Variability in feed intake the first days following weaning impacts gastrointestinal tract development, feeding patterns, and growth performance in nursery pigs

Abstract The present study investigated the effects of voluntary feed intake (FI) the first days after weaning on gastrointestinal development and protein fermentation the first week after weaning and growth performance and feeding patterns during the nursery phase. A total of 144 mixed-sex weaned pigs (24 ± 2 d old; 7.2 ± 0.8 kg body weight [BW]) were allocated to 12 pens with 12 pigs/pen. Each pen was equipped with an electronic feeding station for monitoring individual FI during a 40-d study. Pigs were classified based on their cumulative FI over the initial 3 d after weaning (FId1-3) being above or below their pen median FId1-3 (high = 626 ± 193 g or low = 311 ± 181 g FId1-3). Similarly, weaning BW classes (BW0; high = 7.72 ± 0.59 kg or low = 6.62 ± 0.88 kg BW) were created to study interactions with FId1-3. Two female pigs with either a high or a low FId1-3 per pen (n = 24) were selected for sampling at d6 and were used to study gastrointestinal development and fermentation products in the small intestine. Feeding patterns per day, FI, and growth performance were measured individually. Low FId1-3 pigs had lower (P < 0.05) daily FI during d0 to d8, d8 to d15, and d22 to d28, BW on d15, d22, d29, and d40, and average daily gain during d0 to d8, d22 to d29, and d29 to d40 compared to high FId1-3. High FId1-3 pigs increased (P < 0.05) the number of visits to the feeder between d1 to d13 and d31 to d35, and the time spent per visit only for d1 to d4 (P < 0.05). The daily rate of FI (g/min) was higher (P < 0.05) for High FId1-3 pigs on d6, d8, d9, and d10, and again several days later (d20 to d39). In addition, the high FId1-3 × high BW0 interaction improved daily FI during d18 to d40 compared to low FId1-3 × low BW0 class (P < 0.05). For the sampling on d6, low FId1-3 pigs had a lighter small intestine, colon, and pancreas, and reduced villi length, smaller villi surface area, and a lower number of goblet cells size in jejunum (P < 0.05), while concentrations of lactic acid, histamine, and cadaverine in small intestinal content were increased (P < 0.05). In conclusion, pigs with high FId1-3 became faster eaters with higher FI and growth rates toward the second half of the nursery, which was similar and additive for pigs with higher weaning BW. High FId1-3 was also associated with greater development of the gastrointestinal tract and a reduced protein fermentation 1-wk after weaning.


Introduction
The weaning event and the transition to solid feed remain a challenge to pig husbandry.In effect, it is common that weanling pigs show null or low feed intake (FI), as below maintenance energy requirements, for the first 3 to 5 d after weaning (Le Dividich and Herpin, 1994;Bruininx et al., 2001).The lack of FI and, thereby, nutrients in the intestinal lumen results in disturbances including intestinal villus atrophy, reduced enzyme activity and nutrient absorption (Brooks and Tsourgiannis, 2003;Moeser et al., 2017), increased intestinal permeability (Carey et al., 1994), and inflammation (Lallès et al., 2007) which altogether results in a period of lower bodyweight (BW) gain (Dong and Pluske, 2007).Furthermore, the presence of unabsorbed nutrients in the lumen, especially protein, causes undesired intestinal fermentation which is widely acknowledged as a risk factor for digestive disorders and disease (Kim et al., 2012;Zhang et al., 2020).Proteolytic fermentation by intestinal microbiota includes relevant end-products such as short-chain and branched-chain fatty acids, ammonia, hydrogen sulfide and methanethiol, phenol and indole compounds, and biogenic amines (Zhang et al., 2020).Importantly, high concentration of biogenic amines in the intestinal lumen have been associated with diarrhea, inflammation, and poor performance in pigs (Pieper et al., 2012;Li et al., 2022).Nonetheless, the associations between poor FI early after weaning, gastrointestinal development, protein digestion, and performance remain largely unknown.This is especially the case under the current worldwide trends of reducing dietary N emissions, and the refraining from antibiotics and ZnO in feed.
Several internal and external factors influence FI immediately after weaning and, thereby, the adaptation to solid feed.Piglet birth weight, general health, stress, growth, and gastrointestinal development prior to weaning, and BW and age at weaning are acknowledged as crucial factors to prepare the pig for overcoming the diet shift (Lallès et al., 2007).Visual sorting of pigs by size as a proxy for weaning BW could be a trait used by producers.Weaning BW is predictor for performance up to slaughter (López-Vergé et al., 2018).However, heavy pigs at weaning may be more susceptible to weaning stressors.In fact, a heavier category of pigs at weaning showed lower initial FI after weaning than their lighter counterparts (Bruininx et al., 2001).Faster-growing piglets during the suckling phase were associated with higher sow milk consumption (Tang et al., 2022); however, when lacking familiarity and interest in solid creep feed, heavy pigs seem more susceptible to weaning stress (Tang et al., 2022).Despite extensive research and a wide array of measures developed to enhance FI early after weaning (Wensley et al., 2021), little is known about the natural variance of immediate after weaning FI within a group of pigs and the reasons driving it.In a recent paper, Engelsmann et al. (2023) reported that individually caged pigs with a higher FI d0 to d4 after weaning had improved overall growth performance and villus surface area with more acid mucus-producing goblet cells (GC) but increased concentrations of inflammatory markers, higher risk of diarrhea, and a higher frequency of antibiotic use compared to pigs with lower FI.Thus, a higher FI early after weaning may not be necessarily better for pig health and antibiotic use, while further investigation in this area and group-housing systems is warranted.
We hypothesized that poor FI early after weaning would impair gastrointestinal tract (GIT) function with negative effects on growth performance.Furthermore, high vs. low early after weaning FI would be associated with different feeding patterns and interact with weaning BW.Pigs with high weaning BW perform worse short term but better long term in the nursery.Therefore, the present objectives were to characterize pigs with a high or low early FI (d1 to d3) after weaning to study 1) their individual feeding patterns and performance throughout the nursery phase, 2) their interaction with high and low weaning BW, and 3) their effect on GIT organ weights, histomorphometry, and protein digestion the first week after weaning.

