- Email: [email protected]
Evaluation of Overall Reproductive Performance of Dairy Herds
PHYSIOLOGY AND MANAGEMENT Evaluation of Overall Reproductive Performance of Dairy Herds J.C.B. PLAIZIER,* K. D. LISSEMORE,† D. KELTON,† and G. J. KING* *Department of Animal and Poultry Science, University of Guelph, Guelph, ON, Canada N1G 2W1 †Department of Population Medicine, Ontario Veterinary College, Guelph, ON, Canada N1G 2W1
ABSTRACT Reproductive efficiency was assessed in 106 commercial dairy herds in Ontario. Herds that were similar in historical calving interval or projected calving interval varied widely in reproductive culling rate. The correlation between the reproductive culling rate and the historic calving interval and between the reproductive culling rate and the projected calving interval were –0.23 and were not significant. The coefficient of variation associated with the reproductive culling rate (75.7%) was much higher than that of the historical calving interval (3.8%) or the projected calving interval (4.5%). The repeatability of the reproductive culling rate (0.26) was much lower than that of the historical calving interval (0.81) or the projected calving interval (0.60) because culling rate was not only influenced by the overall herd reproductive performance, but also by the differences in the maximal allowable days open for rebreeding and the estimation of which cows were culled for reproductive failure. Additionally, a high coefficient of variation for the reproductive culling rate is unavoidable because this variable has a binomial distribution and a low mean value (7.5%). These disadvantages are not sufficient to exclude this measure from a comprehensive assessment of overall herd reproductive performance. The projected calving interval and the reproductive culling rate can be combined into the adjusted calving interval and thus interpreted jointly. On 40.5% of the farms providing data for this study, fertility assessment based on the projected calving interval indicated better reproductive performance than that based on adjusted calving interval. Hence, the assessment of herd fertility on the basis of projected calving interval can lead to a different and erroneous conclusion regarding the level of herd fertility than that obtained with the evaluation based on adjusted calving interval.
Received May 30, 1997. Accepted March 3, 1998. 1998 J Dairy Sci 81:1848–1854
( Key words: herd reproductive performance, calving interval, culling, evaluation) Abbreviation key: ACI = adjusted calving interval, CPI = interval from calving to pregnancy, HCI = historical calving interval, HRP = herd reproductive performance, NCR = fraction of cows not culled for reproductive failure, PCI = projected calving interval, RCR = reproductive culling rate, VWP = voluntary waiting period. INTRODUCTION Herd reproductive performance ( HRP) (i.e., reproductive efficiency) of dairy herds influences profitability because it affects milk production, reproductive culling rates ( RCR) , and animal sales. The HRP attained within a dairy herd is influenced by the care and attention provided by the manager, herdpersons, inseminators, and others involved with husbandry, herd health, or feeding management. The trend toward larger herds and increased mechanization, resulting in less contact time with caretakers per cow, often results in reduced breeding efficiency if reproductive management is not strengthened. Thus, a regular analysis of the overall HRP is required to assess whether management is satisfactory or whether an investigation into the causes for poor HRP is necessary ( 5 ) . De Kruif ( 1 ) concluded that, particularly in small herds, the values of fertility indices will be determined not only by HRP, but also by chance, and, in that case, correct interpretation of the data on reproduction will only be possible when based on the situation over a number of years. Because of differences in distribution, the value of some reproductive indices will be more affected by random chance than others, which should be taken into account when these indices are being interpreted. The calving interval is the measure most commonly used to assess the overall HRP in dairy herds (4, 8, 12, 21). The reductions in milk production and profit caused by extended calving intervals have been demonstrated frequently (2, 7, 9, 10, 14, 17, 19). Calving interval can be calculated as the duration of the interval between the two most recent calvings for
1848
OVERALL HERD REPRODUCTIVE PERFORMANCE
all parturient cows in a herd. This measure, which should be referred to as the historical calving interval ( HCI) , has several inherent limitations (8, 21). One major deficiency is that first lactation cows are excluded from the measure as they have not had two calvings. Second, cows culled for reproductive failure also do not contribute to the measure. Thus, an apparently acceptable calving interval might misrepresent actual herd performance because infertile cows and cows with poor fertility are often culled from the herd (8, 15, 21). Hence, a high RCR might result in a low average calving interval, even in herds with serious reproductive problems. Therefore, calving intervals should be interpreted in combination with RCR ( 4 ) . The HCI is based on historical events (i.e., past calvings), and does not reflect recent changes in reproductive performance, which is a serious disadvantage because proactive assessment and planning of reproductive management require current information on reproductive performance. Fetrow et al. ( 6 ) , therefore, described the projected calving interval ( PCI) , a measure that is based on the projected minimum average days open plus a standard gestation length. For the estimation of the projected minimum average days open, it was assumed that all bred cows conceived at their last breeding and that cows not yet bred but past the voluntary waiting period will conceive 10 d after the day of calculation. This measure is an optimistic estimate of the minimum average days open that will be experienced by the herd in the future (6, 8, 21). The PCI is considered to be an improvement over the HCI, because PCI is based on a larger proportion of the herd and provides more current information. Accurate pregnancy diagnosis is essential for the use of the PCI. However, because the PCI is a best estimate, the HCI calculated at a later date will most likely be longer. This difference should be taken into account when the data for PCI are being interpreted. Esslemont ( 4 ) introduced the calving interval adjusted calving rate, which is the proportion of a herd that recalves within 12 mo. The advantage of this measure is that it includes the calving interval and the culling rate in one measure. Following the same logic as used for the calving interval adjusted calving rate, Plaizier ( 1 6 ) and Plaizier et al. ( 1 7 ) introduced the adjusted calving interval ( ACI) . The ACI is calculated by adjusting the PCI for the percentage of the cows in the herd that were culled for failure to conceive; the PCI was divided by the ratio of the cows that were not culled for reproductive failure. Hence, ACI is a more comprehensive measure than PCI. A high RCR reduces PCI but increases ACI. Hence, an
1849
assessment of herd fertility on the basis of ACI could result in a different, more accurate conclusion than that based on PCI. Plaizier et al. ( 1 7 ) estimated financial losses from suboptimal reproductive performance of the herd on the basis of PCI and ACI using data generated by computer simulation. Those researchers ( 1 7 ) found that estimation based on ACI was more accurate than estimation based on PCI, as ACI had a higher correlation with net revenue than did PCI. However, to study the usefulness of ACI for the routine evaluation of reproductive performance of commercial dairy herds, it should be determined whether this measure can be calculated accurately with data that are routinely collected by the farmer. The objectives of this study were to study the distributions of measures of herd fertility and correlations between them and to determine whether the evaluation of herd fertility based on ACI would lead to a different result than an evaluation based on PCI. MATERIALS AND METHODS Data Data were obtained from herds participating in the Ontario Dairy Monitoring and Analysis Program ( 1 1 ) and were collected between March 1, 1990 and February 28, 1992. For each farm included in the study, The RCR, calving to pregnancy interval ( CPI) in days, HCI, and voluntary waiting period ( VWP) (i.e., the duration of the period after calving that cows are deliberately not bred) were recorded. The HCI was calculated from the data provided by the Ontario DHI Corporation in December 1990 (yr 1 ) and December 1991 (yr 2). Year 1 included 106 herds, and yr 2 included 97 herds. The mean herd size was 49.1 in yr 1 and 49.7 in yr 2. In both years, the herd sizes ranged between 21 and 146. Differences between the size of individual herds in yr 1 and yr 2 ranged from –7 and 6 cows. The RCR was calculated for the 12-mo period ending in December 1990 and in December 1991 as the percentage of the herd reported by the farmer as failing to conceive. For all cows culled from the herd, farmers were asked to choose the main reason for removal from the following options: sold for dairy, culled for poor reproduction, culled for mastitis, culled for poor production, culled for feet and legs, died, or culled for other reasons. The HCI was calculated as the herd average for the interval between the two most recent calvings. The CPI considered all breeding during the two 12-mo periods of interest and was calculated as the mean duration of the interval between calving and the following Journal of Dairy Science Vol. 81, No. 7, 1998
1850
PLAIZIER ET AL.
