Effect of dietary protein intake on energy utilization and feed efficiency of lactating sows

Abstract

The objective of the current study was to quantify loss of energy in feces, urine, heat, and milk, to evaluate feed efficiency and to evaluate optimal ratio of dietary CP to energy for lactating sows fed increasing dietary CP. A total of 72 sows were included in the experiment from day 2 after parturition until weaning at day 28. Sows were allocated to 6 dietary treatments formulated to be isocaloric (9.8 MJ NE/kg) and increasing standardized ileal digestible (SID) CP (11.8, 12.8, 13.4, 14.0, 14.7, and 15.6% SID CP). Sows were weighed and back fat scanned within 2 d after farrowing, at days 18 ± 3 and 28 ± 3. Litters were standardized to 14 piglets within 2 d after farrowing and weighed at day 1 or 2 and at days 11, 18, and 28 (within ± 3 d). Feed intake (feed supply minus residue) was registered, and milk, urine, and fecal samples were collected at days 4, 11, and 18 (within ± 3 d). Sow milk yield was estimated from litter gain and litter size, and sow heat production was calculated factorially. On days 4 and 18 (±3 d), sows were enriched with D₂O (deuterated water) to estimate body protein and fat pool size. Overall, sow BW loss, back fat loss, fat and protein mobilization, litter size, and piglet performance were not affected by diets, except for sows fed treatment 5, which had lower ADFI and lower milk production, and a tendency to lower piglet ADG compared with the remaining treatment groups (P < 0.01, P = 0.03, P =0.08, respectively). Relative to GE intake, the energy excreted in urine increased from 3.3% to 5.3% (P < 0.001), whereas energy lost as heat increased numerically from 54.5% to 59.0% with increasing dietary CP. The feed efficiency as evaluated by NE corrected for body mobilization peaked when sows were fed at their requirement (treatment 2; 12.8% SID CP; P = 0.01), whereas the feed efficiency was 1% lower for treatment 1, whereas it was 3% to 6% lower for treatments 3 through 6. In conclusion, energy loss in urine and likely also energy lost as heat increase if the dietary protein to energy ratio is unbalanced, and evaluating feed efficiency of lactating sows by correcting for body mobilization seems to be a promising approach to improve sow feeding in the future.

INTRODUCTION

Improved feed efficiency is important because energy is a critical component of the total costs of pig production. Each percentage unit decrease in feed efficiency for lactating sows corresponds to a loss of €3.4 million annually to Danish swine producers (Helverskov, 2017; DanmarksStatistik, 2018). Dietary energy utilization is dependent on the relative concentration of fiber, fat, ash, and protein (Noblet and Perez, 1993). In animals fed dietary CP above requirements, AA are oxidized leading to a loss of energy in urine and additional heat production (Just, 1982). Few studies are available on utilization of dietary energy in lactating sows (Verstegen et al., 1985; Noblet and Etienne, 1987; Theil et al., 2004), and none have reported the impact of increasing dietary CP on energy utilization.

Lactating sows utilize as much as 70% of dietary CP for milk protein synthesis (Pedersen et al., 2016). Insufficient dietary CP suppresses milk yield and may cause substantial body weight loss (Strathe et al., 2017a). Excess dietary CP reduces energy utilization and feed efficiency of growing pigs (Just, 1982), but it remains unclear how lactating sows respond to diets with increasing protein to energy ratios. Furthermore, when energy demand for lactation exceeds energy intake, lactating sows mobilize body fat and protein, and energy mobilization is substantially higher in modern high-yielding sows nursing 13 to 14 piglets (Strathe et al., 2017b) than those nursing 10 suckling piglets (Theil et al., 2004).

The objectives were 1) to quantify energy utilization in lactating sows fed diets with increasing standardized ileal digestible (SID) CP:energy, 2) to determine feed efficiency when energy contribution from body mobilization is taken into account, and 3) to evaluate the optimal dietary SID CP:energy using the calculated feed efficiency. The hypothesis was that feed efficiency is maximized when the dietary ratio between SID CP and energy is optimal, because energy lost through urine and heat is minimized.

MATERIALS AND METHODS

The housing and rearing of sows and litters were in compliance with Danish laws and regulations for the humane care and use of animals in research (The Danish Ministry of Justice, 1995). The protocols used in the current experiment were reviewed and approved by The Danish Animal Experimentation Inspectorate.

Animals and Housing

A total of 72 cross-bred sows (Danish Landrace × Danish Yorkshire; DanBred, Herlev, Denmark) bred to Duroc (Ornestation Mors, Redsted, Denmark) were used. The experiment was conducted in a commercial Danish sow herd, from February to April 2016 using 12 first parity and 60 multiparous sows. The study was carried out in 4 blocks, with 18 sows per block. One week prior to expected parturition, sows were moved to the farrowing unit and allocated to 1 of 6 treatments according to parity.

Sows were individually housed in farrowing crates (2.7 × 1.8 m) with slatted floors in 4 different rooms with environmental temperature maintained around 20 °C. Within a block, all sows were housed in the same room. Within 48 h postpartum, each litter was standardized to 14 medium- to large-sized piglets (weighing above approximately 900 g). Piglets had access to a rubber mattress-covered area and a heating lamp (VengSystem A/S, Roslev, Denmark). The temperature in the covered piglet area decreased automatically from 34 °C around parturition to 22 °C 15 d after parturition. Sawdust was provided in the covered piglet area before parturition, and a cereal-based diet (10.6% protein kg/DM; Vestjyllands Andel, Ringkøbing, Denmark) was supplied to piglets, beginning at approximately day 14 until weaning at day 28.

Dietary Treatments and Feeding

Six treatments were formulated to contain the same calculated NE per kg feed with decreasing carbohydrate and increasing CP content. All diets contained 40% barley and 6% fiber mix. The decrease in carbohydrate and increase in CP concentration from treatments 1 through 6 were achieved by gradually decreasing the inclusion rate of a low CP and high starch supplementary feed mix (FM1) at the expense of gradually increasing a high CP and low starch supplementary feed mix (FM2; Table 1). The FM1 and FM2 were formulated with cereal grains (barley and wheat) and soybean meal, whereby FM1 and FM2 contained a different ratio of cereal grains to soybean meal.

Table 1.

Ingredient and chemical composition of the supplements and fiber mix, and chemical composition of barley (as fed)

Item	Supplement 1	Supplement 2	Barley	Fiber mix
Ingredient, [%]
Barley	26.0	–	100	–
Wheat	33.8	38.5	–	–
Wheat bran	3.00	–	–	12.0
Soybean meal	23.0	50.0	–	–
Soy hulls	–	–	–	12.0
Sugar beet pulp	–	–	–	72.0
Sugar beet molasses	–	–	–	2.00
Soy oil	5.10	5.00	–	–
Leci E1	–	–	–	2.00
Premix2	4.50	4.50	–	–
Calcium carbonate	0.90	0.84	–	–
Sodium chloride	0.95	0.95	–	–
Monocalcium phosphate	0.33	0.10	–	–
l-Lys	1.73	–	–	–
dl-Met	0.25	0.07	–	–
l-Thr	0.33	–	–	–
l-Trp	0.12	–	–	–
Phytase3	0.04	0.04	–	–
Chemical composition
GE, MJ/kg	16.2	16.6	16.0	17.0
DM, %	85.8	86.2	85.3	86.3
CP, %	19.3	26.8	9.0	9.8
Fat, %	6.5	6.3	2.6	3.5
Starch, %	35.3	25.2	52.7	7.0
Dietary fiber, %	13.7	13.5	16.6	55.9
Lignin, %	2.2	2.0	3.3	3.9
Ash, %	7.6	8.0	2.1	3.7
Lys, %	1.43	1.53	0.36	0.58
Met, %	0.42	0.39	0.15	0.14
Thr, %	0.93	0.99	0.31	0.40
Trp, %	0.26	0.36	0.11	0.12

¹Phospolipids, FFA and triglycerides from rape seed oil.

²Supplied per kilogram of diet: Retinol 8694 IU, Cholecalciferol 1242 IU, α-tocopherol 79.11 mg, phylloquinone 4.14 mg, thiamin 2.17 mg, cyanocobalamin 0.022 mg, riboflavin 5.43 mg, pyridoxine 3.26 mg, biotin 0.43 mg, D-pantothenic acid 16.3 mg, folic acid 1.63 mg, niacin 21.74 mg, 15.57 mg Cu as CuSO₄ ∙ 5H₂O, 103.5 mg Zn as ZnO, 86.94 mg Fe as FeSO4 ∙ 7 H2O, 2.07 mg I as CaI, 43.47 mg Mn as MnO, 0.31 mg Se as Na2SeO3.

