Conservation and improvement of native animal species requires morphological characterization, as well as estimation of genetic and phenotypic parameters of economically significant traits. Understanding the types of associations or correlations between traits is fundamental to developing genetic improvement strategies, because direct selection can be impractical or costly for traits that are difficult to measure or of low heritability.

The Mexican domestic turkey (Meleagris gallopavo gallopavo) is an important reservoir of useful genes with notable adaptive traits¹, that allow it to live and develop in almost any of the country’s many agro ecological zones. In recent years, a number of studies have addressed the morphological and genetic diversity in Mexican backyard turkeys from certain regions: the coast of the state of Oaxaca; Mani municipality in the state of Yucatan; the north-central portion of the state of Chiapas; the state of Michoacán; and Kopala municipality in the state of Puebla^2-9. This is only a small portion of the country, and studies are needed in many more agro ecological regions and production systems to create a functional database.

Only one study to date has included phenotypic correlation estimates for quantitative morphological characteristics such as body weight, breast circumference, tarsus length, body length, height, and leg length in Mexican backyard turkeys⁷. The results suggest that phenotypic correlation estimates depend on the sex of the bird since they were moderate in males, but low in females, and some of the estimates in females were not significant. However, this study used a limited number of observations (<50 per sex) to calculate the estimates. Confirming estimates validity will require additional studies involving larger numbers of observations. The present study objective was to evaluate quantitative and qualitative morphological characteristics of backyard turkeys from twenty-four states in Mexico, and estimate the phenotypic correlations between the quantitative morphological variables.

From 2013 to 2014, morphological data from 248 backyard turkeys were collected by opportunity sampling in 126 rural production units located in 75 municipalities (Table 1) in 24 states of Mexico: Baja California Sur, Campeche, Chiapas, Chihuahua, Coahuila, Colima, Mexico City, Durango, Estado de México, Guanajuato, Hidalgo, Jalisco, Morelos, Nayarit, Nuevo León, Oaxaca, Puebla, Querétaro, Quintana Roo, Tabasco, Tamaulipas, Tlaxcala, Veracruz and Zacatecas. This sampling technique was used because of its low cost, minimal time requirements, relative simplicity and production of data from a wide array of individuals. The sample included juvenile and adult turkeys of both sexes (138 males and 110 females).

Seven quantitative variables were evaluated: body length, wingspan, breast circumference, tarsus length, live body weight, robustness, and solidity. Three qualitative variables were evaluated: plumage color, skin color, and tarsus color. Measurement of quantitative variables was done following the FAO Guidelines for Animal Production and Health¹⁰. Turkeys were weighed with a 10 kg capacity hanging scale (ECO-DIN 10; PEXA; accuracy = ± 25 g). A flexible plastic measuring tape (FIBER-GLASS) was used to collect data on body length, wingspan, breast circumference and tarsus length. The seven quantitative variables were defined as follows:

Body length. The distance between beak base and caudal tip at the level of the uropygial gland, without considering feather length on the tail; measured while extending the neck.

Wingspan. The length from the terminal phalange of one wing to the terminal phalange of the other wing (not including feathers) with the wings completely extended.

Breast circumference. Measured at the level of the crest of the breastbone, extending the measuring tape along the posterior insertion of the wings.

Tarsus length. Metatarsal length measured as the distance from the intertarsal articulation to the metatarsophalangeal articulation.

Body weight. Bird live weight as measured on a hanging scale at the time of the visit to the production unit.

Robustness. (Breast circumference/body length) x 100.

Solidity. (Body weight/body length) x 100.

An analysis of variance (ANOVA) was run for each quantitative trait using the GLM procedure in the SAS statistics package¹¹. In all cases, the statistical model included sex, state, and municipality nested within state. The model was mathematically represented as:

y_ijkl= µ + α_i + β_j + γ_k(j) + ε_ijkl,

where: y_ijkl is the l-th observation of the response variable (body length, wingspan, breast circumference, tarsus length, live body weight, robustness, or solidity); µ is the general mean; α_i is the fixed effect of the i-th sex (i= 1,2); β_j is the fixed effect of the j -th state (j=1,…,24); γ_k(j) is the fixed effect of the k-th municipality nested within the j-th state (k=1,…,75); and ε_ijkl is the l-th random error; y_ijkl ~ N(µ, σ²).

Differences between means of males and females were estimated using the PDIFF option in the GLM procedure. In addition, the CORR procedure was run to estimate the Pearson correlation coefficients among body length, wingspan, breast circumference, tarsus length and body weight. This was done to identify the variables most strongly correlated with body weight. The correlations were estimated independently for males and females, and then for the combined data set of males and females. In response to reports indicating that thoracic perimeter is a good indicator of body weight in different animal species¹², the REG procedure was run to estimate the linear regression coefficient of body weight on breast circumference. Again, this was done independently for males and females, and then for the combined data set of males and females. Finally, the FREQ procedure was applied to calculate the frequency of the categories observed in the field corresponding to each of the studied qualitative traits.

Average weight in the studied animals was 4.6 kg, with a range of 2.5 to 8.7 kg (Table 2). This is similar to the 2.9 to 8.9 kg range reported for male and female turkeys in the state of Michoacán⁸, but generally lower than the 6 to 20 kg range found in the Xochimilco region of Mexico City¹³. Minimum and maximum values for body length and wingspan were similar, although average wingspan (67 cm) was slightly greater than average body length (62 cm). The variation coefficients for body length, wingspan, breast circumference and tarsus length were about two times less than the body weight variation coefficient.

State and sex were significant (P<0.01) sources of variation for all the analyzed quantitative variables. However, municipality affected (P<0.01) body length, wingspan, tarsus length, body weight, robustness and solidity, but not breast circumference.

