stud-Mean breed performance of the progeny from beef-on-dairy matings

Gains through breeding can be achieved through a combination of both between-breed and within-breed selection. Two suites of traits of particular interest to dairy producers when selecting beef bulls for mating to dairy females are calving-related attributes and the expected value of the subsequent calf, the latter usually being a function of expected carcass value. Estimated breed effects can be informative, particularly in the absence of across-breed genetic evaluations. The objective of the present study was to use a large national database of the progeny from beef-on-dairy matings to estimate the mean breed effects of the used beef sires. Calving performance (i.e., gestation length, calving difficulty score, and perinatal morality) as well as calf value were investigated; a series of slaughter-related traits (i.e., carcass metrics and age at slaughter) of the prime progeny were also investigated. Phenotypic data on up to 977,037 progeny for calving performance, 79,903 for calf price and 103,175 for carcass traits (including dairyx-dairy progeny for comparative purposes) were used; sire breeds represented were Holstein-Friesian, Angus, Aubrac, Belgian Blue, Charolais, Hereford, Limousin, Salers and Simmental. Large inter-breed differences existed. The mean gestation length of male calves from beef sires varied from 282.3 d (Angus) to 287.4 d (Lim-ousin) which were all longer than the mean of 280.9 d for Holstein-Friesian sired male calves. Relative to a Holstein-Friesian sire, the odds of dystocia varied from 1.43 (Angus) to 4.77 (Belgian Blue) but, once adjusted for both the estimated maternal genetic merit of the dam and direct genetic merit of the calf for calving difficulty, the range in odds ratios shrunk. A difference of €125.4 existed in calf sale price between the progeny of the different beef breeds investigated which represented over twice the residual standard deviation in calf price within the day of sale - Angus was the cheapest while Charolais calves were, on average, the most expensive calves. Mean carcass weight of steers, not adjusted for age at slaughter or carcass fat, varied from 327.1 kg (Angus) to 363.2 kg (Belgian Blue) for the beef breeds with the mean carcass weight of Holstein-Friesian steer progeny being 322.4 kg. Belgian Blues had, on average, the best carcass conformation with the Herefords and Angus having the worst of all beef breeds. Angus and Hereford steers were slaughtered the youngest of all beef breeds but just 9 d younger than the average of all other beef breeds yet 24 d younger than Holstein-Friesian sired progeny. Clear breed differences in calving and carcass performance exist among beef breeds mated to dairy females. Those breeds excelling in calving performance were not necessarily the best for carcass merit.