³2000 U/kg.

Crystalline l-Lys, dl-Met, l-Thr, and l-Trp were added to FM1 and FM2, to meet the levels recommended for Danish sows nursing 14 piglets for 28 d while having an average feed intake of 6.5 kg/d (Tybirk et al., 2015). Furthermore, additional dl-Met was added to FM1 to meet the requirement for Cys (Ball et al., 2006). The FM1 and FM2 had the same content of supplemented fat, whereas the content of wheat and wheat bran was allowed to vary to obtain constant estimated NE (MJ/kg) concentrations in FM1 and FM2. The 6 dietary treatments were therefore formulated to be constant in NE and SID Lys, whereas SID Met, Met + Cys, Thr, and Trp met or exceeded the recommended levels (Table 2; Tybirk et al., 2015). According to Tybirk et al. (2015) and Strathe et al. (2017a), the optimal dietary CP concentration for lactating sows is equivalent to 13.1% SID CP. In the present study, the aim was to reach 13.1% SID CP between treatments 3 and 4, but in contrast to Strathe et al. (2017a), crystalline l-Lys, dl-Met, l-Thr, and l-Trp were added to meet the recommended levels for these 4 AA. The diets were formulated to meet the Danish recommendations for all other nutrients for lactating sows (Tybirk et al., 2015).

Table 2.

Ingredient and chemical composition (as fed) of experimental diets

Item	Treatment
	1	2	3	4	5	6
Ingredients, %
Barley	54.0	50.5	48.1	46.0	43.5	40.0
Wheat	18.3	18.9	19.3	19.7	20.2	20.8
Wheat bran	2.34	1.94	1.66	1.41	1.13	0.72
Soybean meal	12.4	16.1	18.6	20.8	23.4	27.0
Soy hulls	0.72	0.72	0.72	0.72	0.72	0.72
Sugar beet pulp	4.32	4.32	4.32	4.32	4.32	4.32
Sugar beet molasses	0.12	0.12	0.12	0.12	0.12	0.12
Soy oil	2.75	2.74	2.73	2.72	2.71	2.70
Leci E1	0.12	0.12	0.12	0.12	0.12	0.12
Premix2	2.43	2.43	2.43	2.43	2.43	2.43
Calcium carbonate	0.49	0.48	0.47	0.47	0.46	0.45
Sodium chloride	0.51	0.51	0.51	0.51	0.51	0.51
Monocalcium phosphate	0.18	0.15	0.13	0.11	0.09	0.05
l-Lys	0.93	0.70	0.54	0.40	0.23	0
dl-Met	0.14	0.11	0.09	0.08	0.06	0.04
l-Thr	0.18	0.13	0.10	0.08	0.04	0
l-Trp	0.06	0.05	0.04	0.03	0.02	0
Phytase3	0.02	0.02	0.02	0.02	0.02	0.02
Chemical composition4
DM, %	85.6	85.7	85.7	85.7	85.8	85.8
GE, MJ/kg	16.1	16.2	16.2	16.3	16.3	16.4
NE, MJ/kg	9.8	9.8	9.8	9.8	9.8	9.8
CP, %	14.6	15.6	16.3	16.9	17.6	18.6
SID5 CP, %	11.8	12.8	13.4	14.0	14.7	15.6
Fat, %	4.7	4.7	4.7	4.7	4.7	4.6
Ash, %	5.4	5.3	5.3	5.2	5.2	5.1
Starch, %	40.5	39.2	38.2	37.4	36.5	35.1
Dietary fiber, %	17.4	17.4	17.3	17.3	17.3	17.3
Lignin, %	2.7	2.7	2.7	2.7	2.6	2.6
Lys, %	0.95(0.83)5	0.97(0.83)	0.97(0.84)	0.98(0.84)	0.99(0.84)	1.01(0.85)
Met, %	0.30(0.27)	0.29(0.27)	0.29(0.26)	0.29(0.26)	0.28(0.25)	0.28(0.25)
Thr, %	0.65(0.54)	0.66(0.55)	0.66(0.55)	0.67(0.55)	0.68(0.56)	0.68(0.56)
Trp, %	0.19(0.16)	0.21(0.18)	0.22(0.18)	0.22(0.19)	0.23(0.20)	0.24(0.21)

¹Phospolipids, FFA and triglycerides from rape seed oil.

²Supplied per kilogram of diet: retinol 8694 IU, cholecalciferol 1242 IU, α-tocopherol 79.11 mg, phylloquinone 4.14 mg, thiamin 2.17 mg, cyanocobalamin 0.022 mg, riboflavin 5.43 mg, pyridoxine 3.26 mg, biotin 0.43 mg, D-pantothenic acid 16.3 mg, folic acid 1.63 mg, niacin 21.74 mg, 15.57 mg Cu as CuSO₄ ∙ 5H₂O, 103.5 mg Zn as ZnO, 86.94 mg Fe as FeSO₄ ∙ 7 H₂O, 2.07 mg I as CaI, 43.47 mg Mn as MnO, 0.31 mg Se as Na₂SeO₃.

³2000 U/kg.

⁴Values in brackets are calculated SID values.

⁵SID = standardized ileal digestible, based on calculated values divided by the expected analyzed times analyzed values.

The FM1 and FM2 were produced by DLG (Tjele, Denmark) and the fiber mix, which mainly contained sugar beet pulp, soybean hulls, and wheat bran, was produced by Vestjyllands Andel (Ringkøbing, Denmark). At each meal and for each sow, the 4 dietary components (FM1, FM2, barley, and fiber mix) were weighed, mixed, and distributed automatically by a SpotMix air-assisted feeding system (Schauer Agrotronic, Prambachkirchen, Germany). Feed was mixed with water when distributed and sows had free access to water. First parity and multiparous sows followed 2 separate feeding curves. The feeding curve defined the maximum allowance and the feed supply was reduced according to appetite for individual sows to minimize feed leftovers. First parity sows were fed 2.38 kg feed from day 1 in lactation and increased daily by approximately 0.40 kg to reach 4.78 kg on day 7. From days 8 to 17, feed supply increased by approximately 0.31 kg per d, reaching 7.6 kg/d on day 17, and remained constant until weaning. Multiparous sows were fed 2.38 kg feed at day 1 of lactation and the supply was increased daily by approximately 0.48 kg to reach 5.26 kg on day 7. From days 8 to 17, feed allowance increased by approximately 0.36 kg daily and reached 8.56 kg/d on day 17 and remained constant until weaning. Sows were fed twice daily from days 1 to 9 in lactation, between 0730 and 0830 h and between 1400 and 1500 h. From day 10 until weaning, sows were fed 3 times daily between 0600 and 0900, 1230 and 1500, and 1900 and 2000 h.

Feed leftovers from sows were collected on specific days once a week (days 4, 11, and 18 ± 3) to determine intake of feed and nutrients. The feed leftovers were assumed to represent the same distribution of FM1, FM2, barley, and fiber mix as the supplied meal. To convert wet feed leftovers to as-fed basis, calibration curves were established for treatments 1 and 6, whereas values for treatment 2 through 5 were found by interpolation. On average, 1000 g of wet feed residue corresponded to 425-g feed.

Experimental Procedure

Body weight, back fat, and milk yield.

Sows were weighed and back fat was measured at litter equalization (within 2 d after farrowing), at days 18 ± 3 and 28 ± 3. Back fat was measured using a SonoGrader II (RENCO, MN), on the right side, over the last rib and 63 mm from the backbone (conventionally known as P2 measurement). Piglets were weighed at day 1 or 2, at days 11 ± 3, 18 ± 3, and 28 ± 3. Litter weight gain and litter size were used to estimate milk yield according to Hansen et al. (2012).

Urine, milk, and fecal samples.

