Early Development of Locomotion in the Term Piglet Model: Does Size Matter?

Synopsis Intrauterine undernutrition in humans typically results in low birth weight ([small for gestational age] SGA) and delayed postnatal neuromotor maturation. Since SGA and intrauterine growth retardation are also common in domestic pigs, piglets are premised as models to study delayed motor development. Applied to the locomotor paradigm, however, questions emerge: (i) how to map the developmental time scale of the precocial model onto the altricial target species and (ii) how to distinguish size from maturation effects? Gait data were collected at self-selected voluntary walking speed during early development (0–96 hours postpartum; pp) for SGA- and normal ([appropriate for gestational age] AGA) piglets. Dimensionless spatiotemporal gait characteristics (according to dynamic similarity) become invariant already after 4 hours pp, suggesting rapid postnatal neuromotor maturation. Moreover, dimensionless gait data are largely identical for SGA- and AGA-siblings, indicating that primarily size effects explain absolute locomotor differences. This is further supported by (i) normalized force-generating capacity of limb muscles, (ii) joint kinematics (<10 hours pp), and (iii) normalized ground reaction forces (<5 days pp) being indifferent between SGA- and AGA- piglets. Furthermore, predictive modeling based on limb joint kinematics is unable to discern the majority of SGA- from AGA-piglets (<10 hours pp). All this leads to the conclusion that, although smaller than the AGA piglets in absolute terms, SGA-piglets mature (neuromechanically speaking) just like, and equally fast as their AGA littermates. Yet, it remains a fact that early SGA piglets are reported to be less mobile, less vital, and less competitive than their AGA siblings (even often die before day 3 pp). This conspicuous difference likely results from the energy level (blood glucose and glycogen) and its mobilization being considerably different between the piglet categories during early development.


Introduction
Huma n f etuses a re cla ssified a s sm a l l for gestat iona l age (SGA) when o n, o r be low, th e 10th percent i le of t he fet al Body m a ss (BM) growt h curve ( Batt aglia and Lu bch en co, 1967 ; Beun e et a l.2018 ; se e a l so Ki serud et al. 2018 ). Thi s implies th at, at term birth, infants look 2-4 w eeks y oun g er, and hav e 5-10% less BM comp are d to the av erag e neonat e .Oft en, SGA infa nts suffer from Intra Uter ine Growt h Restr ictio n (o r r etar dation; IUGR), which is lin ke d to s e vere ly in cre asing r isks of mo rb idi ty an d m o rtali t y w it h decre asing BM-percent i le ( Beune et a l. 2018 ;Esi h et al. 2022 ;G ia bicani et al. 2018 ).Consid-er ing t hat, worldwide, 10-15% of a l l preg na ncies ca n be cla ssified a s IUGR (up to 30% in dev elopin g countries), thi s i s far from triv i al ( de Onis et a l. 1998 ;Le e et al. 2013Le e et al. , 2017 ; ;McCowan et al. 2018 ).
IUGR infa nts a re bo rn wi th, amo ngst others, immature bra ins (f ewer neuro ns, hypo myelinatio n; fo r a revie w s ee: Baud an d Ber kan e, 2019 ).This can, during f urt h er deve lopm ent, be lin ke d to n euro-m oto r p roblems and defici ts (p roblems in mov ing , ma inta ining b a lance or posture , et c.; Esih et al. 2022 ).Desp i te the a bov em ention ed high prevalen ce of IUGR ca ses, surpri singly few studies provide ca usal insigh ts in to th e m ech ani stic A dvance A ccess pu blication Jun e 12, 2023 C Th e Auth or(s) 2023.Pu blis h ed by Oxford University Press on behalf of the Society for In tegra tive and Com para tiv e B iology.This is an Open Access article dist ribute d under the terms of the Creative Co mmo ns Attribu tio n License ( https://creativecommo ns.o rg/licenses/by/4.0/), which per mits unrestr ict ed reuse , dist ribut ion, and r epr oduction in any medium, provided the orig ina l work is properly cit ed .aspects of th ese n euro-m otor deve lopm ental problems.This ca l ls fo r mo re fun dam enta l (clinica l) r esear c h t o develop st rateg ies fo r p rev entiv e ca re a nd ap pro priate t reat ment.
Cle arly, bot h from the pract ica l and (espe cia l ly) the ethical point of vie w, res ea rch on huma n neonates a nd ea rly infa nts is fa r from e vident, e ven more s o when these infa nts a r e classified SGA/IUGR.Ther efor e, both a r epr esen ta tive an d re levant m otor p aradig m an d th e prop er mo del sp ecies are indisp en sa ble.