INTRODUCTION
Gains through breeding can be achieved through a combination of selection both between and within breeds.Many studies have documented the extent of within-breed genetic variability for a whole plethora of cattle performance traits including calving performance (Johanson et al., 2011;Vallée et al., 2013;Ahlberg et al., 2016), gestation length (Norman et al., 2009;Johanson et al., 2011), animal value (McHugh et al., 2011;Penasa et al., 2012), carcass merit (Pabiou et al., 2009;Boukha et al., 2011;Judge et al., 2019), age at slaughter (Berry et al., 2017;Pritchard et al., 2021) and meat eating quality (Boukha et al., 2011;Berry et al., 2021).Such information can be used to infer the potential of selection within breeds in the pursuit of genetic gain.For example, the within-breed genetic variation in carcass weight of prime cattle has been reported to be between 22.9 kg and 26.1 kg (Pritchard et al., 2021;Kause et al., 2015); this means that the expected difference in genetic merit between the top and bottom 10% of animals for carcass weight is 80 kg to 92 kg.
Step-changes in genetic merit can, however, be achieved through breed substitution.Making an informed decision on such a tactic requires estimates of breed effects for not only the trait(s) of interest, but also correlated traits that may be important either for the breeder themselves or their customer.Cattle breeds have been compared both in controlled stud-ies (Campion et al., 2009;Keane and Moloney, 2010;Bureš and Bartoň, 2018), through a meta-analyses of controlled studies (Alberti et al., 2008) or from crosssectional analyses of available data sets (Kause et al., 2015;Judge et al., 2021;Huuskonen et al., 2013Huuskonen et al., , 2014)).Many such breed comparisons have been undertaken using purebreds (Phocas and Laloë, 2004;Alberti et al., 2008) or crossbreds (Ahlberg et al., 2016;Pritchard et al., 2021) based on progeny exclusively from beef dams (Phocas and Laloë, 2004;Kause et al., 2015) or from a combination of beef and dairy dams but with no distinction between the dam types (Judge et al., 2019;Pritchard et al., 2021).The data used in many genetic evaluations globally for beef-related traits originate exclusively or predominantly from the progeny of beef dams.The same beef-related phenotypes are also measured on the progeny from dairy dams.Therefore, estimating the breed effects from progeny of only dairy dams is important given the ever-growing interest in the mating of beef bulls to dairy females to generate more valuable calves (Berry, 2021;Basiel and Felix, 2022;Poock and Beckett, 2022).Because of the phenomenon of genotype-by-environment interactions (Falconer, 1989), the observed breed differences estimated using data from beef dams may not materialize in the progeny from beef-on-dairy matings.Moreover, the beef sires used by dairy producers are likely to be a distinct group of sires with particular attributes and may differ from those sires used by beef cow-calf producers.Beef output represents circa.5% of the gross revenue or income in many dairy herds (Berry, 2021).Although many dairy producers sell their surplus calves soon after birth, in some jurisdictions, including Ireland, a large proportion of dairy herds also rear their surplus animals for > 1 yr of age (Berry and Ring, 2021).Prices paid to dairy producers for surplus calves do differ by breed (McHugh et al., 2011) so breed effects for beef traits in the progeny of dairy dams is important to know.
Using data from almost 1.1 million dairy cows, Eriksson et al. ( 2020) compared the calving performance and carcass merit in the progeny of different breeds of beef sires born to dairy cows; the dam dairy breeds investigated in that study were purebred Swedish Red and Swedish Holstein while the beef breeds included Angus, Hereford, Limousin, Charolais and Simmental.Others have also compared the performance of beefxdairy versus dairyxdairy crosses for calving performance (Fouz et al., 2013), growth (Campion et al., 2009;Keane and Moloney, 2010;Huuskonen et al., 2013) and carcass (Campion et al., 2009;Keane and Moloney, 2010;Huuskonen et al., 2013Huuskonen et al., , 2014) ) traits.However, given the growing interest in sex-sorted semen, whether the breed effects on performance differ by gender of the progeny has not been thoroughly investigated.Fur-thermore, whether breed differences can be captured by across-breed genetic evaluations has not been investigated; this is important to provide confidence to producers that breeds can be fairly compared based on sire estimates of genetic merit from across-breed genetic evaluations.Additionally, the extent of variability within breeds relative to that across breeds has not been thoroughly investigated in progeny from beef-ondairy matings.
Therefore, the objective of the present study was to use progeny information to estimate the breed effect of several beef breeds, when mated to Holstein-Friesian dams, for calving performance, calf value and slaughter-related traits and if across-breed genetic evaluations could capture these breed effects; both calving performance and calf value are of interest to the dairy producer as they impact farm costs and revenue, respectively while slaughter traits are of interest to beef producers.Also of interest was if the estimated breed effects were consistent across animal genders.Breed effects from the present study can be used, in tandem with within-breed variability estimates, to help decide on the appropriate breeding strategy for different herds to realize gains in performance.The breed solutions can also be used in bio-economic models to evaluate alternative production systems using breed-specific performance metrics.

