island or locality sampled. UV = Uta‘ Vavau‘ localities. Totals are calculated without White-rumped Swiftlet. Nuku and Ku10 not surveyed.

West Poly- PUtpIe- Purple nesian capped Whtte- Collared Poly- Poly- Wattled IslandiLocality Banded Swamp- Ground- FIUtt- PZKlfiC rumped King- nesian Tongan nesian HotEy- (no. stations) Rail hen Dove Dove Pigeon Swiftlet fisher Triller Whistler Starling eater TOtAS

UV-Taoa ( 16) 0.13 0.06 0.00 1.88 0.69 1.38 0.75 1.88 0.56 3.69 1.38 11.00 UV-Liku Holonga (7) 0.00 0.00 0.00 1.29 1.71 0.14 0.43 1.29 2.14 0.86 1.00 8.71 UV-Moungalafa’ (59) 0.05 0.00 0.05 1.76 1.80 0.15 0.69 0.81 2.12 1.41 1.75 10.44 UV-Neiafu (30) 0.13 0.03 0.00 0.60 0.00 0.70 0.07 1.20 0.07 1.33 0.43 3.86 UV-Tuanuku‘ (24) 0.00 0.46 0.00 1.33 0.25 1.54 0.21 1.50 0.67 1.08 0.38 5.88 UV-Toula (3) 0.67 0.00 0.00 1.67 0.00 1.00 0.67 1.67 1 .oo 1.67 0.00 7.33 Pangaimotu (19) 0.05 0.05 0.00 0.53 0.26 1.78 1.05 0.74 0.47 1.26 0.05 4.47 Kapa (16) 0.00 0.00 0.00 1.31 0.56 0.38 0.19 1.25 0.69 1.38 0.31 5.69 Hunga (20) 0.00 0.00 0.00 1.10 0.65 0.35 0.35 0.95 0.35 1.25 0.25 4.90 Ofu (16) 0.00 0.00 0.00 1.00 0.38 0.81 0.56 1.00 0.00 2.13 0.00 5.06 Utungake‘ ( 12) 0.00 0.00 0.00 1.33 1.00 0.67 0.67 1.42 0.67 1.17 0.08 6.33 Vakaeitu’ (i 1) 0.00 0.00 0.00 1.00 2.09 0.27 0.27 1.55 0.91 0.73 0.18 6.73 Euakafa‘ (22) 0.00 0.09 0.00 0.50 0.59 0.08 0.64 0.36 1.23 0.82 2.18 6.41 Taunga (2>) ’ 0.00 0.00 0.00 0.28 0.04 0.37 0.04 1.16 0.00 1.16 0.20 2.88 Aa‘ (12) 0.00 0.00 0.00 2.25 2.25 0.25 0.33 0.83 0.25 0.50 0.17 6.58 Mafana ( 16) 0.31 0.00 0.00 0.81 0.69 0.69 0.44 1.75 0.00 1.63 0.06 5.69 Eueiki‘ (3) 0.00 0.00 0.00 0.67 1.00 0.10 0.00 0.67 1 .oo 1.33 2.00 6.67 Foelifuka (14) 0.00 0.00 0.00 0.57 0.14 0.29 0.57 1.50 0.64 0.43 0.00 3.86 Foeata (4) 0.00 0.00 0.00 0.50 0.00 0.00 0.25 1.50 0.50 0.00 0.00 Q.75

sus the birds equally in each of the five habitat stations. This is most easily done for speciesthat categories. Rather, we emphasized sampling the vocalize consistently, such as the Purple-capped birds in forested habitats. Fruit-Dove or Tongan Whistler. Our calculations All counts were done between 07:OO and are based on “birds per station,” which is sim- 12:O0. We also recorded all bird observations, ply the mean number of birds seen or heard per regardless of the time of day. Surveys were not station, regardless of age, sex, behavior, or vo- conducted in rainy or windy conditions. The lo- calization. cations of transectswere recorded on air photos Aquatic species, such as herons, ducks, and using acetate overlays, copies of which are migrant shorebirds, as well as seabirds flying available from Steadman. Most islands were sur- overhead, are excluded from the analyses. The veyed on only one day (Table 1). Because of its White-rumped Swiftlet is a silent aerial feeder greater size and diversity of habitats, ‘Uta Va- that is much easier to detect (visually) in open va‘u was surveyed at six localities on 11 differ- habitats than in forests. Therefore we have not ent days. included the transect data for this swiftlet in the The transects in disturbed habitats followed totals for Table 3 or in any subsequentcalcula- existing trails or one-lane dirt roads to facilitate tions. rapid but quiet travel between stations (Ralph et al. 1993). Transects in mature forests generally RESULTS followed topographic contours, which provided The transect data (Table 3, Fig. 3) reveal major the most continuous, linear routes over the ex- inter-island differences in the total abundance of tremely rugged limestone terrain. Stations were birds and in the presence and/or abundance of 100 m apart. At each station we recorded each individual species. Birds are scarceston Foeata bird detected within a 50 m radius during a 5- and Taunga (mean total abundance = 2.75 and min period. Each bird was recorded as being 2.88 birds/station, respectively) and are most heard or seen. Birds detected at distances > 50 abundant on ‘Uta Vava’u and Vaka’eitu (mean m were noted but were not included in our anal- total abundance = 7.87 and 6.73 birds/station, yses. Individual birds recorded at the previous respectively). station were monitored during travel between Among the 12 indigenous species of resident stationsto avoid recounting the same bird at two landbirds that survive on the 16 islands sur- 616 DAVID W. STEADMAN AND HOLLY B. FREIFELD