L o co motio n is co mmo nly co nsidered a co nvenient o p tio n fo r this purpos e.Its de ve lopm ent proce e ds a long we ll-defin ed mileston es that ar e r eadily and easily accessi ble an d qu antifiab le ( Ado lp h and Rob inso n, 2015 ).At the same time, how ev er, loco motio n is, desp i te i ts cyc lical nature , a m ost complex m otor tas k: pro p ulsion must be co mb ined wi th p roper b a la nce-a nd postural contro l ( Ado lp h and Robinson, 2015 ; Ha l lemans et a l.2004 , 2006 ).As suc h, ext er o-and pr o priocep t ive fe e dback, r eflexive r espo nse, cogni tive antici patio n, and adju stments, a s well a s the ba sic rhythmic fe e dforward drive, are a l l integ rate d into one and the same p aradig m.Also, fro m a b roader b iological p ersp ective, the fo cus o n loco motio n i s relevant a s i t co nsti tu tes an impo rtant behavio ral co mpo nent o f many e colog ica l funct ions.
With r egar ds to the pr op er mo del sp ecies, piglets are o ften pu t fo rward as the re levant su bsti tu t e , a s rem arkab le p hysio log ica l and interna l anatomica l simi larit ies exist with huma n infa nts (e.g., Mudd and Dilger, 2017 ;Odle et al. 2014 ).Mo reover, and impo rtant in the context of SGA/IUGR, bre e ding sele ct ion for lar g er litter sizes h a s resu lte d in increase d percentages ( > 20%) of SGA indiv idu a ls, born in mixe d litt er s t oget her wit h o therwise ap pro pr iate for gest at iona l age (AGA) littermates.Below a BM of 1.1 kg at term birth, early mortality increases sharp l y ( F eldpa usch et al. 2019 ), and these SGA piglets also suffer from increased mo rb idi ty.Prolon g ed farrowin g and associated hypoxia are im portan t fact or s in this, but the effect of the increased competitio n fo r nu trients amo ng li tt ermat es cann ot be un derest imate d ( Baxter et al. 2008 ;Muns et al. 2016 ;Pandolfi et al. 2017 ).This co mpeti tio n not only happens during s uckling (acces s t o the t eats), but already starts in utero durin g g esta tion, th us leading to varying levels of IUGR in SGA piglets ( Va l let and Frekling, 2007 ;Vellet et al. 2011 ).
As for human term SGA/IUGR -infants, SGA/IUGRpiglets ap pear yo unger at birth (comp are d to the average BM of AGA piglets, the a bov e-ment ione d 1.1 kg boundary accords to a seeming difference of ∼2 weeks; F eldpa us ch et a., 2019 ).The y are not on ly sma l ler but also s h o w lo wer vi tali ty (th e leve l depen ding on the IUGR s e verity; s ee further), which leads to red uced co mpeti tiven ess re lative to th eir AGA litt ermat es ( Baxter et al. 2008 ;Litten et al. 2003 ;Muns et al. 2016 ;Pandolfi et al. 2017 ).Furt her more, t hey suffer from t he s a me bra in develop mental p ro blems and s ubsequent n euro-m oto r defici ts a s hum an SGA-neon ates a nd infa nts do a n d s h ow, wh en s e v erely IUGR, con spicuous dev i ating h ead m orph ologies ("dolphin"-s hape; Douglas et a l. 2016 ;Ha les et al. 2013 ;Ridders h olm et al . 2021 ).As suc h, the t erm SGA/IUGR-piglet seems to r epr esent the p erfect mo del r equir ed to fill the a bov em ention ed kn owledge gap in th e n euro-m otor maturatio n p roblems o f term huma n SGA/IUGR-infa nts.How ev er, befo re co ncl uding this, so me impo rtant is s ues must be addressed.
First, there is the problem of mapping ontogenetic t ime sca les ( Garwicz et al . 2009 ).Reac hing the s ta ge of in depen dent wa l king ta kes a pproxima te ly on e year in huma ns, a n d even th en, s e vera l we e ks/m onths wi l l p ass before a som e h ow regular pattern emer g es.Wildebeest calves, on the other ha nd, sta nd, wa l k, and even ga l lop wit h t h e h erd wi thin minu t es aft er birt h.In ot her words, the post part um ( pp ) mat urat iona l t im e win dow may differ en orm ous ly b etween sp ecies (i .e., a lt ricia l versus pre cocia l, respe ct i vel y).Piglets, too, ar e pr e cocia l, and it s h ould be ques tioned firs t what the relevant time window for th eir (loco-)m otor ma tura t ion is.Se cond, it s h ould be establis h ed in w hich aspec ts and to what extent SGA-piglets differ in the pr ogr ess of their (loco-)motor ma tura tion from the AGA litt ermat es.If differences exist, it should (third) be questioned what aspects of these differences a re just a matter of size (sma l l v er sus ap pro priat e size), whic h aspects are caused by altered (de layed) n euro-m otor ma tura tion, a nd which a re the resul t o f st i l l other fact or s.
