Association between prepartum body condition score and prepartum and postpartum dry matter intake and energy balance in multiparous Holstein cows

The objectives of this retrospective observational study were to investigate the association between body condition score (BCS) at 21 d before calving with prepartum and postpartum dry matter intake (DMI), energy balance (EB), and milk yield. Data from 427 multigravid cows from 11 different experiments conducted at the University of Florida were used. Cows were classified according to their BCS at 21 d before calving as FAT (BCS ≥4.00; n = 83), MOD (BCS 3.25 to 3.75; n = 287), and THIN (BCS ≤3.00; n = 57). Daily DMI from −21 to −1 and from +1 to +28 DIM was individually recorded. Energy balance was calculated as the difference between net energy for lactation consumed and required. Dry matter intake in FAT cows was lesser than in MOD and THIN cows both prepartum (FAT = 9.97 ± 0.21, MOD = 11.15 ± 0.14, THIN = 11.92 ± 0.22 kg/d) and postpartum (FAT = 14.35 ± 0.49, MOD = 15.47 ± 0.38, THIN = 16.09 ± 0.47 kg/d). Dry matter intake was also lesser for MOD cows compared with THIN cows prepartum, but not postpartum. Energy balance in FAT cows was lesser than in MOD and THIN cows both prepartum (FAT = −4.16 ± 0.61, MOD = −1.20 ± 0.56, THIN = 0.88 ± 0.62 Mcal/d) and postpartum (FAT = −12.77 ± 0.50, MOD = −10.13 ± 0.29, THIN = −6.14 ± 0.51 Mcal/d). Energy balance was also lesser for MOD cows compared with THIN cows both prepartum and postpartum. There was a quadratic association between BCS at 21 d before calving and milk yield. Increasing BCS from 2.5 to 3.5 was associated with an increase in daily milk yield of 6.0 kg and 28 d cumulative milk of 147 kg. Increasing BCS from 3.5 to 4.5 was associated with a decrease in daily milk yield of 4.4 kg and 28 d cumulative milk of 116 kg. In summary, a moderated BCS at 21 d before calving was associated with intermediate DMI and EB pre-and postpartum but greater milk yield compared with thinner and fatter cows. Our findings indicate that a moderated BCS is ideal for ensuring a successful lactation.


INTRODUCTION
The transition into lactation is critically important for dairy cow health, production, and farm profitability given that 75% of all infectious diseases and metabolic disorders emerge during this time (LeBlanc et al., 2006).Consequently, only 56% of cows remain healthy by 60 d after parturition (Santos et al., 2010).The occurrence of health disorders in the first weeks after calving results in less milk production not only during the time of disease, but also during the entire lactation (Rajala-Schultz et al., 1999).Approximately 25% of cullings from dairy farms occur within the first 60 d after parturition due to a poor transition into lactation (Dechow and Goodling, 2008); therefore, the success of the transition period effectively determines farm profitability (Drackley, 1999).
The primary challenge cows face during the transition into lactation is a reduction in feed intake when nutrient demand for the developing fetus and the impending lactogenesis are increasing (Abuelo et al., 2019).Consequently, most cows experience negative nutrient balance, characterized by a sharp decrease in blood minerals, amino acids, and glucose, leading to body fat mobilization in the form of nonesterified fatty acids (NEFA), and accumulation of products of incomplete oxidation of NEFA such as β-hydroxybutyrate (BHB).Both the decrease in calcium and glucose (Galvão et al., 2010;Martinez et al., 2012), and the increase in NEFA (Ster et al., 2012;Casaro et al., 2023) and BHB concentrations (Grinberg et al., 2008) are associated with immune dysregulation and increased risk of postpartum diseases.It is crucial, therefore, to understand why some cows suffer from a sharper reduction in feed intake, which triggers this chain of events.
In the periparturient period, dry matter intake (DMI) can be influenced by various animal, dietary, and management factors (Hayirli et al., 2002;Huzzey et al., 2007;Daros et al., 2021).According to Hayirli et al. (2002), cows with a high body condition score (BCS >3.75) 21 d before calving had the most pronounced decrease in prepartum DMI as a proportion of their body weight (DMI%BW) compared with cows with BCS = 3.25-3.75and cows with BCS <3.25.Daros et al., 2021 also showed that cows with BCS >3.25 at dryoff had lower DMI and feeding time during the early and late dry periods than cows with BCS ≤3.25.Previous work had also shown that cows with BCS >3.75 at dry-off had more reproductive diseases and disorders such as dystocia, retained fetal membranes, metritis and endometritis (Gearhart et al., 1990).Additionally, loss of BCS during the dry period was associated with increased incidence of uterine disease and indigestion and reduced milk yield and decreased pregnancy after first and second AI, with BCS at dry-off accounting for 94% of the variation in BCS change during the prepartum period (Chebel et al., 2018).These studies suggest that cows with high BCS in the early or late dry period experience a sharper reduction in DMI, resulting in lower energy balance (EB), and increased risk of postpartum health disorders.
Although higher BCS during early or late prepartum has been associated with a decrease in prepartum DMI%BW (Hayirli et al., 2002) and absolute DMI (Daros et al., 2021), it is still unclear what is the ideal BCS prepartum that optimizes DMI pre-and postpartum and milk yield postpartum.Furthermore, the association between BCS prepartum and EB pre-and postpartum has not been evaluated.The hypothesis of the current study was that overconditioned cows (BCS >3.75 at 21 d before calving) will have a reduced DMI both in absolute terms and as a percentage of their body weight prepartum and postpartum, which will lead to a lower EB prepartum and postpartum.The objectives of this observational study were to investigate the association between BCS at 21 d before calving and prepartum and postpartum absolute DMI, DMI%BW, EB, and milk yield.

