Risk factors affecting the culling of Iranian Holstein dairy cows

Introduction

Culling is defined as the departure of cows from the herd because of sale, slaughter, salvage, death, age, and fertility (Fetrow et al., 2006). Culling of dairy cows causes financial losses, including decreased milk production, cow selling price, and extra labor costs (Raboisson et al., 2011), and in most cases a higher cost of replacing the culled cows (Fetrow et al., 2006). Culling rate has been rising recently due to sub-optimal health and inadequate welfare, structural factors and management practices (Smith et al., 2000; Thomsen et al., 2006 and McConnel et al., 2008).  Factors that influence culling decisions may include herd, location (state or region), time or cow-related characteristics such as calving season, lactation number, milk production (kg), milk fat (%), milk protein (%) and the breed (Booth et al., 2004). Some studies have discussed risk factors for culling mainly focused on herd characteristics and management style (Shahid et al., 2015). However, very few studies investigated the association between cow attributes and culling risk. Shahid et al. (2015) reported that the risk of culling increased with higher milk fat percent, cows with male calves, cows carrying multiple calves, increased parity, increased calving difficulty score, breed and herd size. They also showed that cows in herds with higher milk yield had lower mortality. Alvasen et al. (2012) and Raboisson et al. (2011) also reported that culling of dairy cows increased with larger herd size and with the proportion of purchased cows but decreased with increasing milk yield. Besides, at the cow level, higher mortality was associated with lower milk yield (Pinedo et al., 2010) and increasing parity (Raboisson et al., 2011). Quantification of the association between risk factors and culling is useful for making culling decisions and helps motivate efforts to reduce or eliminate the risk factors (de Vries et al., 2010). Therefore, the aim of this study was to investigate the effects of risk factors associated with the culling rate in Iranian Holstein dairy cows.

Materials and methods

Data

Data collected by the Animal Breeding Center of Iran, and including all calving records from March 21^st, 2003 to March 20^th, 2012 were used. Records with missing parity number, calving and culling date and production data in each year were omitted. Cows with calving intervals lower than 300 and greater than 600 d, and cows with first calving before the age of 20 or after the age of 40 months and with parity more than 12 were excluded. Also, the data of herds with less than 10 milking cows per year were removed from the study.

Outcome and predictor variables

The outcome variable was the risk of culling regardless of reason, and the predictor variables included the calving season [spring (21 March to 21 June), summer (22 June to 22 September), fall (23  September to 21 December) and winter (22 December to 20 March)], birth type (single or twins),  calving ease scores: [(1 = normal (not assisted), 2 = moderate assistance (farmer), 3 = moderate assistance (veterinarian called as precaution), 4 = difficult (extraction by farm staff) and 5 = very difficult calving (veterinary assistance)]- because the number of cows with score 5 was low (less than 0.2%), categories 4 and 5 were merged as 4 = very difficult, parity (1, 2, 3, 4, 5 and ≥ 6 ),  three geographic (Unesco, 1979) regions [ North and West  (temperate or cold and wet, annual rainfall more than 350 mm and mean temperature between -2.5 to 25 °C ), Central (warm and semi-arid , annual rainfall from200 to 350 mm and mean temperature from 5 to 32.5 °C ), and East and South (Warm and dry, annual rainfall less than 200 mm and mean temperature from7.5 to 35 °C )], age at first calving with 4 categories based on quartiles ( ≤727 d, 728 to 762 d, 763 to 815 d and  ≥816 d), the number of calving cows with 4 categories (≤ 25,  26 to 53, 54 to 125 and  ≥126), and the standardized 305-d milk  production with categories based on quartiles (≤ 7354, 7354 to 8596, 8596 to 9872, and ≥ 9872 kg).

Statistical analysis

The logistic regression was used to analyze the associations between predictors and response variable. The effect of variables of interest on the odds ratio (OR=P/1-P) and risks of culling was estimated by using the logit-link in GLIMMIX procedure (SAS, 2008).

