Key words: colostrum intake, colostrum yield, feed intake, lactation, sow, weaning weight

Transcription

1 Understanding the drivers of improved pig weaning weight by investigation of colostrum intake, sow lactation feed intake, or lactation diet specification 1 A. Craig,* 2 A. Gordon,* and E. Magowan* *Agri-Food and Biosciences Institute, Hillsborough BT26 6DR, UK; and Queens University Belfast, University Road, Belfast BT7 1NN, UK ABSTRACT: Modern sows have low feed intake (FI) during lactation. The main aim of this study was to understand interactions between and separate effects of FI and nutrient density on litter weaning weight (WW). Key drivers of colostrum intake (CIn), piglet survival, WW, and colostrum yield (CY) were also investigated. Sows (n = 82) were offered a High (15.8 MJ/kg DE; 1.3% total lysine) or Normal (15.2 MJ/kg DE; 1.28% total lysine) specification lactation diet at either a High (feed allowance increased by 0.5 kg/d after farrowing until intake reached 10 kg/d) or Low (feed allowance was increased by 0.3 kg/d after farrowing until intake reached 7.5 kg/d) feeding level (2 2 factorial design). A subset of sows (n = 18) were observed during farrowing to collect data on factors affecting CIn. No interactions were found between diet specification and feeding level. Sows on the Low feeding level lost 10.6 kg more BW during lactation than those on the High feeding level (P < 0.001). Sows offered the High specification diet lost 6.4 kg more BW than those on the Normal specification diet (P = 0.018). Diet specification had no effect on ADFI. Between birth and weaning, litters of sows offered the High feeding level grew 326 g/d faster (P < 0.001) and were heavier at 28 d (114 kg; P < 0.001) compared with those of sows offered the Low feeding level (104 kg). Although litters from sows offered the High specification diet had WW similar to that of litters from sows offered the Normal specification diets, their ADG was 190 g/d greater (P = 0.018) between birth and weaning. A regression analysis was completed using data from 192 sows and indicated that FI and lysine intake throughout lactation and DE and lysine intake from 14 to 28 d of lactation were the main drivers of litter WW. Lactation efficiency was 0.65 from 0 to 7 d and decreased to 0.42 from 21 to 28 d. Variation in CIn was mainly explained by 24-h weight, birth weight, and the duration of farrowing. Colostrum yield was significantly correlated (P = 0.004; pseudo R 2 = 54.5%) with litter birth weight. Piglet WW was positively correlated with 3-wk weight (P < 0.001) but negatively correlated with sow parity (P = 0.035), number born alive (P = 0.045), and being female (P < 0001). Out of 45 variables, preweaning piglet survival was positively correlated (P = 0.008) with only 24- to 48-h weight gain. In conclusion, lactation FI and DE and lysine intake in the second half of lactation were the main drivers of litter WW. Key words: colostrum intake, colostrum yield, feed intake, lactation, sow, weaning weight 2017 American Society of Animal Science. All rights reserved. J. Anim. Sci : doi: /jas The authors gratefully acknowledge the Department of Agriculture Environment and Rural Affairs (DAERA) and Pig Regen Ltd. for funding this project. The authors declare that there are no conflicts of interest. 2 Corresponding author: aleslie02@qub.ac.uk Received June 5, Accepted August 14, INTRODUCTION There are a number of sow and litter characteristics that can influence colostrum and milk production. Colostrum production has been correlated with factors such as time to first suckle and gestation length (Declerck et al., 2015), whereas the number of piglets suckling the sow is the greatest driver of milk yield (Toner et al., 1996). Piglet birth weight (BiW) and vitality also improve milk consumption (Algers and Jensen, 1991; King et al., 1997). The increase in sow

2 4500 Craig et al. prolificacy in the last decade has been accompanied by genetic selection for improved feed efficiency in the growing and finishing periods of slaughter generation animals, but this improvement in feed efficiency has reduced the voluntary feed intake (FI) of sows (Bergsma et al., 2009). The ADFI of lactating sows in papers published from 2008 to 2014 was 5.7 kg/d (Craig et al., 2015a). However, sows are capable of consuming more, as intakes of 7.7 kg/d over a 28-d lactation have been noted (Craig et al., 2015b). As high FI result in high nutrient intakes, increasing the energy density of the diet may enable sows with low intakes to meet the challenge of greater production demands. A number of studies have improved litter weaning weight (WW) by increasing energy and lysine levels in lactating sow diets (Park et al., 2008; Yang et al., 2009; Walsh et al., 2012). Highenergy and lysine diets can also reduce sow weight and back fat loss (Park et al., 2008). Litter WW in these studies was <90 kg/sow; however, modern sows can produce up to 110 kg of litter weight in litters of over 12 pigs (Craig et al., 2015b). This level of production was achieved at high FI using high-nutrient-specification diets. The aim of this study was to understand the interactions and separate effects between FI and the nutrient density of lactation diets. A small colostrum study was also conducted to measure factors that affect colostrum intake (CIn), piglet WW and preweaning survival, and also colostrum yield (CY) of sows rearing large litters. MATERIALS AND METHODS The present study was conducted under the regulations of the Department of Health, Social Services and Public Safety of Northern Ireland in accordance with the Animals (Scientific Procedures) Act 1986 (Parliament of the United Kingdom, 1986). Animals Sows representing parities 2 to 6 (n = 82) were selected and allocated to treatment balancing for parity and BCS. Sows were PIC F1 cross (Large White Landrace). PIC 337 was the terminal sire used. Gestation Feeding and Management During the first 28 d of gestation sows were kept in groups of 4 in free-access cubicles with a m pen at the rear. After 28 d, sows were moved into a large dynamic group, where they were fed through a Nedap electronic sow feeder (Nedap