Effect of prepartum source and amount of vitamin D supplementation on lactation performance of dairy cows

The objectives of this experiment were to determine the effects of supplementing 25-hydroxyvitamin D 3 (calcidiol, CAL) compared with vitamin D 3 (cholecal-ciferol, CHOL) at 1 or 3 mg/d in late gestation on production outcomes of dairy cows. One hundred thirty-three parous and 44 nulliparous pregnant Holstein cows were enrolled in the experiment. Cows were blocked by parity and previous lactation milk yield (parous) or genetic merit (nulliparous) and assigned randomly to receive 1 or 3 mg/d of CAL or CHOL in a 2 × 2 factorial arrangement of treatments (CAL1, CAL3, CHOL1, and CHOL3). Treatments were provided to individual cows as a top-dress to the prepartum diet from 250 d in gestation until parturition. The prepartum diet had a dietary cation-anion difference of −128 mEq/kg of dry matter. Production and disease were evaluated for the first 42 d in milk, and reproduction was evaluated to 300 d in milk. Incidence of postpartum diseases did not differ among treatments. Feeding CAL compared with CHOL increased yields of colostrum and colostrum fat, protein, and total solids, resulting in an increased amount of net energy for lactation secreted as colostrum (CHOL = 7.0 vs. CAL = 9.0 ± 0.7 Mcal). An interaction between source and amount was observed for milk yield: CAL3 increased milk yield compared with CHOL3 (CHOL3 = 34.1 vs. CAL3 = 38.7 ± 1.4 kg/d) but milk yield did not differ between CAL1 and CHOL1 (CHOL1 = 36.9 vs. CAL1 = 36.4 ± 1.4 kg/d). Concentrations of serum calcidiol on day of calving and average serum Ca from d 2 to 11 postpartum were positively associated with milk yield in the first 42 d in milk. Interactions between source and amount of vitamin D were also observed for pregnancy after


INTRODUCTION
Nutrition and metabolism of vitamin D for periparturient cows has significant implications for the transition period.The negative associations of hypocalcemia with immune function, risk of disease, lactation performance and reproduction are documented (Wilkens et al., 2020).As vitamin D is a key contributor to Ca, many attempts have been made to prevent hypocalcemia with vitamin D over the years and, although most attempts have not been successful, the amount and source of vitamin D can influence Ca homeostasis of transition cows (Vieira-Neto et al., 2017;Rodney et al., 2018).Meanwhile, scientists also have identified many extracalcemic functions of vitamin D (Norman, 2008) with implications for overall lactation performance for cows; thus, insufficient vitamin D potentially has widespread negative effects (Eder and Grundmann, 2022).Conversely, oversupply of vitamin D has potential for negative effects because of the toxicity potential of vitamin D (Littledike and Horst, 1982).Despite the importance of vitamin D for dairy cows, data to guide evidence-based recommendations for vitamin D supplementation of dairy cows remain sparse.
Vitamin D refers to a group of seco-steroid molecules.Cholecalciferol (CHOL; vitamin D 3 ) is endogenously synthesized in the skin of dairy cows upon UV B light exposure and readily converted to calcidiol (CAL; 25-hydroxyvitamin D 3 ) by hepatic 25-hydroxylases (Horst and Reinhardt, 1983).Calcidiol is a precursor to the active vitamin D metabolite, calcitriol, which exerts pluripotent biological activity via vitamin D receptors present in nearly every cell type of the body (Norman, 2008).Physiological processes affected by vitamin D signaling with relevance to transition cows include Ca and P transport, host defense, inflammation, and mammary development (Zinser et al., 2002;Merriman et al., 2015;Wilkens et al., 2020).
The National Academy of Science, Engineering, and Medicine (NASEM, 2021) recommends that dairy cows receive 0.7 mg/d (28,000 IU/d) of CHOL, largely to support Ca and P homeostasis but also to support immune-related outcomes of vitamin D. The recommendations for vitamin D reflect an estimation of adequate intake because too few experiments have evaluated the effect of CHOL on production outcomes of dairy cows.Nonetheless, CHOL is routinely supplemented in the diets of dairy cows in the United States at a rate of 0.75 to 1.25 mg/d because it is relatively inexpensive and provides potential benefits (Weiss, 1998;Nelson et al., 2016).In some instances, CHOL is supplied at more than 5 times the recommendation (Holcombe et al., 2018) despite data showing that excessive CHOL supplementation will depress feed intake and feed efficiency (Montgomery et al., 2004).Therefore, experiments are needed that evaluate the effect of CHOL on production outcomes of transition cows.
Supplemental CAL was recently introduced as an alternative source of vitamin D for ruminants.Supplemental CAL in the diets of dairy cows is much more effective at increasing serum CAL concentration compared with CHOL (Rodney et al., 2018;Poindexter et al., 2020).Supplementing 3 mg/d CAL compared with 3 mg/d CHOL increased colostrum IgG and ECM yield (Martinez et al., 2018b).Similarly, supplementing 3 mg/d CAL increased colostrum, milk, and ECM yields compared with cows fed 0.625 mg/d CHOL (Silva et al., 2022).Collectively, these data indicate promising potential for supplemental CAL in prepartum diets.However, the effective dose of CAL in prepartum diets is unknown.We hypothesized that feeding 1 or 3 mg/d of CAL prepartum would benefit lactation performance compared with equivalent amounts of CHOL.Thus, the objectives were to test the effects of CAL versus CHOL at 1 and 3 mg/d in late gestation on production outcomes in Holstein cows.