Ethical approval
The internal animal welfare body of Trouw Nutrition R&D reviewed and approved the experimental protocol and followed the Central Committee Animal Experimentation (CCD, The Hague, The Netherlands) approval with animal use protocol no.AVD2040020184665.

Animals, housing, and experimental design
A total of 144 mixed-sex pigs (Hypor Libra × Hypor Maxter, Hendrix Genetics B.V., Boxmeer, The Netherlands) were used in a 40-d study at the Trouw Nutrition R&D Swine Research Centre (St.Anthonis, The Netherlands).Pigs were selected from 20 litters excluding pigs that previously received medication or had poor growth during the suckling phase.From d5 all litters were offered a piglet milk replacer (i.e., crude protein [CP] = 20%, ether extract [EE] = 15%, lactose = 36%, net energy [NE] = 13.8MJ/kg, standardized ileal digestible (SID) Lys = 16 g/kg; Milkiwean Babymilk; Trouw Nutrition Nutreco Belgium NV, Ghent, Belgium) and from d12 onwards they were offered a standard commercial creep feed (i.e., CP = 19%, EE = 10%, lactose = 15%, NE = 11.7 kcal/kg, SID Lys = 14 g/kg; Milkiwean Precoce; Trouw Nutrition Nutreco Belgium NV, Ghent, Belgium) until weaning.Pigs were weaned at 24 ± 2 d of age (7.16 ± 0.80 kg BW) and allocated to two rooms with 6 pens/room and 12 pigs/pen.The allocation was based on BW at weaning (BW d0), litter origin, and sex with 5 males and 7 females per pen.Each pen (3.0 m × 1.67 m) was equipped with an electronic feeding station (EFS, Schauer Agrotronic, GmbH, Austria) having 15 cm feeder space.The EFS was activated by the pig's ear tag Radio-Frequency IDentification (RFID, MS Tag Round HDX STF-YELOPPRPB1, MS Schippers BV, Hapert, The Netherlands) enabling monitoring individual FI and feeding pattern.Room lighting and EFS lights were switched on between 0600 and 2200 hours and temperature was set at 30 °C at weaning and gradually reduced to 25 °C on d28 until the end of the nursery.
FI classes were defined based on cumulative FI over the initial 3 d (FId1-3) after weaning.The FId1-3 criterion was chosen based on literature (Le Dividich and Herpin, 1994;Bruininx et al., 2001) as having negative effects on the GIT and performance (Spreeuwenberg et al., 2001;Engelsmann et al., 2023).Within each pen, pigs were classified for FId1-3 as above (High n = 58; or below (Low n = 53; 311 ± 181 g FId1-3) their pen median FId1-3 after removing the two pigs per pen that were selected for euthanasia and sampling on d6 (see below) and other removed pigs.The FI on d0 was not recorded because the EFS was continuously open without ear tag registration for adaptation.Besides the FId1-3 class, a similar classification was created based on weaning BW, and pigs were classified as High (n = 56; 7.72 ± 0.59 kg BW) or Low weaning BW (n = 55; 6.62 ± 0.88 kg BW) relative to their pen median for further comparison.The choice for studying average high and low BW classes instead of four quartiles was made 1) to evaluate if being an average heavy or a light pig was already meaningful in its interacting with initial FI following weaning, and 2) so equivalent BW distribution was present within FId1-3 classes.
At d 6 after weaning, two female pigs per pen (n = 24 in total) were selected for euthanasia and tissue sampling.One female pig per pen had a High FId1-3 and the other female pig per pen a Low FId1-3 (highest and lowest in the pen, respectively) to study the effect of initial FI on GIT, including organ weights, jejunum histomorphometry, small intestine fermentation products and CP content as detailed later at sample collection.Only females were selected to reduce variation and minimize sample size since females showed more adverse responses in GIT development caused by weaning stress than their male counterparts (Pluske et al., 2019).
Two experimental diets were formulated and produced at an experimental feed plant (ForFarmers, Heijen, The Netherlands) and were fed during d0 to d15 and d15 to d40, respectively, to all pigs.Diets were a wheat-barley-soybean meal-based diets formulated to meet or exceed the nutrient requirement of weanling pigs (NRC, 2012).During the entire period, pigs had ad libitum access to feed and water (one nipple per pen).

