Hatching asynchrony as a parental reproductive strategy in birds: a review of causes and consequences

Bird parents may choose among various breeding strategies. From an evolutionary point of view the outcome of their choice is believed to provide the highest reproductive success in a given environment. However, some of the mechanisms underlying parental reproductive decisions are still poorly understood, for example the phenomenon of hatching asynchrony. Resultant age and size hierarchy among nest mates often lead to competitive disadvantages, starvation and death of last-hatched siblings. Although synchronised broods generally result in higher fledging success, altricial bird species typically hatch asynchronously. Various hypotheses were proposed and tested to explain this strategy. A quarter of a century has passed since the last review of research on hatching asynchrony. In the meantime the number of studies investigating the issue has considerably grown, same as the array of hypotheses. Our aim is to present a comprehensive synthesis of the research on 20 hypotheses presented until recently as well as to indicate the blind spots in the reviewed area—9 out of 20 formulated hypotheses still lack any experimental approach that would confirm or reject the assumptions.


Introduction
Bird parents may choose among various breeding strategies, such as large vs. small clutches, open vs. cavity nesting or synchronous vs. sequential hatching of their young (Clark and Wilson 1981;Hansell et al. 2000). From an evolutionary point of view the outcome of their choice is believed to Communicated by S. Bouwhuis. provide the highest reproductive success in a given environment. Having said that, some of the mechanisms underlying parental reproductive decisions are still poorly understood. A good example is the phenomenon of hatching asynchrony (hereafter HA). In numerous bird species females start incubation before clutch completion, which results in sequential hatching of nestlings (Enemar and Arheimer 1989;Lord et al. 2011) and, consequently, pronounced size hierarchy among siblings (Forbes and Glassey 2000;Johnson et al. 2009). Last-hatched smaller nest-mates often experience competitive disadvantages and suffer more starvation than the rest of the brood (Wiebe 1996;Ostreiher 1997;Clotfelter et al. 2000;Pettifor et al. 2001;Ostreiher 2001;Jeon 2008;Soler et al. 2020but see Krebs 1999Węgrzyn 2012). And even if they manage to fledge, they seem to pass their initial developmental handicap to subsequent generations (Song et al. 2019) or they pay delaying fitness costs later in life (Vedder et al. 2021). Indeed, experimental studies confirmed that the hatching sequence has the potential to influence nestling phenotypes (Mainwaring et al. 2010;Mainwaring and Hartley 2013). Despite the fact that experimental synchronisation of broods generally results in higher fledging success (Nilsson 1995) altricial bird species typically hatch asynchronously (Clark and Wilson 1981;Lessells and Avery 1989). As it frequently leads to the death of the smallest nestling and to reduced fledging success, the question arises why birds (and other taxa) practice HA. Various hypotheses were proposed and tested in the quest for an answer (reviewed in : Magrath 1990;Clark and Wilson 1981;Stoleson and Beissinger 1995;Stenning 1996).
A quarter of a century has passed since the last review of research focusing on HA was presented by Stenning (1996). In the meantime the number of studies investigating HA have considerably grown. Also, the array of hypotheses aiming to explain sequential hatching have become wider-20 hypotheses can be found nowadays in scientific literature. Some of them shed a new light on the causes and consequences of HA while others still await confirmation.
The aim of the current review is to present a comprehensive synthesis of what has been learnt about HA since the first attempt to explain this curious behaviour by Lack (1947) and what areas are still the blind spots that await for testing. To make it lucid we have divided all studies into two groups, namely, hypotheses explaining HA and size hierarchy among siblings as (i) an adaptation, and (ii) a by-product of the onset of incubation before clutch completion. Within each group we present the hypotheses according to the date of their first publication rather than their relevance. We believe that such an approach provides maximum objectivity of our survey. We report both observational and experimental approaches to ensure an in-depth review of the studies exploring HA. We focus on studies released after the last review of HA by Stenning (1996), however we include older papers if they are crucial to outline a particular hypothesis or to show a wider perspective of the reviewed area. We also propose a new hypothesis, which may provide the rationale for nest-mates size and condition hierarchy during the post-fledging period, as opposed to to-date hypotheses that mainly refer to selective pressures operating directly in the nest.

Hatching asynchrony and sibling size hierarchy as an adaptation
The below presented group of hypotheses unanimously assume that sibling age and size hierarchy is beneficial despite costs connected with increased probability of the death of youngest nest-mates. The range of possible advantages is wide, both for nestlings and parents. Numerous researchers attempted to test different assumptions and the results of their rigid search are presented hereafter.
1. The Brood Reduction Hypothesis (Lack 1947(Lack , 1968) argues that HA is advantageous in unpredictable environments-when food is abundant all nestlings can survive but under food-shortage parents can produce a reduced brood at the expense of the death of the last-hatchlings in asynchronous broods. In synchronous broods, due to an increased competition between equal nest-mates adult birds are supposed to experience even lower breeding success. Thus HA is assumed to present an adaptive parental strategy for secondary adjustment in brood size to fit prevailing food conditions. Numerous studies have aimed to test this hypothesis but produced mixed results.
The major support for Lack's prediction still comes from observational studies indicating that last-hatched sibs have lower growth and/or survival rates than their older sibs (e.g., Malacarne and Cucco 1991;Martín-Vivaldi et al. 1999;Theofanellis et al. 2008;Sockman 2008;Weatherhead 2008, 2012;Ortiz-Catedral et al. 2013; Sockman 2018 but see Stier et al. 2015). Interestingly, the survival of marginal nestlings was affected not only by their social rank but also by family size and the interaction of rank and core nestlings number (Forbes 2011). Additionally, some bird species decrease egg mass with the laying order, which together with HA result in pronounced size and competitive hierarchy among nestlings, suggesting complex adaptation for brood reduction in an unfavourable environment (Heg and van der Velde 2001;Groothuis et al. 2005;Bosman 2014;Lahaye et al. 2015). Yet, other descriptive studies report quite the opposite-HA can be a tool to minimize sibling asymmetries. For example, Eagle Owl Bubo bubo L. females may produce the nestlings of the larger sex as the last hatchlings in asynchronous broods, which reduces size differences between nest-mates and prevents siblicide (Penteriani et al. 2010). Also in Chinstrap penguins Pygoscelis antarctica F. no connection was found between HA and nest-mates size hierarchy, early brood reduction in unfavourable circumstances or improved growth of the surviving chick (Moreno et al. 1994). However, it should be stressed that the line of evidence presented above is correlational and thus cannot prove or reject the hypothesis.
As it comes to experimental studies testing The Brood Reduction Hypothesis, a number of them manipulated only hatching spread but not food availability. As the brood reduction hypothesis predicts the advantage of HA under food limitation, not surprisingly some research found no differences between reproductive success of synchronous and asynchronous broods (Hébert 1993) while others even reported greater fledging rate of synchronous broods (reviewed in Amundsen and Slagsvold 1991). However, the experimental approach devoid of food manipulation seems not to provide a reliable test of Lack's hypothesis.
Less frequent studies that experimentally tested the effect of HA under food shortage rather rejected than supported the hypothesis. The importance of food manipulation in testing The Brood Reduction Hypothesis was elegantly presented in the study by Hébert and McNeil (1999). The authors found that without food limitation survival rate of first-hatched chicks, total brood loss, and fledging success were higher in asynchronous broods as predicted by the hypothesis, however no differences were found in regard to above factors between food-stressed asynchronous and synchronous broods. Similarly, under food limitation no differences were demonstrated between synchronous and asynchronous broods of House Wrens Troglodytes aedon V. in terms of fledging success (Harper et al. 1992). Also, offspring mortality was similar in synchronous and asynchronous broods of the Great Tit Parus major L. when food was limited (Podlas and Richner 2013). But the opposite results have also been found. Under adverse food conditions asynchronous broods were more productive in Blackbirds Turdus merula L. (Magrath 1989) and Yellow Warblers Setophaga petechial L. (Hébert 1993). However, it should be mentioned that the latter experiment involved only the manipulation of hatching spread, while food limitation occurred fortuitously in the second year of the study. Consequently, this parameter was not entirely controlled and other seasonal effects may have occurred.