METHODS
All data were extracted from the Irish Cattle Breeding Federation (www .icbf.com);therefore, it was not necessary to obtain animal ethics committee approval before conducting this study.Only data between 2018 and 2022, inclusive were considered further.Estimated breeding values (EBVs) values from the national multibreed genetic evaluations in the year 2017 were available for direct and maternal calving difficulty, direct gestation length, and direct perinatal calf mortality, as well as the carcass traits of carcass weight, conformation, and fat score.Estimated breeding values of all animals in Ireland, irrespective of breed, are expressed relative to a common base.The EBV of each animal in the data set was calculated as sum of 1/2EBV of the sire plus 1/4EBV of the maternal grandsire; the EBV of each dam for her maternal genetic effect was simply based on her sire.
For all traits investigated in the present study, only progeny from dams recorded to be > 87.5% Holstein-Friesian were retained and parities > 15 were discarded.Dam parity number was recoded as: 1, 2, 3, 4-6, 7-10, and > 10.No dam calved younger than 660 d of age, nor calved > 2 years from the median age at calving within parity.The sire, dam and maternal grand-sire (for the purposes of EBV calculation) of each animal had to have been recorded.All progeny had to be sired by a purebred AI or natural mating bull; for the purposes of this study, Holstein-Friesian was considered a single breed and, although a dairy breed, was included in the present study not just for comparative purposes, but also to help better estimate effects in the statistical model.The sire breeds considered in the present study were Holstein-Friesian, Angus, Aubrac, Belgian Blue, Charolais, Hereford, Limousin, Salers and Simmental; all other sire breeds, which represented < 1% of the data, were collapsed into a single group but retained to help estimate the effects in the statistical model but the associated model solutions for this breed group are not reported.

Phenotypes
Calving assistance in Ireland is subjectively scored by producers on a 4-point scale of: (1) no assistance, (2) slight assistance, (3) considerable assistance, and (4) veterinary assistance (including caesareans).Only herd-years that recorded some variability in the recorded calving assistance scores in that year were retained.The 4-point scale was dichotomized into calving assistance (score 1 versus scores ≥ 2 combined) or calving dystocia (scores 1 and 2 combined versus scores 3 and 4 combined).Gestation length records between 271 and 300 d were retained.Because it is a legal requirement to record the death of all animals in Ireland, it was possible to define a perinatal mortality trait as calf died at birth or within the first 24 h of birth.Calving dystocia and perinatal mortality information was available on 1,736,595 singleton calves born in 8,506 dairy herds all of which had ≥ 50 calving events in that year.
Many calves in dairy herds, not retained for herd replacements, are traded as young calves at public livestock auctions (McHugh et al., 2010); the price paid per calf was available for the present study.Female dairysired calves were not considered.Price information was available on 637,178 calves traded singly in Irish livestock auctions younger than 42 d of age between the years 2018 and 2022, inclusive.
Carcass-related information recorded in Ireland include (cold) carcass weight (kg), carcass conformation, and subcutaneous carcass fat cover score.Both carcass conformation and fat score are mechanically graded on a 1 (poor conformation; little fat cover) to 15 (excellent conformation; large fat cover) point scale; more details are given by Kenny et al. (2020).Price paid per kg carcass is also recorded.Age at slaughter was also available for all animals; only steers and heifers slaughtered between 10 and 36 mo of age were retained in the present study whereas bulls had to be slaughtered < 2 years of age (and not sire any progeny) thereby representing bulls for meat production as opposed to for breeding purposes.Furthermore, only animals that resided in the farm from where they were slaughtered for ≥ 100 d were retained and animals with > 2 inter-herd movements during their lifetime were discarded.Following edits, slaughter-related information was available on 560,746 prime cattle born in Irish dairy herds.
Contemporary group was defined as herd-year-season of calving for the 4 calving traits and herd-year-sex-season of slaughter for the 3 carcass traits; sex was defined as either heifer, steer or young bull.The contemporary groups were defined for each trait separately using an algorithm routinely used in Irish national genetic evaluations (Berry et al., 2013).The contemporary group for age at slaughter was based on that proposed by Berry et al. (2017) for age at slaughter and was based on the herd-sex-season of entry onto the final farm where animals of different ages were grouped together.A maximum of 60-d range was allowed between the start and end of each contemporary group, and each contemporary group had to have at least 10 records for the calving traits and at least 5 for the slaughter traits.The contemporary group for the calf price traits was herd-date of sale (McHugh et al., 2010) with an edit imposed of having ≥ 5 records per contemporary group to be retained for the analysis.A further edit was imposed in that no more than 75% of the average breed composition of the animals in an individual contemporary group could be made up of any one sire breed.
Following all editing, dystocia data were available on 1,389,670 births from 768,738 cows in 6,885 dairy herds; of these, 865,920 also had a valid recorded gestation length.A total of 79,903 of these calves had a recorded auction price of which, 52,979 were beefxdairy calves.Slaughter information was available on 103,175 cattle of which 72,547 were beefxdairy crosses.

