Feeding the modern sow to sustain high productivity

Selection for hyper‐prolific sows has increased the litter size by more than 50% during the last three decades, and proper nutrition of the female pigs has concomitantly changed due to improved prolificacy and productivity of gilts and sows. This review summarizes the physiological characteristics and nutritional challenges associated with feeding modern hyper‐prolific sows during the gilt rearing period and during gestation, transition, and lactation periods. The review presents up‐to‐date knowledge of the energy and lysine requirements of female pigs and focuses on how nutrition may increase fat gain and limit protein and weight gain in the gilt rearing period and in early and mid‐gestation. In late gestation, fetal and mammary growth should be considered and during the transition, colostrum yield and farrowing performance need to be optimized. Finally, milk production should be optimized and body mobilization should be minimized in the lactation period to achieve high feed efficiency in hyper‐prolific sows.


| INTRODUCTION
Modern sows are characterized by their high prolificacy as illustrated by the increase in the number of total born piglets in Danish sows from 12.9 in 2000 to 19.6 piglets per litter in 2020 (Danish Pig Research Centre, 2001-2020. Modern sows commonly wean 33-35 piglets per sow per year and herds with the highest productivity now wean more than 40 piglets per year per sow. However, the increase in litter size is accompanied by a decrease in the average birth weight of piglets (Moreira et al., 2020) and a decrease in the amount of colostrum consumed per piglet (Devillers et al., 2007;Vadmand et al., 2015), which, in turn, increases the vulnerability and reduces the growth potential of pigs born from modern sows . Pig breeds are genetically selected for lean meat production and high feed efficiency during the growing-finishing stage, but this concomitantly affects the genetics, physiology, productivity, and feed efficiency of reproductive females. Feeding the reproductive females is a discipline that deviates substantially from that of feeding growing pigs because traits such as sufficient body fat and colostrum production, and optimal farrowing and lactation performance are substantially more important for female reproductive pigs than the traditional traits in growing pigs, where the focus is on maximizing gain and feed efficiency. This review summarizes the physiological characteristics of growing gilts and reproductive sows during gestation, transition, and lactation, and revises the current knowledge within these physiological stages with respect to energy and lysine requirements.

| Physiological and nutritional aspects
Age and live weight at the onset of puberty, longevity, and lifetime reproductive output are important aspects when considering nutrition and management of growing gilts. Nutritionally, growing gilts represent a challenge because pigs have genetically been selected for lean meat and for high feed efficiency, and these traits have favored rapid growth of muscle tissues during growing/finishing periods with minimal fat retention. However, in contrast to growing/finishing pigs, the nutrition of growing gilts should aim for increased fat retention at the expense of protein retention, to avoid excess muscle mass and insufficient body fat. For reproductive females, excess muscle mass is expensive due to the extra feed required for maintenance energy requirements (0.440 MJ ME/kg 0.75; Dourmad et al., 2008). If gilts are mated at 170 kg at their first service, as compared with 140 kg, their daily maintenance requirement for energy is roughly equivalent to 0.2 kg more feed needed each day throughout life, which compromises their feed efficiency. Delaying gilt breeding to a greater age/weight has also other consequences than maintenance requirements because litter size is clearly favored by breeding at a higher weight (0.4 extra total born piglet for each 10 kg extra; , but it is also associated with negative consequences for breeding success when they are remated later for their second parity . Furthermore, gilts mated at a high live weight will most likely experience compromised longevity due to locomotory problems. In contrast to growing/finishing pigs, increased body fatness is required for reproductive females, because sufficient body fatness may be favorable for the onset of puberty and future reproductive output (J. S. Kim et al., 2015;Knecht et al., 2020;Maes et al., 2004). With modern lean genotype gilts, the focus should, therefore, be to ensure sufficient backfat thickness at an optimal age and weight to maximize the sow's lifetime reproductive output. The recommendations regarding gilts' age, weight, and backfat thickness at first service vary. Close and Cole (2000) recommended a breeding age of 220-230 days, at a weight of 130-140 kg and backfat thickness within 18-20 mm. Rozeboom (2015) suggested a target weight of 135-150 kg with less attention paid to backfat thickness and age. In Denmark, the current recommendation for the first service is a range of 140-160 kg live weight and 13-15 mm backfat thickness, which should be reached at an age of 220-240 days. This latter recommendation is valid for modern hyper-prolific sows to ensure large litter size in first parity without compromising sow longevity and feed efficiency.

