Bali starling TNBB (Balen, 1999)

Bas van Balen in 1999 published his PhD thesis at Wageningen University (NL) called 'Birds on fragmented islands: persistence in the forests of Java and Bali' also published as 'Tropical Resource Management Papers No 30' (see sources below). Below article presents the abstract and relevant details on the Bali starling in the West Bali National Park - TNBB, including tables and graphics.

balen1999-balistarlingImage: Bali starling (drawing by Bas van Balen, p.58)


This study describes, analyses and provides suggestions for the amelioration of the impact of age-long deforestation on the distribution of forest birds on the islands of Java and Bali (Indonesia). The first section deals with colonisation and extinction processes of forest birds in a number of remaining forest patches on Java. In the regenerating forest of the Krakatau Islands colonisation and extinction of land birds appear to follow vegetation succession, and therefore seem to affect the monotonie change as predicted by MacArthur & Wilson's equilibrium theory of island biogeography. Extinction of forest birds in the Bogor botanical gardens appears to mirror closely the condition of bird communities in the surroundings of this isolated woodland patch. Distribution patterns of forest birds across 19 highly scattered forest fragments ranging from six to 50,000 hectares show clearly that the ability of birds to survive in surrounding habitat reflects the ability to survive in these patches. To show this, four ecological groups of forest birds have been distinguished: (1) forest interior birds, (2) forest edge birds, (3) woodland birds and (4) rural/urban birds. Nestedness patterns (in which species are found in all fragments larger than the smallest one in which it occurs) are found to be strongest for species restricted to forest interior and edge, weaker for secondary growth, and weakest for rural and urban bird species. A large number of forest interior species appear to be absent from most patches smaller than 10,000 ha, and most are entirely absent from forest patches smaller than 100 ha. In the second section of this thesis the conservation status of three globally threatened, high-profile birds is analysed. The endemic, endangered Javan hawk-eagle Spizaetus bartelsi, traditionally considered amongst the most vulnerable forest dwellers appears to survive in 137-188 breeding pairs in often small and isolated rainforest patches; its survival is explained by (a) juvenile dispersal capabilities, (b) broader niche widths and (c) rather opportunistic feeding. Partly protected by local taboos on hunting, the vulnerable green peafowl Pavo muticus has survived many centuries of human pressure; nowadays at least 1000 birds are scattered across numerous subpopulations. The wild population of the endemic, critically threatened endemic Bali starling Leucopsar rothschildi collapsed since its discovery in 1910 to near extinction in 1990, due to habitat loss and popularity amongst bird-keepers world-wide; despite various conservation measures (captive breeding, awareness programmes, etc.) an intricate web of socio-economic factors prevents the species from emerging from this precarious situation.

Key words: conservation, threatened bird species, Java, Bali, Krakatau, Indonesia, forest fragmentation, extinction, colonisation, Bali starling, Leucopsar rothschildi, Javan hawk-eagle, Spizaetus bartelsi, green peafowl, Pavo muticus, nestedness, captive breeding.

Chapter 5: Status and distribution of the endemic Bali starling Leucopsar rothschildi (S. van Balen, IW. A. Dirgayusa, I M.W. Adi Putra & H.H.T. Prins Oryx [accepted])


(p.59) The Bali starling Leucopsar rothschildi is a passerine endemic to the dry monsoon forest of the island of Bali, Indonesia. Habitat conversion and excessive trapping for the pet trade brought the species to the verge of extinction in the 1980s. The species is critically endangered because of 1) an extremely small population size, 2) restriction to a small area, 3) illegal trapping, and 4) small amount of suitable habitat left within its natural range. An intricate web of factors prevents the Bali Starling from emerging from this precarious situation.


The Bali starling (Bali or Rothschild's mynah) Leucopsar rothschildi represents a monotypic genus which is endemic to the island of Bali (Indonesia). Habitat destruction and capturing for the pet trade brought the species to the verge of extinction. In 1990 and, after a temporary recovery, in 1998 the species was at an all time low number of less than 15 individuals, restricted to the Bali Barat National Park (van Balen & Gepak 1994; I. Z. Mutaqin verbally 1998).

On account of its restricted range, its extremely small numbers in the wild, and pressures on the last free-ranging birds, the Bali starling is considered critically endangered according to the newest IUCN threat categories (Collar et al. 1994). In this paper we made a reconstruction of the species' historical distribution and decline. We also report current population trends and evaluate the possible factors that induced the population crash and those that are likely to inhibit recovery of the population.

Methods - Literature study

A number of reports on Bali starling distribution have been consulted (see van Balen 1995a). Figures obtained between 1974 and 1984 on the population size of the Bali starlings have been interpreted with caution, as these refer to partial inventories, i.e., (p.60) counts of only a restricted number of roosts. For the re-assessment of the population estimates between 1974-1984, the intensity of surveying, coverage, dispersal behaviour of the Bali starling and contemporary weather conditions have been taken into account.

Population censuses

In 1984 the Indonesian Department of Forest Protection and Nature Conservation (PHPA), in co-operation with BirdLife International (then ICBP) started monitoring the wild population on a regular basis. The census method was standardised (van Balen 1995b), based on a counting method already in use by the Bali Barat reserve managers (I M.S. Adi verbally). Until 1991 one annual census of the wild population was made at the end of the dry monsoon (September-October), when the birds associate in roosting flocks. During the recovery of the wild population (see below) a second annual post-breeding census was introduced, immediately after the last young birds had fledged (May or June) and coinciding with the first weeks of the dry monsoon. This provided data on population dynamics, in particular the relationship between fledging success and population size. Each census extended over a period of four days. Dusk and dawn counts lasting 2.5 hours each were made by 9-18 teams of two to three experienced park wardens. Each team was posted at a strategic position near roosting sites, or along flight routes to and from these sites. These sites were identified during reconnaissance visits one or two weeks prior to each census. Numbers of in-going and out-going birds and passers-by, flight direction, times of observation and weather conditions, were recorded on standard forms. To monitor dispersion and range expansion, which might occur with increasing population size, we conducted surveys outside the presently known range. These concentrated along the north and south coast of west Bali and preceded the pre-breeding census when the birds were maximally dispersed.


