Genetic analysis of fertility traits of Holstein dairy cattle in warm and temperate climate

The edited data set for the estimation of heritability, genetic and phenotypic correlations of fertility traits contained up to 23,402 records from 10,894 cows calved between 2001 and 2015. Heritability estimates for success in first service (FS), gestation length (GL), number of inseminations (NI), insemination outcome (IO), calving interval (CI), calving birth weight (CBW) and days open (DO) were low and ranged between 0.016 (DO) and 0.123 (GL). Repeatability of fertility traits was estimated to vary from 0.021 (FS) to 0.411 (IO). The genetic correlations between DO × CI, DO × NI and CI × NI were positive and nearly perfect (0.98, 0.88 and 0.88, respectively), while those between DO × IO and CI × IO were negative (-0.98 and -1, respectively). Further, the phenotypic correlations between DO × CI, DO × NI, CI × NI, CBW × IO and SF × IO were 0.99, 0.83, 0.83, 0.99 and 1, respectively, while those between DO × IO, CI × IO, GL × IO and NI × IO were -0.99, -0.99, -0.99 and -1, respectively. Overall genetic parameters imply a good practical management in heat stress conditions will be essential for improving fertility efficiency.


Introduction
Fertility traits are considered very important because of their impact on the economy of dairy cattle breeding.Economic losses due to fertility problems are mainly caused by low dairy yield, prolonged calving intervals, increased insemination costs, few calves per cow per year, increased culling, high replacement costs and shorter reproductive lifespans (González-Recio & Alenda, 2005;Abe, Masuda, & Susuki, 2009).Nonetheless, for many years, genetic improvement programs worldwide did not include reproductive performance, since the selection was mainly focused on milk yield.An exception was Scandinavia, whose selection indices included not only milk yield, but also health and reproductive traits (Miglior, Muir, & Van Doormaal, 2005).Now, the increase in milk yield without considering the reproductive performance is a problem, because it produced an important decline in the reproductive efficiency over time (Pryce, Royal, Garnsworthy, & Mao, 2004;Melendez & Pinedo, 2007).Fertility in lactating dairy cows is also very sensitive to season, especially in hot climates.Global warming and the breeding of selected animals that are more and more sensitive to environmental effects have made this phenomenon, named heat stress, particularly relevant even in temperate areas (Nardone, Ronchi, Lacetera, Ranieri, & Bernabucci, 2010;Ferreira, 2013).Decreasing heat tolerance may be another of the reasons for decline in reproductive efficiency.So, one way to counteract this decline is through genetic selection (Pszczola, Aguilar, & Misztal, 2009).It is difficult to determine which traits must be included in the genetic evaluation of fertility, since they have very low heritability values, i.e., close to 0.1 (Thaller, 1998;Jamrozik, Fatehi, Kistemaker, & Schaeffer, 2005).However, over the last decade, reproductive traits have increasingly been included in the selection indices for reproductive traits in genetic evaluations in different countries, thus highlighting the importance of including fertility in improvement programs of dairy cattle (Miglior et al., 2005).Pozveh, Shadparvar, Shahrbabak, and Taromsari (2009) estimated genetic parameters for calving interval (CI), days open (DO), and gestation length (GL) in traits collected by the Animal Breeding Center of Iran from 1980 to 2004 on a data set including fertility records from 6000 cows.However, these traits and other economically important traits, such as number of insemination (NI), insemination outcome (IO), success in first service (SF) and calving birth weight (CBW) were not considered in heat stress condition.The objectives of this study were to estimate heritability, repeatability, genetic and phenotypic correlations of fertility traits of Holstein dairy cattle in warm and temperate climate.

Reproductive data and editing procedure
The data were collected from Holstein dairy population located in the north of Iran.The edited data set contained up to 23,402 records from 10,894 cows calved during 2001 to 2015.Original data file for reproduction traits consisted of insemination records that were matched to pedigree, lactation, and calving performance information to calculate the traits of interest.The fertility traits selected for this study were success in first service (SF), gestation length (GL), number of inseminations (NI), insemination outcome (IO), calving interval (CI), calving birth weight (CBW) and days open (DO).The SF and IO as binary traits and the NI as categorical trait were considered whereas the GL, CI, CBW and DO were determined as continuous traits.Insemination outcome was defined as 1 = successful if cow became pregnant at insemination time and 0 = failure.Gestation length was measured as an interval from the last insemination to subsequent calving; GL was considered between 240 and 290 d.Days open was defined as the number of days between calving and conception; DO was limited to between 45 and 350 d.Calving interval was defined as the number of days between 2 consecutive calving events.CI records were limited to be between 285 and 640 d.Number of services was defined as the number of inseminations within a lactation; If NI was greater than 10, then NI was assigned to 10. SF was a binary trait defined as 1 = successful if cow became pregnant at first insemination and 0 = failure.Also, CBW was required to be between 20 and 60 kg.Subsequently, cows without pedigree information were excluded.