| Feed and energy intakes
Increasing the daily feed intake of a single diet concomitantly increases the daily intake of crude protein (CP), lysine (Lys), and energy with the same proportion. Increased feed intake is associated with greater average daily gain (ADG) of growing gilts (Klaaborg et al., 2019;Klindt et al., 1999;Strathe et al., 2019) even when feed intake is increased at a fixed daily intake of CP and Lys (Thingnes et al., 2015). These studies indicated that the energy supply is the major limiting factor for growing gilts. Changing the feeding strategy from restricted feeding (max 2.52 kg/day) to semi-ad libitum (i.e., feeding close to ad libitum but the feed is provided in meals to avoid spillage; here max 3.27 kg/day) in gilts from 47 kg for the following 12 weeks increased the final body weight by 16 kg irrespective of whether the diet contained low, intermediate, or high dietary CP and Lys concentrations . Gilts fed the lowest feed supply gained on average 3.7 mm of backfat during the 12 weeks, while gilts fed semi-ad libitum gained on average 5.9 mm, and this was also not affected by dietary CP and Lys levels. In gilts with a mean initial weight of 62 kg, increasing the maximum feed allowance from 2.9-3.25 kg/day using a diet with 12.2% CP and 5.6 g SID Lys/kg led to an 8 kg increase in live weight and 0.6 mm increase in backfat thickness after 11 weeks of treatment (Klaaborg et al., 2019). Van Vliet et al. (2016) also demonstrated that ad libitum feeding with either low-or high-CP diets increased body fat content as well as backfat thickness at first service compared with restricted feeding using intermediate CP levels. Results, therefore, illustrate that diets low in CP and Lys fed ad libitum are superior in terms of favoring body fat at the expense of protein accretion. Having leaner genotypes, these feeding regimes provide opportunities to ensure sufficient backfat thickness at first service without excess weight gain. A further argument for focusing on the body fat pool at first service is the discovery of the appetite regulation hormone, leptin, which is synthesized within the adipose tissue. As reviewed by Barb et al. (2008), leptin affects the neuroendocrine axis, being responsible for upregulation of gonadotropin-releasing hormone secretion by the hypothalamus, thus affecting the secretion of luteinizing hormone, which is a modulator of the onset of the first estrus. Furthermore, leptin secretion seems to be affected by estradiol concentrations in the later part of the rearing period (Qian et al., 1999), with estradiol affecting the response to leptin around timing of first estrus (Barb et al., 2005). The reproductive benefits of leptin, as described by Barb et al. (2008), may be compromised by genetic improvements that have led to leaner gilts. Thus, nutritional manipulation of growth to favor storage of backfat during the rearing period is of great interest, especially because achieving sufficient body fat pool at an older age (or weight) may negatively affect sow longevity.

| Lysine intake
Reducing the daily supply of CP and Lys will decrease the growth rate of gilts (Cia et al., 1998;Gill, 2006;Jørgensen & Sørensen, 1998). By diluting a diet to achieve a lower Lys to energy ratio at a fixed feed intake, Cia et al. (1998) found increased storage of backfat and a drastic decrease in ADG. A recent study found that providing semi-ad libitum feeding with access to either a low lysine/protein-feeding regime (5.6 g Lys/kg and 11.8% CP from 47 to 110 kg followed by 4.8 g Lys/kg and 10.8% CP from 110 kg onwards) or a high lysine/ protein feeding regime (9.8 g Lys/kg and 15.0% CP from 47 to 110 kg followed by 6.5 g Lys/kg and 12.1% CP from 110 kg onwards) increased both the ADG and body fat mass after 12 weeks of treatment  even though backfat thickness was not affected. Controlling the daily intake of Lys may be an efficient way to attenuate the ADG (Cline et al., 2000), and ideally, it should be done in combination with an increased energy supply. A link between low CP (and Lys) and behavioral disorders in terms of ear and tail biting has been observed (Meer et al., 2017). It could be that providing diets very low in CP and Lys needs to be accompanied by increased feeding levels because the growing gilts will try to meet their daily nutrient requirements and growth potential by increasing their feed intake.

| Nutritional flushing
Flushing of gilts before first mating is commonly applied and leads to more eggs being released (Flowers et al., 1989), which in turn increases litter size . The effect of flushing may depend on the energy status and backfat of the gilt as Bruun et al.
(2021) showed a tendency for an interaction between backfat thickness and parity, suggesting that leaner gilts were most sensitive to flushing when mated in their second heat. The impact of flushing multiparous sows on the number of eggs being released seems not to have received much scientific attention, but in general, it is believed that flushing is most effective in young and lean sows/gilts as shown for first and third parity sows (Kirkwoord & Thacker, 1989).

