1. Introduction
The poultry industry is, in terms of the production of local breeds in a sustainable way, a currently booming sector [
1,
2,
3]. The pressure to meet the growing demands for poultry products (meat and eggs) has led to the implementation and improvement of new techniques and work tools to improve productivity, such as the sexing of chickens in farms [
4], as early in the life of the individuals as possible. This allows sex segregation and the withdrawal of those individuals without zootechnical interest from the production cycle (e.g., males in egg layer farms), with the consequent economic and logistical benefit that this implies for the poultry industry.
Apart from the economic interests from food related industries, the knowledge of the relative distribution of sex in fowl populations is considered to be of a special relevance in the case of studies following an evolutionary, ecological, and behavioral perspective. In this context, the determination of the sex of individuals is a key element in the design and planning of conservation and breeding programs [
5,
6,
7] for threatened or endangered species or breeds for which sex distribution may be biased or what is the same, whose characteristics may depart from a Hardy–Weinberg equilibrium [
8].
Vent sexing, venting, or Japanese examination is the most used traditional method of determining sex in one-day-old chickens, along with the sexing from the primary and secondary wing feathers [
9]. Qualified and trained professionals are able to appreciate the subtle morphological differences that exist between the genital eminences and local folds of males and females, by visual examination of the cloaca of the animal in the first hours after breaking the shell (first 12–26 h) and before the chicks start eating [
10,
11]. However, this method is complex and reaching an efficient enough sexing performance of operators requires investing in time, training, and economic resources.
The advances in genetics and reproduction have enabled the obtention, through guided crosses, of offspring for whom the sex of the individuals can be determined based on specific phenotypic characters at the time of birth or a few days later (auto-sexing and semi auto sexing), what simplifies the process. These methods rely on the growth rate of the feathers of the outer edge of the wing (primary feathers and secondary feathers or coverts) [
9,
10,
12,
13] or the color patterns in the first days after hatching [
14]. This methodology is not usually used in commercial practice, since its value is debatable given the time needed to be able to perform chicken sexing after their birth as differences in appearance are commonly emphasized as the animals grow [
11]. Other examples of noninvasive sexing techniques in chicks are the monitoring of the degree of opacity of the eggs during their incubation (with those eggs displaying a higher opacity resulting in male chickens) [
15] and the identification of possible behavioral differences depending on the sex [
16,
17,
18,
19].
On the other hand, there are invasive techniques to determine the sex of chickens, such as the Kizawa method or proctoscope sexing and molecular sexing. In the first, the gonads can be seen through the intestinal wall, using a cannula equipped with an optical augmentation system and a light bulb that is introduced through the rectum of the animal [
10,
11]. This method has been used successfully for day-old chicks and has the advantage of requiring less training and skill than the Japanese method. The molecular method, on the other hand, implies the identification of chromosomes by karyotype or the biochemical characterization of the genetic information of the animal, thus reserving this methodology of sexing for poultry strains of high economic interest [
20,
21].
The growing concern for animal welfare has placed the industrial practice of eliminating male chickens in laying strains as one of the greatest ethical problems in modern poultry [
22]. In this context, the use of accurate and reliable techniques for sex determination of endangered fowls is of special importance in breeding programs, especially in captive ones displaying monogamous behavior, with problems or absence of copulation, and with a low hatching rate [
23]. In view of this situation, numerous research works are being carried out to develop a technique of sexing of chickens during the first days of embryonic development, since it has not yet developed sensitivity, and hence, there is no suffering [
24].
The first in ovo sexing techniques available for industrial application are invasive in nature, as they require sampling for the production of a hormonal or genetic profile, which in turn ends up affecting the posterior hatchability of the egg. In addition, its application is limited only to the period of sexual differentiation of the embryo [
25,
26].
New tendencies of in ovo sexing systems focus on the lack of contact required (noninvasive methods) and early determination (as they allow the determination of sex on day 3.5 of incubation). One of the most remarkable discoveries makes use of Raman spectroscopy to analyze the spectrum of the circulating blood of the extraembryonic vessels, with a 90% accuracy. This technique is in its early stages of development and refinement before it can be implemented at a commercial scale in the incubation plants [
25].
Out of this commercial scale, endangered fowl populations may benefit from early sex confirmation, even if eliminating one sex or the other is not the aim, but to implement and design better population management strategies to preserve genetic diversity, when sex distribution across the population may be biased. Contextually, local fowl populations have additional issues regarding the fact that standardized methods may not be appropriate to determine sex, as sex may be conditioned by marked inner external features typical from each breed or variety [
27].