Climate data
Daily climate records were obtained from the most nearby meteorological station located at the same altitude as the farm studied.The major climatic variables directly affecting livestock are temperature, humidity, air movement and radiation (Konig, Chongkasikit, & Langholz, 2005).Attempts to combine environmental parameters in one single index have had limited success except for the temperature-humidity index (THI) (Kadzere, Murphy, Silanikove, & Maltz, 2002) A daily THI was computed using the following Equation 1 (National Research Council [NRC], 1971): where: T is mean daily temperature in degrees centigrade and RH is the mean daily relative humidity as a percentage.
Insemination records were merged with daily temperature-humidity index.THI on the day of the insemination, 1 d prior and 1 d after insemination were studied as independent variables.

Genetic analysis and statistical models
The reproduction traits (SF, GL, NI, CI, CBW and DO) were analyzed with the Equation 2 as follow: Also, trait of IO was analyzed with the following Equation 3: where: P is the observed trait of SF, GL, NI, CI, CBW and DO; Q is the observed trait of IO; μ is the mean of trait; Parity is the fixed effect of parity in 5 classes; YS is the fixed effect of year-season of calving in 14 and 4 classes, respectively; DYS is the fixed effect of dystocia score (1 = no problem to 5 = caesarean); β 1 and β 2 are linear and quadratic regression coefficients of dependent variable (P, Q) on days in milk effect, age of calving or temperature-humidity index effect; X DIM as continuous variable representing days in milk, in weeks ranged from 15 to 105; X AC as age of animal at calving in months, ranged from 20 to 135; X THI as continuous variable representing temperature-humidity index; Animal is the random additive genetic effect; pe is the random permanent environmental effect and e is the random residual effect.(Co) Variance components were estimated by AI-REML in DMU software package (Madsen & Jensen, 2013) using an animal linear mixed model; univariate threshold models were also carried out for the binary traits.Heritability was estimated as the ratio of the additive genetic variance to total phenotypic variance; and repeatability, as the ratio of the sum of the additive genetic variance plus permanent environmental variance to phenotypic variance, as described by Falconer and Mackay (2001), according Equation 4and 5: Genetic and phenotypic correlations between traits were estimated using a series of bivariate animal linear mixed models.The covariance structure for the models was defined as Equation 6and 7: where: A is the numerator relationship matrix; I was an identity matrix of appropriate order; is the Kronecker product;  0 = variance and covariance matrix of random additive genetic effects;   2 = animal additive genetic variance for trait i;   2 = additive genetic variance for trait j; =   = additive genetic covariance between traits i and j;  0 = variance and covariance matrix of random permanent environmental effects;   2 = permanent environmental variance for trait i;   2 = permanent environmental variance for trait j;   =   = permanent environmental covariance between traits i and j;  0 = variance and covariance matrix of residual effects;   2 = residual variance for trait i;   2 = residual variance for trait j; and   =   = residual covariance between traits i and j.

Climatic conditions in the North of Iran
Climatic conditions in the north of Iran (Sari city) could be characterized as mild, and generally warm and temperate.The rain in Sari fallen mostly in the winter, with relatively little rain in the summer.The annual rainfall averaged 690 mm.The lowest precipitation was in June, with an average of 23 mm and the highest occurred in December with an average of 98 mm.The mean of temperature and humidity of present study was 18.26 ± 7.79°C and 75.19 ± 9.29%, respectively.The THI was lowest in January and February (mean of 56), which was associated with the winter season, and highest in June through September (mean of 81), which was associated with the summer season and heat stress condition.

Descriptive statistics
In this study, the mean and standard error for DO was 140.36 days (± 76.16).Also, mean of calving interval was 415.99 days (± 79.62).In accordance to the means of DO and CI traits, average of GL was 278.2 days (± 5.58).Moreover, mean of number of inseminations in this herd was 2.73 (± 1.94).Also, the averages of SF and IO traits of herd were low (0.32 ± 0.003 and 0.31 ± 0.001, respectively).Finally, the mean of calving birth weight was 40.4 kg (± 6.08).

Heritability, repeatability and variance components
Variance components and estimated heritability and repeatability for fertility traits are shown in Table 1.Heritability of GL was higher than others.Estimated heritability for DO and SF was low.However, DO as interval trait showed greater additive genetic variance related to SF as binary trait.Estimated additive genetic variances for CI and DO (11.34 and 7.23,respectively) was much greater than other traits.Additive genetic variance for SF was the least among all assessed traits.Also, permanent environmental variance for CI and DO (11.01 and 7.47, respectively) was the highest and for SF (0.001) was the least among all traits.Likewise, estimated repeatability for DO and SF was lower and for IO was higher than others (Table 1).