Experimental Design and Sample Size
A retrospective longitudinal study was performed using the data from a total of 427 multigravid cows from 11 different experiments conducted at the University of Florida Dairy Unit, located in the city of Hague, Florida.This was a convenience sample; therefore, a priori sample size calculation was not performed.A posteriori sample size was calculated using JMP Pro (Version 15.SAS Institute Inc., Cary, NC) based on prepartum DMI as a % of BW between cows with BCS ≥4.00 and cows with a BCS <4.00 (Hayirli et al., 2002).Based on an anticipated difference of 0.15% in DMI as a % of BW with a pooled standard deviation of 1.07%, a total sample size of 402 cows would be sufficient for achieving 80% power at an α level of 0.05.Individual experiments were approved by the University of Florida Animal Research Committee.
The 11 experiments originating these data were conducted between 2007 and 2017.During the study periods, the University of Florida Dairy Unit milked approximately 500 Holstein cows with an average milk production of 10,500 kg/cow and had a twice-daily milking schedule.The freestall beds and walking alleys were cleaned twice a day and sand was added on top of the freestall beds twice a week to ensure they remained clean and dry.In addition, the barns were equipped with fans over the stalls and soakers over the feed line.Fans ran continuously and water soakers turned on automatically for 1 min at 4-minute intervals when ambient temperatures exceeded 18°C.Lights were on in the barns for 14h/d at an intensity of 150 lx throughout all the studies, thus all cows were exposed to the same photoperiodic conditions.The housing facility separated primiparous and multiparous cows.Vaccinations and treatments for common diseases or disorders were administered according to standard operating procedures developed in collaboration with veterinarians from the University of Florida's College of Veterinary Medicine, Food Animal Reproduction and Medicine Service (FARMS).The stocking density (cows/stalls) was maintained between 80 and 100%.
All cows with a BCS measurement at 21 d before calving that had DMI data until calving or until 28 DIM were included in the prepartum and postpartum analyses, respectively, with the exception of 10 cows from Thompson et al. (2014), which received an intramammary dose of Streptococcus uberis at 5 DIM, hence, their postpartum data were excluded.To account for any conditional effect of treatments, any treatment that affected DMI or milk yield were added to the models.Greco et al. (2014) 2014) evaluated the effects of feeding diets with distinct DCAD levels during the last 21 d before calving.Feeding negative DCAD (−130 mEq/kgDM) resulted in lesser prepartum DMI than feeding positive DCAD (+130 mEq/kgDM).In the remaining studies, if the DCAD was not reported, it was calculated assuming the same absorption for the strong ions with the following formula: DCAD = [(mEq of K + + mEq of Na + ) − (mEq of Cl − + mEq of S 2− )].
To account for the effect of DCAD, the variable 'DCAD' was created.Because the effect of DCAD on DMI is linear, the variable included 3 levels: positive (POS = DCAD >0 mEq/kgDM, n = 94), partially acidified (PNEG = DCAD between 0 and −100 mEq/ kgDM, n = 229), and fully acidified (FNEG = DCAD < −100 mEq/kgDM, n = 104).Six of the studies were conducted during the hot months (June to October) to investigate the effect of evaporative cooling during the dry period on production measures (do Amaral et al., 2009(do Amaral et al., , 2011;;Tao et al., 2011Tao et al., , 2012;;Gomes, 2014;Thompson et al., 2014).For these experiments, cows were provided with shade only or with shade plus evaporative cooling with fans and soakers.The average environmental temperature during the 3 weeks before calving for these experiments was 26.9°C ± 2.0°C and a temperature humidity index (THI) of 77.7 ± 2.8.The categorization of heat stress abatement applied in these 6 previous experiments was maintained, resulting in cows being classified as hot with evaporative cooling or hot without evaporative cooling.In the remaining studies, cows were enrolled from December to May, with an average environmental temperature of 17.6°C ± 3.4°C and a THI of 63.8 ± 8.9 (Greco, 2014;Lopera et al., 2018;Martinez et al., 2018;Zenobi et al., 2018;Bollatti et al., 2020).The researchers chose a prepartum cutoff of THI ≥70 as the midpoint between the traditional (THI ≥72) and revised (THI ≥68) THI cutoffs for lac-tating dairy cows (Armstrong, 1994;Zimbelman et al., 2009).Therefore, cows were categorized as hot with evaporative cooling when the average THI during the last 3 weeks prepartum was ≥ 70 and cool when the average THI for the last 3 weeks prepartum was < 70.The variable heat stress abatement was created to account for any conditional effect of heat abatement: cool (CL; n = 213), hot without evaporative cooling (HT; n = 84), and hot with evaporative cooling (HTEV; n = 130).The THI was calculated using the following formula (Dikmen and Hansen, 2009): THI = 0.8° ambient temperature + ((relative humidity/100) × (ambient temperature -14.3)) + 46.4 The meteorological data obtained from The Weather Underground, Inc. (The Weather Underground Inc., 2016) for the city of Hague, Florida was used to calculate THI.A total of 30 cows participated in more than one study.

Measurement of Dry Matter Intake
Daily dry matter intake (DMI) was measured with a system with individual feeding gates (Calan Gates, American Calan Inc., Northwood, NH) recording the amount of feed consumed by each cow.For the current study, DMI data collected from −21 to −1 d before calving and from 1 to 28 DIM were used.DMI on the day of calving (d 0) was excluded due to inconsistent measurements resulting from parturition and pen moves.Chemical composition of diets used in each experiment included in this study can be found in Pérez-Báez et al. (2021).

