The conservation of critically endangered flightless birds in New Zealand

IBIS 137: S181-Sl90 The conservation of critically endangered flightless birds in New Zealand M. N. CLOUT 81 J. L. CRAIG Centre for Conservation Biology, School of Biological Sciences, Tamaki Campus, University of Auckland. Private Bag 9201 9, Auckland, New Zealand The current status of Kakapo Strigops habroptilus and Takahe Porphyrio mantelli is described along with recent developments in programmes for their conservation. Both species were (at different times) thought to be effectively extinct, and both have been temporarily reprieved by the discovery of new populations. Population declines have continued, with Kakapo now reduced to less than 50 individuals and Takahe to about 150. Kakapo are especially at risk: 87% of the remaining birds are over 14 years old and only 17 females are known. Research on relict populations of both species has identified predation and competition from introduced mammals as major threats. Both species have high rates of egg infertility and low survival of young. Increasingly intensive management of both Kakapo and Takahe over recent years has included translocation to predator-free island refuges, supplementary feeding to encourage breeding, clutch manipulation, captive rearing and predator control. All known Kakapo have now been transferred to three island refuges, where the overall rate of population decline has slowed and supplementary feeding has apparently encouraged more frequent breeding attempts. Takahe conservation has concentrated largely on attempts to increase the population in Fiordland, New Zealand, through clutch manipulation and release of captive-reared young, but birds have also been released on four islands, which now hold 19% of the total population. The relict Fiordland populations of both Kakapo and Takahe were confined to apparently suboptimal habitat. Both species have successfully adapted to novel environments and foods when translocated, and the populations which now exist present improved opportunities for intensive management using a range of conservation techniques to enhance productivity and survival. Recent population trends of Kakapo and Takahe are reconstructed, and the contribution of research to their conservation is reviewed. The terrestrial avifauna of the New Zealand archipelago evolved in the absence of land mammals and is characterized by a high level of endemism and a large number of flightless species (Bell 1991). As a consequence of human colonization of the archipelago, many endemic birds are now either extinct or endangered (Holdaway 1989). The attributable causes of decline include hunting, collecting for specimens, habitat loss or degradation and competition and predation by introduced mammals. The latter factor remains the major threat to surviving species. Because of the scale and severity of the impacts on its avifauna and the critical state of many surviving endemic species, New Zealand experience in bird conservation is of special interest to ornithologists involved in species recovery programmes elsewhere. The lessons learned in New Zealand may be relevant to the management of endangered species in other parts of the world. In this article we describe the status of two critically endangered flightless birds in New Zealand, the Kakapo Strigops habroptilus and Takahe Purphyria rnantelli, and we review recent developments in programmes for their conservation. We have selected these species for close attention because S181 of their critical state, the well-documented role of research in their conservation and some close parallels in their conservation history and the threats they face. Both species were (at different times) thought to be effectively extinct and both were at least temporarily reprieved by the discovery of new populations. We reconstruct recent population trends of the Kakapo and Takahe, consider the relative merits of intensive and extensive management of both species and review the contribution of scientific research to their recovery. THE KAKAPO The Kakapo (Fig. 1) is a nocturnal. flightless parrot. It is entirely herbivorous and is by far the largest parrot in the world, with adult males weighing up to 3.6 kg (D.V. Merton, pen. comm.). Kakapo were originally very abundant and widespread in a range of forest and scrub habitats throughout New Zealand (Best & Powlesland 1985). A marked contraction in their distribution began after Polynesian settlement, but they were still very common in Fiordland (Fig. 2) and some other parts of the South Island at the time of

S182 M. N. CLOUT & J. L. CRAIG IBIS 137 h, Little Barrier Island 1 Tiritiri Island Island Maud Figure 1. The Kakapo. a ground-nesting parrot in New Zealand. Photo: P. Morrison and Department of Conservation, New Zealand. Island European settlement in the middle of last century (Williams 1956. Butler 1989). The diet of Kakapo consists of leaves, leaf bases, buds, roots, fruits and seeds of a variety of plant species (Best & Powlesland 1985). Both sexes are solitary and occupy individual home ranges of 15-50 ha (Best & Powlesland 1985, Moorhouse & Powlesland 1991). They breed irregularly, at intervals of 2-4 years (Powlesland et al. 1992) and are lek breeders (Merton et ai. 1984). Male Kakapo in breeding condition call and display from traditional track-and-bowl systems, which they construct, maintain, and defend and to which females come for mating. Females nest in hollows at ground level and lay clutches of one to five eggs, which they incubate alone for about 25 days (Powlesland et al. 1992). Nestlings are raised by the female alone and remain in the nest for a further 10 weeks. Normally two or less are raised to independence. Kakapo routinely forage and roost on or near the ground and are vulnerable to predation by introduced mammals (Best & Powlesland 1985). This predation is apparently heaviest on nestlings and nesting females, possibly explaining the adult sex ratio which is biased heavily towards males. Demise of the Fiordland population A major decline in Kakapo abundance in Fiordland was first noted late last century, following the spread of introduced Ship Rats Rattus rattus and mustelids. By the middle of this century, Kakapo were apparently reduced to