FI, feeding patterns, and growth performance
The EFS recorded feed disappearance and start and end time of each visit per pig using a RFID tag.Failed identification or incorrect recordings of feed disappearance are main sources of errors in the data collected by EFS (Eissen et al., 1998).Therefore, feed disappearance data per visit were screened before further calculations of feeding patterns as proposed by Bruininx et al. (2001Bruininx et al. ( , 2003)).Furthermore, weekly manual weighing of left-over feed per pen was compared to the EFS raw data to ensure adequate functioning of the EFS as per standard operational procedures.The data curation process allowed for individual monitoring of FI pattern variables.Data included the daily number of visits to the trough (we only report visits with feed disappearance > 0 g), time per visit, and amount consumed per visit from d1 after weaning onwards.Variables were used to calculate FI per day and the rate of FI as FI per visit divided by visit time per day (in g/ min).
At dissection, measurements included BW, organ weights including pancreas, liver, and for the GIT sections full and empty weights of the stomach, small intestine, cecum, and colon including the rectum.The content weights were calculated by difference and the weights from the empty organs and their contents were expressed as g/kg BW at euthanasia.A homogeneous 50 g content sample from the stomach, pooled together small intestine sections (duodenum, jejunum, and ileum content), and pooled content from the colon were collected, snap frozen on dry ice, and stored at −20 °C until further analysis.The pH was determined in the contents of the stomach, pooled together small intestine content, and colon (pH meter WTW 3110 ProfiLine, 2023 Thermo Fisher Scientific Inc., Waltham, MA).Furthermore, a 3-cm tissue sample from jejunum was obtained from 30 cm caudal to the pylorus and fixed in a cassette including 4% buffered formaldehyde for histomorphometry and GC counting and measuring by the laboratory GD animal health (Royal GD, Deventer, The Netherlands).

Histomorphometry and GC in jejunum
After arrival of the tissue samples at the laboratory, they were dehydrated and embedded in paraffin.Tissue blocks were sectioned at 2 μm, mounted on a glass slide, and stained with the combined Alcian Blue (pH 2.5)/PAS procedure.For each batch of slides stained, an Alcian Blue/PAS-positive control slide was included.Microscopic images of representative cross-sections of each tissue were captured by a microscope (Olympus BX41) connected to a digital camera (Olympus Dp26) and analyzed using Olympus cellSens Dimension version 1.12 software.Of each sample, 10 randomly selected well-oriented intact paired villus-crypt units were measured.Villus length (axis top to villus-crypt junction) and crypt depth (from villus-crypt junction to the base of villus) were used to calculate villus length to crypt depth ratio.Goblet cell size and counts of total GC and neutral (PAS positive) and acid mucin (Alcian Blue positive) producing GC were determined and analyzed in five representative well-oriented villi.The GC size and number were reported in μm 2 , as area per villus surface area (in μm 2 ), and as percentage of the villus surface area.The average of each villus:crypt ratio, and the total, neutral, and acid mucin GC characteristics were calculated and reported per animal.The coefficient of variation (CV, %) for each variable was calculated on animal level with five measurements per animal.All morphometric measurements were performed by a European College of Veterinary Pathologists board-certified veterinary pathologist.