| Physiological and nutritional aspects
Pregnant sows are physiologically comparable to growing/finishing pigs during early-and mid-gestation because nutrients are mainly utilized by the sow body (for maintenance) and maternal growth, while only limited amounts of nutrients are utilized for reproductive purposes (Sola-Oriol & Gasa, 2017). This changes gradually with the progress of gestation, and in the last third of gestation, pregnant gilts and sows devote substantial energy and amino acids (AA) towards fetal growth, growth of placenta, conceptual fluids and membranes, and mammary growth (National Research Council, 2012;Noblet et al., 1990). As parturition approaches, mammary growth becomes important and the sow starts to produce colostral proteins , which will be discussed further in the section for transition sows. As is the case in growing gilts, nutrition of gestating gilts and sows in early-and mid-pregnancy should not aim at maximizing the maternal growth of the reproductive females, but merely attempt to keep the body weight at a fairly low level to avoid excessive muscle mass, because it is associated with extra feed needed for maintenance, lower feed efficiency, and reduced longevity.

| Feed and energy intakes
Data on the impact of feeding strategies on the performance and energy utilization of pregnant sows is sparse, and most knowledge is based on experiments focusing on reproductive output like litter size at birth and piglet birth weight. However, these traits do not really reflect the consequences of improper feeding, because the sows highly prioritize the nutrient allocation towards their offspring and use their body as a buffer of nutrients in case the feeding strategy or dietary composition is suboptimal. A Danish study tested a simplified feeding strategy (2.4 kg/day throughout gestation) in comparison with a traditional feeding strategy (low supply in early and midgestation [2.1 kg/day], followed by higher supply in late gestation [3.3 kg/day]; Nielsen and Danielsen, 1983), that is, from Day 84 of gestation and onwards. The simplified strategy compromised litter size at birth (10.2 liveborn) as compared with the traditional feeding strategy (10.9 liveborn). It should be emphasized that the study was performed with low-prolificacy sows. In a follow-up study performed with hyper-prolific gestating gilts and sows fed a common gestation diet (containing 13.32 MJ ME/kg), the impact of two different feeding levels supplied to gestating gilts (2.4 and 3.3 kg/day) and three different feeding levels supplied to multiparous gestating sows (2.4, 3.3 and 4.2 kg/day) during the last third of gestation was tested (Sørensen, 2012). No zootechnical variables (e.g., backfat, live weight, weight gain) or indicators of alterations in body composition were reported for the gilts or sows and only a minor increase in mean birth weight of piglets was observed in response to a higher feed supply in sows (1.34, 1.36, and 1.37 kg/liveborn piglet, respectively, for 2.4, 3.3, and 4.2 kg feed/day; p < 0.05). The corresponding values in gilt progeny were mean birth weights of 1.23 and 1.25 kg/liveborn piglet when gilts were fed 2.4 and 3.3 kg/day. These findings emphasize that gilts and sows use their own body to buffer inadequate nutrient supply because 2.4 kg/day is insufficient to cover the energy requirement in late gestation (Noblet et al., 1990). It should also be mentioned that both studies were performed with sows crated throughout gestation and taking into account that sows can no longer be crated during all of pregnancy, it has become common practice to supply more feed to pregnant sows (0.3-0.5 kg/day) to cover energy expenditure for physical activity (Noblet et al., 1993) in group-housed sows. It should be stated that although sows have increased energy requirements as gestation progresses, there is a lack of data showing that phase feeding gestating sows increases (re)productive responses.

| Lysine intake
Several studies have been performed to elucidate the Lys requirement of pregnant sows by quantifying nitrogen retention using nitrogen balance. The Lys requirement increases as gestation progresses because more Lys is steadily needed for reproductive purposes (Sola-Oriol & Gasa, 2017). A recent study (Ramirez-Camba et al., 2020) reported that maximal Lys utilization efficiency was achieved in pregnant gilts at 7.2, 9.1, and 13.5 g SID lysine/day in THEIL ET AL. | 519 early-, mid-, and late-gestation, respectively. They also found that maximal retention of Lys was achieved with a higher Lys supply, namely at 8.5, 10.5, and 20.9 g SID lysine/day in early-, mid-, and late-gestation, respectively. Finally, they reported the overall efficiency of SID Lys utilized for retention in pregnant gilts, which was highest in early gestation (0.65), lowest in mid-gestation (0.38), and intermediate in late gestation (0.52). This change most likely reflects that pregnant gilts still prioritize maternal gain in early gestation, which declines as gestation progresses, whereas Lys retention increases in late gestation, leading to a greater Lys efficiency. Shi et al. (2016) reported SID Lys requirements of 14 g/day from Days 0 to 80 of gestation, and 21 g/day from Day 81 of gestation until farrowing due to increased mammary and fetal growth in late gestation. Samuel et al. (2012) reported somewhat lower requirements of 9.4 g/day in early gestation and 17.4 g/day in late gestation, but it should be emphasized that they reported total Lys instead of SID Lys and their study focused on multiparous sows fed semisynthetic diets. Woerman and Speer (1976) reported an average Lys requirement of 7.5 g/day (total lysine) for the entire gestation, while Dourmad and Etienne (2002) reported a Lys requirement of 10.5 g SID/day, no information on the gestation stage was provided in this study. Most likely, the SID Lys requirement does not exceed 8-10 g/day in early gestation and increases to 18-24 g/day in late gestation, but this deserves to be studied in hyper-prolific sows. As mentioned above, it should be stressed that although sows have increased lysine requirements with the progress of gestation, there is a lack of data showing that phase feeding gestating sows increases (re)productive responses.