Another line of evidence supporting an adaptive function of brood reduction in asynchronous broods emerges from experiments in which dead runts were replaced (Husby 1986). The experimental treatment resulted in decreased growth and higher mortality of siblings, indicating that in adverse conditions the elimination of runts enables to secure the survival of core offspring. Somewhat different outcomes emerge from a study by Forbes and Glassey (2000), which demonstrated that the presence or absence of marginal offspring in experimentally manipulated broods had no effect upon the growth of core offspring.
The Brood Reduction Hypothesis assumes that HA is advantageous if runts are efficiently reduced when food runs low, yet under favorable conditions all nestlings can fledge independent of hatching order. This proved to be true in Eurasian Kestrels Falco tinnunculus L.-at fledging, only in nests without access to supplemented food was the body condition of last-hatched chicks lower than that of its siblings, but in nests with access to supplemented food no such difference was detected (Massemin et al. 2002). Similar findings were reported in a study on White Storks Ciconia ciconia L.-the reproductive value of last laid eggs doubled when females had access to extra food (Djerdali et al. 2016). A comparative analysis of parent-offspring communication across 143 bird species revealed that in predictable and good environments, chicks in worse condition beg more, and parents preferentially feed those chicks but in unpredictable and poor environments, parents ignore intensive begging of marginal offspring and instead rely on size cues or structural signals of quality (Caro et al. 2016).
Other studies, however, found that mortality of the last-hatched siblings is frequent even if there is more than enough food for all the chicks (Brommer et al. 2004;Kontiainen et al. 2010). HA and the resulting size hierarchy were detrimental for the growth of nestlings in both good and bad years in Collared Flycatchers Ficedula albicollis T. (Szöllősi et al. 2007). The same study revealed that parental survival was not related to the size hierarchy in the broods, and the average body mass growth of the nestlings was slower in broods with a high initial size variance. An interesting outcome emerged from a study on Australian Pelicans Pelecanus conspicillatus T., which showed that senior chicks did not differ in growth rate between nests in which their younger sibs survived and nests in which brood reduction occurred (Johnston 2018).
Despite the fact, that The Brood Reduction Hypothesis was proposed over 60 years ago, in the light of the above cited studies Lack's assumption still looks only partly supported. However, most studies have examined the consequences of HA only in short-term perspective, using for example the number of fledglings as the measure of possible benefits. Studies of post-fledging survival of individuals from synchronous and asynchronous clutches would be needed to see the issue in a wider perspective.
2. The Insurance Hypothesis (Dorward 1962) is another special case of Lack's conception of parents adjusting brood size to resource availability. It explains the adaptive function of HA in bird species with brood reduction resulting from obligate siblicide. In such species parents typically lay two eggs but nestlings hatch asynchronously and the older nestling soon kills its younger sib, except when the first egg fails to hatch or produces a weak chick. Elimination 1 3 of the nest-mate is facilitated by nestlings size asymmetry ensured by HA. Size hierarchy is often further emphasized by decreasing egg size with laying order (Slagsvold et al. 1984 but see Poisbleau et al. 2011). HA reduces the risk of injuries of the older nest-mate and parental investment in raising the brood before its reduction. The hypothesis proposes that the second egg is laid as an insurance against the first egg failure. The last-hatched nestling has therefore only a replacement value. Theoretical models predict that the efficiency of an insurance strategy is enhanced, among other factors, by HA which ensures that offspring are reared in environments that are at least partially uncorrelated (Forbes and Lamey 1996). The Insurance Hypothesis differs from The Brood Reduction Hypothesis mainly in the fact that it presents HA as a parental response rather to unpredictable brood size after hatching than to unpredictable resources. The hypothesis was tested and confirmed, among other studies, in the Masked Booby Sula dactylatra L. (Anderson 1990), the American White Pelican Pelecanus erythrorhynchus G. (Evans 1997), the Brown Booby Sula leucogaster B. (Tershy et al. 2000), the Nazca booby (Clifford and Anderson 2001;Humphries et al. 2006), the Bearded Vulture Gypaetus barbatus L. (Margalida et al. 2004) and the Sandwich Tern Sterna sandvicensis L. (Stienen and Brenninkmeijer 2006), demonstrating that the important premise for laying the insurance egg is low hatching success. In the environment that allows parents to rear only one chick, HA leading to instant elimination of the younger sibling when the survival of the older one is assured, has undoubtedly an adaptive value. Predictions of the hypothesis are supported by the fact that marginal and core offspring are functionally equivalent in the absence of sibling interactions, and that core offspring incur no detectable costs from behaving siblicidally (Humphries et al. 2006). The marginal chicks, however, were reported to have higher levels of testosterone, even during times of no attack, because the core chicks represented such a threat (Ferree et al. 2004). The insurance value of the second egg was experimentally confirmed by Cash and Evans (1986). In American White Pelicans, a typical obligate species, parents that were deprived of their insurance coverage by the removal of the marginal egg had lower breeding success (due to hatching failure of the core egg) than control pairs with two eggs. Parents with experimental three eggs fared no better than control parents with two eggs, indicating that one extra egg produced insurance coverage, however a second insurance egg yielded no additional benefit. Studies on Black Eagles Aquila verreauxi L. and Masked Boobies revealed that the chance of the younger chick to be the only survivor is approximately 22%, which is the insurance reproductive value of the marginal chick to the parents (Gargett 1977;Kepler 1969). At the same time the cost of producing one additional egg is relatively low and accounts for about 2.5% of the female body weight. The insurance value of marginal progeny seems to exceed the cost of its production and thus it is expected to improve parental fitness. The ideal experiment to test the hypothesis completely would investigate both breeding success of pairs with and without the marginal egg and the future fitness of the females that lay one vs. two eggs. However, such an approach is not feasible as in obligate siblicial species females typically lay two eggs. Thus, the only way to estimate the fitness consequence of laying an extra egg is to look at the cost of its production.
Interestingly, The Insurance Hypothesis was also confirmed in species practicing facultative siblicide and rearing more than two sibs but experiencing hatching problems (Wiebe 1996;Forbes et al. 1997). Another remarkable breeding strategy involving an insurance egg was reported in Southern Rockhopper Penguins Eudyptes chrysocome chrysocome F., which present a reversed hatching asynchrony (Poisbleau et al. 2011). Contrary to other avian models, laying and hatching sequences do not coincide in this species. Females lay two eggs, of which the second one is heavier and contains higher levels of androgens. The second-laid egg hatches before the smaller first-laid egg and the lasthatched sibling soon dies of starvation. Noteworthy, the dominant nestling does not kill its smaller nest-mate directly but instead it monopolizes food resources to the extent that is detrimental to the subordinate sibling.
Although existing data are not very broad, the consistent outcomes of different studies well support The Insurance Hypothesis.