TABLE 1. Intervals of projected calving interval (PCI) and adjusted calving interval (ACI) corresponding to herd reproductive performance (HRP) scores. HRP Score
PCI
1 2 3 4 5 6 7
16.0–16.6 15.4–15.9 14.7–15.3 14.1–14.6 13.4–14.0 12.7–13.3 12.0–12.6
ACI (mo) 17.9–18.8 17.0–17.8 16.0–16.9 15.0–15.9 14.0–14.9 13.1–13.9 12.0–13.0
breeding that resulted in a confirmed pregnancy. Cows that did not establish a pregnancy were not included in this measure. The PCI, which was expressed in months, was calculated from the CPI using the following formula: PCI = (CPI + 280)/30.5. The ACI, which was also expressed in months, was calculated by dividing the PCI by the fraction of the cows that were not culled for reproductive failure ( NCR) : ACI = PCI/NCR. The NCR is based on RCR: NCR = (100 – RCR)/100.
Analyses The distributions of all fertility indices were tested for normality using the univariate procedure of SAS (18). If these distributions were not normal, then the data were transformed in an attempt to obtain normal distributions. Transformations included log, square root, inverse, and inverse square root. If transformation resulted in normal distributions, then Pearson regression coefficients were computed. Otherwise, Spearman rank correlation coefficients were computed. Correlation coefficients were calculated among the different measures of HRP for each herd in yr 1 and 2 ( 1 8 ) to estimate their repeatabilities. The ranges between the lowest and the highest PCI and ACI were divided into seven equally spaced intervals. These intervals were used to give herds an HRP score based on their PCI and a HRP score based on their ACI (Table 1). For each farm in each year, the Journal of Dairy Science Vol. 81, No. 7, 1998
HRP score that was based on the ACI was subtracted from the HRP score that was based on the PCI to assess whether evaluation of the overall herd reproductive performance that was based on ACI would lead to different conclusions than that based on PCI. RESULTS AND DISCUSSION Fertility indices did not have normal distributions in both years. Normal distributions were not achieved by the transformations used. Hence, Spearman rank correlation coefficients were computed between fertility indices (18, 20). The summary statistics for reproductive performance (Table 2 ) indicated that the herds included in the survey had a wide range in HRP, which was similar to that observed in comparable surveys (3, 13). The mean HCI and mean PCI were similar to the 13.1 mo average for HCI for Ontario dairy herds (13). The mean HCI and mean PCI were nearly equal, which was not expected because PCI was the best estimate of the future average calving interval. Hence, PCI was expected to be lower than HCI. The RCR varied from 0 to 30% among herds, which resulted in a high coefficient of variation for RCR compared with those of HCI and PCI. The wide ranges in HCI, PCI, and RCR are explained by differences in HRP among farms, by differences in the maximum allowable days open for breeding, and by the estimation of which cows were culled for failure to conceive. An additional factor that contributed to the high coefficient of variation of RCR between farms was that RCR is a binomial variable, and its average is low. Hence, RCR has a high coefficient of variability within each farm (20), which was demonstrated by the correlation coefficient between RCR in yr 1 and RCR in yr 2 (i.e., the repeatability of RCR) compared with the repeatabilities of HCI and PCI (Table 3). The repeatability of HCI was higher than that of PCI. This result can be attributed to the differences in the calculation of these intervals. The HCI is based on the
TABLE 2. Summary statistics for historical calving interval (HCI), projected calving interval (PCI), reproductive culling rate (RCR), adjusted calving interval (ACI), and voluntary waiting period (VWP). Measure
X
SD
Minimum
Maximum
CV
HCI, mo PCI, mo RCR, % ACI, mo VWP, d
13.2 13.1 7.5 14.2 54.3
0.6 0.6 5.2 1.1 8.8
11.9 11.9 0 12.0 40
15.9 16.6 30 18.9 90
(%) 4.5 4.6 69.3 7.7 16.2
OVERALL HERD REPRODUCTIVE PERFORMANCE TABLE 3. Spearman rank correlation reproductive indices in yr 1 and 2.
coefficients
1851
between
Measure1
Correlation coefficient
P
HCI RCR PCI ACI
0.81 0.26 0.60 0.29
0.001 0.01 0.001 0.05
1HCI = Historical calving interval, RCR = reproductive culling rate, PCI = projected calving interval, and ACI = adjusted calving interval.
occurrence of two subsequent calvings that occurred, while PCI is based on one calving and the best estimate for the subsequent calving. The uncertainty of when the next conception will occur, combined with abortions, errors in pregnancy diagnosis, and culling of pregnant cows results in higher variability and lower repeatability of PCI than of HCI. Scatter plots of RCR versus HCI and RCR versus PCI appear in Figures 1 and 2, respectively. Within short ranges of PCI and HCI, RCR varied widely, which was also demonstrated by the correlation coefficients between RCR and HCI and between RCR and PCI (Table 4). This coefficient was low and negative for RCR and HCI, indicating that high values of HCI were associated with low values of RCR. The correlation between RCR and PCI was not
Figure 1. Relationship between historical calving interval (HCI) and culling rate for reproductive failure (RCR).