At days 4 ± 3 (week 1), days 11 ± 3 (week 2), and days 18 ± 3 (week 3), sows were fitted with a urinary balloon catheter (Teleflex medical, Kamunting, Malaysia) for total collection of urine for a 6-h period following the morning feeding. Initially, the urine bladder was emptied, and the catheter was blocked by a stopper and urine was collected quantitatively in a bucket at 2, 4, and 6 h after feeding by removing the stopper. The collected urine was weighed and a pooled sample was stored at −20 °C until analysis. On these same days, a total of 50- to 60-mL milk were collected and filtered through gauze for debris before storing at −20 °C until analysis. For milk collection, piglets were removed and the sow administered with 2-mL oxytocin (Intervet Danmark A/S, Ballerup, Denmark) i.m. to induce milk let down. A fresh grab fecal sample was collected by rectal stimulation and stored at −20 °C until analysis.

Deuterium (D2O) enrichment. 

At days 4 and 18 ± 3, sows were administered i.m. in the neck (18G, 40-mm needle, 10-mL syringe) 0.0425 g per kg BW of a 40% solution of deuterium oxide (SIGMA-ALDRICH, MO) while the remaining 60% consist of saline (9-mg NaCl/mL; B. Braun Melsungen AG, Melsungen, Germany). A urine sample was drawn prior to enrichment to measure D₂O background level and another urine sample was collected 6 to 7 h post-D₂O administration to determine the D₂O equilibration, and these were used to assess the total D₂O space as described by Theil et al. (2002). Based on the measured D₂O space, total pool size of protein and fat at days 4 and 18 were calculated according to Rozeboom et al. (1994) and used to assess the daily mobilization of protein and fat from days 4 to 18.

General management and health.

Piglets were given an injection of iron (0.5 mL; Solofer Vet., Pharmacosmos A/S, DK-4300 Holbæk, Denmark) 3 to 4 d after parturition and were supplied iron in the water distributed by the drinking nipple (1%; Opti-Jern, R2 Agro-Nutriscan, DK-8722 Hedensted, Denmark) throughout lactation. Piglets were tail docked and males castrated 3 to 4 d after parturition, and castration was done under pain relief following i.m. administration (0.1 mL) of Melovem (Dopharma Research B.V., NL-4941 VX Raamsdonksveer, The Netherlands). Sows and piglets were monitored daily by farm staff for health condition and were treated when needed in compliance with accepted on-farm protocols.

Chemical Analyses

A sample of barley, FM1, FM2, and fiber mix was collected for chemical analysis. Barley, FM1, FM2, and fiber mix were analyzed in duplicate for GE, DM, ash, N, crude fat, total dietary fiber, starch, lignin, and AA. The mixed diets were analyzed in duplicate for DM, ash, N, and crude fat. Fecal samples were analyzed in duplicate for DM and lignin, and single determination of GE was performed. Prior to chemical analysis, feed samples were ground to 0.5 mm using an ultra-centrifugal mill (Model ZM200; RETSCH, Haan, Germany), and fecal samples were freeze dried and subsequently ground to 0.5 mm. Feeds and feces were analyzed for DM and ash by drying for 20 h at 103 °C in a forced air oven (for DM), followed by combustion at 525 °C for 6 h (for ash). Gross energy was determined using an adiabatic bomb calorimeter (Model 6300; Parr Instrument, Moline, IL), with benzoic acid used as a calibrating standard. Nitrogen was analyzed according to the Dumas method (Hansen, 1989) using a Vario Max CN Element analyzer (Elementar Analysensystem GmbH, Langenselbold, Germany), with aspartic acid used as a calibrating standard. Analyses for AA and crude fat were conducted by Eurofins Steins Laboratorium A/S (Vejen, Denmark) according to the Official Journal of the European Union (EU; 152/2009). Starch was analyzed by enzymatic colorimetry as described by Bach Knudsen (1997). Total dietary fiber including lignin was analyzed by enzymatic, chemical, and gravimetric determination of soluble and insoluble fibers according to Bach Knudsen (1997). Urine samples (as is) were analyzed for N using a Kjeldahl (KjelTec 2400, Hillerød, Denmark) according to the official method of AOAC (2012). Urine samples were analyzed after ultrafiltration for D₂O according to the method described for plasma by Theil et al. (2002). Sow milk was analyzed in triplicate for DM, fat, protein, and lactose concentrations with infrared spectroscopy using a Milkoscan 4000 (Foss MilkoScan, Hillerød, Denmark) and for N according to the Dumas method (Hansen, 1989).

Calculations

The SID values of dietary CP were calculated based on analyzed total dietary CP content and estimated SID values (AgroSoft WinOpti.Net, AgroSoft A/S, Tjørring, Denmark). The allotment of FM1, FM2, barley, and fiber mix was recorded for each meal, which allowed the daily feed supply to be calculated. Once per week (on days 4 ± 3, 11 ± 3, and 18 ± 3), energy intake, energy output, energy retention (negative retention is equivalent to mobilization), and energy utilization were quantified or estimated as described in detail below. Initially, the intake of feed and GE (GE_intake) were quantified as follows:

$$\text{ADFI }\left[\text{kg}/\text{d}\right]=\text{ feed supply }\left[\text{kg}/\text{d}\right]–\text{ feed residue }\left[\text{kg}/\text{d}\right],$$ (1)

$${\text{GE}}_{\text{Intake}}\left[\text{MJ GE}/\text{d}\right]=\text{ ADFI }\left[\text{kg}/\text{d}\right]\times {\text{ GE}}_{\text{Diet}}\left[\text{MJ GE}/\text{kg}\right].$$ (2)

The daily losses of energy via feces (E_Feces) and urine (E_Urine) were quantified using marker technique and total collection, respectively. The DM digestibility was calculated as the apparent total tract DM digestibility using lignin as a marker as described by Stein et al. (2007). Undigested DM was calculated from ADFI and DM digestibility, and the fecal GE output was then quantified by multiplying undigested DM with the analyzed GE concentration in feces,

$$\text{Fecal GE output }\left[\text{MJ}/\text{d}\right]=\text{ Undigested DM }\left[\text{kg}/\text{d}\right]\times {\text{ GE}}_{\text{Feces}}\left[\text{MJ GE}/\text{kg}\right].$$ (3)

The 24-h urine production was estimated from the urine production quantified during a 6-h period. The GE concentration in urine was estimated using analyzed urinary N and assuming that energy in urine from sows contains 120 kJ per g N in urine (r = 0.97, n = 104; Theil, 2002). It was assumed that the urine production and the energy loss via urine were constant throughout the day and urine GE output was thus calculated as follows:

$$\text{Urine GE output }\left[\text{MJ}/\text{d}\right]=\text{ Urine production }\left[\text{kg}/\text{6h}\right]\times \text{ 4 }\left[\text{6h}/\text{d}\right]\times {\text{ GE}}_{\text{Urine}}\left[\text{MJ GE}/\text{kg}\right].$$ (4)

Milk production was estimated according to Hansen et al. (2012), using litter weight gain (kg/d) and litter size as inputs. Gross energy in milk was estimated using energy values of 23.9, 39.8, and 16.5 kJ/g (Weast, 1977), respectively, for protein, fat, and lactose, where the milk protein content was calculated from milk N × 6.38, and fat and lactose concentrations were obtained by Milkoscan. The fat concentration was multiplied by 0.90, because milk fat measured by Milkoscan overestimates the milk fat content by 10% when analyzed by the classical method for milk fat (Krogh, 2017). Heat energy produced due to maintenance (HE_M) processes was estimated by assuming that lactating sows require 0.482 MJ/(kg^0.75 × d), and heat energy due to milk production was estimated by assuming a k_l of 0.78 (Theil et al., 2004). The daily heat production due to maintenance [HE_M] and milk production [E_Milk] and the daily milk GE output were estimated as described by Feyera and Theil (2017) using the following equations:

$${\text{HE}}_{\text{M}}\left[\text{MJ}/\text{d}\right]=\text{ Sow metabolic BW }[{\text{kg}}^{0.\text{75}}\left]\times 0.\text{482 }\right[\text{MJ }\times {\text{ kg}}^{-0.\text{75}}\times {\text{ d}}^{-\text{1}}],$$ (5)

$$\text{Milk GE output }\left[\text{MJ}/\text{d}\right]=\text{ Milk production }\left[\text{kg}/\text{d}\right]\times {\text{ GE}}_{\text{Milk}}\left[\text{MJ GE}/\text{kg}\right],$$ (6)

$${\text{HE}}_{\text{Milk}}\left[\text{MJ}/\text{d}\right]=\left({\text{E}}_{\text{Milk}}\left[\text{MJ GE}/\text{d}\right]/0.\text{78}\right)-{\text{E}}_{\text{Milk}}\left[\text{MJ GE}/\text{d}\right].$$ (7)