Estimation of breed effects
All association analyses were undertaken in ASREML (Gilmour et al., 2009) to quantify the association between sire breed and the traits of interest.Sire, dam and contemporary group were all included as random effects in all models.The binary dependent variables (i.e., calving dystocia, calving assistance, and perinatal mortality) were all modeled using a logit link function accounting for the binominal distribution of the errors; linear models were used in the association analyses of the other continuous and normally-distributed response variables.The fitted model to all traits was: Y ijkl = Parity i + age j + sex k + breed l + e ijkl

Berry et al.: BEEF BREED PERFORMANCE
where Y ijkl represented the (calving, price or carcass) trait being investigated, Parity i represented the parity of the dam (i = 1, 2, 3, 4-6, 7-10, 10+), age j represented age at calving centered within parity (each class was 6 mo in duration ± 2 years relative to the median age per parity), sex k represented calf sex (male or female) or carcass type (heifer, steer or young bull) in the case of slaughter traits, and breed l represented sire breed; for the analysis of calf price, age at sale was also included as a covariate in the model.Two-way interactions between sire breed and both parity and calf sex (or carcass type for slaughter traits) were also explored.Direct EBV of the calf as well as the maternal EBV of the dam (just for calving dystocia and assistance), for the dependent variable under investigation was also considered as a covariate in the models in a supplementary series of analyses to investigate if the across-breed EBVs can capture the breed effects.Because of the size of the data set used, statistical significance of the interaction terms was declared at P < 0.001.Predicted marginal means were based on a third parity Holstein-Friesian dam calving at the average age for that parity.

RESULTS
The mean incidence of calving assistance, dystocia and perinatal mortality in the edited data set was 9.44%, 1.51% and 1.69%, respectively with a mean (standard deviation) gestation length of 279.3 (4.97) d; when restricted to just beef-on-dairy matings, the respective mean incidence was 14.10%, 2.40% and 1.91% with a mean (standard deviation) gestation length of 282.3 (5.8) d.The raw mean (standard deviation) carcass weight, conformation score, fat score and age at slaughter of the entire data set was 319.9 (45.6) kg, 5.33 (1.46) units, 8.59 (1.80) units and 740.54 (109.8)d, respectively.

Calving performance traits
The association between sire breed and gestation length differed by calf gender (Table 1).The predicted marginal mean breed effects for gestation length of female calves born to a third parity Holstein-Friesian dam varied from 280.9 d (Holstein-Friesian) to 287.4 d (Limousin).Sire breed was still associated with gestation length even after adjusting for calf EBV for gestation length although the strength of the breed effect in the statistical model reduced from a Type III F-statistic of 160.41 to 16.07; this manifested itself as the range in mean sire breed effects (Table 1) reducing by half varying from 283.1 d (Angus) to 286.0 d (Limousin).The association between phenotypic gestation length and its EBV did not differ (P = 0.20) by sire breed.
Sire breed was associated with all of calving assistance, dystocia, and perinatal mortality but the associations did not differ (P > 0.01) by calf gender.The association between each trait and its respective EBV was consistent across sire breeds.The marginal predicted probabilities of calving assistance, dystocia, and perinatal mortality of a female calf born to a third parity Holstein-Friesian dam is in Table 2; the odds ratios for each of the 3 calving traits per sire breed relative to the Holstein-Friesian breed is in Figure 1 with or without correction for the respective EBV of the calf and the dam.Sire breed remained associated both with calving assistance and dystocia (but not perinatal mortality) even after adjusting for EBVs but the estimated Type III F-statistic reduced from 88.06 to 31.35 (i.e., less significant) for calving assistance and from 43.73 to 17.69 (i.e., less significant) for calving dystocia once calf and dam EBV were included as covariates in the statistical model.
The odds of calving assistance relative to Holstein-Friesian varied from 1.29 (Angus) to 5.2 (Belgian Blue) but once adjusted for both the maternal EBV of the dam and direct EBV of the calf, the range in odds ratios shrunk varying instead from 1.45 (Angus) to 2.46 (Aubrac).A similar conclusion was evident for calving dystocia with the range in odds ratios being 1.43 (Angus) to 4.77 (Belgian Blue) with no adjustment for genetic merit, shrinking to a range of 0.95 (Simmental) to 1.84 (Aubrac) once adjusted for the maternal EBV of the dam and direct EBV of the calf.Without adjustment for EBV, the odds of both assistance and dystocia was greater for all beef breeds relative to the Holstein-Friesian.While the odds for some breeds were, nonetheless, high, the actual manifestation of the biological difference in expected incidence between breeds, even without adjustment for genetic merit, was actually small (Table 2) with the expected incidence of dystocia per breed for a beef-breed female calf born to a third parity Holstein-Friesian dam ranging from 0.8% to 2.8% and from 0.7% to 2.0% for perinatal mortality.