Figure 2. Relationship between projected calving interval (PCI) and reproductive culling rate (RCR).
significant ( P = 0.67). Thus, herds with similar PCI and HCI could vary widely in RCR, which illustrates that differences in the overall reproductive performance between farms can be expressed as differences in PCI and as differences in RCR. Herds with similar rates of estrus detection and conception can vary in PCI and RCR because of differences in the duration of the rebreeding period (the maximal allowable number of days open or number of inseminations for breeding) and the VWP. In herds with a short period before rebreeding, when cows were not bred after two or three inseminations or a limited number of days open (e.g., <200 d), calving intervals were short even when estrus detection rates or conception rates were poor. In these herds, the variation in reproductive performance was expressed as variation in RCR. If breeding continued after many inseminations and many days open, then variation in estrus detection and conception rates was expressed as the variation in calving intervals and not in RCR. Thus, if reproductive performance is to be assessed, both calving interval and RCR should be considered by evaluating PCI and RCR separately or by evaluating the ACI. The advantage of using the ACI is that one measure can be used for a comprehensive assessment. Financial losses caused by suboptimal herd fertility can be estimated more accurately on the basis of ACI than on the basis of PCI (17). The correlation coefficient between HCI and ACI was only 0.29 (Table 4). This result is not surprising Journal of Dairy Science Vol. 81, No. 7, 1998
1852
PLAIZIER ET AL. TABLE 4. Spearman rank correlation coefficients between historical calving interval (HCI), reproductive culling rate (RCR), projected calving interval (PCI), adjusted calving interval (ACI), and voluntary waiting period (VWP). HCI HCI RCR PCI ACI VWP 1Level
1
RCR –0.23 1
(0.001) 1
PCI
ACI
VWP
0.55 (0.0001) –0.03 (0.67) 1
0.19 (0.006) 0.73 (0.0001) 0.59 (0.0001) 1
0.29 –0.15 0.25 0.07 1
(0.0001) (0.03) (0.0004) (0.30)
of significance.
because these measures encompass different time periods and the correlation between HCI and PCI and between HCI and RCR (the components of ACI) were positive and negative, respectively. Hence, the correlation between ACI (i.e., the combined effect of PCI and RCR) and HCI was small. The correlation between RCR and ACI was higher than between PCI and ACI, indicating that more variation in ACI is explained by RCR than by PCI. This difference was expected because the coefficient of variation of RCR was much higher than that of PCI. A positive correlation was observed between VWP and HCI and between VWP and PCI, which indicated that part of the variation in HCI and PCI was explained by VWP. Therefore, for a comprehensive assessment of reproductive performance using PCI or HCI, the VWP should be considered. In contrast, the correlation be-
tween ACI and VWP was not significant in both years, which partially was caused by the higher variation of ACI than of HCI and PCI. This absence of a significant correlation does not imply that VWP should not be considered for the interpretation of ACI. The frequency distributions of the HRP scores based on PCI and ACI (Figure 3 ) show that these distributions were skewed. Because the duration of a calving interval has a biological minimum but not a biological maximum, distributions were expected to be skewed. The frequencies of scores 5 and 6 were highest. Figure 4 shows the distribution of the HRP score based on PCI minus the HRP score based on ACI. If this difference is 0, then PCI and ACI indicated the same level of HRP. A positive value resulted when PCI indicated a higher level of HRP than ACI, and values were negative when ACI indicated a higher level of HRP. For about one-half of the farms
Figure 3. Frequency distributions of herd fertility scores based on projected calving interval (PCI) and adjusted calving interval (ACI).
Figure 4. Difference between herd fertility score based on projected calving interval (PCI) and herd fertility score based on adjusted calving interval (ACI).
Journal of Dairy Science Vol. 81, No. 7, 1998
1853
OVERALL HERD REPRODUCTIVE PERFORMANCE
(44.3%), PCI indicated the same level of reproductive performance as did ACI (Figure 4). However, for 46.3% of all evaluations, PCI indicated better reproductive performance than did ACI. For 14.3% of all evaluations, the difference between the PCI HRP score and the ACI HRP score was ≥2. In 3 of the 26 herds that had this difference, the data from both yr 1 and 2 were affected. Differences between the PCI HRP score and the ACI HRP score are explained by differences in RCR. A positive value for this difference was caused by a high RCR. For 9.4% of all evaluations, the value for this difference was negative, which was caused by a low RCR, ranging from 0 to 4.6%, compared with that for other herds. The large difference between the PCI HRP score and the ACI HRP score in yr 1 and yr 2 on only a few farms indicates that RCR can vary widely between years. As mentioned earlier, this wide variation is attributed to differences in HRP and reproductive culling policy between years but also to the distribution of RCR. The evaluation of herd fertility based on PCI could, therefore, lead to an assessment that is different from that based on ACI. The assessment using ACI might be more accurate than that using PCI because of the consideration of cows that were culled for failure to conceive. The ACI can be calculated using routinely collected reproductive data. However, as discussed earlier, the accuracy of RCR might differ among farms, and different interpretations of how RCR should be calculated may exist. Many cows are not culled for one specific reason, but have primary, secondary, and tertiary reasons for removal (5, 9), and failure to conceive might not be the only reason for culling. Cows with low milk production or with disease or other problems might be at increased risk of being culled for reproductive failure. Farmers might not continue breeding problem animals as long as other normal herdmates exist, thus obscuring differences between reasons for removal. Because of the need to include the reproductive culls in the evaluation of reproductive performance, this aspect should be addressed through industry extension programs. A definition of RCR that does not allow differences in interpretation should be adopted (22). The following calculation is proposed: RCR = total number cows culled for reproductive failure/total number cows available for rebreeding. The total number of cows culled for reproductive failure includes those cows that were rebred at least once but failed to conceive plus any that the operator intended to rebreed but did not. The total number of cows available for rebreeding is the number of cows that could have been rebred, including all of the cows
TABLE 5. Adjusted calving interval (ACI) corresponding to values for projected calving interval (PCI) and ratio of cows not culled for reproductive failure (NCR) and interpretation of ACI.1 PCI NCR
12
12.5
13
13.5
14
0.95
12.6 (E) 13.3 (G) 14.1 (CI) 15.0 (SI) 16.0 (SP)
13.2 (G) 13.9 (G) 14.7 (CI) 15.6 (SP) 16.7 (SP)
13.7 (G) 14.4 (CI) 15.3 (SI) 16.3 (SP) 17.3 (SP)
14.2 (CI) 15.0 (SI) 15.9 (SI) 16.9 (SP) 18.0 (SP)
14.7 (CI) 15.6 (SI) 16.5 (SP) 17.5 (SP) 18.7 (SP)
0.9 0.85 0.8 0.75
1E = Excellent, G = good, CI = could improve, SI = should improve, and SP = serious problem.
