EFFECT OF NON-GENETIC FACTORS ON LONGEVITY TRAITS IN SIMMENTAL COWS

. The effects of fixed non-genetic factors (farm, season of birth, year of birth, total number of lactations) and a continuous non-genetic factor (age at first conception) on the expression and variability of longevity traits such as age at culling, length of productive life, days in milk and cow efficiency index were investigated in 2548 Simmental cows in three farming areas. Based on the model used for the analysis of the effects of non-genetic factors, including the environment and cow age at first conception, on the expression and variability of longevity traits, the overall means for age at culling, length of productive life, days in milk and cow efficiency index were 2445.21 ± 17.49 days, 1562.55 ± 17.71 days, 1094.17 ± 12.28 days and 58.68 ± 0.32%, respectively. The effect of farming area, year of birth and lactation group on longevity traits was very significant (P<0.01), whereas the effect of season of birth was significant (P<0.05). Age at first conception had a highly significant (P<0.01) effect on age at culling, length of productive life and cow efficiency index, and no significant effect on days in milk (P>0.05). Based on the model used, the coefficients of determination (R 2 ) were very significant (P<0.01) for all longevity traits, and ranged from 0.898 for age at culling to 0.959 for days in milk.


Introduction
Research into phenotypic and genetic variability of production traits has an immense practical importance as this variability underlies annual and generational selection effects.
Selection work in dairy herds in the last 50 years has strongly focused on milk performance traits. This directselection based on milk performance traits has led to deterioration of fertility, longevity and immunity traits, which have a critical role in profitable milk production. As reported by the United States Department of Agriculture, until 1994, the official selection index used in the selection of dairy cattle included only milk performance traits in calculating the aggregate genotype (Van Raden, 2004).
In the last decade of the 20th century and the first decade of the 21st century, many studies (Allaire and Gibson, 1992;Dekkers, 1993;Smith et al., 2000;Harder et al., 2006;Orpin and Esslemont, 2010) showed a negative effect of a high level of involuntary cullings of animals from the herd on the economic efficiency of milk production on farms. Longevity as a function of animal's constitution and resistance to diseases, along with high productivity and good composition of milk, is an important trait for the economic efficiency of milk production (Pogačar et al., 1998). Consistently with the results of these studies and to solve problems regarding involuntarycullings, a large number of countries have included functional traits in their models for evaluating the breeding value of animals (Sewalem et al., 2008).This approach has led to breeding programmes which have adopted a more balanced approach, with the selection focus moved from milk performance traits to functional traits (Miglior et al., 2005). In terms of milk production, functional traits have today become crucial in defining modern breeding programmes aimed at obtaining productive cost-efficient animals. Their basic effect on the economic efficiency of milk production is not achieved through direct increase in animal productivity, but is evidencedby production cost reduction through animal health, resistance, vitality and ability to last a number of lactations (Bogdanović et al., 2012).
From its earliest beginnings, research on longevity traits has faced difficulties with both their proper interpretation and understanding the processes which affect them. The main problem in the analysis of longevity traits is that the exact value for this group of traits can be obtained only after the animal has been culled, due to which the generational interval is considerably prolonged and the selection effect reduced. In addition, functional traits often have low variability (occurrence of mastitis, calving ease, diseases of some organs, etc.), which significantly complicates their analysis and proper interpretation of results. As reported by Fuerst and Sölkner (1997), many studies have shown that the true productive lifespan cannot be used as an indicator of cow's biological capacity. Therefore, careful attention has been given to functional longevity (yield adjustment) as a better trait to be used in routine genetic evaluation. In Austria, the breeding value of the functional length of productive life has been routinely evaluated since June 1995. These programmes are based on the Cox proportional hazards model.
Longevity and lifetime milk production are significantly affected by a number of non-genetic factors, either discontinuous i.e. fixed (farming area, season Effect of non-genetic factors on longevity… 231 of birth, year of birth, season of calving, year of calving, total number of lactations and their interactions) or continuous i.e. non-categorical factors (age at first conception or age at first calving). In many herds, major reasons for culling include low milk production (30-35%), impaired physiological functions (30-40%) andudder diseases (10-15%). Low production is the main culling reason for cows after the first or second lactation, whereas older cows are commonly culled for infertility.
Cattle breeding practices generally involve the use of linear methods and models combining fixed parameters (year, farm, season, lactation) and random variables (age at first conception or age at first calving, genetic effect of father, genetic effect of an individual animal, etc.), which can be either mutually dependent (related) or independent, depending on the trait analysed, with or without interactions. The model selected is essentially the breeding value of the individual (Bogdanović et al., 2003).
The objective of the present research was to contribute to a better understanding of variability of longevity traits in Simmental cattle in Serbia caused by different non-genetic effects. This research can serve as the basis of initial endeavours in including this group of traits in the breeding programme for Simmental cattle in Serbia, as well as in evaluating the breeding value of bulls for these traits.