Morphological development of the gonads appears to have been conserved through evolution across animal groups. However, in vertebrates, sex has been suggested to be determined by a surprising variety of mechanisms including chromosomally (CSD) or environmentally linked ones (temperature dependent sex determining (TSD) mechanisms). Therefore, some authors concluded that although the basic genetic pathway controlling the morphological differentiation of the gonads appears to have been conserved, the genetic mechanisms triggering sex determination may involve an important diversity of signs [
28].
Sexual dimorphism comprises the differences between males and females of the same species, such as in color, shape, size, structure, behavior, or cognition that can be attributed to the inheritance of sexual genetic material [
29]. An important finding to have emerged from poultry studies is that sexual dimorphism at the molecular level can occur well prior to sexual dimorphism at the morphological level. Genes such as
DMRT1,
ASW, and
FET1, for example, all show sexually dimorphic expression as early as day 3.5, well prior to the histological onset of gonadal sex differentiation at day 6.5 [
30]. The W chromosome represents less than 2% of the chicken genome and contains only a few coding sequences, in contrast with the Z chromosome, which mainly includes the genes influencing feather color and growth patterns [
31,
32]. In this context, criss-cross inheritance magnitude of sex-linked traits is grossly determined by the crossing direction and the presence of certain dominant alleles in male or female genotypes [
14,
33].
In this regards, sexual dimorphism in avian species has been reported to affect several secondary phenotypical traits that can be noticeable even from newly laid eggs. These secondary phenotypical traits range from egg size [
34,
35], feather color, morphology and distribution [
36,
37], appendicular skeleton dimensions (tarso-metatarsus length) [
38], head length and size [
39,
40], tail inclination or lateralization [
41], or even behavioral patterns or cognitive processes [
39,
42].
When we consider sex determination through secondary external features, this inner diversity related to sex may join the implicit diversity present in the gene pool of genetically closer populations such as breeds or varieties [
27]. This specific context makes sex determination more challenging, given the lack of specific protocols to determine sex for each population, but more relevant as certain better productive features may be linked to either one or the other sex in specific local breed or variety populations [
43,
44,
45], Then, the choice of one method or another will ultimately depend on the degree of accuracy or certainty desired, the expected performance of the technique (number of individuals sexed/h), the availability of qualified and experienced professionals, the possible effects of sex selection for ‘unwanted’ or less profitable animals, and economic and financial conditions [
46,
47].
For these reasons, the aim of this study is to determine which method combinations may more successfully determine sex across the four varieties of Utrerana endangered hen breed. This information will be processed to tailor noninvasive early specific models to determine sex in local chicken populations. Identifying sex proportions in endangered populations can contribute to the improvement and progress of the genetic management tasks carried out in such populations, as an alternative or complement to more standardized techniques, which may have been proved at a commercial scale, but which may result inefficient given the implicit diversity found in related local populations.
3. Results
ICC and 95% IC aimed at testing for the intersexer reliability across methods, which proved to be highly reliable as there was statistically significant highly good to excellent agreement between the three sexers’ judgements across the 10 methods tested (
Supplementary Table S1).
Shapiro Francia’s W’ test showed that assigned sex by the 10 methods in the three assignation criteria, was normally or almost normally distributed (
p > 0.05). Descriptive statistics are shown in
Supplementary Table S3. Eighty-one chicks were submitted to the 11 sex assignation methods and then tested for their real sex. Sex assignation and signs related per each of the 10 methods used in Utrerana hens is shown in
Supplementary Table S2.
First, a chi-square independence test showed the variety substantially conditioned the ability to succeed or fail when assigning sex for Method 1.2: Egg width test, Method 2: English test, Method 6: Combs and Method 9: Legs (
p < 0.05), while it did not condition the results obtained from the other methods. Differences in the medians suggested Method 1.2: Egg width test was more likely to be able to assign franciscan and white animals correctly. Method 2 was more likely to correctly assign sex in white and black animals, by contrast, method 6 was more likely to assign sex correctly in black Utrerana hens. As all the records for method 6 reported the same value for assigned sex (2: female), given the animals’ combs had not developed at any of the moments of evaluation, method 6 was not considered in the following methods aimed at addressing differences (
Table 1 and
Table 2).
