Introduction
Poor reproductive performance is a major problem on many dairy farms throughout the United Kingdom (UK) and has been identified in two farm surveys in Northern Ireland (1997 and 2003) as the single most important problem in dairy herd management (AgriSearch Farm Surveys, personal communication). The decline in reproductive performance of the dairy herd has been widely reported with various studies around the world citing conception rates to first insemination of approximately 40% (Butler, Reference Butler1998; Royal et al., Reference Royal, Darwash, Flint, Webb, Woolliams and Lamming2000; Mayne et al., Reference Mayne, McCoy, Lennox, Mackey, Verner, Catney, McCaughey, Wylie, Kennedy and Gordon2002), though this figure tends to be higher in grass-based production systems (Buckley et al., Reference Buckley, O'Sullivan, Mee, Evans and Dillon2003; Grosshans et al., Reference Grosshans, Xu, Burton, Johnson and Macmillan1997). At herd level, decreased individual cow fertility, with more services per conception, has been reported in cows producing more than the herd average (Wicks and Leaver, Reference Wicks and Leaver2004) but Darwash et al. (Reference Darwash, Lamming and Woolliams1997) found no such phenotypic effects of yield on fertility. Hoekstra et al. (Reference Hoekstra, Van der Lught, Van der Werf and Ouweltjes1994) reported negative phenotypic and genetic correlations between milk production and fertility traits. In addition to the decline in traditional measures of fertility, Royal et al. (Reference Royal, Pryce, Woolliams and Flint2002) have reported genetic correlations between yield and endocrine parameters such as commencement of luteal activity. While some studies attribute the decline in reproductive performance to the direct genetic effect of increased genetic merit for milk production (Buckley et al., Reference Buckley, Dillon, Rath and Veerkamp2000; Kadarmideen et al., Reference Kadarmideen, Thompson and Simm2000, Pryce and Veerkamp, Reference Pryce, Veerkamp and Diskin2001; Royal et al., Reference Royal, Pryce, Woolliams and Flint2002) and specifically the increasing proportion of Holstein genes (Buckley et al., Reference Buckley, Dillon, Rath and Veerkamp2000 and Reference Buckley, O'Sullivan, Mee, Evans and Dillon2003), others cite the indirect effects resulting from higher milk production (Berry et al., Reference Berry, Buckley, Dillon, Evans, Rath and Veerkamp2003) or negative energy balance in early lactation as the major cause of the decline (Butler, Reference Butler and Diskin2001). In addition to the direct financial cost, estimated at over £500 million per annum in the UK (Lamming et al., Reference Lamming, Darwash, Wathes and Ball1998), infertility can result in increased management complexity with spread calving patterns and therefore extended breeding seasons with prolonged heat observation and infertility intervention for problem and late calving cows. In Northern Ireland, there is a diverse range of milk production systems ranging from compact spring-calving systems to all-year-round calving systems, but the majority of herds are autumn/winter calving and in these herds infertility causes increased management complexity through the inability to achieve a compact calving pattern (Mayne et al., Reference Mayne, McCoy, Lennox, Mackey, Verner, Catney, McCaughey, Wylie, Kennedy and Gordon2002).
Calving index (the average interval between successive calvings for a group of cows, ignoring cows who fail to conceive) has traditionally been used as the primary measure for assessment of fertility performance in dairy herds as it combines the various contributory factors into one index. Recent results from a fertility study in Northern Ireland have reported a mean calving index of 407 days with a range between herds from 359 to 448 days (Mayne et al., Reference Mayne, McCoy, Lennox, Mackey, Verner, Catney, McCaughey, Wylie, Kennedy and Gordon2002). While information on calving index is readily available from milk records, data analysis is problematical since it only relates to cows that conceive and calve again (Olori et al., Reference Olori, Meuwissen and Veerkamp2002; Veerkamp et al., Reference Veerkamp, Dillon, Kelly, Cromie and Green2002). In the study of Mayne et al. (Reference Mayne, McCoy, Lennox, Mackey, Verner, Catney, McCaughey, Wylie, Kennedy and Gordon2002), 28% of cows were removed or sold of which almost 27% were sold for infertility. Removal rates varied from 17 to 49% between herds, making the comparison of fertility performance using calving index extremely difficult. Fertility performance has also been assessed using conception rate, but this varies according to the method of calculation (non-return rate, pregnancy rate or calving rate).
Kirkland et al. (Reference Kirkland, Ingram, Patterson, Steen, Comerford, Mayne and Keady2003) used an alternative approach for assessing fertility performance in suckler herds. They assessed the proportion of cows that subsequently produced a further calf (re-appeared) within 390 and 450 days of a previous calving with results of 52% and 71%, respectively. The InCalf Project in Australia adopted an alternative approach to take account of variable culling rates for the assessment of fertility performance in dairy herds (Morton, Reference Morton2000 and Reference Morton2001). They have developed in-calf rate, an index that assesses the proportion of cows intended for rebreeding post calving that have conceived, or are back in-calf, by the end of a specific period, either 6 weeks from the start of breeding (in seasonal calving herds) or within 100 days of calving (in all-year-round calving herds). Hence, cows destined for culling or sale are not included in the calculation, so this method of assessing fertility performance gives an early indication of the proportion of cows due to calve again within approximately 380 days of the previous calving.
