Morphological variation in brain through domestication of fowl

Abstract

Great variations in the size, shape, color, feather structure and behavior are observed among fowl breeds. Because many types of domestic fowls have been bred for various purposes, they are ideal to assess the relationship between brain morphology and avian biology. However, little is known about changes in brain shape that may have occurred during fowl domestication. We analyzed the brains of red jungle fowl and domestic fowl to clarify differences in the brain shape between these breeds, as well as the shape changes associated with size enlargement using three‐dimensional geometric morphometrics. Principal component and multivariate regression analyses showed that ventrodorsal bending, anteroposterior elongation and width reduction were significantly correlated with brain size. According to the size‐dependent analysis, the red jungle fowl brain has an intermediate shape between the brain of young broilers and that of large domestic fowl and adult broilers. After the size effect is removed, geometric morphometric analyses show that the brain of red jungle fowl is different from that of domestic fowl, with large round cerebral hemispheres. Significant correlations exist between the skull length and brain volume among fowl, while the brain volume relative to the skull length is distinctly larger in red jungle fowl compared with domestic fowl. The distinct brain shape and increased relative brain size of red jungle fowl may be driven by the presence of large, rounded cerebral hemispheres.

Introduction

Recent genetic evidence suggests that green jungle fowl (Gallus varius) and grey jungle fowl (G. sonneratii) may have contributed to the genetic make‐up of domestic chicken (Nishibori et al. 2005; Eriksson et al. 2008; Sawai et al. 2010; Tixier‐Boichard et al. 2011), but domestic fowl are considered to have originated mainly from red jungle fowl (G. gallus) of Southeast Asia (Akishinonomiya et al. 1996; Miao et al. 2013). The main reasons for the domestication of fowl include uses in agricultural or manufactured products, and esthetic reasons or companionship (Mehlhorn & Rehkämper, 2013). Chicken is the most widely used poultry bird, and various breeds have been domesticated for commercial uses as well as non‐commercial applications, including ceremonial and entertainment purposes (Somes, 1988; Smith & Daniel, 2000). Great variations in the size, shape, color, feather structure and behavior are observed among fowl breeds (Hayashi et al. 1982; Schütz et al. 2004; Castañeda et al. 2005; Eriksson et al. 2008; Endo et al. 2012). These morphological or behavioral characteristics have been developed by artificial selection for esthetic or societal/personal needs.

Because many types of fowls have been bred for various demands, domestic fowl are best used to assess the relationship between brain morphology and avian biology. Hence, it is important to observe the main difference between wild fowl and domestic fowl, and to understand the morphological variation of domestic fowl. Domestic fowl are notable because of the variations in their body size, while they also exhibit high variability of brain volume composition (Frahm & Rehkämper, 1998; Rehkämper et al. 2003; Tiemann & Rehkämper, 2012). However, the brain shape of domestic fowl has never been fully investigated, and the quantitative relationships among the brain shape, brain size and breed are poorly understood. The brain shape of wild birds is related to brain size (Marugán‐Lobón & Buscalioni, 2009; Kawabe et al. 2013), and the changing pattern of the brain shape in accordance with the change of brain size is indicated among wild birds (Kawabe et al. 2013). Therefore, brain size is essential for discussions of wild avian brain morphology. In addition, Kawabe et al. (2015) observed the changing pattern of brain shape similar to that of wild birds through the growth of broiler chicken. However, is the brain of domestic fowl, which exhibits considerable difference in size, also regulated mainly by brain size? It is likely that the degree of shape variation of domestic fowl correlates with brain size/growth. Thus, we investigated to what degree the brain shape of domestic fowl is constrained by brain size/growth, and determined the changing pattern of the brain shape of domestic fowls independent of their brain size. The use of closely‐related taxa (i.e. domestic fowl) removes the relationship between brain shape and size independent of the phylogenetic effects, and it contributes to solving the relationship between avian brain morphology and avian biology.

