Breeding efficiency and its correlations with production traits in Holdeo (Holstein Friesian × Deoni) cattle

Introduction

A dairy cow returns milk only in proportion to how often she calves. Breeding efficiency condenses a cow's whole reproductive history into one figure by setting what she achieved against what an optimum cow would have achieved, and it is used in Indian herd evaluation because it summarises age at first calving and every subsequent calving interval in a single number that can be compared across animals and across herds. Two formulae are in common use and they do not measure quite the same thing. The method of Wilcox (1957) [21] works only from the calving intervals a cow completed. The method of Tomar (1965) [22] adds age at first calving to the denominator, so it penalises a heifer that entered the herd late even if she calved regularly afterwards. Raina et al. (2016) [16] and Singh et al. (2016) [14] applied both to the same crossbred cows and obtained 82.31 ± 0.97 percent by Wilcox and 98.14 ± 1.09 percent by Tomar, a gap of nearly sixteen points, which shows how much the choice of formula matters. Holdeo is the 50 percent Holstein Friesian × Deoni half-bred developed at the Cattle Cross Breeding Project, Parbhani, where crossing began in 1972 (Aglave et al., 2010) [1] . Milk production in this herd and its non-genetic sources of variation have been reported by Wondifraw et al. (2013) [18] and Mote et al. (2014) [8] , reproductive traits by Zewdu et al. (2015) [20] and lactation curve models by Thorat et al. (2019) [16] . Breeding efficiency has not been reported, and neither has its relationship with the production traits. That relationship decides whether a breeding programme built on milk records will carry reproductive performance along with it or leave it behind. Published estimates suggest the latter.

International Journal of Advanced Biochemistry Research https://www.biochemjournal.com Worku et al. (2021) [19] obtained heritabilities of 0.19 ± 0.03 for lactation milk yield and 0.06 ± 0.01 for calving interval in Karan Fries cattle, and Vinothraj et al. (2016) [16] obtained 0.222 ± 0.101 for calving interval and 0.177 ± 0.115 for dry period in Jersey × Red Sindhi crossbreds, where the standard error is as large as the estimate. The present study estimated breeding efficiency in Holdeo cattle by both methods, tested it against period of calving, season of calving and age at first calving group, and examined its correlations with the production traits.

Materials and Methods

Source of data

Traits studied

Five production traits were studied: lactation milk yield (LMY, kg), the trait recorded as peak milk yield (PMY, kg), days required to attain peak milk yield (DRPMY, days), lactation length (LL, days) and dry period (DP, days). Breeding efficiency (BE, percent) was the sixth trait.

Estimation of breeding efficiency

Breeding efficiency was computed by two methods. The method of Tomar (1965) [22] , as given by Singh et al. (2016) [14] , is:

and the method of Wilcox (1957) [21] is:

where, N = total number of calving intervals n = total number of parturitions AFC = age at first calving, days ΣCI = sum of all calving intervals, days D = interval from first to last parturition, days 365 = optimum calving interval, days 1020 = optimum age at first calving in cattle, days

The two expressions differ in one respect that governs everything reported below. Age at first calving appears in the Tomar denominator and does not appear in the Wilcox expression at all. Because D is itself the sum of the calving intervals, the Wilcox formula reduces to 365 divided by the mean calving interval, and is therefore blind to how late the heifer entered the herd.

Statistical analysis

Least squares means for breeding efficiency were obtained by the procedure of Harvey (1990) [5] using the MAUSTAT (2000) [7] statistical package, fitting the model:

where, Pᵢ = fixed effect of the ith period of calving (i = 1 to 5) Sⱼ = fixed effect of the jth season of calving (j = 1 to 3) Aₖ = fixed effect of the kth age at first calving group (k = 1 to 5) eᵢⱼₖₗ = random error Subclass means were compared by Duncan's multiple range test as modified by Kramer (1957) [6] . Heritability was estimated from the paternal half-sib analysis as h² = 4σ²s/(σ²s + σ²w), where σ²s is the sire component of variance, and repeatability as the intraclass correlation among repeated records of the same cow (Becker, 1984; Falconer and Mackay, 1996) [4] . Correlations among the six traits were tested against tabulated values of r and were declared significant at p < 0.05 and highly significant at p < 0.01.