Reproductive performance is influenced by a large number of factors, including management practices (such as prolonged calving patterns, increased herd size, heat detection rate and insemination technique) and nutritional factors (such as energy balance in early lactation), but the decline in dairy cow fertility is also associated with a period of increased genetic capacity for milk production, attained through substitution of the British Friesian by the North American Holstein. Whilst there is evidence of a negative genetic correlation between milk production and fertility performance (Dematawewa and Berger, Reference Dematawewa and Berger1998; Pryce et al., Reference Pryce, Coffey, Brotherstone and Woolliams2002; Berry et al., Reference Berry, Buckley, Dillon, Evans, Rath and Veerkamp2003), there is a lack of consensus in the literature regarding the effects of increasing production and genetic merit for production on dairy cow fertility at farm level. Hence, the aim of the current study was to examine the relationships between genetic merit and milk production on the fertility performance of dairy cows in a range of dairy herds, representative of production systems in Northern Ireland. In addition to examining the associations between genetic merit and milk production on calving interval and conception rate, alternative analytical techniques for assessing fertility will also be considered.
Material and methods
Selection of participating herds
In autumn 1998, a 3-year monitoring study was initiated involving 19 herds of Holstein/Friesian cows across Northern Ireland, representing a total population of approximately 2500 cows. Herds were identified by local DARD advisers as having a good track record of data recording, a willingness and enthusiasm to participate in the study and being geographically spread throughout Northern Ireland. Twenty herds were selected as representative of those throughout the region and included a wide range of herd size, concentrate input, feeding methods, genetic merit and level of milk production, details of which are provided in Mayne et al. (Reference Mayne, McCoy, Lennox, Mackey, Verner, Catney, McCaughey, Wylie, Kennedy and Gordon2002). One of these herds was later omitted from the study due to farmer illness and not replaced. Between herds, both season and compactness of calving varied considerably, ranging from compact autumn and spring calving herds through to prolonged winter calving and all-year round calving herds.
Collation of data
Collaboration between farmers, local veterinarians, Institute technical staff, milk recording agencies and pedigree breed societies allowed the collation of a comprehensive range of data on herd management, genetic merit, productive and reproductive performance. Fertility data were recorded by the individual farmers including details of all calvings, heats, services and removal of cows from the herd, as described previously (Mayne et al., Reference Mayne, McCoy, Lennox, Mackey, Verner, Catney, McCaughey, Wylie, Kennedy and Gordon2002). Each record included a unique cow number and event date, where appropriate. Calving records included details on lactation number, calving difficulty score and calf type. Calving difficulty score was defined using a scoring system from 0 to 5, where 0 = unobserved and unassisted, 1 = observed and unassisted, 2 = assisted without calving aid, 3 = assisted with calving aid, 4 = vet assisted and 5 = caesarean. Calf type was defined as heifer, bull, twin heifers, mixed twins or twin bulls. Data were collated on a series of spreadsheets that were used to regularly update a central masterfile containing fertility, pedigree and production data using SPSS software (Statistical Packages for the Social Sciences, 2002). This facilitated validation of the data pertaining to years 1, 2 and 3 of the study which included the records of cows which calved in the year from 1 August 1998, 1 August 1999 and 1 August 2000, respectively, and which subsequently calved or were removed from the herds. Data continued to be collated for a year after completion of year 3 and cows that had neither calved again nor been removed were followed up after this time.
Cows were milk-recorded monthly by milk-recording technicians from two milk recording agencies, United Dairy Farmers (UDF) or HerdTech. Data provided included individual cow test-day milk yields (kg) with butterfat (BF) and protein (P) composition (g/kg), 305-day milk, butterfat and protein yields (kg) with butterfat and protein composition (g/kg), and complete lactation milk, butterfat and protein yields (kg) with butterfat and protein composition (g/kg). For individual lactations, test-day records were used to establish peak yield (kg/day) and interval from calving to peak yield (days), protein nadir (g/kg) and interval from calving to protein nadir. Test-day records were also used to calculate 100-day milk, butterfat and protein yields (kg) for each lactation, and butterfat and protein concentrations (g/kg). Milk energy output was calculated from milk yield at 100 days, 305 days and over the complete lactation using the formula:
![milk\,energy\,output\,(MJ) = \,yield\,(kg)\times [0.0376\,(BF\,g/kg)\, + 0.0209\,(P\,g/kg) + 0.948]\,](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20160203071335850-0498:S1751731107257921_eqnu1.gif?pub-status=live)
Tyrrell and Reid (Reference Tyrrell and Reid1965).
