Timing of Seasonal Breeding in Birds, with Particular Reference to New Zealand Birds*

Review Reprod. Fertil. Dev., 1995, 7, 1-19 Timing of Seasonal Breeding in Birds, with Particular Reference to New Zealand Birds* J. F. Cockrem Department of Physiology and Anatomy, Massey University, Palmerston North, New Zealand. Abstract. A model to explain the timing of seasonal breeding in birds is presented. It is assumed that, despite the wide range in egg-laying seasons, there are common physiological mechanisms which underlie seasonality in birds and that most, if not all, birds are photoperiodic. Birds are considered to possess an internal rhythm of reproduction which is synchronized with seasonal changes in the environment by external factors, particularly the annual cycle of daylength. The rhythm consists, at least in part, of regular changes in the photoperiodic response between states of photosensitivity and photorefractoriness. Avian breeding seasons effectively start in autumn when birds become photosensitive, regardless of when egg-laying occurs. The timing of breeding is then influenced by the rate of increase of hypothalamic 'drive' and by the sensitivity of the hypothalamus and pituitary gland to inhibitory feedback from gonadal steroids. If sensitivity is high, gonadal growth will not occur untii the threshold daylength for photostimulation is exceeded after the winter solstice. Egg-laying then starts in late winter, spring or summer. Alternatively, steroid feedback may be relatively low and gonadal growth may be sufficiently rapid once the birds become photosensitive that breeding occurs in late autumn or winter. The time of egg-laying in birds may also be strongly influenced by supplementary information, such as social cues, food availability, temperature and rainfall and, in some species, this information is more important than daylength in determining the timing of breeding. The review also includes the first summary of the breeding seasons of New Zealand birds. The pattern of egg-laying is exactly the same in native birds, in birds introduced to New Zealand and in other Southern hemisphere birds from similar latitudes, with a broad peak of egg-laying occurring from September to December. In addition, annual cycles of steroid hormone concentrations in the North Island brown kiwi, the yellow-eyed penguin and the kakapo are consistent with results from many studies on Northern hemisphere birds. This model for the timing of breeding in birds can be applied to New Zealand birds and it is concluded that the physiological control mechanisms for the timing of seasonal breeding in New Zealand birds are similar to those of other birds. Extra keywords: photoperiodism, photorefractoriness, egg-laying, breeding season, steroid feedback, photosensitivity, kiwi, penguin, kakapo. Introduction Birds occupy a wide range of habitats worldwide and exhibit a great diversity of social systems and breeding patterns. The timing of breeding in each species is related to seasonal changes in the local environment, and to the relative lengths of incubation and parental care. However, a common feature of reproduction in almost all species is that breeding is seasonal. Marshall (1936) suggested that birds, and other vertebrates, possess an internal rhythm of reproduction which is synchronized with the environment by external factors, particularly the annual cycle of daylength. External factors could also strongly modulate the progression of the annual cycle. The timing of breeding in birds has since been reviewed (see Lofts and Murton 1968; Murton and Westwood 1977; Follett 1984) and a unifying scheme for the photoperiodic control of avian breeding cycles has been developed (Nicholls et al. 1988). In this review, a model for the timing of seasonal breeding in birds is presented which incorporates the photoperiodic concepts of Nicholls et al. (1988) and the proximate factors which control reproduction described by Wingfield (1980). This model is also applied to New Zealand birds. The New Zealand avifauna includes many unique species such as the kiwis (Apteryx sp.) and the kakapo (Strigops habroptilus). Since the breeding seasons of New Zealand birds had not previously been reviewed, it was not clear whether New Zealand birds show unique features in their patterns of seasonal breeding. The breeding seasons of native *Presented as a lecture at the 25th Annual Conference of the Australian Society for Reproductive Biology, Dunedin, New Zealand 1993. 1031-3613195/010001$10.00

2 J. F. Cockrem New Zealand birds are, therefore, summarized here and these are compared with the breeding seasons of introduced birds in New Zealand and with those of birds from similar latitudes in the Southern and Northern hemispheres. The physiology of the reproductive cycles of three New Zealand birds is described demonstrating that New Zealand birds are similar to other birds in the timing of their breeding seasons. Physiological Control of Seasonal Breeding Ultimate and Proximate Factors Breeding occurs throughout the year in very few species of birds. Instead, the discrete period of reproduction in most species suggests that environmental conditions are appropriate for breeding at only one time of year. This was recognized by Baker (1938a), who suggested that seasonal breeding is controlled by two types of environmental information called ultimate and proximate factors. Ultimate factors select individuals that produce young at the optimum time for survival. The most important ultimate factor for birds is the availability of food for feeding young and for postfledging survival. The timing of breeding in each species of bird probably evolved so that the hatching of eggs coincides with the maximum availability of food for the young birds. Ultimate factors do not, however, regulate the precise timing of breeding each year. The annual reproductive cycle must start well in advance of the time of hatching, and other factors called proximate factors regulate gonadal development and the temporal progression of the breeding cycle. Baker (1938b) suggested that temperature and daylength might be the main proximate cues for the timing of breeding seasons in boreal and temperate zones, and rain and/or intensity of illumination within the tropics. Baker (1938b) also noted that an internal rhythm could be important in determining the onset of the breeding season well before external proximate cues which stimulate reproduction become effective. These ideas remain central to our understanding of the timing of breeding in birds. Proximate factors are used directly by birds to regulate the timing of breeding. Wingfield (1980) subdivided these factors into four types of information: initial predictive; essential supplementary; synchronizing and integrating; and modifying. The primary source of initial predictive information for many birds is the annual cycle of daylength (Follett 1984). This is absolutely constant from year to year and provides a stable reference point for the entrainment of reproductive rhythms. Almost all of the birds that have been studied experimentally have been shown to be photoperiodic; i.e., the endocrine pathways controlling reproduction can respond to changes in daylength. Other factors providing important initial predictive information for some birds include endogenous circannual rhythms of reproductive function and rainfall. Initial predictive information brings birds into a physiological state in which breeding can begin but usually does not induce the final stages of the nesting phase and egg-laying. Complete gonadal development, mating and egg-laying are stimulated by supplementary information, such as social cues, possession of a territory, and availability of particular foods. These factors obviously vary from year to year and between locations and determine the exact timing of breeding each year. However, the scheme whereby initial predictive information stimulates gonadal growth and supplementary information stimulates egglaying does not fully explain the progression of avian annual cycles. Supplementary information may activate the hypothalamo-pituitary-gonad axis of photosensitive birds before initial predictive information has switched-on the reproductive system. For example, in the kestrel (Falco tinnunculus) the presence of the mate can enhance luteinizing hormone (LH) secretion in winter, well before the spring breeding season (Meijer and Schwabl 1989). Once nesting activity begins, a complex series of events (nest building, copulation, oviposition, incubation, feeding of young) must occur in the correct order. The external and internal factors which regulate the organization of these events are termed synchronizing and integrating information. They include both internal physiological processes and behavioural interactions between individuals within a pair. Proximate factors that can disrupt the reproductive cycle are termed modifying information and include adverse weather, loss of nest site, predation and the loss of a mate. Photoperiodism The effects of daylength on reproduction have been studied in greater detail than those of any other factors controlling seasonal breeding in birds. This is largely because of the relative ease with which daylength can be manipulated under controlled conditions. Indeed, avian studies have provided fundamental concepts that underlie knowledge of photoperiodism in animals in general (see Nicholls et al. 1988; Konishi et al. 1989). Photoperiodic responses have been demonstrated in about 70 species of bird (Wingfield and Farner 1993), and the physiology of photoperiodism in birds has been regularly reviewed (Lofts and Murton 1968; Murton and Westwood 1977; Farner and Gwinner 1980; Follett and Robinson 1980; Follett 1984; Nicholls et al. 1988). The most obvious photoperiodic response in birds is the dramatic stimulation of gonadotrophin secretion and gonadal growth in some species by long daylengths. For example, Japanese quail (Coturnix coturnix japonica) transferred from a short photoperiod (8 h light, 16 h dark; 8L:16D) to a long photoperiod (20L:4D) show a large increase in plasma gonadotrophin concentrations during

