Animal 15 (2021) 100218 Contents lists available at ScienceDirect Animal The international journal of animal biosciences Lysine requirements of finishing boars and gilts: A meta-analysis P. Aymerich a,b,⇑, M.D. Tokach c, S.S. Dritz d, J. Gasa b, J. Coma a, D. Solà-Oriol b a Vall Companys Group, Lleida 25191, Spain b Animal Nutrition and Welfare Service, Department of Animal and Food Sciences, Universitat Autònoma de Barcelona, Bellaterra 08193, Spain c Department of Animal Sciences and Industry, Kansas State University, Manhattan 66506, KS, USA d Department of Diagnostic Medicine/Pathobiology, College of Veterinary Medicine, Kansas State University, Manhattan 66506, KS, USA a r t i c l e i n f o a b s t r a c t Article history: The expected increase in boar (pig entire male) production while societal concerns for castration Received 26 August 2020 increase requires good estimations of their nutrient requirements. In this work, a meta-analytical Revised 23 February 2021 approach was used to overcome the inconsistent results between studies that compared lysine require- Accepted 25 February 2021 ments of boars and gilts. For this meta-analysis, data from 14 different studies analysing the effect of Available online 23 May 2021 increasing dietary lysine on growth performance of finishing pigs, 70–100 kg average body weight, were extracted from 11 publications. Those studies represented 128 different treatments (53 for boars Keywords: and 75 for gilts). Diets were reformulated based on NRC (2012) ingredient values to calculate standard- Female Male ized ileal digestible lysine to net energy ratio (SID Lys:NE) and daily SID Lys intake using average daily Pig feed intake (ADFI). As expected, no evidence for differences in ADFI (P = 0.303) was observed between Sex boars and gilts. However, boars grew faster (P < 0.001) and had higher gain to feed (G:F; P < 0.001). The Uncastrated divergent effect of SID Lys:NE on average daily gain (ADG) and G:F was analysed in a quadratic poly- nomial model showing different parameters for each sex (P < 0.001). Although performance between sexes was similar at low SID Lys:NE, differences were greater at higher SID Lys:NE. Furthermore, broken-line linear, broken-line quadratic (BLQ) and quadratic polynomial (QP) models were fitted to each sex to determine SID Lys:NE and SID Lys daily intake requirements to maximize ADG and G:F. Overall, QP models showed the best fit, and reported that to reach maximum ADG 0.88 (95% CI: [0.82–0.94]) or 1.01 (95% CI:[0.91–1.11]) g SID Lys/MJ, NE was required for gilts and boars, respectively. However, boar ADG was best fitted by BLQ using SID Lys daily intake as independent variable, with the requirement for maximum ADG at 24.2 (95% CI:[21.3–27.2]) g SID Lys/day. The three models reported wide confidence intervals for the requirements at maximum performance, and consequently those were overlapped when comparing boars and gilts. Maximum boars’ productive performance when diet- ary lysine was not limiting was 116% of gilts, and at those levels the amount of SID Lys intake required per kg gain was similar between both sexes. Thus, because ADFI and Lys efficiency of gain was similar, the requirement differences were driven by the increased growth rate and gain to feed ratio between boars and gilts. In conclusion, the present study confirmed a greater productive response of boars com- pared to gilts when increasing dietary lysine. Ó 2021 The Authors. Published by Elsevier B.V. on behalf of The Animal Consortium. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Implications and a similar lysine efficiency, they require more lysine than gilts. Therefore, in some price contexts, feeding boars and gilts, different Feed cost is the largest cost in swine production, and amino feeds tailored to their requirements could improve feed efficiency acids are considered the second major cost after energy. This work and reduce feed costs, while reducing the environmental impact of provides evidence that boars and gilts have a different response to swine production, as less nitrogen would be excreted. increasing dietary lysine, as a reference of dietary amino acid con- tent. Because boars have a greater growth performance potential Introduction ⇑ Corresponding author at: Vall Companys Group, Lleida 25191, Spain. Increasing societal pressure in some countries to stop surgical E-mail address: paymerich@vallcompanys.es (P. Aymerich). castration of male pigs will increase the relative production of https://doi.org/10.1016/j.animal.2021.100218 1751-7311/Ó 2021 The Authors. Published by Elsevier B.V. on behalf of The Animal Consortium. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). P. Aymerich, M.D. Tokach, S.S. Dritz et al. Animal 15 (2021) 100218 entire male pigs in future years (Bee et al., 2015). Male pigs that are to articles from 2000 onwards to limit differences related to not castrated have improved feed efficiency and leaner carcasses genetic improvements and nutrition advances. In addition, only than barrows (Dunshea et al., 2013), and better feed efficiency than studies in which initial BW was between 50 and 85 kg and final gilts (Cámara et al., 2014) because of a greater lean deposition BW was between 85 and 120 kg were considered. Publications potential (King et al., 2000; Giles et al., 2009). Several studies have were excluded when there were less than four dietary lysine levels shown that growth differentiation between boars and gilts starts or when dietary SID Lys was not constant throughout the study. around 40–70 kg body weight (BW) (Campbell et al., 1988; Other reasons for exclusion were not meeting amino acid ratios Moore et al., 2013; Cámara et al., 2014). Because boars have a lea- (ideal protein) after reformulation, based on NRC (NRC, 2012), or ner (Carabús et al., 2017) and higher growth, and lysine is the first only providing mixed sex data. limiting amino acid for lean tissue