Data were also provided by the milk-recording agencies on the pedigree details of individual cows involved in the study and these were validated through the co-operation of herd-owners and Holstein UK (HUK). A database was established for all cows that included the name of both sire and maternal grandsire and individual cow predicted transmitting abilities (PTAs) for milk, butterfat and protein yield, butterfat and protein concentration, and £PIN (expressed on a PIN2000 base), collated from the August 2002 proof run. Since PTAs take account of both herd effects and a cow's previous production, and these were not available for all cows, an alternative approach for examination of genetic merit effects was used instead, viz. pedigree index (PDX). The PDX for milk, butterfat and protein yield, and butterfat and protein concentration was calculated using sire and maternal grandsire PTAs from HUK's August 2002 proof run using the formula:
This formula assumed that maternal grand dams had zero PTA for each trait. Where no maternal grandsire was identified, a cow's PDX was estimated to be half the sire PTA, assuming that both maternal grandsire and maternal grand dam had zero PTA for each trait. Pedigree index (£PIN), the economic genetic index used in the UK for milk production traits that allows ranking according to genetic merit was calculated using the formula:
Collation of fertility data and definitions
Interval to first service (IFS) was defined as the interval from calving to first recorded service. Calving rate to first service (CR-1) was defined as the proportion of cows that conceived and subsequently produced a calf to the first insemination, including also cows that had a positive pregnancy diagnosis prior to removal from the herd in the event of either sale or death. Conception rate to second insemination (CR-2) and first or second insemination (CR-1/2) was calculated on the same basis. Minimum and maximum calving indices were based on gestation lengths of 260 and 299 days respectively equating to a gestation length of approximately 280 ± 20 days. For gestation lengths < 260 days the animal was assumed to have aborted and a calving index was not calculated unless a viable calf was produced. For gestation lengths >299 days it was assumed that the animal conceived to a subsequent non-recorded service, nominally 282 days prior to subsequent calving. While calving index (CI) has traditionally been used to assess herd reproductive performance, variable removal rates between herds (Mayne et al., Reference Mayne, McCoy, Lennox, Mackey, Verner, Catney, McCaughey, Wylie, Kennedy and Gordon2002) and between years within a herd (unpublished data from present study) made direct comparisons of herd fertility difficult. Therefore alternative techniques of assessing herd fertility performance were used, viz, re-appearance rate (RR) and in-calf rate (ICR).
In this study, each calving record was followed by either a subsequent calving (normal or abortion) or removal of the cow from the herd. Re-appearance rate was defined as the proportion of cows with a subsequent calving within 365 days (RR-365) or 400 days (RR-400) of the previous calving. The overall re-appearance rate (RR-overall) was the proportion of calvings that were eventually followed by a subsequent calving and was essentially the inverse of removal rate. The in-calf rate was assessed using the method of Morton (Reference Morton2001) which examines the proportion of cows in year-round calving herds intended for rebreeding that are back in-calf within 100 days of calving (ICR-100). This method was selected due to the predominance of prolonged calving patterns in Northern Ireland dairy herds. For statistical analysis, cows intended for rebreeding were retrospectively classified as having at least one service or culled for infertility, signifying an intention to rebreed. In-calf status at 100 days was determined by the date of last service, based on either subsequent calving or pregnancy diagnosis prior to removal from the herd.
Data analysis
Prior to analysis of the dataset, a 99% confidence interval was established for all linear parameters based on the mean ± 3 standard deviations to filter off extremes of data over which there was some doubt. This included the reproductive parameters (IFS and CI), productive parameters (those concerned with minimum milk protein concentration, maximum yield, 100-day production, 305-day production and total lactation production) and genetic merit parameters (those based on PDX). Records with reproductive parameters outside the 99% confidence interval range were omitted from the dataset, while only complete groups of productive and genetic merit parameters (listed in parentheses above) that fell inside the 99% confidence interval range were submitted for data analysis.
Analysis of the data was conducted using the GenStat statistical programming software (GenStat, 2000), employing a three level random intercepts multilevel model, where year was nested within cow, which in turn was nested within herd. Continuous response variables were analysed using the REML command, while binary variables were analysed using the GLMM procedure. A univariate analysis was carried out relating the full range of reproductive variables, viz. IFS, CI, RR-365, RR-400, RR-overall, ICR-100, removal rate, CR-1, CR-2 and CR-1/2 to both genetic merit and production variables. For continuous response variables the results were reported in the form of a slope and standard error, while for the binary variables they were reported using odds ratios and 95% confidence intervals. The significance level of each relationship was calculated using a Wald test. Where explanatory variables were continuous, the results were reported for a meaningful increase on the scale of the particular variable in question. The relative effect of genetic or production variables on both the continuous and binary measures of fertility was examined by observing the effect of a standard deviation increase in each genetic merit or production variable where significant associations occurred. In addition, a similar separate analysis was employed to examine the effect of season of calving, lactation number, calving difficulty and calf type on three key reproductive variables, viz. CR-1, ICR-100 and RR-400. Records were categorised into a number of groups, and one of these groups was designated as the reference category for odds ratio analysis (OR = 1.0) where an OR >1.0 indicates increased likelihood and an OR < 1.0 indicates decreased likelihood.