Seasonal Breeding in New Zealand Birds 3 the first week of photostimulation (Follett and Robinson 1980), and mallards (Anus platyrhynchos) can lay eggs within three weeks of transfer from short to long days (Cockrem, unpublished observations). The common pattern of breeding for birds in mid latitudes is for egg-laying to start in spring. This could simply be explained by suggesting that reproduction in these birds is switched-on by increasing daylengths after the winter solstice. However, the peak of egg-laying in both Southern and Northern hemisphere birds is not centred on the longest day, nor is it symmetrical (Fig. 1). Instead, the peak of egg-laying occurs one month before the longest day. Photoperiodic birds that are switched-on in spring but stop breeding when days are still long must have lost the ability to respond to the long days. This loss of response is termed photorefractoriness, and most of the birds that have been studied experimentaliy undergo a form of photorefractoriness during their annual cycle. The 'standard' annual cycle for birds breeding in the spring is that they become photosensitive in the autumn, are photostimulated by increasing daylengths in spring, lay eggs, and then lose their response to long days and become photorefractory by mid summer. Photorefractoriness is gradually lost and the birds become photosensitive again. However, this scheme does not accommodate the many birds that do not start breeding in spring. The annual cycles of such birds cannot be explained by invoking a simple switch-on of the reproductive system by long days. The physiological basis of photoperiodism in these other birds remained largely unexplained until the introduction of important new concepts in a review of photorefractoriness by Nicholls et al. (1988). These authors defined several physiological processes which they suggested can account for the events of photoperiodically-controlled annual cycles in all seasonally-breeding vertebrates, regardless of the actual time of breeding by each species. Nicholls et al. (1988) defined photorefractoriness as a tendency to terminate reproduction which is induced by long days. It is important to note that photorefractoriness is started by the same long days that cause photostimulation; i.e., long days both induce gonadal growth and lead to the switching-off of reproduction. Photorefractoriness develops in the brain at the level of the hypothalamus or above, and it takes at least several weeks to develop. This phenomenon varies in intensity between species, ranging from a decreased probability of breeding to a complete shutdown of the hypothalamo-pituitary-gonadal axis. Photorefractoriness can be so weak that it is merely a readiness to terminate reproduction that is manifest when daylengths are subsequently reduced (e.g. Japanese quail). In many non-passerine birds it can be a partial reduction in intrinsic neural activity in the hypothalamic centres that control gonadotrophin-releasing hormone (GnRH) release. GnRH release is normally further reduced by steroid feedback. Photorefractoriness can also be so strong that there is a complete loss of neural activity controlling GnRH even in the absence of steroids, as in most temperate-zone passerines such as the starling (Sturnus vulgaris). Long days initiate photorefractoriness as well as stimulating gonadal development, and the threshold daylength for photorefractoriness is longer than that for photostimulation. For example, in the starling a photoperiod of 11L:13D will induce slow gonadal growth. The gonads then remain large and photorefractoriness does not develop (Falk and Gwinner 1988). Conversely, a photoperiod of 13L:llD induces gonadal development followed by the regression that is associated with photorefractoriness. The rate of development of photorefractoriness is directly related to the ambient daylength once the threshold daylength for the induction of photorefractoriness is exceeded (Nicholls et al. 1988). The threshold daylength for the development of refractoriness is particularly important for species that breed at high latitudes where conditions are suitable for breeding for only a short time each year. These species are likely to have a higher threshold than birds at lower latitudes and are likely to experience a relatively rapid onset of refractoriness once this threshold is exceeded. This mechanism provides for a short period of egg-laying which is appropriate to these conditions. The most dramatic form of photorefractoriness is seen in birds such as the rook (Corvus frugilegus) (Lincoln et a!. 1980) and the starling (Dawson and Goldsmith 1982) which start breeding in spring and then finish egg-laying before the summer solstice when daylengths are still increasing. Experimental studies on some of these birds have shown that the reproductive system cannot be switched-on once gonadal regression is complete, even when very long photoperiods are used. These birds are said to be absolutely photorefractory (Nicholls et al. 1988). There is also another form of photorefractoriness known as relative refractoriness. In Japanese quail held outdoors at 52'N, rapid gonadal development starts in spring when daylengths reach about 11.5 h. Gonadal regression, however, occurs in summer when daylengths fall below 14.5 h (Nicholls et al. 1988). At this time, the birds are clearly photorefractory to the stimulatory effects of the long summer daylengths. However, the birds are relatively photorefractory rather than absolutely photorefractory because if moved to a very long daylength (20 h), full gonadal growth occurs again. Photorefractoriness in birds is followed by a gradual recovery of photosensitivity. Photosensitivity is the ability to undergo gonadal growth in response to an increase in daylength and, in many species, photosensitivity will not develop unless the birds experience relatively short

J. E Cockrem 1 I I I I I I I I A M J J A S O N D J F M Month Fig. 1. Annual distribution of egg records for (a) New Zealand birds and for other birds of similar latitudes in the (b) Southern hemisphere and (c) Northern hemisphere. Records of egg-laying for New Zealand birds breeding on the mainland and offshore islands were obtained for 81 native species (76% of all breeding species) and for 25 introduced species (68% of all species) (Anon. 1985). The total number of records for each month was calculated and expressed as a percentage of all the records for native species (-) and for introduced species (---). Data for other birds of similar latitudes were calculated from Baker (1938b), with data for Northern hemisphere birds plotted on a Southern hemisphere time scale. The total number of records at each latitude for each month was calculated and expressed as a percentage of all the records at that latitude: (b) 3030 S (-) and 40-50"s (---);(c) 30-40"N (-) and 40-50"N (---).

Seasonal Breeding in New Zealand Birds 5 daylengths (Nicholls et al. 1988). In natural conditions, most species, therefore, become photosensitive again in autumn. The recovery of photosensitivity also proceeds in some species regardless of daylength, e.g. mallard (Lofts and Coombs 1965). Nicholls et al. (1988) made the important point that the action of short days to end photorefractoriness is a positive one which switches on the reproductive axis. This switch-on is clear in species such as the rook that exhibit autumn sexuality (a resurgence of reproductive activity which usually does not lead to breeding). In many other species these changes are more subtle and can only be detected by appropriate experiments. Nicholls et al. (1988) suggest that the loss of photorefractoriness and the regaining of photosensitivity is a very significant time in the control of annual reproductive cycles. These changes occur under short daylengths regardless of whether an animal actually breeds in autumn (a short day species) or in spring (a long day species) - in all cases the reproductive system is switched-on under short days. Ball (1993) suggested that changes in hypothalamic 'drive' independent of steroid feedback are the fundamental factors regulating the cycle of photosensitivity and photorefractoriness. The importance of steroid feedback varies between species and the timing of breeding is determined by the rate at which photosensitivity is regained, by the sensitivity of the hypothalamus to inhibitory feedback from gonadal steroids, by the timing of an increase in hypothalamic 'drive', and by the need for photostimulation. If steroid feedback is low and photostimulation by long daylengths is not needed, breeding can occur in autumn or winter, shortly after photosensitivity is regained. Conversely, if steroid feedback is high and there is a dependence on photostimulation, breeding will not occur until spring or summer. Gonadal steroids inhibit the activation of the hypothalamo-pituitary-gonadal axis and oppose the stimulatory effects of long daylengths (known as photoperiodic drive). Nicholls et al. (1988) suggest that sensitivity to steroidal feedback gradually increases under short daylengths. Low plasma concentrations of gonadal steroids could then account for the winter suppression of gonadotrophin secretion in species that breed in the spring. Whether or not species which breed in spring show a marked increase in gonadotrophin secretion in response to castration depends on the requirement of the species for photostimulation. For example, photosensitive whitecrowned sparrows (Zonotrichia leucophrys gambelii) show little response when castrated on short days (Wingfield et al. 1980), whereas starlings treated similarly exhibit a rapid and large increase in concentration of gonadotrophins (Goldsmith and Nicholls 1984). Species that have a low sensitivity to steroid feedback will undergo gonadal growth if hypothalamic 'drive' increases once they become photosensitive. In other species in which feedback sensitivity is moderate to high, reproduction will be delayed until steroid feedback is overcome, either by increasing photoperiodic drive or when the gradual increase in GnRH release (hypothalamic 'drive') is sufficient to stimulate gonadal development. Winter breeding usually does not occur in species that start breeding in spring, either because steroid feedback is too high or because there is insufficient hypothalamic 'drive'. These birds have been photosensitive since autumn, and gonadal maturation would eventually occur under short daylengths (Nicholls et al. 1988). Instead, the birds are photostimulated and the reproductive system is switched-on as daylengths increase in spring. Nicholls et al. (1988) call this a premature switch-on, since it occurs in advance of the eventual gonadal maturation which would occur independently of photoperiod. The stimulatory effects of long daylengths are accompanied by the initiation of photorefractoriness, so the length of the breeding season depends on the relative rates at which photostimulation and photorefractoriness develop (Nicholls et al. 1988). Model for the Physiological Control of Seasonal Breeding A model explaining the timing of seasonal breeding in birds is presented below. This model incorporates the photoperiodic concepts (photosensitivity, photorefractoriness and steroid feedback sensitivity) of Nicholls et al. (1988) as well as the subdivision of proximate factors into initial predictive and supplementary information described by Wingfield (1980). A comparison of two seabirds within the same genus provides a good example of why a single model should be applied to all birds. The Westland petrel (Procellaria westlandica) and the black petrel (I? parkinsoni) are medium-sized petrels that breed on, or close to, the New Zealand mainland. They both have dark plumage and are difficult to distinguish from each other except for the larger size of the Westland petrel. However, despite their similarities they breed at exactly opposite times of the year (Table 1) - the Westland petrel breeds in late autumn and the black petrel breeds in late spring. It seems most unlikely that such closely related birds could have fundamentally different mechanisms for timing their breeding seasons. Instead, the present model assumes that all birds time their breeding seasons using similar physiological mechanisms and has the following four main points. (I) The timing of breeding in birds is determined by interactions between an internal rhythm of reproduction and external information from the environment, especially daylength. (2) Avian breeding seasons effectively start in autumn, regardless of when egg-laying occurs.