Material and methods
The effect of non-genetic factors on the phenotypic expression and variability of longevity traits was analysed in 2548 Simmental cows born from 1995 to 2008, which were housed in three farming areas: 1. Zlatiborski Suvati (Mt. Zlatibor Pastures) Dairy Cow Farm on Mt. Zlatibor (n=502), loose housing system, 2. Dairy cow farm at the Dobričevo farm in Ćuprija (n=956), tie-stall housing system, and 3. Private dairy cow farms in the region of Kotraža (n=1090), tie-stall housing system.
The distribution of data on cows across classes of major non-genetic effects is presented in Table 1. Based on the data obtained from birth records of cows housed in three farming areas, the following longevity traits were analysed: -age at culling (days), -length of productive life (days), -days in milk, -cow efficiency index (%).

Preparation of data for statistical analysis
Longevity traits were calculated using data obtained from birth records of cows, as follows: Age at culling -by summing up the length of productive life in days and age at first calving.
Length of productive life -by subtracting age at first calving in days from age at culling. Days in milk -by summing up the lengths of all full lactations during the cow's lifetime i.e. productive life. Cow efficiency index (%) as a relative measure of longevity, calculated by dividing the length of productive life by cow's age at culling, and multiplying by 100.

Non-genetic effect of the environment
The effect of the following non-genetic factors on lifetime production traits was analysed: Farming area (farm): The research was conducted at three farming locations, two in the uplands (dairy cow farm on Mt. Zlatibor, which employed loose housing system, and private farms, which used tie-stall housing systems, in Kotraža) and a farm at Dobričevo in the low lands, which employed tie-stall housing system.

Evaluation of non-genetic effects of the environment
The effect of non-genetic environment-related factors on longevity traits (arithmetic means (X), standard errors of arithmetic means (Sx), significance of the effect of non-genetic factors and coefficients of determination (R 2 )) was analysed using the general linear model of the Statistica 6.0 statistical software. This procedure ensures a simultaneous analysis of a number of different effects, either categorical factors (farming area, total number of lactations, season of birth and year of birth) or continuous factors (age at first conception).
The analysis of the effect of some non-genetic factors on lifetime production traits was performed by the following model: Further analysis of longevity traits presents coefficients of determination (R 2 ) for each trait analysed, which are proportions of residual variance i.e. variance of the model divided by 100.

Results and discussion
The results of the analysis of the effect of non-genetic environment-related factors on longevity traits (cow's age at culling, length of productive life, days in milk and cow efficiency index) are presented in Table 2. Depending on the effect of non-genetic factors (farming area, season of birth, year of birth, total number of lactations and cow's age at first conception), the table presents arithmetic means and standard errors of arithmetic means (X±Sx),the significance of the effect of non-genetic factors and coefficients of determination (R 2 ) for all longevity traits analysed.
The average age at culling and length of productive life in all tested cows were 2445.21±17.49 and 1562.55±17.71 days, respectively. The total number of days in milk was 1094.17±12.28, and cow efficiency index was 58.68±0.32%.
Farming area or farm commonly has a significant effect on longevity traits due to housing system, nutrition, animal care, climate, age structure, herd size and other factors associated with farm operation and management practices. The interaction of these factors is characteristic of each farm. Hence, there are differences across herds and farms even when the genetic potential for milk production is similar. The effect of farming area on all longevity traits in this research was very significant (P<0.01). The highest expression of these traits was observed at the farm on Mt. Zlatibor as the result of loose housing system and grazing during the summer season, whereas their lowest values were recorded for cows at the Dobričevo farm, which employed tie-stall housing without grazing. The values of these traits in cows in the Kotraža region were higher than Dobričevo cows, as the result of more relaxed farming conditions, including a small number of animals and more extensive farming, which favoured the longevity traits analysed. The effect of season of birth on longevity traits in cows is the result of changes in climate, primarily temperature and precipitation, leading to changes in housing system (barn-based and pasture-based) and cow health as major factors determining the expression of these traits. Longevity traits were significantly affected by season of birth (P<0.05). The highest values were recorded for cows born in the summer season, and the lowest for cows born in autumn. The most possible reason lies in the fact that calves born in the summer season were initially raised in pastures, as opposed to calves born in autumn which were raised in poorer barn-based conditions immediately after calving, which adversely affected their immunity and health in general, as well as the expression of longevity traits.