Results
A total of 7747 calving records from 4217 different cows were recorded across the 19 herds in the period from 1 August 1998 to 31 July 2001, with 2476, 2559 and 2712 cows calving in years 1, 2 and 3 of the study, respectively. This allowed the establishment of a comprehensive database on a wide range of fertility, milk production and genetic merit parameters. The mean ± s.d. IFS and CI across the entire dataset were 85.2 ± 39.3 and 405.9 ± 71.0 days, respectively, and the mean CR-1 was 40.3%, but there was large inter-herd variation for all reproductive parameters (IFS, herd range = 69.9–126.0 days; CI, herd range = 372.0–446.8 days; CR-1, herd range = 16.6–61.1%). The average RR-365, RR-400 and RR-overall were 24.0% (herd range = 4.2–44.9%), 44.5% (herd range = 21.8–64.9%) and 72.8% (herd range = 62.0–80.3%), respectively. The average ICR-100, based on cows intended for rebreeding was 46.0% (herd range = 16.4–70.8%). The range in fertility, milk production and genetic merit parameters seen in the filtered dataset used for the more detailed statistical analysis is presented in Table 1.
Number of calving records (n) and the basic statistics for a range of parameters in the data set used for statistical analysis
The association of changing genetic merit on the calculated PDX for a range of reproduction parameters are presented in Table 2, where the association of a £10 increase in £PIN, 100 kg increase in milk yield and 0.1% increase in fat or protein concentration is presented. The main associations were that increasing £PIN and PDX for milk yield caused increases (P < 0.001) in IFS, CI and RR-overall and a decrease in removal rate (P < 0.001). Increasing genetic merit for milk yield was associated with an increase in RR-400 (P < 0.05), CR-1 (P < 0.05) and CR-1/2 (P < 0.05). Increasing genetic merit for fat composition was associated with significant decreases in RR-400 (P < 0.001), RR-overall (P < 0.001) and calving rate (P < 0.05 or less), and a increase in removal rate (P < 0.001). Increased genetic merit for protein composition was associated with a decrease in IFS (P < 0.05), CR-1 (P < 0.05) and RR-overall (P < 0.001).
Effect of changing pedigree index on a range of fertility parameters, where the effect of a pre-determined change in each pedigree parameter (that could be reasonably achieved in practice at farm level) was assessed against its effect on a range of continuous and binary fertility parameters
† Dependent (fertility) variable mean from random effects multi-level model for continuous variables with a response curve.
‡ Independent (genetic merit) variable mean from random effects multi-level model for binary variables where odds ratio analysis was conducted.
The association between milk production parameters, as reflected by 100-day and 305-day production, on a range of reproduction parameters is presented in Tables 3 and 4, respectively. A 1000 kg increase in 100-day milk yield or 1000 MJ increase in milk energy output was associated with a 4.49 (P < 0.001) or 1.38 days (P < 0.01) increase in calving index, respectively, while the same increase in 305-day yield and milk energy output increased calving index by 7.79 (P < 0.001) and 2.64 (P < 0.001) days, respectively. The other main associations of increases in 100-day yield and milk energy output were significant (P < 0.05 or greater) reductions in CR-1, CR-1/2, ICR-100, RR-400 and removal rate, while increases in 305-day yield and milk energy output also decreased RR-365 (P < 0.001) but had no association with removal rate.
Effect of changing 100-day milk production on a range of fertility parameters, where the effect of a pre-determined change in each 100-day yield parameter (that could be reasonably achieved in practice at farm level) was assessed against its effect on a range of continuous and binary fertility parameters
† Dependent (fertility) variable mean from random effects multi-level model for continuous variables with a response curve.
‡ Independent (production) variable mean from random effects multi-level model for binary variables where odds ratio analysis was conducted. Associations are significant where the value of 1.000 falls within the 95% confidence interval (c.i.).
Effect of changing 305-day milk production on a range of fertility parameters, where the effect of a pre-determined change in each 305-day yield parameter (that could be reasonably achieved in practice at farm level) was assessed against its effect on a range of continuous and binary fertility parameters
† Dependent (fertility) variable mean from random effects multi-level model for continuous variables with a response curve.
‡ Independent (production) variable mean from random effects multi-level model for binary variables where odds ratio analysis was conducted. Associations are significant where the value of 1.000 falls within the 95% confidence interval.
A 1-g/kg increase in both 100- and 305-day protein concentrations was associated with an increase in RR-400 (P < 0.01). Increased 100-day protein concentration was also associated with decreases in IFS and CI (P < 0.001), and an increase in RR-365 (P < 0.05) and ICR-100 (P < 0.001). Increased mean milk fat concentration over 100 days had no significant association with any fertility parameters while an increase in mean milk fat concentration over 305 days was associated with increased removal rates (P < 0.05).
The association between nadir protein concentrations and maximum daily yields on the range of fertility parameters is presented in Table 5. Each 1-g/kg increase in milk protein nadir was associated with an increase in both RR-400 and ICR-100, which increased by 6.3 (P < 0.001) and 6.6% (P < 0.001), respectively. Each 10-kg increase in peak daily yield was associated with an increase in both IFS (1.59 days; P < 0.05) and CI (6.58 days; P < 0.001) and a 14.9% decrease in CR-1. Each 10-kg increase in yield was also associated with a decrease in RR-365 (9.6%; P < 0.05), RR-400 (14.2%; P < 0.001) and ICR-100 (18.3%; P < 0.001).