M. W. Adi Putra took semi-structured interviews with local people living within the historical range of the Bali starling. Fifty respondents of 25 years and older were questioned about the occurrence, numbers and time of extirpation of Bali starlings in their area. Additional information was requested about habitat, seasonality in distribution and roosting sites.

Historical synopsis - Pre 1935. Early distribution and numbers

Stresemann (1912, 1913) collected the first Bali starling known to science near Bubunan, from where its range ran in westerly direction (Figure 5.1). Its habitat was described as "dry savannah and shrub woodlands", and "tall and dense forest" in the 1920s (von Plessen 1926; van der Paardt 1926), historically restricted to a narrow belt of dry monsoon climate (Walker et al. 1980). Monsoon forests exist where there are (p.61) several dry months (each with <60 mm rainfall) annually. They are mostly deciduous, although perhaps only briefly, have less climbers and epiphytes, and are less species rich (Whitmore 1984). In the beginning of this century this forest type was largely undisturbed in the unpopulated north-western part of Bali, and only sporadically interspersed with alang-alang (Imperata cylindrical) fields in particular along the north coast (von Plessen 1926). In the 1920s the Dutch colonial government leased out 600 ha for kapok (Ceiba pentandra) and coconut (Cocos nucifera) in the Teluk Terima area, in the center of Bali starling distribution and now National Park territory.

balen1999-p61-fig5.1Figure 5.1 Distribution of Bali starlings, 1911 - 1998 (1: Bubunan; 2: Gerokgak; 3: Negara; 4: Prapat Agung Peninsula; 5: Gilimanuk; 6: Teluk Terima). (p.61)

Reports from 1914-1926 mention the low densities of Bali starlings in the northern coastal areas of western Bali in February-July, but large numbers in September- November, towards the end of the dry monsoon, throughout NW Bali, including Gilimanuk on the west coast. Hundreds of Bali starlings visiting an abandoned papaw (Carica papaya) plantation in the Teluk Terima area (von Plessen 1926), and flocks of 30 and more birds were seen throughout the entire northwest of Bali in October- November. They foraged solely on fruiting shrubs of Lantana camara (Van der Paardt 1926) which, ironically, is a naturalised pest from tropical America (Backer & Bakhuizen van den Brink 1965). Bali starlings occurred in small numbers in the higher parts of the hills in the interior of west Bali (von Plessen 1926). The north coast of west Bali was the centre of the species' distribution, and numbers declined sharply in easterly and southerly directions. Although he travelled extensively throughout the (p.62) rest of Bali, von Plessen (1926) failed to see the species anywhere else. These observations were confirmed by reports from local people.

West & Pugh (1986) gave a rough population estimate of 300 to 900 birds for wild Bali starling at the time of the discovery. They based this number on the reduction of the species' range, assuming a density similar to those reported over the last decade. Although the Bali starling appears to have always been scarce, this estimate is likely to be far below the real numbers, considering the starling's range which was larger than was thought before, and the huge numbers of birds that were captured and exported between the 1960s and 1980s (see Table 5.1).

1940-1974. 'The Exodus'

Forests on Bali and Java continued to shrink rapidly during this period (Smiet 1992; Whitten et al. 1996) and settlements started to enter the Bali starling's range. In 1947 the Bali Barat game reserve was established theoretically securing a large part of the species' range. However, the Margasari social forestry project was established in 1970 within the game reserve. No information is available from the wild in this period, which is characterised by a general lack of ornithological activities throughout the Indonesian archipelago (Junge 1953; Jepson 1997). The trade in wild Bali starlings reached its height in the early sixties and seventies when hundreds of birds were put on overseas transports (van Bemmel 1974; Seibels & Bell 1993). Table 5.1 lists a number of important events in Bali starling trade (undoubtedly the "tip of the iceberg"). Many Bali starlings were imported even after the species was put on Appendix 1 of CITES in 1970, and received protected status by Indonesian law in 1971.

Table 5.1 Selected events in the history of Bali Starling Leucopsar rothschildi trade

1928 First five Bali Starlings exported to Europe, which bred in captivity three years later (Ezra 1931)
1960-1962 Large imports appeared in Europe: one single assignment of 50 birds reported from UK (Harrison 1968)
1969 Birds found in eleven countries throughout the world, e.g., USA, Sri Lanka, Hong Kong, Belgium (Spilsbury 1970)
1969 Official reports of 55 birds refused, and another shipment entering the USA despite the banned importation after the Endangered Species Act in the same year (King 1974)
early 1970s Hundreds of birds transported into Europe, the majority in hands of malevolent aviculturists (Morrison 1981). Illegal railway transport of 200 Bali Starlings Confiscated in Jakarta (Kamil Oesman verbally 1994)
1972 30 birds counted in Singapore bird park (Morrison 1981)
1976 15 birds in Surabaya bird market for export to Singapore through Jakarta (Suwelo 1976)
1979 Nine birds in seen in Singapore bird park; never less than 19 encountered in the local shops for sale (Morrison 1981)
1982 16 seen together in cages in Denpasar, the property of one man; trappers seen active in one ofthe main roosts in Bali Barat (Ash 1984)

1974-1983 - The decline

Accounts of local people indicate that the conversion of monsoon forest to agricultural land had a negative impact on Bali starlings. Nevertheless, there are reports of Bali starlings occurring in man-made environments and even of Bah starlings breeding in coconut groves (Hayward et al. 1980). Bali Barat received the status of national park in 1982.

Fifty years after its discovery the first information on Bali starling numbers became available. Time and budget constraints resulted in estimates that are extrapolations from censuses of known roosts rather than comprehensive censuses. Population estimates were made almost annually and Tables 5.2 and 5.4 give population estimates between 1974-1980. It should be remarked that spring counts were most likely boosted by fledglings, whereas the autumn counts were taken after the dry season, when most of the annual mortality had occurred.

Interestingly Bali starlings were found in fair numbers at places (notably Manistutu) where von Plessen (1926) failed to observe any. The figures also show that on the periphery of distribution, where most of the poaching took place, numbers rapidly shrunk. Interviews with local people made clear that in the 1960s the Bali starling largely disappeared from the southern part of its range and in the 1970s from the northeast.