Fermentation products in small intestine
The content of volatile fatty acids (VFA) and other fermentation products in whole small intestinal pooled contents was analyzed including acetic, butyric, propionic, succinic, valeric, lactic, iso-butyric, and iso-valeric acid.Analyses were conducted at MasterLab (Boxmeer, The Netherlands) as described elsewhere (Van Hees et al., 2019).Briefly, 1 g of homogeneous digesta per sample was diluted with 2 M sulfuric acid, thoroughly mixed and centrifuged.The supernatant was analyzed for VFA by HPLC on a BioRad Aminex HPX-87H using a 0.005 mol/L sulfuric acid eluent at a flow rate of 0.7 mL/min.Biogenic amines contents, i.e., cadaverine, histamine, putrescine, spermidine, spermine, and tyramine, in small intestinal content were analyzed according to Rosier and Van Peteghem (1988).In brief, a 2 g homogeneous sample was extracted by adding 10 mL trichloric acetic acid 5%, and derivatized with dansyl chloride and NaOH.After derivatization, 0.1 mL extract was mixed with 5 mL cyclohexane hexane.After vortexing and centrifuging (180 × g for 5 min at 20 °C), the cyclohexane layer was collected, dried using N 2 at 45 °C, and resuspended in 1000 μL methanol/acetonitrile (50:50).Thereafter, the extract was analyzed using a Spherisorb C8, 5 µm 250 × 4.6 mm (BGB Analytik, Harderwijk, the Netherlands) on a Thermo Quantis liquid chromatography-tandem mass spectrometer (Heated Electrospray Ionization source) with an Ultimate 3000 pump and auto-sampler (Thermo Fisher Scientific).The elution buffer was a gradient with acetonitrile, water, and formic acid (0.3%).Some analysis was below detection limit (BDL, i.e., <10 mg/L for VFA and < 100 μg/kg for biogenic amines).Data for butyric, iso-butyric, propionic, valeric, and isovaleric acid were BDL in all samples and therefore were not reported.For the other VFA, in case some samples were BDL but not all, the detection limit value was assigned to the BDL samples to run statistical analysis adequately.Acetic acid and succinic were given a value of 8 mg/L in n = 3 and n = 14 samples, respectively.Tyramine had n = 20 BDL (out of n = 24 in total) and data was, therefore, not reported as it could not be statistically analyzed.

CP and lysine content
The first intention was to conduct CP and Lys digestibility measurements from ileum content but digesta amounts were insufficient for sample processing and analysis in several pigs, likely because of Low FId1-3 selection criteria.In situ, it was decided to pool the content of the small intestine sections as a whole sample per pig (amounts ranged between 25 and 200 g of fresh sample).Therefore, digestibility values are not reported.Small intestine and colon content samples were dried at 70 °C to constant weight in a Memmert oven (Memmert GmbH) to obtain the dry matter concentration (Masterlab, Boxmeer, the Netherlands).After drying, the samples were milled on a Retsch ultra-centrifugal mill with a 0.75-mm sieve.CP was analyzed as nitrogen (N) by the combustion method (method 990.03;LECO FP 528 MI, USA) using the LECO Nitrogen analyzer and CP was calculated as N × 6.25 as in feed.Insoluble ash was measured as in feed (EC regulation 152/2009, appendix III N).The amino acid analyzer was used to determine the total Lys content in digesta according to an in-house method (Nutreco-MasterLab, Boxmeer, Netherlands) based on NEN-EN-ISO 13903.Briefly, a 0.5-to 1-g dried sample was hydrolyzed with a HCl solution and then evaporated into a film evaporator before being taken up by a buffer solution.Samples were then separated on a cation exchanger with a gradient step and postcolumn derivatization with ninhydrin; Lys was then measured at 440 and 570 nm using a flow colorimeter.All samples were analyzed in duplicate on a BIOCHROM 30 + amino acid analyzer.

Statistical analysis
The normality of data was checked by visual assessment of residual plots and PROC UNIVARIATE (SAS Inst.Inc., Cary, NC).The PROC GLIMMIX procedure was used to analyze growth performance, feeding patterns, and dissection data with dist = gaussian (normally distributed data) link = identity or dist = gamma link = log (nonnormal data i.e., biogenic amines).The statistical model for growth performance and EFS data included the fixed effects of FId1-3 class (High or Low), time (d) and their three-way interaction, and as random effects the intercept (subject = animal) as repeated factor and the pen.The weaning BW was retained as a covariate when P < 0.10.Furthermore, to compare and visualize the effects of weaning BW the model was repeated including the weaning BW class (High or Low) and interactions.For the dissection data, the models included FId1-3 class as a fixed effect and pen and room as random effects.The mean separation was done by the PDIFF option and adjusted by the Tukey-Kramer test.The PROC PLM procedure with Slice option was used to report the fixed effect P values within time interaction.The experimental unit was the pig.The P values < 0.05 were considered significant for all data, and P ≤ 0.10 were considered trends.
In addition, PROC REG was used to describe linear relationship and coefficient of determination between FId1-3 and overall ADG and ADFI variables.