J. F. Cockrem Table 1. Breeding season specifics of the Westland petrel (Procellaria westlandica) and the black petrel (Procellaria parkinsonr] which are endemic to New Zealand Data from Marchant and Higgins (1990) Westland petrel Black petrel Weight (kg) 1.2 0.7 Breeding location Westland Hauraki Gulf Return to colony mid February-late March October Egg-lay ing mid Mayaarly June November-early January Leave colony November-December May-July (3) The annual reproductive cycle can be divided into two phases - one from the start of the annual cycle to egg-laying, the other from egg-laying to the start of the cycle. (4) Variation in the timing and length of avian breeding seasons results from variation in: the relative importance of initial predictive and supplementary information; the duration and type of photorefractoriness; the rate of recovery of photosensitivity; the photoperiodic threshold for photostimulation; the strength of negative steroid feedback; and the latitudes at which birds live. This model presumes that most, if not all, birds are photoperiodic (capable of responding to changes in daylength), regardless or whether or not they actually use changes in daylength as a timing cue. Birds living in equatorial regions are the least likely to be photoperiodic, since their annual cycle of daylength is of very low amplitude. However, several studies have shown that various tropical birds are photoperiodic (e.g. Gwinner and Dittami 1985), and even the domestic chicken responds to daylength (Sharp 1984). Photoperiodic birds are, by definition, photosensitive for part of the year, and almost all of the birds that have been studied also undergo photorefractoriness during their annual cycle. The only reported exceptions are some pigeons in the genus Columba with breeding seasons that tend to be symmetrical about the summer solstice (Murton and Westwood 1977). Yet, even these birds might become photorefractory if subjected to experimental schedules based on current knowledge of photorefractoriness. Interactions Between an Internal Rhythm of Reproduction and External Factors Marshall (1936) suggested that seasonal breeding in vertebrates resulted from an internal rhythm of reproduction which was synchronized with seasonal changes in the environment by external factors. Marshall (1936) described the internal rhythm as an alternation of periods of rest and activity, and noted that this rhythm could be greatly modified by 'exteroceptive' factors. Rowan (1926) and Bissonette (1930) showed that changes in photoperiod could affect the development of the reproductive system in birds, and Marshall (1936) identified light as the principal environmental factor that regulated the progression of vertebrate annual cycles. The concept of an inherent reproductive cycle synchronized and modified by 'exteroceptive' factors was used in a review of avian breeding seasons by Baker (1938b) and applied by Roberts (1940) in a study of the gentoo penguin (Pygoscelis papua). Roberts (1940) performed an elegant analysis of factors in the physical and psychic environments that could promote or delay the breeding cycle and remains relevant. The basic relationships between daylength and an internal rhythm were subsequently investigated experimentally in studies of the house sparrow (Passer domesticus) and the starling (see Bartholomew 1949; Burger 1953). The internal rhythm of reproduction could, in part, result from a systematic variation in the response of the reproductive axis to daylength. Starlings are reported to exhibit a circannual rhythm of testicular growth and regression under a 12L:12D photoperiod (see Gwinner 1986). Gwinner and Wozniak (1982) suggested that this rhythm resulted from the neuroendocrine system responding differently to the same photoperiod at different phases of a circannual cycle. However, Dawson and McNaughton (1993) found no evidence for a circannual rhythm of photosensitivity and photorefractoriness in castrated starlings under 12L:12D. Intact birds may undergo species-specific periods of photosensitivity and photorefractoriness which are properties of the central nervous system but are not driven by a circannual clock. Alternating periods of photosensitivity and photorefractoriness could constitute the endogenous 'programme' which Wingfield (1993) suggested is possessed by song sparrows (Melospiza melodia rnelodia). This 'programme' would usually be entrained to one year by external information from the environment, particularly the annual cycle of daylength. The current model provides a physiological basis for the internal rhythm of reproduction in birds originally proposed by Marshall (1936). The Start of the Annual Reproductive Cycle The change from photorefractoriness to photosensitivity in autumn occurs when the reproductive system is switchedon each year (Nicholls et al. 1988), even though this switch-on may not be apparent until much later. Mid autumn is also the time when few birds start their annual

...-... Seasonal Breeding in New Zealand Birds Start of annual cycle r'/ No photorefractoriness / Steroid feedback high t Non-photoperiodic information Photosensitive (photorefractoriness ended) A - ------- Steroid feedback l No gonadal growth / 1 Gonadal growth to functional size Breeding on decreasing or short daylengths I Photoperiodic stimulation of gonads Photoperiodic stimulation of gonads... 0.---..- of daylength......"./...... -....... Fig. 2. Model for the progression of avian breeding seasons from the start of the annual cycle to the time of egg-laying. The annual cycle is considered to start in autumn, whereas egg-laying -. may start at any time from late autumn until late summer in the following year (see text for details). period of egg-laying. The breeding seasons of birds, therefore, effectively start in autumn, regardless of when overt changes begin in the hypothalamo-pituitary-gonadal axis. The Two Phases of the Annual Reproductive Cycle The progression of the annual reproductive cycle of birds can be conveniently divided into two phases (see Figs 2 and 3). The first phase (Fig. 2) extends from the start of the annual cycle until egg-laying. In birds of tropical regions, non-photoperiodic information may be the primary source of initial predictive information, and breeding may occur independently of daylength (pathway on the left in Fig. 2). The breeding seasons of tropical species are more varied than those of other birds (Immelman 1971), but nevertheless individual species tend to have discrete breeding seasons related to the annual pattern of rainfall (see Vernon 1978; Wilkinson 1983; Dittami and Gwinner 1985; Dittami 1986). In some species, the period of the reproductive cycle may be determined primarily by the internal rhythm of reproduction (e.g. the nine-month breeding cycle of the sooty tern, Sterna fuscata; Ashmole 1963). In the pathway for tropical species, as in all the other pathways in the model, supplementary information, such as social cues and the availability of particular foods, supports the final stages of ovarian development that lead to egglaying. Supplementary information is usually essential for egg-laying, although the relative importance and types of such information vary greatly between individuals and species. Birds of lower latitudes may not become photorefractory under natural conditions because daylengths never exceed the threshold necessary to induce photorefractoriness. In these species, and some pigeons, photoperiodic stimulation of the gonadal axis will potentially lead to egg-laying whenever the natural daylength exceeds the speciesspecific threshold for photostimulation. If this threshold is relatively low, the availability of appropriate supplementary information may be more important than daylength in regulating the timing of breeding, especially at low latitudes where the photoperiod may always be stimulatory. The third pathway leading from the start of the annual cycle applies to most of the birds of mid latitudes and