deposition, boars need more lysine intake than gilts to maximize growth performance Data sources (Aymerich et al., 2020). Some modelling studies have assumed a lower average daily feed intake (ADFI) for boars than for gilts The literature search was performed on April 18, 2020, by (NRC, 2012; Dunshea et al., 2013; van der Peet-Schwering and searching with PubMed and Web of Science using prespecified Bikker, 2018), other work has not found evidence of these ADFI dif- search terms like lysine and pig and boar or gilt (Supplementary ferences (Moore et al., 2013; Rikard-Bell et al., 2013; Aymerich Material S1). References in the identified eligible articles were et al., 2020). Furthermore, those works reported greater differences checked and some authors were contacted for further information in average daily gain (ADG) than the assumed inputs in the models if data were only available as figures. Finally, a publication when sufficient lysine was available. For instance, NRC (2012) detected after revising cross-references in the selected articles assumed that entire male ADG was only 2.8% greater than gilts was also included, although not published in a peer-reviewed jour- between 75 and 100 kg BW whereas Aymerich et al. (2020) nal (Moore et al., 2015). reported on average a 9.2% greater ADG in the period 70–105 kg and O’Connell et al. (2006) a 17.6% greater ADG. Available studies comparing lysine requirements of boars and Study selection gilts to maximize growth performance have indicated inconsistent results. Whereas Moore et al. (2013) reported that boars and gilts A total of 1 473 publications were initially identified after maximized ADG at the same lysine level, Rikard-Bell et al. (2013) removing the duplicates between both data sources. To determine and Aymerich et al. (2020) found different requirements for each whether studies fulfilled the eligibility criteria, first, the title was sex. O’Connell et al. (2006) reported different responses depending checked to discard studies focusing on other topics. Afterwards, whether pigs were housed in pairs or in groups. In addition, most the abstracts of the remaining studies were carefully analysed, of these studies have modelled the requirements for each sex and and if sufficient data were not provided there, the full text was also posteriorly compared the estimates without considering the asso- checked. At the end, there were only eleven publications that ful- ciated standard error. Recent works have shown the wide confi- filled all inclusion criteria, including nine studies for boars and dence intervals of those estimates, which make the comparison thirteen for gilts, of which eight included results for both sexes. between sexes difficult (Aymerich et al., 2020). It could be that dif- ferences in BW range or genetic lines are partly responsible for Data abstraction these disagreements. In addition, the different models used to describe the response to the dietary standardized ileal digestible The data from the selected studies represented 6 654 pigs (1 lysine to net energy ratio (SID Lys:NE) in each study might also 588 boars and 5 066 gilts) in 128 different treatments (53 for boars be responsible for the inconsistencies (Pesti et al., 2009). In a situ- and 75 for gilts), each one used as an experimental unit. Diet infor- ation of lack of clarity in individual studies, meta-analysis is con- mation in each study was entered into a single database to refor- sidered a reasonable and powerful tool to improve the mulate the diets based on NRC (2012) ingredient values. This understanding of the response (Kelley and Kelley, 2019) and deter- approach was preferred instead of using the dietary values pro- mine nutritional requirements (van Milgen et al., 2012). Therefore, vided by the studies because of the large impact of different energy the aims of this meta-analysis were to compare the response to evaluation systems and digestibility values. The authors acknowl- increasing dietary lysine between finishing boars and gilts (65– edge that this approach might miss some information only known 100 kg BW), and, if different, determine dietary SID Lys:NE and by the other researchers but are confident that it is a more objec- SID Lys intake requirements for each sex to maximize growth tive approach. For all studies except King et al. (2000), differences performance. in diet SID Lys content between published and reformulated com- position were 0.05% SID Lys. Next, a database was created includ- ing (1) study reference (first author name, year of publication, Material and methods locations and experiment number), (2) animal characteristics (sex, genetics, initial BW and final BW), (3) experimental design This study did not require ethical approval as the data were col- characteristics (pigs per treatment, replicates per treatment), (4) lected from studies already published in the literature. The diet characteristics (main energy and protein sources, crude pro- methodology outlined by Kelley and Kelley (2019) for meta- tein, metabolizable energy, net energy, SID Lys and SID Lys:NE, analysis in nutrition research was used as a reference to develop after reformulation), (5) growth performance (ADG, ADFI and gain a protocol for obtaining meta-data (Supplementary Material S1). to feed), and (6) measures of variability for the response variables Thus, the following section was divided into study eligibility, data (standard error of the mean, standard error of the difference, stan- sources, study selection, data abstraction, and statistical analyses. dard deviation or the coefficient of variation). The latter were recorded to provide a measure of the consistency of the means to Study eligibility be used as weights in the regression model (St-Pierre, 2001). All variability measures were transformed to standard error of the Only randomized controlled dose–response experiments ana- mean (SEM). Each row in the dataset corresponded to a treatment lysing the effects of dietary lysine on growth performance of finish- of one specific sex. The ADFI and SID Lys content of the diet was ing boars or gilts were included. The literature search was limited used to calculate daily SID Lys intake. 