q MEAN ABUNDANCE 0 SPECIES RICHNESS

9 9

ISLAND/LOCALITY FIGURE 3. Relative abundanceand species richness of indigenous landbirds on islands in Vava’u, Tonga. Based on data in Table 3. “UV-” refers to localities on ‘Uta Vava’u. Number of stationsin parentheses. veyed, 11 are widespread and at least locally (West Polynesian Ground-Dove, Purple-capped common in the Vava’u Group. Seven of these Fruit-Dove, Pacific Pigeon, Tongan Whistler, (Purple-capped Fruit-Dove, Pacific Pigeon, Wattled Honeyeater) show a trend of increasing White-rumped Swiftlet, Collared Kingfisher, abundance with less habitat modification (Fig. Polynesian Tiiller, Polynesian Starling, and the 6). The opposite trend is true for Banded Rail nocturnal Common Barn-Owl) certainly or prob- and White-rumped Swiftlet. The Purple Swamp- ably occur on most or all islands surveyed (Ta- hen, Collared Kingfisher, Polynesian Tiiller, and ble 4). These six species(White-rumped Swiftlet Polynesian Starling are habitat generalists of excluded) make up 83 to 100% of the birds in similar relative abundance regardless of habitat each of the four localities where mean total type (Fig. 6). Considering feeding guilds of ar- abundance is fewer than four birds/station (Table boreal species,both frugivores and passerinein- 3). sectivores show the general trend of increasing Species richness per transect varies from four abundance with less habitat disturbance (Fig. 7). (Foeata) to nine species (four localities on ‘Uta Vava’u). Species richness is lowest in Village SPECIES ACCOUNTS and greatest in the three forest categories (Fig. Throughout the Species Accounts, refer to Ta- 4). The mean relative abundance is inversely re- bles 3 and 4 and Figure 6 for details of the dis- lated to habitat disturbance, again with lowest tribution and relative abundance of individual values for Village and a stepwiseincrease in less species. disturbed (older) habitat categories (Fig. 4). In Gallus gallus (Chicken, moa). A non-native general, this pattern among habitats holds for is- gamebird introduced prehistorically through lands of varying sizes (Fig. 5). The patterns of most of Oceania (Steadman 1993), the Chicken relative abundance among habitats for individual is abundant in villages in Vava’u. Feral birds speciesare highly variable (Fig. 6). Five species occur regularly in all habitat categories. LANDBIRDS OF VAVA’U, TONGA 617

n MEAN TOTAL ABUNDANCE 0 SPECIES RICHNESS 11

10 g 9 8 I= 8 - cnv,zrn rYw 7 W= III wO 6 ofy 22 5 o- zy 4 =a 2” 3 f 2 z 1

0 - Village (n=26) Open Plantation Submature/Dis- Mature Forest (n=9) Plantation/Early turbed Mature (n=73) Successional Forest (n=74) Forest (n=148) HABITAT FIGURE 4. Total relative abundance(mean ? SE) and species richness of indigenous landbirds in the five habitat categories(defined in Methods; see also Table 2).

Gallirallus philippensis (Banded Rail, veku). be shy because of hunting. Specimens: ‘Uta Va- This terrestrial rail thrives in disturbed habitats va‘u (Layard 1876). Sight records: ‘Uta Vava‘u (Engbring and Ramsey 1989), favoring thick (Gill 1990). grassy areas during the day and venturing into Gallicolumba stairii (West Polynesian more open places to forage in the early morning Ground-Dove, tu). A forest obligate in Tonga, and late afternoon. We did not record the Band- the West Polynesian Ground-Dove is difficult to ed Rail on uninhabited islands. Its absence on see in the understory but is readily detected by the four smallest islands (Table 4) probably re- its call (described by Beichle 1991). We en- sults from the scarcity of grassesand sedgeson countered only one population of West Polyne- these goat-infested islands. Our relative abun- sian Ground-Dove (six different calling and ap- dance values for the Banded Rail probably are parently territorial birds, two of which we also underestimates as this species vocalizes irregu- saw), in the relatively large remnant of mature larly. Specimens: ?‘Uta Vava‘u (Layard 1876). forest at Mo’ungalafa on ‘Uta Vava‘u. This area Sight records: ‘Uta Vava‘u, Koloa (Gill 1990). was being cleared for agriculture in July 1995. Porphyrio porphyrio (Purple Swamphen, ka- Given the abundance of rats, cats, dogs, and pigs Zue). This large rail prefers wetlands, especially those with wooded margins. Thus islands such on ‘Uta Vava‘u today, the prospect for survival as ‘Uta Vava’u (which features the royally pro- of ground-doves there is slight. Elsewhere in tected Ngofe Marsh, where Gill [I9901 saw 15 Tonga, the West Polynesian Ground-Dove sur- swamphens) and Hunga (with a smaller marsh) vives only on the volcanic islands of Late, Hun- probably sustain the core population of the Pur- ga Ha’apai, and perhapsHunga Tonga and Tofua ple Swamphen in Vava’u. We recorded this spe- (Rinke 1991, Rinke et al. 1992). Populations of cies only on three islands and never in mature this West Polynesian endemic have been extir- forest. Our relative abundance values for the pated on ‘Eua, Lifuka, and Nomuka’iki (Stead- Purple Swamphen probably are underestimated man 1993, 1997b, 1998). No specimens of this as this speciesvocalizes only irregularly and can species have ever been collected in the Vava’u 618 DAVID W. STEADMAN AND HOLLY B. FREIFELD

13 a. Large island (Uta‘ Vavau:’ 85 k&) b. Medium-Sized Islands (5-10 km)’ I 12 1 11 10 9 8 7 6 5 4 3 2 1

v 1 (n=lS) 2 (n=2) 3 (n=41) 4 (n=37) 5(n=41) 1 (n=O) 2 (n=7) 3(n=42) 4(n=6) 5(n=O) :

!? 13 13 c. Small Islands (0 6-5.0 k&) d. Very Small Islands (co.6 km)’ 2 12 12 a 11 11 MEANABUNDANCE z 10 10 9 49 0 SPECIES E 8 RICHNESS =; 7 6 6 5 5 4 4 3 3 2 2 1 1 0 0 1 (n=S) 2(n=O) 3(n=39) 4(n=18) 5(n=21) HABITAT CATEGORY FIGURE 5. Relative abundanceand speciesrichness among habitat categories on islands in different size classes.Numbers in parenthesesfollowing habitat category refer to the number of stations in that habitat for that island size class. Large island (a) is ‘Uta Vava‘u. Medium-sized islands (b) are Pangaimotu,Hunga, and Kapa. Small islands (c) are Ofu, ‘Utungake, Vaka’eitu, ‘Euakafa, and Taunga.Very small islands(d) are Mafana, A‘a, ‘Eueiki, Foelifuka, and Foeata. Data are from Tables 1 and 3.