BW and BCS
The weight of the cows was measured daily for 232 cows and weekly for 195 cows using a digital scale (Afi-Weight, S.A.E.Afikim).For cows weighed weekly, their daily weight was estimated by interpolation.Prepartum body weight was not available for 3 cows in 12 d.The daily weight measurements were used to calculate the DMI%BW, net energy for lactation (NE L ) required for maintenance (NE L R), and EB.Body condition score was evaluated weekly prepartum (from −21 to −1 d before calving) and postpartum (from 0 to 28 DIM) using a 1-to-5 scale, with 1 being emaciated and 5 being obese, according to Ferguson et al. (1994).The cows were categorized according to their BCS at 21 d before calving as FAT (BCS ≥4.00; n = 83), MOD (BCS 3.25 to 3.75; n = 287), and THIN (BCS ≤3.00; n = 57) as previously described (Gearhart et al., 1990;Hayirli et al., 2002).Others have used different cut-offs (Daros et al., 2021); however, we opted to follow the classification by Hayirli et al. (2002) because of the negative effect of BCS ≥4.0 on postpartum health (Gearhart et al., 1990) and because cows with BCS between 3.25 and 3.75 at calving had similar milk yield postpartum (Roche et al., 2007).The agreement among BCS observers was not evaluated.

BWC and BCSC
Body weight change (BWC) was calculated prepartum as the difference between BW at d −1 and BW at d −21 before calving and postpartum as the difference between BW at 28 DIM and BW at 1 DIM.Prepartum body condition score change (BCSC) was calculated as the difference between BCS at d −1 and BCS at d −21 before calving.Only 16, 122, and 21 cows had their BCS recorded at 28 DIM for FAT, MOD, and THIN, respectively; therefore, BCSC postpartum was not evaluated.

Milk Yield
Cows were milked twice a day, and milk production was recorded using milk meters (AfiFlo; S.A.E.Afikim).The concentrations of milk components such as fat, true protein, and lactose were also recorded either daily or weekly.For weekly measurements, daily milk components were calculated by interpolation.

Statistical Analysis
All continuous data were analyzed by ANOVA with linear mixed models using the MIXED procedure of SAS version 9.4 (version 9.4; SAS/STAT, SAS Institute Inc.) with estimations performed by the method of restricted estimated maximum likelihood.The cow was the experimental unit.After model fitting variables were evaluated for the distribution of the residuals, and all models met ANOVA assumptions.The data were divided into 2 periods, prepartum and postpartum, which were analyzed separately.All models included the random effect of study (1 to 11).
For prepartum and postpartum DMI, DMI%BW, NE L R, NE L C, EB, BW, and Milk the full models contained the fixed effects of BCS 21 d before calving (FAT, MOD, THIN), day (−21 to −1 in the prepartum and 1 to 28 in the postpartum), heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS 21 d before calving and day, BCS 21 d before calving and heat stress, BCS 21 d before calving and DCAD, and BCS 21 d before calving and FA.All postpartum models also contained the fixed effect of morbidity (yes, no) and the interaction between BCS 21 d before calving and morbidity.The variable morbidity (yes, no) was created, which included ketosis, milk fever, retained placenta, metritis, mastitis, displaced abomasum, or lameness recorded from calving to 28 DIM.Detailed description of diagnostic criteria for each health disorder can be found in previous publications (Pérez-Báez et al., 2019a;Pérez-Báez et al., 2019b;Perez-Baez et al., 2021).Models included the random effect of cow nested within study.The repeated statement consisted of time (day) with the subject being cow nested within study.The covariance structure with the smallest Akaike information criterion was selected.Autoregressive covariance structure resulted in the best fit for all responses.The fixed effect of BCS 21 d before calving, day, their interaction, and heat stress were forced in all models.When morbidity, season, DCAD, FA or their respective interactions had a P > 0.10, the effect was removed from the model and the model was rerun.When a main effect of BCS 21 d before calving was observed (P ≤ 0.05), posthoc For milk and cumulative milk produced in the first 28 DIM, the models contained the fixed effects of BCS 21 d before calving or BCSC as linear and quadratic covariates (BCS and BCS x BCS or BCSC and BCSC x BCSC, respectively), heat stress (CL, HT, HTEV), morbidity (yes, no), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between linear BCS 21 d before calving or BCSC and heat stress, BCS 21 d before calving or BCSC and morbidity, BCS 21 d before calving or BCSC and DCAD, and BCS 21 d before calving or BCSC and FA.When morbidity, season, DCAD, FA or their respective interactions had a P > 0.10, the effect was removed from the model and the model was rerun.The formula generated by the models was used to generate the predicted milk and cumulative milk production in 28 d by increasing BCS 21 d before calving from 2.5 to 4.5 or by increasing BCSC from −1.5 to 0.75.
To test if BCS 21 d before calving was linearly or quadratically associated with prepartum and postpartum DMI, models contained the fixed effects of BCS 21 d before calving as linear and quadratic covariates (BCS and BCS x BCS), heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn) and the interactions between linear BCS and heat stress, linear BCS and DCAD, and linear BCS and FA.Postpartum model also contained the fixed effect of morbidity (yes, no) and the interaction between linear BCS and morbidity.When morbidity, season, DCAD, FA or their respective interactions had a P > 0.10, the effect was removed from the model and the model was rerun.The formula generated by the models was used to generate the predicted DMI prepartum and postpartum by increasing BCS 21 d before calving from 2.5 to 4.5.

RESULTS
A total of 427 cows contributed data for the prepartum statistical analyses (FAT = 83; MOD = 287; THIN = 57).A total of 280 cows contributed data for the postpartum statistical analyses (FAT = 44; MOD = 192; THIN = 44).The distribution of BCS within categories can be found in Supplemental Figure 1.The full statistical models can be found in Supplemental Table 1.Supplemental material can be found in xxx.