3. The Sibling Rivalry Reduction Hypothesis (Hamilton 1964) argues that parents create asynchronous hatching patterns to minimize sibling competition, which results in less energy spent on sibling rivalry. Evidence suggests that parents work less with asynchronous than with synchronous broods in some species (Fujioka 1985;Mock and Ploger 1987) but not in others (Bryant and Tatner 1990). Size hierarchy should maintain a stable dominance of older nestmates and reduce fighting (Hahn 1981;Mock and Ploger 1987;Viñuela 1999Viñuela , 2000. The hypothesis was confirmed in a study of Blue-footed Boobies Sula nebouxii M.-E. (Osorno and Drummond 1995), which revealed that asynchronous broods showed less aggression and food consumption than synchronous ones. Some support for the hypothesis also comes from the study on Linetts Carduelis cannabina L. (Drachmann et al. 2000), which demonstrated higher growth rates of asynchronously hatched broods. In Linett nests HA increased with brood size, indicating the importance of reducing competition between siblings. One study, which clearly demonstrates that HA reduces the energy expended by nestlings in competition was published by Massemin et al. (2003). Daily energy expenditures of the Eurasian kestrel nestlings (measured using doubly-labelled water) was similar between the first-hatched nestlings in asynchronous broods and the size-matched nestlings from synchronous broods, indicating the same level of energy expended by the largest siblings on competition. At the same time daily energy expenditures of the last-hatched siblings were 35% lower, which led to a decrease in a the total brood energy requirements in asynchronous broods, and consequently in lower parental food delivery rates. Interestingly, a study on the Black Kite Milvus migrans B. showed that moderate HA reduced sibling rivalry, however large size differences between sibs increased aggression and resulted in cainism (Viñuela 1999).
A study on the American Robin Turdus migratorius L. looked for the relation between HA and parental provisioning of the broods, but it lent no support to the tested hypothesis (Slagsvold 1997). Another way of assessing the cost of breeding is to investigate subsequent survival of parents. A study of Blue Tits Cyanistes caeruleus L. using such an approach revealed no parental benefits from hatching asynchrony (Slagsvold et al. 1994(Slagsvold et al. , 1995. However, when the sexes were analyzed separately, the same study suggested that females benefited rather from synchronous broods, with the opposite results for males. This notion led to The Sexual Conflict Hypothesis presented later on in the review. (Hussell 1972) proposes that a breeding couple may decrease their workload during the most energetically demanding nestling stage by spreading out the ages of offspring, so that not all nestlings reach the highest energy demand simultaneously. Theoretical approaches using mathematical models predicted that due to HA parents may obtain the peak load reduction of 1-5%, depending on the study (Bryant and Gradiner 1979;Mock and Schwagmeyer 1990). It appears doubtful whether this level of reduction provides a substantial evolutionary force for HA. An empirical study of the European Beeeater Merops apiaster L. revealed that HA provided negligible benefits to bird parents (Lessells and Avery 1989). Experimental studies brought mixed results. In some species synchronous broods presented increased consumption which resulted mainly from more pronounced sibling rivalry (Fujioka 1985;Mock and Ploger 1987;Wiebe and Bortolotti 1994;Viñuela 1999;Hébert and McNeil 1999;Gilby et al. 2011). However, in others hatching synchrony had no detectable effect on the parent body mass (Amundsen and Slagsvold 1991) or the rate of nestling provisioning (Slagsvold 1997;Stoleson and Beissinger 1997). An experimental study by Siegel et al. (1999) using the doubly labelled water technique to measure parental energy expenditure demonstrated in an elegant way that the reduction in peak load due to HA was marginal and statistically insignificant. However, the authors of the study pointed out that HA shortened the duration of high parental energy expenditure. In other words, the workload of parents of asynchronous broods was better balanced-they took care of a brood for a longer period but in fewer days they had to work close to the peak load level. Thus HA may potentially change the distribution of parental effort required throughout the nestling period. If this is the case, it may be beneficial especially for individuals in poorer condition. A study on Zebra Finches Taeniopygia guttata V. (Mainwaring et al. 2014) demonstrated that parental feeding effort was a little, but significantly, lower for parents raising asynchronous broods. However, the authors analysed the overall provisioning feeding effort, leaving the question of peak load reduction unanswered. Similar results were obtained in Ring-billed Gulls Larus delawarensis O. (Hébert and McNeil 1999). On the other hand, in Blue-throated Beeeaters Merops viridis L. the provisioning rate did not vary with the degree of asynchrony in the brood (Stader 1994). Thus, hatching asynchrony did not appear to reduce the work load of the parents, at least with respect the number of provisioning visits.

The Peak Load Reduction Hypothesis
5. The Larder Hypothesis, also known as The Ice-box Hypothesis (Alexander 1974) proposes that during a time of relative plenty a female could store food for a dominant nestling in the form of a subordinate one. Not much support for this hypothesis comes from bird studies. In siblicidal raptors cannibalism was occasionally observed (Wiebe 1996;Margalida et al. 2004), however the relatively small size of dead nestlings and/or the abundant food available at the nest rather discounts this conception as the major evolutionary drive of HA. Additionally, most studies on siblicidal species report that dead subordinate chicks were not eaten after they had been killed (Anderson 1990;Magrath 1990;Ploger 1992;Humphries et al. 2006). (Slagsvold and Lifjeld 1989) argues that different patterns of hatching span (synchronous vs. asynchronous) may impose various energy requirements on each parent sex and each sex can optimise its energy input at the mate's cost. As the start of incubation mostly depends on female decision, she can choose the most advantageous option for her. The hypothesis was first formulated based on a study of Pied Flycatchers Ficedula hypoleuca, P. in which Slagsvold and Lifjeld (1989) observed that females rearing asynchronous broods are in better condition than those rearing synchronous ones. Thus the authors concluded that HA benefited females and resulted from their control over the onset of incubation. Two possible advantages for females were put forward. The first assumption was that in asynchronous broods, where some nestlings hatch earlier than if all hatched together, provisioning period of the males is extended because males start feeding the young just after hatching. Incubating and brooding are less demanding than is food collecting, so females may save their energy at the expense of male workload. Moreover, the male often provides a portion of the food directly to the female during brooding period, so she may control the division of the food between hatchlings and herself as well as increase the male investment through her demand behaviour. The second premise that HA may be beneficial for females based on the fact, that the advanced start of nestling provisioning by the male is asynchronous broods would prevent him from allocating his efforts to other activities that are conflicting with the interests of the female, for example attracting a second female. The authors assumed that HA may lower the male's chance of becoming polygynous and thus provide the female with the exclusive access to the feeding effort of the male. Unfortunately, the authors did not investigate the relation between male monogamy and HA. The idea behind the hypothesis was supported solely by the fact that Flycatcher females rearing asynchronous broods had higher body masses at the time of fledging than females rearing the synchronous ones. However, it should be noted that the study recorded neither parental provisioning effort nor female weight at the beginning of breeding attempt. Thus it is equally likely that mothers of synchronous broods lost more weight or that females in initial poorer condition for some reason decided on synchronous broods. Additionally, the authors were not able to demonstrate a disadvantage of asynchronous hatching to males, which appears to be a necessary condition to confirm the hypothesis. Few other experimental tests failed to confirm The Sexual Conflict Hypothesis. Hébert and Sealy (1993a) showed that in the Yellow warbler both sexes had the same interests and benefited from HA. Similar results were reported in a study of the Pied flycatcher (Amundsen 1993). Finally, experiments on the Green-rumped Parrotlets Forpus passerines L. revealed no difference in nestling provisioning or parental survival between synchronous and asynchronous treatments (Stoleson and Beissinger 1997).