In this paper, we spe cifica l ly address these questions.For this purpose, we int egrat e results of our former studies on the locomotion of term AGA piglets and SGA piglets (sp at iotempora l gait data: Vanden Hole et al . 2017Hole et al . , 2018a ; musc le m orph om etrics an d fib re typ ing: Vanden Hole et al. 2018bHole et al. , 2019a ; ;ener g etics: Vanden Hole et al. 2019b ) with new data on join t kinema tics and single-limb dynamics.The in terpreta tion of this new synthesi s i s m at c hed a gains t the ou tco me o f Bayesian pre dict ive m ode ling ( Mie l ke et a l.2023 ).

Conceptual framework
To an sw er the question s posed, w e re ly on th e basic control sch em e f or volunta ry locomotion ( Fig. 1 ; e.g., Gri l lner, 2011 ; Gri l ln er an d El Manira, 2019 ; Nishikawa et al. 2007 ).In ten t iona l locomot ion init iates as a simple, g rade d driv e, descendin g from the brain to the spinal Central Pa ttern Genera tors (CPGs; 1 in Fig. 1 ).Th ese CPG n etwor ks convert t his dr ive into coordinat ed , synergi stic mu sc le-t endon act ivat ion p atterns for a l l limb joints ( 2 in Fig. 1 ).Both levels (brain and spinal) can be subj e ct to external and peripheral fe e db ack.It is, how ev er, the interactio n o f the coo rdina ted m uscle activa tio n wi t h t he inter nal and exter nal mecha nics a nd with the inertia of the system ( 3 in Fig. 1 ) th at fin a l ly resu lts in the observable, in other words direct ly me asurable , ext erna l col le ct ive o utp ut from the en tire in tegra ted n eurom echanical locom ot or syst em ( 4 in Fig. 1 ): i .e ., the sp at iotempora l gait data, in the joint kin ematics, an d th e single-limb groun d reactio n fo rces.As such, any observed ontogenetic change in any of the col le ct ive o utp ut variables of a specific loco motio n task (f or insta nce, wa l king at s elf-s ele cte d voluntary spe e d on level ground) must reflect al teratio n wi t hin t he integ rate d n eurom e chanica l loco moto r system: (i) neurome chanica l ma tura tion a t any level of the system (cent ra l or periph eral n ervou s system, mu scle-tendo n uni ts, m orph olog ica l aspe cts, etc.) and/or (ii) pure size effects.
W hen co mpar ing t he exe cut ions of specific (loco)m otor tas ks o f similar bu t different-sized a nim al s, size effects can be excluded by normalizing according to dynamic similari ty p rinci ples ( Alexander, 1977( Alexander, , 2003( Alexander, , 2005 ; Alexa nder a nd Jayes, 1983 ).In concep t, mo tor tasks are dy namic a l ly simi lar when a l l lin ear dim en sion s (of th e m ovem ents, as we ll as of the musculos ke letal system itse lf) c an be sc a le d by a single factor α (i.e., in concept, anim al s should scale isometrica l ly), a l l t im e-re late d aspe cts of th e m otor be havior c an be sc a le d by an oth er factor β, an d a l l f orces ca n be sca le d by a third facto r γ .This p roced ure generates th e dim ensio nless exp ressio n o f a l l re levant m e chanica l variab les (morp homet rics, kinemat ics, a nd dyna mics).Wh en th e (loco)m otor be havio r fo r two different-sized anim al s i s dyn amica l ly simi lar (i .e., ident ica l behavior in dimensionless me asures), t asks for the muscles (the "mot or s") a re simila r t oo (i .e., wo rk do ne per uni t m a ss and per unit distance).
Re vers ed, howe ver, t he for mer re a soning al so implies tha t, for iden t ica l m otor tas ks, a ny difference in col le ctive o utp ut th at rem a ins a f ter nor ma lizat ion according to dynamic similari ty p rinci ples must reflect intrinsic n eurom e chanica l chan g es.Wh en fram ed in an ontogenetic cont ext, suc h c han g es a re lik ely t o point t o neuromotor ma tura tion.We o p ted fo r sel f-sele cte d vol u ntary spe e d as the relevan t iden t ica l motor t asks dur ing t he ear ly deve lopm ent of SGA-an d AGA-n e onata l piglets (se e a lso, f or insta nce, Vaugha n et al. 2003 ).The collective o utp ut of the in tegra ted n eurom echanical locom ot or syst em (sp at iotempora l gait da ta, join t kinema tics, and single-limb dynamics) is assembled and normalized according to dynamic simil arit y principles to answer the specific r esear ch questions as form ula ted in the Int roduct ion.

Experimental approaches
D et ails on the specific in "Methods" section are presented in our former studies and wi l l on ly be summarize d un les s the res ul ts p r esented her e con cern ed addi tio nal an d unpu blis h ed data.In thi s ca se, a l l r equir ed details will be pro vided.I nsti tu tio nal and n ation al guide lin es for th e ca re a nd use of a nim al s were fol lowe d, and a l l exper iment a l proce dures invo l ving anim al s were approved by the Ethical Co mmi ttee o f Anima l E xper iment atio n, Universi ty o f Antwerp, Belgi um (a pproval n umbers 2015-26 and 2017-25).