Additive genetic and phenotypic correlations
As shown in Table 2, estimates of additive genetic correlations varied from -1 for CI and IO to 0.98 for DO and CI.Phenotype correlation estimates ranged -1 for NI and IO to +1 for SF and IO.
The low heritability in this study suggested that improvement of fertility traits in cows could be achieved by improving reproductive managements such as successful detection of heat, timely insemination, feeding practice for growing and postpartum animals and controlling heat stress.The repeatability estimate for DO in Holstein (0.034) was lower to those reported by Demeke et al. (2004), Estrada-León et al. (2008) Kadarmideen, Thompson, and Simm (2000) obtained lower value (0.049) and some authors (Estrada-León, Magana, & Segura-Correa, 2008;M'hamdi et al., 2010; Ríos-Ultrera, Calderón-Robles, Rosete-Fernández, & Lagunes-Lagunes, 2010a) report higher repeatability values (0.120-0.180).Repeatability for GL was estimated to be 0.181, which was lower than estimated repeatability by Johanson et al. (2011) (0.54).For NI, the repeatability estimate was 0.115, higher than those reported by Kadarmideen et al. (2000), Demeke et al. (2004), Estrada-León et al. (2008) and M'hamdi et al. (2010) which were in a range between 0.022 and 0.08.The estimate of repeatability for SF in this study (0.021) was lower than obtained result in the Holstein breed by Jamdar and Eskandarinasab (2014) (0.077).For IO, the repeatability value found for Holstein dairy cattle (0.411) was higher than those reported by Ríos-Ultrera, Calderón, Rosete, and Lagunes (2010b) and Zambrano and Echeverri (2014) which were 0.03 in Brown Swiss cattle and 0.076 in Holstein-Jersey, respectively.Finally, the estimated repeatability for CBW was 0.142, which was lower than 0.29 obtained by Johanson et al. (2011).The low repeatability estimates obtained for most traits suggest that fertility traits are strongly influenced by temporary environmental factors.It would thus be possible to improve fertility performance through improvement in herd management.This fact suggests that in making decision for culling cows, reproduction performance should take less weight in comparison with production traits, which are considerably more repeatable.The genetic correlation between DO and CI in this study was close to 1 (0.98).This value was similar to those reported by González-Recio and Alenda (2005), Gredler, Fürst, and Sölkner (2007) and Ghiasi et al. (2011), who reported a nearly perfect genetic correlation (0.99, 0.98, and 0.99, respectively).The genetic correlation between DO and NI was high and favorable (0.88).Nearly equivalent results were reported by González-Recio and Alenda (2005) and Pozveh et al. (2009): 0.94 and 0.83, respectively.The joint analysis of CI and NI indicates that the genetic correlation between these 2 traits was 0.88, consistent with those reported by González-Recio and Alenda (2005) and Eghbalsaied (2011) (0.89 and 0.81, respectively).These results suggest that these reproductive traits (DO × CI, DO × NI and CI × NI) are almost genetically equivalent, i.e., they are influenced by the same genes.This is known as pleiotropic effect.The genetic correlation between DO and GL was -0.2.The result obtained in this study was lower than -0.36 reported by Eghbalsaied (2011).For Holstein cows, the genetic correlation between DO and SF was estimated -0.48, which was lower than values obtained by González-Recio and Alenda (2005) and Ghiasi et al. (2011) (-0.94 and -0.83, respectively).The genetic correlation between CI and GL was -0.04 that higher value was reported by Eghbalsaied (2011) (-0.07).The genetic correlation between GL and NI (-0.09) was lower than obtained result by Eghbalsaied (2011) (-0.88).Also, the genetic correlation of GL and CBW was obtained low and near zero (0.007), which was lower than value reported by Johanson et al. (2011) (0.52).For Holstein cattle, the genetic correlation between CI and SF that we obtained was medium and negative (-0.55), similar to the finding obtained by González-Recio and Alenda (2005), who reported a correlation of -0.59 between these 2 traits.Likewise, the genetic correlation between NI and SF was middle and negative (-0.49) that higher value was reported by Kadarmideen et al. (2003) (-0.92).The genetic correlation between NI and IO obtained in this study was nearly high and negative (-0.77), similar to the result reported by Ghiasi et al. (2011) (-0.73).These results suggest that these traits (CI × SF, NI × SF and NI × IO) are not genetically favored, as could be logically expected.The genetic correlation of SF and IO was 0.36, whereas González-Recio and Alenda (2005) and Ghiasi et al. (2011) reported higher values (0.94 and 0.83, respectively).These results suggest that these 2 reproductive traits were essentially the same indicator of fertility and may be originally the same in terms of genetic source.The genetic correlation between CI and IO was -1.The result obtained in this study is similar to those obtained by González-Recio and Alenda (2005), Ghiasi et al. (2011), who both reported a correlation of -0.99 between CI and insemination