The Sexual Conflict Hypothesis
7. The Exploitation of Mate Hypothesis (Slagsvold et al. 1995). Few years after the first formulation of The Sexual Conflict Hypothesis Slagsvold found in Blue tits an exactly opposite pattern to that previously observed by him in Pied Flycatchers-females seemed to benefit from synchronous broods in terms of survival to the next breeding season (Slagsvold et al. 1994). The reverse was true for males, which suggested sexual conflict. This time the theory was called The Exploitation of Mate Hypothesis although it involved exactly the same idea as The Sexual Conflict Hypothesis proposed earlier by the same author. In his subsequent study Slagsvold suggested that also American robin females seemed to benefit more from synchronous broods (Slagsvold 1997)-the authors deducted that females contributed more to experimental asynchronous broods because they tended more often than males to feed the young that remained in the nest when the brood had partially fledged. However, it should be mentioned that the sample size of the latter study was insufficient (due to heavy predation) and the results obtained by the author were not statistically significant. Also the study seems to suffer from other methodological problems-for example, video recordings of parental feedings were conducted at any time of the day, which may have accounted for the observed differences in parental provisioning. Similarly, there is no evidence that synchronous broods, which appeared to be more beneficial for females, were the most common in the population. Thus, despite the information given in the abstract, this study does not really prove the hypothesis proposed by its authors. Moreover, studies on other species did not demonstrate female preference towards feeding the smallest nestlings (Smiseth et al. 1998(Smiseth et al. , 2003. In conclusion, there is very little support for the assumption that sexual conflict is an important evolutionary force in shaping HA in birds. 8. The Sex Ratio Manipulation Hypothesis (Slagsvold 1990) assumes that mothers may influence the fitness of sons and daughters by assigning them certain positions in laying order and thus impose a secondary sex ratio adjustment because the survival of the first-and last-hatched siblings differs in asynchronous broods. Females may do so for several reasons. In its initial version the hypothesis referred to sexually size dimorphic species. Because in bird species with sexual size dimorphism the nestlings of the larger sex may be more vulnerable if food is limited (reviewed in: Sheldon 1998), to keep the sex ratio at fledging close to unity the larger sex may be placed at favoured positions, i.e. as firsthatched siblings in asynchronous broods. HA may also be used as a maternal tactic to compensate growth rate between sexes, and the slower developing sex may be hatched first to compensate their growth rate disparities (Legge et al. 2001;Badyaev et al. 2002Badyaev et al. , 2006Magrath et al. 2003;Darolová et al. 2008). Above examples show that the combination of HA with a specific positioning of sexes in laying order allows to keep the sex ratio at fledging close to unity, which otherwise would be biased toward males or females. Alternatively, females may use HA to produce the favoured sex earlier in the laying sequence, thus giving them a competitive advantage over younger siblings. For example, a study on the cooperatively breeding Carrion Crows Corvus corone corone L. showed that male offspring are more favoured in the first breeding attempts because they are the philopatric sex which provide parents with help during subsequent breeding attempts (Canestrari et al. 2012). Later on female offspring are more favoured because they disperse and decrease within-group competition for resources. Accordingly, crow females may allocate offspring sex along the hatching sequence in a way that provide the highest fledging probability to sons in the first breeding attempt and to daughters in the following ones. In this case a secondary sex ratio adjustment with the use of HA creates a bias in sex ratio in a given breeding attempt.
An experimental synchronisation of hatching in Yellowlegged Gulls Larus michahellis N. revealed that survival of male chicks in synchronised broods was reduced compared to that of their sisters and that of male chicks in control (asynchronous) broods (Saino et al. 2011). However, quite the opposite results were obtained in the study of Lesser Black-backed Gulls Larus fuscus L.-in experimentallycreated synchronous broods there was no difference in survival of the sexes, while sex-biased mortality occurred in asynchronous broods (Bradbury and Griffiths 1999). Similar findings were reported in a study on the House wrenbecause the effect of hatching order was sex specific, females hatching their eggs asynchronously biased first-laid, firsthatching eggs toward sons and late-laid, late-hatching eggs toward daughters, whereas females hatching their eggs synchronously distributed the sexes randomly among the eggs of their clutch (Bowers et al. 2011). In all cited cases the adaptive possibility of giving one sex priority before the other well explains the evolution of asynchronous hatching in birds. Having said that, other studies demonstrate that, despite the potential advantages, offspring sex ratio is not manipulated in asynchronously hatched broods (Arnold and Griffiths 2003;Budden and Beissinger 2004;McDonald et al. 2005;Desfor et al. 2007). On the other hand, the opposite scenario may also be true-females may adjust the sex-ratio of offspring in relation to laying order to reduce the often strong effects on HA on condition and/or survival of male and female offspring (Badyaev et al. 2002). The discrepancies among cited studies may result not only from the level of adaptiveness of the sex ratio manipulation in different species and under different environmental conditions but also from different patterns of parental care (discussed in: Hasselquist and Kempenaers 2002). Thus, it is worth to keep in mind that HA is not the only factor that may be involved in the sex ratio manipulation in birds.
9. The Dietary Diversity Hypothesis (Magrath 1990) argues that parents may bring a great variety of food to different aged nestlings and thus are not forced to search for only some special food types. Asynchronous hatching serves here to increase parental feeding efficiency when a particular limiting food resource is the constraint. This hypothesis applies only to species in which nestlings depend on a specific food during some nesting stages and parents may be not able to deliver this food to the entire brood at the same time. Unfortunately, to our knowledge, there are still no studies directly testing this hypothesis. Some support comes from a study by Barrientos et al. (2016), showing that HA takes place more often in a specialist vs. a generalist tit species. Another line of evidence emerged from a study of Blue-throated bee-eaters, which demonstrated that parents of asynchronous broods brought insects of different, and presumably appropriate sizes, to their broods (Stader 1994). Although the study did not test whether differentiating food types facilitated provisioning, it is possible that this represents a reduction in the search time or the effort for bird parents. On the other hand, Pied Flycatcher parents of asynchronous broods brought large items suitable for the size of the older nestlings despite the fact that runts had difficulty swallowing such a prey, which led to their increased mortality (Slagsvold and Wiebe 2007). The study suggests that larger prey were more profitable to bring to the nest than smaller prey, thus undermining the benefits of prey size diversification. Another study on Pied flycatchers and Blue tits showed that, despite some plasticity, the feeding patterns are fairly innate and fixed (Wiebe and Slagsvold 2014), which seems somewhat counter to The Dietary Diversity Hypothesis.
10. The Offspring Quality Assurance Hypothesis (Amundsen and Slagsvold 1991) was proposed as an extension of Lack's hypothesis. It argues that although asynchronous broods may produce fewer fledglings, but these are of a higher quality than the more numerous offspring from synchronous broods. The rationale for this assumption comes from the fact that nestling recruitment is severely affected by its condition at fledging time (Bortolotti 1986;Tinbergen and Boerlijst 1990;Slagsvold et al. 1995 but see Kontiainen et al. 2010). Again, some studies supported the hypothesis, while other found the opposite results (reviewed in Amundsen and Slagsvold 1991). Recent studies also brought mixed results. For example, in Great tits experiencing harsh conditions nestling mass and size were reduced in asynchronous broods compared to synchronous ones (Podlas and Richner 2013). HA reduced the mean nestling weight in Great tits also in a study by Barrientos et al. (2016). Similarly, an experiment on Collared flycatchers showed that HA and the resulting size hierarchy proved to be detrimental for the growth of nestlings in both good and bad years (Szöllősi et al. 2007). However, the opposite results were also published. For example, HA was not correlated with nestling quality both in terms of body mass and immune response (Martínez-Padilla and Viñuela 2011). Mainwaring et al. (2012) found no difference in the survival of the Zebra finch nestlings from asynchronous and synchronous broods, however the study was conducted on domesticated population with unlimited access to food. Surprisingly, a similar study on domesticated Zebra finches demonstrated a lower survival among synchronized broods (Rutkowska and Cichoń 2005), supporting The Offspring Quality Assurance Hypothesis. Apart from these discrepancies, studies on captive birds not necessarily reflect principles operating under natural circumstances. For example, a study that compared hatching patterns in domesticated, captive wild and wild free-living populations of the Zebra finch revealed that the latter started incubation after clutch completion and hatched eggs synchronously, while the former ones initiated incubation after laying the first egg, which resulted in asynchronous hatching (Gilby et al. 2013). This leaves the question open whether our knowledge on HA gained on captive populations is not misleading. A 6-year long study on a wild population of the Yellow-headed Blackbird Xanthocephalus xanthocephalus B. (Forbes et al. 2002) found that in the most stringent year the average mass of nestlings in synchronized broods was lower than that of both core and marginal nestlings in asynchronous broods. In a good year, the average mass of nestlings in synchronized broods fell between that of core and marginal nestlings in asynchronous broods. Thus, synchrony resulted in a sacrifice of offspring quality under the conditions of ecological stringency, which was in line with the hypothesis. Above examples emphasize the difficulties with drawing any coherent conclusions.