Results
The diets were in accordance with the formulated values except for Lys in phase 1 diet which was 1.1 g/kg below the calculated value (Supplementary Table S1).

General performance traits
In accordance with our internal standard operation procedure, nine pigs were removed from the study since they had poor adaptation after weaning with less than 80 g FI for the first 3d.In total, three pigs died, and 21 pigs required antibiotic medication including n = 12 (22.6%)from Low FId1-3 and n = 9 (15.5%) from High FId1-3 (P = 0.338).

Feeding patterns
In Figure 2, the data from the EFS are reported as FI (A), number of visits (B), average time spent per visit (C), and the rate of FI (D) per day.Different feeding patterns throughout the nursery phase were observed between High and Low FId1-3 pigs.High FId1-3 pigs had significantly higher daily FI later in the nursery, i.e., d20, d27 to d29, and d31 to d40 (P < 0.05; Figure 2A).High FId1-3 pigs increased (P < 0.05) or tended to increase (P ≤ 0.10) the number of visits to the feeder (five visits per day more on average) on several days after weaning, i.e., between d1 and d13 (d1 to d6, d9, d10, and d13) and later in the nursery, i.e., between d31 and d35 (Figure 2B).On the other hand, the time spent per visit was only significantly higher for High FId1-3 pigs on d1 to d4 (P < 0.05) and tended to be higher on d5 and d39 (P ≤ 0.10), while Low FId1-3 pigs tended to spend more time per visit on d16 (P ≤ 0.10; Figure 2C).The daily rate of FI was higher for High FId1-3 pigs early after weaning, i.e., on d6, d8, d9, and d10, and again later in the nursery with several occasions being significantly higher (11 d) or showing trends (6 d) between d20 and d39.
Four periods with different daily FI patterns could be distinguished during the nursery phase (Figure 2A).The first week after weaning, i.e., d1 to d8, showed clear differences in FI matching the FId1-3 selection criteria.Afterward, a plateauing FI period d9 to 14 followed by a steady increase of FI for d14 to d26 for both FI classes.Finally, for d27 to d40, pigs from High FId1-3 class had increasing daily FI of around 100 g/d higher than pigs from Low FId1-3 class.Early FI × weaning BW Pigs classified as above (High) or below (Low) their pen median weighed 7.72 ± 0.59 kg and 6.62 ± 0.88 kg BW at weaning, respectively.When studying the FId1-3 × weaning BW interaction for overall performance traits (Table 1), no differences were observed (P > 0.05).The BW0 class had no effect on cumulative FId1-3.However, High FId1-3 pigs were 0.2 kg heavier at weaning (P = 0.05, SEM = 0.11) than Low FId1-3 pigs (Table 1).High FId1-3 and High BW0 pigs, both showed increased performance including higher final BW (d40), ADG d0 to d40, and ADFI d1 to d40 (P < 0.01).The FE d0 to d40 was not affected by FId1-3 or BW0 class (P > 0.05).
The data for daily FI and the interaction between FId1-3 × BW0 class × day are reported in Figure 3. Early after weaning, BW0 alone did not affect (P > 0.05) daily FI.The FId1-3 alone explained the observed higher daily FI during d1 to d5.From d18 onwards, BW0 class significantly affected FI with higher intakes for High BW0 pigs.However, for most of these days the effect appeared as an interaction.For d18 to d40, High FId1-3 and High BW0 pigs had higher daily FI compared to Low FId1-3 and Low BW0 pigs.High FId1-3 and Low BW0 pigs and Low FId1-3 and High BW0 pigs were often intermediate and not significantly different from each other.High FId1-3 with Low BW0 and Low FId1-3 with High BW0 were not different from low FId1-3 with Low BW0 for d29, d33, d35, d40 (P > 0.05).