Sample Size Calculation
A 2-tailed sample size was calculated using α = 0.05 and β = 0.20 with the Power procedure of SAS (SAS/ STAT version 9.4; SAS Institute Inc.).The sample size was based on expected difference of 2.0 kg/d of ECM between cows fed CHOL and CAL based on the findings of Martinez et al. (2018b).Based on records from the University of Florida dairy herd, the standard deviation for ECM yield is 4.5 kg in the first 100 DIM.The sample size calculation indicated 41 cows per treatment would be sufficient to detect a difference in ECM yield of 2.0 kg/d between main effects or their interaction term.

Selection and Management of Cows
The experiment was conducted at the University of Florida Dairy Unit (Gainesville, FL) from August 2017 to June 2018.The University of Florida Institutional Animal Care and Use Committee approved all procedures involving cows in the experiment (protocol number 20171002).Parous cows received dry cow therapy (SpectraMast-DC and Orbeseal; Zoetis) and were dried off at approximately 220 d of gestation.All cows received J-Vac (Boehringer Ingelheim) at approximately 220 and 248 d of gestation and 28 d postpartum.Cows also received ScourGuard 4KC (Zoetis) at 248 d of gestation.One hundred seventy-seven apparently healthy dry Holstein cows (133 parous and 44 nulliparous) were enrolled in the experiment.Only 130 parous and 43 nulliparous cows contributed to the data analyses.Of the 177 cows enrolled, 3 (2 parous, 1 nulliparous) were found to be open several days after enrollment, and 1 parous cow calved with twins on the day of treatment commencement.These 4 cows were excluded from all analyses.Descriptive statistics of cows used in the experiment are shown in Supplemental Table S1 and Supplemental Figure S1 (https: / / original -ufdc .uflib.ufl.edu/l/ IR00011936/ 00001).
Pregnant, nonlactating cows at 242 ± 3 d of gestation were moved to the experimental freestall barn to acclimate to the facilities.Cows were trained for the first 2 d to use individual feeding gates (Calan Broadbent feeding system, American Calan Inc.).Data collected from 247 to 250 d gestation were used as covariates, and measurements starting at 250 d of gestation until parturition were used for statistical analyses of prepartum data.Prepartum cows were housed together in a freestall barn with sand-bedded stalls, and each cow was randomly assigned to an individual feeding gate for the entire prepartum period between movement to the barn and parturition.Immediately following parturition, cows were moved to a second freestall barn with no individual feeding gates and housed together for the first 42 DIM.The experimental pens were equipped with 2 rows of fans (1 fan/6 m) placed above the beds and a water soaker line with nozzles was placed above the feed bunk for cooling of cows.Fans and water spraying were controlled by thermostats and activated when ambient temperature reached 18°C.
Cows were fed a prepartum TMR with a DCAD of −128 mEq/kg of DM (Supplemental Table S2; https: / / original -ufdc .uflib.ufl.edu/l/ IR00011936/ 00001) once daily at 0630 h.The amounts of feed offered to individual cows prepartum were adjusted daily to result in between 5 and 10% refusals, which were weighed once daily, before the morning feeding.The postpartum TMR (Supplemental Table S2) was fed thrice daily at approximately 0700, 1100, and 1500 h.Postpartum cows were fed as a group and feed intake not recorded.Diets were sampled and analyzed as described in the companion article (Poindexter et al., 2023).

Experimental Design and Treatments
The experiment followed a randomized complete block design with cow as the experimental unit.Weekly cohorts of prepartum cows at 242 d of gestation were blocked by parity (0 vs. 1 vs. >1) and previous lactation 305-d milk yield (parous cows) or PTA of ECM (nulliparous cows) and, within each block, randomly assigned to receive 1 of the 4 treatments in a 2 × 2 factorial arrangement.Randomization was accomplished using the random number generator in Excel (Microsoft Corp.).
Treatments were 2 sources of vitamin D (CHOL or CAL) fed at either 1 mg or 3 mg per day as a top-dress to the TMR.Therefore, the 4 treatments were 1 mg/d CHOL (CHOL1), 3 mg/d CHOL (CHOL3), 1 mg/d CAL (CAL1), and 3 mg/d CAL (CAL3).The vitamin D supplements were prepared by diluting CHOL (Rovimix D3 500; DSM Nutritional Products) or CAL (HyD 1.25%; DSM Nutritional Products) into 100 g of corn meal.The vitamin D treatments were administered as a top-dress dispensed onto the TMR once daily from 250 d gestation until calving.Research personnel were not blinded to treatments.The treatments were provided in addition to CHOL included in the prepartum TMR at a rate of 0.045 mg/kg of DM.The postpartum TMR for all cows was formulated to provide CHOL at a rate of 0.052 mg/kg of DM.