scattered individuals living on the high slopes of a few steep valleys in Fiordland (Butler 1989). Following the realization of their extreme rarity, six birds were captured in the 1960s for possible captive breeding. All proved to be males and all died in captivity (Merton 1985. Butler 1989). The discovery that Kakapo are lek breeders (Merton et al. 1984) led to the abandonment of captive breeding efforts. From 1974. the focus of Kakapo conservation shifted to attempts to find female birds and establish a population on a predator-free island (Butler 1989). Between 1974 and 1978, 18 Kakapo were found in Fiordland, all of which were adult 4 Fiordland - 0 100 200km Figure 2. Locations where Kakapo or Takahe have occurred recently in New Zealand. males (Merton 1987, Butler 1989). Since there was no evidence of breeding. all 18 were presumably present in 1974. Two of them were translocated to Maud Island (Fig. 2) in 1974, another in 1975 and two more in 1981. The first two survived for 3 years, when one was accidentally killed and the other disappeared. The two transferred in 1981 both disappeared shortly after release (Butler 1989). The bird translocated in 1975 still survives (R.J. Moorhouse, pers. comm.), having been transferred to Little Barrier Island in 1982. The Kakapo left in Fiordland declined to extinction by 1989 (Fig. 2), their numbers dropping from 15 to 6 between 1977 and 1981. The three or four birds remaining in 1987 (Butler 1989) had all gone by 1989 (D.V. Merton. pers. comm.). The sole known survivor of the Fiordland population is the male on Little Barrier Island. It was already adult in 1975 and is now well over 20 years old. Decline of the Stewart Island population In 1977, when it was becoming apparent that the relict Fiordland population consisted of only males (and was hence functionally extinct), a new population of Kakapo, including females, was discovered in a large area of shrubland and forest in southern Stewart Island (Fig. 2). It has been estimated that, on discovery, this population contained 100-200 birds (Powlesland 1989). Stewart Island is free of mustelids but contains Feral Cats Felis catus. Shortly after the discovery of the Stewart Island

1994 CONSERVATION OF FLIGHTLESS BIRDS S183 160 140 120 100 80 Stewart Island Translocated from Stewart Island Translocated from Fiordland * Breeding season 60 40 20 Figure 3. Trends in minimum estimated population of Kakapo. 1977-1993 (estimates refer to end of the calendar year). population, an episode of severe predation by cats on Kakapo was recorded. Kakapo remains were identified in 6 of 225 cat droppings collected from 1977 to 1979 (Karl & Best 1982). and the corpses of 15 Kakapo killed by cats were found in 18 months during 1980-1982 (Best & Powlesland 1985). In 1981-1982, the cat predation rate on adult radiotagged Kakapo reached 56% per annum (H.A. Best in Moorhouse & Powlesland 1991). The favoured explanation of this surge in predation was that one or more particular cats had learned to hunt Kakapo and were doing all of the killing (Butler 1989). This is supported by the fact that predation apparently ceased soon after cat control was instigated in 1982. No Kakapo were found dead after 1983. although cats continued to use the area inhabited by the remaining birds (Butler 1989). The decline in occupancy of track-and-bowl lek sites by displaying male Kakapo is a further indication of the scale of loss and provides a further means of estimating the size of the Stewart Island population in 1980. before the main episode of cat predation. Powlesland et al. (1992) noted that 62% of track-and-bowl systems were occupied in the 1980-1981 breeding season but only 17% in the next breeding season in 1984-1985. Since it is known from radiotagging studies that all adult males boom at track-and-bowl systems in a breeding season (Moorhouse & Powlesland 1991, Powlesland et al. 1992) and that at least 25 adult males remained on Stewart Island after 1984-1985 (i.e. the number translocated after that time), the 1980 population of adult males can be crudely estimated as 91 individuals (62 x 25/17). The transfer of ten adult males (and one subadult) to Little Barrier Island in 1982 therefore accounts for only a fraction of the decline in occupancy between the two breeding seasons. Assuming that the adult sex ratio was reflected in the 40 males and 19 females caught by 1983 (Merton et al. 1984). the number of adult females can be estimated as 43. yielding an adult population of approximately 134. The total population on Stewart Island in 1980, therefore, may have been about 140, including subadults. This was apparently reduced by predation to little more than the 60 individuals transferred from Stewart Island from 1981 onwards. Transfers of birds from Stewart Island to predator-free islands were undertaken because of grave concerns for the future of the species in the face of losses due to cat predation (Butler 1989). One male and three females were translocated to Maud Island in 1980 and 1981. One female died shortly after transfer, but the others survived. In 1982, Stoats Mustela enninea invaded Maud Island and the Kakapo were transferred to Little Barrier Island (3083 ha), from which cats had been eradicated (Veitch & Bell 1990). A further 18 birds (including seven females) were transferred directly from Stewart Island to Little Barrier Island in 1982 (Butler 1989). It was eventually decided to transfer all remaining Stewart Island Kakapo to predator-free islands. This amounted to a decision to translocate an entire species. From 1987 to 1992 all 37 Kakapo known to remain on Stewart Island were translocated (D.V. Merton, pers. comm.). Thirty (20 males, 10 females) were released on nearby Codfish Island (1396 ha), which had been cleared of predatory Weka Gallirallus australis and Australian Brushtail Possums Trichosurus vulpecula (Veitch & Bell 1990). Five (three males, two females) were released on Maud Island (309 ha), which is often classed as mammal free although it is only 900 m offshore and has been temporarily colonized by Stoats twice in the past 12 years. The last two males taken from Stewart Island were released in 1992 on Mana Island (217 ha), which is now mammal free, following the eradication of House Mice Mus musculus in 1989.