Livestock Management, Groenlo, Netherlands) until d 108 of gestation. Sows were offered 2.5 kg/d of a gestation diet (12.9 MJ/kg DE, 148 g/kg CP, 7.0 g/kg total lysine) in early gestation, and from d 85 they were offered 3 kg/d of the same gestation diet. Lactation Feeding and Management Sows moved into farrowing accommodation at d 108 (±1 d) of gestation and were housed in crates with an enclosed heated creep area for piglets at the front. Temperature in the farrowing rooms and creep areas was electronically controlled, and daily temperatures were recorded. Temperature was set at 19 C for farrowing and reduced to 17.5 C after farrowing was completed. The temperature of the creep boxes was set at 30 C and reduced to 23 C when the piglets were 1 wk old. Sows had access to a wet and dry feeder and were offered 3 kg/d of their respective lactation diet from d 108 up to and on the day of farrowing. Feed allowance was recorded individually on a daily basis, and the daily allowance was weighed and offered manually across 2 meals. Feed allocation was stepped up only if sows had eaten their previous allowance and left the trough sufficiently clean. Feed disappearance was recorded as FI. Sows were induced to farrow with 2 ml of Planate (cloprostenol, Intervet/Schering-Plough Animal Health, Boxmeer, Netherlands) on d 114 of gestation as is common practice on commercial farms in Northern Ireland. Piglets had their teeth clipped, tails docked, and an iron injection administered within the first 12 h of birth. Piglets also received ear tattoos to allow recordings of individual animals. Cross fostering was completed within 24 h of farrowing and used to standardize litters to 13 or 14 piglets. Cross fostering occurred within and between treatments as gestation management of the sows was standard. Mortalities were recorded after cross fostering had been completed, and piglets that died were not replaced. No creep feed was offered to the piglets, and sow troughs were sufficiently high to deter piglets from consuming much, if any, sow feed. However, piglets had free access to water via nipple drinkers. Piglets were tagged, corresponding to their ear tattoo, during wk 2, and they were weaned at 28 d. Experiment 1. Dietary Treatments. Dietary treatments began at d 108 of gestation, on entry to the farrowing crate. Dietary treatments were arranged in a 2 (High or Normal specification diet; Tables 1 and 2) by 2 (High or Low feeding level) factorial design. In the High feeding level, sow feed allowance was increased by 0.5 kg/d after farrowing until intake reached a maximum of 10 kg/d. Intake was capped at 10 kg/d to reduce the likelihood of sows overeating and becoming sickened, leading to food refusal and reduced lactation intake. The target ADFI over the 28-d lactation period was 7.5 kg. In the Low feeding regime, sow feed allowance was increased by 0.3 kg/d

3 Sow factors that improve pig weaning weight 4501 Table 1. Formulated and actual composition of experimental diets Formulated Actual Item Normal High Normal High DM, % DE, MJ/kg CP, % Oil B, % Oil A, % Total fiber, % Total lysine, % Total methionine, % Total threonine, % Total tryptophan, % Total valine, % after farrowing and capped at 7.5 kg/d. The target ADFI over the 28-d lactation period was 6 kg. The feed was manufactured by Devenish Nutrition (Belfast, UK). It was offered in small-pellet form (2 mm in diam.). Diet analysis was performed by Sciantec Analytical (North Yorkshire, UK). The DE was calculated as ( CP) + ( oil B) ( ash) ( NDF). Crude protein was determined using the Dumas method. Neutral detergent fiber was determined by enzymatic gravimetry. Measurements Table 2. Ingredients of experimental diets Ingredient Normal High Wheat, g/kg Maize, g/kg Soy (full fat), g/kg Soy (high protein), g/kg Soy oil, g/kg Vitamin and mineral mix, 1 g/kg l-lysine, g/kg l-threonine, g/kg l-valine, g/kg l-methionine, g/kg Tryptophan, g/kg Premix provided (per kg of finished feed) million IU vitamin A, million IU vitamin D 3, 0.12 g vitamin E, g vitamin K, g vitamin B 1, g vitamin B 2, g vitamin B 6, 0.03 mg vitamin B 12, g biotin, g calcium pantothenate, 0.02 g nicotinic acid, g folic acid, 0.2 g choline chloride, g iodine, g selenium, 0.1 g iron, g manganese, g copper, and 0.1 g zinc. The back fat depth, BCS, and BW of the sows were measured at transfer to the farrowing accommodation (108 d of gestation) and at weaning. Back fat depth was measured at the P 2 position (65 mm from the midline at the level of the last rib) with an ultrasonic back fat scanner (Pig Scan-A-Mode back fat scanner, SFK Technology, Smorum, Denmark). Body condition score was assigned using a 5-point scale, and half scores were also used, with 1 being thin and 5 being very fat. As BCS can be subjective, the same stock person consistently recorded all BCS measurements. Sow weight was recorded at 108 d of gestation, and the BiW of the piglets, including stillbirths and mummified piglets, was subtracted to calculate empty weight (as per Cottney, 2012). Piglets were individually weighed at birth and 7, 14, 21, and 28 d (weaning). Any piglets that died before weaning were weighed, and their gain was taken into account when calculating litter weight and ADG. Litter and piglet ADG, within-litter CV, and mean piglet weight for each litter was also calculated. Feed efficiency and DE efficiency were calculated for each treatment for 0 to 7 d, 7 to 14 d, 14 to 21 d, 21 to 28 d, and 0 to 28 d of lactation. The lactation efficiency data were analyzed using REML as per the experimental design of a 2 (diet specification) by 2 (feed level) factorial design. Sow was the experimental unit used, and data were blocked for time period (or replication). There was no difference between treatments, so treatments were combined, and a REML analysis was used to compare the effect of stage of lactation on efficiency measures. Statistical Analysis Treatment effects on sow performance and litter performance were assessed by analyzing the data as per the experimental design of a 2 (diet specification) by 2 (feed level) factorial design using REML. Sow or litter was the experimental unit used, and data