Dry Matter Intake and DMI%BW
There was a main effect (P < 0.01) of BCS 21 d before calving for DMI prepartum, but no interaction was observed (P = 0.25) between BCS and day (Table 1, Figure 1).The main effect showed that FAT cows had lesser DMI than MOD (P < 0.01) and THIN (P < 0.01) cows, and MOD cows had lesser DMI than THIN cows (P < 0.01).When the linear and quadratic effects of BCS 21 d before calving were tested for DMI prepartum, a quadratic effect (P = 0.04) of BCS on DMI prepartum was observed (Supplemental Figure 2A).There was a main effect (P < 0.01) of BCS 21 d before calving for DMI postpartum, but no interaction was observed (P = 0.34) between BCS and day (Table 1, Figure 1).The main effect showed that FAT cows had lesser DMI than MOD (P < 0.01) and THIN (P < 0.01) cows.No difference (P = 0.16) in DMI postpartum was observed between MOD and THIN cows.When the linear and quadratic effects of BCS 21 d before calving were tested for DMI postpartum, a linear effect (P < 0.01) of BCS on DMI prepartum was observed, but no quadratic effect (P = 0.13) was observed (Supplemental Figure 2B).
There was a main effect (P < 0.01) of BCS 21 d before calving for DMI%BW prepartum, but no interaction was observed (P = 0.57) between BCS and day (Table 1, Figure 2).The main effect showed that FAT cows had lesser DMI%BW than MOD (P < 0.01) and THIN (P < 0.01) cows, and MOD cows had lesser DMI%BW than THIN cows (P < 0.01).There was a main effect (P < 0.01) of BCS 21 d before calving for DMI%BW postpartum, but no interaction was observed (P = 0.55) between BCS and day (Table 1, Figure 2).The main effect showed that FAT cows had lesser DMI%BW than MOD (P < 0.01) and THIN (P < 0.01) cows, and MOD cows had lesser DMI%BW than THIN cows (P < 0.01).

Net Energy for Lactation Required, NE L C, and EB
An interaction (P < 0.01) was observed between BCS 21 d before calving and day for NE L R prepartum (Table 1, Supplemental Figure 3).The interaction showed that FAT cows had greater NE L R than MOD and THIN cows from d-21 to d-1, and that MOD cows had greater NE L R than THIN cows from d-21 to d-1.There was a main effect (P < 0.01) of BCS 21 d before calving for NE L R postpartum, but no interaction was observed (P = 0.21) between BCS and day (Table 1, Supplemental Figure 3).The main effect showed that FAT and MOD cows had greater NE L R than THIN cows (P < 0.01), but no difference (P = 0.48) in NE L R was observed between FAT and MOD cows.
There was a main effect (P < 0.01) of BCS 21 d before calving for NE L C prepartum, but no interaction was observed (P = 0.30) between BCS and day (Table 1, Supplemental Figure 3).The main effect showed that FAT cows had lesser NE L C than MOD (P < 0.01) and THIN (P < 0.01) cows, and MOD cows had lesser NE L C than THIN cows (P < 0.01).
There was a main effect (P < 0.01) of BCS 21 d before calving for L C postpartum, but no interaction was observed (P = 0.55) between BCS and day (Table 1, Supplemental Figure 3).The main effect showed that FAT cows had lesser NE L C than MOD (P < 0.01) and THIN (P < 0.01) cows.No difference in NE L C postpartum was observed between MOD and THIN cows (P = 0.66).
There was a main effect (P < 0.01) of BCS 21 d before calving for EB prepartum, but no interaction was observed (P = 0.40) between BCS and day (Table 1, Figure 3).The main effect showed that FAT cows had lower EB than MOD and THIN cows (P < 0.01), and MOD cows had lower EB than THIN cows (P < 0.01).There was a main effect (P < 0.01) of BCS 21 d before calving for EB postpartum, but no interaction was observed (P = 0.29) between BCS and day (Table Prepartum and postpartum periods were analyzed separately. 3 Full models contained the fixed effects of BCS (FAT, MOD, THIN), day (−21 to −1 in the prepartum and 1 to 28 in the postpartum), heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS and day, BCS and heat stress, BCS and DCAD, and BCS and FA.Postpartum models also contained the fixed effect of morbidity (yes, no) and the interaction between BCS and morbidity.Posthoc P-values were adjusted for multiple comparisons.

a-c
Within response, columns with different superscripts differ (P ≤ 0.05).

A-B
Within response, columns with different superscripts tended to differ (0.05 < P ≤ 0.10).
1, Figure 3).The main effect showed that FAT cows had lower EB than MOD and THIN (P < 0.01) cows, and MOD cows had lower EB than THIN cows (P < 0.01).