2 P. Aymerich, M.D. Tokach, S.S. Dritz et al. Animal 15 (2021) 100218 Statistical analyses a model with a general positive variance–covariance structure was fitted. Subsequently, the complex and simpler models were com- A reduced dataset with the eight experiments including both pared using BIC, and the one with the lowest BIC was selected. boars and gilts was analysed to compare performance of both Over parametrization was assessed analysing the correlation sexes. The highly suspected heterogeneity was corroborated com- between random parameters in the model, and if present, the paring null models and heterogeneity models using the Bayesian model was tested with and without each parameter, selecting information criteria (BIC). Initially, growth performance of both the model with the lowest BIC. Only the best fitting model, the sexes was compared in a linear mixed model including sex as a ones reporting the lowest BIC, was finally considered (Supplemen- fixed effect and study as a random effect using the nlme package tary Table S2). 95% confidence intervals (CIs) of the fixed effects (Pinheiro et al., 2019) of R (R Core Team, 2019). In addition, the were calculated for the parameter estimates of the models. The SEM of the specific variable tested was specified as a variance CIs of the parameters representing the level at which maximum covariate using the VarPower function in the weight statement performance was achieved in BLL and BLQ models were used to (Pinheiro and Bates, 2000), representing the inverse of the variance compare requirements of boars and gilts. Furthermore, the CI at as suggested by St-Pierre (2001). Finally, model validity was exam- the inflection point in the QP models was estimated using the delta ined using standardized residual scatterplots to observe whether method in the msm package (Jackson, 2011). the distribution of residuals was more homogeneous after account- ing for the random study effect and the variance covariate adjust- Results ment. The sex effect significance was determined by the F-test in the ANOVA when P-value was 0.05. Least square means and Study characteristics SEM were estimated with the emmeans package (Lenth, 2020). Fur- thermore, the differential response to SID Lys:NE between boars The 14 studies included in the meta-analysis are summarized in and gilts was analysed in a quadratic polynomial function Table 1. Initial BW ranged from 49.6 to 84.1 kg, whereas final BW (Y i ¼ LQP þ BQP  X i þ AQP  X 2i ) in which the parameters LQP , BQP from 86.2 to 120 kg. The minimum SID Lys:NE was 0.38 for gilts and AQP were interacted with sex. Posteriorly ANOVA with F-tests and 0.45 for boars whereas the maximum was 1.23 g/MJ for both was used to assess the significance of the interactions, and only sexes. Net energy ranged from 9.7 to 11.7 MJ/kg, but the diets with the ones that reduced BIC were considered for the final model. higher energy density (>10.6 MJ NE/kg) were only from gilts stud- Next, each sex was modelled independently to determine the ies (Main et al., 2008; Shelton et al., 2011). The differences in NE response to dietary lysine on growth performance using the com- within some studies were related to formulating diets based on plete dataset. Regression models to predict growth performance, digestible energy or metabolizable energy, instead of NE, and to ADG or gain to feed (G:F), depending on SID Lys:NE and SID Lys using different ingredient composition tables than NRC (2012). daily intake were built using the nlme package (Pinheiro et al., Median year of publication was 2008, and of the 14 studies, five 2019). Broken-line linear (BLL), broken-line quadratic (BLQ), and were from Australia, four from Europe, four from North America quadratic polynomial (QP) models were built following Robbins and one from South America. Finally, the median number of repli- et al. (2006). Furthermore, the quadratic parameter of the BLQ cates per study was six and seven, for boars and gilts, respectively. model (ABLQ ) was included as a function of BBLQ and the require- ment at maximum performance (RBLQ Þ: Sex differences ABLQ ¼ BBLQ =ðRBLQ  2Þ The effect of sex on growth performance and carcass composi- Initially, a random component was included for all fixed effect tion variables is reported in Table 2. The effects of sex on BW were parameters using a diagonal variance–covariance structure to not statistically analysed because for some studies, BW data were determine which did not account for the between subject model only reported as mixed sex, while others only reported the study variability. The parameters which had a near zero standard devia- average initial and final BW. Nevertheless, on average initial BW tion (SD) were removed starting by the one with the lower SD and was 68.9 and 69.1 kg, average BW was 85.6 and 85.4 kg, and final Table 1 Summary of the studies included in the meta-analysis to predict the effect of sex on the growth performance response to dietary lysine of finishing boars and gilts. Average BW (kg) Dietary Dietary range Ingredients used to treatments increase the dietary Study Publication – Experiment (Exp.) Sex Initial Final SID Lys:NE1 NE2 SID Lys 1 Aymerich et al. (2020)3 Boars/Gilts 69.6 105.5 5 0.63–1.00 10.6–10.6 SBM + Lys + AA 2 Cline et al. (2000) Gilts 53.6 116.4 5 0.62–1.16 10.1–10.6 SBM 3 Kill et al. (2003)3 Gilts 66.3 95.5 4 0.65–0.92 10.5–10.6 SBM 4 King et al. (2000)3 Boars/Gilts 80.0 120.0 6 0.45–0.92 10.1–10.5 SBM + Lys + AA 5 Main et al. (2008) – Exp. 2 Gilts 59.8 86.2 6 0.50–0.92 11.1–11.6 SBM + AA 6 Main et al. (2008) – Exp. 3 Gilts 78.4 102.9 6 0.38–0.74 11.4–11.7 SBM 7 Moore et al. (2013) – Exp. 23 Boars/Gilts 49.6 103.3 