were blocked for time period (or replication). The tests were run using an α value of 0.05 (5%). No covariates were used. Milk yield was calculated as piglet gain 4.2 (Van der Peet-Schwering et al., 1998). There was no interaction between time and treatment, so milk yield data were analyzed using ANOVA (unbalanced designs) with time period as the treatment structure and yield as the variate. All analyses were performed using Genstat (version 16; Genstat, VSN International Ltd, Hempstead, UK). Regression Analysis Combining the raw data from Craig et al. (2015b) and the current trial (n = 192 sows), individual sow FI and actual DE and lysine intakes were calculated for various time points (0 to 7 d, 7 to 14 d, 14 to 21 d, 21 to 28 d, 0 to 14 d, 14 to 28 d, and 0 to 28 d) using the values from the analyzed diet samples and daily FI. Both trials used animals of the same maternal and terminal genetics and were kept under the same husbandry and housing conditions. Two analyses were completed: 1) A metaregression analysis fitted using the REML method

4 4502 Craig et al. was used to compare the individual effects of all variates. 2) A multivariate regression analysis employing the method of forward selection with backward elimination was used to select the best model for WW from the list of all possible explanatory variables. Experiment was included as a random model in both analyses. Experiment 2: Colostrum Intake. A subset of sows from the Low feed allowance treatments representing both diets (n = 18) were observed during farrowing to collect data on sow and piglet parameters that may affect piglet CIn and WW. At each farrowing 2 sows (1 from each dietary treatment) that farrowed on the same day with similar BCS and back fat depth at P 2 were selected and were supervised through the duration of farrowing. Manual assistance at birth was given only when deemed necessary, and piglets were given no assistance after birth other than to save them from crushing. Measurements The time of the first piglet s birth was recorded as the start of farrowing, and birth interval was recorded for each subsequent piglet. At birth each piglet was observed and given a score for how long it took the piglet to attempt to breathe (<5 s, 5 to 10 s, or >10 s). Whether the cord was broken or still intact at birth, whether the piglet was born in a caul, and whether the piglet had an assisted birth were also recorded. Piglets were then weighed (including stillborn and mummified pigs), dried with a paper towel, numbered according to birth order, and placed back behind the sow. Time to suckle was recorded for each piglet. The birth of the last piglet was recorded as the end of farrowing. Piglets were weighed 4 h after farrowing had ended, and their rectal temperature and crown to rump were measured. Piglets were also weighed at 24 h from the end of farrowing, and place on the udder (posterior, mid, and anterior) was recorded. Piglets were weighed again at 48 h and 5, 7, 14, 21, and 28 d. Colostrum intake was calculated at 24 h using the following formula (Theil et al., 2014b): CIn = WG BiW D WG/D WG/BiW, where WG = weight gain between birth and 24 h, BiW = birth weight, and D = time from first suckle to 24 h. Colostrum yield for each sow was calculated as the sum of the CIn values for each pig in the litter. Statistical Analysis To find the variables that explained most of the variation in CY, CIn, and WW, a stepwise regression analysis employing the method of forward selection with backward elimination was used to select the best model for each response variable from the list of all possible explanatory variables (see Table 3). In each case this model was then refitted using linear mixed model methodology with time period as a random effect and all other explanatory variables as fixed effects. A further backward elimination procedure was then employed so that all remaining variables in the model were significant at the P < 0.05 level. To find the variable that best predicated if a piglet would survive, the same testing strategy was employed, except that it was implemented in terms of generalized linear and generalized linear mixed model methodology using a binomial distribution and logit link function in each case. RESULTS Impact of Dietary Treatments There were no significant interactions between diet specification and feeding level. Table 4 reports the effects on sow performance. Average lactation length was 28.4 d. Numbers after cross fostering (13.5) or at weaning (13) were not significantly different between treatments. Neither diet nor feed level affected litter CV at birth (0.175) or weaning (0.180). Feed level or diet specification had no effect on sow BCS loss or back fat loss. However, sows on the Low feeding level lost more weight (10.6 kg; P < 0.001) than those on the High feeding level. Sows on the High specification diet also lost more weight (6.4 kg; P = 0.018) than those on the Low specification diet. Lactation feeding level significantly affected ADFI, with sows on the High feeding level consuming 0.9 kg more per day on average over the 28-d lactation period (6.0 kg vs. 6.9 kg; P < 0.001). On the basis of the actual dietary concentrations of DE and lysine, the High feeding level delivered a greater daily intake of DE and lysine (P < for both). However, diet specification had no significant effect on DE or total lysine intake, although energy intake was numerically 3 MJ/d DE higher and lysine intake was numerically 0.62 g/d higher when the High specification diet was offered compared to the Normal specification diet. Table 5 details litter performance. Litter weight after cross fostering was, on average, 20 kg and was not significantly different between treatments. Between birth and weaning, litters of sows offered the High feed level grew 326 g/d faster (P < 0.001) than those of sows offered the Low feeding level. This difference resulted in heavier litter weights at the end of wk 2, 3, and 4 of lactation (P = 0.048, 0.004, and <0.001, respectively). The WW of litters from sows offered the High feed level was 9% greater (114 kg, P < 0.001) than that of those offered the Low feed level (104 kg).