Considering the results from wild birds and growth of chick (Kawabe et al. 2013, 2015), it is possible that the brain shape of domestic fowl is highly influenced by brain size and ontogenetic stages. If this is the case, then it is necessary to exclude allometric shape changes from analyses when comparing brain shapes among breeds or it would lead to misjudgments regarding brain morphology. Hence, in this study, we examined the relationships between brain shape and size among domestic fowl, and compared them by excluding size allometry. Comparing the brain shapes among domestic and wild fowl after removing the effect of size from shape variation may help to clarify the intrinsic differences in brain shape between domestic and wild fowl. Additionally, a major aim of this study was to establish whether the similarities in brain shape among various ontogenetic stages can be recognized after removing the effect of size. Therefore, we prepared a data set of broiler chickens containing various ontogenetic stages to elucidate the effects of ontogenesis on brain shape among fowl.

Materials and methods

Specimens and computed tomography (CT) scanning

Our sample comprised 25 skull specimens of adult domestic fowl from eight breeds and three broiler chicks (Table 1). The specimens are stored at the University Museum, the University of Tokyo.

Table 1.

Species analyzed and measurements sorted by brain volume

Breed	Day	Brain volume (mm3)	Log CS	Skull length (mm)
Red Junglefowl		2741.71	3.78	51.40
		2821.35	3.78	54.32
		2823.78	3.79	51.15
		2893.19	3.79	55.83
		3066.45	3.81	55.71
Silky Fowl
Ornament and egg breed/ancient breed of Asia*		2626.50	3.81	56.67
Kurokashiwa
Ornament breed/native Japanese chicken/dense black plumage/long‐crower†		2686.35	3.80	61.40
		2696.69	3.80	60.50
		2858.79	3.83	68.05
O‐Shamo
Game breed/Japanese chicken/aggressive fighter†		3315.01	3.90	74.92
		3364.48	3.90	81.10
		3774.28	3.94	83.90
Nagoya Cochin
Egg and meat breed/orange buff color/established in Japan† , ‡		3267.97	3.86	73.62
		3305.55	3.87	71.01
		3813.37	3.91	67.33
Lueang Hang Khao
Game breed/Thai native chicken/black plumage§		3349.81	3.88	–
		3471.40	3.90	–
		3656.44	3.90	77.71
Pra Dhu Hang Dum
Game breed/Thai native chicken/black plumage¶		3465.12	3.92	73.96
		3773.16	3.93	–
		3795.98	3.94	–
		3822.10	3.95	–
Broiler	4	1157.00	3.51	31.04
	2	1226.33	3.53	29.54
	8	1549.03	3.61	34.96
	28	2508.00	3.77	51.16
	116	3752.97	3.94	81.70
	94	4349.36	3.99	86.37

CS, centroid size.

Skull length, from the anterior point of the premaxilla to the posterior surface of the occipital condyle.

The sources of data are as follows: *Ekarius (2016); ^†Oka et al. (2015); ^‡Tsudzuki (2003); ^§Molee et al. (2016); ^¶Dorji et al. (2012).

The avian brain essentially fills the endocranial cavity, so it is considered that avian brain endocasts can be used as representations of brains (Jerison, 1973; Iwaniuk & Nelson, 2002). However, in recent years, it has been suggested that the endocast morphology does not necessarily reflect the brain morphology in the early ontogenetic stage of some archosaurs. Therefore, the use of endocast morphology as a proxy of brain morphology in archosaurs must be done with caution.

The samples were subjected to CT scanning at the National Museum of Nature and Science (Ibaraki, Japan) using an X‐ray CT scanner for laboratory animals (Hitachi Aloka Medical, Tokyo, Japan). Acquisition parameters were 50 kV, 0.5 mA, a pixel size of 0.10–0.25 mm and a slice thickness of 0.15–0.20 mm. We subsequently prepared virtual endocasts from acquired CT images using Amira (v 5.3.2, Mercury Computer Systems, San Diego, CA, USA; Fig. 1). Details of the methods used to prepare and examine the brain models were provided by Corfield et al. (2008).

Figure 1.

Representative virtual endocasts in the lateral aspect.