11. The Tasty Chick Hypothesis (Christe et al. 1998) proposes that ectoparasites could play a role in promoting the establishment and maintenance of a nestling size hierarchy within broods. If the immune system of the runts is poorly developed then parasites may favour such chicks thus reducing the parasitic pressure on other siblings in the brood. It was tested on Corsican Blue Tits Cyanistes caeruleus ogliastrae H. but the results were not consistent with the hypothesis (Descamps et al. 2002). The authors found that that larvae of Protocalliphora spp. F. did not aggregate on a single chick and also the experimental removal of the runts had no adverse effect on the condition of the remaining nest-mates. Quite the opposite, nestlings from reduced broods without the smallest chicks did better than those from control nests when exposed to parasitic pressure. However, another study on the same population of Blue tits brought the opposite results (Simon et al. 2003). The authors in an elegant experiment with radio-labelled inulin monitored nestling blood turnover and they found that the blood loss attributed to Protocalliphora parasitism was significantly correlated with chick mass and size rank within the brood, with the smallest and lowest-ranking chicks losing more blood than their older siblings. Although the study was in line with the tested hypothesis, the last-hatched chicks are mostly the weakest ones and, consequently, the outcomes of the study cannot be interpreted as supporting the hypothesis.
Another way of testing whether runts are more prone to parasite attacks is to sanitize experimental nests and check which nestlings benefit from the treatment compared to control parasited nests. Such an experiment, conducted on the Mountain Bluebird Sialia currucoides B., revealed that parasite removal disproportionately benefited middle-ranked nestlings, which was not in line with the hypothesis (O'Brien and Dawson 2009). The authors also reported that, contrary to some other studies, last-hatched nestlings in asynchronously hatching broods were no less immunocompetent than their older nest-mates.
The Tasty Chick Hypothesis relies on the fundamental assumption that the last-hatched chicks have an impaired ability to resist parasites because of their poorer immunocompetence. This was directly tested in a study by Roulin et al. (2003) using four different species. The authors found that, in line with the prediction, the junior chicks in broods of the Barn Owl Tyto alba S. showed a lower humoral immune response than their older siblings, however the cellmediated immune response of senior chicks in broods of the Great Tit was not significantly greater than that of their junior siblings. Similarly, they obtained mixed results for the distribution of parasites between runts and older siblings. In accordance with the hypothesis the fly Carnus hemapterus N. infested last-hatched chicks more than other nest-mates in both the Barn owl and the Kestrel broods but ticks Ixodes ricinus L. were distributed randomly with respect to hatching rank in the Barn owl and Great Tit broods. A random distribution of ticks was found in the Yellow-legged gull (Bosch and Figuerola 1999). Moreover, louse-flies Crataerina melbae R. infested heavily older nestlings but not runts in Alpine Swifts Apus melba L. (Roulin et al. 2003). Valera et al. (2004) experimentally tested the preference of the fly C. hemapterus N. towards the nestling size in European beeeaters by exposing pairs of nestlings of different size to a given number of carnid flies. They found exactly opposite results to what was reported by Roulin et al. (2003) for Barn owls and Kestrels, namely parasites consistently and nonrandomly preferred larger nestlings. Similarly, studies on Starlings Sturnus vulgaris L. (Walter and Hudde 1987;Liker et al. 2001), the American kestrel (Dawson and Bortolotti 1997) and the Barn owl (Roulin 1998) revealed that C. hemaptoperus flies preferentially infested the largest nestlings in the nest. Importantly, these studies were conducted on nestlings prior to feather development. The contrary results obtained by Roulin et al. (2003) can be explained by the fact that in their study the larger nestlings were feathered while the smaller ones were still featherless. As feathers protect nestlings from carnid flies (Kirkpatrick and Colvin 1989;Dawson and Bortolotti 1997;Roulin 1998;Liker et al. 2001), their preference towards younger siblings appears to stem simply from the lack of feathers.
Another set of studies that contradict the proposal of The Tasty Chick Hypothesis demonstrated positive relationship between body condition and parasitic load, thus indicating that parasites prefer good-quality hosts (Blanco et al. 1997;Darolová et al. 1997;Dawson and Bortolotti 1997). This shows that the distribution of ectoparasites within-broods is generally not governed by the rank-related variation in host defence of chicks.

Hatching asynchrony and sibling size hierarchy as a by-product of the selective pressure on incubation schedule
The common ground for the set of hypotheses presented below is that HA itself is not advantageous for birds. It is the early start of the incubation which benefits avian families and HA is an inescapable consequence of such incubation pattern. Alternatively, there are some incubation constraints that result in HA. The premises for above assumptions come mainly from the observations that HA is often detrimental for the youngest siblings and in many cases parents seem to undertake strategies counteracting HA despite early start of incubation. For example, several studies have found that the rate of development of eggs within a clutch is not the same and the last-laid eggs require shorter periods of incubation to hatch than the earlier laid ones (Muck and Nager 2006;Boonstra et al. 2010;Hadfield et al. 2013). This may be due to the increase of maternal yolk androgens with the laying order within a clutch (Lipar et al. 1999;Royle et al. 2001), which in turn enhances the development of the hatching muscle necessary to crack the egg shell (Lipar 2001;Lipar and Ketterson 2000) and leads to earlier hatching (Eising et al. 2001). Another line of evidence that birds may try to minimise the effects of HA comes from the studies that show that in some species females increase egg volume with laying order which may increase competitive abilities of younger nest-mates (Rosival et al. 2005;You et al. 2009). Finally, females are observed to feed the last-hatched nestlings preferentially (Budden and Beissinger 2009;Ryser et al. 2016), which suggest efforts to reduce size hierarchy within siblings. Thus the hypotheses presented hereafter seek for the explanation of why early start of incubation may be beneficial for birds despite possible costs of HA.
1. The Nest Failure Hypothesis (first proposed by Husell 1972 and then extended by Clark and Wilson 1981) suggests that HA is a non-selected consequence of incubation starting before the clutch completion as a response to nest predation. The benefits of doing so may lie in the shortening of nesting stage and thus minimizing predation risk. The early start of incubation to shorten the nesting period would be compatible with other strategies involved by birds under high predation pressure, for example, the evolution of faster nestling growth to minimise exposure time to predators (Bosque and Bosque 1995;Remes and Martin 2002). In some species females enhance development rate of embryos in last eggs by increasing egg volume and/or androgen deposition in laying order (Hillström 1999;Rutkowska and Cichoń 2005;Schwabl 1996). In such a scenario the offspring from the last-laid eggs may hatch only slightly later than their siblings and, generally, the entire incubation period is shortened. Additionally, some studies on species under high predation pressure show that the runts can catch up with the older siblings before the latter ones are able to fledge (Węgrzyn 2012). Altogether, in the above optimal circumstances early incubation and HA may actually increase nest success by shortening the risky nesting period without being detrimental for the youngest nestlings. Some support for the hypothesis comes from a study on American robins, which showed that in the population under a predictably low level of predation, a considerable number of females postpone the start of incubation until some days after clutch completion (Rowe and Weatherhead 2009).