Egg-laying I \ No photorefractoriness Absolute photorefractoriness Relative photorefractoriness J. E Cockrem 1 Breeding ends when daylength no longer stimulatory (or when supplementary information absent) 1 1 I Changing photoperiodic threshold Breeding ends Breeding I ends Photosensitivity regained 1 Photosensitivity regained (short days may be needed) Start of annual cycle Fig. 3. Model for the progression of avian breeding seasons from the time of egg-laying to the start of the annual cycle (see text for details). 1 northern latitudes. These birds become photosensitive in autumn and the timing of breeding is then influenced by the sensitivity of the hypothalamus and pituitary gland to inhibitory feedback from gonadal steroids and by the rate of increase of steroid-independent hypothalamic 'drive'. If sensitivity is high, gonadal growth will not occur until the threshold daylength for photostimulation is exceeded after the winter solstice. Photoperiodic drive then overcomes the steroid feedback and the reproductive system is switched-on. This pathway applies to most birds that start egg-laying in late winter, spring or summer. The actual time of egg-laying in these birds may be strongly influenced by supplementary information. The alternative pathway from the end of photorefractoriness applies to birds in which the sensitivity to steroid feedback is relatively low. If hypothalamic 'drive' steadily increases as birds become photosensitive, gonadal growth may be sufficiently rapid that breeding occurs in late autumn or winter on decreasing or short daylengths. Autumn daylengths may also act directly to photostimulate the reproductive system to varying degrees (e.g. gonadal growth in the house sparrow can be stimulated by only 10 h of light, daily; Bartholomew 1949). In other species, hypothalamic 'drive' and gonadal growth increase more slowly and are overtaken by increasing photoperiodic drive in spring or summer (pathway on far right in Fig. 2). Species in which gonadal growth occurs in autumn include the white-crowned sparrow (Farner et al. 1966), the house sparrow (Hegner and Wingfield 1986), the grey partridge (Perdix perdix) (Sharp et al. 1986) and both the great tit (Parus major) and the willow tit (P montanus) (Silverin et al. 1989). Many birds that breed in spring have a resurgence of sexual behaviour in autumn which may well reflect the onset of increasing hypothalamic 'drive' and gonadal growth. Indeed, autumn sexuality can occasionally progress to egg-laying (e.g. house sparrow, Summers-Smith 1963). The first phase of the annual reproductive cycle culminates in egg-laying. Some birds lay only a single clutch of eggs, whereas others may lay up to five clutches spread over many months. Egg-laying leads into the second phase of the annual cycle which extends to the start of the next cycle (Fig. 3). Birds that do not become photorefractory will stop egg-laying when the daylength is no longer stimulatory (pathway on the left in Fig. 3). In addition, in species of lower latitudes that depend on

Seasonal Breeding in New Zealand Birds 9 supplementary information, breeding may finish because appropriate information is no longer available. Egglaying finishes in most other birds because they become photorefractory (middle and right pathways in Fig. 3). Birds that start breeding in spring and finish egg-laying before the summer solstice are likely to show absolute photorefractoriness (gonadal regression occurs and the reproductive axis cannot be switched-on). These species generally need to be exposed to shorter daylengths for photorefractoriness to be lost. Photosensitivity will be regained as days shorten in autumn, the reproductive system is switched-on and the annual cycle begins again. Birds that continue egg-laying through summer, or that do not start egg-laying until after the summer solstice, probably become relatively photorefractory rather than absolutely photorefractory (i.e., the gonads regress but gonadal growth can still be stimulated by exposure to very long days). These birds are photostimulated when daylength reaches the species-specific photoperiodic threshold (also known as critical daylength), but lose their response to long days when the daylength is still above the original threshold. This can be viewed as an increase in the photoperiodic threshold (Phillips et al. 1985), so that as daylength decreases in autumn it falls below the new, higher threshold and breeding stops. As photorefractoriness is gradually lost following the breeding season, this threshold effectively decreases so that when photosensitivity is regained the photoperiodic threshold has returned to its previous level. In these birds, as for those that become absolutely photorefractory, the regaining of photosensitivity is accompanied by the switching-on of the reproductive axis and the resumption of the annual cycle. The model outlined in this review can be applied in a speculative scheme to the Westland petrel and the black petrel (see Table 1). These petrels, like many others, migrate to and from the breeding colonies and have precisely-timed breeding seasons. Another petrel that breeds in Australasia, the short-tailed shearwater (PufJinus tenuirostris), travels from breeding colonies in Australia to the northern Pacific Ocean, yet arrives back at the breeding colonies within just 11 days each year. Marshall and Serventy (1959) suggested that the timing of the annual cycle of migration and reproduction in this species was regulated by a circannual rhythm. There is experimental evidence that circannual rhythms influence both the direction and distance of migration in some birds that migrate from Europe to Africa (Gwinner 1986), and circannual rhythms might contribute to the annual cycle of migration in the Westland petrel and the black petrel. It is postulated that the annual reproductive cycle of both petrels starts in autumn when the birds become photosensitive, and that the primary difference between the two species is in the strength of negative steroid feedback and in the rate of increase in hypothalamic 'drive' once birds become photosensitive. Under this model, steroid feedback in the Westland petrel is low. The egg-laying season is short and precisely timed, and supplementary information must be of relatively little importance for the timing of breeding. Gonadal growth occurs in autumn once the birds are photosensitive, and the Westland petrel might also be directly photostimulated. Male house sparrows held under a 10-h photoperiod can reach full breeding condition after 46 days (Bartholomew 1949), and daylength on the breeding grounds of the Westland petrel is 13.6 h in mid February when the birds start arriving. If the threshold for photostimulation is 10 h or 11 h in the Westland petrel, photostimulation by autumn daylengths could contribute to gonadal growth. Most importantly, these daylengths could also initiate processes leading to photorefractoriness and the regression of the reproductive system in June after egg-laying. Photorefractoriness would then be prolonged, maintaining the reproductive system in a completely inactive state until photosensitivity is regained the following autumn. It is suggested that in the black petrel, there is high steroid feedback so that gonadal growth is delayed until late spring and egg-laying does not occur until early summer. These birds become absolutely photorefractory once egg-laying is complete, photosensitivity is regained 2-3 months later in autumn and the annual cycle is complete. Egg-laying is less tightly synchronized than that in the Westland petrel and extends over two months. Supplementary information is, therefore, likely to be more important in determining the timing of egg-laying in the black petrel than in the Westland petrel. Variation in Breeding Seasons The timing and length of breeding varies between individuals of the same species, between breeding locations for a single species, from year to year for one species at the same location, and between species. In the third part of the model, it is suggested that the diverse breeding seasons of birds can be accommodated within the scheme outlined in Figs 2 and 3 by considering variation in the parameters of the model. The timing of the breeding season can be very precise and may vary little from year to year for a species at one location (e.g. Westland petrel). Conversely, individuals within a population may start egg-laying over a prolonged period each year (e.g. North Island brown kiwi, Apteryx australis; Potter 1989). In the case of opportunistic breeders, Immelman (1971) suggested that they can breed at any time of the year although recent data indicate that such birds are more likely to breed at certain times of the year (Hahn 1993). The range in breeding patterns can be partially explained by the relative importance of initial predictive and supplementary information. Birds that