| Physiological and nutritional aspects
The transition period is the period around parturition where sows shift from late gestation to early lactation. Although this term is often used nowadays, there is no consensus on the duration of the period.
Most tend to agree that the transition period includes the last 5-7 days of gestation and the first 3-5 days of lactation, and it is characterized by a dramatic change in nutrient requirements from day to day . Sows become catabolic during this period because of the imbalance between intake and output of nutrients due to rapid fetal growth and mammary development, the farrowing process, colostrum production, and the onset of copious milk production after parturition. Thus, transition sows undergo profound metabolic changes . Sows are physically moved from gestation stalls, where they are grouphoused, into farrowing pens where they are crated individually. This may give rise to some behavioral concerns in addition to nutritional challenges. As sows approach farrowing, they undergo physiological changes due to changes in reproductive hormones. This period is also characterized by the transition from intrauterine to extrauterine life for the offspring, and piglet survival is prominently challenged during birth and the first few days postpartum. Regardless of the dramatic shift in nutrient requirements and several physiological changes during the transition period, nutrition of transition sows has received little scientific attention. The aim of transition feeding should be to meet the rapid changes in daily nutrient requirements of the sows to improve farrowing performance and colostrum yield (CY), which are of paramount importance for minimizing stillbirth rate (SR) and early piglet mortality, respectively.

| Farrowing duration and piglet survival
Farrowing is likely one of the most stressful and painful phases of the reproductive cycle in sows and is certainly life-defining for the offspring. Piglets that experience an extended birth interval or extended farrowing duration most likely have an increased risk of being stillborn or born less vigorous (Feyera et al., 2018) and may not be able to compete with littermates, therefore increasing their chances to die shortly after birth or grow considerably slower. The positive correlation between farrowing duration (FD) and SR is well established (Holm et al., 2004;Peltoniemi et al., 2014). Farrowing is an energy-demanding process (Vallet et al., 2013) and insufficient energy status of sows during parturition will slow down the farrowing process (Feyera et al., 2018). A longer farrowing with less intense uterine contractions is distressful and compromises piglet survival due to an increased rate of umbilical rupture and meconium-stained intrapartum stillbirths (Olmos-Hernandez et al., 2008). Therefore, a fast farrowing achieved by high-intensity uterine contractions is crucial to optimize piglet survival, and sow nutrition during the transition period plays a vital role.

| Feed and energy intakes
Sow energy requirements increase during the transition period  because of the exponential growth of the fetuses, extensive mammary growth in late gestation, nest-building, farrowing process, and colostrum production. The importance of sow energy status on the farrowing kinetics and piglet survival was demonstrated recently using a multicatheterized sow model and strategic feeding with the elevated intake of fiber before the onset of farrowing (Feyera et al., 2018). This study revealed that prolonged FD increases the SR and that energy status, as indicated by plasma glucose and the time from the last meal until the onset of farrowing, was a recognized key component governing this process. Sows that initiate their farrowing shortly after their last meal had high concentrations of plasma glucose to fuel the farrowing process, leading to a minimal need for farrowing assistance and, ultimately, reduced SR.
Infusion of additional energy using a 10% glucose solution from the onset of nest-building until 24 h postpartum reduce the farrowing assistance from 21.0% to 9.0% (p = 0.01) and SR from 16.1% to 7.4% (p < 0.05; Nielsen et al., 2021). This finding supports the importance of energy supply and energy status of sows at farrowing for the farrowing process and survival of the neonates (Feyera et al., 2018).
Shorter FD, lower farrowing assistance, lower SR, and increased piglet survival within 24 h after birth have been reported in sows supplemented with extra energy on the day of farrowing (van Kempen, 2007;Oliveira et al., 2020). Energy from cornstarch (32%), fed from Day 85 of gestation until farrowing, reduced FD and SR when compared with energy originating from soybean oil or fish oil at 3% inclusion (Y. Yang, et al., 2019). Conversely, supplementation of 5% soybean oil during the same period as in the study of Y. Yang, et al. (2019) led to lower SR when compared with an 11.3% cornstarch supplemented group (Quiniou et al., 2008). Recently, we reported shorter FD, minimal farrowing assistance, and lower SR when sows were fed 3.7 to 4.1 kg/day during the last week of gestation, whereas both inadequate and excess energy affected the farrowing process negatively . In support of this, increased feed intake from 3.0 to 3.2 kg/day (29-31 MJ NE/day) and fed from Day 90 of gestation until farrowing reduced the FD, although the SR was barely affected (L. Q. Che et al., 2019). Other studies also reported a disadvantage of excess feed intake on FD and SR, which was most likely because the birth canal becomes blocked physically with undigested fiber when sows are overfed (Gourley et al., 2020;Liu et al., 2020;Miller et al., 2004).
Physical blockage may also be part of the explanation for farrowing problems at low feed intakes or low intake of fiber because constipation prevents piglets from being born in due time. Therefore, available data suggest that adequate energy supply during the transition period boosts the energy status of the sows and ensures a successful farrowing, which in turn maximizes the number of piglets being born alive.