The GLIMMIX procedure fits statistical models to data with correlations or no constant variability and where the response is not necessarily normally distributed and assume normal (Gaussian) random effects.

The logistic regression model used for analysis was:

where the response variable was risk of culling (y_i), takes on the values either 0 or 1, and the predictor variables were calving season (x_i1), twins (x_i2), calving ease (x_i3), standardized 305-d milk yield (x_i4), age at first calving (x_i5), region and number of calving cows (x_i6), parity (x_i7), and ε_i  is the error term.

Results and discussion

Descriptive statistics

After editing, 329,772 calving records were available for the years 2003-2012. The number of herds increased from 300 in 2003 to 730 in 2012 and median of calving cows increased from 47 in 2003 to 61 in 2012. Descriptive statistics for the number of calving cows, number and percentage of culled cows, and their distribution are shown in Table 1. The highest percentage of calving cows during different year was found in parity 1, and the highest percent of culling occur in parities 1 and 2. Between seasons, the highest calving occurred in summer and fall, and the highest culling in winter. The percentage, number of culled cows and the ratio of culling are shown in Fig. 1 to 3.

Table 1. Descriptive statistics and distribution of studied variables in Iranian Holstein dairy cattle through the years.

¹1=North and West (temperate or cold and wet, annual rainfall more than 350 mm and mean temperature from -2.5 to 25 °C), 2= Central (warm and semi-arid, annual rainfall from 200 to 350 mm and mean temperature from 5 to 32.5 °C), and 3= East and South (Warm and dry, annual rainfall less than 200 mm and mean temperature from 7.5 to 35°C).

²1 = normal (non-assisted); 2= moderate assistance (farmer); 3= moderate assistance (veterinarian called); 4= very difficult.

Parity

The risk of culling (P< 0.001) increased with parity. The OR of culled cows were 1, 1.32, 1.55, 1.62, 1.67 and 1.88 for parities 1 to 6, respectively (Table 2). These results agrees with de Vries et al. (2010), who reported increases in culling rate with parity; with the risk of culling for culled cows being 1, 1.51, 2.14, 2.68, 3.11, and 3.46 for parities 1 to 6, respectively. Studies showed that higher mortality was associated with increasing parity (Miller et al., 2008; Pinedo et al., 2010; Raboisson et al., 2011).  The reason for increased culling rate with parity may be the incidence of certain diseases in older cows (Shahid et al., 2015). 

In other studies, culling risks also increased linearly with parity in dairy cattle in the United States (Pinedo et al., 2010), and in Danish herds risk of mortality was higher for parity ≥3 compared with cows in the first and second parity (Thomsen et al., 2004; Miller et al., 2008). Thomsen and Houe (2006) showed that the most often reported disposal code in parity 3 and ≥ 4 was death. Also, Thomsen et al. (2004) and Miller et al. (2008) reported that higher parity cows were at a greater risk for live culling and death. Azizzadeh (2011) showed that in Holstein-Friesian dairy cows in Iran, the risk of culling significantly increased with parity and cows in parity ≥5 were at greater risk for culling compared parity 1, 2, 3 or 4. Ansari-Lari et al. (2012) and Mohammadi and Sedighi (2009) also showed that the age was one of the most important factors for culling Iranian Holstein cows, and that culling rate significantly increased in old cows.  In contrast, Harris (1989) and White and Nichols (1965) reported that there was no significant difference in mortality or culling among cows at different ages. This occurrence may be due to the differences in management or differences in policy and decisions for culling.

Figure 1. Number of culled Iranian Holstein cows through the years

Figure 2. Percentage of culled Iranian Holstein cows through the years

Figure3. Culling ratio of Iranian Holstein cows through the years.