DMI, BW, and BCS
Prepartum intake for each cow was calculated daily from 250 d of gestation until calving based on the weekly DM content measured at 105°C of the ingredients and the respective composition of diets.Cows were weighed and body condition scored using a 1 to 5 scale (Ferguson et al., 1994) on the day of enrollment and then once weekly prepartum, in the morning before feeding, until calving.Postpartum, immediately after each milking, cows were weighed on a walk-though scale (AfiWeigh, S.A.E.Afikim) and body condition was scored once weekly.

Characterization and Diagnosis of Diseases and Disorders
A complete physical examination of all cows was performed at 4, 7, and 12 DIM.In addition, cows were observed daily for the first 42 DIM, and any abnormal symptom or substantial decrease in milk yield resulted in cows undergoing further physical examination for diagnosis of clinical disease.During labor, cows were monitored for progress at 1-h intervals.If no progress was being made between intervals, assistance was provided and recorded as dystocia.Milk fever was characterized by a recumbent cow that responded to intravenous administration of a solution containing 10.8 g of Ca as Ca borogluconate (Cal-Dex CMPK injection, Agrilabs).A plasma Ca concentration <1.5 mM immediately before Ca treatment was used to confirm the diagnosis of milk fever.Retained placenta was characterized by the retention of fetal membranes 24 h after calving.Metritis was diagnosed based on transrectal palpation of a flaccid enlarged uterus with the presence of watery, fetid, reddish/brownish discharge.All cows had rectal temperature measured at 1, 2, 3, 4, 5, 8, 11, 13, and 15 DIM, and those with temperatures >39.5°C were classified as having fever.At every milking, cows were examined for signs of clinical mastitis by the herd personnel immediately before milking based on presence of abnormal milk in one or more quarters.Displaced abomasum was diagnosed by percussion and auscultation of the flanks and confirmed during surgical intervention for correction by omentopexy.Cows with more than one clinical disease event were classified as having multiple diseases.Morbidity included milk fever, retained placenta, metritis, mastitis, or displaced abomasum recorded from calving to 42 DIM.Removal from the herd was determined for the first 300 DIM.

Blood Sampling and Processing
Blood was sampled at 250 d gestation and at 0, 1, 2, 3, 4, 5, 8, and 11 DIM by puncture of the coccygeal blood vessels into evacuated tubes (Vacutainer, Becton Dickinson) containing no anticoagulant agents for serum separation.All assays followed the initial random-ization with blocks such that samples from each block were analyzed in the same assay.Serum sampled at 250 d of gestation before the start of treatment was used as a covariate for all serum assays.Serum calcidiol, Ca, and P concentrations were measured as described in the companion paper (Poindexter et al., 2023).Concentrations of BHB and free fatty acids (FA) were measured using an automated biochemical analyzer (RX Daytona, Randox Laboratories Ltd.) with respective kits for each analyte according to the manufacturer's instructions.Intra-and interassay coefficients of variation were 1.9 and 7.9% for BHB, and 1.24 and 12.8% for FA.

Measurements of Colostrum, Yields of Milk, and Milk Components
Cows were milked by machine into individual buckets within the first 6 h after calving, and yield was automatically measured and manually recorded for the first and second milkings (AfiFlo; S.A.E.Afikim).Duplicate samples were collected during the first and second milkings and diluted 1:1 with skim milk.Samples of skim milk and diluted colostrum samples were analyzed in duplicate for concentrations of fat, true protein, lactose, SNF, TS, and SCC at the Southeast DHI laboratory (Belleview, FL).Concentrations of IgG in colostrum were quantified using a radial immunodiffusion assay by the Saskatoon Colostrum Company (Saskatoon, SK, Canada).
Cows were milked twice daily at 0600 and 1800 h and yields of milk were recorded automatically (AfiFlo) for the first 42 DIM.Samples of milk were collected from sequential milkings once weekly for analyses of concentrations of fat, true protein, lactose, and SCC at the Southeast DHI laboratory (Belleview, FL).Milk yield from each sampling was used to calculate the final concentrations of milk components.Yields of milk corrected for 3.5% fat content and for energy, and the net energy content of milk were calculated as follows:

Reproductive Management and Reproductive Responses
All cows were subjected to the double Ovsynch protocol (Souza et al., 2008) for first AI, and timed AI was performed at 80 ± 3 DIM, approximately 14 to 16 h after the final GnRH treatment.Cows that returned to estrus were re-inseminated on the same day and considered to be nonpregnant to the previous insemination.Pregnancy was diagnosed 32 d after each AI based on the presence of an amniotic vesicle with an embryo with heartbeat by transrectal ultrasonography.Nonpregnant cows had the estrous cycle resynchronized for timed AI with the Ovsynch protocol to be re-inseminated 10 d after the nonpregnancy diagnosis.Pregnant cows at 32 d were re-evaluated for pregnancy at 74 d after AI.For statistical analyses, the diagnosis at 74 d after AI was used to determine whether a cow became pregnant to the first or subsequent AI.Interval to pregnancy up to 300 d postpartum was recorded.Cows determined not eligible for breeding, or that were sold or died, or that remained nonpregnant by 300 d postpartum were censored on the respective dates.Responses measured included proportion of cows receiving AI, days postpartum at first AI, pregnancy per AI (P/AI) at first AI, P/AI at all AI, and interval to pregnancy.