5 Sow factors that improve pig weaning weight 4503 Table 3. Simple statistics for the 45 variables used in colostrum study 1 Variable Minimum Maximum Mean SD CIn CY WW Srvl Sow X X X X Diet (normal or high) X X X X Parity X X X X BCS at farrowing X X X X Sow weight, kg X X X X Total born X X X X Live born X X X X Stillborn X X X X Total litter weight, kg X X X X Live litter weight, kg X X X X Avg birth weight, kg X X X X Avg live birth weight, kg X X X X Birth CV X X X X Wean CV X X X X Pigs after cross fostering X X X X Pigs weaned X X X X Farrowing, min X X X X Birth order X X X Birth interval, min X X X Cord broken (Y/N) X X X Born in caul (Y/N) X X X Time to breath (1, 2, 3) X X X Individual pig birth weight, kg X X X Sex (boar/gilt) X X X Time to suckle, min X X X Length, cm X X X Rectal temperature at 4 h, C X X X Pig weight at 4 h, kg X X X Pig weight at 24 h, kg X X X 0 24-h weight gain, g X X Colostrum intake, g X X Colostrum yield, kg X X Teat position (top, mid, rear) X X Pig weight at 48 h, kg X X h weight gain, g X X 0 48-h weight gain, g X X Pig weight at 5 d, kg X X Pig weight at 7 d, kg X X Pig weight at 14 d, kg X X Pig weight at 21 d, kg X X Pig weight at 28 d, kg X Pig ADG B-7 d, g X X Pig ADG at 7 14 d, g X X Pig ADG at d, g X X Pig ADG at d, g X X 1 An X indicates a variable was included in a particular analysis. CIn = colostrum intake, CY = colostrum yield, WW = weaning weight, Srvl = survival. The ADG of litters suckling sows offered the High diet specification was 190 g/d greater (P = 0.018) than that of litters of sows on the Normal diet specification. However, litter WW was statistically similar (111 vs kg, respectively; P > 0.05). There was no interaction between treatment and time point on milk yield, so treatments were combined when comparing milk yield at different time points. Milk yield for each sow was calculated as litter gain 4.2 (Van der Peet-Schwering et al., 1998). Milk yield increased significantly (P < 0.001; SEM = 0.152) between d 7 (10.2 L) and d 14 (13.6 L). Yields on d 21 (15.4 L) and d 28 (15.7 L) were not significantly different. Sows did not peak in milk yield until d 26, and yield

6 4504 Craig et al. Table 4. Sow performance when using a Normal or High specification diet at a High or Low feeding level Normal High P-value Item High Low High Low SED Diet Level Interaction Farrowing BCS Weaning BCS BCS change Farrowing P 2, mm Weaning P 2, mm P 2 change, mm Empty weight, kg Weaning weight, kg Weight change, kg < Lactation length ADFI, kg < Lys intake, g/d < DE intake, MJ/d < Feed intake efficiency DE intake efficiency No. birth No. weaned Birth weight CV Wean weight CV Number after cross fostering. was 15.8 L at peak yield (Fig. 1). Sows had an average milk yield of 12.8 L/d over the 28-d lactation period. When the milk yield of the subset of 18 sows used in the colostrum study was measured, peak milk yield was on d 22 (15.1 L/d). These sows had an average milk yield of 12.1 L/d over the 28-d lactation period. Regression Analysis Table 6 reports the minimum, maximum, SD, and average of each variable used in the regression analysis. The P-value of the equation relating the variable to the litter WW is also reported. In the univariate analysis, 14- to 21-d DE intake (P = 0.034), 21- to 28-d DE intake (P = 0.046), 14- to 28-d DE intake (P = 0.014), and 0- to 28-d lysine intake (P = 0.011) were all positively correlated with litter WW. When the variates were combined, those that explained most of the variation in litter WW (kg) = (0-28d lysine intake, g) (14-28d Lysine intake, g) (14-28d DE intake, MJ) (0-28d FI, kg) The resultant equation was (psuedo R 2 = 9.4%). Litter WW (kg) = (0- to 28-d lysine intake, g) (14- to 28-d lysine intake, g) (14- to 28-d DE intake, MJ) (0- to 28-d FI kg) Lactation Efficiency There was no significant effect of treatment on lactation efficiency (Table 4). Lactation efficiency in terms of FI (kg) and DE intake (MJ) declined throughout lactation (Fig. 2). Feed intake efficiency declined from 0.65 to 0.42 between the first and last weeks of lactation. Similarly, DE efficiency declined from 0.42 to 0.27 between the first and last weeks of lactation. Colostrum Intake Out of 29 possible variates (Table 3), those that best explained the variation in CIn at 24 h were 24-h weight (24hW; P < 0.001), BiW (P < 0.001), and duration of farrowing in minutes (DF; P < 0.001) in combination. The equation was as follows: CIn (g) = (24hW, kg = 1271) (BiW, kg = 1070) (DF, min = ). Pusedo R 2 was Out of 17 possible variates (Table 3), the variable that best explained the variation of CY of the sow was litter weight at birth (LW; P = 0.004). The equation was as follows: CY (g) = (LW, kg = 286.8). Puesdo R 2 was