Results and Discussion

Least squares means for breeding efficiency by the Tomar and Wilcox methods are given in Tables 1 and2, the analysis of variance for both in Table 3, heritability and repeatability estimates in Table 4, the correlation matrix in Table 5 and the comparison with published values in Table 6.

Breeding efficiency by the Tomar method

Breeding efficiency by the Tomar method averaged 72.55 ± 0.44 percent (Table 1 [16] and Singh et al. (2016) [14] the pattern is instructive. By the Wilcox method Holdeo scored 83.02 percent against their 82.31 percent, a difference of less than a point. By the Tomar method Holdeo scored 72.55 percent against their 98.14 percent, a shortfall of 25.59 points.

Once the formulae are set side by side the reason is plain.

On the calving intervals they actually completed, Holdeo cows performed as well as published crossbred herds. What separates them is the age at which they entered the herd: 1642 days against a Tomar optimum of 1020. The entire 25.59-point shortfall is the cost of that delay, and none of it is a shortfall in the calving intervals themselves. This is a reason to report both figures rather than choose between them. The Wilcox estimate says how well the cows cycled once they started; the Tomar estimate says how much of a cow's potential lifetime was spent waiting for the first calf. For a herd whose first calving is two years later than it should be, the second question is the more important one, and a study that reported only the Wilcox figure would have concluded that reproductive performance was satisfactory.

Heritability and repeatability

Heritability and repeatability estimates for the eight production and reproductive traits are given in Table 3. Every estimate was negative. Heritability ranged from -0.0537 ± 0.0202 for lactation length to -0.0583 ± 0.0331 for gestation period, and repeatability from -0.0437 ± 0.0175 to -0.0483 ± 0.0225 for the same two traits. Negative estimates arise when the sire component of variance computed from the analysis of variance falls below zero, which happens when the true value is small relative to the sampling error of a data set with few sires or few daughters per sire. The parameter cannot itself be negative. Steinsaltz et al. (2020) [15] have shown that truncating such estimates at zero introduces its own bias and that they are better reported than suppressed, which is why they are given here in full. Two features of Table 3 go beyond the usual explanation. First, seven of the eight heritability estimates lie within 0.0003 of one another, and the eighth within 0.0046. Eight biologically distinct traits, measured in different units and governed by different physiology, do not return the same variance ratio to three decimal places. Second, the difference between the heritability and the repeatability estimate is exactly -0.0100 for all eight traits without exception. Heritability and repeatability are different functions of different variance components and there is no mechanism by which their difference would be constant across eight traits.

Taken together these two features point to a computational artefact rather than to eight independent findings of zero heritability. The most likely explanation is that the grouping variable used in the analysis of variance carried no real sire structure, in which case the between-group and within-group mean squares are equal apart from rounding and the resulting intraclass correlation is fixed by the design rather than by the data. Whatever the cause, the estimates cannot be read as evidence about the genetic architecture of this population, and neither can any covariance component derived from the same analysis.

Published estimates show what the traits look like when the design supports them. Worku et al. (2021) [19] obtained 0.19 ± 0.03 for lactation milk yield, 0.06 ± 0.02 for lactation length and 0.06 ± 0.01 for calving interval in Karan Fries cattle. Vinothraj et al. (2016) [16] , working with 732 lactation records accumulated over 30 years, obtained 0.222 ± 0.101 for calving interval and 0.177 ± 0.115 for dry period.

Deshpande and Bonde (1981) [3] obtained 0.405 ± 0.023 for peak yield on 1346 crossbred cows. Reproductive traits are weakly heritable everywhere, and resolving them takes more sires and more daughters per sire than a single research herd usually holds.