Twin and Dystocia

Twinning and calving difficulty (P< 0.001) increased the risk of culling, with the OR being  1.09 for twinning, and 0.89, 1.22 and 1.49 for dystocia score of 2 to 4, respectively (Table 2). Although with increases in the degree of calving difficulty (categories 3 and 4) the risk of culling increased, unexpectedly, the risk of culling in category 2 (farmer assistance) decreased. McConnel et al. (2008) reported that reproductive problems and dystocia were associated with increased mortality levels. Dystocia can increase the risk of reproductive tract diseases, decrease the dry matter intake after calving, and change the hormonal status, resulting in reduced milk, milk fat and milk protein production (Barrier and Haskell, 2011; Atashi et al., 2012), and therefore, can increase the risk of culling.

 Shahid et al. (2015) reported that mortality risk was greater for cows with calving difficulty scores 3, 4 and 5, respectively, compared with cows without calving difficulty. Dematawewa and Berger (1998) showed that cows with calving difficulty score 5 had 4% higher mortality than cows with score 1(calving without assistance), and Bicalho et al. (2007) reported about 20% higher risk of culling with more severe calving difficulty of scores to 3 and 4 compared with scores 1 and 2. The lowest culling risk of cows recorded in category 2 may be due to presence of an assistant at parturition, provision of parturition confinement, better treatment and nutrition and therefore produce more milk compared with the other groups (Eaglen et al., 2011) and error in classification by the farmer.

 de Vries et al. (2010) reported that twinning cows had higher calving difficulty scores and were more likely to be culled.  Also, Shahid et al. (2015) and Atashi et al. (2012) showed that twinning cows had a higher odds ratio for dystocia, and mortality risk was 7% greater among the cows that had twins than cows with singleton birth. Bicalho et al. (2007) also concluded that cows that calved twins had 42% higher hazard of death or culling than those with singletons. Higher risk of mortality among twinning cows  may be due to the problems such as increased dystocia, retained placenta, metritis, displaced abomasum, and ketosis (Bell and Roberts, 2007; Fricke, 2001).

Region

The region had a significant effect (P< 0.001) on culling risk (Table 2), where the cows in category 2 and 3 (warm and dry and semi-arid regions) had a higher risk of culling than cows in category 1 (temperate or cold and wet region). Smith et al. (2000) reported different culling rate according to region in the US and showed that culling percentage for Holstein herds was higher in the Southern regions than the Northern ones. They also reported that herds in the Southern region had a higher percentage of cows leaving the herd (36.3%) than herds in the Southern region (34.5%) (P < 0.05). Alvasen et al. (2012) studied the regional differences in mortality rate and reported that the length of the legislated pasture season may be the reason of such differences. Incidence of disease, differences in physical environment, nutrition, and management factors and weather condition, other management practices such as feeding  a total mixed diet, balance rations ( McConnel et al, 2008), different feeds, housing and price of feeds and higher cost for maintenance may be associated with different level of culling in regions in this study.

Season

Our results showed that cows calved in hot seasons (spring and summer) had a higher (P< 0.001) risk of culling than cows calved in cool seasons (fall and winter). The OR for culling in summer, fall and winter were 0.87, 0.79 and 0.77, respectively (Table 2). Calving season played a significant role in contributing to the likelihood of a cow being culled in a study by Hadley et al. (2006). These results are in agreement with the results of Dechow and Goodling (2008) and Alvasen et al. (2012) who reported that the unfavorable effect of season on cow survival was more pronounced during the hottest calving months. Also, Pinedo and de Vries (2010) and Miller et al. (2008) reported that cows calved during spring and summer had a greater risk of death and live culling compared with the cows calving in fall and winter. Diseases can also increase the risk of involuntary culling in dairy cows. Madadzadeh et al. (2013) showed that some diseases like claw lesions occurred with greater frequency during the warm season and therefore could increase the risk of culling. In contrast to our results, Pinedo et al.  (2014) reported no clear association between culling and calving season, and Booth et al. (2004) and Hadley et al. (2006) showed that cows calving during summer and autumn had reduced culling risk compared with those calving in winter, and that the culling risk for spring-calving cows was not different from that for winter-calving cows. Climatic differences characterizing the season affect cow health, milk production level, feed availability, milk price, and profitability, and thus can affect the variation in culling risk (Hadley et al., 2006). Seasonal variations in milk production, conception and milk price also can affect culling rate in dairy herds (Delorenzo et al., 1992).