If food is not abundant, the runts may suffer from a delayed development compared to their older nest mates and thus they may not be able to fledge as soon as their sibs. But even in such a scenario HA may be advantageous under high predation pressure, because at least part of the brood may successfully escape the attack. Interestingly, a study on House wrens revealed that in broods with a pronounced size hierarchy at the time of fledging, older nestlings wait for runts if there is no urgent need to leave the nest (Bowers et al. 2013). Thus the part of the brood is ready for emergency fledging in a shortest possible time, but if there is no rush all members of the family benefit from maximum fitness. Wilson (1981, 1985) extended the original idea of HA being a result of predation pressure and they modelled the effects of nest failure assuming that the risk to eggs and chicks may be different. If the nestlings are more prone to predation than the eggs, then synchronous hatching is more productive than asynchronous one. In the opposite scenario HA will maximize reproductive output because asynchrony may lead to the earlier fledging of some young and thus it minimizes overall risk of nest failure (Murray and Bertram 1994). The model predicts that most species should begin incubation before clutch completion. Hussell (1985) modified the above model by assuming that HA depends on nest failure at extreme periods of nesting cycle, i.e. egg-lying and fledging. A high risk of predation during the egg-lying period should result in early incubation and HA. Importantly, all models to date assumed a single nesting attempt per season. The model was updated again by Lee and Lima (2017), this time with the focus on multiple nesting attempts and their effects on expected patterns of HA. The rationale of doing so was in the fact, that numerous bird species attempt two or more nests within a season (Roper 2005). Despite some differences in quantitative and qualitative expectations of HA obtained from the new model, the general idea is that a higher risk at the egg stage favours greater hatching asynchrony, whereas a higher risk at the nestling stage promotes synchrony.
The predictions of the hypothesis cannot be directly tested experimentally, however nest failure rates can be measured for egg and nestling stages and subsequently the HA predicted by a model may be compared to the observed degree of asynchrony. The studies using such approach brought mixed results, for example in Snow Buntings Plectophenax nivalis L. predictions matched the observations (Hussell 1985), while in other species, such as the Boat-tailed Grackles Quiscalus major V., the Least Flycatcher Empidonax minimus B. B. and the Yellow Warbler, HA was smaller than predicted by the model (Bancroft 1985;Briskie and Sealy 1989;Hébert and Sealy 1993b). Another mathematic model proposed by Murray and Bertram (1994) demonstrated that HA results in a higher probability that at least one chick will leave the nest, but at the same time the expected number of fledglings in asynchronous broods is smaller than in synchronous ones. Thus the fitness benefits for parents starting incubation earlier seems equivocal. Additionally, some studies show that there is a considerable intraspecific variation in incubation and hatching pattern, even within the same population breeding in the same site-a number of females hatch their eggs synchronously while the others do the opposite (Eguchi and Yamagishi 2002;Kontiainen et al. 2010). Such findings lends no support to the tested hypothesis because predation pressure is expected to be similar for all females nesting in a given environment.
2. The Hurry-up Hypothesis (Clark and Wilson 1981) argues that HA is a consequence of incubation starting before the clutch completion as a response to the ending of the breeding season. Earlier incubation allows to hatch the brood in somewhat shorter time and thus to avoid a decline in food availability later in the breeding season. If The Hurry-up Hypothesis is correct, then it is predicted that hatch spreads will increase with date of clutch initiation. There is evidence that in Pied Flycatchers hatching spread increased with time of season (Slagsvold 1986) resulting in more asynchronous broods as the breeding season progresses. Similar findings were reported in other studies (Mead and Morton 1985;Hébert and Sealy 1993a, b;Hébert and McNeil 1999;Both and Visser 2001;Kontiainen et al. 2010;Barrientos et al. 2016). Some studies also reported that asynchronous broods were more productive later within seasons (Bowers et al. 2016). On the other hand, Bryant (1978) documented a seasonal decline in the food supply of the House Martin Delichon urbica L., but found no correlation with the hatch spread. A study on Tree Swallows Tachycineta bicolor V. reported that later-laying individuals did not show patterns of incubation onset different from early laying individuals (Ardia et al. 2006). Other authors investigated the species that lay multiple clutches during a breeding season and they found no variation in the degree of asynchrony (Beissinger and Waltman 1991;Waltman and Beissinger 1992;Ellis et al. 2001). Interestingly, a study on Blue tits showed that females of this species have evolved plastic responses to environmental cues-they start incubation prior clutch completion towards the end of breeding season resulting in asynchronous broods, yet they may delay incubation until after clutch completion in wait for caterpillars at the beginning of the season, which leads to synchronous hatching (Stenning 2008;García-Navas and Sanz 2011). In sum, it seems that The Hurry-up Hypothesis explains the HA in some species, while it does not in others.
3. The Hormonal Hypothesis (Mead and Morton 1985) explains an early onset of incubation, resulting in HA as a consequence of the fact that the end of ovulation and the start of incubation may be controlled by the same hormone, namely prolactin. Because ovulation precedes the laying of an egg by one day, the hormone responsible for its termination should be released at the laying of a penultimate egg. If the same hormone triggers the incubation, it should occur before clutch completion. Under this constraint the synchrony could only evolve if its selective advantage outweighed the costs of separate hormones. The weakness of the above hypothesis lies in the fact that (i) much intraspecific variation in the onset of incubation can be observed (Harper et al. 1994) and (ii) early incubation does not always fall on the penultimate egg (Magrath 1990). Even within the same species hatch spreads vary with clutch size, which is inconsistent with predictions (Hébert and Sealy 1992). Another issue is that the hormonal mechanisms of bird reproduction are not yet understood in detail. For example, it is unknown whether prolactin is the cause or consequence of the incubation activity (Balthazart 1983). However, The Hormonal Hypothesis may also be considered from another perspective. Assuming that the release of a hormone needed to form a brood patch is an external stimulus, for example provided by a mate (Drent 1975), the incubation activity of a female may arise gradually during laying. In such a scenario, it does not have to happen on the penultimate egg, despite the hormonal mechanism involved. To the best of our knowledge, this hypothesis has not been yet tested. (Arnold et al. 1987) suggests that HA is the result of a decline in the viability of the first laid eggs if they are not incubated within a few days following their laying. Females may face a trade-off between maximization of embrio survival and growth and survival of late hatched offspring (Arnold 2011). Therefore, to increase the probability of hatching of all eggs in a clutch, females are forced to start incubation before clutch completion. In unincubated eggs embryos are most susceptible to exposure to high and moderate temperatures, i.e. above physiological zero but below optimal development temperatures (Webb 1987), however they remain viable at low temperatures (O'Connor 1984). Below physiological zero (25 °C) no development occurs and the eggs can remain viable for an extended period of time (White and Kinney 1974;Wilson 1991;Decuypere and Michels 1992). Optimal development occurs at incubation temperatures, typically 36-38 °C (Webb 1987;Rahn 1991). Exposure to temperatures above physiological zero, but below normal incubation temperature, can cause abnormal development and embryo mortality (Wilson 1991;Deeming and Ferguson 1991). As a result, cool ambient temperatures may facilitate the onset of incubation once the clutch complete, while in warm climates birds may be constrained to commence incubation during laying to maintain egg viability. Thus an advanced onset of incubation (and thus increased HA) may be expected when laying occurs at 1 3 high ambient temperatures. This assumption is supported by studies showing that HA is correlated with latitude in passerines (Clark and Wilson 1981;Hussell 1985). Another line of evidence is brought by the observations that the larger clutch size accelerated the timing of full incubation (Wang and Beissinger 2009). Experimental studies on the Black kite confirmed that delayed onset of incubation reduced the viability of first-laid eggs, particularly when the ambient temperature during the laying period was high (Viñuela 2000). Similar findings were reported in other studies for House Sparrows Passer domesticus L. (Veiga 1992;Veiga and Viñuela 1993) and Green-rumped Parrotlets (Stoleson and Beissinger 1999). Ardia et al. (2006) compared the onset of incubation in the Tree Swallow at two locations at the extremes of their breeding range, Alaska and Tennessee. In the latter location more females initiated incubation prior to clutch completion than did females in the former one. In both sites the early onset of incubation leading to HA was predicted by the proportion of daily temperatures above physiological zero during laying. Additionally, female Tree swallows increase the amount of time spent incubating eggs during the laying stage in response to experimental heating of nest boxes (Ardia et al. 2009). Similarly, the HA was affected by the mean ambient temperature during laying and the first five incubation days in Treecreepers Certhia familiaris L. (Enemar 1995). At higher temperatures, more incubation occurred during the laying period, resulting in a prolonged hatching spread. Aldredge et al. (2014) reported that Florida Scrub Jays Aphelocoma coerulescens B. began incubation earlier relative to clutch completion when laying larger clutches or when ambient temperatures increased. An experimental approach revealed that in Blue tits heating of nest-boxes during the laying phase resulted in increased incubation prior to clutch completion, resulting in a greater HA (Vedder 2012).