10 J. E Cockrem have very regular breeding seasons must rely primarily on initial predictive information (generally daylength) and will respond little to local environmental factors. Other birds that have more variable breeding seasons must be much more strongly influenced by supplementary information from the environment, such as local variations in weather, food availability and nesting habitat. This concept has been developed by Wingfield et al. (1992) into a model which takes into account the relative importance of the two types of information as a function of seasonality in the environment and variability between years in environmental conditions. Many birds will delay breeding or not breed at all if the food supply is inadequate. For example, Hirons et al. (1984) found that gonadal development in the tawny owl (Strix aluco) was correlated with bodyweight and food supply. The proportion of pairs that attempted to breed each year varied from 0% to 85% (Hirons 1985), indicating a very strong role for supplementary information in this species. In another species, the great tit, a five-year study showed striking differences between years in the amplitude of seasonal changes in plasma androgen concentrations (De Laet et al. 1985). The maximum concentrations during the non-breeding season in one year exceeded the maximum concentrations in the breeding season in another year. These variations were suggested to be related to environmental factors (i.e. supplementary information), including population density and food availability. Progressive changes during the annual cycle from photosensitivity to photorefractoriness are an important feature of the model. Some birds do not become photorefractory at all, either because they are apparently incapable of becoming refractory or because the natural daylength is never long enough to initiate photorefractoriness. The daylength at which photorefractoriness develops varies between species, as does the need for shorter daylengths to end photorefractoriness. The time when photorefractoriness develops within a species can also be affected by supplementary information, such as social stimulation and temperature. In the white-crowned sparrow, the duration of maximum testis size before photorefractoriness and gonadal regression starts can vary from 10 days in birds exposed to artificially-induced long days to 44 days in free-living birds that renest (Wingfield and Farner 1993) and, in the Japanese quail, low ambient temperatures can accelerate gonadal regression (Wada et al. 1990). The strength of negative steroid feedback influences the timing of gonadal growth in birds after they become photosensitive in the autumn. This varies both between and within species. In Britain there is a local, non-migratory race of British starlings and a migratory race of Continental starlings. Bullough (1942) reported that gonadal growth in the British starlings started 'precociously' in autumn, whereas in the Continental birds it was delayed until late winter. Spontaneous gonadal growth under short (8 h) daylengths has since been shown experimentally in the British starlings (Nicholls et al. 1988), and steroid feedback sensitivity must be much greater in the Continental race than in the British race of starlings. There can be marked variation within local populations as well, as shown by the occurrence of egg-laying in a female mallard transferred indoors in early autumn to short days (8L: 16D; Cockrem, unpublished observations). This bird laid a full clutch of eggs under short days, suggesting it had a much lower feedback sensitivity than other mallards. The existence of such variation provides an explanation for the reports of occasional egg-laying by free-living mallards in early winter in New Zealand, before the winter solstice and 2-3 months in advance of the usual start of egg-laying. Finally, the photoperiodic threshold for photostimulation (critical daylength) is an important parameter which affects the timing of breeding, especially as most birds start breeding on increasing daylengths between the winter and summer solstices. The start of egg-laying and the photoperiodic threshold both tend to increase with latitude. For example, the threshold for egg-laying in the black swan (Cygnus atratus) which breeds at 28"s in Australia is about 10 h whereas in the Bewick's swan (C. columbianus) which breeds at 65"N the threshold is about 16 h (Murton and Westwood 1977). Latitude also affects the timing of breeding within species, since the amplitude of the annual cycle of daylength increases with latitude. For example, in the house sparrow a daylength of 10 h is photostimulatory (Bartholomew 1949), so at latitudes where daylength never falls below 10 h (0-30") the timing of gonadal growth in species with a low threshold for photostimulation may be determined largely by an internal rhythm of photorefractoriness and photosensitivity. In addition, the importance of supplementary information for the timing of breeding is inversely related to latitude. The threshold for photostimulation can also vary within a species, as shown by Silverin et al. (1993) in the great tit in Europe. The great tit is distributed over a wide latitude range and egg-laying starts later in the more northern populations. This pattern was reflected in differences between the populations in the threshold daylength for the induction of an increase in the concentration of LH and for the induction of rapid testicular growth (Table 2). Furthermore, the population in Norway is suggested to have been established for 4 0 years, indicating that selection for different photoperiodic thresholds could account for the differences between populations in egg-laying times. In another example, Donham (1979) showed that mallards held on a game farm for 20 generations laid eggs at least one month

Seasonal Breeding in New Zealand Birds Table 2. Photoperiodism in the great tit Parus major in Europe Adapted from Silverin et al. (1993) Location Latitude Date of first egg Photoperiodic threshold (h) for: (ON) increase in LH rapid testicular growth Italy 45 22 March 10 >11 Sweden 58 26 April 11 11-12 Norway 70 3 May 11.5 >I2 earlier than wild mallards and had large testes for a longer period of time. Selection for egg-laying in the game-farm mallards may have advanced the onset of egg-laying by reducing the threshold daylength for photostimulation and by shortening the period of photorefractoriness. In addition, there can be considerable variation within a population, as shown by differences between colour morphs of the feral pigeon (Columba livia) in the timing and extent of gonadal growth and regression each year (Murton et al. 1973). Hence, it is clear that the timing of breeding in a particular species should not be viewed as a fixed characteristic. Instead, it may vary between birds and between years, and it may also change relatively rapidly if local environmental conditions are altered. Breeding Seasons of New Zealand Birds The New Zealand Avifauna A listing of the birds of the New Zealand region (mainland, offshore and sub-antarctic islands, Ross Dependency area of Antarctica) gives a total of 305 species (Anon. 1985). This total contains 268 species of native birds and 37 species of birds that have been introduced to New Zealand by humans over the past 150 years. Of the native birds, 92 species (34%) are endemic (i.e. found only in New Zealand). However, the level of endemism is much higher when breeding species are considered. There are 138 species of native birds breeding in New Zealand and on close offshore islands, and 92 (67%) of these are endemic. This high level of endemism and a relatively high incidence of flightlessness and of melanism (Baker 1991) are distinctive features of the New Zealand avifauna. Egg-laying Seasons of New Zealand Birds Avian breeding seasons include periods of pair formation, nest building, egg-laying, incubation and rearing of the chicks. The breeding season of an avian species is often taken as the period from egg-laying until the end of parental care. An alternative is the egg-season (the months of the year when living eggs are found). However, Baker (1938b) recognized that the best measure of the breeding season is the egg-laying season, defined as the months of the year when eggs are laid, and this measure is used below. The egg-laying seasons of nine species of native New Zealand birds are shown in Fig. 4. and illustrates two points. Firstly, the timing of the start of egg-laying varies markedly between species. Egg-laying can start in June (in early winter) in the case of the North Island brown kiwi, or it can start as late as January (in mid summer) in the kakapo. Secondly, the duration of egg-laying varies from just over one month (yellow-eyed penguin, Megadyptes antipodes) to eight months (North Island brown kiwi). The pattern of egg-laying of all New Zealand birds is shown in Fig. 1. Although there have been numerous studies of individual species, this review presents the first summary of the breeding seasons of New Zealand birds. Clearly, native New Zealand birds are seasonal breeders. There is a broad peak of egg-laying from September to December with a maximum in November, although some eggs are laid in every month. Interestingly, the pattern of egg-laying for introduced birds is exactly the same as that for native birds (X2 = 5.07; d.f., 11; P = 0.928), even though the proportions of introduced birds and native birds in different taxonomic groups are not the same. This striking result indicates that the pattern of egg-laying of the native species is not unique; rather, it appears to be a function of the New Zealand environment since the same pattern is evident in birds that have been brought to New Zealand from other countries. This conclusion is supported by comparing the clutch sizes of introduced European passerines in New Zealand and Europe. Niethammer (1970) found that clutch sizes are smaller in New Zealand and noted that the breeding season of blackbirds (Turdus merula) was longer in New Zealand. Reproductive parameters such as clutch size have not previously been considered for the native New Zealand avifauna as a whole. However, Woinarski (1985) compared clutch sizes and the length of the breeding seasons of small, insectivorous Australian birds with these parameters in similar European and North American birds. Australian birds had smaller clutch sizes and longer breeding seasons than Northern hemisphere birds. In another analysis, Wyndham (1986) used a derived measure (the number of equally good months for breeding) to compare Australian birds with birds from other regions of the world (excluding New Zealand).

J. F. Cockrem North Island brown kiwi Grey-faced petrel Great-spotted kiwi t Little blue penguin t Fantail Little-spotted kiwi Yellow-eyed penguin t I t Weka Kakapo t A M J J A S O N D J F M Month Fig. 4. Egg-laying periods (starting in April in autumn) of nine New Zealand birds. For each species, the period of laying is represented by a bar. Data on egg-laying for individual species were obtained as follows: North Island brown kiwi, grey-faced petrel, yellow-eyed penguin (Marchant and Higgins 1990); great-spotted kiwi (J. A. McLennan, personal communication); little blue penguin (Kinsky 1960); fantail (Blackbum 1965); little-spotted kiwi (Jolly 1989); weka (Beauchamp 1987); and kakapo (Powlesland et al. 1992). Wyndham (1986) found that the breeding seasons of Australian and African birds were, on average, two months longer than those of birds in various regions of the Northern hemisphere. The data from Australia and New Zealand together are consistent with the concept that the breeding seasons of birds in the Australasian region are dependent on physical and ecological characteristics of the region rather than being unique to the native species of the region. The pattern of egg-laying in New Zealand birds can be compared with data from Baker (19386) for other birds of similar latitudes. New Zealand covers a latitude range of 34-47"s and data for the latitude ranges of 30-40"s and for 40-50"s (Baker 19386) are shown in Fig. 1 (New Zealand birds were not included in the analysis). The pattern of egg-laying in New Zealand birds is very similar to that for other Southern hemisphere birds, with a peak of egg-laying occurring in the other birds from September to December and maxima occurring in November (30-40"s) and November-December (40-50"s). This reinforces the conclusion that, despite the high level of endemism, the seasonal pattern of egg-laying in New Zealand birds is not characteristic of these birds. Northern hemisphere birds have a peak of egg-laying in the same month (on a Southern hemisphere time scale) as New Zealand birds and Southern hemisphere birds of similar latitudes, and the peak of egg-laying in birds of mid latitudes is late spring in both hemispheres. In contrast, the pattern of egg-laying is not the same. At 30-40 S the greatest percentage of egg records in one month is 18.8%, whereas at 30-40"N the equivalent figure is 33.8%. There are four months in which the percentage of egg records exceeds 10% for native New Zealand birds as well as for other birds at 30-40 S, whereas this percentage is exceeded in only three months for birds at 30-40 N. Furthermore, there are no months in which no eggs are laid at this latitude in the Southern hemisphere whereas there are two months in which no eggs are laid in the Northern hemisphere. These results extend those of Wyndham (1986) and show that the breeding of birds of mid latitudes is seasonal in both hemispheres but spread over a greater portion of the year in the Southern hemisphere. The mid latitudes in the Southern hemisphere are largely ocean, whereas in the Northern hemisphere they include continental North America, Europe and central Asia. The annual extremes of