| Lysine intake
The supply of the optimal amount of CP in the diet of gestating sows is important to meet the increasing demand for fetal and mammary growth in late gestation. However, if the intake of dietary CP is above the requirement, it is used as an energy source which may be detrimental to the farrowing process (Tydlitát et al., 2008). In that last study, a linear increase in FD was reported in response to increased dietary CP concentrations (4.5, 5.8, 6.5, and 8.6 h in sows fed 13%, 15%, 18%, and 21% CP, respectively). Hence, there was a considerable increase in FD from 4.5 to 8.6 h when more CP was utilized as fuel. Concomitantly, that study demonstrated greater SR in sows fed 18% or 21% CP (16.1% and 19.8%, respectively; p < 0.05) than in sows fed 13% or 15% CP (8.8% and 11.2% SR, respectively).
In another study, the greater number of liveborn piglets was greater in sows fed 13.5% CP from mating until farrowing as compared with sows fed 17% (Sabioni et al., 2007). In further support, Pedersen et al.
(2020) reported a numerically greater SR (8.2%) in sows fed a standard lactation diet with 16% dietary CP during the last week of gestation as compared with sows fed a standard diet lower in dietary CP (12% CP) which had an SR of 6.4%. Increasing intake of AA either throughout gestation (Shi et al., 2018)  with sows fed 20.0 g/day. In general, it appears that oversupply of CP compromises the FD much more than oversupply of one or few individual AA. Therefore, to improve the farrowing performance, it is highly recommended to examine explicitly the dietary concentrations of CP and the first limiting AA, which is normally Lys, to maximize the utilization of dietary CP and AA for reproductive purposes, and in turn, minimize the utilization of CP as a fuel. The mechanism whereby high dietary CP affects FD and SR seems to be associated with the oxidation of protein, which is an energetically unfavorable process . Therefore, to decrease the adverse effects of high dietary CP on FD and SR, determining the CP and Lys requirements of transition sows is needed.

| Colostrum production
Colostrum is the first secretion of mammary glands and plays a vital role in the short-term survival of piglets. Production of sufficient amounts of colostrum is a big challenge nowadays due to the continuous selection for large litters. It was widely accepted that colostrum was produced before the farrowing started (Csapó et al., 1996;Hartmann et al., 1997;Jönsson, 1973). Today, we know that the majority of colostral lactose and fat is produced after the first piglet is born (Feyera et al., 2019). The capacity of the mammary glands to produce colostrum likely depends on the number and efficiency of functional mammary epithelial cells developed during late gestation (Farmer, 2018;Head et al., 1991;Rezaei et al., 2016), and mammary development during this period is sensitive to nutritional manipulation.
However, data on the impact of nutritional manipulation during the transition period on mammary growth and consequently on colostrum production is lacking. Over the past 13 years, we have conducted 14 different experiments on colostrogenesis in an attempt to improve sow CY through different dietary interventions, including dietary concentrations of fat, fiber, protein, energy, feeding level (i.e., daily energy supply), glucose infusion, and body condition. It is evident that over the past 13 years, the number of liveborn piglets has increased from 12.5 (experiment 1; Exp.1) to 21.7 (Exp. 14) due to genetic selection for hyper-prolificacy ( Figure 1a). Concomitantly, the sow CY has increased from 5.3 to 6.9 kg per sow (Figure 1b), and that increase seems to be associated with improved sow productivity because the litter size exceeds the number of functional mammary glands. In support, Krogh