Calf price
Sire breed was associated with calf price (Table 3) but the association differed by calf gender.The manifestation of this interaction was not a re-ranking of sire breeds by gender but instead a change in the scale of the breed difference in calf price between genders.The mean difference in calf price between genders varied from €17.4 (Belgian Blue) to €35.6 (Limousin); in all instances, the male calf was worth more than the female calf.The lowest calf value was for the Holstein-Friesian male calf (€72.54) with the Charolais calves being the most valuable.The standard deviation in calf price after accounting for the fixed effects in the statistical model was €60.71 of which €24.66 was due to the sire and €43.04 was due to the herd-date.Ignoring gender from the statistical model, the mean difference in breed effects between the Salers (lowest value of the beef breeds) and Charolais (highest value) was €126.2 which represents > 5 times the standard deviation of the within breed sire effects estimated in the present study.Nonetheless, the mean difference between breeds differed by year albeit the difference between the least valuable beef breed and the most valuable beef breed per year varied from €112.74 (in the year 2020) to €129.62 (in the year 2018); irrespective of year, the least valuable breed was always the Salers and the most valuable was always the Charolais.

Slaughter traits
The estimated standard deviation in sire effects from the mixed model was 27.6 d, 10.8 kg, 0.35 units and 0.34 units for age at slaughter, carcass weight, conformation score and fat score, respectively.Marginal mean sire breed effects for age at slaughter and carcass price are in Table 4 while those for carcass weight, conformation and fat score are in Table 5. Considering only the beef breeds, the range in mean age at slaughter per breed was ≤ 1 mo albeit with a distribution with Angus, Aubrac and Hereford being in the early age group and the remaining breeds (including the Holstein-Friesian) being in a later slaughter age group.The range in inter-breed differences for carcass weight of the beef breeds varied from 34.0 kg (heifers) to 38.8 kg (bulls) with the Angus-sired progeny, on average, having the lightest carcasses and Charolais and Belgian Blues being the heaviest; the carcasses of the Holstein-Friesian-sired animals were the lightest of all breeds investigated.The mean breed effects of the beef breeds for conformation varied from 5.3 (Hereford) to 7.3 or 7.5 (Belgian Blue) in the steers and heifers and from 6.3 (Hereford) to 8.4 (Belgian Blue) in bulls; the mean conformation score of the Holstein-Friesian-sired steers and bulls was 2.2 to 2.4 units inferior to the mean of all the beef breeds (Table 5).The Belgian Blue produced the leanest carcasses with the carcasses of the traditional beef breeds (i.e., Angus and Hereford) being the fattest; excluding the Belgian Blue, the Holstein-Friesian was the (joint) leanest breed.
The range in price per kg for the different beef breeds relative to the mean of all beef breed effects was just 2-3% (Table 4) with the greatest difference in breed effects existing for steers and the least for bulls.The price per kg of the Holstein-Friesian was lower than all beef breeds.Considering both the mean breed price per kg (Table 4) and the mean breed carcass weight (Table 5), the range in mean overall carcass worth per beef breed represented 10 to 13% of the average of the beef breed effects.Carcasses from both the Angus and Hereford generated the least carcass revenue of the beef breeds investigated but both received a greater total carcass income than the Holstein-Friesian carcasses; bonus payments by processors for some Angus and Hereford cattle were included in the price paid.Revenue, however, is not synonymous with value or profit since no adjustment was made in the present study on the  purchase price of the animal or indeed the associated costs of production.Sire breed remained associated with carcass weight, conformation and fat score even after adjusting for the EBV of the animal itself for the respective trait; nonetheless, the significance of the breed effect in the statistical model reduced considerably for each with the Type III F-statistic reducing from 137.8 to 26.53 for carcass weight, from 1401.3 to 97.2 for carcass conformation and from 414.3 to 30.46 for carcass fat.This adjustment for genetic merit manifested itself as a halving of the difference in carcass conformation and fat score marginal means between the Angus and Belgian Blue breeds with the difference between both breeds when adjusted for EBV of carcass weight being 63% of the difference without adjustment.Moreover, the association between EBV and the respective carcass trait differed by breed, and the range in regression coefficients on EBV were often quite large.The regression coefficients for phenotypic carcass weight on its EBV varied from 0.41 (Simmental) to 0.82 (Holstein-Friesian).The regression coefficients for phenotypic carcass conformation on EBV for conformation varied from 0.18 (Belgian Blue) to 0.86 (Charloais).With the exception of Belgian Blue where the regression coefficient of phenotypic carcass fat on its EBV was 0.18, the regression coefficients for the remaining breeds varied from 0.61 (Hereford) to 1.61 (Salers).