1983-1994 - Bali Starling Project

From 1983 to 1994 the species has been the subject of a conservation programme, called the Bali Starling Project (BSP). Its range had shrunk to the fire-induced open shrub and savannah woodland, found below an elevation of 150-175m in the north-eastern part of peninsular Prapat Agung. The open woodlands were dominated by Acacia leucophloea trees with an undergrowth of Lantana camara and Eupatorium shrubs and Imperata cylindrica grass, and intersected by moister and more densely forested valleys with dominant trees Grewia koordesians, Vitex pubescens, Borassus and Schoutenia. This vegetation type might however be sub-optimal habitat for the Bali starling in which it had been 'pushed' by poaching pressure. Movements between roosts and foraging sites increased towards the end of the dry season when food resources ran short. Distances of up to 5-6 km were recorded when part of the birds dispersed in groups of 20-30 into the open mixed forest edge and flooded savannah woodland in the southern parts of the Prapat Agung peninsula. Breeding (December-April) was restricted to an area of 500 ha and nest sites were taken in trees in the above-described valleys. Non-breeding and immature birds also roosted in this area, but covered larger distances (up to 2-3 km) to forage. Table 5.3 gives the results of twelve censuses conducted in 1984-1994. Each of these was based on six counts on consecutive days. The original estimates of totals have been revised by taking into account possible movements between roosts and double counts. The censuses show that Bali starlings have disappeared from areas where they occurred only ten years before.

After a crash of the Bali starling population that started in the early 1980s and which almost completed the extinction of the species in the wild, the population (p.64) recovered in 1991-1992. This was mainly due to improved and effective protection of the park, in addition to favourable weather conditions, which greatly enhanced breeding success (van Balen & Gepak 1994; see Chapter 6 of this thesis). Figure 5.2 shows numbers of the wild population and numbers of birds that fledged. After 1992 breeding was good, but did not result in an increase of the population. Evidence of bird catchers in the park indicated that poaching was the main cause.

Table 5.2 Distribution and numbers of Bali Starling numbers across west Bali, 1974-1980

Year 1974 (1) 1975 (2) 1976 (3) 1977 (4) 1979 (5) 1979 (6) 1980 (7)
Month Oct Feb Sep Jan Mar/Apr ? Aug
1. Cekik - - - - 2 2 2
2. Sumber Batok / Klampok 40 40 - x 4 - -
3. Tegal Bunder / Sawo Kecik 2 20 35 x - 5 10
4. Teluk Lumpur - - - - - 36 25
5. Prapat Agung 2 18 0 x 25 39 43
6. Batu Licin - - - - - - -
7. Lampu Merah 13 13 - x 26 7 16
8. Teluk Kelor 42 - - x 9 26 41 ?
9. Batu Gondang [32] 36 21 x 11 11 10
10. Teluk Brumbun 1 12 - - - 12 6
11. Tanjung Kotal - - - - - 2 21
12. Tanjung Gelap - 9 - - - - -
13. Teluk Terima 6 7 10 x - 7 6
14. Banyuwedang 13 15 82 x 7 18 19
15. Krapeyak - - - - - 2 -
16. Sumber Batok - - - x - 5 8
17. Sumberrejo - - 5 - - - -
18. *Munduk Tumpang - - - - - - -
19. *Pulaki - 5 - - - 7 -
20. *Grokgak - - - - - 2 11
21. *Tegalunan - - - - - 2 -
22. *Yeh Embang - - - - - 5 9
23.*Penyaringan - - - - - 28 22
24. *Manistutu - - 25 - - - -
Estimated Totals 100 100 200 125 110 175 200

Sources: 1. Sungkawa et al. 1974; 2. Natawiria et al. 1975; 3. Suwelo 1976; 4. Sieber 1978; 5. De Iongh et al. 1979; 6. Hayward et al. Oxford Expedition, unpublished data 1980, 7. van Helvoort et al. 1985. *outside Bali Barat National Park

Table 5.3. Distribution and mean numbers of Bali Starlings, 1984-1994

Year 1984 1988 1989 1990 1991 1991 1992 1992 1993 1993 1994 1994
Month Aug Oct Sep Oct May Sep Jun Oct Jun Oct Jun Oct
2 Sumber Batok / Kelampok - - - 4 - - - - - - - -
3 Tegal Bunder / Sawo Kecik 48 5 2 - - - - - - 2 - -
5. PrapatAgung 18 7 - - - 20 9 - - - - -
6. Baru Licin 2 2 - - - - - - - - - -
7. Lampu Merah - 2 3 - - - - - - - - 2
8. Teluk Kelor 8 12 5 5 4 4 31 19 18 14 8 3
9. Batu Gondang 6 - 13 2 27 10 7 18 4 13 19 10
10. Teluk Brumbun - 3 - 3 7 - 6 4 19 4 6 2
11a Tanjung Kotal / Kelompang 2 - - - - - 2 4 4 2 7 8
12. Tanjung Gelap 2 - - - - - - - - - - -
Total 86 31 23 14 38 34 55 45 45 33 40 25
N valid counts 6 5 6 5 5 6 5 6 5 5 5 5

Sources: B.E. van Helvoort & M.N. Soetawidjaya, unpubl. Data; van Helvoort, 1990; van Balen & Soetawidjaya, 1989-1991; van Balen et al. 1990; van Balen & Jepson, 1992, van Balen & Dirgayusa, 1993-1994

1994 to date, Post-BSP

(p.66) In 1994 the Bali Starling Project was continued by the Bali Barat National Park management as sole executors of the field programme. The dwindling population was monitored closely though the 1994-1995 and 1995-1996 censuses were limited to once a year (Suryawan 1996; PHPA/BirdLife International-IP, 1997). On two occasions in 1998 six birds were released (I.Z. Mutaqin & Samedi verbally 1998).