outcome.Also, the genetic correlation obtained in this study between DO and IO was high and negative (-0.98).This is consistent with the results reported by González-Recio and Alenda (2005) and Ghiasi et al. (2011) (-0.99 in both reports).The estimated genetic correlations among CI × IO and DO × IO indicated that selection for cows with high insemination outcome could lead to shorten DO and CI.Therefore, they could be used as one of the best indicators for cow fertility.This would enable efficient selection for better reproductive performance.The phenotypic correlation between CI and DO in this study was high and positive, with value of 0.99.González-Recio and Alenda (2005), Ghiasi et al. (2011), Eghbalsaied (2011) and Zambrano and Echeverri (2014) reported similar estimates for Holstein dairy cattle (0.91, 0.95, 0.99 and 1 respectively).Likewise, the phenotypic correlation between NI and CI was high (0.83).El Amin, Simerl, and Wilcox (1986), Kadarmideen et al. (2000), Ageeb and Hayes (2000), González-Recio and Alenda (2005) and Ghiasi et al. (2011) reported lower values (0.01, 0.69, 0.05, 0.68 and 0.70, respectively).Similarly, the result obtained in this study regarding the phenotypic correlation between NI and DO was high (0.83).This result was higher than reported by El Amin et al. ( 1986), Ríos-Ultrera, Calderón-Robles, Rosete-Fernández, and Lagunes-Lagunes (2010c) and Ghiasi et al. (2011) (0.01, 0.56 and 0.73, respectively).The reason of these high correlations may be due to the high use of estrus synchronization programs in warm and temperate climate condition.The phenotypic correlation between IO and NI obtained in this study was perfect and negative (-1), and it was higher than results obtained by González-Recio and Alenda (2005) and Ghiasi et al. (2011) (-0.75 and -0.73, respectively).The obtained higher and negative correlation may also be explained by the factor discussed above.The phenotypic correlation between IO and CI was -0.99, similar the values reported by González-Recio and Alenda (2005) and Ghiasi et al. (2011) (-0.91 and -0.95, respectively).Likewise, the phenotypic correlation between IO and DO, the value obtained in this study was -0.99, which is consistent with the results reported by González-Recio and Alenda (2005) and Ghiasi et al. (2011) (-1 and -0.98, respectively), which indicate that the degree of phenotypic association between these 2 traits is high and negative.In relation to the phenotypic correlation between IO and SF, the value obtained in the present study was perfect and favorable associated, presenting 1 in the population studied.This result was higher than findings of González-Recio and Alenda (2005) and Ghiasi et al. (2011) who reported 0.61 and 0.55, respectively.In relation to the phenotypic correlation between GL and CI, the value obtained in this study was 0.18, which was higher than reported value by Pozveh et al. (2009) (0.002).As for phenotypic correlation between NI and GL, the value obtained was 0.15.This result was positive and higher than obtained value by Eghbalsaied (2011) (-0.13).The phenotypic correlation between GL and DO was 0.13, unlike the lower value reported by Pozveh et al. (2009) (0.003).The phenotypic correlation between SF and NI was -0.17, which is lower than result obtained by Kadarmideen et al. (2003), González-Recio andAlenda (2005) and Ghiasi et al. (2011) (-0.69, -0.76 and -0.7, respectively).Also, phenotypic correlation between SF and CI was low and negative associated with value of -0.07.Kadarmideen et al. (2003), González-Recio and Alenda (2005) and Ghiasi et al. (2011) reported phenotypic correlation values which were moderate and negative (-0.54, -0.54 and -0.53, respectively).Similarly, phenotypic correlation between SF and DO was low and negative (-0.09) which was lower than results of González-Recio and Alenda (2005), and Ghiasi et al. (2011) (-0.61 and -0.55, respectively).Low phenotypic correlations between SF × NI, SF × CI and SF × DO may be due to the fact that present study was carried out in region with warm and temperate climate, unlike most of the studies which were carried out in zones with a subtropical climate.

Conclusion
According to the results, this study confirms that reproductive traits present low heritability, < 10% in most cases, in Holstein dairy cattle, suggesting that fertility is affected mostly by the environment.Therefore, good management of the fertility traits in heat stress condition must be considered in order to improve reproductive efficiency.The high and negative genetic correlations for DO × IO and CI × IO, and high and positive genetic correlations for DO × CI, DO × NI, and CI × NI in this study suggest that these reproductive traits are genetically equivalent, i.e., they are influenced by the same genes.This obviously favors the selection of these traits, as we can predict what will happen to several of the reproductive traits after performing selection on one of them.In this manner, we can integrate information on different traits to propose more efficient selection strategies.

Table 3 .
Heritability reported by several authors for fertility traits.