DISCUSSION
Being able to properly estimate true breed effects from a cross-sectional data set is not trivial.In the present study, only data within contemporary groups where one breed did not predominate was retained in an attempt to disentangle the breed effect from the herd-sex (i.e., management) effect.Therefore, the breed means reported in the present study may not match up with national statistics where some breeds may only exist in some contemporary groups and those contemporary groups may operate a very different style of herd management which may be specific to a given breed (type).For example, the raw mean age at slaughter for Belgian Blues, Charolais, Salers and Simmental sired     progeny before any restriction was placed on the size and breed diversity of contemporary groups was 770 d, 766 d, 790 d and 752d, while these values were 753 d, 740 d, 704 d and 722 d post-editing; the change in values was relatively small for the other breeds with the larger observed change in breed effects no doubt impacted by the smaller population representation of these breeds in the data set.Also of note was that covariates like age at slaughter, carcass weight or carcass fat were not considered in the statistical model in the present study in an attempt to emulate the system of production.For example, carcass weight in many studies is often adjusted for age at slaughter either as a covariable in the model (Eriksson et al., 2004;Kause et al., 2015) or as a correlated trait (Pritchard et al., 2021) so as to fairly compare breeds; however, it may not be biologically sensible to statistically adjust to a common age across all breeds with the same being true for carcass weight, but also not sensible to adjust all animals with a single covariate.Despite all this, the difference between breed effects documented in the present study are generally consistent in direction, albeit sometimes different in magnitude, compared with other cattle breed comparison studies (Alberti et al., 2008;Keane and Moloney, 2010;Fouz et al., 2013;Pritchard et al., 2021).In an analysis of the UK national catle database, Pritchard et al. ( 2021) reported a mean difference in carcass weight of purebred Hereford and Angus (i.e., traditional breeds) prime cattle (i.e., bulls, steers and heifers) and that of the continental breeds of Limousin, Charolais and Simmental of 69.6 kg while in the present study, the equivalent value (from half-breds) was 18.8 to 23.4 kg depending on gender; using the same scoring system as that in the present study the difference in carcass conformation between the 2 breed types investigated by Pritchard et al. (2021) was 3.2 units with the difference (again in half-breds) being 1.0 to 1.2 in the present study.Also based on the breeds investigated in the present study, albeit with no Herefords, Alberti et al. (2008) reported a mean difference between purebred Angus bulls and the average of Limousin, Charolais and Simmental of 28 kg carcass weight and 0.3 units of conformation score; the equivalent values for bulls in the present study was 23.6 kg and 1.1 units with Angus being lighter and poorer conformed than the continental breeds in both studies.Similarly, comparing purebred Hereford and Angus heifers and bulls combined against Charolais, Limousin and Simmental heifers and bulls, Kause et al. (2015) reported the latter group to be 27.7 kg heavier with 1.9 conformation units greater than that the former group of purebreds.Irrespective of the scale of the inter-breed difference in the present study, much of the ranking of beef breeds (as well as relative to Holstein-Friesian) are consistent with reported elsewhere but the present study puts a value on these relative differences where the dam was exclusively a Holstein-Friesian.The present study, however, also includes other beef breeds like the Aubrac and Salers that are not often included in such studies heretofore.The former point stresses that the reported breed effects in the present study being from dairy dams is important and should not be interpreted to represent actual breed effects.This is because, the sires used on dairy females may be a specific selection of the sires available from that breed; the beef sires chosen by dairy producers for mating are likely to be easier calving and shorter gestation.Given the known genetic associations between calving performance and carcass merit, sires genetically less predisposed to calving difficulty will, on average, be genetically lighter with possibly less well conformed carcasses (Eriksson et al., 2004).Breeding schemes and breeding goals are, of course, population specific and the results reported in the present study are really just applicable to the genepool available to Irish producers over the years investigated in the present study.Of the data used in the present study, 98% originated from sires born in the 10 years before and including 2018.