Statistical Analysis
Normality of residuals and homogeneity of variance were examined for each continuous dependent variable analyzed after fitting the final statistical model.Responses that violated the assumptions of normality (e.g., colostrum yield) were subjected to power transformation according to the Box-Cox procedure (Box and Cox, 1964) using the TRANSREG procedure of SAS (version 9.4, SAS/STAT, SAS Institute Inc.).
Data were analyzed by ANOVA with the MIXED procedure of SAS.Data for pre-and postpartum periods were analyzed separately.The statistical models for single measures included the fixed effects of source of vitamin D (CHOL vs. CAL), amount of vitamin D (1 vs. 3 mg), parity (nulliparous vs. parous), and the interactions between source and amount; source and parity; amount and parity; source, amount, and parity; and the pretreatment covariate value.The model also included the random effect of block.For responses with repeated measures, the statistical models were the same as previously described but also included the fixed effects of day of measurement, and the interactions between source and day; amount and day; parity and day; source, amount, and day; source, amount, parity, and day; and the random effect of cow nested within source and amount of vitamin D. For yields of milk and milk components, previous lactation 305-d ECM (parous cows) and predicted ECM from genomic PTA (nulliparous cows) were normalized to generate relative centering values to use as covariates.The associations of serum Ca, P, BHB, and FA with serum CAL were Poindexter et al.: EFFECT OF VITAMIN D ON COW PERFORMANCE evaluated using mixed models with parity as a fixed effect and block as a random effect.
Binary data were analyzed by logistic regression with the GLIMMIX procedure of SAS (SAS/STAT).The statistical models included the fixed effects of source of vitamin D (CHOL vs. CAL), amount of vitamin D (1 vs. 3 mg), parity (nulliparous vs. parous), the interactions between source and amount, and the random effect of block.
The Kenward-Roger method was used to compute the approximate denominator degrees of freedom for the F tests in the statistical models.When an interaction was significant, pairwise comparisons among treatments were performed after adjusting by the method of Tukey.Statistical significance was considered at P ≤ 0.05, and tendency was considered at 0.05 < P ≤ 0.10.

Prepartum DMI, BW, and Body Condition
All cows (n = 173) contributed to the full prepartum analyses.Dry matter intake in the last 21 d of gestation did not differ between sources of vitamin D (Table 1).However, 3 mg tended to decrease (P = 0.08) DMI compared with 1 mg (1 mg = 10.6 vs. 3 mg = 10.2 ± 0.2 kg/d).The effect of the amount was more apparent in parous than in nulliparous cows (parity × amount, P = 0.07) because the amount of vitamin D did not affect DMI in nulliparous cows (Figure 1A) but 3 mg decreased intakes of parous cows compared with 1 mg as the day of parturition approached (Figure 1B).

Lactation Performance
An interaction between source and amount of vitamin D was observed for milk yield in the first 42 DIM.Feeding CAL3 prepartum increased (P < 0.05) milk yield in the first 42 DIM compared with CHOL3 (Table 4), whereas milk yield was similar for CAL1 and CHOL1.In nulliparous cows, the interaction between source and amount appeared to be from a reduction in milk yield for cows fed CHOL3 compared with other treatments (Figure 2A), whereas in parous cows, the interaction appeared to be driven by an increase in milk yield for cows fed CAL3 compared with other treatments (Figure 2B).
Feeding CAL tended to produce more (P = 0.09) fat as a percentage of milk (CHOL = 4.47 vs. CAL = 4.65 ± 0.09%; Table 4).Feeding CAL resulted in greater (P = 0.05) fat yield compared with CHOL (CHOL 1.43 = vs.CAL 1.54 ± 0.05 kg/d; Table 4), but the amount of CAL or CHOL did not affect fat yield (Table 4).Concentrations of protein in milk did not differ among treatments; however, CAL3 and CHOL1 increased (P = 0.03) protein yield in the first 28 DIM compared with CAL1 and CHOL3 (Table 4).In contrast to fat, CAL tended to decrease (P = 0.09) concentrations of lactose in milk (CHOL = 4.65 vs. CAL = 4.71 ± 0.03%) compared with CHOL.An interaction between source and amount of vitamin D was observed for lactose yield because CAL3 and CHOL1 resulted in greater (P = 0.04) lactose compared with CAL1 and CHOL3 cows (Table 4).
Overall, CAL tended to increase (P = 0.06) ECM in the first 28 DIM compared with CHOL (CHOL = 36.3vs. CAL = 39.0 ± 1.1 kg/d).The interaction between source and amount of vitamin D was not significant for ECM, but the pattern was similar to milk yield in that most of the difference between sources resulted from the difference between CAL3 and CHOL3 (Table 4).
The effect of disease incidence on production and its interaction with vitamin D was also evaluated.Cows with clinical disease postpartum produced less (P < 0.001) milk compared with cows without clinical disease.Incidence of disease did not affect the influence of vitamin D treatments on milk yield in the first 42 DIM but the numerical difference in least squares mean was more pronounced in cows with disease than in cows without disease (Table 5).Likewise, cows that  2 Source = effect of vitamin D source (CHOL vs. CAL); Amt = effect of amount of vitamin D per day (1 mg vs. 3 mg); Source × Amt = interaction between source and amount of vitamin D; Parity = effect of parity (nulliparous vs. parous).
experienced clinical disease postpartum produced less (P < 0.001) ECM compared with cows that did not experience disease.However, a tendency (P = 0.07) for interaction between disease and vitamin D was ob-served for ECM because, compared with CHOL3 and CAL1, CHOL1 and CAL3 increased ECM to a greater extent for cows with disease than for cows with no disease (Table 5).