that eventually became pregnant plus all of those culled for reproductive failure. This value corresponds to the total number of cows in the herd minus any voluntary culls that the operator never intended to rebreed and involuntary culls that were removed for reasons other than reproductive culls. The RCR is calculated for the preceding 12 mo. Some general guidelines for the interpretation of ACI are given in Table 5. Assessment of ACI will indicate if reproductive performance needs to be improved and if causes for suboptimal herd fertility need to be investigated. However, ACI cannot identify the causes underlying poor fertility. Problem identification and resolution involve a holistic evaluation of the entire production system including past records, clinical examinations, and management practices. CONCLUSIONS Differences in overall reproductive performance among herds can be expressed as differences in calving interval (i.e., PCI or HCI) and as differences in RCR. The correlation between these measures is low. Hence, for a comprehensive evaluation of overall HRP, both a measure of calving interval and RCR need to be considered. Disadvantages of RCR are that by nature of its distribution it has a high coefficient of variation and low repeatability and that herd operators vary in their perception of which cows were culled for reproductive failure. Hence, the adoption of a uniform definition of RCR is essential. These disadvantages do not provide sufficient reason to ignore RCR when overall HRP is being evaluated. Adjusted calving interval, which is based on PCI and RCR, provides a more accurate assessment of overall HRP than does PCI alone. As shown by the differences Journal of Dairy Science Vol. 81, No. 7, 1998
1854
PLAIZIER ET AL.
between PCI HRP score and ACI HRP score, evaluation of HRP on the basis of PCI and this evaluation on the basis of ACI can lead to different conclusions. REFERENCES 1 De Kruif, A. 1978. Factors influencing the fertility of a cattle population. J. Reprod. Fertil. 54:507–518. 2 Dijkhuizen, A. A., J. Stelwagen, and J. A. Renkema. 1985. Economic aspects of reproductive failure. Financial loss at farm level. Prev. Vet. Med. 3:251–263. 3 Dohoo, I. R., S. W Martin, A. H. Meek, and W.C.D. Sandals. 1983. Disease, production and culling in Holstein-Friesian cows. I. The data. Prev. Vet. Med. 1:321–334. 4 Esslemont, R. J. 1992. Measuring dairy herd fertility. Vet. Rec. 131:209–212. 5 Etherington, W. G., J. Fetrow, B. E. Seguin, W. E. Marsh, L. D. Weaver, and C. L. Rawson. 1991. Dairy herd reproductive health management: evaluating dairy herd reproductive performance-part II. Comp. Contin. Educ. Prac. Vet. 13: 1491–1502. 6 Fetrow, J., D. McClary, R. Harman, K. Butcher L. Weaver, L. E. Studer, J. Ehrlich, W. Etherington, W. Guterbock, D. Klingborg, J. Reneau, and N. Williamson. 1990. Calculating selected reproductive indices: recommendations of the American Association of Bovine Practitioners. J. Dairy Sci. 73:78–90. 7 Funk, D. A., A. E. Freeman, and P. J. Berger. 1987. Effects of previous days open, previous days dry and present days open on lactation yield. J. Dairy Sci. 70:2366–2373. 8 Gaines, J. D. 1989. The role of record analysis in evaluating subfertile dairy herds. Vet. Med. 84:532–543. 9 Holmann, F. J., C. R. Shumway, R. W. Blake, R. B. Schwart, and E. M. Sudweeks. 1984. Economic value of days open for Holstein cows of alternative milk yields. J. Dairy Sci. 67: 636–643.
Journal of Dairy Science Vol. 81, No. 7, 1998
10 James, A. P., and R. J. Esslemont. 1979. The economics of calving intervals. Anim. Prod. 29:157–162. 11 Kelton, D. 1995. Monitoring and investigating the relationship among health, management, productivity and profitability on Ontario dairy farms. Ph.D. Diss., Univ. Guelph, Guelph, ON, Canada. 12 King, G. J. 1993. Reproductive performance and problems. Pages 531–565 in Reproduction in Domesticated Animals. Vol. B 9. Elsevier Sci. Publ. B.V., Amsterdam, The Netherlands. 13 Lissemore, K. D., K. E. Leslie, P. I. Menzies, S. W. Martin, A. H. Meek, and W. E. Etherington. 1992. Implementation and use of a microcomputer-based management information system to monitos dairy herd performance. Can. Vet. J. 33:114–120. 14 Olds, D., T. Cooper, and F. A. Thrift. 1979. Relationship between milk yield and fertility in dairy cattle. J. Dairy Sci. 62: 1140–1141. 15 Peters, A. R., and P.J.H. Ball. 1995. Reproduction in Cattle. 2nd ed. Butterworth & Co. Publ., London, England. 16 Plaizier, J.C.B. 1996. A study on the relationship between economic and reproductive efficiency in Ontario dairy herds using computer simulation. Ph.D. Diss., Univ. Guelph, Guelph, ON, Canada. 17 Plaizier, J.C.B., G. J. King, J.C.M. Dekkers, and K. Lissemore. 1997. Estimation of economic values of indices for reproductive performance in dairy herds using computer simulation. J. Dairy Sci. 80:2775–2783. 18 SAS® User’s Guide, Version 6 Edition. 1990. SAS Inst., Inc., Cary, NC. 19 Schmidt, G. H. 1989. Effect of length of calving intervals on income over feed and variable costs. J. Dairy Sci. 72:1605–1611. 20 Snedecor, G. W., and W. G. Cochran. 1978. Statistical methods. 7th ed. Iowa State Univ. Press, Ames. 21 Upham, G. L. 1991. Measuring dairy herd reproductive performance. Bovine Pract. 26:49–56. 22 Weaver, L. D. 1992. Reproductive health programs. Pages 99–109 in Large Dairy Herd Management. H. H. Van Horn and C. J. Wilcox, ed. Am. Dairy Sci. Assoc., Champaign, IL.