Group (Amadon 1943), where our sight/sound demic is a canopy- and subcanopy-dwelling fig records are the first evidence of its presence. specialist (Engbring and Ramsey 1989, Stead- Ptilinopus porphyraceus (Purple-capped man 1998). The reasonsfor its extirpation in Va- Fruit-Dove, kulukulu). This vocally conspicuous va’u are not clear. The Many-colored Fruit-Dove frugivore is very common and widespread in apparently occurred in Vava‘u in the 1870s but Vava’u. It occurs in all habitats, but is consis- was rare and not collected or seen by Layard tently more abundant with increased maturity of (1876). In Ha’apai in 1995-1996, the only in- the forest, an observation corroborated by data dividuals of this fruit-dove detected were on To- from Ha’apai (Steadman 1998). The distinctive fua (a large, forested volcanic island) and in call of the Purple-capped Fruit-Dove is given all Pangai Village on Lifuka, where four pairs were day (and often in the night) from the forest can- observed foraging in the banyan Ficus obliqua opy or high in isolated trees. Fruit-doves con- (Steadman 1998). Large Ficus trees (F. obliqua, gregate in fruiting trees (especially Ficus spp.) F. pro&z) are much more common in Vava’u but maintain separatebreeding territories. Spec- than in Ha’apai, yet we never heard the distinc- imens: ‘Uta Vava‘u (Layard 1876, Townsend tive vocalization of Many-colored Fruit-Doves and Wetmore 1919); WSSE ‘Uta Vava’u, Kapa, in spite of being especially attentive for it. Ovaka (Ripley and Birckhead 1942). Sight re- Ducula pac@ca (Pacific Pigeon, lupe). This cords: ‘Uta Vava’u, Pangaimotu, ‘Utungake, large, vocal frugivore is abundant and conspic- Kapa (Gill 1988, 1990). uous in the canopy of mature forests in Vava’u. Ptilinopus perousii (Many-colored Fruit- The relative abundance of the Pacific Pigeon in- Dove, manuma‘a). This West Polynesian en- creases sharply in mature forests regardless of LANDBIRDS OF VAVA’U. TONGA 619

2.0 Porphyria

$ 3.5

z 3.0

: 2.5 9 2.0 a 15 9 1.0

2 0.5 a z 0.0 1

HABITAT CATEGORY FIGURE 6. Relativeabundance (mean ? SE) of indigenouslandbirds in five habitatcategories. 1 = village, 2 = open plantation,3 = woodedplantation/early successional forest, 4 = submature/disturbedmature forest, 5 = matureforest. island size. The Pacific Pigeon is less territorial man 1993). There is no record of the Blue- than the Purple-capped Fruit-Dove, with one to crowned Lorikeet for Vava’u during the 20th four birds typically encountered. That it has sur- century, although Layard (1876) had described vived 3,000 years of hunting in Tonga (Burley this species (his Coriphilus fringillaceus) as 1996, Steadman 1989, 1993, 1997a) is at least “very abundant” in Vava’u in the 1870s. Layard in part becauseof an excellent ability to disperse (1876) also “. . . heard of a small parrot with between islands (Diamond 1974, Franklin and two long feathers in its tail, which formerly ex- Steadman 1991, Steadman 1997b). Specimens: isted in the group [Vava’u], but has become ‘Uta Vava’u (Layard 1876, Amadon 1943). quite extinct.” The identity of this parrot is un- Sight records: ‘Uta Vava’u, Pangaimotu, ‘Utun- known but might possibly be Charmosyna ama- gake, Koloa, Kapa (Gill 1990). bilis (Red-throated Lorikeet), confined today to Vini australis (Blue-crowned Lorikeet, hen- Fiji (Pratt et al. 1987). The Blue-crowned Lori- ga). This West Polynesian endemic (Amadon keet consumes nectar and pollen gathered with 1942) is no longer found on many of the islands its brushy-tipped tongue. Every individual of that once made up its range (Rinke 1985, Stead- this speciesrecorded by Steadman in Ha’apai in 620 DAVID W. STEADMAN AND HOLLY B. FREIFELD

widespread and common in Vava’u because FRUGIVORES caves (required for roosting and nesting) are nu- merous on certain islands. As stated earlier, White-rumped Swiftlets are not amenable to au- ditory censusing from within the forest, there- 2.5 fore our data on relative abundance contain a 2.0 major sampling artifact. Although their recorded

p 1.5 abundance is highest in open, disturbed habitats

f 1.0 (typical group size of one to five birds), swiftlets forage silently above the forest canopy as well,

: 0.50.0 T where they cannot be censusedaccurately. Spec-

Y imens: ‘Uta Vava’u (Layard 1876, Townsend i and Wetmore 1919). Sight records: ‘Uta Vava’u,

4 - Pangaimotu, ‘Utungake, Okoa, Koloa, Kapa

2 4.55.0 - PASSERINE INSECTIVORES (Gill 1990). T 3 4.0 - Halcyon chloris (Collared Kingfisher, sikota).

3.5 - Conspicuous on nearly every island surveyed,

3.0 - the Collared Kingfisher is a habitat generalist that feeds primarily on large insects and small 2.5 - lizards, although it occasionally takes small fish 2.0 - over the reef at low tide. Usually encountered as 1.5 single birds or pairs, the mean abundance of 1.0 Collared Kingfishers is highest in open planta- 0.5 tion but the small number of stations (9) in that 0.0 habitat and the large standard error preclude 1 2 3 4 5 stating that this species in fact prefers this hab- HABITAT CATEGORY itat over others. Specimens: ‘Uta Vava’u (Lay- FIGURE 7. Relative abundance(mean 5 SE) of fru- ard 1876, Townsend and Wetmore 1919). Sight givores and passerine insectivores in habitat catego- records: ‘Uta Vava’u, Pangaimotu, ‘Utungake, ries. Frugivores are Gallicolumba,Ptilinopus, Ducula, Okoa, Koloa, Kapa (Gill 1990). and Aplonis. Insectivores are L&age, Pachycephala, and Foulehaio. Habitat categoriesare as in Figure 6. L.alage maculosa (Polynesian Triller, sikiviu). An omnivorous gleaner of the subcanopy and occasionally the understory, the Polynesian 1996 was either perched in or flying between Triller is common and widespread in Vava‘u, coconut trees. Given the abundance of coconut where typical group size is two to three birds. It trees in Vava’u, poor quality habitat seems un- is the only species that we recorded unequivo- likely to account for its absenceor extreme scar- cally on each island surveyed. The relative city there. Rats and pesticides have been sug- abundance data reveal no well-defined habitat gested to account for population declines of preference. Specimens: ‘Uta Vava’u (Layard Blue-crowned Lorikeet elsewhere, although the 1876, Townsend and Wetmore 1919); WSSE evidence is equivocal (Steadman 1998). Speci- ‘Uta Vava’u, Ovaka, Kapa, ‘Euakafa (Mayr and men: ‘Uta Vava’u (Layard 1876). Sight records: Ripley 1941). Sight records: ‘Uta Vava’u, Pan- ‘Uta Vava’u (Layard 1876). gaimotu, ‘Utungake, Okoa, Koloa, Kapa (Gill Tyto alba (Common Barn-Owl, Zulu). Bam- 1990). owls are common in Vava’u. Night surveys Clytorhynchusvitiensis (Fiji Shrikebill,fuiva). would probably record them on virtually every This species is a gleaning understory omnivore island. One or two species of non-native rats that probably is extirpated from Vava’u. The (Rattus exulans, R. r&us) occurs on each island only record of Fiji Shrikebill from Vava‘u is two we visited, providing ample food for the Com- specimenscollected on ‘Uta Vava’u in the 1860s mon Barn-Owl. Sight records: ‘Uta Vava’u (Finsch and Hartlaub 1870). In Ha’apai, this (Layard 1876, Gill 1990). West Polynesian endemic and forest obligate Collocalia spodiopygia (White-rumped Swift- seems to have disappearedby 1991 from eight let, pekepeka). This small aerial insectivore is islands where it was recorded in 1925 (Mayr