| Feed and energy intakes
In a study by Heo et al. (2008), sows were fed slightly increased levels of energy (39.9, 41.7, and 42.6 MJ ME/day) from Day 80 of gestation until weaning. Accordingly, elevated energy intake increased the fat (p = 0.02) and lactose (p = 0.04) concentrations in colostrum. However, these findings were in contrast with those reported by Feyera, Skovmose, et al. (2021), who found no impact of feeding levels on colostral content of lactose in spite of a great range of feed supply (from 1.8 to 5.0 kg/day). Moreover, that latter study indicated that increased feed supply lowers the colostral fat concentration, which was in contrast to the study by Heo et al. (2008) but in line with what is normally observed for sows during lactation as discussed below. Most likely, limited feed intake in late gestation stimulates fat mobilization and in turn increases the colostral fat content. In a study by Y. Yang, et al. (2008), sows were fed increasing dietary energy density (13.7, 13.9, and 14.2 MJ ME/kg) but similar energy supply on a daily basis from Day 80 gestation until weaning, and they reported no impact of energy concentration on colostrum composition.
With respect to CY, sows tended to have greater CY (4.0 vs. 3.5 kg, respectively; p = 0.07) and piglets had a greater colostrum intake (239 vs. 200 g/piglet, respectively; p = 0.02) when sows were fed 4.5 kg/day compared with sows fed 1.5 kg/day from Day 108 of gestation until Day 3 of lactation (Decaluwé et al., 2014). In support of this, a dose-response study where sows were fed 1.8, 2.4, 3.1, 3.7, 4.3, or 5.0 kg/day during the last week of gestation showed that CY was lowest in sows fed only 1.8 kg/day . Fitting a curvilinear response in the latter study indicated that CY was maximized at 3.1 kg/day, whereas the farrowing process was maximized at 3.7-4.1 kg/day. This indicates that sows prioritize colostrum above the farrowing process, but it most likely also reflects that colostrum production is sensitive to an optimal supply of energy and to feed composition, whereas the farrowing process is mainly sensitive to an optimal supply of energy. The fact that CY was not linearly increased with increasing feed intake in late gestation might suggest that a high feed intake above 3.1 kg could impair the proliferation of milk-producing mammary cells even though data are lacking on the proliferation of mammary cells during the transition period. Therefore, it seems that both low and high feed intake have adverse effects on CY, in which the mode of action could be through the limited supply of dietary nutrients at a low feed intake and compromised proliferation of mammary cells at a high feed intake . Decaluwé et al. (2014) reported greater lactose and lower CP concentrations in the colostrum of sows fed 4.5 kg/day compared with sows fed 1.5 kg/day. As glucose is the major precursor for lactose synthesis, the greater lactose output observed at a high feeding level may suggest that more glucose was available and taken up by the mammary glands, which allowed more lactose to be produced. This, in turn, would increase both the CY and colostral concentration of lactose although colostral lactose is the least variable colostral component and is hardly influenced by the diet (Decaluwé et al., 2013).

| Lysine intake
Several studies reported differential impacts of AA intake on colostrum production while studies looking at CP level is very limited. In the study by Jang et al. (2014), sows were fed 11%, 13%, 15%, and 17% CP throughout gestation, and no evidence of dietary CP on colostrum composition was reported (CY was not reported in that study). Supplementation of the leucine metabolite β-hydroxy-βmethylbutyrate (HMB; 2.5 g/day) from Day 108 of gestation until weaning increased colostrum intake (512 vs. 434 g/piglet, respectively; p < 0.05) when compared to the control group, while colostrum composition was not affected . Interestingly, this latter study showed reduced piglet mortality during the colostral period (0.0% vs. 4.8%; respectively; p < 0.05) in response to HMB supplementation. This finding confirmed the importance of high colostrum intake on early piglet survival. Nissen et al. (1994) also observed an improved weight gain in piglets suckling sows fed a diet containing HMB during the last 4 days of gestation. A linear increase in DM and CP contents of colostrum was reported in sows fed increasing levels of Lys (0.46%-0.74%) from Days 30 to 110 of gestation (Zhang et al., 2011) or from Day 80 of gestation until weaning (Y. Yang et al., 2008;Y. X. Yang, et al., 2009), but again, the CY was not estimated in these studies.
Supplementation of 25 g/day arginine from Day 30 of gestation until weaning decreased lactose (4.2% vs. 3.9%, respectively; p < 0.05) and increased DM (21.2% vs. 23.1%, respectively; p < 0.05) contents of colostrum at 12 h after farrowing without affecting CY (Krogh, Oksbjerg, Purup, et al., 2016). Neonatal survival and piglet growth largely depend on the quantity of colostrum and not the quality, but because it is both time-consuming and labor-intensive to weigh piglets at birth and 24 h after the onset of farrowing, it is much more common to study the dietary impact on colostrum composition than on CY. L. Che et al. (2020)