The Egg Viability Hypothesis
The Egg Viability Hypothesis is also supported by another line of evidence. Some studies have demonstrated that incubation during the laying period reduces the risk of microbial infection in early-laid eggs (Cook et al. 2005a, b). Using the Domestic Chicken Gallus gallus domesticus L. eggs exposed in tropics, Cook et al. (2003) determined that the trans-shell microbial infection of egg contents occurred within 3 days of exposure, significantly reducing egg viability. The subsequent experimental design for the potential for microbial infection involved two experimental groups (cleaned and uncleaned) and a control group (unmanipulated) of newly laid Pearly-eyed Thrasher Margarops fuscatus V. eggs (Cook et al. 2005b). The experimental eggs were placed for five days in two sites: cool and hot, while the control eggs were left in the original nests under parental care. Experimental eggs became infected, in contrast to the eggs incubated by parents, suggesting that incubation limits infection. Exposure to infection and high ambient temperatures resulted in a dramatic reduction in egg viability. The study provided the first evidence that microbes can infect unincubated eggs of a wild bird, and that infection and ambient temperature act independently to reduce hatching success. A similar experimental approach was conducted in a temperate environment (Wang et al. 2011). This time the study revealed that eggshell microbial loads did not increase with exposure to ambient conditions, were not reduced by disinfection with alcohol, and were unaffected by parental incubation. In contrast to studies performed in the tropics, the authors found little evidence that the temperature or microbial mechanisms of egg viability decline were important at temperate-zone. The conclusion of the study was that delaying the onset of incubation until the penultimate or last egg of a clutch in a temperate zone may maintain hatching synchrony without a large trade-off in egg viability. The study was conducted on three species: the Western Bluebird Sialia Mexicana S., the Tree swallow, and the Violet-green Swallow Tachycineta thalassina S. Interestingly, in the American robin about 20% of females delay the onset of incubation up to four days beyond clutch completion without any fitness costs associated with such a strategy (Rowe and Weatherhead 2009). Thus, egg viability appears to be somewhat independent of the incubation pattern, at least in certain species. (Bollinger et al. 1990) explains the early incubation of the clutch resulting in HA in terms of protection of eggs from conspecifics or predators that destroy single eggs within a clutch in unattended nests. For example, egg predation by conspecifics is common among gulls (Parsons 1976), while Ruddy Turnstones Arenaria interpres L. are important predators of tern eggs (Brearey and Hildén 1985;Morris and Wiggins 1986). Ruddy turnstones pecked Sooty Tern Sterna fuscata L. and Common Tern Sterna hirundo L. eggs in unattended nests (Crossin and Huber 1970;Bollinger et al. 1990) and Arctic Tern Sterna paradisaea P. eggs were usually lost only when adults were not present at their nests (Skipnes 1983). This suggests that incubation appears to prevent egg predation. Consequently, Common terns were reported to initiate incubation after chasing Ruddy Turnstone from previously unincubated clutches (Thompson and Raveling 1987;Bollinger et al. 1990). Incubation may reduce predation when parents do not perceive another species as a potential egg predator, which seems to be the case with terns and Ruddy Turnstones (Parkes 1971). Bollinger et al. (1990) reported that in Common Terns egg predation coincided with a short period in which turnstones were present in the colony and within this time egg predation decreased with increased constancy of incubation. Interestingly, in other colonially nesting birds, predation of the first-laid eggs exceeded that of eggs laid later (Parsons 1976;Verbeek 1988), supporting the hypothesis that consistent incubation helps to protect eggs from predators.

The Egg Protection Hypothesis
Alternatively, some bird species breeding in extreme environments may need to protect their eggs from detrimental weather conditions, which leads to the initiation of incubation after the first egg is laid. Such a situation takes place, for example, in the Gentoo Penguin Pygoscelis papua F. (Williams and Croxall 1991). Interestingly, females counteract a size hierarchy between siblings by producing the second of the two eggs larger than the first one.
6. The Brood Parasitism Hypothesis (Stoleson and Beissinger 1995) assumes that parents may compromise the survival of last-hatched nestlings by initiating incubation early to prevent their clutch from brood parasitism and to minimise the probability of rising young they did not sire. The early start of incubation, just after laying the first egg, may also be beneficial in parasited broods because own young will hatch first and be competitively superior (Ricklefs 1993). The hypothesis assumes that unincubated nests are more prone to receive parasitic eggs than incubated ones, thus the ideal experimental test would involve comparison of brood parasitism between nests protected by incubating parents and those with the eggs left alone. Such approach was made by Beissinger et al. (1998) but they did not detect parasitic eggs in experimental nests of the Green-rumped Parrotlet. Since then the hypothesis has not been tested in other species practicing intraspecific brood parasitism. As it comes to interspecific brood parasitism it seems that the early incubation may not prevent the host from receiving a parasitic egg. A study by Fiorini et al. (2009) revealed that the Shiny Cowbird Molothrus bonariensis G. parasitism is effective when females lay they eggs both during host egg laying and early incubation. The observational findings were confirmed by authors in experimental approach. The above line of evidence suggests that HA is an unlikely byproduct of early incubation used as protection against brood parasitism.
7. The Limited Breeding Opportunities Hypothesis (Beissinger and Waltman 1991) seems an extension of The Egg Protection Hypothesis, however it explains HA in hole-nesting bird species that do not defend territories. For these species, consistent occupancy of a cavity may be the most effective way for a single individual to defend a cavity against intruding pairs. This strategy is employed by Green-rumped Parrotlets. The female constantly occupies the nest cavity and starts the incubation with lying the first egg (Waltman and Beissinger 1992). The onset of incubation before clutch completion may be favoured under limited breeding opportunities because early incubation would prevent the nest site and eggs from intrusion of other pairs through occupancy. Some support for the above assumption comes from the authors' observations that invading pairs indeed destroyed unguarded eggs and young. Intense competition among potential breeding pairs for limited nesting sites, leading to destruction of eggs followed by a cavity takeover, was also observed in other species (Johnson and Kermott 1990;Pribil and Picman 1991). Thus early incubation, resulting in HA, may be the best way to protect own eggs from competitors for a nest site. As the proportion of breeders to non-breeders, or nest site availability, is often not easy to estimate, the above hypothesis has not been tested to-date by any researchers.