Body Weight, BWC, and BCSC
There was a main effect (P < 0.01) of BCS 21 d before calving for BW at d-21 (Table 1).The main effect showed that FAT cows had greater BW than MOD (P < 0.01) and THIN (P < 0.01) cows, and MOD cows had greater BW than THIN cows (P < 0.01).An interaction (P < 0.01) was observed between BCS 21 d before calving and day for BW prepartum (Supplemental Figure 4).The interaction showed that FAT cows had greater BW than MOD and THIN cows from d-21 to d-1, and MOD cows had greater BW than THIN cows from d-21 to d-1.There was a main effect (P < 0.01) of BCS 21 d before calving for BW at calv-ing (Table 1).The main effect showed that FAT cows had greater BW than MOD (P < 0.01) and THIN (P < 0.01) cows, and MOD cows had greater BW than THIN cows (P < 0.01).There was a main effect (P < 0.01) of BCS 21 d before calving for BW postpartum, but no interaction was observed (P = 0.23) between BCS and day (Supplemental Figure 4).The main effect showed that FAT cows had greater BW than MOD (P < 0.01) and THIN (P < 0.01) cows, and MOD cows had greater BW than THIN cows (P < 0.01).There was a main effect (P < 0.01) of BCS 21 d before calving for BW at d 28 (Table 1).The main effect showed that FAT cows had greater BW than MOD (P < 0.01) and THIN (P < 0.01) cows, and MOD cows had greater BW than THIN cows (P < 0.01).
There was a main effect (P < 0.01) of BCS 21 d before calving for BWC prepartum (Table 1, Supplemental Figure 5).The main effect showed that FAT cows had greater BW loss than MOD (P < 0.01) and Cows were categorized according to their BCS at 21 d before calving as FAT (circles; BCS ≥4.00; n = 83), MOD (squares; BCS 3.25 to 3.75; n = 287), and THIN (hexagons; BCS ≤3.00; n = 57).Prepartum (d-21 to d-1 relative to calving) and postpartum (d1 to d28 relative to calving) periods were analyzed separately.For prepartum DMI, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.35).For postpartum DMI, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.34).Both models were adjusted for the effects of heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS and heat stress, BCS and DCAD, and BCS and FA.Postpartum models were also adjusted for the effect of morbidity (yes, no) and the interaction between BCS and morbidity.Error bars represent SEM.Symbols next to 'Pre' or 'Post' represent differences (P ≤ 0.05) between groups in the posthoc comparisons, + represents difference (P ≤ 0.05) between THIN and MOD, * represents difference (P ≤ 0.05) between THIN and FAT, and × represents difference (P ≤ 0.05) between MOD and FAT.P-values were adjusted for multiple comparisons.THIN (P < 0.01) cows.No difference in BWC prepartum was observed between MOD and THIN cows (P = 0.80).There was a tendency for a main effect (P = 0.07) of BCS 21 d before calving for BWC postpartum (Table 1, Supplemental Figure 5).The tendency for the main effect showed that FAT cows tended to lose more BW postpartum than THIN cows (P = 0.08).No difference in BWC postpartum was observed neither between FAT and MOD (P = 0.13), nor between MOD and THIN (P = 0.71) cows.
There was a main effect (P < 0.01) of BCS 21 d before calving for BCSC prepartum (Table 1).The main effect showed that FAT cows had greater BCS loss than MOD (P < 0.01) and THIN (P < 0.01) cows, and MOD cows had greater BCS loss than THIN cows (P < 0.01).A Sankey diagram of the differences in BCSC between FAT, MOD, and THIN can be found in Supplemental Figure 6.

Milk Yield
There was a main effect (P < 0.01) of BCS 21 d before calving for milk yield, but no interaction was observed (P = 0.30) between BCS and day (Table 1).The main effect showed that MOD cows produced more milk than THIN cows (P < 0.01).No difference in milk yield was observed neither between FAT and MOD (P = 0.55), nor between FAT and THIN (P = 0.18) cows.A quadratic effect (P = 0.02) of BCS at 21 d before calving on daily milk yield was observed (Figure 4A).Increasing BCS from 2.5 to 3.5 was associated with a predicted increase in daily milk yield of 6 kg.On the other hand, increasing BCS from 3.5 to 4.5 was associated with a predicted decrease in daily milk yield of 4.4 kg.A quadratic effect (P = 0.03) of BCS at 21 d before calving on cumulative milk in 28d was observed (Figure 4B).Increasing BCS from 2.5 to 3.5 was associated with a predicted increase of 146 kg cumulative milk in  and postpartum (d1 to d28 relative to calving) periods were analyzed separately.For prepartum DMI%BW, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.57).For postpartum DMI%BW, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.55).Both models were adjusted for the effects of heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS and heat stress, BCS and DCAD, and BCS and FA.Postpartum models were also adjusted for the effect of morbidity (yes, no) and the interaction between BCS and morbidity.Error bars represent SEM.Symbols next to 'Pre' or 'Post' represent differences (P ≤ 0.05) between groups in the posthoc comparisons, + represents difference (P ≤ 0.05) between THIN and MOD, * represents difference (P ≤ 0.05) between THIN and FAT, and × represents difference (P ≤ 0.05) between MOD and FAT.P-values were adjusted for multiple comparisons.28d.On the other hand, increasing BCS from 3.5 to 4.5 was associated with a predicted decrease of 116 kg cumulative milk in 28d.
A quadratic effect (P = 0.01) of prepartum BCSC on daily milk yield was observed (Supplemental Figure 7A).Increasing prepartum BCSC from −1.5 to 0 was associated with a predicted increase in daily milk yield of 26 kg.On the other hand, increasing BCSC from 0 to 0.75 was associated with a predicted decrease in daily milk yield of 6 kg.A quadratic effect (P = 0.04) of prepartum BCSC on cumulative milk in 28 d was observed (Supplemental Figure 7B).Increasing BCSC from −1.5 to 0 was associated with a predicted increase of 551 kg cumulative milk in 28d.On the other hand, increasing BCSC from 0 to 0.75 was associated with a predicted decrease of 109 kg cumulative milk in 28 d.