622 DAVID W. STEADMAN AND HOLLY B. FREIFELD

1933, Rinke et al. 1992). By 1996, it could not ica papaya, Psidium guajava, and Lantana ca- be found on two other islands where it was re- mara (F&&e 1987, Steadman 1998; pers. ob- corded in 1991 (Steadman 1998); eight other serv.). Specimens: ‘Uta Vava’u (Layard 1876, small islands where it was found in 1991 were Townsend and Wetmore 1919); WSSE ‘Uta Va- not visited by Steadman in 1996. Elsewhere in va‘u, ‘Euakafa (Mayr 1942). Sight records: ‘Uta Tonga, two small populations of the Fiji Shrike- Vava‘u, Pangaimotu, ‘Utungake, Okoa, Koloa, bill may survive on the islets of ‘Eueiki (off Kapa (Gill 1990). Tongatapu) and Kalau (off ‘Eua), although it Foulehaio carunculata (Wattled Honeyeater, once occurred on both Tongatapu and ‘Eua (Rin- fuleheu). A West Polynesian endemic, this can- ke 1987, Steadman 1993). The only large pop- opy dweller is a noisy, conspicuous, aggressive ulations are on the volcanic islands of Niuato- nectarivore/insectivore (Whitmee 1875, Eng- putapu, Tafahi, Kao, Tofua, and Hunga Ha’apai. bring and Ramsey 1989, Steadman 1998). The The small populations of shrikebills that may Wattled Honeyeater occurs in Vava’u in groups still exist on nonvolcanic islands in Ha’apai are of two to six birds (usually three or four), much vulnerable to extinction, probably within years as on ‘Eua (Rinke 1984). The highest numbers or decades. Population declines probably result are sustained by the two mature forest habitat from a blend of deforestation, understory clear- categories, a pattern found as well in American ance by pigs and goats, and predation by cats Samoa (Freifeld 1998). Outside of the forest, the and rats. Specimens: ‘Uta Vava’u (Finsch and Wattled Honeyeater is attractedto isolated flow- Hartlaub 1870). Sight records: none. ering trees such as coconuts and Erythrina var- Pachycephala jacquinoti (Tongan Whistler, iegata. It is absent or scarce on many small is- hengehenga).Strictly territorial and vocally con- lands in Vava’iu; the only island < 0.5 km2 spicuous, the omnivorous Tongan Whistler is where we recorded it regularly is the largely for- common in the understory of mature forests. ested ‘Eueiki. Specimens: ‘Uta Vava’u (Layard Rinke et al. (1992) found this Tongan endemic 1876); WSSE ‘Uta Vava‘u, Kapa, ‘Euakafa, to be the most common passerine in the mature Maninita (Mayr 1932a). Sight records: ‘Uta Va- forests on Late Island and at Liku Holonga on va‘u (Gill 1990). ‘Uta Vava’u. Our data are corroborative, with the Tongan Whistler being the most common DISCUSSION speciesin the two largest tracts of mature forest FACTORS INFLUENCING COMMUNITY (UV-Liku Holonga, UV-Mo’ungalafa). Overall, COMPOSITION it is much more abundant in mature forest than Our knowledge of the landbird communities in in disturbed mature forest, wooded plantation/ Vava‘u lacks direct prehistoric perspective. early successionalforest, or open plantation. The Based upon data from elsewhere in Tonga (‘Eua Tongan Whistler is absent from village habitat. and five islands in Ha’apai), however, we know We also note that it occurs in successionalhab- that a typical island avifauna in Tonga included itats only if they are adjacent to a tract of mature at least 25-30 speciesof landbirds when humans forest. Specimens: ‘Uta Vava’u (Layard 1876, first arrived (Steadman 1993, 1995, 1998). Thus Townsend and Wetmore 1919); WSSE ‘Uta Va- we assume that many species of landbirds have va‘u, Kapa A’a, ‘Euakafa (Mayr 1932b). Sight become extinct in Vava’u during the past three records: ‘Uta Vava’u, Pangaimotu, ‘Utungake, millennia of human occupation. The extant for- ‘Euakafa, A‘a, Kapa (Rinke 1986, Gill 1990, est-obligate species in Vava‘u, such as the Pa- Rinke et al. 1992). cific Pigeon, West Polynesian Ground-Dove, and Aplonis tabuensis (Polynesian Starling, misi). Tongan Whistler, are remnants of a once much This highly frugivorous omnivore is common larger set of species that probably required for- virtually throughout Vava’u. Typically encoun- ested conditions. The present-day variation in tered in groups of two or three birds, the Poly- landbird communities among habitats and is- nesian Starling is a generalist in habitat and food lands in the Vava’u Group is the product of habits in Tonga. The fruit it consumes in Tonga long-term anthropogenic influences on island includes the native species Ficus spp., Grewia landscapes and on individual species. Interpre- crenata, Rhus taitensis, Pleiogynium timoriense, tation of our survey data rests on understanding Guioa lentiscifolia, Alphitonia zizyphoides, and how these processeshave operated. Cryptocarya turbinata, and the non-native Car- The mean relative abundance of landbirds is LANDBIRDS OF VAVAU,’ TONGA 623 greater in mature forest than in any other habitat Whistler, the presenceof a well-developed shrub (Fig. 4). This has been documented as well in layer may be a critical habitat component. The nearby Samoa (Bellingham and Davis 1988, understory plants in less-disturbed forests may Freifeld 1998) and Cook Islands (Franklin and sustain a richer prey base for insectivorousbirds. Steadman 199 1). The dearth of autecological in- A similar concept may apply for frugivores not formation about Polynesian landbirds constrains restricted to the canopy, such as the Polynesian us to speculative discussion of the requirements Starling (Fig. 6), that is, mature forest provides underlying the apparent habitat preferences of a greater diversity of food plants in the under- particular species.Nevertheless, we explore five story and subcanopy as well as in the canopy. factors that may help to explain the patterns of Human activities (deforestation, cultivation) habitat preference, relative abundance, and dis- restrict the extent of structurally complex for- tribution that we observed. ests.The escarpmentsthat separatethe limestone 1. Seasonally variable associations between terraces in Vavdu restrict human activities and bird and plant taxa. Frugivores may occupy cer- thus act to preserve bands of relatively intact tain forest types seasonally in continental envi- forest between plantations. If deforestation and ronments to exploit particular food resources agriculture were as feasible on the escarpments (Levey 1988, Innis 1989, Levey and Stiles as on terraces, Vava’u probably would lack ma- 1992). Because many forest trees in Tonga seem ture forest altogether.Under such conditions, the to reproduce seasonally, birds exploiting their West Polynesian Ground-Dove certainly would fruits may be expected to frequent selectedhab- be extirpated, and the Pacific Pigeon and Tongan itats at a particular time of the year. If this is the Whistler would be rare or gone. In contrast, case for the Pacific Pigeon and the Purple- some native forest birds seem to have adapted capped Fruit-Dove, for example, our determi- well to the spread of agricultural and disturbed nation that these columbids are more abundant forests. The Polynesian Tiiller and Polynesian in mature forest than in other categories(Fig. 7) Starling are more abundant in wooded planta- may only be part of an annual cycle of intra- tion/early successionalforest and submatureidis- island