Afterwards, an exact McNemar’s test determined that there was a statistically significant difference in the proportion of animals whose sex had been correctly assigned using each of all the 10 remaining methods except for method 2: egg width test and method 9: legs (
p > 0.05) for assignation criteria 1 with respect to real sex. Assignation criteria two only reported statistically significant results for methods 2 (English test), 5 (down feathers), 7 (wing fan), and 10 (behavior/coping styles) (
p < 0.05) compared to real sex. However, for assignation criteria three, results were similar, thus statistically significant (
p < 0.05), except for method 1.2 (egg width test), method 9 (legs length), and method 4 (tail inclination) compared to real sex (
Table 3).
The logistic regression model was statistically significant, χ2(10) = 22.974, p < 0.001. Given the three assignation criteria comprised the same variables (methods), they all explained 33.1% (Nagelkerke R2) of the variance in confirmed sex, presenting a likelihood of correctly classifying 86.7% of males and 61.1% of female cases, resulting in an overall predictive percentage of 75.3%.
When studying assignation criteria 1, the probability of attributing sex correctly occurring based on a one unit change in method 10 (behavior/coping styles), method 4 (cloaca), and method 3 (tail inclination), when all other methods are kept constant was statistically significant (
p < 0.05) 7.176, 4.235, and 0.156 times greater, respectively, to assign males correctly as opposed to females (
Table 4).
When studying assignation criteria 2, the probability of attributing sex correctly occurring based on a one unit change in method 10 (behavior/coping styles), method 3 (tail inclination), and method 4 (cloaca), when all other methods are kept constant was statistically significant (
p < 0.05) 7.176, 6.420, and 4.235 times greater, respectively, to assign males correctly as opposed to females (
Table 5).
When studying assignation criteria 3, the probability of attributing sex correctly occurring based on a one unit change in method 3 (tail inclination), method 4 (cloaca), and method 10 (behavior/coping styles), when all other methods are kept constant was statistically significant (
p < 0.05) 6.420, 0.236, and 0.139 times greater, respectively, to assign males correctly as opposed to females (
Table 6).
4. Discussion
The genetic background behind sexual dimorphism comprises the differences between males and females of the same species, such as in color, shape, size, structure, behavior, or cognition that are caused by the inheritance of one or the other sexual pattern in the genetic material [
29]. An important finding to have emerged from the chicken studies is that sexual dimorphism at the molecular level can occur well prior to sexual dimorphism at the morphological level. Genes such as DMRT1, ASW, and FET1, for example, all show sexually dimorphic expression as early as day 3.5, well prior to the histological onset of gonadal sex differentiation at day 6.5 [
30]. However, the high costs involved in the implementation of molecular techniques makes the development of early sex assignation techniques become especially relevant, even more in the case of local breeds, whose market opportunities are still developing, thus have not reached a sufficient level of profitability as to ensure their survival and satisfy the increasing market demands.
These differences not only attain species but also certain domestic breeds or even varieties [
57]. Given the varieties may present different sexual dimorphism patterns, there is a relatively strong possibility that sex determination methods may not be valid across the same varieties of a particular breed. The conditioning effect of the variety on the success to determine sex is supported by the results obtained at the chi-square test of independence and differences in the median, which reflects the significant relationships existing between the categorical sex variable and some of the methods of sexing employed (
Table 1 and
Table 2).
Egg width, the English test, combs and leg length may present significantly moderately high differences in the suitability to successfully determine sex across the varieties of the Utrerana hen (white, partridge, black, and franciscan). In these cases, a bias may occur towards either of the two sexes as shown by median values shown in
Table 2. The rest of tests are appropriate to determine sex for all the individuals, with independence of the variety that they belong to, as results may not be affected by such factor.
When we assessed the results for the differences across the three different assignation criteria, results varied. This finding may rely on the bias occurring due to the particular dimorphic characteristics which may differ across the different varieties (
Table 3).
No difference seems to exist across varieties when egg length was compared. However, the egg length test was able to significantly detect differences between females and males when the assignation criteria found in literature (first assignation criteria) was used and when such criteria was reversed (second assignation criteria) (that is if a male was reported in literature to present a certain characteristic while a female did not, we considered the opposite possibility, that is males lacked the characteristic and females presented it).
Contrastingly, egg width reported statistically different results across varieties as shown in
Table 1 and
Table 2.