Table 5.4 Population estimates of the wild Bali Starlings

Year Month Population Estimate     References
  Pre Pre Breeding Post
1974 Oct 100 - - Sungkawa et al. 1974
1975 Feb - - 68-144 Natawiria 1975
1976 Sep 175 [+25] - - Suwelo 1976
1976 Dec/Jan - >127   Sieber 1978
1977 Aug/Sep 110 - - Alikodra 1978
1979 Mar/Apr - - 150-200 de Iongh 1982
1980 Aug >207(+22) - - J.R. Hayward et al. (Oxford Expedition unpublished data
1981 Oct 254 - - I M. Sutaadi (Bali Barat NP), unpublished data
1983 Oct 142 - - J Rustandi (PPA/Denpasar), unpublished data

Causes of decline - Habitat conversion

Conversion into coconut and kapok plantations, and settlements replaced most of the former habitat of the Bali starling. Dry hill and lowland forest shrunk from an original area of 3,550 km2 to 600 km2 (MacKinnon et al. 1982). Agricultural lands and a village form enclaves within BBNP and it is estimated that about one third of the ca 10,000 ha of suitable Bali starling habitat in the park has been converted into settlements and plantations (Wind 1991). Also today development projects for tourism and the improvement of Bali's infrastructure pose a threat to BBNP's already affected integrity. Outside the national park, the areas inhabited by people along the north and south coast of west Bali had Bali starlings until as recently as the mid-1980s. Development of these areas is, however, proceeding very rapidly and most of it may be unsuitable already.

Breeding season

balen1999-p67-fig5.2Figure 5.2 Numbers of wild birds and fledglings in 1989-1997; in 1989-1990, 1994-1995 and 1995-1996 no spring censuses have been conducted (after PHPA/BirdLife 1997) (p.67)

(p.67) Early observations of the species' erratic occurrence suggest seasonal movements. Von Plessen (1926) observed Bali starlings in the mountains between May and July, coinciding with the end of the breeding season, before the birds normally disperse from their breeding grounds. This may suggest the occurrence of birds breeding in the hills, as is supported by several nesting pairs of Bali starlings found far inland on foot hills north and east of Negara until the mid-1980s (M.W. Adi Putra this paper). Especially towards the end of the dry season the birds seemed to concentrate in coastal areas, where they had been absent during the wet season (van der Paardt 1926; von Plessen 1926). Since then most of the Bali starling's former habitat has been irreversibly lost to agriculture. This raises the question whether and to what extent the present habitat fragmentation has disrupted the movements between nesting grounds and 'wintering' quarters. It may be necessary to restore habitat ranging from the coastal area to the interior hills, and take this into account in reforestation projects that are planned for Bali. A large part of the above-mentioned areas has the status of Protection Forest and has been proposed as an extension of ca 58,000 ha to the national park (MacKinnon et al. 1982). At present the unclear legal status has resulted in weak protection and general deterioration of the habitat due to human activities.

Illegal trapping

The losses resulting from trapping have been severe during the past three decades. Even in the early 1970s trading took place at a large scale despite national and international bans. At present capturing is mainly to meet the demand for wild-caught birds within Indonesia. Capturing the wild starlings was traditionally accomplished with the use of birdlime and decoys. Birds were also taken out of their nest holes, preferably done during the night. However, more recently mistnets are used and the poachers are reportedly well equipped with telescopes and walkie-talkie's. Poachers (p.68) are known to come from the enclave village and mainland Java. Prices of up to US$ 130 for a live bird in 1982 have been reported (Ash 1984), and nowadays prices of US$ 2000 or more are being paid for the birds on the black market (PHPA/BirdLife IP 1997). Prices as these are strong incentives to catch Bali starlings from the wild that make stopping the illegal trapping and trade extremely difficult.

The importance of efficient protection of the park has been made clear on one hand by the increase of Bali starling numbers as the consequence of a temporary new life blown into the protection and management of Bali Barat in the late 1980s. A second increase in 1991-1992 was also the result of a temporary relief from poaching and favourable weather conditions. On the other hand during the period the zero growth of the population during the otherwise successful breeding seasons of 1992-1993 and 1993-1994 was the result of largely uncontrolled poaching.

Factors limiting recovery - Climate

The rainy season in the north west part of Bali has from December to April, and the dry season from June to October (Sandy 1987). Shifts by several months are known, e.g., in 1991 and 1993, when the rainy season extended until June/July. Figure 5.3 shows the rainfall pattern of this part of Bali. Bali starlings breed only during the wet season. The onset of the rains triggers the development of young leaf sprouts and foliage-consuming caterpillars, items which form the main diet of the nestlings during the first weeks (Cahyadin 1992).

A breeding cycle can be completed in 57 days (van Helvoort et al. 1985). This means that in a wet season of 6-8 months three broods can be raised, but fewer in shorter rainy seasons. Whilst poaching and habitat destruction have been the major factors to the Bali starling's decrease in the wild, the unpredictability of the climate in the region present an extra risk factor that has to be taken into account. The long droughts associated with El Nino Southern Oscillations (ENSO) are known to have an especially deleterious impact on the populations of both seabirds and land birds (Schreiber 1994). Indeed, the ENSO of 1982/83 and that of 1986-1987 coincided indeed with 50% collapses in numbers of wild Bali starlings. Prolonged ENSO influenced periods of drought were also prevalent during the years 1989 through 1990, when Bali starling numbers nearly dropped to zero. In these years hardly any breeding took place and only two fledglings were observed in the first quarter of 1989 (van Balen et al. 1990).

Day and night temperatures differ to the extent that dew formation is substantial. It is believed that during the dry season the Bali starlings in the driest areas largely depend on this when no other water resources are available (van Helvoort 1990). Circumstantial evidence shows that the starlings sometimes take brackish seawater in the mangrove bushes along the coast (I M. Suta Adi verbally). Distribution of Bali starlings in the dry season is believed to be influenced by the distribution of available water sources, and attraction to waterholes makes them vulnerable to capturing.

balen1999-p69-fig5.3Figure 5.3 Average rainfall in northwest Bali (data taken from the Gerokgak Weather Station (after Sandy, 1987)). (p.69)

Nest hole availability

Due to a long history of timber collection and local forest fires, there is a limited number of large trees offering suitable nest sites in peninsular Prapat Agung. Therefore, in 1984-1986, 96 nest boxes and excavated logs were installed to offset the presumed shortage of natural nesting sites in the park (van Helvoort et al. 1986). However, only black-winged starlings Sturnus melanopterus, geckos Gecko sp., ants and wild bees were found to use these (van Balen et al. 1990). Further provision of artificial nests has been discontinued until a fuller understanding of the Bali starling nest hole selection process is established. Moreover, numbers of wild starlings are very low now, and nest hole availability may become a limiting factor only when the population recover substantially.