Correlations among the traits Correlations among the production traits

The correlation matrix is given in Table 5. Lactation milk yield showed large negative correlations with the trait recorded as peak milk yield (r = -0.715, p < 0.01) and with lactation length (r = -0.668, p < 0.01), a positive correlation with dry period (r = 0.285, p < 0.01) and no significant association with days to peak (r = 0.148). Peak milk yield correlated positively with lactation length (r = 0.225, p < 0.05) and dry period (r = 0.224, p < 0.05). Days to peak correlated positively with lactation length (r = 0.239, p < 0.05) and dry period (r = 0.247, p < 0.05). Lactation length and dry period correlated positively (r = 0.285, p < 0.01).

The negative correlation between lactation milk yield and lactation length is explicable in this herd, and the explanation is a warning rather than a finding. Over the five periods lactation yield rose from 990.87 to 1651.77 kg while lactation length fell from 289.30 to 260.94 days. A correlation computed on records pooled across 26 years without removing that trend will return a negative coefficient between the two traits even where the withinperiod association is positive. Worku et al. (2021) [19] obtained genetic correlations above 0.80 among 305-day yield, lactation yield and lactation length in Karan Fries cattle, and phenotypic correlations that were also positive and high. The present coefficient measures the herd's history, not the relationship between the traits. The negative correlation between lactation milk yield and the trait recorded as peak milk yield is not explicable in the same way. Across the thirteen subclass means of this study the recorded peak yield equals lactation milk yield divided by lactation length to within 1.33 percent, so the two traits rose together across periods, from 990.87 kg and 3.38 kg in P 1 to 1651.77 kg and 6.38 kg in P 5 . Between the period means they correlate at + 0.97. A coefficient of -0.715 cannot be reconciled with that, nor with the positive coefficient of + 0.225 reported between peak yield and lactation length in the same matrix. Deshpande and Bonde (1981) [3] found peak yield positively and highly significantly correlated with total lactation yield on both the genetic and the phenotypic scale in 1346 crossbred cows. This estimate should be re-derived before it is used.

Correlations of breeding efficiency with the production traits

Breeding efficiency correlated positively with dry period (r = 0.609, p < 0.01) and lactation length (r = 0.255, p < 0.01), and showed no significant association with lactation milk yield (r = 0.169), peak milk yield (r = -0.135) or days to peak (r = -0.048).

The coefficient for dry period is the largest in the matrix and it runs against the formula that defines the trait. Dry period is a component of the calving interval, and the sum of the calving intervals is the The reported + 0.609 is consistent with the second of these patterns and not with the first. It reflects confounding by age at first calving rather than any relationship between udder rest and reproductive efficiency, and it should not be read as a reason to lengthen the dry period. Age at first calving should be fitted as a covariate before the correlation is estimated.

The absence of a significant association between breeding efficiency and lactation milk yield is the practically important result and it is not affected by any of the difficulties above. Selection on milk records alone would not be expected to move breeding efficiency in this population. Reproductive performance will not arrive as a by-product of selection for yield; it has to be recorded, selected for and managed on its own account.

Implications for the herd

Three conclusions follow, in descending order of confidence.

The first is arithmetic and does not depend on any genetic parameter. Every month by which first calving is delayed costs 0.75 percentage points of breeding efficiency by the Tomar measure, and the herd's age at first calving of 1642 days is roughly two years later than the 1020 days the formula treats as optimum. On the calving intervals the cows actually completed they performed as well as published crossbred herds. The whole of the deficit sits in the rearing period before first calving, which is also the part of the production cycle a farm controls most directly, through heifer nutrition, growth monitoring, breeding on body weight rather than age, and heat detection in maiden heifers.