Gastrointestinal tract development
The general results of the pigs selected for euthanasia and sampling are reported in Table 2. Pigs from the Low FId1-3 class showed lower FI 24 h prior to euthanasia than the High FId13 class (P = 0.035), however, there were no differences in FI 2 to 12 h prior to euthanasia (P > 0.05).Regarding the GIT measurements, the relative weight of (in g/kg BW) the small intestine (P < 0.001), colon plus rectum (P = 0.021), and pancreas (P = 0.028) were lower for Low FId1-3 pigs compared to High FId1-3 pigs.Other organ weights and digesta content weights were not different between FId1-3 classes (P > 0.05).
Data from jejunum histomorphometry and GC measurements are reported in Table 3.Low FId1-3 pigs had reduced villus length (P = 0.004), villus surface area (P = 0.043), and a tendency for reduced crypt depth (P = 0.056).In addition, Data LSMEANS of individual pigs per FId1-3 class as above or below their pen (n = 12 pens) median for FId1-3 and BW0 class.The means for FId1-3 × BW0 interaction were reported as important information but the interaction with P > 0.10 was removed from the statistical model.LSMEANS within a row without a common superscript differ at P < 0.05.
x,y LSMEANS within a row without a common superscript tend to differ at P ≤ 0.10.Low FId1-3 pigs had an increased variability (CV %) in GC (P = 0.026) and acid mucus GC (P = 0.042) surface area per villus as percentage of total villus surface area and reduced size of acid mucus GC (P = 0.029).There was no effect of FId1-3 class on neutral mucus GC (P > 0.05).

Protein content and fermentation products
The pH of digesta, CP and Lys content in different GIT sections and the VFA and biogenic amines concentration in small intestinal digesta are presented in Table 4.There were no differences in pH, CP content, and Lys content between FId1-3 classes across GIT sections (P > 0.05).Regarding VFA, the Low FId1-3 class increased lactic acid (P = 0.045) and tended to increase succinic acid (P = 0.085) concentrations in small intestinal contents while acetic acid was not different between FId1-3 classes (P = 0.366).For the biogenic amines, cadaverine and histamine significantly increased (P = 0.002 and P = 0.001, respectively) and putrescine tended to be increased (P = 0.075) in Low FId1-3 class compared to High FId1-3 class.Of note, the content of biogenic amines on d6 after weaning was overall highly variable with high CV for all of them, i.e., cadaverine 50%, putrescine 44%, spermidine 42%, spermine 72% and more importantly for histamine 148%.

Discussion
The present study demonstrated the negative associations of low FI early after weaning on GIT function, feeding patterns, and growth performance in weanling pigs.Although, arguably, the two weaning BW classes represented a relatively minor difference in weight when compared to commercial conditions (1.08 kg BW difference), the high weaning BW class was found to act additively to high early FI improving performance toward the second half of the nursery.

Early FI, feeding patterns, and growth performance
The weaning transition-associated stressors are multiple and may reduce performance after weaning (Dong and Pluske, 2007;Moeser et al., 2017;Van Kerschaver et al., 2023).A clear dip was not observed in the present study, but pigs plateaued their FI and ADG between d5 and d14.Intriguingly, FId1-3 or BW0 had no effect on FI during such period nor the recovery from it.However, high FId1-3 pigs reduced FI to a larger extent than their counterparts, which may indicate some temporary negative consequences of high initial FI.As it seems, one or more factors overruled the effects of FId1-3 and BW0 classes while being responsible for the anorexigenic phenotype observed during the plateau period.Beyond altered nutrient intake and GIT functions, which are discussed later, other factors such as the microbiota-gut-brain axis and hypothalamic pituitary adrenal response can influence the voluntary FI and could help to explain the present results (Pelleymounter et al., 2000;Salfen et al., 2003;Leon et al., 2019;Margolis et al., 2021).There is evidence that at weaning, enteric generation of neurons and glial cells (non-neuronal cells) is accompanied by a functional maturation of intestinal neural circuits (Roberts LSMEANS within a row without a common superscript differ at P < 0.05.et al., 2007;Koenig et al., 2011).However, the medium-tolong term consequences of disrupted nutrient intake on such microbiota-gut-brain axis maturation are largely unknown and deserve future investigation as these could affect feeding behavior in pigs.

Feeding patterns
From our FI patterns data, pigs from High FId1-3 were visiting the feeder more often and spent more time per visit resulting in higher FI the first week after weaning, however, the pattern changed over time.The differences in FI mostly disappeared in the following 2 wk and reappeared again toward the second half of the nursery.Part of the FId1-3 effect may result from GIT development that increases growth and, thus, metabolic needs (Quiniou and Noblet, 2012;Li and Patience, 2017).Otherwise, the effect could be associated with a genetic variability in FI (Fàbrega et al., 2003;Gao et al., 2021) or other unknown factors related to dominance, feeding behavior, and learning.Unfortunately, behavior was not evaluated, and thus such discussion is speculative.The number of studies investigating individual FI and FI pattern associations to performance for nursery pigs housed in EFS are scarce (Bruininx et al., 2001(Bruininx et al., , 2002(Bruininx et al., , 2003)).Data in grow-finishing pigs indicate that the increasing FI is associated with increasing daily feeding duration in the first few weeks (Hyun and Ellis, 2000;Rauw et al., 2006).However, in time, the number of visits and feeding duration decrease with age whereas FI rate increases (Hall et al., 1999;Hyun and Ellis, 2000;Rauw et al., 2006).This is in contrast with the present study in nursery pigs where the number of feeding visits and time spent feeding were rather constant from d4 onwards.However, in general, grow-finishing pigs achieve higher FI in time through faster feeding rate (Hall et al., 1999;Kavlak and Uimari, 2019) which is in accordance with our general observations and for High FId1-3 pigs.Therefore, for nursery pigs, higher FI seemed to be the consequence of a faster FI rate rather than an effect of number of visits or the time spent per visit.The coefficient of variation (CV %) for each variable, calculated on animal level (five measurements per animal) was also analyzed but only reported when P < 0.10.