Postpartum Concentrations of Blood Metabolites
As reported in the companion article (Poindexter et al., 2023), an interaction (P < 0.001) between vitamin D source and amount was observed for serum CAL concentrations, with the greatest differences observed on day of calving (Table 6).Feeding CAL also increased (P = 0.03) postpartum serum Ca concentrations and postpartum serum P concentrations (Table 6).Serum BHB increased (P < 0.001) after parturition to a maximum serum BHB concentration at 4 DIM (Supplemental Figure S3A).However, an interaction between source and amount was observed for serum BHB because the increment in serum BHB was greater (P = 0.04) in CHOL3 and CAL1 than in CHOL1 and CAL3 (Table 6).An interaction between amount of vitamin D and DIM (P = 0.002) and source was observed for serum FA in the first 8 DIM (Supplemental Figure S3B).Maximum serum FA concentrations for CHOL1, CAL1, and CAL3 were observed at 0 DIM, whereas maximum serum FA concentration for CHOL3 was at 2 DIM (Supplemental Figure 3B).Of the serum variables that were measured only postpartum, serum Ca was associated (P = 0.01) with the concentration of serum CAL on day of calving; however, it was the serum Ca concentrations from d 2 to 11 postpartum that were associated with serum CAL, not those on d 0 to 1 postpartum (Supplemental Table S4; https: / / original -ufdc .uflib.ufl.edu/l/ IR00011936/ 00001).
The relationship between serum CAL concentrations on day of calving and yields of milk and ECM were evaluated to gain insight into the effect of vitamin D treatment that was observed for milk and ECM yield.Milk yield increased (P = 0.04) proportionately with   serum CAL concentrations such that milk yield increased 21 ± 10 g for each 1 ng/mL increment in serum CAL (Figure 3A).Likewise, milk yield increased (P < 0.001) proportionately with the average concentrations of serum Ca from d 2 to 11 postpartum such that milk yield increased 3.1 ± 0.5 kg for each 0.1 mM increment in serum Ca (Figure 3B).When milk yield was modeled as a function of serum CAL, serum Ca, and the interaction of serum CAL by serum Ca and parity, an interaction (P = 0.002) between serum CAL and serum Ca was observed for milk yield (P = 0.002) because yield did not change with the serum Ca concentration when serum CAL concentration were larger (i.e., 215 ng/mL; Figure 3C), whereas milk yield increased with serum Ca concentration when serum CAL concentration were smaller (i.e., 56 to 106 ng/mL; Figure 3C).Milk yield was not associated with average serum Ca concentrations on d 0 and 1 postpartum (Supplemental Figure S4).

Reproductive Performance
Of the 173 cows in the experiment, 154 received at least 1 insemination and 153 had at least one pregnancy diagnosis performed.Of the 19 cows not bred, 15 were deceased and 4 were sold before the end of the 72-d voluntary waiting period (Table 7).Four cows were voluntarily removed from the breeding program by farm management before the cutoff of 300 DIM.Days to first insemination were similar among treatments with an average of 73.3 DIM.An interaction (P = 0.02) between vitamin D source and amount was observed for pregnancy after first AI because pregnancy rates were greater at 32 d for CHOL3 and CAL1 than for CHOL1 and CAL3 (Table 7).Overall, pregnancy per AI and the percentages of cows pregnant by 300 DIM did not differ among treatments (Table 7).Median days open did not differ between source or amount of vitamin D (Table 8).