DISCUSSION
The objective of this study was to investigate the association between BCS at 21 d before calving and peripartum DMI, EB, and milk yield.We observed that the higher the BCS at 21 d before calving, the lower DMI and EB prepartum and postpartum.A quadratic association between BCS at 21 d before calving and milk yield showed that milk yield increased as BCS increased from 2.5 to 3.5 and decreased as BCS increased from 3.5 to 4.5.
The current study demonstrated that although all cows experienced a decline in prepartum DMI, FAT cows had the lowest prepartum DMI.It is not clear why fatter cows ate less than their herdmates.Factors such as reduced rumen capacity because of excessive abdominal fat accumulation (Szura et al., 2020), adipose tissue control of satiety through leptin signaling (Schwartz et al., 1996;Riosa et al., 2022;Bradford et al., 2006), and satiety induced by persistent low-grade inflammation caused by excessive adiposity (Contreras et al., 2015;Kuroda and Sakaue, 2017;Kuhla, 2020) could have contributed to the observed reduction in DMI.Independent of the cause, the marked reduction in DMI was sufficient to result in a more pronounced negative EB. Indeed, FAT cows not only had the lowest EB but also were in negative EB since −21 d before calving.These results were further supported by the changes in BCS and BW, where FAT cows lost 0.30 units of BCS and 31 kg of BW during the prepartum period, indicating that FAT cows were in a more pronounced negative EB and were mobilizing more adipose tissue than their herdmates.
The lesser DMI and EB observed in FAT cows prepartum was maintained during postpartum.Furthermore, although FAT cows lost more BCS and BW than their herdmates during prepartum, they still had a higher BCS and were heavier at calving.Therefore, it is possible that the adipose-related metabolic or inflammatory regulation of feed intake may contribute to the decrease in postpartum DMI.Additionally, regardless of what led to the reduction in prepartum DMI, it is possible that the inflammatory state prepartum was sustained postpartum, leading to a reduction in DMI postpartum.Exacerbated mobilization of adipose tissue is associated with greater cellular death (Scalia et al., 2006;Casaro et al., 2023), which triggers systemic inflammation (Roh and Sohn, 2018;Gong et al., 2020) and has been proposed as the link between lipid mobilization and systemic inflammation in periparturient dairy cows (Casaro et al., 2023).Recently, we showed that heavier cows lost more weight prepartum and had higher blood fatty acids, and that blood fatty acids were positively correlated with cell death, and IL-1β concentration at calving (Casaro et al., 2023).Systemic inflammation is characterized by greater peripheral blood pro-inflammatory cytokines such as IL-1β, which as previously mentioned, can exacerbate the depression in feed intake (Kuhla, 2020).Therefore, it is possible that the more pronounced negative EB observed in FAT cows during the prepartum period led to persistent systemic inflammation, triggering a vicious cycle of DMI depression, lipid mobilization, and inflammation, which could help explain the reduced DMI and lower EB observed in FAT cows postpartum.
In addition to elucidating if overconditioned cows experienced a decrease in peripartum DMI and EB, the present study aimed to determine what was the ideal prepartum BCS to maximize milk production.We observed a quadratic association between BCS at 21 d before calving and milk yield, which showed that milk yield increased as BCS increased from 2.5 to 3.5 and decreased as BCS increased from 3.5 to 4.5.The lower milk yield in thinner cows indicate that these cows may not have had enough body energy reserves to maximize milk production.In addition to supplying long-chain fatty acids for the production of milk fat, released NEFA from adipose tissue during early lactation offers an energy source for non-mammary tissues, thereby preserving glucose for mammary lactose synthesis and enhancing milk production (Bauman and Currie, 1980).Roche et al. (2007) also reported a quadratic association between BCS and milk yield, indicating that as BCS increased from 2.5 to 3.5 60 d cumulative milk yield increased by 264 kg.It seems that despite THIN cows having the greatest DMI and EB, their underconditioned state may be limiting lactation performance.The lower milk yield in fatter cows may be simply due to a reduction in nutrient availability to the mammary gland due to the lesser DMI (Hristov et al., 2005).These findings indicate that a prepartum BCS ranging from 3.25 to 3.75 is optimal for maximizing milk production.
The main limitation of the current study is the lack of information about reproductive performance.Furthermore, due to the observational nature of the current study, it is not possible to establish a cause-and-effect relationship between BCS at 21 d before calving and DMI, EB and milk yield.Therefore, controlled trials need to be conducted to confirm our findings.

CONCLUSION
This retrospective observational study provides insights in the relationship between BCS 21 d before calving and prepartum and postpartum DMI, EB, and postpartum milk yield.The present study demonstrated that FAT cows had lesser DMI and were in a lower EB both prepartum and postpartum when compared with MOD and THIN cows.Furthermore, DMI was also lesser for MOD cows compared with THIN cows prepartum, but not postpartum, but EB was lesser for MOD cows compared with THIN cows both prepartum and postpartum.Increasing BCS at 21 d before calving from 2.5 to 3.5 was associated with an increase in milk yield, indicating that thinner cows may not have enough body energy reserves to maximize milk production, whereas increasing BCS from 3.5 to 4.5 was associated with a decrease in milk yield, indicating that fatter cows may have excessive losses of body reserves and suppression of DMI, which may impair milk production.Our findings indicate that having a moderated BCS prepartum is ideal for ensuring a successful lactation.
evaluated the effects Casaro et al.: ASSOCIATION BETWEEN BODY CONDITION SCORE AND INTAKE of supplementing diets containing low amounts of fatty acids with either saturated free fatty acids (SFA, n = 14) or Ca salts containing essential fatty acids (EFA, n = 17) against no fat supplementation (control, n = 21) on lactation performance.Cows that were fed EFA showed lesser DMI than control and EFA, therefore, a variable named 'FA' was created to account for the differences in DMI because of EFA supplementation.Lopera et al. (2018) evaluated the effects of feeding diets with 2 levels of negative dietary cation-anion differences (DCAD) during the last 42 or 21 d of gestation on performance.Reducing the level of DCAD from −70 to −180 mEq/kgDM decreased prepartum DMI.Furthermore, Martinez et al. ( Casaro et al.: ASSOCIATION BETWEEN BODY CONDITION SCORE AND INTAKE Casaro et al.: ASSOCIATION BETWEEN BODY CONDITION SCORE AND INTAKEcomparisons with Tukey adjustments were performed between the groups (FAT vs. MOD, FAT vs. THIN, MOD vs. THIN).For prepartum BCSC, prepartum and postpartum BWC, BW at d −21, BW at calving, and BW at 28 d, the models contained the fixed effects of BCS 21 d before calving (FAT, MOD, THIN), heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS 21 d before calving and day, BCS 21 d before calving and heat stress, BCS 21 d before calving and DCAD, and BCS 21 d before calving and FA.Postpartum models also contained the fixed effect of morbidity (yes, no) and the interaction between BCS 21 d before calving and morbidity.When morbidity, season, DCAD, FA or their respective interactions had a P > 0.10, the effect was removed from the model and the model was rerun.