| Physiological and nutritional aspects
The balance between piglet demand for milk and the sow's capacity to produce milk is a result of complex regulation. The piglets' demand for milk is crucial in determining milk production, as illustrated by the 30% increase in litter growth when six glands were alternately suckled by two groups of six piglets at 30 min intervals instead of only one group of six piglets (Auldist et al., 2000). At the same time, the capacity of the sow to produce milk may also limit milk production. Indeed, growth rates of artificially reared piglets may exceed 400 g/day during the initial 23 days of postnatal life (Boyd et al., 1995), while the growth of sow-reared piglets averages 250 g/day. The regulation of mammary gland growth during lactation plays a central role in this balance between piglet demand and sow capacity to produce milk . Indeed, there is a 100% increase in the total amount of mammary gland DNA during the first 3 weeks of lactation (S. W. Kim et al., 1999), a positive correlation between mammary gland weight and piglet growth (S. W. Kim et al., 2000), and a two-thirds reduction in mammary gland weight within 1 week after weaning (Ford et al., 2003). Numerous factors, such as systemic and local hormone levels, nutrient supply by the blood, cell proliferation and apoptosis, and the capacity of suckling piglets to remove milk from the glands are involved in this regulation . Accordingly, the level of milk production is a key factor to consider when feeding modern lactating sows. The supply of energy and Lys are considered the first and second key nutritional elements for pigs and they need to be understood to be met for optimal performance level while minimizing the loss of body reserves and maximizing feed efficiency. In this context, it is central to understand the nutritional relationships between feed intake, mobilization of body reserves, and milk production when feeding lactating sows.

| Feed and energy intakes
Milk production is the highest prioritized trait in lactating sows, and sows are capable of mobilizing substantial amounts of body reserves to sustain milk production while compensating for an insufficient dietary nutrient supply (King & Dunkin, 1986). Feed intake is stimulated by greater litter size and greater litter gain (Eissen et al., 2000). However, in lactating sows, the energy and Lys requirements for milk production and maintenance commonly exceed their dietary intake  and this is even more pronounced for modern hyper-prolific sows. Accordingly, the imbalance between the intake of dietary nutrients and nutrients required for maintenance and milk production determines how much energy needs to be mobilized from body reserves. Consequently, a high feed intake capacity and optimal composition of the feed are crucial variables, especially for high-prolific and high-performing sows. Feed intake is controlled by several other parameters than milk yield, such as feed composition, body condition, parity, ambient temperature, and genotype (Eissen et al., 2000). Table 1 illustrates the energy and Lys balances in sows with the 25% lowest litter growth, average litter growth, and the 25% highest litter growth during 28 days of lactation.
It can be seen that the intakes of energy and Lys are insufficient to cover the demand for both low-and high-performing sows. The outputs of energy and Lys are 40%-50% greater in the best performing sows than in sows having the poorest performance.
Intakes of energy and Lys also increase with performance level, but THEIL ET AL. | 523 T A B L E 1 Sow performance and the balance of energy and lysine in low-, average-, and high-performing sows during 28 days of lactation. Note: Low-and high-performing sows were characterized as the 25% sows with the lowest or highest litter weight gain from Days 2-28 in lactation. Data originated from four different studies with high-prolific sows (Krogh, Bruun, et al., 2017;Krogh, Oksbjerg, et al., 2017;Pedersen et al., 2016;Zhou et al., 2018). a Estimated based on average litter gain and litter size according to Hansen et al. (2012). b Intake was calculated to feed intake multiplied by the content in feed; average energy content was 13.6 MJ metabolizable energy/kg feed in the diets and average lysine content was 6.7 g standardized ileal digestible lysine/kg feed. c Energy requirement for maintenance: 0.460 MJ ME/kg BW 0.75 (Dourmad et al., 2008); Lysine requirement for maintenance was assumed to be 2.5 g/day (National Research Council, 2012). d Milk output was calculated as milk yield multiplied by milk composition; Milk yield was estimated based on litter size and litter gain (Hansen et al., 2012); Milk energy content was calculated as milk fat, g/ kg × 0.0389 MJ/g + lactose, g/kg × 0.01654 MJ/g + milk protein, g/kg × 0.0239 MJ/g (Hansen et al., 2012); Milk lysine content was calculated assuming that milk lysine constituted 7% of milk protein content (Krogh, Oksbjerg, et al., 2017). e The efficiency of energy and lysine for milk production was assumed to be 78% (Theil et al., 2004) and 88% (Hojgaard, Bruun, & Theil, 2019), respectively.
their intake is only around 20% greater in high-performing sows compared to the lowest-performing sows. However, as previously discussed, feed intake is affected by several factors, and average feed intake across the lactation period may exceed 8 kg/day and reach 11 kg/day at peak lactation (Thingnes et al., 2012). In a recent Danish study, it was shown that sows with a high milk yield were characterized by having both a high feed intake and a high body mobilization . Thus, focusing on achieving high feed intake during lactation seems to be central when feeding lactating sows but we lack the understanding of how milk production may be stimulated around farrowing. Although sows may buffer insufficient feed intake by increasing body mobilization, milk production is attenuated by an insufficient supply of feed (and energy), as indicated by a reduced piglet gain with increased sow body weight loss during lactation (Figure 2a) in response to insufficient energy supply (Figure 2b). Milk production appears to be compromised when sow body weight losses exceed 10% and energy is deficient by more than 15 MJ/day. This is in agreement with the study by Clowes et al. (2003), who found that piglet performance was compromised when mobilization of body protein exceeds 10% of the body pools at parturition (Clowes et al., 2003).
F I G U R E 3 Relationships and Pearson correlations (r) between sow energy balance during the lactation period and the concentration of milk fat, lactose, and milk protein. Energy balances were calculated based on a data set consisting of 114 treatment means from 31 published studies with lactating sows that all reported feed intake and composition, backfat and body weight change, litters size and growth, and milk composition during the lactation period. The average, minimum, and maximum values for each included study are illustrated in Table 2.
| 525 T A B L E 2 Average, minimum, and maximum levels of milk fat, milk lactose, milk protein, and the calculated energy balance from studies reporting feed intake and composition, backfat and body weight change, litter size and growth, and milk composition during the lactation period.  sustaining productivity in modern sows. To achieve high feed efficiency of lactating sows, it is crucial that they produce as much milk as possible directly from the feed and minimize the amount of milk being produced from body reserves (Pedersen et al., 2016;Theil et al., 2020).