Weaning body weight
The effects of FId1-3 and weaning BW were found to be independently but additively improving performance when both were high.The importance of weaning BW is well acknowledged and commonly used as predictor for performance up to slaughter weight (López-Vergé et al., 2018;Montoro et al., 2020).The BW0 classes did not affect FI following weaning (d1 to d3).Contrasting, Bruininx et al. (2001), reported that a class of low weaning BW (6.7 kg BW) like the one in the present study (6.6 kg BW), showed higher initial (24h) FI compared to higher BW counterpart classes (i.e., 7.9 kg and 9.3 kg BW).However, in the same study, the heavier pigs consumed more feed than lighter ones from d8 to the end of the experiment, which is in alignment with our findings from d17 onwards.Bruininx et al. (2001) also reported that lighter pigs had more daily visits and a lower FI per visit the first week after weaning compared to heavier pigs.This contrasts with the present study where no differences between BW0 classes were observed.Probably, differences in classification methodology or genetic progress can explain some of the differences.We classified pigs based on pen median for the FId1-3 and BW0 classes while Bruininx et al. (2001) used the complete batch distribution.Classifying the pigs of our study as high or low FId1-3 across all pens yielded the same interpretation of results (data not shown).The approach chosen here intentionally combined and balanced the pen effect as it is known that group interactions in a pen can influence FI (Figueroa et al., 2013;Laskoski et al., 2021).In general, weaning BW in the present study was relatively high, and a lower weaning BW i.e., <6 kg may have caused greater differences or different results.In the present study, weaning BW ranged from 5.4 to 8.9 kg BW but the BW0 classes high and low were the average within pigs above and below the pen median.Hence, a limitation of current study is that the two BW0 classes were not extreme and possibly do not serve to study the lightest pigs in a batch.Instead of studying the more extreme differences, we demonstrated that an average high or average low BW pig at weaning showed differences in growth performance.This was especially the case when interactions with high and low FId1-3 classes were taken into account.Furthermore, the patterns here are limited to the data collected by the EFS and it is possible that different feeding systems or management factors yield different results (Bruininx et al., 2003;Laskoski et al., 2021).The present study pointed out the relevance of weaning BW and initial FI as independent factors influencing feeding behavior and performance early and late in the nursery phase, and how that had an impact on GIT development.