5 0.55–1.12 10.0–10.4 SBM + Lys + AA 8 Moore et al. (2015)3 Boars/Gilts 63.6 100.5 7 0.60–1.12 9.7–10.2 SBM + AA 9 Moore et al. (2016)3 Boars 60.1 105.1 5 0.46–1.06 9.9–10.3 SBM + CM + MM + Lys + AA 10 O’Connell et al. (2006) – Exp. 1 Boars/Gilts 60.0 91.0 6 0.68–1.12 9.8–10.3 SBPC + AA 11 O’Connell et al. (2006) – Exp. 2 Boars/Gilts 81.0 102.0 8 0.59–1.23 9.7–10.4 SBPC + AA 12 O’Connell et al. (2006) – Exp. 3 Boars/Gilts 80.0 99.0 6 0.59–1.09 9.9–10.4 SBPC + AA 13 Rikard-Bell et al. (2013)3 Boars/Gilts 60.0 90.0 5 0.55–1.11 9.8–10.3 SBM + MM + AA 14 Shelton et al. (2011) – Exp. 3 Gilts 84.1 110.5 6 0.46–0.78 11.0–11.3 SBM + AA Abbreviations: SID Lys = standardized ileal digestible lysine; SBM = soybean meal; Lys = synthetic lysine; AA = synthetic amino acids different to Lys; CM = canola meal; SBPC = soybean protein concentrate; MM = meat meal. 1 Standardized ileal digestible lysine to net energy ratio (SID Lys:NE, g/MJ) calculated after reformulating the diets (NRC, 2012). 2 Net energy (NE, MJ/kg) calculated after reformulating the diets (NRC, 2012). 3 Reported backfat thickness measurements. 3 P. Aymerich, M.D. Tokach, S.S. Dritz et al. Animal 15 (2021) 100218 Table 2 slope (BQP , P < 0.001), intercept (LQP , P < 0.001) and a different The effects of sex (boars vs. gilts) on growth performance and carcass composition of quadratic parameter (AQP , P = 0.074) for each sex. These differences finishing pigs from the eight studies included in the meta-analysis that studied both boars and gilts. suggested that boars require more SID Lys:NE to reach maximum performance, and that gilt performance is significantly reduced Item Boars Gilts P-value when fed at high SID Lys:NE. Summarizing, both ADG and G:F n (observations) 48 48 – models confirmed a different response to increasing SID Lys:NE Initial BW, kg1 68.9 69.1 – between boars and gilts. Average BW, kg1 85.6 85.4 – Final BW, kg1 102.4 101.6 – Average daily gain, g 1 000 ± 33.0 889 ± 33.1 <0.001 Lysine requirements by sex Average daily feed intake, g 2 448 ± 55.2 2 433 ± 55.2 0.303 Gain to feed, g/kg 408 ± 12.0 365 ± 12.0 <0.001 Backfat thickness, mm2 11.0 ± 0.98 11.7 ± 0.98 <0.001 The different response to SID Lys:NE between boars and gilts suggested fitting models to determine the requirements to maxi- Least square means ± SEM. 1 mize performance for each sex separately. The best fitting BLL, For BW, the value represents the arithmetic mean of all the observations. 2 Replicates were 33 and 32, for boars and gilts, respectively. BLQ and QP regression equations to describe the effect of SID Lys:NE and SID Lys intake in boars and gilts ADG and G:F and each model BIC are reported in Table 3. In addition, the final models in response to SID Lys:NE and the observations from each study are illustrated in Fig. 3. Similarly, Fig. 4 illustrates the models explain- ing the effect of SID Lys intake on ADG and G:F and all the obser- vations included in the meta-analysis. Quadratic polynomial models were the best fitting ones except for boars ADG that was best predicted by BLQ using SID Lys intake as explanatory variable. Nevertheless, the differences in BIC between boars BLQ and QP for both ADG and G:F were at most two units with SID Lys:NE as explanatory variable. As expected, the three models reported different SID Lys:NE and SID Lys intake to maximize growth performance, with the lowest being always the BLL. Although the 95% confidence intervals of the requirement (RBLL ) were overlapped between sexes, the slope of the effect of increasing SID Lys:NE on ADG was greater in boars than in gilts. Besides, maximum performance (LBLL ), for both ADG and G:F was greater in boars than in gilts regardless of the explana- tory variable used. On average, boars’ maximum performance using BLL was 16% greater than gilts. Broken-line quadratic models did neither report different SID Lys:NE nor SID Lys intake require- ments (RBLQ ) because of the wide confidence intervals. However, boars’ requirement was always numerically greater than gilts. For instance, boars required 0.926 g SID Lys/MJ NE (95% CI: [0.826, 1.026]) or 24.2 g SID Lys/d (95% CI: [21.3, 27.2]) to maxi- mize ADG whereas gilts required 0.872 g SID Lys/MJ NE (95% CI: [0.728, 1.016]) or 22.6 g SID Lys/d (95% CI: [19.1, 26.2]). The QP models reported the highest requirement to reach the maximum performance, calculated at the inflection point of the function. Regarding boars, maximum ADG and G:F were reached Fig. 1. Variation in average daily feed intake (ADFI) of finishing boars and gilts at 1.01 (95% CI:[0.91–1.11]) and 1.06 (95% CI:[0.97–1.15]) g SID within each study and sex. Error bars represent the minimum and maximum Lys/MJ NE, whereas gilts at 0.88 (95% CI:[0.82–0.94]) and 0.91 reported values while the vertical dashed and solid lines represent the mean of each (95% CI:[0.84–0.98]) g SID Lys/MJ NE, respectively. Table 4 shows sex. the dietary SID Lys required to reach different relative target per- formances (95–100%) in 10.47 MJ NE/kg diets. To increase one point of relative performance between 95 and 99% of the maximum BW was 102.4 and 101.6 kg, for boars and gilts, respectively. As performance, dietary SID Lys had to increase around 0.03–0.05%, expected, boars compared to gilts had greater ADG (P < 0.001) with whereas to increase from 99% to 100%, it required 0.12–0.13% no evidence for differences in ADFI (P = 0.303). Therefore, boars higher dietary SID Lys. Although QP and BLQ SID Lys:NE models fit- had an improved G:F (P < 0.001). In addition, Fig. 1 shows the ted similarly for boars, QP reported a greater requirement for max- variation in ADFI between treatments within each study by sex imum ADG that could be related to maximum ADG being slightly and visually confirms the similar ADFI between boars and gilts. greater in the QP than in the BLQ (1 063 vs. 1 054 g). Finally, as Regarding carcass composition, only backfat thickness was for the BLL, QP