Environmental and Lactational Variables Affecting Prolactin Concentrations in Bovine Milk 1 J. P. McMURTRY, 2 P. V. MALVEN, C. W. ARAVE,8 R. E. ERB, and R. B. HARRINGTON Department of Animal Sciences Purdue University West Lafayette, IN 47907

higher milk prolactin. Individual cows tended to have characteristic prolactin, but this tendency was eliminated by statistical adjustment of the data for environmental and l~ctational variables.

Abstract

Relationships were examined between prolactin concentrations in bovine milk and various environmental and lactational variables. Prolactin was quantified by radioimmunoassay in 1,316 milk samples from two experiments. Environmental temperatures preceding milking, stage of lactation, daily milk yield, and dominance rank of the cow were correlated significantly with milk prolactin concentration. Stepwise multiple regression analysis was used to determine, in order of importance, the variables having significant independent effects on milk prolactin. Ambient temperature extremes, high and low, exerted the greatest effects and were each associated with elevated concentrations of prolaetin. In Experiment I, c o n d u c t e d throughout the year, the variable representing maximum temperature preceding milking accounted for 14.3% of the variation in milk prolactin. In Experiment II, conducted during late fall and early winter, the minimum temperature preceding milking accounted for 21.1% of the variation. Although earlier stages of lactation and larger daily milk yields were associated with higher prolactin concentrations, inclusion of lactation stage in the stepwise regression model tended to eliminate almost all the variance of prolactin previously associated with daffy yield. Dominance measurements indicated that more submissive cows had

Introduction

Previous research has demonstrated by radioimmunoassay the pituitary hormone prolac-

Received July 8, 1974. 1Journal Paper No. 5576, Purdue University Agricultural Experiment Station. Supported in part by a grant GB-43473 from the National Science Foundation. mAddress: Department of Biochemistry and Biophysics, University of Hawaii, Honolulu 96822. 8Address: Department of Dairy Science and Industry, Utah State University, Logan 84321.

tin (PRL) in milk from cattle, sheep, goats, and rats (10, 11, 12). While radioimmunoassay procedures allowed accurate quantification of immunoreactive PRL in small volumes of milk, they could not resolve whether milk PRL was still in the biologically active form of the pituitary. However, a series of papers published by Iwamura in Japan helped to resolve this question. The initial observations, published in 1943, were that extracts of bovine colostrum contained prolaetin-like biological activity (5). Subsequent research with a different bioassay substantiated prolaetin-like activity (6). Amino acid analyses of the prolaetin-like protein were used to estimate a molecular weight similar to that for pituitary prolaetin (7). In addition to demonstrating PRL in milk, the previous studies with radioimmunoassays showed that concentrations of milk PRL could be made to increase or decrease in direct proportion to experimentally induced changes in blood plasma concentrations of PRL (11, 12). By measuring milk PRL we should be able to estimate plasma PRL averaged over time. This would be more stable than the widely fluctuating plasma PRL concentrations. Furthermore, the latency required for transfer of plasma PRL into milk (11, 12) should insure that removal of milk does not affect PRL in the sample. This is another advantage of milk sampiing over blood sampling for PRL measurement since PRL increases induced by stress are less rapid in milk. Even though studies of plasma PRL in lac-

181

182

McMURTRY E T AL

tating cows were characterized by large variation, variables which had significant effects on plasma PRL included parturition, stage of lactation, milk yield, parity, and season (8, 9, 16, 17). We examined the influence of these and other variables on PRL concentrations in bovine milk. Materials and Methods

Procedures for Experiment I (Exp. I). Monthly samples of milk fi-om all cows in the Purdue University dairy herd were obtained 10 times from July, 1972, to May, 1973. Milk was collected at one afternoon milking (1400 h) and one morning milking (0200 h) and pooled to form for each cow a composite monthly sample for analysis. Other milk constituents and PRL were measured on 1,078 milk samples from 135 Holstein-Friesian cows each sampled at least twice. Parturition had occurred in all cows 5 to 350 days prior to sampling. All cows were managed under a system of loose housing in open unheated buildhags.