| Lysine intake
The supply of protein and essential AA is critical for milk production.
Studies using high-prolific lactating sows have focused on the dietary requirements of essential AA or protein (Hojgaard, Bruun, & Theil, 2019a, 2019bHojgaard, Bruun, Strathe, et al., 2019;Strathe, Bruun, Geertsen, et al., 2017). From these studies, it is clear that if sows receive insufficient dietary Lys on a daily basis, the milk production decreases as indicated by daily compromised litter gains.
At the observed performance level (around 13 weaned piglets and 3 kg daily litter gain), a daily Lys supply above 55 g SID Lys had no effect on piglet ( Figure 4a) or litter (Figure 4b) growth. The efficiency of dietary Lys used for milk synthesis was estimated from the same studies (Figure 4c), showing efficiencies of 82% when intake and requirements were balanced, decreased to 50% when Lys intake exceeded the requirement for maintenance and output in milk. Most recently, it was found that hyper-prolific sows were able to utilize dietary SID Lys with an efficiency as high as 88% (Hojgaard, Bruun, & Theil, 2019b). This illustrates the importance of adjusting feed composition according to the observed feed intake level and milk production potential of sows to maximize productivity and feed efficiency concomitantly. This is challenged by the large variations in intake and requirements among individuals and herds, but it is a central aspect to optimize feeding of lactating sows. Currently, sow lactation diets are formulated based on a recommended feeding curve and an ideal AA profile (Hojgaard, Bruun, & Theil, 2019a) without paying attention to the production level. Including herdspecific information such as feed intake and litter gain, or even more simple criteria such as litter size of the nursing litter, would make it possible to optimize sow performance and feed efficiency, while minimizing the amount of feed per kilogram of piglet produced.

| CHALLENGES/FUTURE IMPLICATIONS
Research is needed to understand the importance of body fat pools in gilts with respect to their future reproductive output and to study diet composition and feeding strategies favoring fat retention. During for growing gilts and multiparous sows during early and midgestation when body condition needs to be restored should aim at reducing maternal growth and protein retention and favor fat retention to counteract the genetic strive for protein retention.
Nutrition during the transition and lactation periods needs to be understood better because these periods greatly influence litter size and litter weight at weaning and hence, overall sow productivity.
Clarifications of feed efficiency for growing gilts, and sows during gestation, transition, and lactation are also needed to minimize the nutrient input and maximize nutrient utilization for productive and reproductive outputs. Such practices also minimize the negative impact of pig production on the environment with respect to climatic and environmental pollution.

ACKNOWLEDGMENT
Acknowledgment statement is not applicable as the paper is a review.
Also, no funder has supported this study.

CONFLICTS OF INTEREST
The authors declare no conflicts of interest.