reported boars maximum ADG and G:F to be abstracted from the studies because it was the most frequently 116% of gilts. reported parameter. As expected, backfat thickness was greater in gilts than in boars (11.7 vs. 11.0 mm, P < 0.001). To assess whether the response to increasing SID Lys:NE dif- Discussion fered between boars and gilts, a quadratic model including the interaction between sex and each of the three parameters was built The increasing importance of entire male production requires and each interaction was evaluated. The best fitting models to pre- good estimations of their nutrient requirements. As lysine is the dict ADG and G:F accounting for the interactions are shown in first limiting AA for protein deposition (NRC, 2012) and boars have Fig. 2. Best fitting models for both ADG and G:F included a different a greater potential for protein deposition than gilts (King et al., 4 P. Aymerich, M.D. Tokach, S.S. Dritz et al. Animal 15 (2021) 100218 Fig. 2. Best fitting quadratic polynomial models to predict average daily gain (a) and gain to feed (b) from dietary standardized ileal digestible lysine to net energy ratio (SID Lys:NE) for boars (─) and gilts (- -). The regression equations for growth were Y ¼ 623  X 2 þ 1295  X þ 371 and Y ¼ 944  X 2 þ 1669  X þ 184, for boars and gilts, respectively. The regression equations for feed efficiency were Y ¼ 238  X 2 þ 522  X þ 145 and Y ¼ 362  X 2 þ 665  X þ 74, for boars and gilts, respectively. Each plot is based on 96 observations from eight different studies. Table 3 Parameter estimates of different models to predict average daily gain (ADG, g) and gain to feed (G:F, g/kg) from the dietary standardized ileal digestible lysine (SID Lys) to net energy (NE) ratio and SID Lys daily intake for boars (B) and gilts (G). Explanatory variable (Y i ) Response variable (X i ) Sex Parameter estimates1 BIC2 Broken-line linear3 LBLL3 UBLL3 RBLL3 SID Lys:NE ADG B 1 045 [979, 1 112] 851 [618, 1 084] 0.736 [0.702, 0.770] 581 (g/MJ) G 913 [871, 956] 380 [231, 530] 0.715 [0.654, 0.776] 795 G:F B 429 [408, 450] 196 [149, 244] 0.882 [0.822, 0.941] 478 G 371 [352, 390] 136 [78, 195] 0.736 [0.665, 0.806] 667 SID Lys/day ADG B 1 060 [988, 1 135] 20.3 [12.7, 27.9] 21.9 [20.4, 23.4] 587 (g/day) G 915 [873, 957] 14.4 [8.9, 19.8] 19.2 [17.7, 20.7] 793 G:F B 431 [408, 454] 7.31 [5.53, 9.10] 22.9 [21.1, 24.7] 484 G 371 [352, 390] 4.95 [2.93, 6.96] 19.5 [17.8, 21.2] 667 Broken-line quadratic4 LBLQ4 BBLQ4 RBLQ4 SID Lys:NE ADG B 109 [239, 458] 2 040 [1 084, 2 997] 0.926 [0.826, 1.026] 574 (g/MJ) G 490 [296, 685] 977 [395, 1 558] 0.872 [0.728, 1.016] 791 G:F B 117 [23, 210] 607 [371, 842] 1.033 [0.910, 1.156] 468 G 174 [83, 265] 476 [205, 747] 0.824 [0.704, 0.944] 663 SID Lys/day ADG B 79.2 [191, 349] 81.5 [52.3, 110.6] 24.2 [21.3, 27.2] 573 (g/day) G 424 [181, 668] 43.5 [16.2, 70.9] 22.6 [19.1, 26.2] 789 G:F B 127 [24, 230] 22.9 [12.7, 33.1] 26.6 [22.7, 30.5] 478 G 174 [65, 283] 18.0 [5.7, 30.3] 21.8 [18.2, 25.4] 665 Quadratic polynomial5 LQP5 BQP5 AQP5 SID Lys:NE ADG B 260 [71, 450] 1 589 [1 138, 2 040] 787 [1 063, 511] 567 (g/MJ) G 407 [278, 536] 1 172 [860, 1 485] 664 [854, 473] 765 G:F B 131 [57, 206] 567 [391, 743] 267 [372, 163] 458 G 165 [106, 224] 463 [326, 599] 254 [337, 171] 648 SID Lys/day ADG B 235 [42, 429] 64.6 [46.2, 83.1] 1.27 [1.73, 0.81] 581 (g/day) G 393 [250, 536] 46.5 [32.4, 60.5] 1.02 [1.38, 0.66] 789 G:F B 147 [59, 235] 20.7 [12.4, 29.1] 0.375 [0.573, 0.177] 487 G 207 [157, 257] 14.3 [9.8, 18.7] 0.308 [0.412, 0.205] 673 1 Estimate [95% confidence interval]. 2 Bayesian information criteria. 3 Y ¼ LBLL  U BLL  ðRBLL  X i Þ; if X i  RBLL Broken-line linear: i . Y i ¼ LBLL ; if X i > RBLL  4 Y i ¼ LBLQ þ BBLQ  X i  BBLQ  X 2i = RBLQ  2 ; if X i  RBLQ Broken-line quadratic: . Y i ¼ LBLQ þ BBLQ  RBLQ =2; X i > RBLQ 5 Quadratic polynomial: Y i ¼ LQP þ BQP  X i þ AQP  X 2i . 2000; Giles et al., 2009), this study focused on the differential deposition as input parameters. Although these approaches might response of boars and gilts to increasing SID Lys:NE ratio. In the provide reliable estimations, it is necessary to validate the values last decade, the publication of different nutrient requirements for provided by these models with experimental data to determine boars and gilts evidenced the increasing concern on this issue. whether those approaches are valid or need further revision. For instance, NRC (2012) provided different requirements for boars Consistent with the literature, this meta-analysis confirmed and gilts, whereas the previous version did not (NRC, 1998). Simi- that boars grow more rapidly and with a better feed efficiently larly, more recent work also gave different requirements for each than gilts, and without evidence of differences in ADFI, similar to sex (van der Peet-Schwering and Bikker, 2018). These publications results reported by Cámara et al. (2014). The variation in ADFI were based on models which consider ADFI and ADG or protein between studies could be related to the differences in dietary NE 5 P. Aymerich, M.D. Tokach, S.S. Dritz et al. Animal 15 (2021) 100218 Fig. 3. Best fitting broken-line linear (BLL, ─), broken-line quadratic (BLQ, - - -) and quadratic polynomial (QP,   ) models to predict boars (a) and gilts (b) growth performance from the dietary standardized ileal digestible lysine to net energy ratio (SID Lys:NE). Each number represents a different study, as summarized in Table 1. The boar’s plots (a) are based on 53 observations from nine different studies, and the gilt’s ones (b) on 75 observations from 13 different studies. concentration or to the feed intake potential of the genotype used. Commonly, models with a quadratic shape are preferred for The reduction in BFT for boars might be the result of less energy being ‘‘biologically meaningful” (Remmenga et al., 1997) or available for fat deposition as a greater fraction is used for protein because they better represent the ‘‘diminishing marginal produc- deposition (Moore et al., 2013). Nevertheless, some publications tivity”. However, the concept of a nutritional requirement consists have not reported evidence for different BFT between boars and in assuming that a plateau is reached, and therefore, models which gilts (Gispert et al., 2010; Moore et al., 2013; Trefan et al., 2013) combine a plateau but with an ascending quadratic part (BLQ) whereas others have (Cámara et al., 2014; Aymerich et al., 2019). might be a good combination of both concepts. However, the In addition, this study confirmed a different response to dietary greater standard error for the requirement estimate of BLQ models SID Lys:NE between both sexes, with boars showing a positive as outlined by Pesti et al. (2009) might rise concerns around the reaction up to higher SID Lys:NE levels. precision of that estimate. Generally, the best fit in this study The different requirements to maximize ADG and G:F reported was provided by the QP models, probably related to accounting by BLL, BLQ and QP models might be the result of the model itself for the reduction in performance at dietary lysine above the (Pesti et al., 2009), but also for which fixed effect parameters, a ran- requirement. This reduction could be the result of a reduced energy dom component was included in the model (Robbins et al., 2006). available for fat deposition when fed high CP diets, as part of it is Moreover, as the meta-data only included studies in a specific BW used to deaminate excess amino acids (Bender, 2012) or of a neg- range, it was not considered necessary to account for the variation ative correlation between dietary SID Lys and ADFI (Aymerich in requirements related to BW as implemented by van Milgen et al. et al., 2020). (2012). Moreover, because the BLL and BLQ did hardly ever include The outputs of the QP models showed that to increase the per- a random component for the requirement parameter (RBLL and formance of both boars and gilts from 99 to 100%, the required RBLQ), it could be concluded that there was almost no variation in increase in dietary SID Lys was of similar magnitude to that the dietary lysine requirement between the studies included. The required to increase performance from 96 to 99%. Therefore, in different slope before reaching the plateau in the BLL models for some price contexts, it might not be economically feasible to feed ADG suggested that although the requirement estimate was not finishing boars and gilts at their maximum performance. Nutrition- different for those models, the marginal efficiency was greater ist can use the QP equations provided to decide the most optimal for boars. The low requirement for boars ADG could be related to diet for their production goal and whether it is feasible or not to the requirement underestimation of BLL model outlined by some feed boars and gilts separately. For instance, to reach 99% of max- authors. imum ADG, it would be necessary to feed diets with 3.75 or 3.20 g SID Lys/Mcal NE for boars and gilts, respectively. Similarly, in the 2 6 P. Aymerich, M.D. Tokach, S.S. Dritz et al. Animal 15 (2021) 100218 Fig. 4. Best fitting broken-line linear (BLL, ─), broken-line quadratic (BLQ, - - -) and quadratic polynomial (QP,   ) models to predict boars (a) and gilts (b) growth performance from the standardized ileal digestible lysine (SID Lys) daily intake. Each number represents a different study, as summarized in Table 1. The boar’s plots (a) are based on 53 observations from nine different studies, and the gilt’s ones (b) on 75 observations from 13 different studies. Table 4 Dietary standardized ileal digestible lysine (%) required to reach target average daily gain (ADG) and gain to feed (G:F) in relation to their maximum for finishing boars and gilts using quadratic polynomial models.1 % of maximum performance Item 95 96 97 98 99 100 ADG Boars 0.79 0.82 0.85 0.89 0.94 1.06 Gilts 0.65 0.68 0.71 0.75 0.80 0.92 G:F Boars 0.81 0.84 0.88 0.92 0.98 1.11 Gilts 0.67 0.70 0.73 0.77 0.82 0.95 1 Standardized ileal digestible lysine (SID Lys) calculated assuming a diet with 10.47 MJ net energy/kg. 500 kcal NE/kg diet example, boars required between 0.14 and in NRC (2012) were the result of assuming an ADG of 922 and 0.16% more dietary SID Lys than gilts to reach between 95and 897 g/d, and protein deposition at 156 and 144 g/d when mod- 100% of maximum ADG or G:F. elling boars and gilts, respectively. On the contrary, our models Models describing the effect of SID Lys intake could be more reported greater differences in ADG between both sexes that useful for practical feed formulation when feed intake is well explain the differences in SID Lys requirements. characterized, but because feed intake is difficult to predict, SID The SID Lys intake models might also be used to calculate the Lys:NE is usually preferred. Moreover, in this meta-analysis, BLQ efficiency of use of SID Lys per kg gain at the maximum perfor- was the best fitting model for boars ADG when using SID Lys intake mance. Main et al. (2008) found a constant efficiency of around as explanatory variable, but gilt fit was rather poor. The BLQ pre- 20 g total ileal digestible Lys/kg gain for grow-finishing pigs at dicted a requirement for maximum ADG at 24.2 g and 22.6 g SID the level at which growth performance was maximized. In the pre- Lys/d, for boars and gilts, respectively. Both values were relatively sent study, considering the SID Lys intake and the performance at high compared to NRC (2012) between 75 and 100 kg BW, 19.0 and the breakpoint (BLL and BLQ) or the inflection point (QP), there 18.4 g SID Lys/d, for boars and gilts, respectively. The lower values were some differences between models, but the value was similar 7 P. Aymerich, M.D. Tokach, S.S. Dritz et al. Animal 15 (2021) 100218 between sexes. Sex related variation in lysine efficiency was Author contributions between 20.7and 21.0 for BLL, 22.7 