7 Sow factors that improve pig weaning weight 4505 Table 5. Litter performance when sows were offered A Normal or High specification diet at a High or Low feeding level Item 1 High Low High Low SED Diet Level Interaction Normal High P-value Litter BiW, kg Litter W at 1 wk, kg Litter W at 2 wk, kg Litter W at 3 wk, kg Litter WW, kg < ADG B-1 wk, kg ADG at 1 2 wk, kg ADG at 2 3 wk, kg ADG at 3 4 wk, kg < ADG B-W, kg < BiW = birth weight, WW = weaning weight. Out of 44 possible variates (Table 3), those that best explained the variation in piglet WW were 3-wk weight (3wkW; P < 0.001), sex (P < 0.001), sow parity (P < 0.035), and number of live-born pigs (LB; P < 0.045). Puesdo R 2 was The equation was as follows for boars: WW (kg) = (3wkW, kg = 1.166) (parity = ) (LB = ). The equation was as follows for gilts: WW (kg) = (3wkW, kg = 1.166) (parity = ) (LB = ) Out of 45 possible variates (Table 3), the variable that best explained the variation in piglet survival to weaning was 24- to 48-h weight gain (P = 0.008; constant = 2.374, slope = ). DISCUSSION Impact of Diet Specification and Feed Allowance In this trial, sows on the Low feeding level ate 0.9 kg less than those on the High feeding level. Sows on the High feeding level were expected to eat more than an average of 6.9 kg/d, and therefore, the target of 7.5 kg/d was not achieved. It could be that the high oil content in the diets reduced voluntary FI, as has been found before when using high-fat diets (van den Brand et al., 2000; Quiniou et al., 2008; Park et al., 2008). The High specification diet contained 37% more oil B than the Normal specification diet. However, it is surprising that if oil affected the intake that there was no interaction between diet and feeding level. Despite the smaller than expected increase in ADFI, a 15% increase in FI was still observed between the High and Low feeding levels, which resulted in a significant improvement in litter performance. Sows on the High feeding level had litters that were 3 kg heavier at 2 wk of age, 7 kg heavier at 3 wk, and 10 kg heavier at weaning. Sows on the Low feeding level reached their maximum allowance (7.5 kg/d) around d 15 of lactation, after which the difference in litter growth between the High and Low feeding levels became greater. Sows on the Low feeding level lost 10.6 kg more weight than those on the High feeding level. In fact, sows on the High feeding level actually gained weight slightly. Eissen et al. (2003) also reported that sows with lower lactation FI had increased BW losses. Sows on the High specification diet lost more BW than those on the Low specification diet (P = 0.018). Although these sows lost BW, they did not lose significantly more body condition or back fat. A 10-kg difference in weight would not have been noticeable on a BCS scale of 1 to 5, with half points. Also, if muscle degradation was the cause of the weight loss, it would not have affected back fat depth as that measures only fat cover. The High specification diet delivered an increase of 0.63 MJ/kg DE and 0.2 g/kg total lysine. Which improved litter increase ADG between 14 and 21 d and also from birth to weaning compared to that of the Normal specification diet but did not significantly impact litter WW. However, the High specification diet did not increase sow average daily intake of DE or lysine or affect ADFI, likely because the actual difference in the DE and lysine density of the diets was less than expected as the diets were formulated to have a difference of 0.8 MJ/kg DE and 0.6 g/kg total lysine. The final diets, therefore, may not have been divergent enough to significantly improve litter performance. The Normal specification diet, although a higher specification than is usually fed commercially, was formulated for sows rearing 13 pigs to 8.5 kg at 28 d using calculations from Whittemore et al. (2003). The High specification diet was formulated to

8 4506 Craig et al. Table 6. Variables used in regression analysis and their relation to litter weaning weigh (WW) in the univariate analysis Figure 1. (a) Milk yield of all sows (L/d). (b) Milk yield of 18 sows measured for colostrum yield (L/d). ). Variable 1 Minimum Maximum Average SD P-value Litter WW, kg d 0 7 FI, kg d 7 14 FI, kg d FI, kg d FI, kg d 0 14 FI, kg d FI, kg d 0 28 FI, kg d 0 7 DE intake, MJ d 7 14 DE intake, MJ d DE intake, MJ 607 1, d DE intake, MJ 228 1,350 1, d 0 14 DE intake, MJ 679 1,399 1, d DE intake, MJ 1,019 2,543 1, d 0 28 DE intake, MJ 2,071 3,889 3, d 0 7 Lys intake, g d 7 14 Lys intake, g d Lys intake, g d Lys intake, g 192 1, d 0 14 Lys intake, g 559 1, d Lys intake, g , d 0 28 Lys intake, g 1,390 3,153 2, FI = feed intake. ascertain if sows were capable of using the extra nutrients to increase litter WW, and the feeding levels were designed to look at the interaction between diet and feed allowance. In this study, an extra 0.9 kg of FI did improve litter performance. This outcome is in agreement with an extensive review on lactation nutrition by Boyd et al. (2000, p. 1643) that states that nutrient specification is critical, but secondary to feed intake. Indeed, previous studies have shown that a 0.4 kg/d (Eissen