Seasonal Breeding in New Zealand Birds climate experienced by birds of these latitudes are likely to be less in the Southern hemisphere. The longer breeding seasons of the Southern hemisphere birds are, therefore, expected given that conditions are usually favourable for breeding over a longer period of time each year. Reproductive Endocrinology of New Zealand Birds Annual cycles of reproductive behaviour and egg-laying have been reported for all New Zealand birds, but an annual gonadal cycle has been reported for only one native species, the weka (Gallirallus australis) (Carroll 1963). However, there are recent descriptions of annual cycles of plasma hormone concentrations for three endemic species - the North Island brown kiwi, the yellow-eyed penguin and the kakapo. North Island Brown Kiwi The North Island brown kiwi is one of three currently recognized species of kiwi. Kiwis are flightless and nocturnal, and feed largely on invertebrates. Female kiwis have two functional ovaries (Kinsky 1971) and lay very large eggs in relation to their size. The eggs are incubated by the male alone for 70-90 days. Egg-laying in free-living North Island brown kiwis extends from June to February (McLennan 1988), with a peak occurring in September. Annual cycles of plasma concentrations of sex steroids were described in kiwis in Northland by Potter and Cockrem (1992). Egg-laying in these kiwis occurred from July to February, with a peak in September. There were clear annual cycles of plasma testosterone and oestradiol in male kiwis (Fig. 5). Steroid concentrations increased in April (oestradiol) and in May (testosterone), indicating that the reproductive system of these birds is switched-on in autumn, several months before the start of egg-laying. Data from studies of other avian species in which plasma testosterone concentrations and testis size have been measured together (see Wingfield and Farner 1993) indicate that the changes in testosterone concentrations in the kiwi probably reflect 10-100-fold changes in testis weight. The oestradiol concentrations are relatively high for male birds which might be related to the fact that the male kiwis alone incubate the eggs in this species. Female kiwis had low concentrations of testosterone and an annual cycle of oestradiol concentrations which was very similar to that in males (Fig. 6), indicating that the reproductive axis of females is also switched-on in autumn. Seasonal changes in ovarian weight are likely to be as marked as the postulated changes in testis size in the male. Yellow-eyed Penguin The yellow-eyed penguin is a rare penguin that breeds on the South Island and some sub-antarctic islands. Egg-laying occurs in September and October and is highly synchronized between birds. Indeed, in one year all eggs are usually laid within four weeks. Also, if eggs are lost, females do not lay replacement clutches. The incubation and chick-rearing periods are long, and young penguins do not fledge until February and March (Darby and Seddon 1990). J F M A M J J A S O N D Month Fig. 5. Plasma concentrations of testosterone, oestradiol and progesterone in male North Island brown kiwi in Northland. Values are meanrts.e. (n = 1-8); where n c 3, individual values are shown. Sample sizes are indicated across the top of figure. From Potter and Cockrem (1992). Plasma steroid concentrations were measured in yelloweyed penguins on the Otago Peninsula by Cockrem and Seddon (1994). Testosterone and oestradiol concentrations were elevated for only a short period of time before egg-laying in September (Fig. 7), indicating that there is a similarly short period when the gonads are large. These data were obtained from birds of unknown sex. However, other data from the same study showed that the peak concentration of testosterone occurred in male penguins and the peak concentration of oestradiol occurred in female penguins. A decrease in steroid concentrations and, by inference, gonadal regression, occurred in spring when daylength was still increasing, indicating that these penguins become absolutely photorefractory.

J. F. Cockrem J F M A M J J A S O N D Month Fig. 6. Plasma concentrations of testosterone, oestradiol and progesterone in female North Island brown kiwis in Northland. Values are meanf s.e. (n = 2-16); where n < 3, individual values are shown. Sample sizes are indicated across the top of figure. Where sample sizes for testosterone differ from other hormones, these are given in parentheses. From Potter and Cockrem (1992). Kakapo The kakapo is an endangered parrot and, like the kiwi, is flightless and nocturnal, but entirely herbivorous. The reproductive cycle of the kakapo is unusual in that the species is a lek breeder (Merton et al. 1984). Males gather in groups to display to females but, after copulation, play no part in incubation of the eggs or in chick rearing. Historically, the kakapo was thought to breed only every 3-4 years and breeding was suggested to occur during the years in which the availability of fruit from forest trees was highest. However, Cockrem (1989) suggested that the kakapo could breed annually, and this was confirmed on Little Barrier Island when breeding occurred in both 1990 and 1991 after the introduction of supplementary food. Egg-laying occurs from late January to March (in late summer). In years when egg-laying does occur, it is synchronized between females, all eggs being laid over a 4-5-week period (D. V. Merton, personal communication). L a o l - - - -... J F M A M J J A S O N D Month Fig. 7. Annual changes in plasma concentrations of oestradiol, testosterone and progesterone in yellow-eyed penguins. Samples were collected monthly from September 1985 to August 1986 (n = 8-33). From Cockrem and Seddon (1994). Since it is not practical to regularly collect blood samples from kakapo, an alternative approach to the study of reproductive endocrinology in the kakapo has been necessary. As a result, concentrations of oestradiol and testosterone were measured in the droppings from kakapo on Little Barrier Island (Cockrem and Rounce 1995). There was a clear annual cycle in the ratio of testosterone:oestradiol in droppings (Fig. 8). The ratio reached a peak in February at the time of copulation and maximum sexual behaviour, then decreased rapidly reaching low values in April. The ratio of testosterone:oestradiol increased again in spring when male sexual behaviour was starting again. As most of the kakapo on Little Barrier Island are male, although the droppings studied by Cockrem and Rounce (1995) could not be ascribed to individual kakapo, most were assumed to have been from males. These data indicated that male kakapo undergo an annual cycle of plasma concentrations of testosterone and, hence, of gonad size, with peak hormone concentrations and maximum gonad size occurring in late summer.