and 24.6 for BLQ, and 24.1 and 24.7 g SID Lys/kg gain for QP, for boars and gilts, respectively. P. Aymerich: Conceptualization, Data curation, Formal analysis, Thus, relevant differences in SID Lys efficiency of utilization for Investigation, Methodology, Software, Validation, Visualization, growth at maximum performance between sexes were only Writing-original draft. reported by BLQ. Unexpectedly, boars did not require more SID M.D. Tokach: Conceptualization, Data curation, Project admin- Lys per kg gain, although protein deposition represents a greater istration, Resources, Supervision, Validation, Writing – review & fraction of their growth. These results would be further supported editing. by Heger et al. (2009) that suggested that there was no evidence of S.S. Dritz: Formal analysis, Methodology, Resources, Software, differences on SID Lys utilization between pigs with different Supervision, Writing – review & editing. protein deposition potential. Future studies might aim to compare J. Gasa: Conceptualization, Supervision, Writing – review & SID Lys digestibility and maintenance requirements of finishing editing. boars and gilts. J. Coma: Conceptualization, Funding acquisition, Supervision, The relative maximum performance of boars for the different Writing – review & editing. SID Lys:NE models was around 115–116% of gilts, for both ADG D. Solà-Oriol: Conceptualization, Funding acquisition, Project and G:F. If there are no differences in SID Lys efficiency for growth administration, Supervision, Writing – review & editing. between sexes, then the requirements of boars relative to gilts would be directly related to their relative performance. Therefore, Declaration of interest assuming finishing boar SID Lys requirement to be around 115% of gilts might be useful for practical feed formulation. Dunshea et al. The authors declare no conflict of interest. (2013) suggested that SID Lys:NE requirements of boars relative to gilts might be 108% from 50 to 95 kg BW and 114% from 95 to 125 kg BW when modelled with InraPorc (van Milgen et al., Acknowledgements 2008) using previously published data (Quiniou et al., 2010) as inputs. Thus, the relative requirements were the result of the The authors would like to acknowledge K. L. Moore for provid- observed differences in performance, which were small during ing the numeric values reported in figures in Moore et al. (2016). the first 42 days in the grow-finishing facilities. In addition, the single phase diet used in that study might have been initially lim- Financial support statement iting boars’ growth. Summarizing, this study provided evidence of a different This research was supported by the Catalan Industrial Doctor- response of finishing gilts and boars to increasing dietary lysine. ates Plan of the Department for Universities and Research, Ministry In this meta-analysis, the maximum performance of boars rela- of Business and Knowledge of Generalitat de Catalunya (Ref: 2017 tively to gilts when dietary lysine was not limiting was around DI 046). 115–116% between 70 and 100 kg BW. Thus, the requirements of boars can be expected at around 115% of gilt requirements until further studies compare the efficiency of use of lysine between References sexes. However, basing boar requirements on gilts requires good estimates of gilts’ dietary lysine requirements. The equations pro- Aymerich, P., Gasa, J., Bonet, J., Coma, J., Solà-Oriol, D., 2019. The effects of sire line, vided in this work, especially the quadratic polynomial, can be sex, weight and marketing day on carcass fatness of non-castrated pigs. Livestock Science 228, 25–30. used to evaluate the effects of different dietary strategies in boars Aymerich, P., Soldevila, C., Bonet, J., Farré, M., Gasa, J., Coma, J., Solà-Oriol, D., 2020. and gilts in the body weight range studied. Interrelationships between sex and dietary lysine on growth performance and carcass composition of finishing boars and gilts. Translational. Animal Science 4, txaa129. Supplementary material Bee, G., Chevillon, P., Bonneau, M., 2015. Entire male pig production in Europe. Animal Production Science 55, 1347–1359. Bender, D.A., 2012. The metabolism of ‘surplus’ amino acids. British Journal of Supplementary data to this article can be found online at Nutrition 108, S113–S121. https://doi.org/10.1016/j.animal.2021.100218. Cámara, L., Berrocoso, J.D., Sánchez, J.L., López-Bote, C.J., Mateos, G.G., 2014. Influence of net energy content of the diets on productive performance and carcass merit of gilts, boars and immunocastrated males slaughtered at 120kg BW. Meat Science 98, 773–780. Ethics approval Campbell, R.G., Taverner, M.R., Curic, D.M., 1988. The effects of sex and live weight on the growing pig’s response to dietary protein. Animal Production 46, 123– Not applicable. 130. Carabús, A., Sainz, R.D., Oltjen, J.W., Gispert, M., Font-i-Furnols, M., 2017. Growth of total fat and lean and of primal cuts is affected by the sex type. Animal 11, Data and model availability statement 1321–1329. Cline, T.R., Cromwell, G.L., Crenshaw, T.D., Ewan, R.C., Hamilton, C.R., Lewis, A.J., Mahan, D.C., Southern, L.L., 2000. Further assessment of the dietary lysine None of the data were deposited in an official repository. requirement of finishing gilts. Journal of Animal Science 78, 987–992. Dunshea, F.R., Allison, J.R.D., Bertram, M., Boler, D.D., Brossard, L., Campbell, R., Crane, J.P., Hennessy, D.P., Huber, L., De Lange, C., Ferguson, N., Matzat, P., Author ORCIDs Mckeith, F., Moraes, P.J.U., Mullan, B.P., Noblet, J., Quiniou, N., Tokach, M., 2013. The effect of immunization against GnRF on nutrient requirements of male pigs: a review. Animal 7, 1769–1778. P. Aymerich: https://orcid.org/0000-0003-4766-9521. Giles, L.R., Eamens, G.J., Arthur, P.F., Barchia, I.M., James, K.J., Taylor, R.D., 2009. M.D. Tokach: https://orcid.org/0000-0002-9621-362X. Differential growth and development of pigs as assessed by X-ray computed S.S. Dritz: https://orcid.org/0000-0001-6371-0729. tomography. Journal of Animal Science 87, 1648–1658. Gispert, M., Oliver, M.