ABSTRACT Reproductive efficiency was assessed in 106 commercial dairy herds in Ontario. Herds that were similar in historical calving interval or projected calving interval varied widely in reproductive culling rate. The correlation between the reproductive culling rate and the historic calving interval and between the reproductive culling rate and the projected calving interval were –0.23 and were not significant. The coefficient of variation associated with the reproductive culling rate (75.7%) was much higher than that of the historical calving interval (3.8%) or the projected calving interval (4.5%). The repeatability of the reproductive culling rate (0.26) was much lower than that of the historical calving interval (0.81) or the projected calving interval (0.60) because culling rate was not only influenced by the overall herd reproductive performance, but also by the differences in the maximal allowable days open for rebreeding and the estimation of which cows were culled for reproductive failure. Additionally, a high coefficient of variation for the reproductive culling rate is unavoidable because this variable has a binomial distribution and a low mean value (7.5%). These disadvantages are not sufficient to exclude this measure from a comprehensive assessment of overall herd reproductive performance. The projected calving interval and the reproductive culling rate can be combined into the adjusted calving interval and thus interpreted jointly. On 40.5% of the farms providing data for this study, fertility assessment based on the projected calving interval indicated better reproductive performance than that based on adjusted calving interval. Hence, the assessment of herd fertility on the basis of projected calving interval can lead to a different and erroneous conclusion regarding the level of herd fertility than that obtained with the evaluation based on adjusted calving interval.
Received May 30, 1997. Accepted March 3, 1998. 1998 J Dairy Sci 81:1848–1854
( Key words: herd reproductive performance, calving interval, culling, evaluation) Abbreviation key: ACI = adjusted calving interval, CPI = interval from calving to pregnancy, HCI = historical calving interval, HRP = herd reproductive performance, NCR = fraction of cows not culled for reproductive failure, PCI = projected calving interval, RCR = reproductive culling rate, VWP = voluntary waiting period. INTRODUCTION Herd reproductive performance ( HRP) (i.e., reproductive efficiency) of dairy herds influences profitability because it affects milk production, reproductive culling rates ( RCR) , and animal sales. The HRP attained within a dairy herd is influenced by the care and attention provided by the manager, herdpersons, inseminators, and others involved with husbandry, herd health, or feeding management. The trend toward larger herds and increased mechanization, resulting in less contact time with caretakers per cow, often results in reduced breeding efficiency if reproductive management is not strengthened. Thus, a regular analysis of the overall HRP is required to assess whether management is satisfactory or whether an investigation into the causes for poor HRP is necessary ( 5 ) . De Kruif ( 1 ) concluded that, particularly in small herds, the values of fertility indices will be determined not only by HRP, but also by chance, and, in that case, correct interpretation of the data on reproduction will only be possible when based on the situation over a number of years. Because of differences in distribution, the value of some reproductive indices will be more affected by random chance than others, which should be taken into account when these indices are being interpreted. The calving interval is the measure most commonly used to assess the overall HRP in dairy herds (4, 8, 12, 21). The reductions in milk production and profit caused by extended calving intervals have been demonstrated frequently (2, 7, 9, 10, 14, 17, 19). Calving interval can be calculated as the duration of the interval between the two most recent calvings for
1848
OVERALL HERD REPRODUCTIVE PERFORMANCE
all parturient cows in a herd. This measure, which should be referred to as the historical calving interval ( HCI) , has several inherent limitations (8, 21). One major deficiency is that first lactation cows are excluded from the measure as they have not had two calvings. Second, cows culled for reproductive failure also do not contribute to the measure. Thus, an apparently acceptable calving interval might misrepresent actual herd performance because infertile cows and cows with poor fertility are often culled from the herd (8, 15, 21). Hence, a high RCR might result in a low average calving interval, even in herds with serious reproductive problems. Therefore, calving intervals should be interpreted in combination with RCR ( 4 ) . The HCI is based on historical events (i.e., past calvings), and does not reflect recent changes in reproductive performance, which is a serious disadvantage because proactive assessment and planning of reproductive management require current information on reproductive performance. Fetrow et al. ( 6 ) , therefore, described the projected calving interval ( PCI) , a measure that is based on the projected minimum average days open plus a standard gestation length. For the estimation of the projected minimum average days open, it was assumed that all bred cows conceived at their last breeding and that cows not yet bred but past the voluntary waiting period will conceive 10 d after the day of calculation. This measure is an optimistic estimate of the minimum average days open that will be experienced by the herd in the future (6, 8, 21). The PCI is considered to be an improvement over the HCI, because PCI is based on a larger proportion of the herd and provides more current information. Accurate pregnancy diagnosis is essential for the use of the PCI. However, because the PCI is a best estimate, the HCI calculated at a later date will most likely be longer. This difference should be taken into account when the data for PCI are being interpreted. Esslemont ( 4 ) introduced the calving interval adjusted calving rate, which is the proportion of a herd that recalves within 12 mo. The advantage of this measure is that it includes the calving interval and the culling rate in one measure. Following the same logic as used for the calving interval adjusted calving rate, Plaizier ( 1 6 ) and Plaizier et al. ( 1 7 ) introduced the adjusted calving interval ( ACI) . The ACI is calculated by adjusting the PCI for the percentage of the cows in the herd that were culled for failure to conceive; the PCI was divided by the ratio of the cows that were not culled for reproductive failure. Hence, ACI is a more comprehensive measure than PCI. A high RCR reduces PCI but increases ACI. Hence, an
1849
assessment of herd fertility on the basis of ACI could result in a different, more accurate conclusion than that based on PCI. Plaizier et al. ( 1 7 ) estimated financial losses from suboptimal reproductive performance of the herd on the basis of PCI and ACI using data generated by computer simulation. Those researchers ( 1 7 ) found that estimation based on ACI was more accurate than estimation based on PCI, as ACI had a higher correlation with net revenue than did PCI. However, to study the usefulness of ACI for the routine evaluation of reproductive performance of commercial dairy herds, it should be determined whether this measure can be calculated accurately with data that are routinely collected by the farmer. The objectives of this study were to study the distributions of measures of herd fertility and correlations between them and to determine whether the evaluation of herd fertility based on ACI would lead to a different result than an evaluation based on PCI. MATERIALS AND METHODS Data Data were obtained from herds participating in the Ontario Dairy Monitoring and Analysis Program ( 1 1 ) and were collected between March 1, 1990 and February 28, 1992. For each farm included in the study, The RCR, calving to pregnancy interval ( CPI) in days, HCI, and voluntary waiting period ( VWP) (i.e., the duration of the period after calving that cows are deliberately not bred) were recorded. The HCI was calculated from the data provided by the Ontario DHI Corporation in December 1990 (yr 1 ) and December 1991 (yr 2). Year 1 included 106 herds, and yr 2 included 97 herds. The mean herd size was 49.1 in yr 1 and 49.7 in yr 2. In both years, the herd sizes ranged between 21 and 146. Differences between the size of individual herds in yr 1 and yr 2 ranged from –7 and 6 cows. The RCR was calculated for the 12-mo period ending in December 1990 and in December 1991 as the percentage of the herd reported by the farmer as failing to conceive. For all cows culled from the herd, farmers were asked to choose the main reason for removal from the following options: sold for dairy, culled for poor reproduction, culled for mastitis, culled for poor production, culled for feet and legs, died, or culled for other reasons. The HCI was calculated as the herd average for the interval between the two most recent calvings. The CPI considered all breeding during the two 12-mo periods of interest and was calculated as the mean duration of the interval between calving and the following Journal of Dairy Science Vol. 81, No. 7, 1998
1850
PLAIZIER ET AL.