Sp at iotempora l gait data ( Vanden Hole et al. 2017 ;2018a ).Fourteen AGA piglets of normal v italit y and 11 SGA p iglets wi th low vi tali ty fro m 7 li tt er s (AGA = m a ss av erag e ± 1 SD per litter; SGA < litter av erag e-1 SD; no rmal vi tali ty = no rmal resp iratio n, standin g/walkin g at birth; low v italit y = l abor ed r espiration, n o m ovem ent/stan ding) are sele cte d for long itudina l fol low up.D at a are col le cte d from latera l v iew v ide o se quences (50 Hz) of voluntary wa l king (s elf-s ele cte d spe e d), re corde d on t he far m at 0, 1, 2, 6, 8, 24, 26, 28, an d 96 h ours pp .For the present s tudy, s tride ( = cy cle) len gth and frequency, step length (displacement during single-limb co ntact), d u ty facto r (stance t ime/st ride t ime), relati ve limb p h a sing (difference in foot fa l l t imings relative to stride tim e), an d hip height at midstance are sele cte d.L ine ar mixed model s (cla s sical s tatis tics) were fitt ed t o evaluat e the effect of age (fixed facto r) o n the ou tco me variables.To account for the dependence betwe en observat io ns o n li tt ermat es o r wi t hin t he s ame a nimal, ra ndo m facto rs were included for sow and piglet (n ested), plus ran dom s lopes for age and piglet (nested).The starting model was simplified using stepwise backward m ode ling.Post ho c an alysi s wi th Dunnett's co rrection wa s u sed for comparing norma lize d ga it va riables of the different age grou ps wi t h t he control age (i .e., 96 hours pp ).
Joint a ngles ( Mielk e et a l.2023 ; addit iona l ana lyses).Fo r 50 AGA p iglets ( ≥ 0.8 kg at b irth) and 8 SGA p iglets ( < 0.8 kg at birth) within the age interval of 1 to 10 hours pp , stride cy cles w ere sele cte d fro m 10 min o f co n tin uous video r ecor ding (at the farm) of voluntary wa l king (s elf-s ele cte d spe e d) of each indiv idu a l (cross-se ct iona l study).Joint angle profiles are measured thro ugho ut the wa l kin g cy cles.Fo r the p resent study, s h oulder, e l bow, carp a l j o int, hi p, kn ee, an d t ars a l j oint angles are sele cte d.Time p ro files are t ra nsf or med to t he frequency domain (Fourier Series Deco mposi tio n, exponent ia l form; eight harmonics retained fo r p ro file descri ptio n; se e a lso Miel ke et a l.2020 ).This proce dure p rovides fo r each jo int th e m ea n a ngle a n d th e amplit ude (general post ural info rmatio n) and 2 × 8 Fourier coefficients (p ro file "shape" info rmatio n).Fo r the six j oints col le ct ive ly, th e coefficients yie ld info rmatio n o n the int ra lim b and interlim b coo rdinatio n.A Princi ple Co mpo nent A n alysi s i s per for me d to re duce t he var iable count from 96 coefficients to 12 PCs.These components are consider ed coor dination variables.Genera lize d linear modeling (Bayesian s tatis tics) is used for the analyses.
Single-limb ground reactio n fo rces (GRFs; unpublis h ed da ta).F our Kist ler ® squir r el for ce plates wer e built in a wa l kway a lo ng, which AGA p iglets and SGA piglets ( ≥0.8 and < 0.8 kg birthweig ht, respec ti vel y) walked at s elf-s ele cte d voluntary spe e d.A nim al s wer e r ecor ded bet ween d ay 1 and d ay 5 pp (cross-se ct iona l setup).Only single-limb GRFs were used for f urt her analysis.Al though reco rded at 1 kHz, force p ro files were resampled to obtain a fixed number of 50 intervals per st ance.St at ist ica l p a ra metric mapp ing (S PM;Pataky, 2012 ;www.spm1d.org ), taking account o f the dependence of samples at neighboring timepoints, was used to co mpare no rmalized fo rce p ro files (exp ressed in Body weights, BW; see f urt h er) between AGA piglets an d SGA piglets (e.g., Pataky, 2012 ;Pataky et al. 2013Pataky et al. , 2015 ; ;Vanrent erghem et al . 2012 ).To incre ase t he number of useful samples, the first and last half of the stance ph a ses were considere d sep arate ly (n ote that spm1dpackag eV4.0 ena bles to con sider the tw o parts of the fo rce p ro files wi thou t neg lec ting the spa tiotem poral depen den ce between n eighboring tim e m easurem ent points and the smoothness of the complete curve).Vert ica l a nd f ore-a ft gr ound r eactio n fo rces o f fo re-and hind-limbs were considered.