Table 2 shows that the results for egg width are biased towards incorrectly assigning male sex to chickens in franciscan and white varieties, while they were prone to incorrectly assign females in black and partridge varieties, something that did not occur for egg length (χ
2 = 2.503, df = 3,
p = 0.47). Egg width test was able to detect significant differences between females and males when the assignation criteria found in literature (first assignation criteria) was used and when such criteria was reversed (second sex assignation). For the third assignation criteria, such differences were not detected as reported in
Table 3.
The relationship between and the influence of egg measurements on the determination of chicken sex has been reported on only a few occasions in literature. Yilmaz-Dikmen and Dikmen [
58], reported the effects of egg shape index (
p = 0.001), egg length (
p = 0.0018), egg width (
p < 0.01), and volume (
p = 0.004) of the egg had a significant effect on the sex of hatching chick of Super Nick White Layers. This could have been expected as the mathematical methods to compute egg shape index or egg volume reported in literature widely depend on egg width and length parameters. In particular, egg volume has been reported to be significantly different between males and females in some avian species [
35,
58]. For example, using molecular techniques, house sparrow eggs containing male embryos were reported to be significantly larger than those containing female embryos, considering they are laid randomly with respect to laying order [
34]. The same authors speculated that this sexual dimorphism of eggs was adaptive, because male house sparrows were more prone to present a greater variability in condition-dependent reproductive success than females. Such results provided further evidence of the ability of females to detect or control ovulation of either male or female ova and to differentially invest in one sex over the other. This was supported by Mead, Morton, and Fish [
35] with male eggs in Mountain White-crowned Sparrows (
Zonotrichiu leucophrys oriantha) being highly significantly slightly larger than those of female eggs for each of the five consecutive years that the study lasted (
p < 0.01).
Some authors have suggested plumage variety may play an important role. In particular, the greater or lesser accuracy when determining the sex of an individual have been reported to potentially be conditioned by the variety of hen Utrerana for method 1.2 (width of the egg), method 2 (English method), method 6 (degree of development of the crest), and method 9 (leg length). In the particular case of method 6, this would not provide conclusive data for any variety if the tasks of sex assignment are developed before the 1.5 month age, as it happens in our case. Hence, this method was discarded as a method of sexing for subsequent analyses, and it was rather used to confirm sex when the animals reached 1.5 months of age (6 weeks).
For Brown Leghorn, adult plumage is a sexually dimorphic feature, in terms of the kind and distribution of color in the individual feathers, and in one or more of the seven areas in which plumage color differences may distinguish among breeds and varieties. Furthermore, males are characterized by the structure of the feathers of the neck and saddle hackles, and by the presence of the large tail sickles in males. In other cases, such as the White Leghorn, the sexes are to be distinguished only by the structural differences in the hackle feathers and by the large tail sickles of the male. In the case of certain other breeds, the Campines and the Sebrights, for example, the plumage of the male is identical with that of the female both in coloration and in structure. Cocky-feathering in the case of such varieties as the Brown and the White Leghorns can be regarded as a trustworthy indication that within the body there is, or was at the time when the plumage was developed, active functional testicular tissue; henny-feathering as an indication that there is, or was when the plumage was developed, active functional ovarian tissue.
The underlying causes and the likely physiological consequences of the association of certain genes with sex biased expression were not the focus of this study. However, these remain interesting topics for future research. For instance, sex-biased gene expression of various steroid hormones in adrenals may potentially contribute to the extensive sexually dimorphic behavioral and physiological traits observed in chickens [
42]. Chronic [
59,
60] and acute [
55] stress may cause immediate, short- and long-term changes in physiology, behavior, and gene regulation. These responses may vary between individuals as well as between classes of individuals within a single species (for example, between sexes of animals with different coping styles) [
58,
61] and their characterization may greatly increase our understanding of stress effects in general and their differential extent between males and females [
54].
Stress causes cascades of both immediate and long-term changes in physiology, behavior, and gene regulation. However, both short- and long-term responses may vary between individuals as well as between classes of individuals within a single species (for example, between sexes, or between animals with different coping styles) [
61]. Characterization of such differences may greatly increase our understanding of stress effects in general. Chronic stress in chickens (
Gallus gallus domesticus) has been reported to be the case, for example, changes in learning ability, social dominance, feeding behavior, and gene expression [
59,
60], and the extent of these effects were found to differ between the sexes [
62]. Whether similar sex differences are also found in response to acute stress experiences (i.e., stressors acting over short periods) is less well understood [
63].