À., Velarde, A., Suarez, P., Pérez, J., Font i Furnols, M., 2010. J. Gasa: https://orcid.org/0000-0002-5828-7142. Carcass and meat quality characteristics of immunocastrated male, surgically D. Solà-Oriol: https://orcid.org/0000-0001-8365-340X. castrated male, entire male and female pigs. Meat Science 85, 664–670. 8 P. Aymerich, M.D. Tokach, S.S. Dritz et al. Animal 15 (2021) 100218 Heger, J., Křížová, L., Šustala, M., Nitrayová, S., Patráš, P., Hampel, D., 2009. Pinheiro, J., Bates, D., 2000. Mixed-effects Models in S and S-PLUS. Springer Science Individual response of growing pigs to lysine intake. Journal of Animal & Business Media, New York, NY, USA. Physiology and Animal Nutrition 93, 538–546. Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., Team, R.C., 2019. _nlme: Linear and Jackson, C.H., 2011. Multi-State Models for Panel Data: The msm Package for R. Nonlinear Mixed Effects Models_R package version 3.1-139. Retrieved on 26 Journal of Statistical Software 38, 1–29. https://doi.org/10.18637/jss.v038.i08. April 2020 from https://CRAN.R-project.org/package=nlme. Kelley, G.A., Kelley, K.S., 2019. Systematic reviews and meta-analysis in nutrition van der Peet-Schwering, C.M.C., Bikker, P., 2018. Amino acid requirement of research. British Journal of Nutrition 122, 1279–1294. growing and finishing pigs Report 1101. Wageningen Livestock Research, Kill, J.L., Donzele, J.L., De Oliveira, R.F.M., Ferreira, A.S., Lopes, D.C., De Silva, F.C.O., Wageningen, The Netherlands. Barbosa Da Silva, M.V.G., 2003. Níveis de lisina para leitoas com alto potencial Pesti, G.M., Vedenov, D., Cason, J.A., Billard, L., 2009. A comparison of methods to genético para deposição de carne magra dos 65 aos 95 kg. Revista Brasileira de estimate nutritional requirements from experimental data. British Poultry Zootecnia 32, 1647–1656. Science 50, 16–32. King, R.H., Campbell, R.G., Smits, R.J., Morley, W.C., Ronnfeldt, K., Butler, K., Dunshea, Quiniou, N., Courboulay, V., Salaün, Y., Chevillon, P., 2010. Impact of the non- F.R., 2000. Interrelationships between dietary lysine, sex, and porcine castration of male pigs on growth performance and behaviour-comparison with somatotropin administration on growth performance and protein deposition barrows and gilts. Conference at the 61st Annual Meeting of the European in pigs between 80 and 120 kg live weight. Journal of Animal Science 78, 2639– Association for Animal Production, 23–27 August 2010, Heraklion, Crete Island, 2651. Greece, paper 8. Main, R.G., Dritz, S.S., Tokach, M.D., Goodband, R.D., Nelssen, J.L., 2008. Determining R Core Team, 2019. R: A language and environment for statistical computing. R an optimum lysine: calorie ratio for barrows and gilts in a commercial finishing Foundation for Statistical Computing, Vienna, Austria. https://www.R-project. facility. Journal of Animal Science 86, 2190–2207. org/. van Milgen, J., Gloaguen, M., Le Floc’H, N., Brossard, L., Primot, Y., Corrent, E., 2012. Remmenga, M.D., Milliken, G.A., Kratzer, D., Schwenke, J.R., Rolka, H.R., 1997. Meta-analysis of the response of growing pigs to the isoleucine concentration in Estimating the maximum effctive dose in a quantitative dose-response the diet. Animal 6, 1601–1608. experiment. Journal of Animal Science 75, 2174–2183. van Milgen, J., Valancogne, A., Dubois, S., Dourmad, J.Y., Sève, B., Noblet, J., 2008. Rikard-Bell, C.V., Pluske, J.R., van Barneveld, R.J., Mullan, B.P., Edwards, A.C., Gannon, InraPorc: A model and decision support tool for the nutrition of growing pigs. N.J., Henman, D.J., Dunshea, F.R., 2013. Dietary ractopamine promotes growth, Animal Feed Science and Technology 143, 387–405. feed efficiency and carcass responses over a wide range of available lysine levels Moore, K.L., Kim, J.C., Mullan, B.P., 2015. Dietary lysine requirements and feeding in finisher boars and gilts. Animal Production Science 53, 8–17. regimes for finisher pigs. Retrieved on 12 March 2020 from https:// Robbins, K.R., Saxton, A.M., Southern, L.L., 2006. Estimation of nutrient australianpork.infoservices.com.au/items/2011-1034-247REPORT. requirements using broken-line regression analysis. Journal of Animal Science Moore, K.L., Mullan, B.P., Campbell, R.G., Kim, J.C., 2013. The response of entire male 84, E155–E165. and female pigs from 20 to 100-kg liveweight to dietary available lysine. Animal Lenth, Russell, 2020. emmeans: Estimated Marginal Means, aka Least-Squares Production Science 53, 67–74. Means. R package version 1, 5 https://CRAN.R-project.org/package=emmeans. Moore, K.L., Mullan, B.P., Kim, J.C., Dunshea, F.R., 2016. Standardized ileal digestible Shelton, N.W., Tokach, M.D., Dritz, S.S., Goodband, R.D., Nelssen, J.L., DeRouchey, J. lysine requirements of male pigs immunized against gonadotrophin releasing M., 2011. Effects of increasing dietary standardized ileal digestible lysine for factor. Journal of Animal Science 94, 1982–1992. gilts grown in a commercial finishing environment. Journal of Animal Science NRC, 1998. Nutrient Requirements of Swine. National Academic Press, Washington, 89, 3587–3595. DC, USA. St-Pierre, N.R., 2001. Invited review. Integrating quantitative findings from multiple NRC, 2012. Nutrient Requirements of Swine. National Academic Press, Washington, studies using mixed model methodology. Journal of Dairy Science 84, 741–755. DC, USA. Trefan, L., Doeschl-Wilson, A., Rooke, J.A., Terlouw, C., Bünger, L., 2013. Meta- O’Connell, M.K., Lynch, P.B., O’Doherty, J.V., 2006. Determination of the optimum analysis of effects of gender in combination with carcass weight and breed on dietary lysine concentration for boars and gilts penned in pairs and in groups in pork quality. Journal of Animal Science 91, 1480–1492. the weight range 60 to 100 kg. Animal Science 82, 65–73. 9