DISCUSSION
Supplemental CAL is an alternative source of vitamin D for ruminant diets that is reported to have benefits for transition cow health and lactation performance (Martinez et al., 2018a,b;Silva et al., 2022).Because CAL is more effective than CHOL at increasing serum CAL concentrations, we hypothesized that feeding CAL at 1 or 3 mg/d compared with CHOL would improve lactation performance.We observed several outcomes that corroborate previous research (Martinez et al., 2018b;Silva et al., 2022) and support our hypothesis that CAL would improve lactation performance compared with CHOL.For example, CAL increased energy-corrected (A) Effect of serum calcidiol (P = 0.04) and parity (P < 0.001); regression lines represent slope for nulliparous (0.015 ± 0.008, P = 0.07) and parous (0.021 ± 0.007, P = 0.004) cows.(B) Effect of serum Ca (P < 0.001) and parity (P < 0.001); regression lines represent slope for nulliparous (35 ± 3, P < 0.001) and parous (28 ± 2, P < 0.001) cows.(C) Predicted milk yield as a function of serum calcidiol (P = 0.002), serum Ca (P < 0.001), interaction of serum calcidiol by serum Ca (P = 0.002), and parity (P < 0.001).Lines represent change in milk yield as a function of serum Ca using average serum calcidiol concentrations for CHOL1 (serum calcidiol = 56 ng/mL, dashed black line), CHOL3 (serum calcidiol = 63 ng/mL, solid black line), CAL1 (serum calcidiol = 106 ng/mL, dashed red line), and CAL3 (serum calcidiol = 215 ng/ mL, solid red line).From 250 d in gestation until calving, cows received either 1 or 3 mg/d of either cholecalciferol (CHOL) or calcidiol (CAL).Values represent LSM estimated from the mixed effects models, with the raw count indicated in parentheses.
2 Source = effect of vitamin D source (CHOL vs. CAL); Amt = effect of amount of vitamin D per day (1 mg vs. 3 mg); Source × Amt = interaction between source and amount of vitamin D; Parity = effect of parity (nulliparous vs. parous).
colostrum and milk yields compared with CHOL and, overall, CAL3 generally resulted in better performance than CHOL3.In contrast to our hypothesis, CAL1 did not increase milk yield compared with CHOL1.Our data indicate that amount and source of vitamin D fed prepartum affect lactation performance.Prepartum DMI in this experiment was slightly lower compared with previous experiments in the same herd with similar prepartum diets (Martinez et al., 2018b;Bollatti et al., 2020).Moreover, parous cows had similar prepartum intakes as parous cows in this experiment.The farm dealt with a period of poor-quality corn silage that was fed to prepartum cows, which could explain some of the low prepartum intakes and incidence of displaced abomasum postpartum.The prepartum diet had a negative DCAD, which resulted in mean urine pH of 5.8 (Poindexter et al., 2023).The partially compensated metabolic acidosis that results from a negative DCAD decreases DMI (Zimpel et al., 2018), but the amount of vitamin D in this experiment also caused a slight reduction in prepartum DMI.Very large amounts (e.g., 100 mg/d) of CHOL reduced DMI in in feedlot cattle (Scanga et al., 2001).In contrast, feeding 6.25 mg/d of CHOL to feedlot steers for 165 d did not reduce DMI (Korn et al., 2013).Even though 3 mg decreased prepartum DMI, it resulted in a smaller BW change postpartum compared with 1 mg.Whether the effect of vitamin D on BW change was related to intake could not be determined because postpartum DMI was not measured due to space limitations at the research facility.Overall, the smaller than expected prepartum intakes and lack of postpartum intakes present some limitations for the interpretation of production results.
Incidences of postpartum diseases did not differ among treatments in this experiment.In contrast, Martinez et al. (2018a) observed that cows fed 3 mg/d CAL had fewer cases of retained placenta and metritis compared with cows fed 3 mg/d CHOL (Martinez et al., 2018a).Likewise, Wisnieski et al. (2020) reported that serum 25(OH)D 3 concentrations in early lactation were associated with decreased risk for uterine diseases.Thus, an effect of vitamin D treatment on uterine diseases could be expected in the present experiment.However, the experiment was not designed to detect an effect of vitamin D treatment on health outcomes.For example, with a 10-percentage-point difference in disease incidence, the experiment would require 240 cows (120 cows per main effect or interaction term) to have an 80% probability of detecting a statistically significant difference.
We hypothesized that feeding CAL would increase colostrum and milk yield compared with CHOL based on Martinez et al. (2018b).Feeding CAL increased TS and net energy yield of colostrum and ECM yield compared with CHOL.Feeding CAL3 also increased milk yield compared with CHOL3 in the present experiment, which was comparable to the 3.7 kg/d increase reported by Martinez et al. (2018b).Notably, even though milk yield differed between CHOL3 and CAL3, neither were different from CHOL1 and CAL1.Elsewhere, feeding CAL at 3 mg/d increased colostrum yield by 2 kg and ECM by 3 kg/d compared with CHOL at 0.625 mg/d (Silva et al., 2022).Golder et al. (2021) reported that CAL supplementation at 2 mg/d in transition (−21 d to 21 d relative to calving) and 1 mg/d during lactation compared with no CAL did not affect milk yield of cows in an experiment comprising 1,167 cows in 4 dairy herds in Australia and New Zealand.The experimental design and conditions of the experiment reported in Golder et al. (2021) differed drastically from those here but, to date, their experiment has been the most extensive to evaluate the effect of CAL supplementa- tion on production and health of dairy cows.Further experimentation on the amount and duration of CAL is certainly required as promising results from CAL have been observed for lactation performance when fed at 3 mg/d but not when feeding greater amounts (6 mg/d; Weiss et al., 2015) or lesser amounts of CAL as shown here and elsewhere (Golder et al., 2021).
The pleiotropic actions of vitamin D allow for several speculative explanations for the effect of CAL on colostrum and ECM yield.Vitamin D receptors are present in nearly every cell type of the body and exert widespread regulation of transcription and chromatin remodeling activity when activated by calcitriol (Norman, 2008).The prominent role of vitamin D in Ca homeostasis likely contributes at least in part to the effect of CAL3 on lactation performance observed in this experiment.Reduced plasma Ca concentrations at 4 DIM were associated with reduced milk yield, whereas reduced plasma Ca concentrations at 1 DIM were associated with greater milk yields (Neves et al., 2018).Mitigation of hypocalcemia by feeding prepartum diets with a negative DCAD also results in greater milk yield and fewer uterine diseases (Santos et al., 2019).Supplemental CAL increased postpartum serum Ca concentrations compared with CHOL in this experiment.Serum CAL concentrations were also positively associated with postpartum serum Ca, specifically for d 2 to 11 postpartum.Moreover, milk yield was positively associated with concentrations of CAL and Ca in serum.The interaction of serum CAL × serum Ca observed for milk yield gives some indication that increased milk yield resulting from increased serum CAL is mediated in part by Ca.On the other hand, the relationship between milk yield and serum Ca at d 2 to 11 postpartum may be explained by other factors such as inflammation (Horst et al., 2020).Morbidity in the first few days of lactation reduces subsequent lactation performance (Santos et al., 2004;Dubuc et al., 2011).A weak interaction between vitamin D treatment and incidence of morbidity was observed for ECM yield.It is plausible that cows fed CAL3 were not affected by disease to the same extent as other treatments because CAL3 improved resolution of postpartum disease.Vitamin D acts to decrease inflammation (Zhang et al., 2012;Zhao et al., 2012), so improved resolution of disease from increased serum CAL could simultaneously alleviate hypocalcemia associated with inflammation.However, this explanation is highly speculative and the interaction between morbidity and vitamin D treatment is not sufficient to support this theory.
Direct effects of CAL on vitamin D signaling in mammary epithelial cells also conceivably contributed to the effect of CAL on colostrum and ECM yields.Mammary vitamin D receptors are required for normal mammary development in mice (Zinser et al., 2002).Bovine mammary epithelial cells exhibit vitamin D receptor activity when exposed to vitamin D metabolites (Yue et al., 2017;Kweh et al., 2019).In goat mammary epithelial cells, calcitriol increased glucose uptake and expression of genes for Ca and glucose transport in goat mammary epithelial cells (Sun et al., 2016).Mammary epithelial cells also have the capacity to convert CAL to calcitriol via activity of 1α-hydroxylase (Kemmis et al., 2006).The intracrine and paracrine vitamin D signaling in mammary tissue promotes mammary epithelial cell differentiation while inhibiting ductal outgrowth in mice (Kemmis et al., 2006).In theory, dramatically increased concentrations of CAL in the circulation provide more substrate to potentiate vitamin D signaling activity in mammary cells, allowing for greater intracrine vitamin D activity that potentiates capacity of mammary epithelial cells.
The capacity to produce more milk must come from greater DMI, mobilization of energy reserves, or feed efficiency.Dry matter intake and, therefore, energy balance and feed efficiency were not determined for the postpartum period because of space limitations.Martinez et al. (2018b) did not find a difference in feed efficiency between CHOL and CAL, but rather found that feeding CAL at 3 mg/d reduced energy balance compared with feeding 3 mg/d CHOL in the first 42 DIM.Silva et al. (2022) similarly reported that feeding CAL at 3 mg/d decreased energy balance compared with CHOL at 0.625 mg/d.Martinez et al. (2018b) also reported interactions between source of vitamin D and parity for serum indicators of energy balance, in that CAL increased serum concentrations of free FA and BHB compared with CHOL in nulliparous but not parous cows.In contrast, 3 mg/d CAL increased plasma glucose and decreased plasma FA compared with 0.625 mg/d CHOL (Silva et al., 2022).Source and amount of vitamin D affected serum BHB concentrations postpartum, with CHOL1 resulting in the lowest BHB concentration among treatments; however, serum CAL was not associated with serum BHB.Poindexter et al. (2020) showed that serum CAL concentrations were associated with serum BHB in lactating dairy cows fed 1 or 3 mg/d of CHOL or CAL.Wisnieski et al. (2020) reported that serum CAL concentrations at dry-off were predictive of postpartum ketosis.The lack of association of serum CAL with BHB and FA concentrations in the present experiment indicates that the effect of CAL was probably not attributed to lipid mobilization or ketogenesis.
The effects of source and amount of vitamin D on reproductive outcomes were mixed in comparison with lactation and health outcomes.Despite the positive effect of CAL3 on lactation, cows fed CAL3 had the lowest rate of pregnancy per AI at 32 d after first AI among treatments.In contrast, cows fed CHOL3 and CAL1 had the greatest rate of P/AI at 32 d after first AI.Some of the discrepancy may result from numerically fewer CHOL3 cows being eligible for AI at the end of the voluntary waiting period.Cows that were ineligible for AI presumably would have a lower pregnancy rate and, therefore, confounded the outcome for P/AI.Altogether, P/AI for all AI, pregnant by 300 DIM, and median days open did not differ among treatments.In contrast, Golder et al. (2021) reported an interaction between parity and CAL supplementation for median days open in an experiment with 1,036 cows.Supplemental CAL did not affect probability of pregnancy in primiparous cows but increased probability of pregnancy, and hence decreased median days open, in multiparous cows.Martinez et al. (2018a) also observed a tendency for fewer median days open for cows fed 3 mg/d CAL compared with 3 mg/d CHOL.
In contrast to the positive effects of CAL on physiological processes that support lactation, some of the difference observed between CHOL3 and CAL3 may be explained by negative effects of excess CHOL.Dietary CHOL is absorbed by chylomicrons and circulates in lipoprotein fractions of plasma to a greater extent than CAL or cutaneous-derived CHOL (Hymøller and Jensen, 2017).Cholecalciferol also has a greater propensity to accumulate in adipose tissue compared with CAL, whereas dietary CAL has a greater propensity for blood compared with CHOL (Burild et al., 2016).Tissue accumulation of CHOL presents the potential for aberrant responses to excess CHOL without a change in serum 25(OH)D 3 concentration (Fraser, 2021).The total amount of CHOL consumed by CHOL3 cows was nearly 5 times the recommendation for CHOL (NASEM, 2021).In contrast, CHOL1 was comparable to current practices for dairy cows in the United States (Nelson et al., 2016) and within the amount of CHOL generally recognized as tolerable for dairy cows (NASEM, 2021).Cows fed CHOL3 had numerically less milk yield and greater incidences of mastitis and metritis compared with cows fed CHOL1.The data from this experiment do not support the practice of feeding CHOL at a rate of 3 mg/d in the prepartum.