S184 M. N. CLOUT & J. L. CRAIG IBIS 137 Performance of translocated Kakapo Survival of Kakapo translocated since 1982 has been generally high. Of the 22 birds transferred to Little Barrier ISland, at least 14 are known to have survived to 1994 (Greene 1993, D.V. Merton, pers. comm.). Only four birds are known to have died. Three of these were males, two of which apparently died as a result of fights between Kakapo (in successive summers) at the same track-and-bowl system. Yet another male was found injured near this site but survived (R.J. Moorhouse, pers. comm.). The fate of the four Kakapo missing on Little Barrier is unknown, but some probably survive since others have been re-discovered after having been missing for up to 11 years (Greene 1993). The survival rate of translocated Kakapo on Little Barrier has therefore averaged between 97% and 98.5% per annum over the past 12 years. On Codfish Island, at least 25 of the 30 Kakapo released since 1987 are known to survive and none has been found dead. On Maud Island, all five birds released since 1989 survive (along with a captive subadult). Both males released on Mana Island in 1992 have since died (D.V. Merton, pers. comm.). The productivity of translocated Kakapo apparently has been enhanced by a programme of supplementary feeding proposed by Powlesland (1989) and since instituted on Little Barrier Island (James et al. 1991). Maud Island and Codfish Island @.V. Merton, pers. comm.). Supplementary feeding was suggested following the discovery that Kakapo on Stewart Island bred only when protein-rich foods (e.g. seed of mast-fruiting podocarps) were readily available (Powlesland et al. 1992). From 1989, foods such as nuts, apples and sweet potatoes were provided at artificial feeding stations to wild Kakapo on Little Barrier Island. There had been no known breeding attempts by Kakapo in the first 7 years after their transfer to this island (Powlesland et al. 1992). but nesting occurred in 3 of the 5 years after supplementary food was first provided (Fig. 3). and single males fledged from each of two of the eight nests detected. Productivity could have been much higher but for a high proportion (36%) of infertile eggs and the early death or disappearance of chicks from at least three nests (D.V. Merton. pen. comm.). Supplementary feeding was extended to the Codfish Island population in 1992 after the unsuccessful outcome of the first known breeding attempts there. A large crop of podocarp fruit apparently led to nesting by at least four females, but the fruit crop failed before chicks could be raised and some starved (Higham 1992). Three of at least seven known chicks were removed for hand raising, and one of these (a female) survives in captivity. Although all remaining Kakapo are on nominally predator-free islands, they are exposed to periodic invasions by Stoats on Maud Island and the Suspected predation of nestlings by resident Kiore Rattus erulans on Little Barrier and Codfish Islands. Some losses of nestlings on both of the latter islands in the past 3 years have now been attributed to Kiore predation (B.M. Fitzgerald & D.V. Merton. pers. comm.). Translocation of Kakapo to islands where they can be Figure 4. The Takahe. which has been the subject of extensive conservation effort in New Zealand. Photo: Department of Conservation, New Zealand. protected from predation on adults and subjected to intensive management has slowed the rate of population decline (Fig. 3) but has not as yet reversed it. Under the most conservative scenario (Le. no missing birds survive on Little Banier or Codfish Islands) only 48 Kakapo existed at the end of 1993. a decline of 14% since 1989 (Fig. 3). Another male died at a track-and-bowl system on Little Barrier Island in March 1994 @.V. Merton, pers. comm.), bringing known numbers down to 47. The most optimistic scenario (i.e. all missing birds survive) would yield a total population of up to 57 at the end of 1993. The age structure is heavily biased to older birds: only six of 47 known individuals are younger than 14 years. The sex ratio is also biased: 30 males and 17 females. Of these females (including a captive subadult). only six are known to have laid fertile eggs in the past 8 years (D.V. Merton, pers. comm.). The species remains in an extreme crisis. THE TAKAHE The Takahe is a large (1.8-3.3 kg) flightless gallinule (Fig. 4). It is largely herbivorous although some invertebrates are fed to young or used for courtship feeding (Crouchley 1994, J.L. Craig, pers. obs.). Takahe were once widespread throughout New Zealand (Atkinson & Millener 1991) although prior to their rediscovery in 1948. only five specimens were known (all collected in the 1800s) (Williams 1960). The population discovered in 1948 was in an area of alpine tussock grassland in the Murchison Mountains of Fiordland (Fig. 2). Declines in the Fiordland population Estimates of the number of Takahe did not start until 5 years after the population was first discovered in the Takahe Valley-Point Burn area of the Murchison Mountains, and the

1994 CONSERVATION OF FLIGHTLESS BIRDS S185 250 200 150 100 50 Captivity U Islands = Fiordland this dropped to seven in 1968-1969. By 1972-1973 this was reduced to five pairs and then to only three pairs by 1973-1974 (Mills & Lavers 1974). A similar decline from seven to four pairs was recorded for surrounding grasslands that were above the treeline. By 1992 only ten birds remained in the initial study area (Eason & Rasch 1993). At the time of these declines in Takahe Valley, small increases (totalling three to five pairs) were recorded in neighbouring areas. More intensive monitoring in the 1990s has continued to document a decline in Takahe across their whole range. Sixty pairs were known in 1990, but this was reduced to 41 pairs by 1992, a decline of 32%. The 1990-1991 year was classified as a Stoat plague year with a number of adult, juvenile and chick Takahe found dead and partly eaten. The winters of 1991 and 1992 had the highest snowfalls on record when many birds were found dead. The overall decline in the Takahe population prior to the use of islands is illustrated in Figure 5. 