Effect of nadir milk protein concentration and maximum daily yield on a range of fertility parameters, where the effect of a pre-determined change in nadir milk protein concentration or peak milk yield (that could be reasonably achieved in practice at farm level) was assessed against its effect on a range of continuous and binary fertility parameters
† Dependent (fertility) variable mean from random effects multi-level model for continuous variables with a response curve.
‡ Independent (production) variable mean from random effects multi-level model for binary variables where odds ratio analysis was conducted. Associations are significant where the value of 1.000 falls within the 95% confidence interval.
The relative effect of the various uni-variate comparisons of either genetic merit or production variables on the continuous measures of fertility are presented as the value of t in Tables 2–5. The value of t on IFS by AI was 3.95 (P < 0.001), 0.12 (P>0.10) and 9.74 (P < 0.001) days for PDX milk, 100-day milk yield and 305-day milk yield, respectively, while the value of t for CI was 4.16 (P < 0.001), 3.23 (P < 0.001) and 12.78 (P < 0.001), respectively. The effect of standard deviation increase in genetic or production variables on both the continuous and binary measures of fertility indicated that 305-day production and milk energy output had a greater effect on the fertility parameters, with the exception of RR-overall where only genetic merit had significant effects (Table 6).
Comparative effect of genetic merit and production parameters on fertility
The associations between season of calving, lactation number, calving difficulty score and calf type on CR-1, ICR-100 and RR-400 are presented in Table 7. When compared to the 3-month period from August to October, the odds of conceiving to first insemination (CR-1) were significantly lower for cows calving in the months of February to April, while the odds of becoming pregnant within 100 days of calving (ICR-100) were significantly higher for cows calving in November to January, February to April and May to July than for cows calving in August to October. Similarly, cows calving in November to January and February to April had higher odds of calving again with 400 days (RR-400) than cows calving in August to October, while cows calving in May to July had lower odds.
Association between a range of factors on some key fertility parameters
Cows in their first or second lactation had significantly higher odds of conceiving to first insemination than those in their third lactation or more. The findings were similar for both becoming pregnant within 100 days of calving and calving again within 400 days, although the odds of cows in their second lactation becoming pregnant again within 100 days of calving were not significantly different (P < 0.10) from third lactation cows.
The odds of cows with a calving difficulty score of 4–5 calving again with 400 days were significantly lower than cows with a calving difficulty score of 0. In cows intended for re-breeding, cows with calving difficulty scores of 4 or 5 had a similar ICR-100 to those with a calving difficulty score of 0, although there were some significant differences for other scores. Cows that had twins were significantly less likely to calve again within 400 days than cows that had a single heifer calf. They also had a tendency (P < 0.10) towards reduced calving rates to first insemination but this was generally not reflected in a significantly reduced ICR-100 in cows intended for rebreeding.
Discussion
This study was designed to assess the reproductive performance of a range of dairy herds in terms of both genetic merit and milk production, reflecting the diverse range of dairy production systems present throughout Northern Ireland. Fertility performance was generally poor with an average CI of 405.9 days and CR-1 of 40.3%. Both these results were consistent with results from the 1st year of the study (Mayne et al., Reference Mayne, McCoy, Lennox, Mackey, Verner, Catney, McCaughey, Wylie, Kennedy and Gordon2002) where the average CI and CR-1 were 407.2 days and 37.1%, respectively, and the herd ranges were 359 to 442 days and 21 to 66%, respectively. The average CI interval reported by United Milk Records for approximately 1000 herds in Northern Ireland (personal communication) for a similar period was 399 days (herd range 365 to 445 days), similar to the 397 days reported nationally by National Milk Records (Esslemont et al., Reference Esslemont, Kossaibati, Allcock and Diskin2001). The mean calving interval was also consistent with that extrapolated from the average interval to conception (134 and 107 days for two successive years) in a fertility study of seven herds reported by Wicks and Leaver (Reference Wicks and Leaver2004).
Slippage and extension of the calving pattern is a common feature of many dairy herds in Northern Ireland. This is demonstrated by the fact that only 24.0% of cows in this study re-appeared within 365 days of a previous calving, and only 44.5% re-appeared within 400 days. A major cause of this poor performance was the prolonged interval to first service of 85.2 days on average, and the poor calving rate to first service of 40.3% which meant that only 46.0% of cows intended for re-breeding were back in-calf within 100 days of calving, although this varied considerably between herds (herd range: 16.4 to 70.8%). While the mean 100-day in calf rate is less than the 53% reported by Morton (Reference Morton2001) for year-round calving herds in Australia, he reported a similar range of 19 to 73% between herds. Both 100-day in-calf rate and re-appearance rate offer alternatives to calving index as a means of assessing reproductive performance in dairy herds as they avoid the problems associated with cows not calving again and can be used to actively target improved herd fertility performance.