TABLE 1. Intervals of projected calving interval (PCI) and adjusted calving interval (ACI) corresponding to herd reproductive performance (HRP) scores. HRP Score
PCI
1 2 3 4 5 6 7
16.0–16.6 15.4–15.9 14.7–15.3 14.1–14.6 13.4–14.0 12.7–13.3 12.0–12.6
ACI (mo) 17.9–18.8 17.0–17.8 16.0–16.9 15.0–15.9 14.0–14.9 13.1–13.9 12.0–13.0
breeding that resulted in a confirmed pregnancy. Cows that did not establish a pregnancy were not included in this measure. The PCI, which was expressed in months, was calculated from the CPI using the following formula: PCI = (CPI + 280)/30.5. The ACI, which was also expressed in months, was calculated by dividing the PCI by the fraction of the cows that were not culled for reproductive failure ( NCR) : ACI = PCI/NCR. The NCR is based on RCR: NCR = (100 – RCR)/100.
Analyses The distributions of all fertility indices were tested for normality using the univariate procedure of SAS (18). If these distributions were not normal, then the data were transformed in an attempt to obtain normal distributions. Transformations included log, square root, inverse, and inverse square root. If transformation resulted in normal distributions, then Pearson regression coefficients were computed. Otherwise, Spearman rank correlation coefficients were computed. Correlation coefficients were calculated among the different measures of HRP for each herd in yr 1 and 2 ( 1 8 ) to estimate their repeatabilities. The ranges between the lowest and the highest PCI and ACI were divided into seven equally spaced intervals. These intervals were used to give herds an HRP score based on their PCI and a HRP score based on their ACI (Table 1). For each farm in each year, the Journal of Dairy Science Vol. 81, No. 7, 1998
HRP score that was based on the ACI was subtracted from the HRP score that was based on the PCI to assess whether evaluation of the overall herd reproductive performance that was based on ACI would lead to different conclusions than that based on PCI. RESULTS AND DISCUSSION Fertility indices did not have normal distributions in both years. Normal distributions were not achieved by the transformations used. Hence, Spearman rank correlation coefficients were computed between fertility indices (18, 20). The summary statistics for reproductive performance (Table 2 ) indicated that the herds included in the survey had a wide range in HRP, which was similar to that observed in comparable surveys (3, 13). The mean HCI and mean PCI were similar to the 13.1 mo average for HCI for Ontario dairy herds (13). The mean HCI and mean PCI were nearly equal, which was not expected because PCI was the best estimate of the future average calving interval. Hence, PCI was expected to be lower than HCI. The RCR varied from 0 to 30% among herds, which resulted in a high coefficient of variation for RCR compared with those of HCI and PCI. The wide ranges in HCI, PCI, and RCR are explained by differences in HRP among farms, by differences in the maximum allowable days open for breeding, and by the estimation of which cows were culled for failure to conceive. An additional factor that contributed to the high coefficient of variation of RCR between farms was that RCR is a binomial variable, and its average is low. Hence, RCR has a high coefficient of variability within each farm (20), which was demonstrated by the correlation coefficient between RCR in yr 1 and RCR in yr 2 (i.e., the repeatability of RCR) compared with the repeatabilities of HCI and PCI (Table 3). The repeatability of HCI was higher than that of PCI. This result can be attributed to the differences in the calculation of these intervals. The HCI is based on the
TABLE 2. Summary statistics for historical calving interval (HCI), projected calving interval (PCI), reproductive culling rate (RCR), adjusted calving interval (ACI), and voluntary waiting period (VWP). Measure
X
SD
Minimum
Maximum
CV
HCI, mo PCI, mo RCR, % ACI, mo VWP, d
13.2 13.1 7.5 14.2 54.3
0.6 0.6 5.2 1.1 8.8
11.9 11.9 0 12.0 40
15.9 16.6 30 18.9 90
(%) 4.5 4.6 69.3 7.7 16.2
OVERALL HERD REPRODUCTIVE PERFORMANCE TABLE 3. Spearman rank correlation reproductive indices in yr 1 and 2.
coefficients
1851
between
Measure1
Correlation coefficient
P
HCI RCR PCI ACI
0.81 0.26 0.60 0.29
0.001 0.01 0.001 0.05
1HCI = Historical calving interval, RCR = reproductive culling rate, PCI = projected calving interval, and ACI = adjusted calving interval.
occurrence of two subsequent calvings that occurred, while PCI is based on one calving and the best estimate for the subsequent calving. The uncertainty of when the next conception will occur, combined with abortions, errors in pregnancy diagnosis, and culling of pregnant cows results in higher variability and lower repeatability of PCI than of HCI. Scatter plots of RCR versus HCI and RCR versus PCI appear in Figures 1 and 2, respectively. Within short ranges of PCI and HCI, RCR varied widely, which was also demonstrated by the correlation coefficients between RCR and HCI and between RCR and PCI (Table 4). This coefficient was low and negative for RCR and HCI, indicating that high values of HCI were associated with low values of RCR. The correlation between RCR and PCI was not
Figure 1. Relationship between historical calving interval (HCI) and culling rate for reproductive failure (RCR).