Number of calving cows

The OR of culling increased with increasing number of calving cows (Table 2). These results confirmed the findings of Alvasen et al. (2012) who reported that culling rate increased with herd size and the largest herd size group had a greater culling rate than other herd size groups. Furthermore, Thomsen et al. (2006), McConnel et al. (2008) and Pinedo et al. (2010) concluded that mortality increased with herd size, but Batra et al. (1971) did not find any relation between herd size and culling risk. Raboisson et al. (2011) explained that increased mechanization, less personal attention, and greater levels of physiologic stress could explain the higher culling risk in large compared with small herds. According to Norgaard et al. (1999), average milk production, concentrate consumption and less time available to spend on individual cows could also be involved.

Standardized 305-d milk yield

Significant (P< 0.001) decrease in culling risk was observed with increasing standardized 305-d milk yield (OR 1, 0.83, 0.70 and 0.51 for category 1, 2, 3 and 4, respectively) (Table 2). Pinedo et al. (2010) and Shahid et al. (2015) reported similar association between milk yield and culling rate. They showed that at the cow level, higher culling was associated with lower milk yield. Pinedo et al. (2010) investigated the effects of 305-d milk production on live culling rate and death and reported that annualized culling rate decreased with increasing 305-d milk yield. Sanjabi et al. (2013), Davasazetabrizi (2012) and Mohammadi and Sedighi (2009) showed that an important reason for voluntary culling of Holstein dairy cows in Iran was low milk production. Alvasen et al. (2012) stated that milk yield was associated with the level of management, high average milk yields were achieved with good management, and the farmer who is able to obtain a high milk production probably does many other things right, thereby reducing the risk of culling. Also, Dematawewa and Berger (1998) explained that preferential treatment by the farmer for high-yielding cows made their culling risk lower than those for low-producing cows. Beside these, cows with lower milk production are less profitable, and therefore, the risk of culling of these cows increases.  

Age at first calving                   

Cows with the age at first calving of about 24 months had a lower risk of culling with the risk of culling being increased with increasing age at first calving. Heinrichs (1993) suggested that optimum age at first calving in Holstein cows for achieving maximum profit should be 23 to 24 months. Our results are agreement with Zavadilova and Stipkova (2013) who found that the highest risk ratio of culling occurred for the cows that were older at first calving. The reason for this association may be better productive and reproductive performance in cows calving at about 24 months of age. Vukasinovic et al. (2001) showed a slight increase in culling risk in cows that calved very early or, especially, very late. Similar to our results, Ducrocq (2005) and M’hamdi et al. (2010) observed an in increase culling risk with increased age at first calving but Ducrocq (1994) or Ojango et al. (2005) did not find any significant effect of age at first calving on culling risk.

Table 2. Final logistic regression model, odds ratio, 95% confidence interval and P-value for risk factors in Iranian Holstein dairy cattle.

¹ 1=normal (non-assisted); 2 = moderate assistance (farmer); 3 = moderate assistance (veterinarian called); 4 = very difficult.

²1=North and West (temperate or cold and wet, annual rainfall more than 350 mm and mean temperature from -2.5 to 25 °C), 2= Central (warm and semi-arid, annual rainfall from 200 to 350 mm and mean temperature from 5 to 32.5 °C), and 3= East and South (Warm and dry, annual rainfall less than 200 mm and mean temperature from 7.5 to 35°C).

Conclusions

The risk of culling Holstein cows in Iran significantly increased with parity, twinning, calving difficulty and decreased with increase in the standardized 305-d milk yield. Also, the number of calving cows, calving in hot season and different regions had a significant effect on culling rate. The association between individual risk factors and culling rate suggested that better farm management and greater farming skill may reduce the risk of culling dairy cows.