Sexual dimorphism was clearly present in the sample. Least squares means of the quantitative variables showed males had greater (P<0.001) body length (10.4 cm more), wingspan (11.4 cm more), breast circumference (13.8 cm more), tarsus length (2.5 cm more), body weight (2.5 kg more), robustness (9.0 percentage points more), and solidity (2.8 percentage points more) than females (Table 3). This agrees with one previous study that found males to have greater weight than females⁴, and another that found males to have greater body weight, dorsal length, breast circumference, tarsus length and wing length than females⁵. A third study also found that males had higher values than females in terms of breast circumference (54.03 vs 42.55 cm), tarsus length (15.35 vs 12.08 cm) and body weight (5.02 vs 2.94 kg)⁷. The overall trend of males having higher quantitative variable values than females has also been observed in other countries such as Croatia^14,15) and Nigeria^16,17.

The Pearson correlation coefficients estimated separately for males and females showed body weight and breast circumference to be highly phenotypically correlated (P<0.01) in males (r= 0.74) and females (r= 0.71)(Table 4). In contrast, body length was unrelated to wingspan and tarsus length in females, but was correlated to them in males, though at a low value. The estimates for the phenotypic correlations of body weight with body length (r= 0.38), wingspan (r= 0.30), and tarsus length (r= 0.34) were moderate in males. For females, estimated values were moderate between body weight and body length (r= 0.35) and tarsus length (r= 0.34), but low between body weight and wingspan (r= 0.25).

When the data for males and females was analyzed as a set, all the correlation coefficients differed from zero (P<0.0001) and were positive, indicating that if one variable increased all the others did as well (Table 5). In this data set, body weight exhibited high phenotypic correlation with body length, tarsus length and wingspan. The phenotypic correlation estimates were low for body weight with wingspan (r= 0.42) and tarsus length (r= 0.49).

Correlation coefficient values between these variables have varied among a number of studies done in different countries and with different types of domestic turkeys. In one study done using the MH meat production turkey line, body weight and body length were found to have a low correlation (r= 0.38), while body weight was strongly associated (r= 0.85) with breast circumference¹⁸. Native Nigerian turkeys (Meleagris gallopavo) were reported to have a strong correlation between body weight and wing length (r= 0.91), tarsus length (r= 0.97), body length (r= 0.93), and breast circumference (r= 0.89)¹⁹. In a more recent study done in Tunisia²⁰, high correlations were identified between body weight and tarsus length (r= 0.89), and body length (r= 0.90). Another study reported moderate correlations between breast circumference and tarsus length in male turkeys (r= 0.67 and 0.46, respectively), but low correlations in females (r= 0.30 and 0.06, respectively)⁷. Finally, in Nigeria²¹ native turkeys (Meleagris gallopavo) were found to have high correlations between body weight and breast circumference (r= 0.83), body length (r= 0.90), and tarsus length (r= 0.89).

The regression analysis showed the regression coefficient of body weight on breast circumference to differ from zero (P<0.01). Body weight in males increased 143 g for every centimeter increase in breast circumference, and in females it increased 113 g. In the regression analysis of the grouped data, body weight increased 159 g per centimeter gain in breast circumference. The fit (Figure 1) of the linear regression of body weight (BW) on breast circumference (BRC) was run using one equation for males, one for females and a third for the grouped data:

BW= -2.1108 + (0.1427 x BRC) [males]

BW= -1.3874 + (0.1130 x BRC) [females]

BW= -3.1485 + (0.1593 x BRC) [males & females]

Figure 1 Linear regression of body weight (BW) on breast circumference (BRC) in males, females, and males and females grouped 

For males, the standard error was 0.61888 for the intercept and 0.0111 kg for the regression coefficient; in the females they were 0.4561 and 0.01093 kg, respectively; and in the grouped set they were 0.28691 and 0.00577 kg, respectively. By comparison, 12-wk old native Nigerian turkeys were found to increase 106 g in body weight for every centimeter increase in breast circumference²². The difference between this estimated regression coefficient and that in the present results could be due to differences in the average age of the animals in each study.

Thirteen (13) different plumage color phenotypes were documented in the present study, highlighting a wide diversity among the sampled locations. Of the four basic identified colors (white, brown, grey and black), black was the most frequent (30.8 %), followed by white (14.7 %), brown and then grey. Nine combinations of these four colors were identified (color order does not indicate relative predominance), the most frequent being “black with white” (18.1 %), followed by “black with white and brown” (10.2 %), “white with brown” (5.8 %), and “black with white and grey” (4.9 %). On the coast of Oaxaca state, the most frequent colors in turkey plumage were black (29.2 %) and “white with black” (30.3%)², whereas in the state of Yucatan the predominant colors were black and “black with brown and white”⁴. Black was also found to be the predominant color among turkeys in the state of Puebla, both in males (46.7 %) and females (52.2 %)⁷, and among native turkeys in Croatia¹⁵⁾ and Morocco²³.

Among the five skin colors observed in the studied sample, white was the most common (81 %), followed by yellow (11 %), and pink (3.9 %). Of the nine observed tarsus colors, the three most frequent were brown (31.3 %), white (26.4 %) and black (20.5 %), although pink, red, purple, green, yellow and grey were also present. This contrasts with the six tarsus colors documented on the coast of Oaxaca, white being the most frequent², and the four found in Yucatan, pink being most common⁴.

The present results indicated the presence of five main distinguishing aspects among the sampled Mexican backyard turkeys. Sexual dimorphism was notable, with clear morphological distinction between males and females. Breast circumference and body weight were highly correlated in both males and females, but for other traits the magnitude of the phenotypic correlation estimate was sex-dependent. Black and “black with white” turkeys were the most common, most of the turkeys had white skin, and brown was the most frequent tarsus color.