Figure 2. Relationship between projected calving interval (PCI) and reproductive culling rate (RCR).
significant ( P = 0.67). Thus, herds with similar PCI and HCI could vary widely in RCR, which illustrates that differences in the overall reproductive performance between farms can be expressed as differences in PCI and as differences in RCR. Herds with similar rates of estrus detection and conception can vary in PCI and RCR because of differences in the duration of the rebreeding period (the maximal allowable number of days open or number of inseminations for breeding) and the VWP. In herds with a short period before rebreeding, when cows were not bred after two or three inseminations or a limited number of days open (e.g., <200 d), calving intervals were short even when estrus detection rates or conception rates were poor. In these herds, the variation in reproductive performance was expressed as variation in RCR. If breeding continued after many inseminations and many days open, then variation in estrus detection and conception rates was expressed as the variation in calving intervals and not in RCR. Thus, if reproductive performance is to be assessed, both calving interval and RCR should be considered by evaluating PCI and RCR separately or by evaluating the ACI. The advantage of using the ACI is that one measure can be used for a comprehensive assessment. Financial losses caused by suboptimal herd fertility can be estimated more accurately on the basis of ACI than on the basis of PCI (17). The correlation coefficient between HCI and ACI was only 0.29 (Table 4). This result is not surprising Journal of Dairy Science Vol. 81, No. 7, 1998
1852
PLAIZIER ET AL. TABLE 4. Spearman rank correlation coefficients between historical calving interval (HCI), reproductive culling rate (RCR), projected calving interval (PCI), adjusted calving interval (ACI), and voluntary waiting period (VWP). HCI HCI RCR PCI ACI VWP 1Level
1
RCR –0.23 1
(0.001) 1
PCI
ACI
VWP
0.55 (0.0001) –0.03 (0.67) 1
0.19 (0.006) 0.73 (0.0001) 0.59 (0.0001) 1
0.29 –0.15 0.25 0.07 1
(0.0001) (0.03) (0.0004) (0.30)
of significance.
because these measures encompass different time periods and the correlation between HCI and PCI and between HCI and RCR (the components of ACI) were positive and negative, respectively. Hence, the correlation between ACI (i.e., the combined effect of PCI and RCR) and HCI was small. The correlation between RCR and ACI was higher than between PCI and ACI, indicating that more variation in ACI is explained by RCR than by PCI. This difference was expected because the coefficient of variation of RCR was much higher than that of PCI. A positive correlation was observed between VWP and HCI and between VWP and PCI, which indicated that part of the variation in HCI and PCI was explained by VWP. Therefore, for a comprehensive assessment of reproductive performance using PCI or HCI, the VWP should be considered. In contrast, the correlation be-
tween ACI and VWP was not significant in both years, which partially was caused by the higher variation of ACI than of HCI and PCI. This absence of a significant correlation does not imply that VWP should not be considered for the interpretation of ACI. The frequency distributions of the HRP scores based on PCI and ACI (Figure 3 ) show that these distributions were skewed. Because the duration of a calving interval has a biological minimum but not a biological maximum, distributions were expected to be skewed. The frequencies of scores 5 and 6 were highest. Figure 4 shows the distribution of the HRP score based on PCI minus the HRP score based on ACI. If this difference is 0, then PCI and ACI indicated the same level of HRP. A positive value resulted when PCI indicated a higher level of HRP than ACI, and values were negative when ACI indicated a higher level of HRP. For about one-half of the farms
Figure 3. Frequency distributions of herd fertility scores based on projected calving interval (PCI) and adjusted calving interval (ACI).
Figure 4. Difference between herd fertility score based on projected calving interval (PCI) and herd fertility score based on adjusted calving interval (ACI).
Journal of Dairy Science Vol. 81, No. 7, 1998
1853
OVERALL HERD REPRODUCTIVE PERFORMANCE
(44.3%), PCI indicated the same level of reproductive performance as did ACI (Figure 4). However, for 46.3% of all evaluations, PCI indicated better reproductive performance than did ACI. For 14.3% of all evaluations, the difference between the PCI HRP score and the ACI HRP score was ≥2. In 3 of the 26 herds that had this difference, the data from both yr 1 and 2 were affected. Differences between the PCI HRP score and the ACI HRP score are explained by differences in RCR. A positive value for this difference was caused by a high RCR. For 9.4% of all evaluations, the value for this difference was negative, which was caused by a low RCR, ranging from 0 to 4.6%, compared with that for other herds. The large difference between the PCI HRP score and the ACI HRP score in yr 1 and yr 2 on only a few farms indicates that RCR can vary widely between years. As mentioned earlier, this wide variation is attributed to differences in HRP and reproductive culling policy between years but also to the distribution of RCR. The evaluation of herd fertility based on PCI could, therefore, lead to an assessment that is different from that based on ACI. The assessment using ACI might be more accurate than that using PCI because of the consideration of cows that were culled for failure to conceive. The ACI can be calculated using routinely collected reproductive data. However, as discussed earlier, the accuracy of RCR might differ among farms, and different interpretations of how RCR should be calculated may exist. Many cows are not culled for one specific reason, but have primary, secondary, and tertiary reasons for removal (5, 9), and failure to conceive might not be the only reason for culling. Cows with low milk production or with disease or other problems might be at increased risk of being culled for reproductive failure. Farmers might not continue breeding problem animals as long as other normal herdmates exist, thus obscuring differences between reasons for removal. Because of the need to include the reproductive culls in the evaluation of reproductive performance, this aspect should be addressed through industry extension programs. A definition of RCR that does not allow differences in interpretation should be adopted (22). The following calculation is proposed: RCR = total number cows culled for reproductive failure/total number cows available for rebreeding. The total number of cows culled for reproductive failure includes those cows that were rebred at least once but failed to conceive plus any that the operator intended to rebreed but did not. The total number of cows available for rebreeding is the number of cows that could have been rebred, including all of the cows
TABLE 5. Adjusted calving interval (ACI) corresponding to values for projected calving interval (PCI) and ratio of cows not culled for reproductive failure (NCR) and interpretation of ACI.1 PCI NCR
12
12.5
13
13.5
14
0.95
12.6 (E) 13.3 (G) 14.1 (CI) 15.0 (SI) 16.0 (SP)
13.2 (G) 13.9 (G) 14.7 (CI) 15.6 (SP) 16.7 (SP)
13.7 (G) 14.4 (CI) 15.3 (SI) 16.3 (SP) 17.3 (SP)
14.2 (CI) 15.0 (SI) 15.9 (SI) 16.9 (SP) 18.0 (SP)
14.7 (CI) 15.6 (SI) 16.5 (SP) 17.5 (SP) 18.7 (SP)
0.9 0.85 0.8 0.75
1E = Excellent, G = good, CI = could improve, SI = should improve, and SP = serious problem.