Inter-specific competition

The black-winged starling, endemic to Java and Bali and believed to be closely related to the Bali starling, is considered a possible competitor for food resources and nest holes (Sieber 1978; Hartojo & Suwelo 1987). Aggressive behaviour between the two species was observed in the breeding season (Natawiria 1975), but it was black-winged starlings that were chased away by Bali starlings from the latter species' breeding territory (Y. Cahyadin verbally) The black-winged starling is a typical open woodland bird and was reported as being very scarce in west Bali in times when forest cover was still extensive and the Bali starlings numerous (von Plessen 1926). Both species met mainly towards the end of the dry season when mixed flocks were reported foraging on the berries of the Lantana camara shrubs (van der Paardt 1926; (p.70) Natawiria 1975). It has been suggested that because of the recent disappearance or increasing scarcity of certain preferred local fruit trees, and the serious disruption of the closed forest habitat, the more arboreal Bah Starling is being crowded out by the black-winged starling in the changed environment (Sontag 1991, 1992). Ironically, the black-winged starling has itself become scarce in the area as a result of excessive illegal trapping for the bird trade, and competition has become irrelevant.

Small population size

In a very small population inbreeding is inescapable and, as it generally affects fecundity (Senner 1980) and fitness/adaptability (Allendorf & Leary 1986), it frequently increases a population's vulnerability to a changing environment. Moreover, natural disasters, such as extended drought, wild fires and disease epidemics could easily wipe out a small isolated population in one stroke.

A new technique known as the population viability analysis (PVA) has become a central component of a number of recent recovery efforts (Scott et al. 1994). Whilst numbers in the wild were down below 20, calculations at the 1990 Bali Starling PVA Workshop suggested a 100% probability of extinction within a year if no action was undertaken. The (partly) successful release of captive-bred birds (van Balen & Gepak 1994) in April 1990, increased protection of the park, and favourable climatic conditions resulted in a temporary recovery in the following two years.

An effective population size has been suggested of not less than 50 in the short term, and 500 in the long term, below which numbers inbreeding would become unacceptably high (Franklin 1980). An effective population size of 500 birds would require a roughly estimated population of 1000 birds, assuming 50% of the birds are contributing to the breeding pool (Seal 1990). This number, if distributed over several wild sub-populations in secure areas on Bali, is believed to be sufficient to ensure the Bali starling's viability. Natural habitat has become scarce and reforestation as well as effective forest protection throughout its range is needed to provide enough area to support this number (Wind 1991). There are also proposals to establish feral populations of Bali starlings in suitable tourist areas elsewhere on Bali (P.R. Jepson verbally).

Genetic erosion

Van de Paardt (1926) reported that Bali starlings "keep shunning people; as soon as village huts appear, the number of birds decline; they are noticeably pushed back by Man". This is in contrast with more recent accounts of birds breeding in coconut groves adjacent to villages (van Helvoort et al. 1985). It may indicate that the Bali starling has been in an assimilation process before it was pushed back by poaching to its present refuge. Adaptive abilities seen in urbanising forest birds in Papua New Guinea (Diamond 1986), or the European blackbird Turdus merula in European cities may be genetically determined (Gehlbach 1988). Captive stock in USA and Europe originates largely from birds imported in the early seventies, when Bali starlings were still relatively abundant. Within easy reach of villages and human settlements, they were thus more susceptible to poaching. Would it be possible that more Bali starlings (p.71) with, than without, genes for a more adaptive character have been caught away from the wild population in the early seventies? And that in the wild a depauperate population remained lacking this adaptability and doomed to being locked up in a still shrinking habitat patch?

Skewed sex ratio

Poachers often obtain their birds at night during the breeding season, when females are in the nest holes, and males hold guard in a nearby tree, safe from poachers (Sieber 1983; I.Z. Mutaqin verbally). Indeed, a higher proportion of female than male birds is known to have been exported to the USA in the 1960s (R.E. Seibels verbally). The resulting skewed sex ratio in the wild is suggested by field observations of three released, captive-bred females paired to wild males (M.N. Soetawidjaya verbally 1990; Cahyadin 1992; I.Z. Mutaqin verbally 1998), whereas no reverse cases are known. Moreover, breeding activity had ceased almost entirely in the years prior to the 1990 releases (van Helvoort verbally).

The negative impact of sex-skewed capturing in small populations is clear. Franklin (1980) pointed out that an unequal number of sexes would give a smaller effective population and thus increased genetic drift. This then causes inbreeding depression, random change in phenotypes and a decrease in genetic variance.

Weak law enforcement

Protecting the Bali starling from the illegal trade seems straightforward and simple. At the "source" end of the trade chain the birds are found in a very small area. At the 'sink' end the situation is far less surveyable, but huge penalties exist (in theory) for possessing protected wildlife (PHPA/BirdLife IP 1997), and these should effectively discourage people from capturing, trading and keeping illegal birds. Reality is different. Rich and powerful people can afford to keep Bali starlings entirely unpunished. In the early 1990s the Indonesian Department of Nature Conservation was charged with a programme aimed at getting hold of captive birds for the breeding and release programme by giving amnesty to illegal keepers of Bali starlings. Implementation was seriously impeded and made ineffectual by an exhibit of power. At the source end the problems are no less complicated. Lack of law enforcement finds its roots in the absence of an effective bonus system for park wardens. Moreover, though only incidentally illegal bird catchers have been caught and punished, to date no single illegal keeper was ever punished. Morale of the park wardens is understandingly low. Failing to address this problem has caused the creation of a perverse incentive, and keeping the population low has actually become advantageous. For park guards it means that their jobs are guaranteed and for the bird catchers that birds make higher prices in the illegal trade.

