Skeletal variation in bird domestication: limb proportions and sternum in chicken, with comparisons to mallard ducks and Muscovy ducks

Introduction

Through domestication, much morphological diversity has been generated (Darwin, 1868; Herre & Röhrs, 1990). Domesticated animals exhibit phenotypic changes compared to their wild counterparts (Clutton-Brock, 1999). This is the case in domestic birds, as shown in recent works on the skulls of pigeons (Columba livia) and chickens (Gallus gallus) (Young et al., 2017; Stange et al., 2018) and the integument in chickens (Núñez León et al., 2019). In contrast, postcranial anatomy, including limb bones, has rarely been dealt with analytically and globally in domesticated species (see Wayne, 1986 for dogs; Van Grouw, 2018 for an overview for many species), despite having been widely studied in bird evolutionary biology (Middleton & Gatesy, 2000; Dyke & Nudds, 2009; Nudds et al., 2013). Darwin (1868) mentioned that changes in limb length (or lack thereof) in domestic birds can be observed. Especially notable is the reduction of wing length of the mallard duck (Anas plathyrhynchos) due to domestication (Darwin, 1868).

Limb proportions can be of importance in biomechanical processes such as flight ability. Middleton & Gatesy (2000) highlighted the importance of the humerus relative proportion in the forelimb, where flightless birds have relatively longer humeri while flight manoeuvrability is being linked to shorter humeri. The selective pressure for flight is thus expected to influence the relative proportion of the different bones in the forelimb (Mason, 1984). While the wild type red junglefowl (RJF) is generally considered a short distance flier and a non-migratory bird (Bird Life International, 2016), flight capability in its domestic counterpart, the chicken, is breed-dependent (Schippers, Simons & Kippen, 2013). Certain chicken breeds are known to have only limited flight capacity or even completely lost their flight capacity (e.g., ukokkei and Polish breeds) and this is largely attributed to the unfitness of their plumage for such task (Ekarius, 2007; León et al., 2021). Differences in flight capacity have also been reported depending on the weight of the breed: the heavier, the less prone to sustained flight (Darwin, 1868; Schippers, Simons & Kippen, 2013).

It is generally agreed that domestic mallard ducks have lost their flight capacity when compared to their wild counterpart, which in contrast to the RJF, is a migratory bird (Accordi & Barcellos, 2006; Bird Life International, 2016). This change due to domestication has been attributed to a gain in weight and reduction in the proportional wing length (Darwin, 1868; Cnotka, 2006). Muscovy ducks (Cairina moschata) are non-migratory birds (Bird Life International, 2018), but even the domestic forms retain their flight ability (Swatland, 1980). Unlike the case of differences in wing length between domestic and wild mallard ducks, finding osteological differences between domestic and wild Muscovy ducks has proven more difficult (Angulo, 1998; Stahl, Muse & Delgado-Espinoza, 2006).

Another skeletal element of biomechanical importance in birds is the sternum, the site of attachment of the pectoralis muscle, important for producing mechanical work during downstroke and pronating the wing, and the supracoracoideus muscle, which elevates and supinates the wing during upstroke (Biewener, 2011). The sternum also fulfils an important role in respiration as it is the site of attachment of m. obliquus abdominis externus and m. rectus abdominis (Lowi-Merri et al., 2021).