Based on quantified and estimated traits, the total GE output was estimated as follows:

$$\begin{array}{l}\text{Total GE output }\left[\text{MJ}/\text{d}\right]={\text{ Fecal GE output}}_{}\left[\text{MJ}/\text{d}\right]+\text{ Urine GE output }\left[\text{MJ}/\text{d}\right]+{\text{ HE}}_M\\ \left[\text{MJ}/\text{d}\right]+\text{ Milk GE output }\left[\text{MJ}/\text{d}\right]+{\text{ HE}}_{\text{Milk}}\left[\text{MJ}/\text{d}\right].\end{array}$$ (8)

Retained energy (RE) was then calculated as follows (negative values represent energy mobilization):

$$\text{RE }\left(\text{MJ}/\text{d}\right)={\text{ GE}}_{\text{Intake}}\left[\text{MJ}/\text{d}\right]–\text{ Total GE output }\left[\text{MJ}/\text{d}\right].$$ (9)

The daily RE was also estimated from the changes in body composition from days 4 to 18 of lactation. Thus, body pools of protein and fat were estimated using prediction equations developed for Landrace × Yorkshire gilts (Rozeboom et al. (1994), based on BW (kg), D₂O space (kg), and back fat measurements (mm),

$$\text{Body protein }\left[\text{kg}\right]=\text{ 1}.\text{3 }+0.\text{1}0\text{3 }\times \text{ BW }\left[\text{kg}\right]+0.0\text{92 }\times {\text{ D}}_{\text{2}}\text{O space }\left[\text{kg}\right]–0.\text{1}0\text{8 }\times \text{ back fat }\left[\text{mm}\right],$$ (10)

$$\text{Body fat }\left[\text{kg}\right]=–\text{7}.\text{7 }+0.\text{649 }\times \text{ BW }\left[\text{kg}\right]–0.\text{61}0\times {\text{ D}}_{\text{2}}\text{O space }\left[\text{kg}\right]+0.\text{299 }\times \text{ back fat }\left[\text{mm}\right].$$ (11)

The D₂O space was estimated based on D₂O enrichment in urine at baseline (0 h) and after equilibration (6 h after D₂O administration) on days 4 and 18. Protein and fat mobilization between days 4 and d 18 of lactation were estimated by subtracting body pools at day 4 from those derived at day 18. Retained energy (negative RE indicating mobilization) using the D₂O technique was then estimated according to Weast (1977) as follows:

$$\begin{array}{l}{\text{RE}}_{\text{D2O}}\left[\text{MJ}/\text{d}\right]=(\left(\text{Body protein d 18 }–\text{ Body protein d 4 }\left[\text{kg}\right]\right)\times \text{ 23}.\text{9 }\left[\text{MJ}/\text{kg}\right]+(\text{Body}\\ \text{fat d 18 }–\text{ Body fat d 4 }\left[\text{kg}\right])\times \text{ 39}.\text{8 }\left[\text{MJ}/\text{kg}\right])/\text{14 }\left[\text{d}\right].\end{array}$$ (12)

The dietary concentration of ME and metabolizabiliy of the feed (ME:GE ratio) were quantified as follows:

$$\text{ME in feed }\left[\text{MJ}/\text{kg}\right]=\left({\text{E}}_{\text{Intake}}–{\text{ E}}_{\text{Feces}}–{\text{ E}}_{\text{Urine}}\right)\left[\text{MJ}/\text{d}\right]/\text{ ADFI }\left[\text{kg}/\text{d}\right],$$ (13)

$$\text{Metabolizability of feed }\left[\%\right]=\text{ ME }\left[\text{MJ}/\text{kg}\right]\times \text{ 1}00/\text{ GE }\left[\text{MJ}/\text{kg}\right].$$ (14)

The feed efficiency of the sows was estimated by calculating the NE (NEc) corrected for energy mobilization obtained from estimated energy balances (values derived by equations 5, 6, 9, and 1, respectively),

$$\text{NEc }\left[\text{MJ}/\text{kg}\right]=\left({\text{HE}}_{\text{M}}+\text{ Milk GE output }+\text{ RE}\right)\left[\text{MJ}/\text{d}\right]/\text{ ADFI }\left[\text{kg}/\text{d}\right].$$ (15)

Utilization of ME originating from the feed was estimated as follows:

$$\text{Utilization ofME from feed}\left[\%\right]=\text{NEc }\left[\text{MJ}/\text{kg}\right]\times \text{ 1}00/\text{ ME }\left[\text{MJ}/\text{kg}\right].$$ (16)

Finally, utilization of GE as NE was estimated as follows:

$$\text{Utilization ofNE from feed}\left[\%\right]=\text{NEc }\left[\text{MJ}/\text{kg}\right]\times \text{ 1}00/\text{ GE }\left[\text{MJ}/\text{kg}\right].$$ (17)

Statistical Analysis

Data were analyzed as a randomized design and the effect of sow ADFI, initial measures of BW, back fat, mobilization (i.e., changes in BW, back fat, and body protein and fat pools), milk yield, litter size, and piglet ADG were tested using the following model:

$$Y_{ijk}=\mu +{\alpha }_i+{\beta }_j+{\delta }_k+{\epsilon }_{ijk},$$

where Y_ijk is the observed trait, µ is the overall mean of the observations, α_i is the main effect of the dietary treatment (i = 1, 2, 3, 4, 5, 6), β_j is the main effect of parity (j = 1 or ≥ 2), δ_k is the random effect of block (k = 1, 2, 3, 4), and Ɛ_ijk is the residual random component. Feed intake, milk production and composition, urine production and composition, fecal production and composition, and the energy metabolism (GE output in feces and urine, GE output in milk, and heat production) and total energy retention were tested for each week separately in a similar model, except that actual day in milk (DIM) for sampling was included as a covariate in the model.

To analyze the main effects of treatment and week of lactation on energy metabolism (GE supply, GE intake, ME intake, and total energy retention), energy utilization (milk GE output, heat production, urine GE output, fecal GE output, and total energy output relative to intake), and dietary energy and utilization of dietary energy (ME and NEc in feed, ME:GE, NEc:ME, and NEc:GE ratios) were applied to the following model:

$$Y_{ijklm}=\mu +{\alpha }_i+{\beta }_j+{\gamma }_k+{\delta }_l+t_m+{\epsilon }_{ijklm},$$

where Y_ijklm is the observed trait, µ is the overall mean of the observations, α_i is the main effect of the dietary treatment (i = 1, 2, 3, 4, 5, 6), β_j is the effect of week relative to parturition (j = 1, 2, 3), γ_k is the main effect of parity (k = 1 or ≥2), δ_l is the effect of block (l = 1, 2, 3, 4), t_m is the random effect of sows (m = 1, 2, 3, …, 72), and Ɛ_ijklm is the residual random component. To account for repeated measurements within each sow, week relative to parturition was included as a repeated component.

Statistical analyses were performed using the MIXED procedure of SAS (version 9.3, SAS Institute Inc., Cary, NC). To account for multiple mean comparison, the P values were adjusted using a Tukey test. In addition, linear, quadratic, and cubic effects of dietary protein content were tested within the final model. All estimates were considered significant at P < 0.05 and tendency for significance was accepted at P ≤ 0.10. Mean values are presented as least squares means ± SEM. No interactions (treatment × parity or treatment × week) were observed for any variables (P > 0.05) and were therefore omitted in the final model.

RESULTS

Dietary Treatments and Lactation Performance

Analyzed CP concentration of diets was higher than expected based on the feed formulation and ranged from 14.6% for treatment 1 to 18.6% for treatment 6 (Table 2), whereas the intended contents were 13.1% to 17.7%, respectively. Concentrations of Lys, Met, Met + Cys, and Thr were fairly constant across treatments, whereas Trp increased gradually from treatments 1 through 6, because soybean meal in FM2 increased the total Trp supply above the daily requirement.