Procedures for Experiment 11 (Exp. H). Thirty-four cows from the herd were paired according to sire, age, stage of lactation, size, and milk production. One member of each pair was assigned randomly to one of two groups within which their social rank would be determined. The experiment began in late October of 1972 and lasted for 10 wk. At intervals of 1 to 3 wk, samples of milk were collected at the afternoon milking (1400 h) for PRL analysis. Three or more times per week behavior of cows in both groups was observed for 1 h at feeding time. A behavioral encounter was won if the loser yielded space by either forceful ejection or movement to avoid contact. At the end of each 2-wk period the cows were ranked for dominance from 1 to 17 within each group. A dominance rank (DR) of 1 was given to the most dominant cow. At the end of each 2-wk period selected cows were exchanged between the two social groups. For example, the most dominant cow of one group might be exchanged with the most submissive cow of the other group. The effects of exchange of cows between groups on milk yield and other variables have been reported (1). No effect of these exchanges on milk PRL was detectable. Analyses of milk samples. All PRL analyses were on milk samples which had been stored at 4 C less than 4 days. Two 50 F1 aliquots from each sample of whole milk were radioimmunoassayed for PRL by procedures preJOURNAL OF DAIRY SCXENC~ VOL. 58, NO. 2

viously described (10). The range of PRL measurements in this assay was from 2 to 160 ng/ml. Some samples in which PRL concentration appeared to exceed 160 ng/ral were arbitrarily assigned this value. Analyses of milk composition included determinations of fat (Milko-tester, Foss Electric), protein (19), and solids-not-fat (SNF) (4). In addition, the concentration of somatic ceils in each sample was measured (15).

Measurement o~ environmental temperatures. Climatic records for days on which milk had beert sampled were obtained from the Purdue University Climatology Laboratory. Environmental temperatures were collected at a site 8 km from the buildings housing the cows. Because each monthly sample in Exp. I was a composite from two consecutive milkings, environmental temperatures during the 9 h prior to the first milking (0400 to 1300 h) and during the 9 h between the first and second milkings (1600 to 0100 h) were examined. The maximum and minimum environmental temperatures during each period were used in the statistical analysis and were designated Pre-Hi, Pre-Lo, Interim-Hi, and InterimLo. Because the samples in Exp. II were obtained at the afternoon milking, environmental temperatures were examined only for 0400 to 1300 h. These maximum and minimum temperatures were designated Hi-Temp and LoTemp. In Exp. I the environmental temperatures prior to first milking ranged from the lowest Pre-Lo value of --5 C (1/3/73) to the highest Pre-Hi of 28 C (9/7/72). The Pre values, representing 0400 to 1300 h, covered a wider range of environmental temperatures than Interim values. Because Exp. II was begun late in the fall, environmental temperatures were much lower, ranging from a high of 16 C (11/2/72) to a low of --15 C (1/12/73) for 0400 to 1300 h. Lactation variables. In Exp. I projected 305 day yield and total lactation yield to date were obtained from Dairy Herd Improvement Association records provided by the Dairy Record Processing Center, Raleigh, NC. The compilation of other pertinent lactation and reproductive information associated with each milk sample was aided by the computerized dairy records program (3). The reproductive status of each cow at the time of sampling was included in the statistical analysis. Each cow was assigned one of the following numbers: 0, nonpregnant; 1, pregnant less than 91 days; 2, pregnant from 91 to 180 days; and 3, pregnant more than 180 days.

183

MILK PROLACTIN CONCENTRATIONS

Data analysis. Because of significant simple correlation coefficients between concentration of PRL in milk samples and many environmental and lactational variables, multiple regression analysis was used to determine the independent relationship between each independent variable and the dependent variable milk PRL. Stepwise regression analysis (BMD2R, UCLA Biomedical Library) using the buildup approach was the primary method, but weighted regression analysis using the "tearing down" approach was also used. Since results with each approach were almost identical, we present those from the buildup approach. Regression analyses are summarized in Resuits. In preliminary analyses the first, second, and third powers of each variable were included in the regression model. However, the final analysis included second and third powers only if" they had made significant independent contributions. Stepwise multiple regression allowed estimation of relationships between milk PRL and each variable independently of all other variables in the regression equation for that step of the analysis. Therefore, it was possible to predict and graphically to illustrate the relationship between milk PRL and one independent variable, from its independent regression coefficient while other variables in the regression equation were constant at their mean value. This method illustrates PRL responses to changes in one independent variable, but predictions would not be realized under actual conditions because other highly correlated variables would not remain constant at their mean. This did not negate the use of independent regression coefficients for interpretation of physiological relationships especially when the predicted regression line was not extended beyond the range of actual values

for the independent variable. Results and Discussion

Experiment L Simple correlation coetIicients among milk PRL and selected variables are in Table 1. Coefficients for daily yield and stage of lactation with PRL were of like magnitude but opposite sign. Coefficients between PRL and the three environmental temperatures were positive indicating association between warm weather and high milk PRL. Stage of lactation also was correlated positively with environmental temperatures because a majority of the cows calved during fall and winter when low temperatures prevailed. The expected high correlations among the three variables of temperature and between daily yield and lactation stage are also in Table 1. While correlations in Table 1 suggested overall relationships, this procedure was inadequate to analyze the many variables which might affect milk PRL, especially for significant correlations among these variables. Multiple regression analysis did provide an opportunity to quantify the independent influence of each variable in the stepwise model (Table 2). The first and most important variable included in the regression equation was the third power of Pre-Hi (Pre-Hia). The coefficient of determination (R 2) for this equation was .•43 indicating that 14.3% of the variation in milk PRL could be accounted for by variations in Pre-Hi 3. Whereas the correlation between PreLo and milk PRL was .19 (Table 1), their partial correlation coefficient after including Pre-Hi 3 in the equation was --.27. This negative coefficient explains why in step 2 of the build-up model the first power of Pre-Lo (PreLo 1) entered the regression equation with a negative regression coefficient. To illustrate relationships between milk PRL and both Pre-