Figure 1 .
Figure 1.Pre and postpartum dry matter intake (DMI) in kilograms per day according to body condition score (BCS) in multiparous cows.Cows were categorized according to their BCS at 21 d before calving as FAT (circles; BCS ≥4.00; n = 83), MOD (squares; BCS 3.25 to 3.75; n = 287), and THIN (hexagons; BCS ≤3.00; n = 57).Prepartum (d-21 to d-1 relative to calving) and postpartum (d1 to d28 relative to calving) periods were analyzed separately.For prepartum DMI, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.35).For postpartum DMI, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.34).Both models were adjusted for the effects of heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS and heat stress, BCS and DCAD, and BCS and FA.Postpartum models were also adjusted for the effect of morbidity (yes, no) and the interaction between BCS and morbidity.Error bars represent SEM.Symbols next to 'Pre' or 'Post' represent differences (P ≤ 0.05) between groups in the posthoc comparisons, + represents difference (P ≤ 0.05) between THIN and MOD, * represents difference (P ≤ 0.05) between THIN and FAT, and × represents difference (P ≤ 0.05) between MOD and FAT.P-values were adjusted for multiple comparisons.Figure created using GraphPad Prism version 8.4.3 for Windows, GraphPad Software, San Diego, California, USA.
Figure 1.Pre and postpartum dry matter intake (DMI) in kilograms per day according to body condition score (BCS) in multiparous cows.Cows were categorized according to their BCS at 21 d before calving as FAT (circles; BCS ≥4.00; n = 83), MOD (squares; BCS 3.25 to 3.75; n = 287), and THIN (hexagons; BCS ≤3.00; n = 57).Prepartum (d-21 to d-1 relative to calving) and postpartum (d1 to d28 relative to calving) periods were analyzed separately.For prepartum DMI, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.35).For postpartum DMI, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.34).Both models were adjusted for the effects of heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS and heat stress, BCS and DCAD, and BCS and FA.Postpartum models were also adjusted for the effect of morbidity (yes, no) and the interaction between BCS and morbidity.Error bars represent SEM.Symbols next to 'Pre' or 'Post' represent differences (P ≤ 0.05) between groups in the posthoc comparisons, + represents difference (P ≤ 0.05) between THIN and MOD, * represents difference (P ≤ 0.05) between THIN and FAT, and × represents difference (P ≤ 0.05) between MOD and FAT.P-values were adjusted for multiple comparisons.Figure created using GraphPad Prism version 8.4.3 for Windows, GraphPad Software, San Diego, California, USA.

Figure 2 .
Figure 2. Pre and postpartum dry matter intake as a percentage of body weight per day (DMI%BW) according to body condition score (BCS) in multiparous cows.Cows were categorized according to their BCS at 21 d before calving as FAT (circles; BCS ≥4.00; n = 83), MOD(squares; BCS 3.25 to 3.75; n = 287), and THIN (hexagons; BCS ≤3.00; n = 57).Prepartum (d-21 to d-1 relative to calving) and postpartum (d1 to d28 relative to calving) periods were analyzed separately.For prepartum DMI%BW, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.57).For postpartum DMI%BW, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.55).Both models were adjusted for the effects of heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS and heat stress, BCS and DCAD, and BCS and FA.Postpartum models were also adjusted for the effect of morbidity (yes, no) and the interaction between BCS and morbidity.Error bars represent SEM.Symbols next to 'Pre' or 'Post' represent differences (P ≤ 0.05) between groups in the posthoc comparisons, + represents difference (P ≤ 0.05) between THIN and MOD, * represents difference (P ≤ 0.05) between THIN and FAT, and × represents difference (P ≤ 0.05) between MOD and FAT.P-values were adjusted for multiple comparisons.Figure created using GraphPad Prism version 8.4.3 for Windows, GraphPad Software, San Diego, California, USA.
Figure 2. Pre and postpartum dry matter intake as a percentage of body weight per day (DMI%BW) according to body condition score (BCS) in multiparous cows.Cows were categorized according to their BCS at 21 d before calving as FAT (circles; BCS ≥4.00; n = 83), MOD(squares; BCS 3.25 to 3.75; n = 287), and THIN (hexagons; BCS ≤3.00; n = 57).Prepartum (d-21 to d-1 relative to calving) and postpartum (d1 to d28 relative to calving) periods were analyzed separately.For prepartum DMI%BW, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.57).For postpartum DMI%BW, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.55).Both models were adjusted for the effects of heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS and heat stress, BCS and DCAD, and BCS and FA.Postpartum models were also adjusted for the effect of morbidity (yes, no) and the interaction between BCS and morbidity.Error bars represent SEM.Symbols next to 'Pre' or 'Post' represent differences (P ≤ 0.05) between groups in the posthoc comparisons, + represents difference (P ≤ 0.05) between THIN and MOD, * represents difference (P ≤ 0.05) between THIN and FAT, and × represents difference (P ≤ 0.05) between MOD and FAT.P-values were adjusted for multiple comparisons.Figure created using GraphPad Prism version 8.4.3 for Windows, GraphPad Software, San Diego, California, USA.
Figure 2. Pre and postpartum dry matter intake as a percentage of body weight per day (DMI%BW) according to body condition score (BCS) in multiparous cows.Cows were categorized according to their BCS at 21 d before calving as FAT (circles; BCS ≥4.00; n = 83), MOD(squares; BCS 3.25 to 3.75; n = 287), and THIN (hexagons; BCS ≤3.00; n = 57).Prepartum (d-21 to d-1 relative to calving) and postpartum (d1 to d28 relative to calving) periods were analyzed separately.For prepartum DMI%BW, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.57).For postpartum DMI%BW, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.55).Both models were adjusted for the effects of heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS and heat stress, BCS and DCAD, and BCS and FA.Postpartum models were also adjusted for the effect of morbidity (yes, no) and the interaction between BCS and morbidity.Error bars represent SEM.Symbols next to 'Pre' or 'Post' represent differences (P ≤ 0.05) between groups in the posthoc comparisons, + represents difference (P ≤ 0.05) between THIN and MOD, * represents difference (P ≤ 0.05) between THIN and FAT, and × represents difference (P ≤ 0.05) between MOD and FAT.P-values were adjusted for multiple comparisons.Figure created using GraphPad Prism version 8.4.3 for Windows, GraphPad Software, San Diego, California, USA.