The average calving rate to first insemination in this study (40.3%) is similar to the findings of Royal et al. (Reference Royal, Darwash, Flint, Webb, Woolliams and Lamming2000) who reported a calving rate to first insemination of 39.7%, and similar to that of Butler et al. (Reference Butler, Cherney and Elrod1995) who reported a conception rate to first insemination of 40.9%. However, these results are lower than that of Grosshans et al. (Reference Grosshans, Xu, Burton, Johnson and Macmillan1997) who reported calving rates to first service of 48.5 and 50.0% for first and second lactation cows. They were also lower than that of Buckley et al. (Reference Buckley, O'Sullivan, Mee, Evans and Dillon2003) who reported a conception rate of 49% using rectal palpation at least 56 days after the end of the breeding season and that of Wicks and Leaver (Reference Wicks and Leaver2004) who reported average pregnancy rates of 56 and 63% in 2 years of study. The study of Grosshans et al. (Reference Grosshans, Xu, Burton, Johnson and Macmillan1997) was conducted in the pasture-based system of New Zealand where there is inherently better cow fertility due to the seasonal nature of the production system. This together with the lower production and shorter lactation of New Zealand cows may be the reason for their more favourable results. Similarly, the study of Buckley et al. (Reference Buckley, O'Sullivan, Mee, Evans and Dillon2003) was also conducted on seasonal spring-calving herds, but conception rates based on rectal palpation are likely to give a more favourable result than results based on a subsequent calving used in the present study, i.e. the ultimate assessment of fertility based on pregnancy diagnosis.
Genetic merit and fertility performance
Various reports have suggested that the declining level of fertility in dairy cows may be associated with increased genetic merit for milk production and an increased percentage of Holstein genes (Hoekstra et al., Reference Hoekstra, Van der Lught, Van der Werf and Ouweltjes1994; Royal et al., Reference Royal, Darwash, Flint, Webb, Woolliams and Lamming2000; Harris and Kolver, Reference Harris and Kolver2001). Selection of dairy cattle for higher milk yield potential has generally been accompanied by poorer reproductive performance, where genetic correlations between production and fertility parameters are generally negative (Pryce et al., Reference Pryce, Esslemont, Thompson, Veerkamp, Kossaibati and Simm1998; Buckley et al., Reference Buckley, Dillon, Rath and Veerkamp2000; Kadarmideen et al., Reference Kadarmideen, Thompson and Simm2000; Olori et al., Reference Olori, Meuwissen and Veerkamp2002). Consequently, fertility traits have now been incorporated into breed selection indices in the Republic of Ireland and UK (Veerkamp et al., Reference Veerkamp, Dillon, Kelly, Cromie and Green2002; Wall et al., Reference Wall, Brotherstone, Woolliams, Banos and Coffey2003; Santarossa et al., Reference Santarossa, Stott, Woolliams, Brotherstone, Wall and Coffey2004) making it possible to make genetic improvements to both milk yield and fertility simultaneously, albeit at a slower rate than selecting for each on its own.
In the present study, the main effects of increasing genetic merit, either through increases in £PIN or in the pedigree index for milk, were observed as significant increases in IFS and CI, but also as a significant increase in overall re-appearance rate or decrease in removal rate. The increased IFS and CI may have been in part due to the longer interval to commencement of luteal activity in these cows, previously reported by McCoy et al. (Reference McCoy, Lennox, Mayne, McCaughey, Edgar, Catney, Verner, Mackey and Gordon2006) but is also likely to have been due to management reasons in higher genetic merit herds, where cows have an extended voluntary waiting period of 3 months or more before service (unpublished data from present study). Cows of superior genetic merit are also less likely to display heat at first ovulation (Westwood et al., Reference Westwood, Lean and Garvin2002) with effects on interval to first service. The increase in overall re-appearance rate and decreased removal rate observed for cows with a higher £PIN and genetic merit for milk production in this study may be in part due to the predominance of herds with prolonged winter calving and extended periods of breeding in this study, typical of that currently present in Northern Ireland. The prolonged breeding period, often exceeding six months allows repeat breeding cows, often with higher genetic merit for milk production, to be preferentially retained and served a number of times, despite their lower levels of fertility performance. This is in contrast to seasonal calving systems where such cows are more likely to be removed from the herd. Cows with higher genetic merit for milk fat and protein concentration had a lower overall re-appearance rate and were more likely to be removed from the herd, possibly due to the greater importance of milk yield.
There are conflicting reports on the effect of increasing genetic merit for milk production on conception rate. Hoekstra et al. (Reference Hoekstra, Van der Lught, Van der Werf and Ouweltjes1994) found that an increased proportion of Holstein genes and hence higher genetic merit for milk production had a detrimental effect on conception rate, as assessed by non-returns. In the present study, there was no significant effect of £PIN on calving rate, but each 100 kg increase in genetic merit for milk production was associated with a 0.5% increase in calving rate to first insemination and this contributed to a 0.8% increase in 400-day re-appearance rate. These findings are consistent with Buckley et al. (Reference Buckley, O'Sullivan, Mee, Evans and Dillon2003) who found that the proportion of Holstein genes had no effect on pregnancy rate to first service, but also found that higher milk production was associated with improved conception rates to first service. In the present study, the positive association between genetic merit and calving rate may have been due to a number of factors including the delayed interval to first insemination.
