Correlations between first and successive records on four traits as a basis for culling dual purpose cattle

In order to determine whether first records could be used as a basis for culling, records on milk yield, days in milk, days open and 4-month calf weight from 882 cows in 12 dual purpose herds were to estimate the correlation between first and successive records up to the fourth. All herds were kept on grazing with variable proportions of cultivated pastures and supplementary feeding. Ten of the herds were milked by hand with calf suckling.

Two data sets were used: the first consisted of unadjusted (UA) performance data to simulate the simplest selection method available to farmers. The second analysis was based on the deviations of each cow's records (A) from contemporary means based on a minimum of ten records, with herd, calving year, season and number being taken into account. The correlations between first and successive records ranged from 0.59 to 0.72 (UA) and from O.24 to 0.47 (A) for milk, but were in the range 0 to 0.27 (UA) or 0.23 (A) for the other three traits. It was concluded that first, unadjusted records could be used to cull poor performers for milk yield. More records were desirable in the case of the other traits but, nevertheless, economically reasonable limits should be established on each farm for first calf cows with long intervals from calving to conception, beyond which their retention in the herd should not be acceptable.

Dual purpose cattle produce over 90% of the milk marketed in Venezuela and their potential for the development of sustainable cattle production systems throughout the Latin American tropics is now well recognised. Earlier research drew attention to the extremely wide variation in levels of performance between individual cows in commercial herds and, consequently, to the importance of culling unproductive animals (Vaccaro et al 1992). The present study arose in response to a request from the farmers who cooperated in the earlier work, to determine whether culling for the traits of economic interest might be carried out on the basis of first records.

Data were obtained from 12 dual purpose herds in the Venezuelan lowlands which cooperate in the Dual Purpose Cattle Research Project run from the Central Univeristy of Venezuela. All farms use grazing systems but they differ widely with respect to the intensity of production, as reflected in the mean performance statistics shown in Table 1. They are situated at less than 350 m above sea level with an average annual temperature of about 26 ºC. The length of the dry season varies from three to seven months/year. Seven farms use high proportions of native grasses, with or without maize and sorghum crop residues, little or no feed supplementation, and milking is carried out once daily by hand with the calf at foot. The other five farms use high proportions of cultivated pastures, some supplementary feeding and milk twice daily with the calf in three cases and without the calf in the rest. Calves are used to stimulate milk let down and are reared by restricted suckling in ten herds, and artificially in the other two. Cows are mainly crossbreds with varying degrees of zebu, criollo and European dairy breed inheritance. Based on their external appearance 18% were classified as zebu-type, 44% as approximately half zebu-half European breed and 38% as higher grade European crossbreds.

Table 1: Variation between farms and overall levels of performance in the herds studied. Adjusted mean values, according to farm*

Trait	Min	Max	Overall (±SE)

Milk yield (kg/lactation)	556	2972	1396 (38)
Days in milk	153	286	225 (5)
Days open	98	189	139 (6)
Calf weight (kg)**	61	87	71 (1)
Calf mortality (% 0-4 months)	1.7	26.1	8.9 (1.5)

* Least squares adjusted means based on 201 to 1071 observations per farm (Vaccaro et al 1995)
** Corrected for sex and age at 4 months.

Records on four traits were used in the study: milk yield per lactation, days in milk, days open and calf weight at four months of age. Milk yields were estimated from weekly or monthly samples and data included all lactations, irrespective of length. Days open were calculated from calving intervals where these were complete, or otherwise from expected calving dates based on pregnancy diagnoses carried out by experts. Calf weights were available from the 10 herds which used restricted suckling. They were corrected for sex and age at four months, with zero assigned to those cows whose calves died.

The data base consisted of records of cows which had completed first and at least second lactations under the supervision of the Project in the period 1990-1994. The cows with two, three and four records available for each of the traits ranged from 882 to 105 as shown in Table 2. Numbers of cows with five or more lactations were too small to justify inclusion.

Data were analysed using intra-class correlations between first and subsequent records up to the fourth. Since performance data processing is poorly developed throughout the region, a first analysis was based on unadjusted records such as farmers would generally have to hand. In the second analysis, the correlations were based on the deviations of each cow's record from contemporary means, making use of data calculated routinely by the Project in order to estimate the breeding values of cows in the cooperator herds. Contemporary means were based on a minimum of 10 cows' records and took into account herd (12), calving year (1990-1994), calving season (wet, dry), and calving number (first, later).

Table 2 shows the correlations obtained using the unadjusted records. The values for milk yield remained moderately high up to the fourth lactation, with variation in first lactation performance accounting for between 35% and 52% of variation in subsequent ones. The coefficients for the other traits were much lower, but for days in milk and calf weight, the correlations between first and later records remained significant and positive up to the third lactation. For days open, only the correlation between first and second records was significant, with a relatively low value (0.27).

Table 2: Correlations between first and subsequent records, using unadjusted data

	Calving number
Trait	1 vs 2	1 vs 3	1 vs 4

Milk yield	.72 **	.65 **	.59 **
(n)	(882)	(404)	(148)
Days in milk	.27 **	.18 **	-.06 NS
(n)	(882)	(404)	(148)
Days open	.27 **	.02 NS	-.01 NS
(n)	(722)	(333)	(125)
Calf weight	.20 **	.21 **	.01 NS
(n)	(549)	(268)	(105)

The correlations based on deviations from contemporary means are shown in Table 3. They follow the same pattern as the actual records, although the values are generally lower. In the case of milk yield, significant positive coefficients were found between deviations obtained in the first lactation and all subsequent lactations up to the fourth, with values of between 0.47 and 0.24. For the rest of the traits, the correlations were lower and ceased to be significant after the second (days open, calf weight) or third (days in milk) lactations.