et al., 2003) or 1 kg/d (Koketsu et al., 1997) increase in lactation FI can significantly improve litter WW. It is recognized that the impact of diet was not truly measured in this study as the actual intake of DE and lysine was not significantly different between the Normal and High diet specifications. For this reason, a detailed regression analysis was conducted to investigate the effect of DE and lysine intake on litter performance. Data from this trial were combined with that of Craig et al. (2015b) to correlate DE and lysine intake at different stages of lactation with litter WW. Across the 2 trials, sows were offered lactation diets that ranged from 13.6 to 15.8 MJ/kg DE and a lysine content of 8.8 to 13 g/kg. The DE intake from d 14 to 21, d 21 to 28, and therefore d 14 to 28 was positively correlated with litter WW. Lysine intake from 0 to 28 d was also correlated with litter WW. However, when all variables were combined, those that explained most of the variation in litter WW were the FI and lysine intake through the whole of lactation (0 to 28 d) and DE and lysine intake from d 14 to 28, although R 2 was low (9.4%). This analysis indicates that sow nutrient intake in the second half of lactation is a more significant driver of litter WW than intake in the first half of lactation. A similar result was found by Koketsu et al. (1997) in which litter WW was significantly correlated with intake during wk 2 and 3 in a 21-d lactation period. A sow in the first half of lactation has not yet reached peak intake, so any reduction in intake or food refusal will have minimal impact on her overall intake. Sows that refuse food in the second half of lactation may not have high voluntary intake and are not able to maintain the high intakes needed to sustain milk production and litter growth. Therefore, a drop in late lactation would have a greater detrimental effect on the resultant WW. These regression analyses demonstrate the importance of high FI, especially in late lactation. Lactation energy efficiency can be expressed as the ratio between the DE intake and the output in litter growth. Lactation feed efficiency can be expressed as the ratio between the feed consumed and the litter output. Bergsma et al. (2009) calculated sows as having an average efficiency of over 65%. That is greater than reported in this study (average of 49%), but Bergsma et al. (2009) included energy that was available through body tissue mobilization in the calculations. In this study, efficiency was not significantly different between

9 Sow factors that improve pig weaning weight 4507 Figure 2. Feed efficiency and DE efficiency during lactation. treatments but was significantly different between time points. Sows were more efficient at the beginning of lactation. That is when their milk production is steadily increasing and outstripping the nutrition provided in feed. Efficiency was the lowest at the end of lactation when sow intakes are the highest but milk production remains steady or is dropping off. Although lactation efficiency as a whole has been studied previously (Bergsma et al., 2009), this study appears to be the first to note the energy efficiency of sows during different weeks of lactation. Effect of Colostrum Intake on Weaning Weight Colostrum intake by piglets is crucial to survival and performance (Decaluwé et al., 2014). Therefore, CY of the sow and piglet CIn are vitally important. In this study, CIn at 24 h was significantly correlated with 24hW, BiW, and duration of farrowing when in combination. It was not surprising that CIn was correlated with BiW or 24hW as larger piglets are more vigorous and can extract more colostrum from the teat (Milligan et al., 2002; Quiniou et al., 2008; Devillers et al., 2007). Also, 24hW is likely to be a direct effect of CIn itself. However, it has not been noted before that the duration of farrowing had any effect on piglet CIn, although it is often recorded as a variable. It could be that sows with longer farrowings are more likely to be lying with a prominent udder, which is more accessible to piglets, than those who farrow quickly and stand up to eat or drink. However, none of the piglet vitality scores (time to breathe, time to first suckle, or the status of the cord or caul) were found to be significantly correlated with CIn, unlike in the study of Devillers et al. (2007). Colostrum yield was not affected by dietary treatment, as diets did not differ in the nutrient intake of sows. Colostrum yield of the sow was best explained by the weight of the litter at birth. This result is in agreement with a study by Vadmand et al. (2015), who first reported a positive correlation between litter weight and CY, which had not been noted previously (Devillers et al., 2007). Vadmand et al. (2015) speculated that the relationship between CY and litter size could be similar to the relationship between milk yield and litter size or possibly an effect of hormones