F orce-genera ting ca paci ty, fib re type co mposi tio n, g lucose, and g ly cog en lev e ls ( Van den Hole et al. 2018b ;2019a ;2019b ).Extens or mus cles of the front-and hindlimb were disse cte d at one b o dy side from AGA-and SGA-indiv idu als at 0, 4, 8, and 96 hours pp (crossse ct iona l setup, at least 4 indiv idu als per timepoint).Mass and mean fibre length were measured to determin e th e p hysio log ica l cross-se ct iona l area p erp endicular to the muscle fibres, h en ce in depen dent o f co nt ract i le status on th e fibre length (m a ss/densi ty/fib re len gth; den sity = 1060 kg/m 3 cf.Méndez and Keys, 1960 ).These ar eas wer e summ ed per limb an d a muscle s tres s of 0.3MPa (cf .e.g., Wi l liams et a l.2008 ) was use d to est ima te force-genera ting ca pacity.Sam ples for fib re typ in g w ere taken from t he ot her b o dy side .B lo o d glucos e le ve ls an d gly cog en lev e ls in th e liv er w ere det ermined .B lo o d glucose was me asured wit h a Onet ouc h U l tra gl ucose m eter, imm ediate ly after the piglets were human e ly ki l le d usin g an ov erdose of anaesthetics a nd exsa nguin ation.G lycogen concen tra tion s w ere determine d spe ct ro pho tomet rica l ly using the protocol by Theil et al. (2011) on a sample of liver tis s ue.
Bayesia n P rob abi list ic Pre dict ive Modeling ( Mielke et al. 2023 ).D at a col le cte d for the an alysi s o f the jo int angles was used (see a bov e).For each of the wa l king sequences, sp at iotempora l data, ne ck a ngles, linea r measures of the dig it ize d seg m ents, an d sex were added to t he dat abas e. G enera lize d linear models were t raine d on 34 variables (sex, 9 sp at iotempora l, 12 postura l, and 12 coo rdinatio n variables) o f 294 strides by AGA piglets

Basic coordination patterns and maturation rate
The following results build on and r einterpr et part of the lo ngi tudinal spa tiotem po ral gai t data o f Vanden Hole et al. (2017 ;2018a ) in the context of th e fram ewor k of the presen t pa per.
To estimate the time window in which piglets are subj e ct to loco moto r develop ment, we used the sp at iotempo ral gai t c haract eri stics a s col le cte d for the 14 AGAand 11 SGA-piglets ag ed betw een 0 and 96 hours pp (cf."Methods" se ct io n).In absol u t e t er ms (m/s), t h e se lfsele cte d voluntary spe e d increases slightly during early deve lopm ent ( Fig. 2 A).SGA piglets wa l k slightly more slowly t han t he AGA p iglets, bu t fro m day 2 o nward (24 hours p p ), volunta ry wa l king be com es in different fr om the r efer ence at 4 days o f develop ment (96 hours pp ; Fig. 2 A).This spe e d increase is caused by the strideand step len gth s becomin g lar g er with ag e. SGA-piglets use s h orter strides an d steps than AGA piglets, and differences wit h t he r efer ence age r emain unt i l 24 hours, respe ct i vel y, 26 ho urs p p ( Fig. 2 B and C).Stride frequen cy an d d u ty facto r a re simila r f o r both catego ries, and only duty factor differs from th e 96 h ours pp reference at age 0 and 2 hours ( Fig. 2 D and E).R el ative limb ph a sing does not ch an g e with ag e an d is in different to the p iglet catego ries.Absol u te sp at iotempora l st ride c haract eris tics s ugges t t hat, dur ing e ar ly deve lopm ent ( < 96 h ours pp ), locom otor ma tura tion ha ppens fast after birth, essent ia l ly within one to two days.P rior to exa mining which of the ont ogenetic c han g es can be att ribute d to size, it is useful to quantify how size actua l ly differs betwe en p iglet catego ries and ages.Fig. 3 A s h ows a c lear, consist ent difference (of ∼50%) in BM between the SGA-and AGA-p iglets.Mo reover, wi thin each piglet category, BM hardly chan g es ov er the first 28 hours pp ( Fig. 3 A, Vanden Hole et al. 2018a ).Also, when con siderin g a BM-ran g e present in th e tim e window, in cluding th e r efer ence age of 96 hours pp , front an d hin d part m a ss es (s ee "Methods" s e ct io n) co nsti tu te n ear ly fix ed fraction s of BM ( Fig. 3 B; m a sses respe ct i vel y e qua l to 0.41BM 1.02 an d 0.38BM 1.16 ; in th e case of the f ront par ts, th e expon ent is effe ct i vel y indiffer ent fr om 1), and gr owt h patter ns of SGA-and AGApiglets appear very similar ( Fig. 3 C).Based on all this, we presume that the d ata c an be sufficiently well appro ximated b y an isomet ric g rowth m ode l , whic h is an as s umpt ion and re quirem ent of n orma lizat ion according to the dynamic similarity p rinci ple (see co nceptual fram ewor k).Note that, even when applied to anim al s differin g con sidera b l y more in b o dy build and p osture t han t he p resent p iglets (e.g., different taxa, different ontogenetic s ta ges), norma lizat ion according to dynamic simil arit y y ield s very u seful an d re li able results (e.g ., Aerts et al. 2000 ;Alexander, 1977  For the purpose of scaling m otor be havior according dy namic simil arit y, sele ct ion of a mech ani stically relevant dimensional scaling variable is necessary.We o p ted for a dynamic measure for funct iona l limb length d u ri n g wa l king, i .e ., hip height h at midstance (with, in practice, t he dist ance "hoof-t ai lb a se" a s a proxy).Thi s choice i s b ase d on the rat iona le t hat t his me asur e dir e ctly lin ks wit h t h e inverted pen d ul um mech ani sm of the walking gait (cf.Alexander, 1977Alexander, , 2003Alexander, , 2005 ; Alexan der an d Jay es, 1983 ).Scalin g the temporal aspects is f urt her obtained by dividing by h/ g (with g g ravitat iona l acceleratio n; e.g., Ho f, 1996 ), while fo r ces ar e expr essed in BW.