movement that cannot be represented in turbed mature forest than in mature forest (Fig. our seasonally limited data set. 6d, e). For three reasons,however, we reject this sug- 3. Intensity of predation by humans and other gestion of much seasonalmovement among hab- mammals (rats, cats, dogs, and pigs). Tongans itats in Vava’u. First, there are no strong intra- have hunted birds for several thousand years island environmental gradients that fruiting phe- with slings, snares,bird lime, thrown rocks, and nology seems to track, not even on the largest bows and arrows (Gifford 1929, Steadman 1995, and highest island (‘Uta Vava‘u). Second, al- 1997a, Burley 1996). The species most often though the fruiting phenology of certain forest pursued today (with guns, snares,thrown rocks) trees seems to be fairly synchronouswithin and are the Purple Swamphen, Banded Rail, Pacific between islands in Vava’u, many species of na- Pigeon, and Purple-capped Fruit-Dove. Even for tive trees were not setting flowers or fruit at the these four species, however, current hunting is time of our study even in the habitats where not intense and may have little effect on distri- landbirds were most abundant (Alphitonia zyzi- bution or population size. Nevertheless, the sta- phoides, Cryptocarya turbinata, Pleiogynium ti- tions designated as mature forest tend to occur moriense, Elattostachys falcata, and the asea- farther from villages and other sustainedhuman sonal Ficus spp. are notable exceptions). Finally, activity than other habitat categories.For species year-round censusesof some of the same species subjected to hunting, distance from human hab- of frugivorous birds across a comparable range itation may reduce hunting pressureand thus im- of forest types in American Samoa do not reveal prove habitat quality. Mortality of birds from seasonal movements or systematic changes in toxic chemicals is a form of indirect human pre- abundance between habitat types (Freifeld dation that has not been studied in Tonga. Pes- 1998). ticides are used on crops regularly in Vava’u, 2. Vegetation structure. The relative overall with an unknown influence on food webs in- abundance of both frugivores and insectivores volving birds. consistently increases with forest maturity (Fig. Various speciesof non-native mammals occur 7). For understory birds such as the Tongan on each of the 16 islands visited. These include 624 DAVID W. STEADMAN AND HOLLY B. FXEIFELD the prehistorically introduced Pacific rat Rattus day on Tongatapu and Niuafo’ou. In 1995, we exulans, dog Canisfamiliaris, and pig Sus scro- observed it regularly in Neiafu Village on ‘Uta fa, as well as the historically introduced black Vava’u, the most urbanized place in the Vava’u rat R. rattus, house cat Felis catus, horse Equus Group; this small population of bulbuls should caballus, goat Capra hircus, and cow Bos tau- be eradicated before it becomes established and rus. The islands inhabited by people tend to spreads. Vava’u does not yet have populations have most or all of these species.The uninhab- of three other non-native speciesthat are estab- ited islands have rats (usually R. rattus) and lished elsewhere in Tonga: Rock Dove Columba sometimes goats or pigs. Understory birds prob- livia, European Starling Sturnus vulgaris, and ably are subjected to more predation by non- Jungle Myna Acridotheres fuscus (Rinke 1986, native mammals than canopy species,a situation 1987, Gill 1988, 1990, Steadman, pers. observ.). exacerbated when goats or pigs damage the un- In Tonga’s capital of Nuku’alofa, the Red- derstory vegetation. This sort of damage may be vented Bulbul and European Starling are now especially harmful to the West Polynesian the two most common speciesof birds. This un- Ground-Dove, Fiji Sluikebill, and Tongan Whis- fortunate condition may apply as well to all of tler. Tongatapu, the largest island (325 km*) in Ton- 4. Ability to dispersebetween islands. A good ga. Unless the spread of non-native species is ability to dispersebetween nearby islands allows contained, such could be the eventual fate of populations to come and go as circumstances other Tongan islands. change on any one island. This is not long-distance dispersal but is within the Vava’u Group, COMPARISONS WITH LANDBIRD among islands within sight of each other (see COMMUNITIES IN HAAPAI‘ distancesin Table 1). Species more strictly con- Steadman (1998) conducted methodologically fined to mature forest, especially understory identical surveys of landbirds in the Ha’apai birds, tend to be poor dispersers.For example, Group of Tonga in 1995-1996. The Vava’u and the West Polynesian Ground-Dove, Many-col- Ha’apai groups (Fig. 1) share 10 species of in- ored Fruit-Dove, and Fiji Shrikebill are three digenous landbirds today. Five species of land- Tongan species with relatively poor inter-island birds (Swamp Harrier Circus approximans, dispersal abilities (i.e., speciesthat we have nev- Many-colored Fruit-Dove, Blue-crowned Lori- er seen flying over the ocean anywhere in Oce- keet, Pacific Swallow Hirundo tahitica, and Fiji ania) and a strong preference for mature forest Shrikebill) currently occur in Ha’apai but not in (Steadman 1998). These three species already Vava‘u. Both the Swamp Harrier and Pacific are gone or rare on nonvolcanic islands in Va- Swallow prefer to forage near fresh water or va‘u and throughout Tonga. brackish wetlands, the scarcity of which ac- The Pacific Pigeon is unusual in preferring counts for their absence or extreme rarity in Va- mature forest yet being an excellent disperser va’u. Two of the three remaining species(Many- over water (Diamond 1974). Rarity also may af- colored Fruit-Dove, Fiji Shrikebill) are extirpat- fect the relationship between distribution and ed or rare on nonvolcanic islands throughout observed dispersal of species. With all else Tonga. On the 13 nonvolcanic islands Steadman equal, a rare speciesis less likely to be recorded surveyed in Ha’apai, the Many-colored Fruit- crossing water than a common species.The Ton- Dove and Fiji Shrikebill occurred on only one gan Whistler, however, is common on forested and zero islands, respectively. These species islands within sight of each other in Vava’u, but seem to require mature forest, although Many- was never recorded over water during our nu- colored Fruit-Doves can exist in limited num- merous inter-island boat trips. bers without mature forest if large Ficus trees 5. Interactions (disease, competition, preda- are available. We are unable to explain why the tion) with non-native birds. This topic is unstud- nectarivorous Blue-crowned Lorikeet thrives on ied in Tonga. The chicken (native to Southeast certain islands in Ha‘apai yet is absent through- Asia) is the only non-native speciesof bird firm- out Vava’u, where there is abundant seemingly ly established in Vava’u. This domesticate has suitable habitat (open plantation, wooded plan- lived in Tonga since human colonization (Stead- tation/early successionalforest). man 1993). The Red-vented Bulbul Pycnonotus Two forest obligates, the West Polynesian cafer (native to SoutheastAsia) is abundant to- Ground-Dove and Tongan Whistler, are the only LANDBIRDS OF VAVA’U, TONGA 625