8. The Adult Predation Hypothesis (Magrath 1988) assumes that early onset of incubation resulting in HA allows to shorten nesting period and thus reduces the time when adults are more exposed to predation. However, the danger during incubation, but not nestling period, favours synchronous hatching (i.e. delayed incubation), even when other factors would favour asynchrony (Hussell 1972). Magrath (1988) modelled the optimal onset of incubation by incorporating a nest content components and a female survival component and he found that the optimal degree of HA involved a trade-off between current and future reproduction, and varied considerably with survival probabilities during incubation and between breeding attempts. The hypothesis and the model was first tested by Stoleson and Beissinger (2001) using the field data on the breeding attempts and predation on adult females in a population of the Green-rumped Parrotlets. The results obtained in the study confirmed the hypothesis. As female survival probabilities during incubation and between breeding attempts are difficult to obtain, to our knowledge no other study aimed at testing the effect of adult predation on HA.
9. The Egg Position Hypothesis (Diez-Méndez et al. 2020) holds that HA may arise if eggs are not repositioned and they remain in the same locations, either centrally or peripherally. A premise for such a prediction comes from the observation that during incubation Great tit eggs located in the center of the nest receive more heat (Boulton and Cassey 2012), and thus they may develop faster than peripheral ones. Such observations are not uncommon, for example, in Chipping Sparrows Spizella passerine B. the difference in temperatures between eggs in the same clutch may reach 4.6 °C (Caldwell and Cornwell 1975) and in mallards even 12.2 °C (Huggins 1941). Moreover, it was demonstrated that even slight differences in incubation temperature may have significant effects on offspring development and quality (DuRant et al. 2013). Consequently, the hatching span would be greater in the less repositioned clutches. This may be particularly the case for species that lay many eggs, as a female brood patch is too small to cover the entire clutch. Such a situation may result in a temperature gradient in the clutch from the warmest eggs located in the center to the coldest ones placed peripherally (Boulton and Cassey 2012). The female may counteract unequal incubation by moving the eggs between the center and the edge so that all of them would acquire the same amount of total heat. Alternatively, if repositioning of eggs does not take place, the offspring from the inner eggs would hatch earlier as the result of better incubation. The hypothesis has been formulated quite recently and it has only been tested by its authors. Their study on Great Tits found that females regularly moved eggs within the clutch and that there was no relation between egg relocation and the HA. Thus, the results did not support the hypothesis. However, the results obtained by Poláček et al. (2017) demonstrated that females did not relocate all eggs equally which supports The Egg Position Hypothesis despite the fact that it was not directly tested in this study.

Discussion
Despite numerous studies on this intriguing pattern of breeding, there seems to be no agreement on the one, ubiquitous explanation of HA. Quite the opposite-the more research is done the more multidimensional this phenomenon appears (Table 1). Only two hypotheses, namely The Insurance Hypothesis and The Egg viability Hypothesis, seem to have been confirmed almost unanimously. Similarly, merely The Hormonal Hypothesis appears definitely rejected, although few authors have sought to test it.
Noticeably, the hypotheses reviewed here are not mutually exclusive and studies testing them suggest that there may be several selective pressures operating within and among species. It seems that striving at all costs to formulate a single hypothesis explaining every aspect of HA falls under the Mencken's remark that for every complex problem there is an answer which is clear, simple and wrong. HA should be viewed as an example of such a complex phenomenon of many possible origins and outcomes. For example, HA may originate from the early onset of incubation to protect the viability of the first laid eggs and/or due to hormonal state of the laying female and later it may facilitate brood reduction under food scarcity or adverse weather conditions preventing parents from provisioning. Simultaneously, due to the shortening of the nesting stage, HA in the above sample-nest may reduce the time of exposure of immobile nestlings to predators and/or help to avoid a decline in food availability if the breeding attempt occurs late in the season. In the same nest HA may decrease sibling rivalry by producing a size hierarchy and for the same reason it may reduce the peak load during the most energetically demanding nestling stage and thus prevent parents from overwork and/or it may enable parents to bring a greater variety of food to different aged nestlings. Whether any or all of the above scenarios happen probably depends on a number of factors, such as a species, a year, a temperature, food availability, the type of territory, parental condition, as well as the magnitude and type of predation pressure, just to name a few. Taking into account the interaction between above-mentioned factors we may expect a numberless environmental and behavioural scenarios. Not surprisingly, this review has presented some studies that brought the opposite results on the causes or consequences of HA even when studying the same species. Thus we infer that HA is such a complex phenomenon that a sole elucidation is unlikely to be true. Rather, we should accept multiple causes and consequences leading to the breeding strategy characterised by unequal chances of siblings to survive. It is not the single explanation but our in-depth understanding of the intricacy of HA that we should try to achieve by formulating and testing numerous hypotheses. A good illustration of a multitude of selective forces, that favour the evolution of hatching asynchrony and may vary between years and environments, is a long-term study on Common terns revealing that in a single species HA may serve the brood reduction or insurance depending on resource abundance (Vedder et al. 2019).
Our paper also reveals the blind spots in the reviewed area-9 hypotheses still lack any experimental approach that would confirm or reject the assumptions (Table 1). For example, The Ice box Hypothesis might be tested by removing dead runts from the experimental nests and comparing the development of core nestlings from the experimental and control nests, from which dead runs were not removed. The Dietary Diversity Hypothesis might be tested by creating experimental synchronous and asynchronous nests of the same species and investigating whether parents rearing asynchronous broods in deed bring more diverse food items and whether it affects brood survival and/or parental condition. An experimental verification of The Hurry up Hypothesis would be possible by creating synchronous and asynchronous broods late in the season and estimating breeding success and/or nestling condition in both types of nests. In turn the role of egg predation as an evolutionary force promoting early incubation (The Egg Protection Hypothesis) can be experimentally studied through creating unattended nests in which subsequent eggs are deposited daily as in the case of synchronous broods and comparing the predation level in such nests and control asynchronous nests where incubation started after the laying of the first egg. Finally, the effect of egg repositioning during incubation on HA (The Egg Position Hypothesis) might be investigated by artificial incubation of eggs with a different degree of egg relocation between nests and comparing the degree of egg relocation and HA. Thus we encourage the investigators to focus on above matters and deliver more experimental tests, as observational studies, although very important, cannot bring definite answers about the functions of asynchronous hatching in bird reproductive strategies. On the other hand, it appears that The Hormonal Hypothesis, The Limited Breeding Opportunity Hypothesis and The Adult Predation Hypothesis cannot be tested experimentally. Additional   Waltman and Beissinger (1992) Hormonal Hypothesis (B) Harper et al. (1994) Hébert and Sealy (1992) Magrath ( 1 3 observational studies would be the only way to look closer at these ideas, and they would be of great value, as to-date only single attempts have been undertaken to investigate these hypotheses.

Conclusions
1. There is no agreement on the one, ubiquitous explanation of HA. Extensive research proves that this phenomenon appears multidimensional. 2. Numerous hypotheses on HA are not mutually exclusive and studies testing them suggest that there may be several selective pressures simultaneously operating both within and among species, depending on ecological and physiological circumstances. 3. The interaction between variety of factors may lead to a numberless environmental and behavioural scenarios under which HA may or may not be advantageous, even within the same species or population. 4. Our in-depth understanding of the intricacy of HA should be achieved by formulating and testing numerous hypotheses without playing favourites among them. 5. Considering the blind spots in the reviewed area there is still need for more research.
Data availability All data supporting the findings of this study are available within the paper. This article has no additional data.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.