Production and fertility performance
While there is much information on the genetic relationships between milk production and fertility (Dematawewa and Berger, Reference Dematawewa and Berger1998; Pryce et al., Reference Pryce, Esslemont, Thompson, Veerkamp, Kossaibati and Simm1998; Kadarmideen et al., Reference Kadarmideen, Thompson and Simm2000; Berry et al., Reference Berry, Buckley, Dillon, Evans, Rath and Veerkamp2003), there is a lack of consensus in the literature as to whether the decline in dairy cow fertility is caused by increasing genetic merit for milk production or indeed increased production itself. In this study, every fertility parameter measured was significantly and adversely affected by a 1000-kg increase in 305-day milk production (or 1000 MJ increase in milk energy output), with the exception of RR-overall and removal rate, indicating the preferential retention of high yielding cows in the herd, despite their poorer fertility performance. An increased interval to first service and a decreased calving rate to first service in higher yielding cows contributed to the prolonged calving index in higher yielding cows, with 22% fewer cows being in-calf within 100 days of calving and 19% fewer re-appearing within 400 days of a previous calving event for each 1000-kg increase in 305-day yield.
In a study of two lines of Holstein cows, a high production and a medium production line, Hageman et al. (Reference Hageman, Shook and Tyler1991) estimated that each 1000-kg increase in 305-day milk yield increased the number of days open, and hence CI by 7.0 and 12.5 days in first and second lactation cows, respectively. Veerkamp et al. (Reference Veerkamp, Dillon, Kelly, Cromie and Green2002) predicted from genetic parameters that under selection for milk yield alone, each 1000-kg increase in yield is associated with a 5- to 10-day prolongation of calving index. Using similar arithmetic procedures for predicting the genetic relationship between production and fertility traits from various studies (Dematawewa and Berger, Reference Dematawewa and Berger1998, Pryce et al., Reference Pryce, Esslemont, Thompson, Veerkamp, Kossaibati and Simm1998, Olori et al., Reference Olori, Meuwissen and Veerkamp2002, Berry et al., Reference Berry, Buckley, Dillon, Evans, Rath and Veerkamp2003, Pryce et al., Reference Pryce, Coffey, Brotherstone and Woolliams2002 and Kadarmideen, Reference Kadarmideen2004), each 1000-kg increase in 305-day milk production was associated with up to a 5-day increase in interval to first service, a 6- to 14-day increase in calving index and a 3 to 6% decrease in calving rate to first insemination. The findings of the present study, where each 1000-kg increase in 305-day milk yield was associated with a 3.2-day increase in interval to first insemination and a 7.8-day increase in calving index, are consistent with the predictions based on the genetic relationships found in the papers listed above. However, the detrimental effect of a 1000-kg increase in 305-day yield on calving rate to first insemination observed in this study (decrease of 13.6%) was greater than predicted from the findings of Pryce et al. (Reference Pryce, Esslemont, Thompson, Veerkamp, Kossaibati and Simm1998), Berry et al. (Reference Berry, Buckley, Dillon, Evans, Rath and Veerkamp2003) and Kadarmideen (Reference Kadarmideen2004) where decreases of 6, 4 and 3%, respectively, would have been expected. The findings from the present study are in contrast to those of Buckley et al. (Reference Buckley, O'Sullivan, Mee, Evans and Dillon2003) where higher yielding cows had higher odds of conceiving to first service. However, in that study, all herds were spring-calving and there was a smaller range in 305-day production. Their statistical analysis also included adjustments for the proportion of Holstein genes and breeding value for milk yield, both of which reduced the odds of cows becoming pregnant to first service at higher levels.
Negative energy balance in the first 3 to 4 weeks post partum is highly correlated with interval to first ovulation (Butler, Reference Butler and Diskin2001). In the present study, parameters relating to early lactation milk production, viz. peak yield, nadir protein concentration and 100-day milk production also had effects on the various measures of fertility performance. Hoekstra et al. (Reference Hoekstra, Van der Lught, Van der Werf and Ouweltjes1994) reported negative genetic associations between production and fertility, particularly protein yield, and Fulkerson et al. (Reference Fulkerson, Wilkins, Dobos, Hough, Goddard and Davidson2001) showed that cows with the lowest milk protein content suffered the most severe and prolonged negative energy balance. Butler (Reference Butler and Diskin2001) cited that timing of the nadir is particularly important, with cows in greater negative energy balance being less likely to exhibit regular oestrous cycles and have poorer fertility performance. Negative energy balance is associated with low oestradiol concentrations in the periovulatory period (Mackey et al., Reference Mackey, Sreenan, Roche and Diskin1999), leading to less overt expression of oestrus (Spicer et al., Reference Spicer, Tucker and Adams1990; Lyimo et al., Reference Lyimo, Nielson, Ouweltjes, Kruip and Van Eerdenburg2000) and greater difficulty in heat detection. In the present study, a 1-g/kg decrease in nadir milk protein was associated with poorer reproductive performance as observed by an increased interval to first service and calving index, and a decreased 100-day in-calf rate and 400-day re-appearance rate. The relationship between milk energy content and subsequent fertility in this study is supported by the association between fertility and 100-day milk production where there was also a negative association with calving rate to first insemination.