Table 2. Arithmetic means (X), standard errors of arithmetic means (Sx), significance of the effect of non-genetic factors and coefficients of determination (R 2 ) for longevity traits
The effect of year of birth on the expression of longevity traits is evidenced through climate conditions across production years and feed quality and quantity. Moreover, farming technology and health management have been improving over the years, and there has been more or less annual selection success. Year of birth in the present experiment had a very significant (P<0.01) effect on longevity traits (age at culling, length of productive life, days in milk and cow efficiency index). Despite improvement in farming technology and health management, longevity traits across years of birth in this research exhibited a decreasing trend. This was associated with increased selection criteria and intensified production, which directly affected the relationship between planned and unplanned cullings and, hence, the expression of longevity traits and milk production profitability. Most cows are culled from production for undesirable reasons, which can be up to 2.5 times more common than planned culling reasons (Pinedo et al., 2010). Major causes of involuntary culling include reproductive disorders, mastitis, hoof and leg diseases and injuries (Nienartowicz-Zdrojewska et al., 2009;Pinedo et al., 2010;Chiumia, 2011;Ansari-Lari et al., 2012;Stojić et al., 2012).
The effect of total number of lactations on longevity traits was very significant (P<0.01). The increase in lactation numbers directly and indirectly led to an increase in longevity traits. The same finding was reported by Petrović D. M. et al., (2004). In 143 Simmental cows housed at the Zlatiborski Suvati (Zlatibor Pastures) farm, the author used the general linear model procedure and determined a very highly significant (P<0.001) effect of total number of lactations on longevity traits such as age at culling, length of productive life, days in milk, and cow efficiency index. The analysis of the effect of lactation number on culling reasons showed that dominant culling reasons varied with lactation. Dominant reasons for culling of first-calversare selection issues (primarily low production) and reproduction problems (Seegers et al., 1998;Stojić et al., 2012). First-calvers account for 20-35% of the total number of animals culled per year (Maher et al., 2008;Pinedo et al., 2010;Chiumia, 2011). First-calvers that become pregnant late for the first time and those that have less expressed standard characteristics are culled earlier due to the higher risk of low production (Dürr, 1997).
Age at first conception had a highly significant effect (P<0.01) on age at culling and length of productive life. The coefficient of linear regression was b xy =0.992 for the first trait, and negative b xy =-0.993 for the second trait. Age at first conception had no significant effect (P>0.05) on days in milk and cow efficiency index.
The effect of age at first conception or age at first calving on longevity traits was examined in a number of studies. In their research on the regression effect of non-genetic factors on longevity traits in Simmental cows, Petrović D.M. et al., (2004;, found a highly significant (P<0.001) effect of age at first conception on age at culling and cow efficiency index (b xy =1.386 andb xy =-0.020), whereas their effect on length of productive life and days in milk was nonsignificant (P>0.05). Considerable research has documented that the risk of culling increases (i.e. animals are culled earlier) with increasing age at first calving (M'hamdi et al., 2010;Raguž, 2012;Zavadilová and Štípková, 2013). When evaluating the effect of age at first calving on functional longevity in Hostein cows in Tunisia, M'hamdi et al., (2010) found that culling risk increased linearly with increasing age at first calving. Specifically, age at first calving of 27 months was set as basal age, with the risk set to 1 for this age. In animals that calved at an average age of 21 months, the risk was 0.96, whereas the risk for the calving age of 39 months was significantly hither -1.43.
Coefficients of determination, which indicate the degree of explanation of variations of longevity traits by the model employed, were very high, above 0.9 i.e. over 90% of variability in longevity traits was induced by the effect ofnon-genetic environment-related factors and age at first conception. Such high coefficients of determination for longevity traits were mostly due to total number of lactations. The lowest coefficient of determination 0.898 (89.8%) was obtained for age at culling, and the highest 0.951 (95.1%) for days in milk. Somewhat lower coefficients of determination for longevity traits, due to fewer non-genetic factors analysed, were reported by Petrović D. M. et al., (2004). In their study on the effect of lactation number and season of calving as fixed non-genetic factors and age at first conception as a continuous factor on longevity traits in 143 Simmental cows, the coefficients of determination ranged from 0.823 for cow efficiency index to 0.894 for days in milk.

Conclusion
Based on the model for analysing the effect of non-genetic environmentrelated factors and age at first conception on the expression and variability of longevity traits, the following conclusions are drawn: The general mean values for age at culling, length of productive life, days in milk and cow efficiency index in 2548 tested cows were 2445.21±17.49 days, 1562.55±17.71 days, 1094.17±12.28 days and58.68±0.32%, respectively.
The effect of farming area, year of birth and lactation group on longevity traits was very significant (P<0.01), whereas the effect of season of birth was significant (P<0.05).
Age at first conception as a continuous factor had a highly significant (P<0.01) effect on age at culling, length of productive life and cow efficiency index, and no significant effect on days in milk (P>0.05).
The coefficients of determination (R 2 ), obtained by the model, were very significant (P<0.01) for all longevity traits and ranged from 0.898 for age at culling to 0.959 for days in milk.
Given that non-genetic factors, both fixed and continuous, had a generally highly significant effect on longevity traits, and were largely responsible for their variations, they should necessarily be included in models for evaluating the breeding value of dairy cows i.e. these traits should be corrected for the effect of these nongenetic factors.