The second concerns the choice of measure. Reporting breeding efficiency by the Wilcox method alone would have shown this herd performing normally. It is the Tomar method that locates the problem, because it is the only one of the two that counts the cost of a late start. Herd evaluations in populations with delayed first calving should report both. The third is a limitation. The heritability and repeatability estimates obtained here are not usable, and the correlations computed alongside them inherit that. Genetic parameters for Holdeo cattle should be re-estimated on a pedigreeconnected data set with an adequate number of sires and daughters per sire, with the fixed effects of period, season and age at first calving removed before the covariances are computed. Until that is done the correlations in Table 4 mark out where to look rather than what to conclude. 3.

Table 2:

Least squares means ± SE for breeding efficiency (percent) in Holdeo cows by the Wilcox (1957) [21] [22] method 72.55% 98.14% Raina et al. (2016) [16] ; Singh et al. (2016) [14] -25.59 points BE, Wilcox (1957) [21] method 83.02% 82.31% Raina et al. (2016) [16] ; Singh et al. (2016) [14] + 0.71 points Age at first calving 1642 d 1020 d (Tomar optimum) Tomar (1965) [22] formula + 622 d h², lactation milk yield -0.0539 0.19 ± 0.03 Worku et al. (2021) [19] , Karan Fries -h², calving interval -0.0540 0.06 ± 0.01; 0.222 ± 0.101 Worku et al. (2021) [19] ; Vinothraj et al. (2016) [16] -h², peak yield -0.0539 0.405 ± 0.023 Deshpande and Bonde (1981) [3] -Published breeding efficiency values are for crossbred cattle at an organised dairy farm.

Conclusion

Breeding efficiency in Holdeo cattle averaged 72.55 ± 0.44 percent by the Tomar (1965) [22] method and 83.02 ± 0.82 percent by the Wilcox (1957) [21] method. By the Tomar method it was affected by period of calving (p < 0.01) and very strongly by age at first calving group (F = 78.64, p < 0.01), falling from 85.73 ± 1.15 percent in cows calving first before 1200 days to 58.89 ± 1.10 percent in those calving after 2100 days, a loss of 0.75 percentage points for every month of delay. By the Wilcox method no effect was significant, because age at first calving does not enter that formula. Compared with published crossbred herds the Holdeo cows matched the Wilcox benchmark almost exactly, 83.02 against 82.31 percent, but fell 25.59 percentage points short of the Tomar benchmark. The shortfall lies entirely in age at first calving and not in the calving intervals the cows completed. Reducing age at first calving towards 30 months is therefore the single change that would do most for reproductive efficiency in this herd. Heritability and repeatability estimates for all eight traits were negative, differed from one another by less than 0.0005 in seven of eight cases, and differed from their repeatability counterparts by exactly 0.0100 in every trait. These features indicate a computational artefact rather than independent estimates, and the correlations computed from the same analysis should be treated as preliminary. Of those correlations, the absence of any significant association between breeding efficiency and lactation milk yield is the one that carries a clear practical message: reproductive performance in Holdeo cattle will not improve as a consequence of selection for milk production.

Acknowledgement

AcknowledgementThe authors thank the Director of Instruction and the staff of the Cattle Cross Breeding Project, Vasantrao Naik Marathwada Krishi Vidyapeeth, Parbhani, for access to the herd records, and the Head, Department of Animal Husbandry and Dairy Science, College of Agriculture, Latur, for the facilities provided.