Higher peak yields were associated with decreased calving rates to first service, equivalent to a 14.9% decrease in CR-1 for each 10-kg increase in peak yield or 15.7% decrease for each 1000-kg increase in 100-day milk production, and these had associated effects on ICR-100, RR-365 and RR-400. This is consistent with previous reports where higher levels of milk production in early lactation were associated with a reduced likelihood of successful pregnancy by day 150 due to a delay in and reduced expression of oestrus at first ovulation (Westwood et al., Reference Westwood, Lean and Garvin2002). The reduced calving rates at higher levels of production were likely to be a consequence of increased negative energy balance leading to metabolic stress and possibly suboptimal uterine conditions for embryo development (Butler, Reference Butler and Diskin2001).
Relative effects of genetic merit and production on fertility performance
Buckley et al. (Reference Buckley, O'Sullivan, Mee, Evans and Dillon2003) found no effect of genetic merit for milk production on pregnancy rate to first service, and only cows with very high genetic merit had decreased submission rates and pregnancy rates within 21 and 42 days of breeding start, respectively. While results from the current study indicated that fertility performance was negatively affected by both genetic merit and level of production, production had the greater negative effect, especially when considered over 305 days (Table 6). These findings indicate that infertility at an individual cow level is predominantly due to environmental effects (including nutrition), a view supported by the findings of Veerkamp and Emmans (Reference Veerkamp and Emmans1995) who cited that indirect selection for feed intake milk yield through selection for milk yield alone, as in the UK through the PIN index, can only provide 40 to 48% of the extra energy requirement for increased milk yield in early lactation. However, Buckley et al. (Reference Buckley, Dillon, Rath and Veerkamp2000) reported no negative effects of genetic merit on reproductive performance in a spring-calving grass-based production system. This, together with unpublished data from the present study pertaining to individual herds suggest that it is possible to get good reproductive performance from high genetic merit dairy cows providing overall fertility management and nutrition is optimised.
Effects of calving season and calving difficulty on fertility performance
There were a number of significant seasonal effects on various fertility parameters, some of which can be explained by the diverse range of calving patterns observed in the 19 participating herds. Compared to cows calving from August to October, cows calving from February to April had a significantly lower CR-1. This is consistent with the findings of Buckley et al. (Reference Buckley, O'Sullivan, Mee, Evans and Dillon2003) who also reported a decreased calving rate to first service in late calving cows, although cows in that study calved from January to May, unlike the present study where the majority of herds calved throughout the winter. In this study, February to April calving cows had a significantly lower CR-1, but had a significantly higher ICR-100. These findings reflect the large influence of spring calving herds where cows were often inseminated in the early post-partum period, resulting in poor calving rates (Mayne et al., Reference Mayne, McCoy, Lennox, Mackey, Verner, Catney, McCaughey, Wylie, Kennedy and Gordon2002), but were also given the opportunity for more chances to conceive within 100 days of calving. The lower CR-1 and higher ICR-100 for February to April calving cows may also have been due, in part, to the shorter interval to first insemination experienced by late calving cows in prolonged winter-calving herds in an attempt to calve them earlier the following year. However, in these herds, where cows calved from May to July, there was a significantly lower RR-400 suggesting that they fell outside the ideal calving season so a smaller proportion of these were intended for re-breeding.
It is widely accepted that dystocia is a major factor affecting the subsequent reproductive performance of dairy cows (Fonseca et al., Reference Fonseca, Britt, McDaniel, Wilk and Rakes1983; Morton, Reference Morton2000). In the present study, the effects of dystocia on CR-1 and ICR-100 were inconsistent, but there was a significantly reduced chance of cows with higher levels of dystocia re-appearing again within 400 days. This may have been due to an increased likelihood of retained foetal membranes and associated metritis which can delay uterine involution, disrupt post-partum hormonal profiles and ultimately lead to a delay in return to cyclicity (McCoy et al., Reference McCoy, Lennox, Mayne, McCaughey, Edgar, Catney, Verner, Mackey and Gordon2006). It is possible that a higher incidence of reproductive problems in older cows (unpublished results from present study) may have contributed to their lower odds of conceiving to first service compared to first or second lactation cows, with associated effects on ICR-100 and RR-400.
Conclusions
The results of this study showed that the fertility performance of dairy cows on-farm is negatively associated with both genetic merit and level of production, and various other factors including season of calving, lactation number and dystocia. Level of production has the greatest negative effect, especially when considered over a 305-day lactation, indicating that infertility at an individual cow level in this study is predominantly due to an inability to meet the nutritional requirements of high genetic merit dairy cows, as evidenced by lower nadir milk protein concentrations. These results also suggest that it is possible to get good reproductive performance from high genetic merit dairy cows providing high levels of fertility management and appropriate nutrition are achieved.
Acknowledgements
This study was co-funded by the Department of Agriculture and Rural Development for Northern Ireland (DARD) and AgriSearch (Northern Ireland). The authors also wish to gratefully acknowledge the following: the 19 farmers who participated in the programme and provided valuable data and assistance throughout the project; staff of the Dairy Unit, Agricultural Research Institute of Northern Ireland, for technical assistance; the milk recording agencies involved – United Dairy Farmers and HerdTech, Animal Data Centre, Holstein UK; and the data preparation staff in the Biometrics Division for input of data.