1963 1973 1983 1993 Figure 5. Trends in minimum estimated population of Takahe. 1963-1993. total area inhabited was not known until the late 1960s (Reid 1971). Reid (1969) estimated that Takahe roamed an area of up to 4000 km2 extending to sea level in some areas. Now, Takahe are known to occur in low densities over an area of about 650 km2 with 530 km2 being classified as a special area with restricted access. Most of the area is covered in alpine grassland and forests. From 1949, the population in the discovery area has been monitored regularly. Reid and Stack (1974), using sightings and signs of feeding, estimated that the total population was probably 435-510 birds but possibly in excess of 560. Mills and Lavers (1974) consider this an overestimate as not all territories are occupied in all areas. Based on the greater knowledge of Takahe habits in the 1990s. Crouchley (1994) suggests that real numbers were more likely to have been half of Reid s estimate. Difficulties remain in estimating absolute numbers. Managers now rely on minimum number known to be alive that are aged 1 year or more (Eason & Rasch 1993). Reliable estimates of numbers in the Takahe Valley-Point Burn study area are known from the mid-1950s. Williams and Miers (1958) began ringing birds here in 1952 and estimated that between 22 and 33 birds inhabited the 10.5 kmz study area. The intensity of study increased in 1972 (Mills & Lavers 1974) and has been maintained since. Numbers in the area have fluctuated between 9 and 47. The number of breeding pairs shows the extent of the variation and the recent marked decline. Williams and Miers (1958) recorded between 10 and 12 pairs on the valley floors in the 1950s. A similar number of pairs (9-11) was recorded by Reid (in Mills & Lavers 1974) between 1963 and 1967, but Reproductive output The periodic flowering of snow tussocks (Mark 1969) seems to influence breeding in Fiordland. For example, in the summers of 1965-1966 and 1966-1967, there was heavy flowering of snow tussocks in Takahe Valley and 12 and 13 pairs, respectively, nested. The 1963-1964 season had light flowering and only three pairs bred (Reid 1967). Weather also influences the ability of birds to renest. In years of heavy snow or cold conditions, breeding can be delayed and many birds that lose a clutch or fail to hatch any of their first clutch do not renest (Eason & Rasch 1993). In good seasons such as 1993-1994 (S. Boyd, pers. comm.), breeding can be up to a month early and pairs have time to renest and raise young if their first clutch is lost or infertile. Few birds have been recorded breeding before age 2 years (Williams 1957, Reid 19671, and not all birds of breeding age are paired in any season. Reid (1967) recorded 72-87% of adults paired in a season. Most pairs (> 79%) lay a clutch of two eggs. Reid (1967) recorded single clutches of three and four eggs, and Eason & Rasch (1993) recorded a pair incubating two clutches at the same time. A small number of pairs are helped by yearling or 2-year-old offspring from previous years (Williams 1957, Reid 1967, Eason & Rasch 1993, Crouchley 1994). Egg infertility varies between 20% and 30% (Williams 1957, Eason & Rasch 1993). Hatching success of fertile eggs is good (> 70 h), but survival to 1 year of age varies between 27% and 71%, dependent on weather and locality (Williams 1957, Eason & Rasch 1993, D. Eason, pers. comm.). Adult survival rates vary between years and localities, ranging from 73% to 88% per annum (Mills 1975). Reasons for declines Predation, weather and competition from introduced herbivorous mammais have all been implicated in the decline

S186 M. N. CLOUT & J. L. CRAIG IBIS 137 in Takahe numbers in Fiordland. A native rail, the Weka, has been observed to take eggs and young chicks although its total impact is considered minimal (Crouchley 1994). Stoats are implicated in the deaths of eggs, chicks, juveniles and adults (Mills & Lavers 1974, Lavers & Mills 1978. Eason & Rasch 1993, Crouchley 1994). with large numbers of Stoats in the area in some years following mouse irruptions linked to mast seeding of beech Nothofagus trees (King 1983). Stoats are also known to feed on carrion, and there is now a belief (Crouchley 1994) that the decline in Takahe during the 1960s and 1970s was due at least partly to high numbers of Stoats. Research on the impact of deer on quality and availability of preferred Takahe foods, such as snow tussocks (Mills & Mark 1977, Lee et al. 1988, Mills et al. 1989). has shown that Red Deer Celvus elaphus and Takahe preferentially take the same food species and that high densities of deer can reduce food availability such that Takahe can no longer survive in an area. Excluding deer and fertilizing tussock lands can enhance breeding and survivorship. Mills (1975) attributed the major decline of birds in Takahe Valley to the impact of deer. Deer control remains an important component of the management of the Fiordland Takahe population. Brood manipulation, monitoring and intensive management Research on the reproductive output of Takahe from the 1950s demonstrated a high level of unsuccessful breeding (Williams 1957, Reid 1967, Mills 1978). Observations that 25-32% of eggs fail to hatch largely due to infertility and that only in good years did most Takahe pairs experience conditions sufficient to allow successful renesting led to a policy of brood manipulation as a way of increasing productivity in all years. Even where pairs hatched two chicks, it was rare that both were reared to independence. Hence a scheme that checked the fertility of eggs in natural clutches was begun in the early 1980s (Crouchley 1994). The aim was to ensure that all nesting pairs had one viable egg. Pairs with two viable eggs were relieved of one, which was then given to pairs whose eggs were not developing. Where a clutch contained one viable and one non-viable egg, the nonviable egg was removed to ensure that the pair directed all of their efforts toward their chick rather than continuing to incubate an egg that would not hatch. Continued monitoring of egg