Within-breed versus between-breed selection
Using the breed effects of the Angus versus the Belgian Blue from the present study, the mean difference between both breeds for carcass weight, conformation and fat score was 3.1 to 5.4 sire standard deviation units; the corresponding metric for age at slaughter was 4.8 standard deviations.For calving assistance and dystocia, the mean difference between the Angus and Belgian Blue breeds represented 2.1 to 2.3 sire standard deviation units.Little difference between both breeds was evident for gestation length but the mean difference in gestation length between the Angus (shortest gestation) and the Limousin (longest gestation) was 2.7 sire standard deviations.Hence, considerable scope exists to achieve a step change in any individual trait by selecting a different breed as opposed to selection within breed.In a well-run progeny-testing breeding scheme, achieving a genetic gain of 0.22 standard deviations per year is expected (Schaeffer, 2006) implying that breed changes for some traits translates to several years of within-breed selection.It should, however, be noted that the inter-sire sire variability estimated in the present study may not reflect the true extent of the intra-breed differences since the sires mated to dairy cows are likely to be highly selected, not just because they are deemed to be genetically elite, but also because they were chosen by dairy producers to be appropriate to their needs.Nevertheless, simply comparing the inter-breed differences to the intra-breed variance for a single trait ignores covariance that exists among traits which are important given the recommendation to producers not to select on individual traits.
No one single breed excelled in all traits.The Angus had favorable calving characteristics but lagged behind other breeds in carcass weight and conformation score while the actual benefit in age at slaughter relative to other breeds was not huge (<1 mo).It should be noted that the breed effects reported in the present study are estimated from first crosses so a) the breed effect reported within is only half that what might be observed if the animal was purebred for that breed and b) any potential heterosis effect is encompassed within the breed effect estimated in the present study (except for the Holstein-Friesian breed) since all animals are firstcrosses and therefore it was not possible to disentangle the heterosis effect from the additive breed effect.Heterosis effects have been reported for calving (Arthur et al., 1989) and carcass (Kenny et al., 2022) traits although a general consensus does not exist (Olson et al., 2009).The size of the heterosis effect is a function of the genetic distance between the parents which here represents the difference between the Holstein-Friesian and the beef breeds.Using genotypes from Irish animals of different dairy and breeds, Kelleher et al. (2017) demonstrated how some breeds like the Belgian Blue are genomically more similar to the Holstein-Friesian in Ireland than other breeds like the Limousin, Charolais and Simmental.Nonetheless, this heterosis effect is a reality in beef-on-dairy production systems and is thus implicit within the reported breed effect.