Three‐dimensional (3D) geometric morphometrics

We employed the method described by Kawabe et al. (2013) and digitized the 3D coordinates of 24 homologous landmarks obtained from brain endocasts (Fig. 2; Table 2). Landmark data were collected using Amira, and 3D coordinate data sets were then subjected to generalized least‐squares Procrustes analysis (Rohlf & Slice, 1990) using the morphoj software package (Klingenberg, 2011). During processing, the effects of size, position and orientation were eliminated by translating, rotating and scaling all of the objects to a common reference, so the remaining data only reflected the shape (Procrustes shape coordinates). Information about the absolute size of the specimen was preserved as the centroid size (CS), which was calculated as the square root of the sum of the squared distances of landmarks from their centroids (Bookstein, 1991).

Figure 2.

3D brain landmarks used for shape analysis in the dorsal (upper) and lateral (lower) views of the red jungle fowl brain (description in Table 2).

Table 2.

Landmarks used and anatomical descriptions

Number	Anatomical description
1	Median anterior tip of olfactory bulb
2	Median junction between telencephalon and cerebellum
3	Median dorsal point of foramen magnum
4	Median ventral point of foramen magnum
5	Median junction between hypophysis and mesencephalon
6	Median ventral tip of hypophysis
7	Median junction between optic nerve and hypophysis
8	Median junction between telencephalon and optic nerve
9	Perpendicular at midpoint between landmarks 2 and 3 to dorsal margin of cerebellum in lateral view
10	Perpendicular at midpoint between landmarks 4 and 5 to ventral margin of mesencephalon in lateral view
11, 18	Most anterior tip of telencephalon, right and left
12, 19	Perpendicular at midpoint between landmarks 11 (18) and 2 to dorsal margin of telencephalon in lateral view, right and left
13, 20	Intersection of telencephalon, cerebellum and optic lobe, right and left
14, 21	Most anterior point of optic lobe, right and left
15, 22	Intersection of telencephalon, optic lobe, and diencephalon, right and left
16, 23	Most lateral point of the widest part of telencephalon, right and left
17, 24	Most lateral point of the widest part of optic lobe, right and left

For profiled anatomical brain structures, refer to Fig. 2.

Analyses

The Procrustes shape coordinates were subjected to principal component analysis (PCA) to explore the patterns of the major variations in brain shape among breeds. The scores obtained for the specimens along the PC axes and log CS values were subjected to correlation analysis to examine the effects of brain size on brain shape. We also performed multivariate regressions to compare the statistical relationships between brain shape and log CS. To remove the overall size‐dependent shape changes, another data set was constructed by calculating the residuals from the multivariate regression line and performing PCA and canonical variate analysis (CVA) on the residual data (hereby size‐adjusted data set). This allowed us to investigate size‐independent shape change.

The shape data and CS were regressed to determine whether size alone was responsible for the differences in shape observed along the PC axes. The coefficients obtained from the regression were vectors that represented the correlations between changes in shape and size. Significant differences were tested using permutation tests. P‐values < 0.05 were considered significant.

Canonical variate analysis was conducted using the size‐adjusted shape data to maximize the separation between groups, i.e. breeds, and to evaluate the brain shape differences that most distinctly separated them. Each CV was a linear combination of variables (shape coordinates), weighted to reflect a distinct mode of shape variation. The geometric morphometric approach preserves the intrinsic geometry of the landmark coordinate data, so the variation in shape along a given CV could be visualized as the displacement of points in 3D space, thereby providing an intuitive approach to the visualization of group differences in shape (Kawabe et al. 2014). CVA and PCA were performed using morphoj.