development and laying patterns has now allowed refinement of techniques (Eason & Rasch 1993, Crouchley 1994) so that chick production is maximized each year. Artificial rearing In the 1960s. eggs were taken from the wild to establish a captive Takahe population. The first eggs were reared under bantams, but the Takahe chicks became behaviourally imprinted on their foster parents, which precluded breeding. A series of experiments overcame this and other problems, but breeding in captivity has still had mixed success until recently. One of the problems for managers of captive breeding has been the difficulty of sexing birds with more than a 90% certainty. Recent work with DNA fingerprinting has produced a sex-specific probe (C. Miller, pers. comm.). Egg manipulations in the wild produced an excess of fertile eggs over those needed to ensure that all breeding pairs had one fertile egg. Warm weather also allowed early breeding by some pairs and offered the opportunity for removing the first clutch and then allowing the pair to renest. A programme of artificial rearing was adopted to provide yearlings for release into the wiid to bolster the declining population. This programme involves artificial incubation and feeding using puppets and recorded sounds to ensure appropriate imprinting (Crouchley 1994). Juveniles are then reared in groups with one or more experienced adult Takahe to learn about natural foods and feeding behaviour. The majority have subsequently been released as yearlings back into Fiordland, and some have been released on islands. Survival of captive-reared birds The initial emphasis for release of captive-reared birds was to establish another population in the Stuart Mountains of Fiordland, an area which once had Takahe and was considered optimal alpine habitat. By July 1993, 58 captive-reared Takahe had been released into the Stuart Mountains. Of the 52 released between 1987 and 1991, only 12 (21%) were known to be alive in early 1994. Two of these were paired and were known to have bred in two seasons. The area is now considered marginal (Eason & Rasch 1993). and no further releases are planned. With the decline of birds in the Murchison Mountains, some captive-reared birds have been released into this original area and success here has been high (Eason & Rasch 1993). Performance of island populations Since 1984, Takahe have been introduced on four offshore islands: Kapiti, Mana. Maud and Tiritiri Matangi (Fig. 2), with each of the last three islands holding four pairs. A total of 21 birds have been introduced, and there are now 34 birds on islands. This is 19% of the total Takahe population. Island birds appear to have adult survival rates similar to or higher than Fiordland birds (76-89%, Eason & Rasch 1993). and, although hatching success has been lower than in Fiordland, yearling survival on islands is much higher (89%. D. Eason, pers. comm.). There is also evidence that most island pairs are able to produce replacement clutches every season. Given that the breeding output of Fiordland populations is enhanced by intensive management through egg manipulations, islands would appear to offer the most easily sustained long-term option for Takahe and are another potential source of fertile eggs for artificial rearing (N. Dawson, unpubl. MPhil thesis, University of Auckland). Why were islands not used earlier? The research by Mills and co-workers led them to postulate that Takahe are adapted to live in tussock grasslands. The

1994 CONSERVATION OF FLIGHTLESS BIRDS S187 corollary of this is that the long-term conservation of these birds should be concentrated in alpine grasslands (Mills et al. 1984. 1988). These authors believe Takahe were once more widespread during earlier ice ages when tussock grasslands were more widespread and more northerly. They attributed recent declines and the shrinkage of distribution largely to climate change. Beauchamp and Worthy (1988) have challenged this view and argue that alpine grasslands represent a refuge area where the birds survived because of low hunting pressure. They argue that Takahe were once widespread throughout New Zealand, in forest as well as grassland ecosystems. Atkinson and Millener (1991) confirmed the widespread subfossil distribution of Takahe and showed that Takahe were a regular component of forest habitats throughout the country. They recommended a shift in management focus away from the alpine grasslands, which may in fact be suboptimal habitat for Takahe. Research on habitat requirements has been an important component of rare species management worldwide. Gray and Craig (1991) discussed the problems of extrapolating from current habitat usage to prescriptions of habitat necessities. They stressed the importance of considering the underlying theoretical assumptions, especially assumptions of optimality. genetic determinism and the failure to consider history. Takahe was one of the examples used to illustrate this, and Gray and Craig concluded that management recommendations for this species may be inappropriate because not all assumptions were considered. This has been recognized in current management, which has established small breeding populations on four islands and is now looking for another large island. Whilst intensive management of the Murchison population (including the release of captive-reared birds) appears to be holding or even reversing the decline in Fiordland, it is the establishment of island populations that has contributed most to the recent increase in Takahe numbers (Fig. 3). Research and experimentation on islands are continuing. Takahe on islands have the potential to lay more clutches in any season than do pairs in Fiordland. Currently eggs are taken from wild Fiordland pairs and hatched at a captiverearing facility, but the supply of eggs vanes markedly between seasons. Four pairs are now retained at the captiverearing facility to provide additional eggs. N. Dawson (unpubl. MPhil thesis, University of Auckland) has researched the costs of raising birds by taking eggs from Fiordland, from islands or by rearing young on islands for subsequent