Table 3: Correlations between first and subsequent records, using deviations from contemporary means*

	Calving number
Trait	1 vs 2	1 vs 3	1 vs 4

Milk yield	.47 **	.38 **	.24 **
Days in milk	.23 **	.17 **	.02 NS
Days open	.18 **	.06 NS	.05 NS
Calf weight	.18 **	.12 NS	.07 NS

The two data sets used in this study were chosen to simulate, on the one hand, the simplest procedure which farmers with performance records could use themselves and, on the other, a theoretically more accurate but also more complex procedure which would generally require technical assistance from outside the farm.

The high correlations for milk yield using unadjusted records (Table 2) is surprising in view of the heterogeneity of the farms and environmental variation to which the cows are subjected, particularly with regard to feed supply and the hand milking procedure with, in most cases, calf suckling. The values indicate that actual first lactation yields may be used with reasonable safety to identify cows which will remain poor performers during most of their productive lives. Thus, the simplest procedure available at farm level should serve as an acceptable basis for culling for milk yield. The results from Table 3 show that the comparative merit for milk yield shown in the first lactation, is also maintained moderately well up to the fourth calving. The difference between the values in Tables 2 and 3 suggests that temporary environmental effects played an important part in explaining the correlations obtained with the unadjusted data. Even though this implies that breeding values are considerably masked by environmental effects, as would be expected in theory, it does not invalidate the decision to cull the worst producers on the basis of their unadjusted first records, because there is a strong probability that they will remain an economic liability during their own lifetimes.

Due to the strong association of lactation length with milk yield (Velásquez 1993) and the ease with which it can be measured, "days in milk" has been considered as an alternative selection criterion where yield cannot be recorded, and "calf weight" is commonly used as an auxiliary selection criterion in the herds used for this study. However, the results in Tables 2 and 3 show that culling for these traits on the basis of first records would not be justifiable, because the correlations are low even though mainly positive. Variation in first records was associated with 7%, or less, of the variation in successive records, whether unadjusted or adjusted data were used.

The same conclusion applies to "days open", where the low correlations in both Tables show that the first interval from calving to conception is poorly related to successive ones. This raises an important practical problem because the variation between cows within herds for "days open" is even wider than for other traits, extremely poor performers are found in all herds and first calf cows have average intervals of 43 days longer than the overall mean (Vaccaro et al 1992, 1995). Given the low correlation between first and subsequent records obtained in this study, it would seem logical to retain cows with long first calving-conception intervals in the herd for another opportunity. There is, in addition, evidence that these cows may include the best milk producers and since the negative correlation between fertility and milk yield appears to have a genetic base, there is a clear need to combine information on both traits before taking selection decisions (Vaccaro et al 1994). Nevertheless, economically reasonable limits for "days open" should be set according to rearing and maintenance costs on each farm, and animals which surpass these limits should leave the herd. Cows which are extremely poor in milk as well as fertility, on the basis of first records, are also obvious candidates for culling.

The results in Table 3 are more closely comparable than those in Table 2 with repeatability estimates as normally defined in the genetic sense. With the main temporary environmental effects reduced, the correlations in Table 3 estimate the similarity between successive records due to genetic and permanent environmental causes. Information on repeatability from mixed crossbred populations in the Latin American tropics is scarce, but the values obtained in this study fall within the ranges reported for milk yield (0.12 to 0.63), lactation length (0.14 to 0.46), and days open or calving interval (0.01 to 0.19) in Gir and Gir crossbreds (Souza et al 1995ab), Pitanguieiras (Lobo and Reis 1989), high European breed grades (Freitas et al 1988; Barbosa et al 1994ab) and different breeds of criollo (Bodisco and Abreu 1981; Fraga et al 1994).

The values obtained in Table 3 may be used as an approximate guide to the upper limits for heritability which have not been quantified precisely for the mixed crossbred populations which are the basis of most dual purpose enterprises in tropical Latin America. The results are compatible with heritability estimates obtained from large samples of specific types of crossbred populations such as the Cuban Siboney and Mambi (López and Ribas 1993) and the Brazilian Pitangueiras (Lobo and Reis 1989). Although the values obtained suggest that the response to selection may be slow in the case of days open and calf weight, the results do not disprove the need to take these traits into account with milk yield in selection programs for tropical dual purpose cattle. They do suggest, however, that first records are an unreliable guide to breeding values, except perhaps for milk yield. In the case of milk, the fact that farmers can use unadjusted first records to cull cows which are extremely poor for this trait is of considerable practical importance. The results also indicate that when cows, such as bull dams, which have an important genetic effect on the herd are being selected, various records rather than single ones should be used because multiple traits are involved. Thus, breeding values for characteristics such as fertility, which are of vital economic importance but low heritability, can be estimated with greater precision. Further research is needed in this area to define optimum selection procedures more clearly.

This research was carried out thanks to the economic support of the International Development Research Center (Canada). Great appreciation is also expressed to the farmers whose cooperation made the study possible.

Velásquez E 1993 Duración de la lactancia y factores que la afectan en vacas de doble propósito. Ingeniero Agrónomo Thesis, Facultad de Agronomía, Universidad Central de Venezuela, Maracay 42p