Seasonal Breeding in New Zealand Birds comparisons are made to emphasize that the breeding seasons of New Zealand birds are not unique. O N D J F M A M J J A S O N D Month Fig. 8. Ratios of concentrations of faecal testosterone:oestradiol in droppings collected from 1989 to 1990 from kakapo on Little Barrier Island. Data are plotted as meanhe. Sample sizes were 3-6 droppings per month except in October 1989 (n = 9) and March 1990 (n = 24). From Cockrem and Rounce (1995). Control of Seasonal Breeding in New Zealand Birds Annual cycles of steroid hormone concentrations have been found in the North Island brown kiwi, the yelloweyed penguin and the kakapo. In all three species, steroid concentrations were maximal during the time of greatest sexual activity, suggesting that endocrine cycles reflect similar cycles of gonadal growth and regression. These data are entirely consistent with those from many studies of Northern hemisphere birds. These results, together with the finding that the seasonal pattern of egg-laying in New Zealand birds is not unique, support the hypothesis that the physiological basis of seasonal breeding in New Zealand birds is the same as that in other birds. The model presented above can be applied to the three New Zealand birds for which there is information on the physiology of their annual reproductive cycle. It is possible that all New Zealand birds are photoperiodic, although at present there is experimental evidence for this in only one species, the grey duck (Anas superciliosa) (Cockrem, unpublished observations). In this model, therefore, the North Island brown kiwi, the yelloweyed penguin and the kakapo will be assumed to be photoperiodic. The basic features of the model for the control of seasonal breeding, as applied to these three New Zealand birds, are summarized in Table 3 and presented in Fig. 9. For each native species, another bird with a similar reproductive cycle is also illustrated. These North Island Brown Kiwi It is suggested that the kiwi becomes photosensitive in autumn, and that sensitivity to steroid feedback is low at this time. Gonadal growth may occur on shortening daylengths and the gonads can remain enlarged through spring and early summer (Fig. 9). The time of egglaying and the number of clutches laid each season varies between birds, and reproduction in individual birds must be influenced by supplementary information. The birds gradually develop a form of relative photorefractoriness, so that gonadal regression occurs by late summer. Finally, the kiwi becomes photosensitive again in mid autumn. The annual cycle of the starling is presented in Fig. 9 to provide a comparison between the North Island brown kiwi and a species for which extensive experimental data are available. The starling is a classical spring breeder, with egg-laying occurring from October to early December followed by a period of absolute photorefractoriness. However, when gonadal steroid feedback is removed by castration it is seen that plasma LH concentrations increase in autumn (Nicholls et al. 1988) at the same time as gonadal growth is beginning in the kiwi. The kiwi is, therefore, analogous to a starling in which the sensitivity to steroid feedback has been lowered and the timing of photorefractoriness is delayed. Yellow-eyed Penguin It is suggested that the penguin becomes photosensitive in the autumn, but that steroid feedback is strong and significant gonadal growth does not occur on short daylengths (Fig. 9). Steroid feedback is eventually overcome by photoperiodic drive from increasing daylengths in spring, leading to a brief egg-laying period which is synchronized between birds. The gonads remain at their maximum size for only a short time before regressing due to the onset of absolute photorefractoriness when daylengths are still increasing. The penguins remain photorefractory for some time and finally regain photosensitivity in the autumn. This model is identical to the one proposed for the rook (Lincoln et al. 1980; see Fig. 9). Rooks in Scotland have a short egg-laying season and lay at the same time Table 3. Postulated physiological control of seasonal breeding in three New Zealand birds Species Egglaying Steroid Photorefractoriness: Relative importance of feedback onset duration supplementary information North Island brown kiwi July-Feb. Low Slow Short Moderate Yellow-eyed penguin Sept.-Oct. High Rapid Long Low Kakapo Jan.-Feb. High Slow Short High

J. E Cockrem North Island brown kiwi Starling Yellow-eyed penguin Rook n 1. Spur- winged goose V -.I. 0.2 1 I I I I I I I Apr. Aug. Dec. Apr.... I I I I I I Apr. Aug. Dec. Apr. Fig. 9. Schematic representation of the breeding seasons of three endemic New Zealand birds (North Island brown kiwi, yellow-eyed penguin, kakapo) - a species from another region with a similar breeding pattern is presented for comparison (starling, rook, spur-winged goose respectively). For each species, the annual cycle of daylength is shown as well as gonadal function (dark area), the egg-laying period (---) and the photorefractory period (horizontal bar). The photorefractory period for the starling only has been determined experimentally - in the others, these periods are speculative. Gonad function for each of the New Zealand birds is based on plasma steroid concentrations in the North Island brown kiwi (383, Potter and Cockrem 1992) and in the yellow-eyed penguin (469, Cockrem and Seddon 1994), and on faecal steroid concentrations in the kakapo (3S0S, Cockrem and Rounce 1995). The scheme for starlings is based on data for British starlings (50"N). The dark curve of gonadal function is taken from plasma steroid concentrations (Dawson 1983), and the stippled area represents the difference between intact birds and the annual cycle of plasma LH in ovariectomized birds (adapted from Nicholls et al. 1988) in which steroid feedback has been removed. The scheme for rooks is based on data from Scotland (57"N), and the curve for gonad function is based on testis weight (Lincoln rt a[. 1980). The scheme for spur-winged geese is based on data from southern Africa (26"S), and the curve for gonad function is based on testis weight (Halse 1985).

Seasonal Breeding in New Zealand Birds 17 (adjusted for the Southern hemisphere) as the yellow-eyed penguins. They have a short peak of maximum testis size and Lincoln et al. (1980) suggested that this is followed by a period of photorefractoriness. The breeding cycles of the endemic yellow-eyed penguin and the Northern hemisphere rook can thus be explained by the same model. Kakapo The kakapo, like the other bird species, probably becomes photosensitive in the autumn. Steroid feedback is likely to be strong, and gonadal growth occurs in late spring or early summer as photoperiodic drive increases (Fig. 9). It is suggested that photorefractoriness in the kakapo is slow to develop, in contrast to the yellow-eyed penguin, and does not switch-off reproduction until late summer when gonadal regression occurs. There is a brief period of photorefractoriness leading to the development of photosensitivity and the start of the annual cycle again. The timing and extent of reproductive activity in the kakapo is variable from year to year. Extensive social displays by male kakapo may start in October, may be delayed until December or January, or may not occur at all. Similarly, egg-laying does not occur each year and supplementary information is clearly very important for breeding in this species. The kakapo can be compared with the spur-winged goose (Plectropterus gambensis), a species of southern Africa, using the data from Halse (1985) on egg-laying and the annual cycle of testis size (Fig. 9). Social organization differs between the two species but, at 26"S, the spur-winged goose lays eggs at the same time as the kakapo and the timing of the cycle of testis size in this goose is directly comparable with the steroid data for the kakapo. The timing of breeding in the kakapo is clearly not species-specific and the model described above for the kakapo can be applied to the spur-winged goose. These results show that the general model for birds can be used to explain the timing of breeding in all three New Zealand species for which physiological data are available. Conclusions In this review, a model to explain the timing of breeding in birds was presented. In this model, the underlying physiological control of avian breeding seasons is assumed to be similar for all birds, despite the differences in the timing that occur. The model is based on the assumption that birds have an internal rhythm of reproduction which is synchronized with seasonal environmental changes by external factors, especially the annual cycle of daylength. The timing of breeding is also modulated to varying degrees by environmental factors. Therefore, variation within and between species in the timing of avian breeding seasons results from variation in the relative importance of initial predictive and supplementary information, and from variation in the nature of the photoperiodic response. The breeding seasons of New Zealand birds were shown to follow the patterns of other birds, and the model of the timing of breeding in birds can be applied to New Zealand birds. It is, therefore, concluded that the basic physiological mechanisms controlling the timing of seasonal breeding in New Zealand birds are similar to those of other birds. This has two important implications for the study and conservation of native New Zealand birds. Firstly, physiological principles derived from detailed experimental studies of a few species of birds, such as the starling and Japanese quail, can be applied to all New Zealand birds. Secondly, it should not be assumed that New Zealand birds are so distinctive that information from physiological and ecological studies of breeding in other birds is not relevant. Instead, such information should be used much more extensively as a guide in the management programmes for endangered New Zealand birds. Acknowledgments I thank Professor Sir Brian K. Follett for introducing me to the physiology of seasonal breeding in birds, Dr K. P. McNatty and Professor D. J. Mellor for their support of the studies of New Zealand birds, and Drs G. F. Ball and A. Dawson for stimulating discussions and comments on the manuscript. References Anon. (1985). 'Reader's Digest Complete Book of New Zealand Birds.' (Reader's Digest: Sydney.) Ashmole, N. P. (1963). The biology of the wideawake or sooty tern Sterna fuscata on Ascension Island. Ibis 103, 297-364. Baker, A. J. (1991). A review of New Zealand ornithology. Curr. Ornithol. 8, 1-67. Baker, J. R. (1938a). The evolution of breeding seasons. In 'Evolution: Essays on Aspects of Evolutionary Biology'. (Ed. G. R. de Beer.) pp. 161-77. (Oxford University Press: London.) Baker, J. R. (1938b). The relation between latitude and breeding seasons in birds. Proc. Zool. Soc. Lond. A 108, 557-82. Ball, G. F. (1993). The neural integration of environmental information by seasonally breeding birds. Am. Zool. 33, 185-99. Bartholomew, G. A. (1949). The effect of light intensity and day length on reproduction in the English sparrow. Bull. Mus. Comp. Zool. 101, 433-76. Beauchamp, A. J. (1987). A population study of the weka Gallirallus australis on Kapiti Island. Ph. D. Thesis. Victoria University of Wellington, New Zealand. Bissonnette, T. H. (1930). Studies on the sexual cycle in birds. I. Sexual maturity, its modification and possible control in the European starling (Sturnus vulgaris). Am. J. Anat. 45, 289-302. Blackburn, A. (1965). Breeding of the North Island fantail. Notornis 12, 127-37. Bullough, W. S. (1942). The reproductive cycles of the British and continental races of the Starling (Sturnus vulgaris L.). Phil. Trans. R. Soc. Lond. B. Biol. Sci. 231, 165-246.