The sternum varies in shape and proportions and is a source of intensive study in poultry research concerned with welfare due to widespread reports of damage ranging between 30 and 100% of commercial hens (Buckner et al., 1949; Fleming et al., 2004; Harlander-Matauschek et al. , 2015; Toscano et al., 2020). Keel damage may include crookedness or bending, as well as fractures (Casey-Trott et al., 2015) and has been observed to occur in different proportions between breeds—potentially also related to wing loading—and between sexes (Kittelsen et al., 2021; Stratmann et al., 2016). Deformations of the keel can be divided into two types: fractures, defined as sharp bends, fragmented sections or shearing of the keel with or without thickened bone (callus) (Fleming et al., 2004; Eusemann et al., 2018; Tracy et al., 2019; Thøfner et al., 2020); and deviations, defined as abnormally shaped bone containing section(s) that vary from an ideally perfect two-dimensional straight plane that has not resulted from fracture (Casey-Trott et al., 2015). Pathological characterization of keel bone fractures in laying hens does not support external trauma as the underlying cause (Thøfner et al., 2020). Therefore, here we use the term ‘crookedness’ to denote exclusively the deviations, not fractures, of the keel, as defined in Casey-Trott et al. (2015). The problem appears somewhat historical as it was referred to by Darwin (1868, p. 282) as ‘generally so much deformed that it is scarcely possible to compare its form strictly in the several breeds’. The crookedness of the keel has been largely studied in the context of its causes, its relation with animal welfare, and its effect on productivity (Blount, 1933; Warren, 1937; Waters, 1949; Buckner et al., 1949; Fleming et al., 2004; Eusemann et al., 2018; Thøfner et al., 2020) and the market value of the domestic chicken (Hyre, 1995). Keel crookedness has been reported to have environmental and genetic origins, appearing mostly when chickens are given the opportunity to perch (Blount, 1933; Warren, 1937; Waters, 1949; Buckner et al., 1949; EFSA, 2015; Gebhardt-Henrich, Toscano & Würbel, 2017). However, whether crookedness of the keel and other alterations of sternum characteristics typically found in domestic chicken are associated with domestication in its initial, less intense phase (Vigne, 2011), or if it appeared only with breed formation and improvement has not been determined (Blount, 1933; Warren, 1937; Waters, 1949; Hyre, 1995; Fleming et al., 2004; Kittelsen et al., 2021). We would expect that wild forms exhibit less deformed and more compact keel bones, i.e., straight and non-notched carina sterni and smooth foramen pneumaticum and pars cardiaca, related with more pronounced use and therefore ossification of the bone. Deformation and increased sponginess, on the other hand, might be related to body size changes due to domestication and breed formation.

In this study, we analysed the limb proportions of three domestic bird species: chicken (G. gallus), mallard duck (A. plathyrhynchos), and Muscovy duck (C. moschata) in order to determine whether there have been changes due to domestication and breed formation. We examined this matter in the context of evolvability, allometry, heterochrony, and selective pressures (artificial selection) that affect flight capacity. In the case of chickens, we examined these changes also among a diversity of breeds. Further, we investigated variation in the sternum by coding discrete anatomical features that characterize its variation between wild (RJF) and domestic chicken, as well as among domestic chicken breeds. Given the previous studies on differences between wild and domestic animals and the importance in locomotion of forelimb and hind limb proportions and sternum characteristics, we aimed to determine whether differences exist between wild and domestic forms of three bird species and to shed light on the possible underlying patterns.

Materials & Methods

Limb bone measurements and sternum character states

We took measurements of the length of the humerus, radius and carpometacarpus for the forelimb and length of the femur, tibiotarsus and tarsometatarsus for the hind limb. For each bone, we measured the length as the maximum length of the bone parallel to the measuring device (Fig. 1), using a digital caliper (Electronix Express 0604CAL6++ LCD Digital Calipers) to the nearest tenth of a mm. All measurements were conducted by one observer (CMHC).

The total length of the forelimb was here defined as the sum of the lengths of the carpometacarpus, radius and humerus, while the total length of the hind limb was defined as the sum of the lengths of the tarsometatarsus, tibiotarsus and femur (Fig. 2). Although the thus estimated limb length does not equal ‘functional limb length’ because cartilage, soft tissue, and many autopodial elements were not considered, this estimate serves as an approximation (e.g., Zeffer, Johansson & Marmebro, 2003). We considered the proportion of a bone to be the length of that bone with respect to the total length of the limb. This ratio was calculated by dividing the length of the bone by the total length of the respective limb.

For the analysis of feature variations in the sternum in chicken, we accounted for five characters (Fig. 3): (1) carina sterni is defined as ‘straight’ when the direction of growth is maintained in a straight line, as opposed to ‘crooked’, in which the growth is tortuous with one or multiple pronounced torsions; (2) carina sterni is defined as ‘notched’ when the direction of the lower margin of the carina in lateral view changes direction, creating a notch (it is defined as ‘non notched’ if this does not happen); (3) pars cardiaca and foramen pneumaticum are described as ‘spongy’ or ‘smooth’ depending on the type of external appearance, either porous or flat, respectively; (4) the caudal end of the trabecula mediana of the sternum is differentiated between ‘fanned’ (end wider than the rest of the trabecula in dorsal view) and ‘straight’ (end as wide as the rest of the trabecula in dorsal view).