As m ention ed ear lier, con ceptually, a ny difference in the col le ct ive o utp ut of the muscu lo-skeleta l system that rema ins a f ter nor ma lizat ion according to dynamic simil arit y principles s h ou ld refle ct int rinsic n euro-m otor chan g es.How ev er, sin ce th e deve lopm enta l p attern of hip heigh t a ppears a b i t mo re in trica te th an th at of BM (se e Fig. 3 C and comp are with 3A), a cautious interpretatio n o f the observed differences a nd cha n g es in h is ne e de d.
Firstly, as for BM, a rather consistent difference bet ween the t wo piglet categories exists in the considered ear ly deve lopm enta l t im e win dow ( < 28 h ours pp ).Except at birth ( i.e., 0 hour pp ) when hip heights are simila r f o r both p iglet catego ries, dynamic hi p heigh ts a t midsta nce a re consistently ∼30% la r g er in AGA piglets ( C in Fig. 3 C).According to the direct s ke letal m easurements of limb length presented in Vanden Hole et al. (2018b) a nd conf or m to t he h ardly ch an gin g BM, the major part of this C (about 80%) can directly be attribut e t o the lar g er size of th e AGA piglets.Th e rema ining differen ce in hip h eigh t a t midstance (about Se cond ly, and cont rary to BM, there is also a developmental chan g e o f the hi p heigh t h a t midst ance in t he considere d t im e win dow (1 h our-28 h ours pp ; P SGA(t) and P AGA(t) in Fig. 3 C).This chan g e is remarkab l y simila r f or SGA-a nd AGA-piglets, bot h in t h e re lative magnitude and in the shape o f i ts pattern ( Fig. 3 C).Because of th e n ear ly constant BM in this time window, the obs erved increas e in h reflects pr imar ily incre asing limb extension, n ot m orph olog ica l limb g rowth (as a ls o e videnced by the dire ct skeleta l limb length m easurem ents at 2, 4 and 8 hours pp by Vanden Hole et al. 2018b ).For both piglet categor ies, t he hip height h at midstance increases by about 28%, most of which by 8 hour s pp.Aft er day 1 pp, no f urt her chan g e in lim b post ure is observed in the considered time window.
Where as t he first-ment ione d consistent difference in hip height between SGA-and AGA-piglets (i .e., C) does reflect a pure size difference (including the sma l l, but systemat ica l ly more flexe d lim b post ure of SGA), the postural chan g e ( P (t) ) reported a bov e, simila r f or SGA -and AGA -p iglets, does not.As a resul t, hi p height h a t midstance, in tended to be u sed a s a (mech ani stica l ly relevant) norma lizat ion variable to remove size effects according to dynamic simil arit y, is to some extent biased by ma tura t iona l effe cts refle cte d in a g radua l ly exten ding limb.Th er efor e (1), withi n p i gle t c at ego ries , differences (with the r efer ence age of 96 hours pp ) that h ave di s appe a red a f ter nor ma lizat ion cannot be claimed to be ex clusiv e ly th e result of th e rem oval of size-effects a nd differences that remain likely reflect ma tura tion effects other than chan gin g lim b post ure.This holds primari ly t rue for the very first hours o f develop ment ( < 24 h ours pp ), wh en m ost of th ese postural chan g es occur (cf. a bov e).W hat fractio n o f thi s i s size rem oval an d w hat frac tio n is d ue to m a s king th e ma tura t iona l postural chan g e cannot be de duce d fro m the p resent an alysi s.