transfer to Fiordland. Costs vary six-fold. In a climate of constrained resources, Dawson urges a re-evaluation of current priorities and a greater emphasis on islands as a source of eggs or young. Future directions in Takahe conservation The primary aims of the Takahe Recovery Plan are to establish a self-sustaining population of over 500 Takahe in Fiordland, to establish populations on at least three predatorfree islands and to promote public awareness. The consequence of these goals is that the survival of Takahe is linked primarily to Fiordland, although all research to date suggests that a large Fiordland population will be possible only with large and continuing financial investment. The Takahe Recovery Plan (Crouchley 1994) considers the islands currently holding Takahe as inadequate for long-term survival of the species and raises the issue of another large island or mainland site in addition to the Fiordland population. Assumptions behind the postulated inadequacy of the currently occupied islands for sustaining viable Takahe populations are (1) that a single population of about 500 is necessary for long-term viability, (2) that inbreeding will be problematic in small island populations, (3) that the behaviour of birds in Fiordland is genetically fixed, normal and representative and (4) that subfossil distributions are irrelevant. It is important to consider these issues for the future of Takahe. The number of 500 was recommended by Franklin (1980) as the number necessary to maintain 90% of genetic variability over 100 generations (see also Lande & Barrowclough 1987. Lande 1988). The number was derived from a model for heterozygosity using data from a laboratory study of bristle number in Drosophilu. Not only is there debate about the importance of heterozygosity w allelic diversity, the applicability of the model for wild populations is also in doubt (Craig 1993). At best, the number 500 was an order-ofmagnitude suggestion only and applies to the total species number not to each population (Lande & Barrowclough 1987). The small size of the Takahe population, the high level of infertility and the high loss of young chicks have prompted some Takahe workers to comment on potential problems of inbreeding (e.g. Reid 1967). The closely related Pukeko Porphyrio porphyrio melanotus has very high levels of inbreeding (Craig & Jamieson 1988). Levels of close inbreeding higher than those assumed acceptable in population modelling (e.g. Sod& 1987) are common in New Zealand birds where population sizes are small (Craig 1991). In addition, infertility in Takahe is confounded by low hatching success that may be related to incubation behatriour (D. Eason, pers. comm.). The feeding behavior of Takahe in Fiordland was elevated to a specific adaptation by Mills et al. (1984), and this coloured management towards maintaining the relict population at all costs. Research by Beauchamp and Worthy (1988). Atkinson and Millener (1991) and N. Dawson (unpubl. MPhil thesis, University of Auckland) has since shown that selective feeding is a characteristic of Takahe in all habitats and that alpine tussock grasslands are but one of these habitats. Estimates of likely population sizes on offshore islands (Crouchley 1994) appear lower than expected from extrapolations using island home ranges. Using Dawson s estimates, islands such as Mana and Tiritiri may be able to support populations of up to 100 Takahe, although a compromise figure of 50 may be more realistic. The present policy of establishing breeding populations on islands will allow testing of this and whether or not forests offer any useable habitat. The establishment of multiple populations will also allow testing of the advantages of a metapopulation approach (e.g. Lande 1992, Craig 1993).

S188 M. N. CLOUT & I. L. CRAIG IBIS 137 Table 1. Contributions of research to Kakapo and Takahe conservation Research finding Management outcome Key references Kakapo Lek behaviour Abandon captive breeding Mertin et al. 1984 Cat predation threat Cat control, translocations to islands Best & Powlesland 1985 Home range size Choice of islands Best & Powlesland 1985 Nutritiodbreeding link Supplementary feeding Powlesland et al. 1992 Takahe Diet and deer competition Deer control Mills & Mark 1977 Stoat predation threat Stoat control Lavers & Mills 1978 Low chick survival Clutch manipulation, captive rearing Mills 1978 Alpine tussock specialists Management in Fiordland Mills et al. 1984 Previous widespread distribution Translocations to islands Atkinson & Millener 1991 CONCLUSIONS Although Kakapo and Takahe are taxonomically and behaviourally very different, there are strong parallels in their general ecology, recent history and potential for conservation management. Both species are giant, flightless herbivores. Both are vulnerable to habitat degradation and predation by introduced mammals. Both were originally widespread and common in a variety of habitats but became restricted to apparently suboptimal habitats in Fiordland by the middle of this century (although for Kakapo there was another population on Stewart Island). The vulnerability of both species to the effects of introduced mammals stems partly from their lack of effective defences against predation and partly from their inherently low reproductive rate and dependence on good quality plant foods to induce breeding and raise young. The latter feature is especially pronounced in Kakapo, where females raise altricial young alone and fail to come into breeding condition in most years. The vulnerability of these two flightless herbivores to the impacts of introduced mammals can therefore be seen as a consequence of their evolutionary history, in which the development of gigantism and flightlessness was linked with a low reproductive rate. The only associated feature which has acted in their favour is the inherent longevity of adult birds of both species. This has led to the persistence of relict populations and is again especially pronounced in Kakapo, where more than 8OYo of surviving birds are over 14 years old, and many of them probably are much older than this. Other