Results

The pairwise comparisons of limb proportions in wild vs. domestic chicken, mallard duck, and Muscovy duck showed that significant differences in limb proportions only occured in the mallard duck (Fig. 4, Table 2). Domestic mallard ducks, compared to their wild counterpart, had proportionally longer humeri and shorter radii and carpometacarpi in the case of the wing, and proportionally longer tarsometatarsi and shorter femora in the case of the hind limb (Fig. 4, Table 2).

Pairwise comparisons of total limb length—as a proxy for body size—between wild and domestic forms in each species showed that the domestic varieties are larger than the wild ones in the mallard duck and the chicken, while we found no significant differences in the Muscovy duck (Fig. 5). In addition, variation of limb length appeared greater in domestic chicken and mallard ducks, compared to their wild counterparts (Fig. 5).

The analyses comparing the log transformed length of the limb bones with the log transformed total length of the limb—as a proxy for body size—showed that the lengths of the limb bones were strongly and significantly correlated with body size in all three species (Fig. 6, Table 3). We found different scaling patterns of limb bone lengths and body size (Table 3). Similar analyses featuring limb bone length vs. body size in ontogenetic samples extracted from Thomas, Sadler & Cooper (2016) and Faux & Field (2017) showed allometric growth in several of the bones during early chicken ontogeny (Fig. 7, Table 4).

Comparisons of limb proportions among chicken breeds revealed significant differences (Fig. 8, Table 5, File S2). A comparison of the mean total forelimb length of the different chicken breeds, showing their size differences, is provided in Table 6.

Our observations on the characters of the sternum show that for most characters, both character states could be found in wild as well as domestic chicken (Table 7). The only exception was that a spongy foramen pneumaticum and pars cardiaca were never reported in wild RJF (Table 7). Further, frequency of character states between wild and domestic were different regarding the carnina sterni, which tended to be more frequently non-notched in domestic chicken, while it tended to be more frequently notched in wild RJF (Table 7). Crooked carina sterni could be found in both, wild and domestic chicken (Table 7). Among the domestic chicken breeds, none of the character states were found to be exclusive for a particular breed (Table 7). However, we also found a hint to an unequal presence of characteristics among breeds (Table 7). Notably, bantam tended to straight carina sterni and fanned trabecula mediana more so than other breeds, and especially the chabo (Table 7). Further, we found that most breeds, except the chabo tended to exhibit a straight, rather than a crooked carina sterni.

Discussion

Our results show that, unlike wild and domestic forms of chicken and Muscovy ducks, domestic mallard ducks had a set of proportions in the limb bones that was significantly different from the wild form (Fig. 4, Table 2). At the same time, we observed that domestic mallard ducks were markedly larger than their wild relatives (Fig. 5). In cases where body size differences between wild and domestic forms are marked, differences in limb proportions between wild and domestic forms might be attributable to allometric scaling. Limb proportions were found to scale allometrically with body size in several limb bones in mallard ducks (Fig. 6, Table 3). The observed difference of limb proportions in wild and domestic mallard ducks (Fig. 4, Table 2) could thus be interpreted as the result of allometric scaling due to particularly pronounced size differences in the wild vs. the domestic sample. Specifically, in the case of the mallard duck’s forelimb, the humerus scales positively allometrically, the radius isometrically, and the carpometacarpus negatively allometrically, leading to proportionally longer humeri and shorter radii in larger specimens (Figs. 2 and 4, Table 3).

Although no ontogenetic study is available for mallard ducks as a comparison, this same tendency for allometric scaling of the limb bones can be extracted from the data available in the literature regarding embryological and post-hatching growth of chicken (Jackson & Diamond, 1996; Thomas, Sadler & Cooper, 2016; Faux & Field, 2017) (Fig. 7). As the body size of the embryo and hatched birds increases during growth, their limb proportions change in a similar tendency (except tibiotarsus and tarsometatarsus in post-hatching growth, Fig. 7, Table 4) as we observe in differently sized adult chickens (Fig. 6, Table 3). Similar scaling relationships were also reported for wing growth in Larus californicus, which shows a progressive decrease in the humerus length relative to the radius and carpometacarpus (Middleton & Gatesy, 2000). This suggests that the differences in proportions of the limbs among adult mallard ducks (static allometry) might be at least partially related to allometric scaling during development and growth (ontogenetic allometry). Although allometric scaling of limb bones was also observed in chicken and Muscovy ducks (Fig. 6, Table 3), the lack of pronounced size differences between wild and domestic forms in these species (Fig. 5) could explain the observed lack of proportional differences of limb bones in the wild vs. domestic forms in these species (Fig. 4, Table 2).