that eventually became pregnant plus all of those culled for reproductive failure. This value corresponds to the total number of cows in the herd minus any voluntary culls that the operator never intended to rebreed and involuntary culls that were removed for reasons other than reproductive culls. The RCR is calculated for the preceding 12 mo. Some general guidelines for the interpretation of ACI are given in Table 5. Assessment of ACI will indicate if reproductive performance needs to be improved and if causes for suboptimal herd fertility need to be investigated. However, ACI cannot identify the causes underlying poor fertility. Problem identification and resolution involve a holistic evaluation of the entire production system including past records, clinical examinations, and management practices. CONCLUSIONS Differences in overall reproductive performance among herds can be expressed as differences in calving interval (i.e., PCI or HCI) and as differences in RCR. The correlation between these measures is low. Hence, for a comprehensive evaluation of overall HRP, both a measure of calving interval and RCR need to be considered. Disadvantages of RCR are that by nature of its distribution it has a high coefficient of variation and low repeatability and that herd operators vary in their perception of which cows were culled for reproductive failure. Hence, the adoption of a uniform definition of RCR is essential. These disadvantages do not provide sufficient reason to ignore RCR when overall HRP is being evaluated. Adjusted calving interval, which is based on PCI and RCR, provides a more accurate assessment of overall HRP than does PCI alone. As shown by the differences Journal of Dairy Science Vol. 81, No. 7, 1998
1854
PLAIZIER ET AL.
between PCI HRP score and ACI HRP score, evaluation of HRP on the basis of PCI and this evaluation on the basis of ACI can lead to different conclusions. REFERENCES 1 De Kruif, A. 1978. Factors influencing the fertility of a cattle population. J. Reprod. Fertil. 54:507–518. 2 Dijkhuizen, A. A., J. Stelwagen, and J. A. Renkema. 1985. Economic aspects of reproductive failure. Financial loss at farm level. Prev. Vet. Med. 3:251–263. 3 Dohoo, I. R., S. W Martin, A. H. Meek, and W.C.D. Sandals. 1983. Disease, production and culling in Holstein-Friesian cows. I. The data. Prev. Vet. Med. 1:321–334. 4 Esslemont, R. J. 1992. Measuring dairy herd fertility. Vet. Rec. 131:209–212. 5 Etherington, W. G., J. Fetrow, B. E. Seguin, W. E. Marsh, L. D. Weaver, and C. L. Rawson. 1991. Dairy herd reproductive health management: evaluating dairy herd reproductive performance-part II. Comp. Contin. Educ. Prac. Vet. 13: 1491–1502. 6 Fetrow, J., D. McClary, R. Harman, K. Butcher L. Weaver, L. E. Studer, J. Ehrlich, W. Etherington, W. Guterbock, D. Klingborg, J. Reneau, and N. Williamson. 1990. Calculating selected reproductive indices: recommendations of the American Association of Bovine Practitioners. J. Dairy Sci. 73:78–90. 7 Funk, D. A., A. E. Freeman, and P. J. Berger. 1987. Effects of previous days open, previous days dry and present days open on lactation yield. J. Dairy Sci. 70:2366–2373. 8 Gaines, J. D. 1989. The role of record analysis in evaluating subfertile dairy herds. Vet. Med. 84:532–543. 9 Holmann, F. J., C. R. Shumway, R. W. Blake, R. B. Schwart, and E. M. Sudweeks. 1984. Economic value of days open for Holstein cows of alternative milk yields. J. Dairy Sci. 67: 636–643.
Journal of Dairy Science Vol. 81, No. 7, 1998
10 James, A. P., and R. J. Esslemont. 1979. The economics of calving intervals. Anim. Prod. 29:157–162. 11 Kelton, D. 1995. Monitoring and investigating the relationship among health, management, productivity and profitability on Ontario dairy farms. Ph.D. Diss., Univ. Guelph, Guelph, ON, Canada. 12 King, G. J. 1993. Reproductive performance and problems. Pages 531–565 in Reproduction in Domesticated Animals. Vol. B 9. Elsevier Sci. Publ. B.V., Amsterdam, The Netherlands. 13 Lissemore, K. D., K. E. Leslie, P. I. Menzies, S. W. Martin, A. H. Meek, and W. E. Etherington. 1992. Implementation and use of a microcomputer-based management information system to monitos dairy herd performance. Can. Vet. J. 33:114–120. 14 Olds, D., T. Cooper, and F. A. Thrift. 1979. Relationship between milk yield and fertility in dairy cattle. J. Dairy Sci. 62: 1140–1141. 15 Peters, A. R., and P.J.H. Ball. 1995. Reproduction in Cattle. 2nd ed. Butterworth & Co. Publ., London, England. 16 Plaizier, J.C.B. 1996. A study on the relationship between economic and reproductive efficiency in Ontario dairy herds using computer simulation. Ph.D. Diss., Univ. Guelph, Guelph, ON, Canada. 17 Plaizier, J.C.B., G. J. King, J.C.M. Dekkers, and K. Lissemore. 1997. Estimation of economic values of indices for reproductive performance in dairy herds using computer simulation. J. Dairy Sci. 80:2775–2783. 18 SAS® User’s Guide, Version 6 Edition. 1990. SAS Inst., Inc., Cary, NC. 19 Schmidt, G. H. 1989. Effect of length of calving intervals on income over feed and variable costs. J. Dairy Sci. 72:1605–1611. 20 Snedecor, G. W., and W. G. Cochran. 1978. Statistical methods. 7th ed. Iowa State Univ. Press, Ames. 21 Upham, G. L. 1991. Measuring dairy herd reproductive performance. Bovine Pract. 26:49–56. 22 Weaver, L. D. 1992. Reproductive health programs. Pages 99–109 in Large Dairy Herd Management. H. H. Van Horn and C. J. Wilcox, ed. Am. Dairy Sci. Assoc., Champaign, IL.