CONCLUSIONS
Feeding CAL compared with CHOL prepartum had positive effects on lactation performance of cows but some outcomes depended on the amount of vitamin D provided.Yield of net energy and TS in colostrum and ECM was increased by CAL compared with CHOL.However, milk yield in the first 42 DIM was affected by an interaction between source and amount of vitamin D. Feeding CAL3 prepartum increased milk yield compared with CHOL3, but milk yield of CAL3 and CHOL3 were not significantly different from that of CAL1 and CHOL1.We speculate that the effect of CAL on lactation performance resulted in part from increased postpartum serum Ca, but the other physiological actions of vitamin D, specifically those in mammary and immune cells, may also have contributed to the differences in lactation performance.Overall, our data indicate benefits for milk production from feeding CAL3 prepartum but show that 3 mg/d CHOL is not beneficial in diets of prepartum to dairy cows.
2 Source = effect of vitamin D source (CHOL vs. CAL); Amt = effect of amount of vitamin D per day (1 mg vs. 3 mg); Source × Amt = interaction between source and amount of vitamin D.3 Prepartum BW and BCS measured weekly from 250 d gestation until calving.4Postpartum BW and BCS measured weekly until 42 DIM.
Poindexter et al.: EFFECT OF VITAMIN D ON COW PERFORMANCE Poindexter et al.: EFFECT OF VITAMIN D ON COW PERFORMANCE Table 2. Incidence of postpartum diseases according to source and amount of vitamin D