advantages held by both Kakapo and Takahe are that they readily withstand transfer and are able to adapt to new foods and prosper in habitats which are very different to those in which they were raised. There is a clear lesson here for the management of other species with relict populations. It is unwise to assume that the habitat in which a rare species survives is necessarily optimal, just because it is the last place where a natural population occurs, or that behaviour in that habitat is optimal or fixed. This lesson has also been learned recently in Hawaii, with the Nene Brunta sandvisensis (J. Black, pers. comm.). The survival rates of both Kakapo and Takahe in recent transfers to islands with markedly different habitat and climate have been high, and their acceptance of alien foods (icluding supplementary food at artificial feeding stations) has often been rapid. Once the philosophical objections to transferring an endangered species from its last known habitat are overcome, there are distinct advantages for conservation and species recovery. Among these are the spreading of risk by managing the species as a metapopulation and the ability to maintain the species at chosen sites where intensive management is logistically feasible. The urgent need for intensive management is starkly evident for both Kakapo and Takahe. In both species there has been a huge wastage of young (nestling Kakapo and juvenile Takahe) caused by predation and starvation. Of 42 eggs known to have been laid by Kakapo since 1981, 15 were infertile, at least 21 hatched but only six or seven young were raised to independence (D.V. Merton, pers. comm.). For Takahe. there has been huge loss among yearlings released into the Stuart Mountains of Fiordland. The focus for both species should now turn to maximizing productivity and survival to adulthood. For Takahe, this may entail a shift in emphasis to building up island populations, whilst reducing the effort put into establishing a new population in Fiordland. Prior to translocation from Stewart Island, Kakapo are known to have nested in only 2 out of 9 years, producing three or four independent young from six known nests (Powlesland et al. 1992). Since translocation, they have nested in 4 of the past 5 years (Fig. 2), producing three independent young from 13 known nests. This suggests a higher potential productivity from the island populations, although none of these recent young are likely to have survived without human intervention through either supplementary feeding or hand raising. Although overall Kakapo numbers are now the lowest they have ever been, there is hope in the fact that 13 of 16 adult females are receiving supplementary food on three different islands, fostering and hand raising of

1994 CONSERVATION OF FLIGHTLESS BIRDS S189 young are known to be feasible and the threat of Kiore predation on nestlings is recognized as potentially manageable (D.V. Merton in litt.). Recent experience has shown that, even with supplementary feeding, female Kakapo do not necessarily come into breeding condition every year (D.V. Merton. pers. comm.). Nevertheless, population recovery is likely to be achieved only by continued intensive management, including supplementary feeding to encourage breeding. fostering or hand raising of surplus young and removing the risk posed by potential nest predators. The latter should entail the transfer of Kakapo from Maud Island, which is vulnerable to invasion by Stoats, and the eradication of Kiore from Codfish Island (1396 ha) and Little Barrier Island (3083 ha). For Takahe, the recent trends in population numbers are also of concern, but the potential for intensive management on islands is high, with breeding pairs now on four islands and an upward trend evident in the combined population of these islands. For the immediate future, the best prospect for recovery in the overall numbers of both species lies with intensive management of all populations, using the full spectrum of intervention techniques available to raise productivity and survival of young. Maximum genetic diversity should be retained in the refugee island metapopulations, including manipulation of matings where necessary. For Kakapo, where there is an evident surplus of males and a risk of further deaths caused by fighting, consideration should be given to the temporary removal of some males to other islands where they can be maintained until needed. Research has contributed significantly to the conservation of both species in the past (Table 1) and should continue to be closely linked with their intensive management. Outstanding problems include the nature of the link between nutrition and breeding and causes of the relatively high rates of egg infertility in both species. The long-term goal for both Kakapo and Takahe should be to restore self-sustaining populations to the New Zealand mainland. This will require technological advances allowing perpetual control of introduced mammals. We thank Daryl Eason, Don Merton and Ron Moorhouse for current information on Takahe and Kakapo and Daryl Eason. Carl Jones, Russell Lande. Don Merton and Jeff Black for their comments on the manuscript. The University of Auckland Research Committee and the University of Auckland School of Biological Sciences supported our research, and the British Ornithologists Union provided assistance to attend the conference, Bird Conservation: The science and the action. REFERENCES Atkinson, I.A.E. & Millener, P.R. 1991. An ornithological glimpse into New Zealand s pre-human past. Proc. Int. Om. Congr. XX: 129-192. Beauchamp, A.J. & Worthy, T.H. 1988. Forum on the decline and conservation of the Takahe. J. R. SOC. N.Z. 18: 103-118. Bell, B.D. 1991. Recent avifaunal changes and the history of ornithology in New Zealand. Proc. Int. Orn. Congr. XX: 195-230. Best, H.A. & Powlesland. R.G. 1985. Kakapo. Dunedin: J. McIndoe & New Zealand Wildlife Service. Butler, D. 1989. Quest for the Kakapo. Auckland: Heinemann Reed. Craig, J.L. 1991. Are small populations viable? Proc. Int. Om. Congr. XX: 2546-2551. Craig, J.L. 1993. Metapopulations: Is management