[Notice t hat t h e m ore flexed lim b post ure in t he e arly s ta ges is probab l y generic in m amm al s (see, for instance, Dewolf et al . 2021 ;Pet er s, 1983 ;S hapiro and Young, 2012 ;Suth er lan d et al. 1980 ;Th e len et al. 1984 ).This ph en om en on (som etim es r eferr ed to as "flexor domin ance") i s att ribute d to a tempora ry, ea rly deficit in exten sor stren gth].
How ev er (2), giv en bot h t h e consisten cy of size differences between SGA-and AGA-piglets and the lar g e resemblance of the pattern of postural chan g es (cf. a bov e), differen ces b etw een p i gle t c at ego ries that have disappea red a fter norma lizat ion were t ru ly on ly there because of the size difference the between SGA-and AGApiglets.
What f ollows a r e the r esul ts o f th e n ormalizat ion.On ly below 2 hours pp , dimensionless voluntary wa l king spe e d ( V/ h g ; wit h V t he absol u te wa l king spe e d) differ fro m that o f the r efer ence age of 96 hours pp ( Fig. 4 A).Moreov er, dimen sionless s elf-s ele cte d wa l king spe e ds do not differ between SGA-and AGA-piglets.Similar ly, dim ensionless stride-and step len gth s ( L/h ; with L the absol u te strideor step length r esp.) ar e indiffer ent to SVA-and AGApig lets, w hi le stat ist ica l differences with the r efer ence age are only a pparen t before 4 hours pp ( Fig. 4 B,  C).Except imm ediate ly pp, dim ensionles s s tride frequency ( F / g /h ; with F the absol u te frequency) is indiffer ent fr om that at 96 hours pp for both SGA-and AGA-p iglets, yet o n av erag e , SGA-piglets seem t o wa l k a t sligh tly low er dimen sion less fre quencies.G iv en the ident ica l dimension less ve locity an d str ide lengt h for bot h categor ies, t h e latter differen ce is probab l y a coin ciden ce of th e lar g e bu t b iolog ica l ly log ica l variabi li ty d ur ing e ar ly deve lopm ent.Duty fact or s and relative ph a ses, dimensionless a s such, h ave a lready be en dea lt with (see a bov e).
R ely ing on the cautious, exten siv e di scu ssio n o f the norma lizat ion vari able prov ided a bov e, th ese dim ensionles s res ults s till leave us t o conc lude t hat t he early n euro-m ot o r d ev el opm ent of AGA-an d SGA-p i glets proce e ds la rge l y identicall y.Mor eover, the neu ro-(l o c o)m ot o r ma tura ti on app ar ently pr oce e ds rather rapidly , i .e ., in < 4 hour s pp

J oint mov ement profiles and inter-and intralimb coordination
The fol lowing resu l ts rely o n the addi tio n al stati stical an alysi s (see "Method s" section) of the joint angle profiles thro ugho u t loco moto r cycles used and di scu ssed in Mielke et al. (2023) for pre dict ive m ode ling purposes.As ar gued a bov e , stride (i .e., cyc le fr equencies) har dly chan g e durin g early dev elop ment o r between p iglet catego ries.Therefo re, fo r the p resent co mpariso n (wi th focus on the time window from 1 to 10 hours pp ), the pr ogr ess of time thro ugho ut a cycle can safely be expressed as (dimensionless) fractio ns o f cycle d uratio n.Joint angles (expressed in radians) are dimensionless as such.A and C) and the hindlimb (B and D) of SGA-and AGA-piglets walking at self-selected voluntary speeds.Color codes are as in Fig. 2 .The number of cycles (# meas.) as obtained from the number of individuals (# ind.) are indicated.Yellow shaded areas r epr esent significant differences between the SGA-and AGA-profiles for the profiles (vertical or fore-aft) marked by the arrow heads.
In addi tio n to sp at iotempora l col le ct iv e varia bles, differences a nd ma tura tio n o f int ra limb coordinat ion co uld ap pe ar at t h e joint leve l. Figure 5 s h ows th e joint angle p ro files fo r the fro nt-an d hin d-limb for AGAand SGA-p iglets wi t hin t h e 1 to 10 h ours pp deve lopmenta l t ime window.Zero-t ime and time "1" acco rd to co nsecu tive t ouc h downs of th e limb.Q ua litati vel y, p ro file shapes are very similar for both piglet categ ories.The av erag e an gle o f the jo in t oscilla tions differ s t o some ext ent for s h oulder, hip, kn ee, an d t ars al j oint angles, refle ct ing the abovem ention ed configurat iona l differences of the funct iona l limb length during ear ly deve lopm ent (see a bov e).How ev er, only the average knee and t ars a l j o int angles o f the cycles o f the SGA piglets differ significantly from those of the AGA p iglets (abou t 11 • and 6 • respe ct i vel y; Fig. 6 ).This links u p wi t h t h e fin dings con cer ning t h e differen ces in dynamic hip height and hind limb posture as dis cuss ed a bov e .S imi larly, on ly for the elbow a nd ca rp a l j oint, ran g es o f motio n seem t o differ (i .e., s ome what lar g er in SGA) between piglet ca tegories ( Fig. 6 ).In terlimb and int ra limb coordinat ion SGA piglets are indifferen t from tha t o f AGA p iglets (cf.coo rdinatio n variables, Fig. 6 ).Based on these results (free of size effects), it seems that also join t kinema tics and coordina tio n o f SGA -and AGA -piglets a re la r g ely ident ica l dur ing e arly deve lopm ent.