Two sows were excluded from the study due to rectal prolapse and very low feed intake, respectively. Sow weight, weight loss, back fat, back fat loss, protein and fat mobilization, and RE_D2O did not differ among treatments (Table 3). Total number of born, live born, and weaned piglets, piglet BW at days 2 and 28, and piglet ADG (days 2 to 28) did not differ among treatments. Total feed intake, piglet ADG (days 2 to 18), and total milk yield (days 2 to 28; P = 0.03) were lowest for treatment 5 but did not differ among treatments 1, 2, 3, 4, and 6.

Table 3.

Body composition and lactation performance over a 28-d lactation period of sows fed varying dietary protein concentrations

Item	Treatment, % SID CP/kg						SEM	Parity		SEM	P-value1
	11.8	12.8	13.4	14.0	14.7	15.6		First	Multi		Trt2	Parity	Linear	Cubic
	1	2	3	4	5	6
Sows
No. of sows	12	12	12	12	12	12		12	60		–	–	–	–
Mean parity	3.2	3.2	3.2	3.3	3.2	3.2		1	3.6		–	–	–	–
ADFI days 2–28, kg/d	6.53a	6.17ab	6.67a	6.38ab	5.67b	6.53a	0.18	5.32	6.52	0.18	<0.01	<0.001	NS	*
Sow BW day 2, kg	266	255	269	269	262	263	8.37	227	272	8.54	0.52	<0.001	NS	NS
Sow BW loss days 2–18, kg/d	0.81	0.59	0.62	0.90	0.44	0.66	0.23	0.87	0.63	0.23	0.72	0.33	NS	NS
Sow BW loss days 2–28, kg/d	0.63	0.85	0.90	0.81	0.72	0.71	0.16	0.77	0.77	0.16	0.86	0.98	NS	NS
Back fat day 2, mm	15.7	16.7	16.1	16.5	14.9	14.2	0.85	16.8	15.4	0.85	0.15	0.11	NS	NS
Back fat loss days 2–18, mm/d	0.19	0.20	0.15	0.17	0.13	0.10	0.05	0.13	0.16	0.05	0.20	0.39	*	NS
Back fat loss days 2–28, mm/d	0.18	0.19	0.15	0.16	0.15	0.12	0.03	0.17	0.16	0.03	0.22	0.54	*	NS
Protein mobilization days 4–18, kg/d3	0.04	−0.01	0.01	0.09	0.01	0.02	0.05	0.02	0.03	0.05	0.51	0.84	NS	NS
Fat mobilization days 4–18, kg/d3	0.82	0.77	0.80	0.75	0.45	0.80	0.22	1.01	0.67	0.22	0.72	0.12	NS	NS
RED2O days 4–18, MJ/d4	33.8	28.6	32.6	31.3	20.5	27.0	7.14	42.4	25.9	6.72	0.72	0.03	NS	NS
Milk yield days 2–28, kg/d5	13.3ab	12.9ab	13.8a	13.0ab	11.9b	13.1ab	0.45	12.3	13.1	0.44	0.03	0.08	NS	NS
Piglets
Total born piglets	19.1	20.8	20.8	19.5	19.1	20.4	1.20	16.7	20.7	1.18	0.78	<0.01	NS	NS
Live born piglets	17.7	19.2	18.6	18.9	17.2	17.2	0.94	14.8	18.8	0.90	0.43	<0.001	NS	NS
Piglets per litter day 2	14	14	14	14	14	14		14	14		–	–	–	–
Piglets per litter day 28	12.8	13.2	13.4	12.8	12.1	12.8	0.45	12.8	12.8	0.46	0.22	0.93	NS	*
Piglet BW day 2, kg	1.68	1.69	1.62	1.80	1.65	1.56	0.07	1.47	1.71	0.07	0.25	<0.01	NS	NS
Piglet BW day 28, kg	8.12	7.62	8.42	8.14	7.37	8.09	0.30	6.55	8.24	0.30	0.13	<0.001	NS	NS
Piglet ADG days 2–18, g/d	230ab	217ab	241a	224ab	191b	234ab	12.8	177	232	13.0	0.03	<0.001	NS	NS
Piglet ADG days 2–28, g/d	248	228	261	243	220	250	11.0	195	251	10.8	0.08	<0.001	NS	NS

P values for linear and cubic treatment effects: NS = not significant; *<0.05; **<0.01; ***<0.001.

^a,bLeast squares treatment means with different superscript letters differ significantly (P < 0.05).

¹No quadratic effects were observed, P > 0.05.

²Treatment.

³Calculated from estimation of body fat and protein pools on days 4 and 18 according to equations by Rozeboom et al. (1994), which are based on BW (kg), D₂O space (kg), and back fat measurements (mm).

⁴Calculated based on fat and protein mobilization and their respective energy values adapted from Weast (1977).

⁵Estimated milk yield from Hansen et al. (2012).

Sow Feed Intake, Milk Composition, Milk Yield, Urine, and Fecal Output

Sow DM intake was highest for treatment 4 and lowest for treatment 3 with the other treatments being intermediate in week 1 (P = 0.04). In weeks 2 and 3, DM intake did not differ among treatments. Estimated milk production decreased linearly (P < 0.01) with increasing dietary CP in week 1, but did not differ among treatments in week 2 or 3. Milk DM did not differ among treatments in weeks 1 and 2, but in week 3 sows receiving treatment 1 had a lower DM content in milk compared with the remaining treatments (P = 0.02). Milk fat, lactose and energy, urine production, and concentration of GE in urine did not differ among treatments in any week (Table 4).

Table 4.

Quantified feed intake, milk, urine and feces production, and composition for weeks 1, 2, and 3 in lactation in sows fed varying dietary protein concentration