Future of the Bali Starling

(p.72) Habitat degradation and excessive capturing brought the Bali starling to the verge of extinction. The hazardous effects of a small population, the intricacy of the law enforcement system and harsh local climatic conditions, have so far prevented the Bali starling from emerging from this precarious situation. BirdLife International has been involved in the Bali starling conservation programme from 1984 through 1994, in an attempt to strengthen the management of the Bali Barat national park - the key objective of later phases in its field programme. This, and major achievements in captive release techniques (van Balen & Gepak 1994; Collins et al. 1998) have not yet resulted in a continued recovery of the wild population of Bali starlings. The main problem has been with security in the park. However, topics such as field warden discipline, the arresting and prosecution of poachers and traders, and the gazetting of conservation areas are all outside the sphere of responsibility of international agencies. It is therefore encouraging that a Species Recovery Plan has been prepared and adopted in close cooperation with all implementing parties in Indonesia (PHPA/BirdLife International-IP 1997). In this document all aspects of Bali starling conservation: protection in the wild, extension programmes, habitat restoration, law enforcement, are addressed and it will provide the guidelines for continuation of the conservation programme.

The Bali Starling Project was directed by the PHPA (Ministry of Forestry), BirdLife International, and the American Zoo and Aquarium Association (AZA, formerly AAZPA). Sponsors were the Art Ortenberg/Liz Claiborne Foundation, AZA, the Jersey Wildlife Preservation Trust, and the Swedish Ornithological Society. Paul Jepson, Paul Loth, Bob Seibels and an anonymous reviewer gave valuable comments on drafts of the paper.

Chapter 6: The captive breeding and conservation programme of the Bali starling Leucopsar rothschildi (S. van Balen & V.H. Gepak 1994; Pp 420-430 in Creative Conservation: Interactive management of wild and captive animals. P.J.S. Olney, G.M. Mace & A.T.C. Feistner (eds). Chapman & Hall, London)


(p.73) The Bali starling Leucopsar rothschildi is a passerine endemic to the island of Bali. Habitat loss and illegal trapping caused the decline of numbers in the wild. About a dozen survived in 1990 in the Bali Barat National Park in the north-western tip of Bali. In 1983 a co-operative captive-breeding programme was initiated to re-stock the numbers in the wild. Birds bred in zoological gardens in the United States and Great Britain supplemented a breeding stock kept in the Surabaya Zoo in East Java. In July 1988 and April 1990, three and twelve captive-bred birds were released in the park. The second attempt succeeded in releasing birds that survived longer than six months: one female was observed in October the same year; another female was rediscovered in the subsequent breeding season, paired to a wild bird three young and successfully raising three young.


The Bali Starling (Leucopsar rothschildi) is a strikingly beautiful silky-white bird with black tips to the flight feathers and blue naked skin around the eyes. It was first described and placed in a monotypic genus by Stresemann in 1912, a year after he discovered it in the dry lowland forest along the coast of northwest Bali (Stresemann 1912). Since its discovery the numbers have declined and its distribution has receded. In the 1920s it occupied roughly 30,000 ha of uninhabited land (van der Paardt 1926; von Plessen 1926; van Helvoort 1990), but with the progressive conversion of forest to agricultural land, by the late 1980s its range had shrunk to less than 4000 ha and the population was restricted to a small part of the Bali Barat National Park in the northwest of the island (van Helvoort 1990). In the last 20 years the decline in numbers has been accelerated by trapping for the international pet trade and an increased demand from aviculturists. By 1990 the total wild population was estimated to be as low as 13 (van Balen & Soetawidjaya 1991). The Bali starling has been included in (p.74) the IUCN Red Data Book since 1966, in the Endangered category, and international trade is prohibited under the Convention on International Trade in Endangered Species (CITES). Since 1970 the species has had absolute protection under Indonesian law.

In 1983 the Indonesian government, represented by the Directorate General of Forest Protection and Nature Conservation (PHPA) formally requested the International Council for Bird Preservation (ICBP; Since 1993: Birdlife-International) to draw up and put into action a conservation project for the Bali starling. The implementation of this project was preceded by a feasibility and preparation period of 1983-1986 (van Helvoort et al. 1986), and by 1987 PHPA, ICBP, the American Association of Zoological Parks and Aquariums (AAZPA) and the Jersey Wildlife Preservation Trust (JWPT) had produced a comprehensive five-year cooperative programme. The agreed overall aim was to restore a viable and self-sustaining population in Bali Barat National Park.

The objectives included: (a) monitoring and protecting the birds in the wild; (b) establishing a captive breading programme in Indonesia with input from other captive breeding programmes in America, Jersey and elsewhere; (c) restocking the wild population; (d) promoting public awareness.

A further three-year plan was agreed in 1992, which continued the original aims and expanded the objectives to include: (a) stopping the illegal capture of birds; (b) reducing the demand for wild-caught birds; (c) establishing new populations within the species' dispersive range from captive stock; (d) continuing to promote an awareness of the cultural and aesthetic value of conserving the Bali starling in the wild; (e) undertaking management-orientated studies of the behaviour and ecology of the species; (f) developing the capability of the Bali Barat National Park to be self-sufficient in conserving the species.

Captive stock world-wide

Fortunately there has been for some time a relatively large captive population in zoos, bird collections and private collections world-wide. This population has been estimated to be in excess of 700 individuals (van Helvoort 1990), but only recently has there been an attempt to co-operatively manage parts of this scattered population. Poor record-keeping and uncontrolled breeding has made any analyses and management difficult. There are two regional studbooks which do provide usable data. One, the American studbook, which is under the auspices of AAZPA and one of (p.75) their Species Survival Plans (SSP), recorded as Of July 1992, 381 birds in 68 participating institutions (Seibels 1992). The other, which is under the auspices of the Federation of Zoological Gardens of Great Britain and Ireland, registered as at the end of 1991, 110 birds in 20 institutions (Fisher 1992). In Europe the European Association of Zoos and Aquaria (EAZA) has approved the setting up of a co-ordinated breeding programme (EEP).