In addition, we investigated the relationship between skull length and brain volume of each domestic fowl and red jungle fowl to assess brain size relative to skull length (hereby relative brain size) using a regression analysis. Skull lengths and brain volumes for individual specimens were averaged for each breed of domestic fowl, while those of red jungle fowl were not. The brain volumes and skull lengths were log‐transformed before the regression analysis. Because of the preservation of specimens, we could only use one specimen from each of Pra Dhu Hang Dum and Lueang Hang Khao for the measurement of skull length. It is assumed that both variables contain some degree of error and that neither variable is independent. Therefore, reduced major axis (RMA) regression was employed (but see Smith, 2009). In addition, RMA seems best in this case because variation in the brain volume and skull length is not expected to be partitioned asymmetrically (Smith, 2009). About 95% prediction limit was used to evaluate a difference of relative brain size between red jungle fowl and domestic fowl.

Results

Original brain shape data

Most explained variance was distributed between the first two dimensions, where PC1 and PC2 explained 45% and 15%, respectively (Fig. 3a). PC1 and PC2 were the only PC axes that accounted for more than 10% of the variance. The variation in shape along the PC1 axis was characterized by ventrodorsal rotation in the posterior part of the brain, especially the cerebellum and optic tectum (Fig. 3a); thus, it was related to the dorsoventral elevation/declination of the cerebellum adjacent to the cerebral hemispheres. Compared with the population that had high scores, the population with low scores on PC1 exhibited a strong V‐shape in lateral view. In addition, a decrease in the PC1 score resulted in anteroposterior elongation, reduction of the telencephalon relative size and decreased width, especially the width of the posterior part. Overall, PC1 primarily distinguished narrow, slender and V‐shaped brains from rounded ones; this degree was not as great as the variance in PC2, but a brain expanded laterally with anteroposterior shortening, and the relative size of the telencephalon increased with the PC2 score (Fig. 3a).

Figure 3.

Results of size‐dependent (a) principal component analysis (PCA) and (b) canonical variate analysis (CVA). Variations in brain shape for each PC and CV score are illustrated by the brain schematics shown outside the graphs.

In the graphs of PC1 and PC2 (Fig. 3a), narrower and longer brains, i.e. adult broilers and other domestic fowl, plot on the lower left (−PC1 and −PC2), whereas rounded brains, i.e. young broilers, plot on the right (+PC1). Red jungle fowl are situated between young and adult fowl on the PC1 axis, and on the upper side of the PC2 axis (Fig. 3a). In particular, the red jungle fowl brain was characterized by a round and relatively large telencephalon and optic lobe. Among domestic fowl, the plots for Kurokashiwa, Nagoya Cochin and Silky Fowl are relatively high in the graph, and their brain configurations were quite similar to each other. O‐Shamo and Lueang Hang Khao had a markedly V‐shaped and elongated brain. Pra Dhu Hang Dum and the adult broiler are plotted on the lower side of the PC2 axis.

There was a significant correlation between PC1 and log CS (Table 3), but no significant correlations with the other PCs (Table 3). Thus, the change in shape along the PC1 axis corresponded strongly with brain size and/or ontogenetic change.

Table 3.

Correlation between PCs from the size‐dependent PCA and CS

	PC1	PC2	PC3
r	−0.88	0.15	−0.11
P	< 0.05	0.4509	0.5909

Canonical variate analysis maximizes the distances among centroids relative to the within‐group variance. CV1 and CV2 accounted for 46% and 24% of the shape variance, respectively (Fig. 3b). The first CV separated red jungle fowl (−CV1) from other domestic fowl (+CV1; Fig. 3b). The latter group had a V‐shaped elongated brain, whereas the former had a rounded shape. CV2 was also related to flexion of the brain at the point between the telencephalon and cerebellum. Red jungle fowl, Pra Dhu Hang Dum and the broiler had high CV2 scores, and O‐shamo and Nagoya Cochin are plotted at the lowest side of the CV2 axis. The plots for Kurokashiwa and Silky Fowl are situated at intermediate locations on the CV2 axis.