through linear regression (Fig. 8). The positive relationships shown, however, are weak (low z values), in spite of the inclusion of islands that span nearly five orders of magnitude of change in A. Relatively low values for z are expected for remote island groups that have depauperate - Regression Line modem avifaunas (Diamond and Mayr 1976). u) 0 We believe that this trend toward low z values O------95% Confidence Interval -I I I I I is reflected in our data from Vava’u, where z = 0.01 0.1 1.0 10 100 0.212 for the less comprehensive S data (Fig. 8a) versus z = 0.142 for the more complete S data (Fig. 8b). Based on our knowledge of modem b. X+A Polynesian avifaunas and the limitations of our surveys, we believe that the S values in Figure ilOd 8b reflect the actual avifaunas of islands more accurately than those in Figure 8a. Both values, however, consider only modem records. If pre- = 5 historic data were available, the values for S un- x doubtedly would be much greater. How z would - Regression Line % ! be influenced by the prehistoric data is specu- 0 - -- - - 95% Confidence Interval lative, although preliminary data from Ha’apai I , I I and ‘Eua suggestthat, in the absence of human 0.01 0.1 1.0 10 100 impact, z would approach zero for Tongan is- ISLAND AREA (log) lands > 1 km2 (Steadman 1998, unpubl. data). FIGURE 8. Species richness vs. island area for in- Except for the large island of ‘Uta Vava’u and dividual islands in Vavdu. Island areas(km? are from the extremely small islands of Ku10 and Nuku, Table 1. Speciesrichness values (X, X + A) are from Table 4. For 8a, F,,,, = 25.1, P < 0.001, 2 = 0.212. the influence of A on S seems to be minimal in For 8b, F,,,, = 32.8, P < 0.001, z = 0.142. Vava’u. Much of the overall positive relationship between S and A may only reflect that it has been more difficult for humans (through species that we recorded in Vava’u but not in both direct and indirect means) to extinguish Ha’apai, where most islands lack submature/dis- populations of birds on large islands than on turbed mature forest as well as mature forest. small islands (Steadman 1995). That habitat The White-rumped Swiftlet has a much broader distribution in Vava’u than in Ha’apai because modification is an important variable is suggest- this species requires caves or rockshelters for ed by examining the points in Figure 8b that roosting and nesting. Such features are absent on deviate the most from the fitted regression line, all islands surveyed in Ha’apai except Tofua and i.e., those that lie outside the 95% confidence to a very limited extent on Nomuka. intervals. Those lying below are the highly disturbed islands of Ofu and Foelifuka. Those lying SPECIES-AREA RELATIONSHIPS above the line are islands retaining much forest The values for speciesrichness (S) versus island in categories4 or 5, such as ‘Utungake, A’a, and area (A) for all islands surveyed in Vava’u pre- ‘Eueiki. scribe an overall positive relationship (Fig. 8). As island area approacheszero, it is intuitive- We use a semi-log rather than log-log plot of ly obvious that one after another resident species species-areavalues in order not to dampen the must drop out as home range requirements are inter-island variation in S (Gilbert 1980). Our not met for even one pair. If all islands in Vava’u use of 1ogA rather than A conforms to standard were still largely forested, however, the mini- practice. This treatment emphasizes the influ- mum size required to sustain the complete mod- ence of very large and very small islands on the em avifauna seems to be only about 1 km2 or relationship between S and A, here expressedas less (Fig. 5). Whether this was true in the past, z, the slope of the line fitted to the points when species richness undoubtedly was much 626 DAVID W. STEADMAN AND HOLLY B. FREIFELD greater, awaits paleontological tests on multiple tion, especially S. Fakaosi, S. Faletau Fotu, L. Muller, islands in Vava’u. N. Prescott, and U. Samani. We also thank C. and L. Matavelea for their continuing cooperation and infor- mation exchange about Tongan birds. M. Goldin as- CONSERVATION IMPLICATIONS sisted in data entry. Comments that improved the The overall species richness and relative abun- manuscript were kindly provided by D. R. Drake, J. dance of land birds in Vava’u today are greater Franklin, A. W. Kratter, D. J. Levey, B. K. McNab, J. in native forests than in other habitat categories. K. Sailer, K. E. Sieving, and M. I. Williams. The remnant patchesof mature forest sustain the LITERATURE CITED remaining populations of species such as the West Polynesian Ground-Dove and Tongan hADON, D. 1942. Birds collectedduring the Whitney Whistler. The populations of the Pacific Pigeon, South Sea Expedition. L. Notes on some non-pas- Purple-capped Fruit-Dove, and Wattled Honey- serinegenera, 2. Am. Mus. Novit. 1176. htADON, D. 1943. Birds collectedduring the Whitney eater also depend on native forests and are most South Sea Expedition.52. Notes on some non-pas- abundant in mature forest. That these species serinegenera, 3. Am. Mus. Novit. 1237. were recorded at lower levels in more disturbed AMERSON, A. B., JR.,W. A. WHISTLER,AND T D. SCHWA- habitats may reflect a more transient use of these NER. 1982. Wildlife and wildlife habitatof Ameri- habitats or a “source-sink” situation (sensu Pul- can Samoa. Il. Accounts of flora and fauna. U.S. Dept. Int., Fish and Wildl. Serv., Washington,DC. liam 1988, Brawn and Robinson 1996) in which BEICHLE, U. 1991. Statusand acousticaldemarcation of breeding populations are sustained only by the pigeonsof WesternSamoa. Notomis 3881-86. less disturbed source areas. BELL~NGHAM,M., ANDA. DAVIS. 1988. Forestbird com- The most vulnerable species of birds already munitiesin WesternSamoa. Notornis 35117-128. have been lost throughout Tonga (and all of BOLICK,L. A. 1995. Conservationand park development at Mount Talau, Vava’u, Kingdom of Tonga. M.A. Polynesia) to anthropogenic predation, habitat thesis,San Diego State Univ., San Diego, CA. loss, and perhapsdisease (Steadman 1989, 1993, BRAWN,J. D., ANDS. K. ROBINSON.1996. The intcrpre- 1995). Land-use by humans is intense on most tation of long-termcensus data. Ecology 77:3-12. islands in Vavdu today, and during our field sur- BURLEY, D. V. 1994. Settlementpattern and Tonganpre- veys some of the last significant tracts of mature history reconsiderationsfrom Ha’apai. J. Polynes. sot. 103:379-411. forest on ‘Uta Vava’u were being cleared for BURLEY,D. V. 1996. Sport, status,and field monuments agriculture. Given the evident importance of main the Polynesianchiefdom of Tonga: the pigeon ture forests for some speciesand the large num- snaring mounds of northern Hdapai. J. Field Ar- ber of extinctions that already have occurred in chaeol.23:421-435. BURLEY,D. V., E. NELSON,AND R. SHUTLER JR. 1995. Tonga, protection of the native forests of Vava’u Rethinking Tongan Lapita chronologyin Hdapai. (as well as those on ‘Eua and the various vol- Archaeol.Oceania 30: 132-134. canic islands) may be essential to preserve what CHRISTOPHER,B. 1994. The Kingdom of Tonga: a ge- little remains of Tonga’s indigenous avifauna. ographyresource for teachers.Friendly Islands Book Finally, the fundamental aspects of behavior Shop, Nuku’alofa, Tonga. DAVIDSON, J. M. 1971. Preliminaryreport on an archae- and ecology are unknown or poorly documented ological survey of the Vava’u Group, Tonga. Roy. for most speciesof Tongan birds. Detailed nest- Sot. New ZealandBull. 8:29-42. ing and feeding studies of individual speciesare DIAMOND,J. M. 1974. Colonizationof explodedvolcanic needed to understandtheir role in terrestrial eco- islandsby birds:the supertrampstrategy. Science 84: systems and their responsesto disturbance and 803-806. DIAMOND,J. M., ANDE. MAYR. 1976. Species-areare- alteration of these systems. lation for birds of the Solomon Archipelago.Proc. Natl. Acad. Sci. 73:262-266. ACKNOWLEDGMENTS DICKINSON,W. R., D. V BURLEY, AND R. SHUTLER JR. Field work and data analysis were funded by the Na- 1994. Impact of hydro-isostaticHolocene sea-level tional Geographic Society (Grant 5 132-93 to-J. Frank- changeon the geologicalcontext of island archaeo- lin, D. R. Drake, and DWS), the Social Sciencesand logical sites, northernHa ’apai Group, Kingdom of Humanities Research Council of Canada (grant to D. Tonga.Geoarchaeol. 9:85-l 11. V. Burley), the Ornithology Endowment of the Uni- DRAKE,D. R., W. A. WHISTLER,T J. MOTLEY,AND C. T versity of Florida Foundation, and the National Sci- IMADA. 1996. Rain forestvegetation of ‘Eua Island, ence Foundation (grant EAR-9714819 to DWS). We Kingdom of Tonga.New Zealand.I. Botany 34:65- thank our field associates(L. A. Bolick, D. R. Drake, 77. J. Franklin, T Motley, D. Smith, and V. C. Steadman) DUPONT, J. E. 1976. SouthPacific birds.Delaware Mus. as well as officials of the Tonga Government who Nat. H&t., Monog. Ser. 3. granted researchpermits and other forms of coopera- ENGBRING,J., AND E L. RAMSEY. 1989. A 1986 survey LANDBIRDS OF VAVAU,‘ TONGA 627