Item	Treatment, % SID CP/kg						SEM	Parity		SEM	P-value
	11.8	12.8	13.4	14.0	14.7	15.6		First	Multi		Trt1	Parity	Linear	Cubic
	1	2	3	4	5	6
No. of sows	12	12	12	12	12	12		12	60		–	–	–	–
Week 1 (days 4 ± 3)
ADFI, kg	4.02ab	4.08ab	3.74b	4.19a	3.95ab	3.91ab	0.10	3.68	4.05	0.10	0.04	<0.01	NS	NS
Milk production, kg/d2	8.14ab	8.24ab	8.31a	7.97ab	8.02ab	7.83b	0.12	8.27	8.04	0.12	0.02	0.09	**	NS
Milk DM, %	18.8	18.7	19.6	19.0	18.8	19.5	0.76	20.2	18.8	0.70	0.68	0.03	NS	NS
Milk fat, %	7.86	7.68	8.97	8.17	7.70	8.47	0.78	9.00	7.95	0.73	0.48	0.09	NS	NS
Milk lactose, %	4.86	4.90	4.79	4.80	4.85	4.87	0.05	4.82	4.85	0.05	0.60	0.58	NS	NS
Milk protein, %	5.39ab	5.56ab	5.21b	5.51ab	5.70a	5.66ab	0.12	5.75	5.46	0.10	0.04	0.02	*	NS
Milk GE, MJ/kg	5.02	4.96	5.36	5.13	5.04	5.29	0.28	5.51	5.05	0.25	0.59	0.04	NS	NS
Urine production, kg/d	7.16	6.37	8.10	12.7	8.11	9.41	2.17	6.88	8.98	2.16	0.10	0.28	NS	NS
Urine GE, MJ/L	0.47	0.47	0.74	0.41	0.62	0.49	0.12	0.72	0.49	0.13	0.23	0.09	NS	NS
Feces production, kg DM/d	0.42a	0.41ab	0.36b	0.43a	0.40ab	0.37ab	0.02	0.38	0.40	0.02	<.01	0.23	NS	NS
Feces GE, MJ/kg DM	17.7	17.7	17.8	17.8	17.7	17.9	0.23	17.7	17.8	0.23	0.99	0.61	NS	NS
Week 2 (days 11 ± 3)
ADFI, kg	6.02	6.34	6.54	5.98	5.56	5.87	0.45	5.07	6.25	0.45	0.27	<0.01	NS	NS
Milk production, kg/d2	13.9	13.5	14.2	13.4	12.6	13.5	0.49	13.3	13.6	0.53	0.27	0.63	NS	NS
Milk DM, %	20.7	19.1	19.2	20.6	19.8	19.0	1.43	20.5	19.5	1.41	0.36	0.24	NS	NS
Milk fat, %	10.3	8.45	8.62	9.97	8.93	8.23	1.46	9.75	8.92	1.44	0.26	0.33	NS	NS
Milk lactose, %	4.75	4.93	4.82	4.76	4.87	4.94	0.11	4.82	4.85	0.11	0.10	0.64	NS	NS
Milk protein, %	4.96ab	4.81b	4.77b	5.12ab	5.37a	5.09ab	0.25	5.21	4.98	0.24	0.02	0.15	*	*
Milk GE, MJ/kg	5.80	5.09	5.17	5.68	5.37	5.08	0.56	5.64	5.30	0.55	0.32	0.29	NS	NS
Urine production, kg/d	9.97	9.84	12.1	12.8	9.11	14.2	1.81	9.57	11.7	1.88	0.21	0.30	NS	NS
Urine GE, MJ/L	0.38	0.45	0.40	0.38	0.65	0.36	0.07	0.50	0.43	0.07	0.05	0.39	NS	NS
Feces production, kg DM/d	0.64ab	0.70ab	0.74a	0.67ab	0.54b	0.62ab	0.05	0.51	0.68	0.05	0.03	<0.001	NS	*
Feces GE, MJ/kg DM	17.4	17.3	17.1	17.1	17.1	17.2	0.15	17.2	17.2	0.16	0.31	0.64	NS	NS
Week 3 (days 18 ± 3)
ADFI, kg	7.98	7.93	8.08	7.98	7.53	8.06	0.21	6.89	8.14	0.20	0.37	<0.001	NS	NS
Milk production, kg/d2	15.3	14.7	15.6	15.1	13.7	15.4	0.69	14.3	15.1	0.73	0.24	0.26	NS	NS
Milk DM, %3	17.4b	18.1a	18.2a	18.6a	18.2a	18.2a	0.25	18.4	17.8	0.24	0.02	0.02	*	NS
Milk fat, %	6.98	7.26	7.27	7.57	7.28	6.96	0.29	7.62	7.13	0.31	0.61	0.16	NS	NS
Milk lactose, %	5.05	4.99	5.03	4.98	5.02	5.05	0.04	5.00	5.02	0.04	0.81	0.66	NS	NS
Milk protein, %	4.46d	4.81bc	4.71cd	5.04ab	5.09ab	5.18a	0.08	4.98	4.87	0.08	<.001	0.23	***	NS
Milk GE, MJ/kg3	4.43	4.60	4.62	4.80	4.74	4.62	0.11	4.80	4.60	0.11	0.17	0.10	NS	NS
Urine production, kg/d	11.7	8.81	10.6	11.0	11.9	11.1	1.69	8.03	11.5	1.60	0.76	0.06	NS	NS
Urine GE, MJ/L	0.34	0.52	0.42	0.41	0.58	0.49	0.07	0.49	0.46	0.08	0.22	0.72	NS	NS
Feces production, kg DM/d	0.93a	0.88ab	0.92a	0.91ab	0.79b	0.91ab	0.03	0.81	0.91	0.03	0.03	0.01	NS	NS
Feces GE, MJ/kg DM	17.5	17.5	17.5	17.5	17.4	17.4	0.21	17.4	17.5	0.22	0.84	0.87	NS	NS

P values for linear and cubic treatment effects: NS = not significant; *<0.05; **<0.01; ***<0.001.

^a–dLeast squares treatment means with different superscript letters differ significantly (P < 0.05).

¹Treatment.

²Estimated milk production for the day of sampling from Hansen et al. (2012).

³Quadratic effect of treatment P value < 0.05.

Metabolism and Utilization of Energy

Urine GE output did not differ in week 1, but increased with increasing dietary CP content in week 2 (ANOVA, P < 0.01; Linear, P < 0.001) and week 3 (ANOVA, P < 0.01; Linear, P < 0.001; Table 5). Metabolizable energy intake differed among treatments in week 1, but did not differ among treatments in week 2 or 3. Milk GE output and total energy retention did not differ among treatments in any week. In week 1, heat production was lowest in treatment 5, highest in treatments 3 and 4 while intermediate in treatments 1, 2, and 6 (P = 0.02). No evidence for difference in sow heat production was observed in week 2, whereas it was highest in week 3 for treatment 4, lowest for treatment 5, and intermediate for treatments 1, 2, 3, and 6 (P = 0.02).

Table 5.

Daily energy input and estimated output and retention in sows fed varying dietary protein concentration (MJ/d)

Item, MJ/d	Treatment, % SID CP/kg						SEM	Parity		SEM	P-value1
	11.8	12.8	13.4	14.0	14.7	15.6		First	Multi		Trt2	Parity	Linear	Cubic
	1	2	3	4	5	6
No. of sows	12	12	12	12	12	12		12	60		–	–	–
Week 1 (days 4 ± 3)
GE intake	63.1ab	63.9ab	58.5b	65.4a	61.6ab	60.8ab	1.57	57.5	63.2	1.52	0.03	<0.01	NS	NS
Urine GE output	2.46	2.73	3.61	4.10	3.77	3.82	0.50	3.57	3.38	0.54	0.06	0.73	**	NS
Fecal GE output	7.39ab	7.20ab	6.35b	7.75a	7.03ab	6.68b	0.24	6.73	7.13	0.22	<0.01	0.11	NS	*
ME intake3	52.4ab	53.3a	46.4b	53.4a	49.2ab	49.5ab	1.69	46.7	51.6	1.54	0.01	0.01	NS	NS
Milk GE output	40.1	40.0	44.7	42.0	39.8	39.3	2.43	44.7	40.2	2.32	0.27	0.02	NS	NS
Heat production	42.9ab	41.9ab	44.3a	44.0a	41.6b	42.1ab	0.82	40.7	43.2	0.78	0.02	<.001	NS	NS
Total energy retention4	−33.4	−29.7	−39.9	−34.8	−36.0	−33.1	2.71	−38.1	−33.6	2.66	0.12	0.12	NS	NS
Week 2 (days 11 ± 3)
GE intake	94.5	99.2	102.3	93.4	86.7	91.2	7.05	79.3	97.6	7.07	0.24	<0.01	NS	NS
Urine GE output	3.00ab	2.90b	3.92ab	4.63ab	4.62ab	4.73a	0.50	3.31	4.11	0.52	<0.01	0.12	***	NS
Fecal GE output	11.2ab	12.1ab	12.6a	11.6ab	9.37b	10.7ab	0.97	8.70	11.8	0.95	0.03	<.001	NS	*
ME intake3	79.9	83.8	85.8	76.9	72.1	75.7	6.37	66.6	81.5	6.38	0.20	<0.01	NS	NS
Milk GE output	77.5	67.8	73.4	76.4	64.9	68.4	7.53	70.1	71.5	7.39	0.31	0.80	NS	NS
Heat production	52.9	49.5	52.4	53.2	49.3	50.2	2.40	47.4	52.0	2.35	0.21	0.01	NS	NS
Total energy retention4	−56.3	−33.4	−45.4	−56.2	−49.8	−42.5	15.7	−57.9	−44.7	15.2	0.39	0.18	NS	*
Week 3 (days 18 ± 3)
GE intake	125.3	124.1	126.3	124.6	117.4	125.2	3.27	108	127	3.09	0.33	<.001	NS	NS
Urine GE output	3.28b	3.61b	4.08ab	4.07ab	5.58a	4.55ab	0.46	3.35	4.39	0.48	<0.01	0.03	***	NS
Fecal GE output	16.3a	15.3ab	16.2a	15.9ab	13.7b	15.8ab	0.58	14.1	15.8	0.63	0.01	0.02	NS	NS
ME intake3	105.8	104.6	106.0	104.5	97.4	104.6	2.74	90.2	106.6	2.58	0.16	<.001	NS	NS
Milk GE output	65.9	67.2	71.7	72.6	64.1	71.0	3.17	66.8	69.1	3.32	0.20	0.51	NS	NS
Heat production	49.0ab	48.9ab	51.3ab	51.8a	47.9b	50.7ab	1.04	45.8	50.8	1.06	0.02	<.001	NS	NS
Total energy retention4	−14.5	−15.9	−16.7	−14.8	−18.1	−17.0	3.12	−21.1	−15.2	3.12	0.96	0.09	NS	NS

P values for linear and cubic treatment effects: NS = not significant; *<0.05; **<0.01; ***<0.001.

^a–bLeast squares treatment means with different superscript letters differ significantly (P < 0.05).

¹No quadratic effects were observed, P > 0.05.

²Treatment.