Figure 3 .
Figure 3. Pre and postpartum energy balance in Mcal per day (EB) according to body condition score (BCS) in multiparous cows.Cows were categorized according to their BCS at 21 d before calving as FAT (circles; BCS ≥4.00; n = 83), MOD (squares; BCS 3.25 to 3.75; n = 287), and THIN (hexagons; BCS ≤3.00; n = 57).Prepartum (d-21 to d-1 relative to calving) and postpartum (d1 to d28 relative to calving) periods were analyzed separately.For prepartum EB, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.40).For postpartum EB, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.29).Both models were adjusted for the effects of heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS and heat stress, BCS and DCAD, and BCS and FA.Postpartum models were also adjusted for the effect of morbidity (yes, no) and the interaction between BCS and morbidity.Error bars represent SEM.Symbols next to 'Pre' or 'Post' represent differences (P ≤ 0.05) between groups in the posthoc comparisons, + represents difference (P ≤ 0.05) between THIN and MOD, * represents difference (P ≤ 0.05) between THIN and FAT, and × represents difference (P ≤ 0.05) between MOD and FAT.P-values were adjusted for multiple comparisons.Figure created using GraphPad Prism version 8.4.3 for Windows, GraphPad Software, San Diego, California, USA.
Figure 3. Pre and postpartum energy balance in Mcal per day (EB) according to body condition score (BCS) in multiparous cows.Cows were categorized according to their BCS at 21 d before calving as FAT (circles; BCS ≥4.00; n = 83), MOD (squares; BCS 3.25 to 3.75; n = 287), and THIN (hexagons; BCS ≤3.00; n = 57).Prepartum (d-21 to d-1 relative to calving) and postpartum (d1 to d28 relative to calving) periods were analyzed separately.For prepartum EB, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.40).For postpartum EB, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.29).Both models were adjusted for the effects of heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS and heat stress, BCS and DCAD, and BCS and FA.Postpartum models were also adjusted for the effect of morbidity (yes, no) and the interaction between BCS and morbidity.Error bars represent SEM.Symbols next to 'Pre' or 'Post' represent differences (P ≤ 0.05) between groups in the posthoc comparisons, + represents difference (P ≤ 0.05) between THIN and MOD, * represents difference (P ≤ 0.05) between THIN and FAT, and × represents difference (P ≤ 0.05) between MOD and FAT.P-values were adjusted for multiple comparisons.Figure created using GraphPad Prism version 8.4.3 for Windows, GraphPad Software, San Diego, California, USA.
Figure 3. Pre and postpartum energy balance in Mcal per day (EB) according to body condition score (BCS) in multiparous cows.Cows were categorized according to their BCS at 21 d before calving as FAT (circles; BCS ≥4.00; n = 83), MOD (squares; BCS 3.25 to 3.75; n = 287), and THIN (hexagons; BCS ≤3.00; n = 57).Prepartum (d-21 to d-1 relative to calving) and postpartum (d1 to d28 relative to calving) periods were analyzed separately.For prepartum EB, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.40).For postpartum EB, effect of BCS (P < 0.01), day (P < 0.01), and interaction between BCS and day (P = 0.29).Both models were adjusted for the effects of heat stress (CL, HT, HTEV), DCAD (POS, PNEG, FNEG), FA (Control, EFA, SFA), season (winter, spring, summer, autumn), and the interactions between BCS and heat stress, BCS and DCAD, and BCS and FA.Postpartum models were also adjusted for the effect of morbidity (yes, no) and the interaction between BCS and morbidity.Error bars represent SEM.Symbols next to 'Pre' or 'Post' represent differences (P ≤ 0.05) between groups in the posthoc comparisons, + represents difference (P ≤ 0.05) between THIN and MOD, * represents difference (P ≤ 0.05) between THIN and FAT, and × represents difference (P ≤ 0.05) between MOD and FAT.P-values were adjusted for multiple comparisons.Figure created using GraphPad Prism version 8.4.3 for Windows, GraphPad Software, San Diego, California, USA.
Figure 4. Daily milk yield (A) and cumulative milk in 28 d postpartum (B) according to body condition score (BCS) 21 d before calving.Panel A: BCS as linear covariate (P = 0.02), BCS as quadratic covariate (P = 0.02).Panel B: BCS as linear covariate (P = 0.03), BCS as quadratic covariate (P = 0.03).The association between BCS and milk yield was analyzed fitting models with BCS as linear and quadratic covariates and controlling for the effects of heat stress, DCAD, FA, season, morbidity, and the interactions between BCS and heat stress, BCS and DCAD, BCS and FA, and BCS and morbidity.Statistical significance was considered at P ≤ 0.05. Figure created using GraphPad Prism version 8.4.3 for Windows, GraphPad Software, San Diego, California, USA.
Casaro et al.: ASSOCIATION BETWEEN BODY CONDITION SCORE AND INTAKE

Table 1 .
Casaro et al.: ASSOCIATION BETWEEN BODY CONDITION SCORE AND INTAKE The LSM and SEM for the associations between body condition score (BCS) 1 at 21 d before calving and prepartum (d-21 to d-1) and postpartum (d1 to d28) dry matter intake (DMI), net energy for lactation consumed (NE L C), NE L required (NE L R), energy balance (EB), body weight (BW), body weight change (BWC), prepartum body condition score change (BCSC), BCS (BCS pre) and BW (BW pre) at −21 d before calving, BCS (BCS calving) and BW (BW calving) at calving, BW at 28 DIM (BW 28 d), and daily milk yield (Milk) 2