of the forest birds of American Samoa.U.S. Dept. dae from Polynesiaand the Solomon Islands.Am. Int., Fish and Wildl. Serv., Washington,DC. Mus. Novit. 516. EVANS,S. M., E J. C. FLETCHER,l? J. LOADER,AND E G. MAYR, E. 1932b. Birds collected during the Whitney ROOKSBY.1992. Habitat exploitationby landbirds South Sea Expedition. XXI. Notes on Thickheads in the changingWestern Samoan environment. Bird (PuchycephaZu)from Polynesia.Am. Mus. Novit. Conserv.Int. 2:123-129. 531. FINSCH,O., ANDG. HAR~AUEI. 1870. Zur ornithologie MAYR, E. 1933. Birds collected during the Whitney der Tonga-Inseln.J. Ornithol. 18:119-140. South Sea Expedition.XXIV Notes on Polynesian FRANKLIN,J., D. R. DRAKE,L. A. BOLICK,D. S. SMITH, flycatchersand a revision of the genus Clytorhyn- ANDT J. MOTLEY.In press. Rain forest vegetation thus Elliot. Am. Mus. Novit. 628. and patternsof secondarysuccession in the Vavdu MAYR, E. 1942. Birds collected during the Whitney Island Group, Kingdom of Tonga.J. Veg. Sci. South Sea Expedition.XLVIII. Notes on the Poly- FRANKLIN,J., ANDM. D. MERLIN. 1992. Species-envi- nesianspecies of Aplonis. Am. Mus. Novit. 1166. ronmentpatterns of forest vegetationon the uplifted MAYR,E., ANDS. D. RIPLEY. 1941. Birds collecteddur- reef limestoneof Atiu, Mitiaro, and Ma’uke, Cook ing the Whimey SouthSea Expedition.XLIV. Notes Islands.J. Veg. Sci. 3:3-14. on genusL.uZuge Boie. Am. Mus. Novit. 1116. FRANKLIN,J., ANDD. W. STEADMAN.1991. The potential PARK,G., J. HAY, A. WHISTLER,T LQVECROVE,AND P for conservationof Polynesianbirds throughhabitat RYAN. 1992. The National Ecological Survey of mappingand speciestranslocation. Conserv. Biol. 5: WesternSamoa: conservation of biologicaldiversity 506-521. in the coastal lowlands of Western Samoa. Dept. FREmD, H. B. 1998. Temporaland spatialvariation in Conserv.,Ministry Ext. Rel. and Trade,Wellington, forest birds on Tutuila Island, American Samoa. New Zealand. Ph.D. diss.,Univ. Oregon,Eugene, OR. PRAY, H. D., l? L. BRUNER,AND D. G. BERREIT.1987. FRITH.H. J.. E H. J. CROME.AND T 0. WOLFE. 1976. The Birds of Hawaii and the TropicalPacific. Prince- I&d of fruit-pigeonsin ’New Guinea.Emu 76:49- ton Univ. Press,Princeton, NJ. 58. PULLIAM,H. R. 1988. Sources,sinks, and populationreg- GIFFORD,E. W. 1929. Tongansociety. Bishop Mus. Bull. ulation.Am. Nat. 132:652-661. 61. RALPH, C. J., G. R. GEUPEL,I? PYLE,T E. mm, AND G~ERT, E S. 1980. The equilibriumtheory of island D. E DESANTE.1993. Handbookof field methods biogeography:fact or fiction? J. Biogeogr.7:209- for monitoringlandbirds. Gen. Tech. Rep. 144, Pa- 235. cific SouthwestRes. Sta., Forest Serv., U.S. Dept. GILL, B. J. 1988. Records of bids and reptiles from Ag., Albany, CA. Tonga.Records Auckland Inst. Mus. 25:87-94. REMSEN,J. V., JR. 1994. Use and misuseof bird lists in community ecology and conservation.Auk 111: GILL, B. J. 1990. Recordsof wildlife from Tonga, es- 225-227. pecially Vavdu. Records Auckland Inst. Mus. 27: REMSEN, J. V., JR., ANDD. A. GOOD. 1996. Misuse of 165-173. data from mist-net capturesto assessrelative abun- Hmo,R.L.,S.M. PLETSCHET,ANLIP HENDRICKS. 1986. dancein bird populations.Auk 113:381-398. A fixed-radiuspoint count method for nonbrccding RINKE,D. R. 1984. Zur biologiedes schuppenkopf-hon- and breedingseason use. Auk 103:593-602. igfressers(Foulehaio cun~ncuZutu).Trochilus 5:96- INNIS, G. J. 1989. Feeding ecology of fruit pigeonsin 129. subtropicalrainforests of South-easternQueensland. RINKE,D. R. 1985. Zur biologiedes blaukappchens(Vini Aust. Wildl. Res. 16365-94. uu_struZis),mit bemerkungenzum statusder maidlor- BIRCH,I? V. 1994. The wet and the dry: irrigationand is (gattungVini). Trochilus6:29-40. agriculturalintensification in Polynesia.Univ. Chi- RINKE,D. R. 1986. The statusof wildlife in Tonga.Oryx cago Press,Chicago. 20:14&151. LAYARD,E. L. 1876. Notes on the birds of the Naviga- RINKE,D. R. 1987. The avifaunaof ‘Eua and its offshore tors’ and Friendly islands, with some additionsto islet Kalau, Kingdom of Tonga.Emu 87:2&34. the ornithology of Fiji. Proc. Zool. Sot. London RINKE,D. R. 1991. Birds of the islandsof ‘Ata and Late, 1876:490-506. and additionalnotes on the avifaunaof Niuafo’ou, LEVEY, D. J. 1988. Spatialand temporalvariation in Cos- Kingdom of Tonga.Notornis 38:131-151. ta Rican fruit and fruit-eatingbird abundance.Ecol. RINKE,D. R., H. ONNEBRINK,AND E. CURIO.1992. Mis- Monogr. 58:251-269. cellaneousbird notes from the Kingdom of Tonga. LEVEY,D. J., ANDE G. STILES.1992. Evolutionarypre- Notomis 39:301-315. cursorsof long-distancemigration: resource avai- RIPIYEY,S. D., ANDH. BIRCKHEAD. 1942. Birds collected ability and movementin Neotropicallandbirds. Am. during the Whitney South Sea Expedition. 51. On Nat. 140:447-476. the fruit pigeonsof the Ptilinopuspurpurutus group. LOVEGROVE,T, B. BELL,AND R. Hay. 1992. The indig- Am. Mus. Novit. 1192. enouswildlife of WesternSamoa: impacts of cyclone SNOW,D. 1981. Tropicalfrugivorous bids and their food Val and a recovery and managementstrategy. Dept. plants:a world survey.Biotropica 13: 1-14. Conserv.,Ministry Ext. Rel. and Trade,Wellington, SOTOSSERRANO, C. 1982. Los pintoresde la expcdicion New Zealand. de Alejando. Malaspina.Real Academia de la His- MAYR, E. 1932a. Birds collected during the Whitney toria, Madrid. South Sea Expedition.XVIII. Notes on Meliphagi- STEADMAN,D. W. 1989. New speciesand records of 628 DAVID W. STEADMAN AND HOLLY B. FFCEIFELD