as flexible as nature? In Olney, P.J.S.. Mace, G.M. & Feistner. A.T.C. (eds) Creative Conservation: Interactive management of wild and captive animals: 50-56. London: Chapman & Hall. Craig, J.L. & Jamieson, I.G. 1988. Inbreeding in a communal bird: A family affair. Am. Nat. 131: 58-70. Crouchley. D. 1994. Takahe Recovery Plan. Wellington: Department of Conservation. Eason. D. & Rasch. G. 1993. Takahe Management 1992193. Department of Conservation Internal report. Wellington: Department of Conservation. Franklin. I.R. 1980. Evolutionary change in small populations. In SoulB, M.E. & Wdcox. B.A. (eds) Conservation Biology: An evolutionary-ecological perspective: 135-149. Sunderland. Mass.: Siauer Associates. Gray, R.D. & Craig, J.L. 1991. Theory really matters: Hidden assumptions in the concept of habitat requirements. Proc. Int. Om. Congr. XX: 2553-2560. Greene. T.C. 1993. Kakapo Breeding Activity on Little Barrier Island October 1992-April 1993. Auckland Conservancy Technical Report Series No. 8. Auckland: Department of Conservation. Higham. T. 1992. The Kakapo of Codfish Island. N.Z. Geographic 15: 30-38. Holdaway, R.N. 1989. New Zealand s pre-human avifauna and its vulnerability. N.Z. J. Ecol. 12(Suppl.): 11-25. James, K.A.C., Waghorn, G.C., Powlesland, R.G. & Lloyd, B.D. 1991. Supplementary feeding of Kakapo on Little Barrier Island. Proc. Nutr. SOC. N.Z. 16: 93-102. Karl, B.J. & Best, H.A. 1982. Feral cats on Stewart Island: Their foods and their effects on Kakapo. N.Z. J. 2001. 9: 287-294. King. C.M. 1983. The relationship between beech (Nothofagus sp.) seedfall and population of mice (Mus rnusculus). and the demographic and dietary responses of Stoats (Mustela erminea), in three New Zealand forests. J. Anim. Ecol. 52: 141-166. Lande, R. 1988. Genetics and demography in biological conservation. Science 241: 1455-1460. Lande. R. 1992. Neutral theory of quantitative genetic variation in an island model with local extinction and colonisation. Evolution 46: 381-389. Lande. R. & Barrowclough. G.F. 1987. Effective population size, genetic variation, and their use in population management. In SoulB:. M.E. (ed.) Viable Populations for Conservation: 87-124. Cambridge: Cambridge University Press. Lavers, R.B. & Mills. J.A. 1978. Stoat studies in the Murchison Mountains, Fiordland. In Proc. Seminar on the Takahe and Its Habitat: 222-233. Invercargill: Fiordland National Park Board. Lee, W.G.. Mills, J.A. & Lavers, R.B. 1988. Effect of artificial defoliation of Midtibbed Snow Tussock, Chionochloa pallens. in the Murchison Mountains. Fiordland, New Zealand. N.Z. J. Bot. 26: 511-523. Mark, A.F. 1969. Ecology of snow tussocks in the mountain grasslands of New Zealand. Vegetatio 18: 289-306. Merton. D.V. 1985. Kakapo. In Robertson, C.J.R. (ed.) Complete Book of New Zealand Birds: 242-243. Sydney: Readers Digest. Merton. D.V. 1987. Epilogue. In Hill, S. & Hill, J. (eds) Richard

S190 M. N. CLOUT & I. L. CRAIG IBIS 137 Henry of Resolution Island-A biography: 318-323. Dunedin: J. McIndoe & New Zealand Wildlife Service. Merton, D.V., Morris, R.D.&Atkinson, I.A.E. 1984. Lek behaviour in a parrot: The Kakapo Strigops habroptilus of New Zealand. Ibis 126: 277-283. Mills, J.A. 1975. Population studies on Takahe. Notornis rnantelli in Fiordland, New Zealand. ICBP Bull. 12: 140-146. Mills, J.A. 1978. Population studies of Takahe. In Proc. Seminar on the Takahe and Its Habitat: 52-72. Invercargilk Fiordland National Park Board. Mills, J.A. & Lavers. R.B. 1974. Preliminary results of research into the present status of Takahe (Notornis rnantelli) in the Murchison Mountains. Notornis 21: 312-317. Mills, J.A. & Mark, A.F. 1977. Food preferences of Takahe in Fiordland National Park, New Zealand. and the effect of competition from introduced Red Deer. J. Anim. Ecol. 46 939-958. Mills, J.A.. Lavers. R.B. & Lee, W.G. 1984. The Takahe-A relict of the Pleistocene grassland avifauna of New Zealand. N.Z. J. Ecol. 7: 57-70. Mills, J.A.. Lavers. R.B. & Lee, W.G. 1988. The post-pleistocene decline of Takahe (Notornis rnante1li)-a reply. J. R. SOC. N.Z. 18: 112-118. Mills. J.A.. Lee, W.G. & Lavers. R.B. 1989. Experimental investigations of the effect of Takahe and deer grazing on Chionochloa pallens grasslands, Fiordland. New Zealand. J. Appl. Ed. 26397-417. Moorhouse, R.J. & Powlesland, R.G. 1991. Aspects of the ecology of Kakapo Strigops habroptilus liberated on Little Barrier Island (Hauturu). New Zealand. Biol. Conserv. 56: 349-365. Powlesland, R.G. 1989. Recovery plan for Kakapo. 1989-1994. Wellington: Department of Conservation. Powlesland, R.G., Lloyd, B.D., Best, H.A. & Merton. D.V. 1992. Breeding biology of the Kakapo Strigops habroptilus on Stewart Island, New Zealand. Ibis 134: 361-373. Reid, B. 1967. Some features of recent research on the Takahe (Notornis rnantelli). Proc. N.Z. Ecol. SOC. 14: 79-87. Reid, B. 1969. Survival status of the Takahe, Notornis mantelli of New Zealand. Biol. Conserv. 1: 237-240. Reid, B. 1971. Takahe-A vanishing species. In Wildlife 1971- A review: 39-42. Wellington: Department of Internal Affairs. Reid, B. &Stack. D.J. 1974. An assessment of the number oftakahe in the special area of the Murchison Mountains during the years 1963-1967. Notornis 21: 296-305. SoulC, M.E. (ed.). 1987. Viable Populations for Conservation. Cambridge: Cambridge University Press. Veitch, C.R. &Bell, B.D. 1990. Eradication of introduced mammals from the islands of New Zealand. In Towns, D.R., Daugherty. C.H. & Atkinson. I.A.E. (eds) Ecological Restoration of New Zealand Islands: 137-146. Conservation Sciences Publication No. 2. Wellington: Department of Conservation. Williams, G.R. 1956. The Kakapo (Strigops habroptilus. Gray). A review and re-appraisal of a near-extinct species. Notornis 7: 29-56. Williams. G.R. 1957. Some preliminary data on the population dynamics of the Takahe (Notornis rnantelli Owen). Notornis 7: 165-171. Williams. G.R. 1960. The Takahe (Notornis rnantelli Owen, 1848): A general survey. Trans. R. SOC. N.Z. 88: 235-258. Williams. G.R. & Miers. K.H. 1958. A five year banding study of the Takahe (Notornis rnantelli Owen). Notornis 8: 1-12.