Contextually, the study by Zappia and Rogers [
64] examined the effect of testosterone on the asymmetry of visual discrimination performance of young chicks. Two-week-old chicks were tested on the pebble floor visual discrimination task. Male chicks were found to have brain asymmetry for visual discrimination learning, since chicks tested binocularly, or tested monocularly using their right eye system, have superior learning performance compared to chicks tested monocularly using their left eye system.
Among zoometric parameters, neonatal tail posture is a sexually dimorphic behavior with females more biased leftwards than males. Prenatal exposure of female chickens to testosterone propionate (TP) but not dihydrotestosterone propionate (DHTP) shifts the population pattern of tail posture to the right. Contrastingly, no effect was found with male pups [
65].
Casual observations suggested that the difference in size between male and female fowls is more marked in the tarso-metatarsus than in other bones of the appendicular skeleton. Plumage usually refers to the shape, size, and appearance of the feathers on a fowl at any specific time. Plumage, therefore, continually changes in juvenile individuals, and becomes fairly consistent in adults. Neck feathers are usually referred to as the “hackles”. The hackle feathers of the adult male are very distinctive and can alone be used to differentiate sex of the chicken. Male hackles are long, pointed, and usually reach down to the wings, while in the female the hackles are less distinctive, rounded in shape, and usually blend in with the other body feathers. The hackle feathers of many-colored males are greatly prized for producing “flies” for fishermen.
The plumage of most adult males and females also differ in the pelvic region, where again, the males have much longer and more pointed tail coverts. Abnormalities in growth of these feathers, or physical damage, can cause disruption of this normal arrangement of the primaries, often leading to a characteristic ruffling or “sticking-out” of one or two feathers. In broiler chickens, this latter condition is sometimes referred to as “helicopter wing”, where one or two primaries on each side of the body stick out at a 25–45° angle compared to their normal plane, which is parallel to the body [
37].
Thyroid hormone has been reported to be associated to much darker black/brown color due to extra melanin deposition in eggshells. Sex hormones also influence feather color. For example, in Brown Leghorn fowls’ sexual dimorphism of feather color is under the control of estrogen although the effect is influenced by thyroid status [
66,
67]. Sex hormones can (but do not always) influence melanoblast differentiation. Genetics rather than hormone balance essentially control the different feather down colors of day-old chicks. However, if estrogen or testosterone levels are altered in the developing embryo, down color can be affected. The authors of [
68] suggest that selection for genes that suppress feather color results in improved feed efficiency in layers, possibly due to better retention of feather cover as fowls age.
The conditioning effect of variety on the accuracy of sexing methods becomes evident by the results obtained at the χ2 independence test and the differences in the medians, which suggest the existing significant relationship between the dicotomic variable of sex and some of the sexing methods employed. In particular, the greater or lesser precision in determining an individual’s sex could depend on the variety of Utrerana hen for method 1.2 (egg length), method 2 (English method), method 6 (degree of development of the crest), and method 9 (length of legs). In the particular case of method 6, this method would not report relevant information as sex assignment tasks are carried out before the age of 1 month, as is our case, the reason why this method was discarded in subsequent analyses, but used for sex confirmation.
Once the influence of the variety of the Utrerana breed on sexing methods in chickens at an early age has been determined, the best accurately performing combination of methods was determined. Given their efficiency regardless of the variety of the individuals, assignation criteria 2 reported the most accurate results when the gender allocation criteria used was the one reversing the information normally provided by literature. Hence, as opposite to literature, if the chick kicks, it will be considered a male; while if the animal remains motionless, it will be considered a female.
The proportion of animals whose sex had been correctly assigned differed significantly between methods, namely, for assignation criteria 1, all methods used except method 1.2 (egg width) and method 9 (leg length), allowed for certain sexing at a statistically significant level. However, in assignation criteria 2, only method 3 (English method), method 5 (general down feathers coloration), method 7 (wing fan), and method 10 (behavior/coping styles) reported these same significant results. In the case of assignation criteria 3, the results are similar to those obtained for assignation criteria 1, with no statistically significant proportion of animals sexed for methods 1.2 (egg width), 3 (tail inclination), and 9 (legs length).
Finally, the percentage of variance explained (75%) in the sex of chickens at an early age is the same for the three assignation criteria considered. This could have been expected as only sex proportions in the population was considered and the same within sample variability occurred. However, although the explanatory power of all assignation criteria is identical, that is they presented an equal potential to explanation of variance, assignation criteria 2 resulted to be the most efficient as it correctly assigns males more frequently.