³GE intake – urine GE output – fecal GE output.

⁴GE intake – milk GE output – heat production – urine GE output – fecal GE output.

Impact of Treatment and Stage of Lactation on Metabolism and Utilization of Energy

The main effects on sow ME intake, NEc (NE corrected for mobilization) in feed and energy utilization (for GE in milk, heat, urine, and feces) are shown in Table 6. Sow ME intake did not differ among treatments but increased from weeks 1 through 3 (P < 0.001). The ME in feed was highest for treatments 1 and 2, intermediate for treatment 3, and lowest for treatments 4, 5, and 6 (P < 0.01). The feed NEc peaked for treatment 2, whereas it was intermediate for treatments 1, 3, 4, and 6, and lowest for treatment 5 (ANOVA, P = 0.01; Linear, P < 0.01; Figure 1). The feed metabolizability (ME:GE) was highest for treatments 1 and 2, intermediate for treatment 3, and lower for treatments 4, 5, and 6 (ANOVA, P < 0.01; Linear, P < 0.001). Feed NEc:GE was highest for treatments 1 and 2, intermediate for treatments 3 and 4, and lowest for treatments 5 and 6 (ANOVA, P = 0.03; Linear, P < 0.01; Figure 2). Milk GE output, heat production, and urine GE output, when expressed relative to GE intake, did not differ among treatments. Milk GE output, relative to GE intake, was greatest for sows in week 2, intermediate in week 1, and lowest in week 3 (P < 0.001). Heat production relative to GE intake decreased (P < 0.001) as lactation progressed from weeks 1 to 3. Urine GE output relative to GE intake was lowest for treatments 1 and 2, intermediate for treatments 3 and 4, and highest for treatments 5 and 6 (P < 0.001), and it was greater in week 1 compared with weeks 2 and 3 (P < 0.001). Total energy output as percentage of GE intake was greater in weeks 1 and 2 than in week 3 (P < 0.01).

Table 6.

Feed efficiency and utilization of dietary energy in sows fed varying dietary protein concentration

Item	Treatment, % SID CP/kg						SEM	Week			SEM	P-value1
	11.8	12.8	13.4	14.0	14.7	15.6		1	2	3		Trt2	Week	Linear	Cubic
	1	2	3	4	5	6
No. of sows	12	12	12	12	12	12		72	72	72
Energy metabolism, MJ/d
GE supply	99.1	96.3	98.1	97.6	95.7	94.2	2.14	62.1C	103B	130A	1.66	0.24	<0.001	*	NS
GE intake	96.4	94.8	96.5	95.7	88.1	90.4	2.67	62.1C	94.6B	124A	1.71	0.09	<0.001	*	NS
ME intake3	81.3	80.1	79.9	79.2	72.4	75.2	2.52	50.7C	79.1B	103.9A	1.76	0.05	<0.001	**	NS
Total energy retention4	−34.0	−26.4	−32.5	−33.9	−33.7	−31.5	5.4	−32.9B	−46.9C	−15.9A	4.68	0.64	<0.001	NS	NS
Energy utilization, % of GE intake
Milk GE output	73.4	64.2	70.7	69.2	71.5	68.2	5.42	66.5B	85.7A	56.2C	4.52	0.58	<0.001	NS	NS
Heat production	54.5	53.6	55.7	56.0	59.0	59.0	2.73	71.0A	57.4B	40.6C	2.17	0.29	<0.001	*	NS
Urine GE output	3.3b	3.5b	4.4ab	4.7ab	5.3a	5.1a	0.43	5.3A	4.2B	3.6B	0.36	<0.001	<0.001	***	NS
Fecal GE output	12.3a	12.1abc	12.1abc	12.2ab	11.6c	11.8bc	0.13	11.4C	12.0B	12.6A	0.09	<0.001	<0.001	***	NS
Total energy output	153.9	152.2	147.6	145.4	153.4	144.2	11.2	156.8A	163.9A	127.7B	8.93	0.96	<0.01	NS	NS
Dietary energy and utilization of dietary energy
ME in feed, MJ/kg5	13.1a	13.1a	13.0ab	12.8b	12.8b	12.8b	0.09	12.9	13.0	13.1	0.07	<0.01	0.07	***	NS
NEc in feed, MJ/kg6	10.3ab	10.4a	10.1ab	9.8ab	9.8b	9.9ab	0.20	9.8B	9.7B	10.6A	0.16	0.01	<0.001	**	*
ME:GE, %7	84.6a	84.4a	83.5ab	82.9b	83.0b	83.0b	0.44	83.2	83.7	83.8	0.36	<0.01	0.19	***	NS
NEc:ME, %8	79.1	79.1	77.8	77.6	76.0	77.2	0.98	76.3B	75.5B	80.8A	0.66	0.14	<0.001	*	NS
NEc:GE, %9	66.7a	66.6a	65.1ab	64.4ab	63.6b	63.8b	1.11	63.1B	63.5B	67.7A	0.87	0.03	<0.001	**	NS

P values for linear and cubic treatment effects: NS = not significant; *<0.05; **<0.01; ***<0.001.

^a,bTreatment least squares means with different superscript letters differ significantly (P < 0.05).

^A–CWeek least squares means with different superscript letters differ significantly (P < 0.05).

¹No quadratic effects were observed, P > 0.05.

²Treatment.

³GE intake GE urine – GE feces.

⁴GE intake – GE urine – GE feces – GE milk – heat.

⁵ME intake / feed intake.

⁶Milk GE output + energy required for maintenance + energy retention / feed intake.

⁷ME in feed / GE in feed.

⁸NEc in Feed / ME in feed.

⁹NEc in Feed / GE in feed.

Figure 1.

Net energy corrected for energy mobilization (NEc, MJ/kg) in feed varying in dietary standardized ileal digestible protein (SID CP).

Figure 2.

Net energy corrected for energy mobilization in percent of GE in feed varying in dietary standardized ileal digestible protein (SID CP).

DISCUSSION

Optimal Dietary SID CP

A previous Danish dose–response trial with a total of 540 sows fed increasing SID CP, carried out in the same herd (i.e., same genetics and management) as the current study, revealed that sows require 13.5% SID CP to reach maximal litter gain (Strathe et al., 2017a). However, in that study, SID CP was increased by including more soybean meal and consequently all essential AA increased along with increasing SID CP and it was expected that this requirement could be reduced slightly by improving the dietary AA profile used. This study was a part of a large scale experiment with 540 sows, aiming at evaluating how much SID CP was required to maximize daily litter gain when crystalline Lys, Met, Thr, and Trp were added, but unfortunately no break-point could be detected because the SID CP content in the lowest group was too high (11.8% SID CP; Højgaard et al., unpublished data). In a follow-up study, carried out in the same herd and under the same conditions as the current study (i.e., addition of crystalline Lys, Met, Thr, and Trp), it was revealed that sows require 12.5% SID CP to reach maximal litter gain (Højgaard et al., 2017). In the latter study, the dietary SID CP was within the range of 9.6% to 15.2% SID CP, which facilitated a clear break-point to be reported. In light of these 2 other Danish studies, the present study indicate that feed efficiency, as evaluated by NEc, is maximized at the same CP inclusion level as that required to maximize litter gain (Højgaard et al., 2017). The present study thus indicate that maximizing energy utilization coincides with maximizing milk production and is likely due to minimal energy being lost via urine and heat.

Sow Mobilization

Throughout lactation, the energy output in milk, heat production, and GE output in feces and urine substantially exceeded the GE intake. An energy output of 157% and 164% relative to intake was found in weeks 1 and 2, respectively, whereas it dropped to 128% in week 3 (Table 6), indicating that high-yielding sows are greatly challenged in terms of adequate energy supply. The RE evaluated across days in milk using calculated energy balances yielded an average mobilization of 26.4 to 34.0 MJ/d (Table 6; mean 32.0 MJ/d for all treatments), whereas the RE_D2O (estimated from the D₂O technique) resulted in 20.5 to 33.8 MJ/d (Table 3; mean 29.0 MJ/d for all treatments). Overall, the mobilization was slightly lower when estimated from the D₂O technique compared with the balances, likely because loss of body protein and loss of body fat were underestimated due to higher gut fill at day 18 compared with day 4. The estimated energy mobilization obtained using the 2 approaches agreed very well for treatments 1 through 4 (less than 10% deviation), which provides confidence that the mathematic estimates (and assumptions) are reasonable, at least for these treatments. For sows fed treatment 5, the mobilization was substantially underestimated using the D₂O technique (20.5 vs. 33.7 MJ/d; respectively), and for sows fed treatment 6, it was slightly underestimated (27.0 vs. 31.5 MJ/d; respectively), and we speculate whether this might be due to different degrees of hydration for these sows at days 4 and 18 (sows fed treatment 5) and due to altered N metabolism (both treatment groups). The D₂O technique and loss of back fat strongly indicated that a substantial proportion of the total energy required by the sows was covered by energy being mobilized from body fat. Our data and those of Strathe et al. (2017b) suggest that body mobilization appears to be an important homeorrhetic function to support milk production in high-yielding sows although it remains unclear as to whether body mobilization is an obligatory process.