J. F. Cockrem Burger, J. W. (1953). The effect of photic and psychic stimuli on the reproductive cycle of the male starling, Sturnus vulgaris. J. Exp. Zool. 124, 227-39. Carroll, A. L. K. (1963). Breeding cycle of the North Island weka. Notornis 10, 300-1. Cockrem, J. F. (1989). Reproductive physiology and the management of the kakapo. Ecology Division Report No. 18. (DSIR: Lower Hutt, N.Z.). Cockrem, J. F., and Rounce, J. R. (1995). Non-invasive assessment of the annual gonadal cycle in free-living kakapo (Strigops habroptilus) using fecal steroid measurements. Auk. (In press.) Cockrem, J. F., and Seddon, P. J. (1994). Annual cycle of sex steroids in the yellow-eyed penguin (Megadyptes antipodes) on South Island, New Zealand. Gen. Comp. Endocrinol. 94, 113-21. Darby, J. T., and Seddon, P. J. (1990). Breeding biology of yelloweyed penguins (Megadyptes antipodes). In 'Penguin Biology'. (Eds L. S. Davis and J. T. Darby.) pp. 45-62. (Academic Press: San Diego.) Dawson, A. (1983). Plasma gonadal steroid levels in wild starlings (Sturnus vulgaris) during the annual cycle and in relation to the stages of breeding. Gen. Comp. Endocrinol. 49, 286-94. Dawson, A., and Goldsmith, A. R (1982). Prolactin and gonadotrophin secretion in wild starlings (Sturnus vulgaris) during the annual cycle and in relation to nesting, incubation, and rearing young. Gen. Comp. Endocrinol. 48, 213-21. Dawson, A., and McNaughton, F. J. (1993). Circannual rhythms in starlings: response of castrated birds to 12-hr days depends on photoperiodic history. J. Biol. Rhythms 8, 141-50. De Laet, J. F., Dhondt, A. A., and De Boever, J. G. (1985). Circannual plasma androgen levels in free-living male great tits (Parus major major L.). Gm. Comp. Endocrinol. 59, 277-86. Dittami, J. P. (1986). Seasonal reproduction, moult and their endocrine correlates in two tropical Ploceidae species. J. Comp. Physiol. B 156, 641-7. Dittami, J. P., and Gwinner, E. (1985). Annual cycles in the African stonechat Saxicola torquata and their relationship to environmental factors. J. Zool. (Lond.) 207, 357-70. Donham, R. S. (1979). Annual cycle of plasma luteinizing hormone and sex hormones in male and female mallards (Anasplatyrynchos). Biol. Reprod. 21, 127345. Falk, H., and Gwinner, E. (1988). Timing of photorefractoriness in the European starling: significance of photoperiod early and late in the reproductive cycle. Biol. Reprod. 39, 1004-8. Farner, D. S., and Gwinner, E. (1980). Photoperiodicity, circannual and reproductive cycles. In 'Avian Endocrinology'. (Eds A. Epple and M. H. Stetson.) pp. 331-66. (Academic Press: New York.) Farner, D. S., Follett, B. K., King, J. R., and Morton, M. L. (1966). A quantitative examination of ovarian growth in the white-crowned sparrow. Biol. Bull. 130, 67-75. Follett, B. K. (1984). Birds. In 'Marshall's Physiology of Reproduction: Reproductive Cycles of Vertebrates'. (Ed. G. E. Lamming.) pp. 283-350. (Churchill Livingstone: New York.) Follett, B. K., and Robinson, J. E. (1980). Photoperiod and gonadotropin secretion in birds. Prog. Reprod. Biol. 5, 39-61. Goldsmith, A. R., and Nicholls, T. J. (1984). Prolactin is associated with the development of photorefractoriness in intact, castrated and testosterone-implanted starlings. Gen. Comp. Endocrinol. 54, 247-55. Gwinner, E. (1986). 'Circannual Rhythms.' (Springer-Verlag: Heidelberg.) Gwinner, E., and Dittami, J. (1985). Photoperiodic responses in temperature zone and equatorial stonechats: a contribution to the problem of photoperiodism in tropical organisms. In 'The Endocrine System and the Environment'. (Eds B. K. Follet, S. lshii and A. Chandola.) pp. 279-94. (Japan Sci. Soc. Press: Tokyo.) Gwinner, E., and Wozniak, J. (1982). Circannual rhythms in European starlings: why do they stop under long photoperiods? J. Comp. Physiol. 146, 419-21. Hahn, T. P. (1993). Integration of environmental cues to time reproduction in an opportunistic breeder, the red crossbill (Loxia curvirostra). Ph.D. Thesis. University of Washington, Seattle. Hake, S. A. (1985). Gonadal cycles and levels of luteinizing hormone in wildspur-winged geese,plectropterus gambensis. J. Zool. (Lond.) 205, 335-55. Hegner, R. E., and Wingfield, J. C. (1986). Behavioral and endocrine correlates of multiple brooding in the semi-colonial house sparrow Passer domesticus. I. Males.Horm. Behav. 20, 294-31 2. Hirons, G. J. (1985). The importance of body reserves for successful reproduction in the tawny owl (Strh aluco). J. Zool. (Lond.) 1, 1-20. Hirons, G. J. M., Hardy, A. R., and Stanley, P. I. (1984). Body weight, gonad development and moult in the tawny owl (Strh aluco). J. Zool. (Lond.) 202, 145-64. Immelmann, K. (1971). Ecological aspects of periodic reproduction. In 'Avian Biology'. Vol. 1. (Eds D. S. Farner and J. R. King.) pp. 341-89. (Academic Press: New York.) Jolly, J. N. (1989). A field study of the breeding biology of the little spotted kiwi (Apteryx owemi) with emphasis on the causes of nest failure. J. R. Soc. N.Z. 19, 4434. Kinsky, E C. (1960). The yearly cycle of the northern blue penguin (Eudyptula minor novaehollandiae) in the Wellington harbour area. Rec. Dom. Mus. (Wellington) 3, 145-218. Kinsky, F. C. (1971). The consistent presence of paired ovaries in the kiwi (Apteryx) with some discussion of this condition in other birds. J. Ornithol. 112, 334-57. Konishi, M., Emlen, S., Ricklefs, R., and Wingfield, J. C. (1989). Contributions of bird studies to biology. Science 246, 465-72. Lincoln, G. A., Racey, P. A., Sharp, P. J., and Klandorf, H. (1980). Endocrine changes associated with spring and autumn sexuality of the rook Corvus frugilegus. J. Zool. (Lond.) 190, 137-53. Lofts, B., and Coombs, C. J. E (1965). Photoperiodism and the testicular refractory period in the mallard. J. Zool. (Lond.) 146, 44-54. Lofts, B., and Murtou, R. K. (1968). Photoperiodic and physiological adaptations regulating avian breeding cycles and their ecological significance. J. Zool. (Lond.) 155, 327-94. Marchant, S., and Higgins, P. J. (1990). 'Handbook of Australian, New Zealand & Antarctic birds. Vol. 1. Ratites to Ducks.' (Oxford University Press: Melbourne.) Marshall, A. J., and Serventy, D. L. (1959). Experimental demonstration of an internal rhythm of reproduction in a trans-equatorial migrant (the short-tailed shearwater Pufinus tenuirostris). Nature (Lond.) 184, 1704-5. Marshall, F. H. A. (1936). Sexual periodicity and the causes which determine it. Phil. Trans. R. Soc. Lond. B Biol. Sci. 226, 423-56. McLennan, J. A. (1988). Breeding of North Island brown kiwi, Apteryx australis mantelli, in Hawke's Bay, New Zea1and.N.Z. J. Ecol. 11, 89-97. Meijer, T., and Schwabl, H. (1989). Hormonal patterns in breeding and nonbreeding kestrels, Falco tinnunculus: field and laboratory studies. Gen. Comp. Endocrinol. 74, 14M0. Merton, D. V., Morris, R. B., and Atkinson, I. A. E. (1984). Lek behaviour in a parrot: the kakapo Strigops habroptilus of New Zealand. Ibis 126, 277-83.