Study area, collection procedures and information procurement

The study was carried out in the region of Central Valleys in Oaxaca, Mexico, which includes the districts of Ejutla, Etla, Ocotlán, Tlacolula, Zaachila, Zimatlán and Center, using the probabilistic sample by conglomerates in two stages (^{Sukhatme, 1970}). The primary units were constituted by the seven districts, and the production units (PUs) inside the districts were considered secondary units. A sample of three primary units (n) was chosen and inside them, 19 secondary units were chosen randomly, representing 16 % of the population. In each secondary unit nine hens were chosen on which 21 morphometric variables were measured, constituting a total sample size of 171 adult hens distributed in the study region, with an estimated inventory of adult hens of 2004 animals. The size of the sample was obtained with an accuracy of 10% and reliability of 90 %.

The information about morphometric measures was obtained from each animal, considering 21 variables described in Table 1, outlined in Figure 1, considering weight and age. To take the measurements, the following were used with each indigenous hen: digital scale, digital Vernier (Mitutoyo, CD-18¨C) and measuring tape.

Figure 1 Morphometric variables evaluated in indigenous hens. 

Univariate descriptive and Pearson correlation analyses were carried out, prior to the analysis of the model’s assumption. The comparisons between males and females were performed through a t-Student test for independent samples. In addition, the multivariate principal component analysis (PCA) technique was used, which allowed reducing the multidimensional space of morphometric variables in a subset of linear combinations (^{Morrison 1976}). In order to avoid the scale effects of the PCA, the variables were standardized. The PCA results were graphed in a biplot that allowed the representation in two dimensions (first and second PC) of the birds and the variables measured simultaneously (^{Johnson, 1998}). The analyses were performed using SAS ver. 9.0 (^{SAS Institute, 2002}).

Morphological features

Table 2 shows the descriptive statistics, means and standard deviations, finding differences (p<0.05) between the dorsal weight and length of males and females: 2.5±0.5 kg, 45.4 cm and 2.0±0.5 kg, 41.4 cm, respectively. Similar results in the variables studied were reported by other researchers in Mexico, Spain and Botswana (^{Jerez-Salas, 2014}; ^{Campo, 2009}; and ^{Badubi et al., 2006}, respectively).

Phenotypical correlations

Table 3 shows the significant phenotypical correlations (p<0.01) between some morphometric characteristics of indigenous hens. The positive correlations were between: live weight with pectoral perimeter (0.62), live weight with tarsus length (0.45), confirming: with higher weight, birds are larger and therefore have a greater breath of the breast area. Likewise, tarsus length with thigh length (0.69) and tarsus length with appendage length (0.67) are associated with the development of extremities with sexual accessories of the head. These results agree with those reported by ^{Alabi et al. (2012)} and ^{Guni et al. (2013)} in South Africa and Tanzania, who found a high positive correlation between the live weight with pectoral perimeter and tarsus length with thigh length, which describe the size of the animal.

Typological description based on PCA

The principal component analysis allowed reducing the dimensionality of the 21 variables considered, showing that five linear combinations explained 80 % of the variation, but it was considered that the first two PCs were the most important, given that they explained 63 % of the total variation (Table 4) and allowed establishing a practical and easy-to-understand hen classification based on their morphological characteristics.

The PC1 component represents the highest amount of variability of the data, explaining 50 % of the total variance, composed by variables of: tarsus length, thigh length, which defined the size of the bird; while chin length, chin width, crest width, beak length, crest length, appendage length, appendage width explain the shape and magnitude of the head (Table 5). That is, it explains the size and shape of the head that is directly related to the sexual dimorphism of the bird; in males, these accessories are developed and attractive, while in females they are smaller. These results are consistent with those by ^{Yakubu et al. (2009}), who reported the PC1 with the variables of height and length of crest, tarsus length and thigh circumference, in a study of local chickens in Nigeria. However, ^{Egena et al. (2014)} in the PC1 found different variables in a population, such as: wing length (0.840), weight (0.826), and body length (0.814); this can be attributed to the ancestral origin and environmental conditions of adaptation in Nigeria.

The PC2 explains another 13 % of the total variability made up by the variables of weight, pectoral perimeter, and dorsal height (Table 5); as a whole, they define the size or build of the bird that reflects the breeding purpose of the hen in the production units, for egg, meat or double-purpose; that is, light hens would have a better aptitude for egg production, while the heavier tend to get more meat, and the double-purpose fulfill both functions. This agrees with results by ^{Yakubu et al. (2009)}, who found in the PC2 the thorax circumference and length of the birds to be related to the size of the animal.

The two principal components originated that explain the size and build of the animal were the basis to establish the phenotypical descriptors as criteria of typology or grouping of the birds, giving rise to three groups according to the dimensionality plan of indigenous hens (Figure 2).

Figure 2 Distribution of indigenous hens in the level of dimensionality with principal components CP1 vs CP2. 

The first group of birds was characterized as heavy, larger than 3 kg; therefore, they have longer extremities, such as the tarsus and length of leg; that is, animals of greater size and build, which are bred in the production units with the aim of obtaining eggs to preserve those genes in the future offspring and obtain meat for self-consumption after concluding their first laying cycle.

The second group of birds was the most predominant, with a weight of 2 kg and medium size; this evidences that it is double-purpose hens in the production units that are preferred for egg extraction; in addition, they are also a good source of meat.

The third group was identified as the light or inferior, with a weight lower than 2 kg, with characteristics of lower size and weight. This suggests that the aim of these birds in the production unit is to obtain a higher amount of eggs in each laying cycle compared to the second group.

As background, it is convenient to note that one of the greatest advances in industrial poultry production is the moment when the breeders differentiated their genotypes as heavy and light hens (^{Nordskog, 1976}; ^{Orozco, 1991}). In our case there is an intermediate genotype, since this kind of rural poultry production is basically double-purpose: meat and egg production.