Size‐adjusted brain shape data

We performed multivariate regression analysis based on the brain shape data using log CS, which accounted for 36% of total variance in brain shape (Fig. 4). The residual shape data obtained from the multivariate regression analysis, i.e. size‐adjusted shape, were then subjected to PCA and CVA. The first two PCs obtained from the size‐adjusted data set accounted for 26% and 18% of the total variation in shape, respectively (Fig. 5a). As the PC1 score decreased and the PC2 score increased, a brain was more V‐shaped and elongated anteroposteriorly (Fig. 5a), as shown by the PCA based on the original data. However, compared with the PCA based on the original data (Fig. 3a), the degrees of flexion and extension of the brain were relatively small. The red jungle fowl plots on the left lower side of the graph, and has a different brain shape compared with domestic fowl even when size effects were removed. Broiler and Pra Dhu Hang Dum had relatively rectilinear‐shaped brains, and they are plotted on the right of the graph (+PC1). The height of the telencephalon differed little from that of cerebellum in the broiler and Pra Dhu Hang Dum. There was relatively high variation but the other fowl are plotted roughly on the upper left, where they had a strong V‐shaped configuration.

Figure 4.

Results of multivariate regression analysis. A scatter plot showing the regression of brain shape vs. log centroid size (CS). Brain schematics illustrate negative and positive extremes.

Figure 5.

Results of size‐adjusted (a) principal component analysis (PCA) and (b) canonical variate analysis (CVA). Variations in brain shape for each PC and CV score are illustrated by the brain schematics shown outside the graphs.

Size‐adjusted CVs also shows significant group separation. The first CV accounted for 45% of the variance in shape (Fig. 5b). As the CV1 score decreased, the telencephalon height increased and became rounded, while the telencephalon and optic lobe increased in size and width. CVA separated red jungle fowl (−CV1, +CV2) from other domestic fowl (Fig. 5b). Red jungle fowl was also clearly separated from domestic fowl by the size‐adjusted PCA (Fig. 5), where both analyses indicated that it had a large telencephalon and optic tectum (Fig. 5). The broiler is plotted on the lower right, and the brain shape of the broiler was more cylindrical and flat compared with other domestic fowl when the effect of size was eliminated. CV2 was correlated with flexion of the entire brain and it accounted for 36% of the variance. Red jungle fowl and Pra Dhu Hang Dum had a high CV2 score with a round brain configuration. O‐Shamo, Nagoya Cochin and the broiler are located at the lower side for CV2; their brains are strongly V‐shaped. In Kurokashiwa and Lueang Hang Khao, brain shape is intermediate and flexion is not significant.

Kurokashiwa was separated from O‐Shamo, Nagoya Cochin, Pra Dhu Hang Dam and broiler chickens (Fig. 5b). The Kurokashiwa brain was considerably smaller than those of O‐Shamo and Nagoya Cochin, and it superficially resembled those of O‐Shamo and Nagoya Cochin (Figs 1 and 3). However, size‐adjusted analyses showed that cerebral hemispheres were relatively higher in Kurokashiwa than those in O‐Shamo and Nagoya Cochin (Fig. 5b). Similar to the unadjusted CVA, Kurokashiwa was situated almost in the center of the graph (Fig. 5b) close to Silky Fowl. However, like the unadjusted analyses, we could not quantitatively compare the brain shape in Silky Fowl with those of other fowl, although the brains of Silky Fowl and Kurokashiwa were average in shape compared with domestic fowl.

In O‐Shamo and Nagoya Cochin, brains were characterized by strong flexion at the midpoint and the slender hindbrain. In addition, plots for O‐Shamo and Nagoya Cochin were always situated close to each other (Figs 3 and 5) with similar brain shapes.

The brain of Lueang Hang Khao differed from that of Japanese domestic fowl, broiler and red jungle fowl (Figs 1 and 5b). It appeared to be similar to that of Pra Dhu Hang Dum based on the results of the unadjusted analyses (Fig. 3), but the size‐adjusted analyses highlighted the difference between them, where the difference in height between the telencephalon and cerebellum in Lueang Hang Khao was large compared with that in Pra Dhu Hang Dum.

The brain shape in broiler chicks was apparently different from that in adults (Figs 1 and 3), but both brain shapes were clustered well by CVA and the size‐adjusted analyses (Figs 3b and 5).