TABLE 1. Simple correlation coefficients among variables in Experiment I (n ~ 1,078). Daily yield Milk prolactin Daily yield Stage of lactation Pre-Hi Pre-Lo

.09d

Stage of lactation --.09 --.58

Pe -Hi" .36 --.13 .29

Pre -Lob

Interim -I-~~

.19 --.12 .29 .88

.32 --.15 .21 .90 .71

Maximum environmental temperature during 9 h before first milking (0400 to 1300 h). b Minimum environmental temperature during 9 h before first milking. Minimum environmental temperature between first and second milkings ( 1600 to 0100 h). Any coefficient > .07 is significant (P < .01 ). JOURNAL OF DAIRY SCIENCE VOL, 58. NO. 2

184

McMURTRY E T AL

T~LE 2. Summary of stepwise multiple regression analysis for Experiment I with milk prolactin as dependent variable.

Variable Pre-Hib Linear Square Cube pre-Lo~ Linear Square Stage of lactation Linear Square Cube

Order of inclusion and magnitude of R= increase (R=~")~ R= 1" more R2 ~" more R2 1' less than .02 than .005 than .005

12.7 C i ¢ (.143) a

5.2 C

2 (.205)

13 137 days 17

3 (.239)

lo (.332) 6.9 C

Interim-Lo t Linear

Square Cube Iatedrn-Hi~ Linear Fat SNF Parity number Reproductive status Ago Pro~ected 305 day production Daffy yield Total yield to date Protein Somatic cell count

Mean for variable

5

7 (.306)

4 (.270)

12.6 C 11 12 (.341) 14 15 16 18

3.8g 8.7~ 2.23 .71 3.7 yr 6,976 kg

19

21 kg

20

3,107 kg

21 22 (.348)

3.4~ 51X10~lm]

"Variables were classified on the basis of the R2 increase resulting from their stepwise inclusion in the model. b Maximum environmental temperature before first milking (0400 to 1300 h). Highest Pre-Hi was 28C. Number denoting the step in the buildup model at which this variable entered the regression equation. a Parentheses contain the actual R' after inclusion of this variable. In order to calculate the R= increase, subtract the actual R= at the previous step from the Ra in parenthesis. * Minimum environmental temperature before first milking. Lowest Pre-Lo was --5 C. t Minimum environmental temperature between first and second mflkings ( 1600 to 0100 h). g Maximum environmental temperature between first and second milkings.

Hi s and Pre-Lo ~, their independent effects as predicted from the multiple regression equation were plotted in the upper part of Fig. I. In plotting each of these effects it was necessary to hold the other variable constant at the average for that variable. With two such highly correlated variables (r = .86, Table 1), such independent effects on milk PRL will not be realized under actual conditions because these temperatures are not independent. Nevertheless, fluctuations in these two temperature variables did account for 20.5% of the JOURNAL OF DAIRY SCIENCE VOL. 58. NO. 2

variation in milk PRL. The next variable in the regression equation (Table 2) was the second power of lactation stage (stage2). The F value for inclusion of this variable in the model was 48.4 (P < .005), and R z was increased from .205 to .239. The relationship between stage of lactation and milk PRL, independent of Pre-Hi s and Pre-Lo', was plotted in Fig. 2. Daily yield and lactation stage were significantly correlated (r ---- --.58, Table 1), and both variables had small but significant correlations with P R L

MILK

PR.OLACTIN

CONCENTRATIONS

185

(Table 1). After removing the variance due to Pre-Hi 3 and Pre-Lo ~ (step 2), the partial / correlation coefficient between daily yield ,. -. 100 and PRL was .18 and that between lactation E stage and PRL was --.20. The inclusion of On stage~ in the regression equation removed alC most all of the PRL variance associated with 50 daily yield, and this variable did not enter Z 0 the equation until step 19 (Table 2). The I-statistical inference that lactation stage 2 (in

Environmental and lactational variables affecting prolactin concentrations in bovine milk.

Relationships were examined between prolactin concentrations in bovine milk and various environmental and lactational variables. Prolactin was quantif...
788KB Sizes 0 Downloads 0 Views