References

1. Aglave PM, Adangale SB, Walkunde TR, Kale VA, Poul SP. Effect of genetic factors on mortality in crossbred cattle Holdeo (HF × Deoni). Veterinary World. 2010;3(1):32-33.
2. Becker WA. Manual of Quantitative Genetics. 4th edn. Pullman, Washington, USA: Academic Enterprises; 1984.
3. Deshpande KS, Bonde HS. Genetic studies on peak yield in Holstein Friesian × Sahiwal cross-bred cattle. The Journal of Agricultural Science. 1981;97(3):707711. doi:10.1017/S0021859600037072.
4. Falconer DS, Mackay TFC. Introduction to Quantitative Genetics. 4th edn. Harlow, UK: Longman; 1996.
5. Harvey WR. User's Guide for LSMLMW and MIXMDL: PC-2 Version. Columbus, USA: Ohio State University; 1990.
6. Kramer CY. Extension of multiple range tests to group means with unequal numbers of replications. Biometrics. 1957;13:13-18.
7. MAUSTAT. Statistical Package. Parbhani, Maharashtra, India: Marathwada Agricultural University; 2000.
8. Mote SS, Chauhan DS, Ghosh N. Effect of environmental factors on lactation yield and lactation length of Holdeo crossbred cattle. Indian Journal of Applied Research. 2014;4(10):53-57.
9. Raina V, Narang R, Malhotra P, Kaur S, Dubey PP, Tekam S, et al. Breeding efficiency of crossbred cattle and Murrah buffaloes at organized dairy farm. Indian Journal of Animal Research. 2016;50(6):867-871. doi:10.18805/ijar.v0iOF.6663.
10. Ratwan P, Chakravarty AK, Kumar M. Evaluation of relative economic value of performance traits in Sahiwal cattle. Indian Journal of Animal Sciences. 2022;92(10):1228-1232.
11. Ratwan P, Kumar M, Chakravarty AK. Bayesian approach for assessment of co-variances and genetic parameters of production and reproduction traits in Sahiwal cattle. Tropical Animal Health and Production. 2024;56(8):299.
12. Robertson A. The sampling variance of the genetic correlation coefficient. Biometrics. 1959;15(3):469485.
13. Roy I, Rahman M, Karunakaran M, Gayari I, Baneh H, Mandal A. Genetic relationships between reproductive and production traits in Jersey crossbred cattle. Gene. 2024;894:147982. doi:10.1016/j.gene.2023.147982.
14. Singh V, Narang R, Singh SG, Kaur A. Comparison of two methods of calculating breeding efficiency of crossbred cattle and Murrah buffaloes. Journal of Animal Research. 2016;6(5):885-889. doi:10.5958/2277-940X.2016.00114.5.
15. Steinsaltz D, Dahl A, Wachter KW. On negative heritability and negative estimates of heritability. Genetics. 2020;215(2):343-357. doi:10.1534/genetics.120.303161.
16. Thorat BN, Thombre BM, Bhoite UY. Comparative study of fitting various lactation curve models in Holdeo crossbred cattle. Indian Journal of Animal Research. 2019;53(11):1538-1542. doi:10.18805/ijar.B-3590.
17. Vinothraj S, Subramaniyan A, Venkataramanan R, Joseph C, Sivaselvam SN. Genetic evaluation of reproduction performance of Jersey × Red Sindhi crossbred cows. Veterinary World. 2016;9(9):10121017.
18. Wondifraw Z, Thombre BM, Bainwad DV. Effect of non-genetic factors on milk production of Holstein Friesian × Deoni crossbred cows. International Journal of Livestock Production. 2013;4(7):106-112. doi:10.5897/IJLP2013.0173.
19. Worku D, Gowane GR, Kumar R, Chandran PC, Verma MR, Alex R. Estimation of genetic parameters for production and reproductive traits in Indian Karan-Fries cattle using multi-trait Bayesian approach. Tropical Animal Health and Production. 2021;53:369. doi:10.1007/s11250-021-02806-z.
20. Zewdu W, Thombre BM, Bainwad DV. Studies on some non-genetic factors affecting reproductive performance of Holstein Friesian × Deoni crossbred cows. African Journal of Agricultural Research. 2015;10(12):1508-1516. doi:10.5897/AJAR2014.8952.
21. Wilcox CH. Debye potentials. Journal of Mathematics and Mechanics. 1957 Jan 1;6(2):167-201.
22. Tomar NS. A note on the method of working out breeding efficiency in Zebu cows and buffaloes. 1965.