Captive breeding programme in Indonesia - Breeding stock

In August-November 1987 the breeding facility already in existence in Surabaya Zoo in east Java was renovated. This facility comprised 29 aviaries with 16 Bali starlings. In November 1987 the captive population was increased with the addition of 37 birds donated by zoos and private collections in the USA and by Jersey Wildlife Preservation Trust. Five of these birds, most of them over 10 years old, died shortly after arrival. In 1992 the breeding stock in Surabaya Zoo consisted of 44 birds, with IS birds (11 male, seven female) born before 1985, 10 birds (four male, six female) born in 1985 or later, and 16 birds (six male, eight female, two unknown) of unknown age, but all born before 1987.


A number of publications on breeding Bali starlings have appeared in the last two decades (Taynton & Jeggo 1988; Partington et al. 1989; see also bibliographies in Seal (1990) and Seibels (1991)). Husbandry used in Surabaya Zoo followed the guidelines given by American zoos (Seibels 1991) and a brief account only is given here.

The breeding aviaries for single pairs were at least 2.5 m high, 2.5 m wide and 4 m deep, well planted with low shrubs and small trees. Breeding results during the first season were disappointing and measures to enhance productivity were taken: (a) aviaries were screened in order to avoid interaction with starlings in adjoining breeding units; (b) in 1989 the old nest-boxes were replaced by boxes that followed a design widely used in the USA (Seibels 1991); (c) disturbance from visitors to the zoo was decreased by closing off the immediate surroundings of the aviaries; (d) birds with poor breeding performance were re-paired; (e) in addition to dry food pellets, fresh local fruits (papaw, bananas) and live food (mealworms, ant pupae) were provided.

Breeding results

Egg hatching during the year was satisfactory. Mortality, however, after hatching remained high and to date an average of only six to nine birds reach maturity each year. The introduction of new nest-boxes in August 1989 resulted in some increase in (p.76) the number of clutches, but the chick mortality stayed high. To date 39 birds have been successfully raised. Figure 6.1 shows the breeding results for 1988-1992. As three pairs that were amongst the most productive of the breeding stock, were stolen in March 1991, the period April 1991 to July 1992 has been omitted from the graph. The stolen birds were retrieved in the second half of 1992.


In order to manage the captive population scientifically, and in particular to minimise inbreeding, carefully maintained studbooks are essential. American and British birds are already registered, and regional studbooks for birds in Indonesia and in Europe are being prepared. Other studbooks should be prepared for birds held in Japan and Singapore.

Taynton & Jeggo (1988) found evidence of increased chick mortality with higher inbreeding levels in Jersey birds, and van Helvoort (1990) reported an inbreeding depression in the American population leading to a reduction in fertility. A recent study (Thohari et al. 1991) indicated an extremely low heterozygosity of certain blood protein types in the captive population held in Indonesia, with as yet unknown implications for the species. The introduction of wild-caught Bali starlings, of which a fair number are still in private hands, in Java and Bali would diversify the captive gene pool.

balen1999-p76-fig6.1Figure 6.1 Breeding results of the Bali starling propagation centre at the Surabaya Zoo, 1988-1992; reliable figures for the period April 1991-June 1992 were not available (I: January-march, II: April-June, III: July-September; IV: October-December; solid bars: number of eggs laid; grey bars: hatchlings; white bars: successfully fledged birds). (p.76)

Minimal viable population

(p.77) It has been tentatively suggested that for a species' long-term survival a minimum effective population of 500 individuals is needed (Franklin 1980). More specifically a recent population viability analysis (PVA) (Seal 1990) considered that to be viable, 1000 birds in the wild and another 1000 in captivity were desirable, with these two populations being managed as a single meta-population. These numbers are based on little empirical data and their feasibility (especially with regard to the wild population) is doubtful, but it is clear that more birds will decrease the risk of genetic deterioration and extinction. The zoos and private collections in the USA can together house only a restricted number of birds. Those registered in the studbook are closely managed and monitored, as are those in the British studbook, but there are many in the widely scattered world captive population which are outside any managed breeding programme. Co-ordination amongst collections is necessary to increase the effective size and viability of the captive breeding programme.

Of the 44 birds at present in the Surabaya zoo only four are successful breeders. Recombination of the pairs, especially among the groups of different origin, is essential to enhance productivity and increase the effective breeding population. The extension of the Indonesian propagation programme to other sites would increase the size of the Indonesian captive population and the number of birds for release. To this end, birds obtained under a one-time amnesty campaign from local private owners in exchange for captive-bred birds brought over from the USA (in June 1992, 17 birds were brought over for this purpose), and those handed over directly with the help of local PHPA officers, are now being registered. To date more than 80 birds have been registered and fitted with transponders. In 1993 they will be placed in the new additional captive breeding centres, or be released into the wild, if considered appropriate.

Release programme

The building of a Pre-release Training Centre (PTC) in the Bali Barat National Park was completed in June 1988. The unit comprises 10 aviaries each 5x3x2m and follows the design of the Captive Propagation Centre in Surabaya zoo. The PTC is located in an off-public site, with restricted access for interested visitors. The aviaries are sufficiently isolated to reduce any habituation to humans, including the bird keeper.

In autumn 1987, when less than 50 birds survived in the wild, a release was felt to be justified. In July 1988 the first group of three birds from the Surabaya zoo was accommodated in the PTC. To adjust the birds to their future environment, they received six week's training which focused on developing skills for foraging in the wild, retaining fear of humans, and gradually accustoming the birds to the boxes in which they were to be transported to the release site. In the field, the birds were released in turns during the first week, in order to maintain the birds close to the release site - the caged birds always attracted the ones already released. This attempt resulted in one known casualty and the disappearance of the other two birds. The extremely dry conditions and strong wind at the time of release, the birds' (p.78) unfamiliarity with the area, and the location of the site distant from any known Bali starling roosting area may have been contributory factors which caused the failure (B. van Helvoort verbally).

During the Bali starling PVA workshop, held in Bogor, Indonesia, and attended by an international group of conservationists, aviculturists, and other experts (see Seal 1990), it was decided that a second attempt to release captive-bred birds into the wild population should be undertaken as soon as possible.

Accordingly, in April 1988, eight captive-bred birds were brought over from Surabaya Zoo to Bali. One bird died shortly afterwards, probably due to stress, and another was considered unfit for release. The birds were given various kinds of wild fruits which were known to have been eaten by Bali starlings. They readily took various arthropods, including scorpions and millipede, and small reptiles, that entered their cages. They showed instinctive reactions towards raptors flying over. In early 1990, two birds confiscated in east Java were added to the PTC group; one of these was considered for release, but the other was assumed to be unfit for release as its malformed bill suggested it had been hand-raised.