Bivariate analysis

Significant correlation was observed between the brain volume and skull length among domestic fowl (Fig. 6). Points for red jungle fowl fell above the 95% prediction interval of domestic fowl.

Figure 6.

Scatter plot of the log‐transformed brain volume and skull length of red jungle fowls and other domestic fowls. Reduced major axis (RMA) regression line (solid) and 95% prediction intervals (dashed) drawn for domestic fowls.

Discussion

Size‐dependent changes

In this study, variations in the brain shape in domestic and wild fowl were determined using 3D geometric morphometric methods. We prepared a data set that also contained different ontogenetic stages to elucidate the ontogenetic‐ and size‐dependent changes in brain shape among domestic fowl. PCA and correlation analysis showed that ventrodorsal bending, anteroposterior elongation and a decrease in width are significantly correlated with brain size and/or age (Fig. 3; Table 3). Thus, expansion of the width and shortening of the hind brain yielded a rounder and more upward‐oriented brain among small/young fowl (i.e. juvenile broilers), whereas shrinking and elongation produced a more elongated and anteriorly inclined brain among large/adult fowl (Fig. 3). The brains of red jungle fowl and medium‐sized fowl were intermediate in shape between small/young and large/adult brains (Fig. 3). In addition, the multivariate regression analysis revealed that brain size accounted for 36% of total variance in brain shape of fowls. Thus, the brain shape of domestic fowl is significantly affected by its size (Fig. 4).

Previously, we reported that brain shape changes considerably according to size or age of birds, with rounder brains in small/young birds, but straighter ones in large/adult birds (Kawabe et al. 2013, 2015). The changing shapes determined in various avian species in our previous studies are consistent with those seen in various breeds of domestic fowl. Domestic fowl have been artificially selected for various purposes, thereby resulting in highly variable phenotypes, including feather or body color (Sunde, 1992; Castañeda et al. 2005; Eriksson et al. 2008; Sheppy, 2011), feather morphology (Bartels, 2003), skeletal morphology (Hayashi et al. 1982; Endo et al. 2012; Kudo et al. 2016) and ornaments (Tixier‐Boichard et al. 2011); however, much variation in brain shape can be explained by size.

According to CVA, the red jungle fowl brain was most rounded with the relatively large telencephalon and optic lobe (Figs 1 and 3). Pra Dhu Hang Dum was characterized by a dorsoventrally high cerebellum, slender hind brain and slightly downward olfactory bulb (Figs 1 and 3b). The plots for the broiler chickens were spread along the PC1 axis, but the CVA results clustered them well (Fig. 3b), where they had relatively flat cerebrum hemispheres. We could not perform a quantitative analysis because there was only one sample of Silky Fowl, but it was plotted roughly in the center of the graph (Fig. 3b). The brain of Silky Fowl had an average shape compared with the other fowl analyzed in this study (Figs 1 and 3b). Lueang Hang Khao differed from Kurokashiwa due to its strictly V‐shaped configuration (Figs 1 and 3b). There were no prominent differences between Nagoya Cochin and O‐Shamo according to PCA and CVA, i.e. Nagoya Cochinc and O‐Shamo were very similar based on visual examinations.

Size‐independent changes

Brain size accounted for 36% of the total variance in brain shape among wild and domestic fowl, so the effects of size on brain shape were removed to examine the potential changes in shape in the fowl brain. Changes in shape according to brain size detected by the unadjusted PCA were also determined by the size‐adjusted PCA and CVA (Fig. 5). A V‐shaped brain was caused by ventral declination of the area of the cerebellum adjacent to the cerebral hemispheres (Landmarks 2 and 13/20). Previous studies demonstrated that the shape of the avian brain varied inter‐ and intraspecifically, as well as showing that the anterior part of the cerebellum was changeable even after the size factor was removed (Kawabe et al. 2013, 2015). Considering all of these results, the area of the cerebellum adjacent to the cerebral hemispheres might be the most variable part of the avian brain.