Different allometric scaling patterns of limbs bones probably have functional implications. For example, it has been suggested that there is an inverse correlation of the proportion of the humerus in the wing with aerial maneuverability in birds (Middleton & Gatesy, 2000). Moreover, flightlessness is reportedly linked to a proportionally long humerus compared to the rest of the long bones of the wing in theropods (Middleton & Gatesy, 2000) and steamer ducks (Tachyeres) (Livezey & Humphrey, 1986).

Experiments on the domestication of the wild mallard duck resulted in bigger body sizes and rapid loss (after three generations) of the ability to fly (Darwin, 1868). Mason (1984) proposed that this loss of flight capacity might be due to the loss of positive selection for flying ability: under the care of humans, mallard ducks are usually granted access to reliable sources of food at specific locations, thus flight becomes less essential for survival. The loss of the capacity to fly may thus be related to the proportion of the limbs, which upon less selection due to lack of necessity to fly may have become significantly different from their wild counterpart. An alternative explanation could be that the selection of the domesticated mallard ducks for specific traits could have pleiotropic effects that lead to a reduction of flight capacity. In certain fancy breeds, limb proportions might even be under directed artificial selection. These hypotheses remain to be tested in future works. Other skeletal, muscular and perhaps neurological changes may also be related to the loss of flight capacity. In the case of the hind limb, we observed in domestic forms of the mallard duck a relative shortening of the femur and elongation of the tarsometatarsus. Gatesy & Middleton (1997) reported an apparent gradient of cursoriality increasing with tarsometatarsal proportion among theropod species with similar tibiotarsus/femur ratios. The pattern we observed in domestic mallard ducks, together with a decrease in flight adaptations (above), is consistent with a more terrestrial, although not necessarily highly cursorial, lifestyle of the domesticated vs. the wild form.

Changes in limb proportions were not observed in the Muscovy duck. Its limb proportions remained indistinguishable from the wild form, as opposed to the mallard duck (Fig. 4, Table 2). Changes in flight habits might not have been similar between the wild and domestic forms of mallard and Muscovy ducks While some populations of wild mallard ducks are migratory (Bellerose & Crompton, 1970; Bird Life International, 2019), their domestic variants rarely fly as much and many breeds lack flight capacity entirely (Smith, Pederson & Kaminski, 1989). Domestic Muscovy ducks, on the other hand, although non-migratory birds (Accordi & Barcellos, 2006), have been reported to be strong flyers (Swatland, 1980), suggesting that their flight habits might not have changed so markedly in domestication as to produce a different morphology. Alternatively, and related to the allometric scaling relationships considered above, it is noteworthy that in contrast to the other two species, limb bones of the forelimb were found to consistently scale isometrically in the Muscovy duck and body sizes were similar in the wild and the domestic groups (Fig. 5, Table 3), suggesting restricted evolvability concerning forelimb proportions in this species. An analogous conclusion was reached when postulating lower evolvability of facial shape in cats vs. dogs when reporting largely isometric vs. allometric growth in their skulls, respectively (Sánchez-Villagra et al., 2017). Further, timing of domestication might influence the differences which we see in the investigated species. While it was estimated that the domestication process has started about 3,500 years before present in the chicken (Peters et al., 2016) and about 2,200 years ago in the mallard duck (Zhang et al., 2018), much less is known about the temporal framework of Muscovy duck domestication. Minimum estimates for the onset of Muscovy duck domestication range between 600–700 years before present (Mason, 1984). Although most mammal domesticates have been estimated to start into this process much earlier (e.g., maybe more than 12,000 years in domestic dogs; Larson & Fuller, 2014), significant limb proportional differences in domestic vs. wild mallard ducks as well as among domestic chicken breeds suggest that the relatively short timeframe of domestication in the investigated bird species has been sufficient for such changes to occur. In contrast, the Muscovy duck’s probably more recent domestication history might be too short a timeframe as to lead to extensive changes of limb proportions.