In April 1990 another six captive-bred birds were transported from Surabaya Zoo to Bali Barat and housed in the PTC. The second attempt was planned for a location in the far north east edge of the Prapat Agung Peninsula in Bali Barat, a short distance form the Teluk Kelor guard house. Here wild Bali starlings were known to roost regularly. A two-compartment simple cage (2x2x2.5m) was built on the site to serve as framing accommodation. On the day the birds were transported to the release site, transponders were inserted, and colour rings were attached. Special heavy-duty rings designed for the Bali starling conservation programme were attached to the two confiscated birds, whilst the other birds had their metal zoo rings. In order to tell the released birds from the wild birds during at least the first weeks, the breast feathers of the birds to be released were dyed red with rhodamine B. Two coloured plastic spiral rings were attached to all the birds each coded with a unique combination of numbers. On April 15 1990 the first four birds were released from the cage where they had been housed during the previous two days. On April 17, three birds were released, followed by two, two, one and one on each consecutive day. Birds unfit for release stayed in the cage to decoy released birds back to the cage, where food and water were provided during the first weeks. The wild caught confiscated bird that was released with the captive-bred birds, and which was expected to act as a sort of guide was a disappointment: immediately after its reluctant take off, it flew away in a direct line and was never seen again.

Daily monitoring of the starlings by telescope (20-60x) from a hide near the release site was maintained during the first weeks, where food pellets, fruits and drinking water were provided. Acceptance of the wild by the released birds went smoothly and soon mixed foraging, communal anting and roosting flocks could be seen. One pair was formed within one week and the wild bird would follow its partner close to the food and drinking water container. The observation of several birds around the site, but not at the drinking place soon after release, suggested that acclimatisation was rapid in some cases.

(p.79) Reading of the ring codes became increasingly difficult, as the birds became more wary in the process of adaptation, making the success rate hard to assess. This was aggravated by unexpectedly high poaching pressure near the sites. Within one month one of the released birds, detectable by its transponder, was rediscovered in the hands of a local bird dealer, and an unknown number of other birds may have been trapped. In early October 1990, however, six months after the release, a marked bird was identified about 8km from the release site. It was observed copulating with a wild bird (in the kapok tree Ceiba pentandra plantation enclave along the main road that cuts through the National Park), but disappeared soon after. In November the same year, another released bird was rediscovered, paired to a wild bird, hi January the following year this pair successfully raised three young in a free hole not far from the release site.

A major decline in numbers of Bali starlings was found during the pre-breeding census of 1990. Even with the 13 released birds, numbers had dropped to some 15 birds largely due to poaching. The following breeding season guarding of the park was increased, but no releases took place, primarily because sufficient were not available from the Surabaya breeding facility. Poaching appeared to be better controlled, though still going on, and the post-breeding censuses completed in June 1991 and June 1992 showed about 35 and 55 birds, respectively (Figure 5.2). Eight occupied nests were located in the Teluk Kelor/Batu Gondang area. Only natural nest holes were used, and again the nest boxes provided a few years ago were ignored. Though a quarantine period in the PTC is common practice before release, and the birds kept in the zoo are examined regularly, there is still a considerable risk of disease transmission. The incidence of atoxoplasmosis in captive Bali starlings in American zoos (Partington et al. 1989) is especially worrying. Following a discussion paper prepared by PHPA and ICBP, an AAZPA team came from the USA in August 1992 to examine the birds held in Indonesian zoos and in the PTC in Bali. A medical quarantine protocol for all birds to be released and for birds held in captive breeding centres has since been developed (Appendix 4 of van Balen & Jepson 1992).

To further reduce the risk of disease transmission, release in the future will be in former, but now empty, Bali starling habitat. This reinfroduction, as opposed to restocking, will involve rather different and more elaborate release techniques, as no resident guides will be available. An intensive field study is being prepared by ICBP and AAZPA, aimed at collecting data on behaviour and breeding success of released birds (M. Collins verbally). A possible release site has been identified on the island of Menjangan, pending more information on diseases in the captive population and full control of poaching in the area. Furthermore, the use of radio-telemetry is being considered and preliminary tests on captive starlings has had promising results (Elbin et al. 1991).


Habitat availability in the present National Park as a limiting factor on the recovery of the Bali starting is currently being investigated by the project. There may not be (p.80) enough suitable habitat in the national park to support more than a three- or fourfold increase in the present population of 55 Bali starlings. Even if attained, this figure would be far below any number suggested for a viable population. Any continuation of the release programme will have to take this into account and the conversion of the plantation enclaves that exist in the Bali Barat National Park into Bali starling habitat must keep pace with an increasing number of starlings.

Considerable time and effort has been put into the captive-breeding programme, but to date its success in terms of contribution to the conservation of the Bali starling has been limited. The recovery of the wild population following the improvement in protection shows that other techniques can be possibly more immediately efficient. However, the potentially deleterious consequences of inbreeding cannot be discounted and the introduction of new genes is justified. Further releases are planned for 1993 and feasibility studies are now being carried out. To avoid disturbance of the present wild population, sites in Bali starling habitat other than the previous ones will be selected.

The Bali Starling Project Phases 1-3 would not have been possible without the assistance and commitment of a large number of people and institutions. The Project is managed by ICBP and financed by AAZPA and the New York Zoological Society, Liz Claibome/Art Ortenberg Foundation, and JWPT. The PHPA head offices, the Bali Barat National Park and Surabaya Zoo were the local partners in the implementation ofthe Project. In particular, M. Noer Soetawidjaja, Slamet Suparto and Made Rasma c.s. were most closely involved in the release programme. Thanks are forwarded to the former Bali Starling Project officer, Bas van Helvoort, for the discussions and support provided during the early stages of the project, and to Paul Jepson and Professor H.H.T. Prins for their valuable comments on an early draft of the paper.


This email address is being protected from spambots. You need JavaScript enabled to view it. research: Godi Dijkman social facebook box white 24