Gastrointestinal development and protein fermentation
It is known that a smooth transition to solid feed and adequate FI levels early after weaning help maintain GIT development, support nutrient absorption, and improve growth (Pluske et al., 1997;Moeser et al., 2017), which concurs with the differences observed between High and Low FId1-3 pigs.LSMEANS within a row without a common superscript differ at P < 0.05.x,y LSMEANS within a row without a common superscript tend to differ at P ≤ 0.10.
In effect, the findings for Low FId1-3 pigs on overall performance and their poor GIT development expose negative consequences for both the short-and medium-term, i.e., the first week after weaning and the end of the nursery, respectively.Low FI was associated with detrimental factors for GIT function including higher protein fermentation products in the small intestine (Pieper et al., 2012;Li et al., 2022); a reduced villus surface area in jejunum (Kelly et al., 1991;Pluske et al. 1996;Moeser et al., 2017); a reduced size and higher variability for acid mucus-producing GC (Deplancke and Gaskins, 2001;Engelsmann et al., 2023); and lighter sections of the GIT and pancreas (Owsley et al., 1986;Pluske et al., 2003).In a recent study (Engelsmann et al., 2023), these authors used a similar classification for initial FI after weaning (d0 to d4) and reported an improved growth performance and increased villus surface area with more acid mucus-producing GC in the small intestine at d28 for high FI d0 to d4 pigs compared to their low FI counterparts.Such findings agree with those in the present study where Low FId1-3 pigs had increased variability in acid mucus GC surface area and reduced size of acid mucus GC.In contrast to our findings at d6, the high FI d0 to d4 pigs did not show improvements on small intestine villus length at d28 (Engelsmann et al., 2023), suggesting that FI effect on morphometric changes on d6 as observed in the current study may be temporary (Marion et al., 2002).Engelsmann et al. (2023) also reported an increased probability for diarrhea (measured by fecal scores), higher antibiotic use, and inflammatory markers in blood for high FI d0 to d4 pigs albeit with their improved growth.However, we did not observe differences in antibiotic use or fecal scores (data not shown).Differences in health status, dietary interventions, or the fact that pigs were individually caged in that study may explain the disagreement between that study and the current one.Although inflammatory markers were not determined in the present study, the increased concentrations of histamine, cadaverine, and putrescine in small intestine digesta from Low FId1-3 pigs suggested GIT microbial dysbiosis with a higher risk for inflammation compared to their High FId1-3 counterparts, which is in line with other research (Kim et al., 2012;Zhang et al., 2020).
Histamine, which can act as a hormone and neurotransmitter, was notably increased (×4.5) in small intestine digesta of Low FId1-3 pigs.Histamine can be produced in the lumen of the gut by microbial decarboxylation of L-histidine or endogenously synthesized by mast cells, enterochromaffin cells, and histaminergic neurons to exert a pro-inflammatory response (Maintz and Novak, 2007;Wang et al., 2018).Therefore, part of the histamine reported in Low FId1-3 pigs is likely endogenous since its increase was greater than the one for other biogenic amines.In fact, histamine is a valid biomarker for gut health in pigs also prior to weaning (Ramsay et al., 2020).In previous studies, elevated histamine levels originating from high protein diets were associated with diarrhea (Pieper et al., 2012) and the induction of Cl -secretion (Kröger et al., 2013) in pigs.Although we did not find significant differences in CP content in whole small intestine or colon at d6, a carryover effect of low FI during first days after weaning on protein malabsorption is expected with the reduced intestinal surface observed.As such, the consequences may impact intestinal microbiota, produce biogenic amines and inflammation (Moeser et al., 2017;Li et al., 2022) in agreement with the poor performance observed for Low FId1-3 pigs.Under healthy conditions, histamine and other biogenic amines are detoxified by enzymes in the intestinal epithelium i.e., histamine N-methyltransferase, monoamine, and diamine oxidases (Hughes et al, 2000).Arguably, the hypothetical loss of such enzyme activity in the villi tips may occur due to damage or shortening of the villi which was observed in Low FId1-3.Hence, reduced surface area and enterocyte maturity would further contribute to the negative effects of histamine.Interestingly, Kröger et al. (2013) demonstrated that pigs adapt to such conditions and reported an increased activity and gene expression of histamine-degrading enzymes by 21 d following weaning when fed a high CP diet (20% vs. 14.5% CP).Nonetheless, even after an eventual adaptation, the disruption of nutrient intake and the presence of undigested protein and protein catabolites immediately after weaning seem detrimental to growth and FI as demonstrated in the present study.

Conclusion
A high FI early after weaning was associated with a greater development of the gastrointestinal tract and reduced protein fermentation in SI digesta at d6 after weaning.Furthermore, such a phenotype appears to be associated with faster eaters and higher FI and growth toward the end of the nursery phase.The effect of weaning BW was additive and independent from FId1-3 with higher weaning BW increasing growth performance from the second half of the nursery.The present study highlighted and reinforced the relevance of weaning BW and initial FI as independent factors influencing feeding patterns and performance over time.In general, it is important to continue investigating the factors affecting the individual and initial FI following weaning due to its relevance for gastrointestinal development, effect on feeding patterns and behavior, and carryover effects on FI and growth performance in later life.

Figure 2 .
Figure 2. Effect of FI d1 to d3 class (FId1-3) × day interaction on FI (A), number of visits (B), average time spent per visit to the feeder (C), and the rate of FI (D) per day 1 . 1 Data presents LSMEANS ± SE for pigs from FId1-3 class as above (High, n = 58) or below (Low, n = 53) their pen (n = 12 pens) median for FId1-3.

Table 2 .
General traits and gastrointestinal measurements for pigs selected based on FI d1 to d3 (FId1-3) 1 1 Data LSMEANS of individual female pigs per FId1-3 class as highest or lowest FId1-3 per pen (n = 12 pens).2 Standard error of the mean.a,b

Table 4 .
The pH, crude protein, and Lys content in different gastrointestinal tract sections, and fermentation products in the small intestine for pigs selected based on FI d1 to d3 (FId1-3) 1 1 Data LSMEANS of individual female pigs per FId1-3 class as highest or lowest FId1-3 per pen (n = 12 pens).CP = crude protein; DM = dry matter; Lys = lysine.a,b