birds (Aves: Megapodiidae,Columbidae) from an STEADMAN,D. W. 1998. Statusof land birds on selected archaeologicalsite-on Lift&a, Tonga. Proc. Biol. islandsin the Hdapai Group, Kingdom of Tonga. Sot. Washinaton102:537-552. Pacific Sci. 52:14-34. STEADMAN,D. WY 1993. Biogeographyof Tonganbirds THOMPSON,C. S. 1986. The climateand weatherof Ton- before and after humanimpact. Proc. Nat. Acad. Sci. ga. New Zealand Meteor. Serv., Wellington, New 90818822. Zealand. STE~MAN,D. W. 1995. Prehistoricextinctions of Pacific TOWNSEND,C. H., ANDA. WETMORE.1919. Reportson islandbirds: biodiversity meets zooarchaeology. Sci- the scientificresults of the expeditionto the tropical ence 267:1123-1131. Pacific in chargeof AlexanderAgassiz, on the U.S. STEADMAN,D. W. 1997a. Extinctions of Polynesian Fish Commission steamer “Albatross,” from Au- birds:reciprocal impacts of birds and people:p. 5 l- gust, 1899, to March, 1900, CommanderJefferson 79. In I? V. Kirch and T. L. Hunt leds.1.Historical E Moser, U.S.N.. Commandina. XXI. The birds. ecologyin the PacificIslands. Yale Univ~Press,New Bull. Mus. Comp: Zool. 63:151r225. Haven, CT WATLING, D. 1982. The birds of Fiji, Tonga,and Samoa. STEADMAN,D. W. 1997b. Historic biogeographyand Millwood Press,Wellington, New Zealand. community ecology of Polynesian pigeons and Wm, S. J. 1875. List of Samoanbirds, with notes doves.J. Biogeog.24:157-173. on their habits.Ibis 1875:436-447.