Across-breed genetic evaluations
The main motivation for across-breed genetic evaluations is to be able to leverage data from crossbred animals when estimating the EBV of individual animals (including sires).The output are EBVs of animals of different breeds which should be comparable on the same scale.The beauty of this outcome is that producers should be able to fairly compare breeds against each other without having to know the actual underlying breed effects.Whether these mean breed effects can be fully captured by across-breed EBVs was explored in the present study.The expected regression coefficient of the phenotype (adjusted for fixed and random effects in the model) on the EBV of the (as calculated in the present study from half the sire's EBV plus quarter the maternal grandsire's EBV) animal is expected to be 0.75; for carcass weight, conformation and fat score (where the phenotype analyzed was in the same units as the EBV), the respective regression coefficients across the entire population was 0.69 (SE = 0.02), 0.61 (SE = 0.01) and 0.75 (SE = 0.03), respectively.However, like reported elsewhere (Connolly et al., 2016), the regression coefficients differed by gender -for example, the regression of phenotype on EBV for carcass weight was 0.74 (SE = 0.03), 0.60 (SE = 0.04) and 0.65 (SE = 0.04) for steers, heifers and bulls, respectively.While a regression coefficient of 0.75 is expected, deviations from the expectation are not unexpected (and have been reported -Martin et al., 2021;Ring et al., 2021).Causes for the deviations from expectations, particularly relevant to the present study include 1) likely parentage errors which are approximately 10-13% in Irish cattle (Purfield et al., 2016), 2) genotype-byenvironment re-scaling effects due to heterogeneous variance as demonstrated by the how the association between phenotype and EBV differed by animal gender and then considering that a large proportion of the data used in national genetic evaluations originate from beef herds, 3) genotype-by-environment interactions due to re-ranking where the genetic correlation between environments including the genetic background of the dam may not be unity, 4) the fact that the statistical model in the present study differs from that used in the genetic evaluation with the latter being able to disentangle the heterosis effects from the additive genetic effects, 5) sampling variation associated with the estimation of breed effects in the present study especially for some breeds with a smaller population size where most of the regression coefficients actually deviated from the expectations, 6) the present study which was based on dairy female matings only investigated a selection of the population so the regression coefficients may not hold (Martin et al., 2021), and 7) the genetic evaluation is not perfect especially for the less populous breeds.Other factors like systematic error in the phenotypes used in this study (e.g., updated carcass classification algorithms) relative to those used to generate the EBVs could also contribute to results deviating from expectations.Nonetheless, although adjusting for the EBV of the animal did not completely remove the breed effects, the significance of the breed effects weakened considerably following adjustment for EBV signifying that these across-breed EBVs can indeed be useful.

CONCLUSIONS
Clear breed differences existed for the range of traits investigated in the present study, the effect of some differing by animal gender; most of the breed effects were reflected in the across-breed genetic evaluations of individual animals.The traditional beef breeds (i.e., Angus and Hereford) tended to be easier calving with shorter gestation but had lighter and less conformed carcasses than the continental beef breeds although they were slaughtered younger.The breed effects, which were Berry et al.: BEEF BREED PERFORMANCE estimated using first-cross data from Holstein-Friesian dams can be useful for modeling the impact of breed substitution in herd breeding programs.Irrespective, the potential of within-breed genetic selection, especially within the framework of a total merit index, to achieve genetic gain should not be discounted.
Berry et al.: BEEF BREED PERFORMANCE Berry et al.: BEEF BREED PERFORMANCE

Figure 1 .
Figure 1.Odds ratios (95% confidence in parenthesis) of the sire breed effect (relative to a Holstein-Friesian sire) for a) calving assistance, b) calving dystocia, and c) perinatal mortality with adjustment for calf genetic merit for the respective trait (hollow squares) or without adjusted (filled squares) Berry et al.: BEEF BREED PERFORMANCE

Table 1 .
Berry et al.: BEEF BREED PERFORMANCE Number of records and predicted marginal means (standard error in parenthesis) in days for gestation length per sire breed not adjusted or adjusted for the gestation length estimated breeding value (EBV) of the calf

Table 3 .
Number of records and predicted marginal means (standard error in parenthesis) for calf price (€) of both female and male singleton calves < 42 d of age

Table 4 .
Number of records and predicted marginal means (standard error in parenthesis) for age at slaughter (d) as well as carcass price (€/ kg) for steer, heifer and bull progeny of a third parity Holstein-Friesian dairy cow mated to sire of different breeds

Table 5 .
Predicted marginal means (standard error in parenthesis) for carcass weight (kg), conformation (scale 1 [poor] to 15 [excellent]) and fat (1[thin] to 15 [fat]) score in steer, heifer and bull progeny of a third parity Holstein-Friesian dairy cow mated to a sire of a different breed Berry et al.: BEEF BREED PERFORMANCE