Feed Efficiency and Net Energy Corrected for Mobilization

The prevailing energy evaluation systems used worldwide, even though they differ between countries/regions, have been developed for growing pigs, and NE in diets for growing pigs increases when fat is added, because NE systems better represent available energy no matter whether energy is retained as protein or fat. In lactating sows, however, maternal energy retention is clearly negative in high-yielding sows because the energy demand for milk is high (Feyera et al., 2017) and because the total energy requirement exceeds the available E intake. Thus, for lactating sows, feed efficiency needs to incorporate the energy contribution of mobilized energy to reveal the true impact of the diet. In this study, the calculated NE based on the feed formulation across all diets was 9.8 MJ/kg. In contrast, the measured NE corrected for mobilization varied across diets, and it peaked at 10.4 MJ NEc/kg for treatment 2, which contained 12.8% SID CP. Thus, the present study indicates that lactating sows are able to utilize dietary energy more efficiently when dietary protein supply relative to energy is optimal. Efficiency for utilizing dietary ME for milk production was previously found to be as high as 0.78 (Theil et al., 2004) which is higher than that reported for growth (k_g is 0.72; Strathe et al. (2010). In contrast, energy stored in the body can be assumed to be used with an efficiency of 0.89 (Noblet et al., 1998). Overall, for growing pigs, the NE constitute on average 50% of the GE (Just, 1982), whereas NE corrected for body mobilization in sows in the present study on average represented 65% of GE. Thus, sows are considerably more efficient in utilizing dietary GE (and ME) for milk than growing pigs are when converting ME into growth. The NE corrected for mobilization represent how much of the energy output can directly be ascribed to the diet, and taking energy mobilization into account may be a promising way to improve feed efficiency of lactating sows in the future by developing appropriate models.

The NEc was lower in weeks 1 and 2 compared with week 3, suggesting that energy utilization could possibly be improved by modifying the feeding curve in early lactation where sows are greatly under supplied. Probably, a more aggressive feeding curve in early lactation would reduce the body mobilization and maybe also improve the milk yield. Another possibility would be to supply sows ad lib access to protein-rich and energy-rich diets, so she could adjust her own SID CP:energy ratios daily and thereby maximize efficiency. In line with that, we recently showed that changing the dietary balance between protein and energy throughout lactation improved the milk yield of sows (Pedersen et al., 2016), but at present, most production herds do not have a feeding system that is capable of implementing this approach.

Sow Performance

Feeding lactating sows with increasing dietary CP to energy ratios did not affect lactation performance and sows were able to maintain a high milk yield across treatments. Sows secreted 58 to 64 MJ/d in milk and their heat production ranged from 47 to 51 MJ/d across treatments. In contrast, Theil et al. (2004) found that sows secreted 42 MJ/d in milk while their mean heat production was only 38 MJ/d. These differences in milk energy output and heat production are explained by the markedly higher sow productivity in the current study (14 vs. 10 suckling piglets). Interestingly, the milk yield was highest for sows fed low CP in week 1, suggesting that sows either were oversupplied with CP in early lactation, or alternatively, that the ideal AA profile in sow feeds is not consistent throughout lactation. Overall, piglet ADG and estimated milk yield for the entire lactation did not reveal a consistent pattern across dietary treatments. Nevertheless, over the 28-d lactation period, a linear increase in CP to NE ratio did not affect performance in this study, and this is highly likely due to the fact that the dietary supply of Lys, Met, Thr, and Trp was kept constant by increased inclusion level of crystalline AA in low CP diets. Of note are sows receiving treatment 5. These sows had a lower milk production, a lower feed intake, and a tendency for lower ADG throughout the study. The lower performance cannot be ascribed to any particular reason regarding the dietary treatment as the same ingredients were used for all treatment groups and 2 supplementary feed mixes (FM1 and FM2) were mixed in different proportions to obtain the dietary CP gradient. Whether the lower feed intake (0.5 to 1.0 kg/d less) for sows fed treatment 5 caused a lower milk production (1.0 to 0.9 kg/d less) or whether the lower milk yield caused the lower feed intake is not known. The lower productivity for sows fed treatment 5 in the current study is therefore considered a coincidence. On average, sows lost 0.63 to 0.9 kg BW per d, corresponding to 17 to 24 kg during the lactation period, which is in line with what has been shown in previous studies with high-yielding sows (Strathe et al., 2017a; Strathe et al., 2017b). The BW loss mostly originated from mobilization of fat, because protein mobilization was near to zero, indicating that sows were fed close to or above their CP requirement, whereas their energy requirement was not met.

Impact of Dietary CP Content on Utilization of Energy

The energy output in milk when expressed relative to GE intake did not change with increasing dietary SID CP, partly because Lys, Met, Thr, and Trp were kept constant in the feed and partly because sows use their body as a buffer of nutrients to maintain their milk production if nutrients from the diet are inadequate (Theil et al., 2012). Moreover, as expected, GE output in urine on a daily basis and relative to GE intake increased when sows were fed SID CP above their requirement. The heat production relative to GE intake increased concomitantly from 53.6% (treatment 2) to 59.0% (treatment 6) when sows were fed increasing SID CP above their requirement. The dietary effect on heat production was, however, not statistically different, likely because heat production was calculated and because factors like feed intake, milk production, and sow BW also affect the total heat production of sows and cause highly variable responses. From a mass balance point of view, it is important to consider the loss of energy in heat (even though the increase was not statistically different), because lactating sows can only utilize their ME for milk and heat production, while insufficient ME supply will be counteracted by additional energy mobilization. In support of our findings, energy loss in urine and heat production increased when growing pigs were fed 75%, 100%, 125%, 150%, and 175% of their CP requirement (Just, 1982). Le Bellego et al. (2001) reported that energy excretion in urine and heat production increased by 40% and 7%, respectively, in 65-kg growing pigs fed increasing dietary CP from 12.3% to 18.9%. The increasing excretion of energy in urine and the fact that body protein mobilization was close to zero indicate that sows in the current study were feed close to or above their protein requirement, and consequently, energy lost in urine increased from 3.3% in treatment 1 (when expressed relative to GE intake) to 5.3%−5.1% for sows fed treatment 5 and 6. The relative loss of energy in urine (i.e., when expressed relative to GE intake) is clearly higher than that reported for growing pigs, where energy excreted in urine can be as low as 2.1% of GE intake (Just, 1982). A study with lactating sows fed diets with 13.9% to 14.9% SID CP and 2 different fat levels showed a somewhat low excretion of energy in urine of 2.4% and 2.6% of GE intake (Theil et al., 2004), but it should be emphasized that milk production was much lower in that study compared with the present study. A recent study revealed a decreased urinary N output when sows were fed decreasing CP concentration and synthetic AA were added to optimize the AA profile (Huber et al., 2015), and based on that study and by assuming 120 kJ per g of urinary N, the urinary energy loss could be estimated as 2.2% of GE intake to 3.4%. The higher GE output in urine in the current study is therefore likely a result of excessive supply of AA relative to that required for milk production.

CONCLUSION

In conclusion, energy lost in urine increases with dietary supply of SID CP above their requirement and the same likely applies for the total heat production. Feed efficiency of sows should be evaluated based on how much of their productivity can be ascribed to energy originating directly from the diet merely than ignoring the substantial energy input coming from mobilized energy. Formulating diets for lactating sows should take into account that sows utilize dietary energy more efficiently for milk production than growing pigs utilize energy for retention. Incorporating this in future models will likely improve feed efficiency of the sows and in turn reduce feed costs and excretion of nutrients to the environment.