Brain shapes of red jungle fowl were clearly different from that of domestic fowl (Figs 1 and 3), even when the effect of size was eliminated (Fig. 5). Multivariate analyses indicated that red jungle fowl had a relatively large telencephalon and optic tectum (Figs 3 and 5). Thus, the large telencephalon and optic tectum might inherently define the brain shape in red jungle fowl.

Although O‐Shamo and Nagoya Cochin show similar brain shapes (Figs 1, 3 and 5), they are different breeds and they differ markedly in their bodily appearance, including skull morphology (Ino et al. 2008), with one a game breed and the other an egg‐type breed. Their brains were similar in size (Table 1), so it was natural that their brain shapes resembled each other. However, even after the effect of size was removed, their brains were still similar, which indicates that a factor other than size and breed type influenced their shape. O‐Shamo belongs to standing‐type because of their specific upright posture, and its elongated neck is one of its typical external morphological characteristics (Endo et al. 2012). Because neck posture and skull morphology are closely related to each other in Aves (Duijm, 1951), the characteristic appearance of O‐Shamo might have an effect on its brain shape.

The anteroposteriorly elongate brain of Pra Dhu Hang Dum differed remarkably from that of Japanese domestic fowl, broiler and red jungle fowl (Figs 3b and 5), while it was superficially similar to that of Lueang Hang Khao (Figs 1 and 3a). Both breeds are game fowl, and the similarity between them might be due to their battle‐related morphological characteristics, such as upward body trunk and long neck. However, when the size effect was removed, the cerebellar height differed between two Thai game fowl. Variation in cerebellar height was related to cerebral ventral flexion and the foramen magnum‐plane angle. Indeed, because Kulemeyer et al. (2009) suggested a possible link between the position of the foramen magnum and head posture in corvids, these differences in brain shape among Thai game fowl may be due to differences in head posture.

The brain shape of broiler showed considerable change through the growth (Fig. 3). The ontogenetic shape changes in broiler observed in this study are consistent with a previous study (Kawabe et al. 2015). They were, however, clustered well by the size‐adjusted analyses (Figs 3b and 5), which strongly suggests that a certain amount of variation in brain shape among the same breed can be explained by size.

Relative brain size

Points for red jungle fowl were situated above the 95% prediction limit of domestic fowl (Fig. 6). This indicates that brain volume relative to the skull length was distinctly larger in red jungle fowl compared with domestic fowl. Thus, relative reductions in brain size have occurred during the process of domestication in domestic fowl. The size‐dependent PCA showed that the brain shape of red jungle fowl was between that of juvenile broilers and domestic fowl along the PC1 axis, which was significantly correlated with brain size (Fig. 3a). After combining these results with the size‐dependent PCA, we found that although the shape was intermediate between chicks and large domestic fowl along the first PC axis, the brain of red jungle fowl actually had the largest relative size. A reduction in relative brain size is a well‐known phenomenon in domestic animals, including mammals and birds (Sossinka, 1982; Ebinger & Löhmer, 1984; Kruska, 1987, 1988, 1996; Jackson & Diamond, 1996; Jensen, 2006); our results are consistent with previous studies. However, it has not been shown previously that the brain changes shape alongside the enlargement that accompanies the relative reduction in size during domestication.

The size‐dependent and size‐adjusted geometric morphometric analyses demonstrated that the brain of red jungle fowl was distinctly different from those of other domestic fowl due to its large round cerebral hemispheres and optic lobes. The greater relative brain size of red jungle fowl might be due to the enlarged cerebral hemispheres. The cerebral reduction in domestic fowl may be related to the relative size reduction that has occurred during the domestication of fowl.

The comparison of brain morphologies among different breeds of domestic fowl could be possible after removing allometric effect from brain shape variation, and it is expected to help clarifying the relationship between brain morphology of domestic fowl and their biology. Based on the results of this study, we hypothesize that the brains of the dwarf domestic fowl exhibit anteroposterior shortening, lateral expansion and rounding of brain shape. We also hypothesize that the brain shape difference between dwarf and large domestic fowl is similar to that between broiler chick and adult. Testing these hypotheses should provide a more holistic understanding of the effects of domestication on brain morphology in fowl.