Although we observed no significant differences in the proportional length of the limb bones in wild vs. domestic chicken (Table 2), examination of separate breeds revealed significant inter-breed differences in terms of size and limb proportions (Table 6, Fig. 8, Table S1). One possible interpretation is for diversification to have taken multiple directions in the case of the limb proportions in domestic birds, but selection has not provoked a unique directionality, maintaining the same mean proportions between domestic and wild forms in all bones, and therefore obscuring the variation that can only be observed when individual breeds are compared. The most notable case of breed-specific specialization is the chabo, also known as Japanese bantam. Chabo is a true bantam breed (i.e., it does not present a large variant) with characteristically short legs (Roberts, 2008). The short leg trait is associated with a dominant allele of gene Cp, which is lethal in homozygosis (Shibuya, Fujio & Kondo, 1972; Shibuya & Kuroda, 1973) and is selected as a distinct characteristic of the breed (Van Grouw, 2018). The Cp gene accelerates differentiation of chondrocytes accompanied by suppression of their proliferation, leading to stunted limb long bone growth (Shibuya, Fujio & Kondo, 1972). Similar phenotypes are also known in some breeds of domestic dogs (e.g., dachshund, corgi) and cats, and represent cases of specialization in breed formation (Rimbault & Ostrander, 2012). As in the case of the mallard duck, the differences in adult limb proportions in chicken might be due to allometric scaling during development and growth (see above). Differences in limb proportions among breeds could thus be related to the different sizes of breeds. This is consistent with the observation of allometric scaling of the bones depending on size in chicken (Fig. 6 and Table 3).

Our observations on the characters of the sternum show that the only feature that appears to be associated with domestication in chicken is sponginess of the foramen pneumaticum and the pars cardiaca (Table 7). Further, the notch in the carina sterni seems to be a rare condition in domestic chicken, while it is the most common condition in the wild form (Table 7). Underlying reasons for both differentiations of character states between wild and domestic may be related to the functionality of the sternum, which is likely reduced in most domesticates compared to the wild state due to a tendency to flightlessness in the first, or to body size changes related to domestication (see above). The RJF develops a crooked keel in half of the observed cases (Table 7). The crookedness of the keel has been shown to be hereditary and influenced by early roosting conditions (Blount, 1933; Warren, 1937; Waters, 1949). Our observations reveal that this is not a feature unique to domestic varieties, as has been previously reported (Kittelsen et al., 2021).

A majority of domestic variants exhibited straight carina sterni, a trait that is even found in all of the studied bantam specimens (Table 7). This is with the notable exception of the chabo, in which half the investigated specimens exhibited a crooked carina sterni (Table 7). A possible explanation might be that the shortness of the legs in the chabo is responsible for a more direct contact with higher pressure on the keel with the perching structure of their environment, as it has been shown to be a determinant factor for its appearance (Blount, 1933; Warren, 1937; Waters, 1949; Pickel, Schrader & Scholz, 2011; Casey-Trott et al., 2015). Behaviour, especially flying and perching, could be an important factor distinguishing breeds but could not be assessed in this study. In summary, these results reveal that the variation of the different characters are not segregated among the breeds.

Conclusions

This study shows how the allometric growth of birds can help us understand the variation in limb proportions of differently sized domestic breeds. Furthermore, we show that significant differences in the proportions of limb bones of wild and domestic mallard ducks exist, whereas no significant differences between wild and domestic forms of chicken and Muscovy ducks were found. However, when individual chicken breeds were analysed, differences between breeds could be observed. We offer possible explanations for these differences such as an extension of the allometric growth of domestic mallard ducks, loss of positive selection for flying ability, and/or pleiotropic effects or evolvability constraints, which could inspire future research. Furthermore, we observed that the condition of crooked keel is not unique to domestic chicken, that some sternal characteristics are more frequent in certain breeds, but that overall there were no keel characteristics that are unique for certain breeds. These findings add new observations to the study of keel characteristics that can be added to the body of knowledge of domestication and poultry welfare. In sum, this study highlights the lack of universal patterns in domestication and breed formation.

Supplemental Information