2
Source = effect of vitamin D source (CHOL vs. CAL); Amt = effect of amount of vitamin D per day (1 mg vs. 3 mg); Source × Amt = interaction between source and amount of vitamin D; Parity = effect of parity (nulliparous vs. parous).

2
Source = effect of vitamin D source (CHOL vs. CAL); Amt = effect of amount of vitamin D per day (1 mg vs. 3 mg); Source × Amt = interaction between source and amount of vitamin D; Parity = effect of parity (nulliparous vs. parous).

2
Source = effect of vitamin D source (CHOL vs. CAL); Amt = effect of amount of vitamin D per day (1 mg vs. 3 mg); S × A = interaction between source and amount of vitamin D; Dis = effect of disease; S × A × D = interaction between source, amount and disease..

Table 1 .
Poindexter et al.: EFFECT OF VITAMIN D ON COW PERFORMANCE Effect of source and amount of vitamin D fed prepartum on measures of feed intake, BW, and BCS 1 From 250 d in gestation until calving, cows received either 1 or 3 mg/d of either cholecalciferol (CHOL) or calcidiol (CAL).

Table 3 .
Poindexter et al.: EFFECT OF VITAMIN D ON COW PERFORMANCE Effect of source and amount of vitamin D fed prepartum on colostrum 1From 250 d in gestation until calving, cows received either 1 or 3 mg/d of either cholecalciferol (CHOL) or calcidiol (CAL).

Table 4 .
Effect of source and amount of vitamin D fed prepartum on lactation performance 1From 250 d in gestation until calving, cows received either 1 mg or 3 mg per day of either cholecalciferol (CHOL) or calcidiol (CAL).

Table 5 .
Effect of source and amount of vitamin D on production according to disease status 1 From 250 d in gestation until calving, cows received either 1 or 3 mg/d of either cholecalciferol (CHOL) or calcidiol (CAL).Cows were categorized as having no disease or one or more of the following diagnoses: displaced abomasum, lameness, mastitis, metritis, respiratory disease, or retained placenta in the first 42 DIM.

Table 7 .
Effect of source and amount of vitamin D fed prepartum on reproduction Item

Table 8 .
Poindexter et al.: EFFECT OF VITAMIN D ON COW PERFORMANCE Cox's proportional hazard model for time to pregnancy in cows fed 2 levels of vitamin D and 2 sources of vitamin D prepartum 1