involved in fetal development or perhaps a combination of these factors. The CY in this study ranged from 3.1 to 6.7 kg, similar to the range found by Theil et al. (2014b). Piglet WW was significantly affected by 3-wk weight, sow parity, sex, and number of pigs born alive. Parity and number of live-born pigs had a negative correlation, indicating that younger sows with smaller litters (excluding parity 1 sows, which were not represented in this data set) have piglets with improved WW. Indeed, this result confirms the correlation between large litters and more low-weight pigs at weaning (Quiniou et al., 2008; Dunshea et al., 2003). It has also been noted before that boars are faster growing in the growing and finishing periods (Kennedy, 1984), but no other papers in the open literature have found a negative correlation for gilts in the preweaning period. Piglet mortality before weaning is a great loss in terms of resources. The majority of fatalities occur in the first 72 h postfarrowing, and preweaning mortality is generally about 12% mortality when stillbirths are excluded (Edwards and Baxter, 2015). In an attempt to determine what most affects a piglet s ability to survive until weaning, 45 variates were analyzed. Surprisingly, it was only the weight gain between 24 and 48 h that most significantly affected survival, although this is not a commonly used variable. Colostrum intake and BiW were included in this model but were not significant. Piglet BiW is widely noted to be a key determinate of survival (Herpin et al., 2002; Baxter et al., 2008; Edwards and Baxter, 2015). However, Baxter et al. (2008) concluded that even low-biw piglets, if vigorous, could survive equally as well as larger littermates. Therefore, perhaps it is not piglet weight per se, but instead, the ability to gain weight, particularly in the second day of life. Teat possession is established in 50% of piglets around 24 h, with the remainder of piglets fighting up until day 4 (De Passillé et al., 1988; Puppe and Tuchscherer, 1999). It is reasonable to suggest that those piglets who gain ownership of a teat within the first day expend less energy fighting and more time consuming colostrum in the early days, which then gives them an advantage to survive until weaning. More in-depth research is required to confirm these correlations. Although this type of work is timeconsuming, it will be valuable to understand what key measures are determinants of piglet survival and growth in large litters and to exploit this knowledge in management practices.

10 4508 Craig et al. Conclusion Increasing sow lactation ADFI from 6 to 7 kg/d improved litter WW by 9% in litters of 13 piglets, achieving an average litter WW of 114 kg. A 0.6 MJ/kg increase in DE and 0.2 g/kg increase in total lysine improved litter ADG from birth to weaning but had no significant effect on litter WW. Feed intake and lysine intake throughout lactation (0 to 28 d) and DE and lysine intake from d 14 to 28 are the main drivers of litter WW. Lactation efficiency was found to decrease as lactation progressed. Colostrum intake of piglets was driven by their BiW, 24hW, and also the duration of farrowing. Colostrum yield of sows was found to be correlated with litter birth weight. Piglet WW was positively correlated with 3-wk weight but negatively correlated with sow parity, number born alive, and being female. Pig survival was closely correlated with only weight gain between 24 and 48 h. Overall, the drivers of high weaning weight within large litters include high FI, especially consistent high nutrition intake of DE and lysine in late lactation. Piglet weight gain during the second day of life appears to be important for survival. LITERATURE CITED Algers, B., and P. Jensen Teat stimulation and milk production in early lactation of sows. Can. J. Anim. Sci. 71: doi: /cjas Baxter, E. M., S. Jarvis, R. B. D Eath, D. W. Ross, S. K. Robson, M. Farish, I. M. Nevison, A. B. Lawrence, and S. A. Edwards Investigating the behavioural and physiological indicators of neonatal survival in pigs. Theriogenology 69: doi: /j.theriogenology Bergsma, R., E. Kanis, M. W. A. Verstegen, C. M. C. van der Peet-Schwering, and E. F. Knol Lactation efficiency as a result of body composition dynamics and feed intake in sows. Livest. Sci. 125(2 3): doi: /j.livsci Boyd, R. D., K. J. Touchette, G. C. Castro, M. E. Johnston, and K. U. Lee Recent advances in the nutrition of the prolific sow. Asian-Australas. J. Anim. Sci. 13(Special Issue): doi: /ajas Cottney, P. D Improving the productivity of the breeding sow herd in Northern Ireland. PhD Thesis, Queens University Belfast, Belfast, UK. Craig, A., W. Henry, and E. Magowan. 2015a. Effect of phased feeding and valine-to-lysine ratio during lactation on sow and piglet performance. J. Anim. Sci. 94(9): doi: /jas Craig, A., Magowan, E. and Gordon, A. 2015b. Review of scientific knowledge on lactation nutrition of highly prolific modern sows. In: Proceedings of the British Society of Animal Science Annual Conference, April 2015, Chester, UK. Decaluwé, R., D. Maes, B. Wuyts, A. Cools, S. Piepers, and G. P. J. Janssens Piglets colostrum intake associates with daily weight gain and survival until weaning. Livest. Sci. 162: doi: /j.livsci Declerck, I., J. Dewulf, S. Piepers, R. Decaluwé, and D. Maes Sow and litter factors influencing colostrum yield and nutritional composition. J. Anim. Sci. 93(3): doi: /jas De Passillé, A. M. B., J. Rushen, and T. G. Harstock Ontogeny of teat fidelity in pigs and its relation to competition at suckling. Can. J. Anim. Sci. 68: doi: / cjas Devillers, N., C. Farmer, J. Le Dividich, and A. Prunier Variability of colostrum yield and colostrum intake in pigs. Animal 1(7): doi: /s x Dunshea, F. R., D. K. Kerton, P. D. Cranwell, R. G. Campbell, B. P. Mullan, R. H. King, G. N. Power, and J. R. Pluske Lifetime and post-weaning determinants of performance indices of pigs. Aust. J. Agric. Res. 54(4): doi: / AR02172 Edwards, S. A., and E. M. Baxter Piglet mortality: Causes and prevention. In: C. Farmer, editor, The gestating and lactating sow. Wageningen Acad. Publ., Wageningen, Netherlands. doi: / _11 Eissen, J. J., E. J. Apeldoorn, E. Kanis, M. W. A. Verstegen, and K. H. de Greef The importance of a high feed intake during lactation of primiparous sows nursing large litters. J. Anim. Sci. 81(3): doi: / x Herpin, P., M. Damon, and J. Le Dividich Development of thermoregulation and neonatal survival in pigs. Livest. Prod. Sci. 78(1): doi: /s (02) Kennedy, B. W Between and within litter variation, sex effects and trends in sire and dam transmitting abilities of performance tested pigs in Ontario. J. Anim. Sci. 59(2): doi: /jas x King, R. H., B. P. Mullan, F. R. Dunshea, and H. Dove The influence of piglet body weight on milk production of sows. Livest. Prod. Sci. 47: doi: /s (96) Koketsu, Y., G. D. Dial, J. E. Pettigrew, and V. L. King Influence of feed intake during individual weeks of lactation on reproductive performance of sows on commercial farms. Livest. Prod. Sci. 49: doi: /s (97)00050-x Milligan, B. N., D. Fraser, and D. L. Kramer Within-litter birth weight variation in the domestic pig and its relation to pre-weaning survival, weight gain, and variation in weaning weights. Livest. Prod. Sci. 76(1 2): doi: / S (02)00012-X Park, M. S., Y. X. Yang, J. Y. Choi, S. Y. Yoon, S. S. Ahn, S. H. Lee, B. K. Yang, J. K. Lee, and B. J. Chae Effects of dietary fat inclusion at two energy levels on reproductive performance, milk compositions and blood profiles in lactating sows. Acta Agric. Scand. A Anim. Sci. 58: Puppe, B., and A. Tuchscherer Developmental and territorial aspects of suckling behaviour in the domestic pig (Sus scrofa f. domestica). J. Zool. 249: doi: /j tb00767.x Quiniou, N., S. Richard, I. Mourot, and M. Etienne Effect of dietary fat or starch supply during gestation and/or lactation on the performance of sows, piglets survival and on the performance of progeny after weaning. Animal 2 (11) Theil, P. K., C. Flummer, W. L. Hurley, N. B. Kristensen, R. L. Labouriau, and M. T. Sorensen. 2014b. Mechanistic model to predict colostrum intake based on deuterium oxide dilution technique data and impact of gestation and prefarrowing diets on piglet intake and sow yield of colostrum. J. Anim. Sci. 92(12): doi: /jas

11 Sow factors that improve pig weaning weight 4509 Toner, M. S., R. H. King, F. R. Dunshea, H. Dove, and C. S. Atwood The effect of exogenous somatotrophan on lactation performance of first-litter sows. J. Anim. Sci. 74(1): doi: / x Vadmand, C. N., U. Krogh, C. F. Hansen, and P. K. Theil Impact of sow and litter characteristics on colostrum yield, time for onset of lactation, and milk yield of sows. J. Anim. Sci. 93(5): doi: /jas van den Brand, H., M. J. W. Heetkamp, N. M. Soede, J. W. Schrama, and B. Kemp Energy balance of lactating primiparous sows as affected by feeding level and dietary energy source. J. Anim. Sci. 78(6): doi: / x Van der Peet-Schwering, C. M. C., J. W. G. M. Swinkels, and L. A. den Hartog Nutritional strategy and reproduction. In: M. W. A. Verstegen, P. J. Moughan, and J. W. Schrama, editors, The lactating sow. Wageningen Acad. Publ., Wageningen, Netherlands. Walsh, M. C., P. A. Geraert, R. Maillard, J. Kluess, and P. G. Lawlor The effect of a non-starch polysaccharidehydrolysing enzyme (Rovabio (R) Excel) on feed intake and body condition of sows during lactation and on progeny growth performance. Animal 6(10): doi: / S Whittemore, C. T., M. J. Hazzledine, and W. H. Close Nutrient requirement standards for pigs. Br. Soc. Anim. Sci., Penicuik, UK. Yang, Y. X., S. Heo, Z. Jin, J. H. Yun, J. Y. Choi, S. Y. Yoon, M. S. Park, B. K. Yang, and B. J. Chae Effects of lysine intake during late gestation and lactation on blood metabolites, hormones, milk composition and reproductive performance in primiparous and multiparous sows. Anim. Reprod. Sci. 112(3 4): doi: /j.anireprosci