Limb dynamics
As m ention ed in "Meth ods" se ct ion, the resu lts presented in this se ct ion were not pu blis h ed bef ore.Va nden Hole et al. (2017Hole et al. ( , 2018a ) ) s h owed that n orma lize d stance d uratio n is ident ica l for SGA-and AGA-piglets and is only different from t he st ance d uratio n at the r efer ence age of 96 hours pp immediately after birth (time "0").Ther efor e, stan ce tim e can be exp ressed fo r a l l cycles as a fraction (with zero-time equal to touch down and time "1" e qua l to lift-o ff o f the limb co nsidered).
Figures 7 A and B present th e n orma lize d sag itta l single-limb GRFs (me dio-latera l forces not considered) fo r the fro nt an d hin dlimb cy cles (cov erin g the entire 96 hours p erio d) of SGA-and AGA-piglets.Only late in stance (from ∼65 to 80% of stance), sma l l differences in the vert ica l co mpo nents o f GRF p ro files o f th e hin dlimb could be det ect ed (ye llow s haded regio n), wi th SGApiglets g eneratin g s ome what lower GRFs.Focussing on the r ecor din gs collected durin g day 1 pp only ( Figs. 7 C, D; remember that ma tura tion most likely proceeds before 4 hours pp , cf. a bov e), th ese differen ces in th e vert ica l co mpo n ents are n o lon g er a pparen t (except for a biolog ica l ly meaningless sma l l fract io n o f t he st ance time; cf.Fig. 7 D), but thi s i s probab l y a n effect of the limited number of avai lable re cor dings.Yet for e-aft fo rces o f the fro ntlimb s h ow in t his e ar ly deve lopm ental p erio d n ow som e sma l l differences betwe en the piglet catego ries d uring a b rief p erio d at the beginning of stan ce (ye llow s hade d reg ion).
In gen eral, m ost of th e dynamic loadin g durin g wa l king is taken by th e fore limbs dur ing e ar ly deve lopm ent, an d m ost differen ces between SGA-and AGApiglets (if present) happen at the level of the hind limbs, g eneratin g s ome what less force (or being unable to provide the force as AGA-piglets) late in st ance.Yet, t hese differ ences r em ain sm a l l .Hence , in general , it can be co ncl uded that (size effects excl uded), d ur ing e arly deve lopm ent, single-limb dyna mics a re la r g ely ident ica l in both piglet categories and that these dynamics don't chan g e with age.

Force-generating capacity and fibre composition of the limb extensors
The fol lowing cross-se ct iona l data a re summa rized from Vanden Hole et al. 2018bHole et al. , 2019a ) ).
The co mpo nent o f the in tegra ted n eurom echanical loco moto r system that fina l ly powers the col le ct ive outpu t, as p res ented and dis cuss ed a bov e (sp at iotemporal gait da ta, join t kinema tics, and limb dynamics), are the muscles.Overa l l, SGA-piglets have a s ome what high er n orma lize d force-generat ing cap acity (physiolog ica l cross-se ct iona l are a * 0.3 MPa/BW) t han AGApiglets in both the frontlimb an d hin d limbs (se e Figs. 8 A an d B).Moreover, in div idu a ls se em to be relat i vel y stron g er at birth than at the r efer ence age of 96 hours pp .Th e fib re type co mposi tio n (fa st-slow) i s ident ica l for SGA-and AGA-piglets and does not chan g e durin g ear ly deve lopm ent (see Fig. 8 C).Apparen tly, a t the m uscula r level (a nd size effe cts exclude d), SGA piglets do e qua l ly wel l (or even slight ly better) t han t heir AGA litt ermat es.

Size and BM: Bayesian probabilistic predictive modeling
Ta king a l l the a bov e into con sidera tion, it a ppe ars t hat, al though SGA-p iglets are, in absol u te terms, defini tely sma l ler than the AGA-piglets, they mature neuromechanica l ly spea king just li k e a nd e qua l ly fa st a s their AGA-litt ermat es.This conc lusion seems to be confirmed by an alterna tive a pp roach (Bayesian p robab ilistic m ode ling) in which m a ss o r size are p re dicte d from (norma lize d) locomotor kinematics ( i.e. , from col le ctive o utp ut of the entire int egrat ed n eurom echanical loco moto r system; cf .a bov e).As m ention ed in the "Methods" se ct ion, th ese m ode ls a re tra ined a nd va lidate d on data (kin ematics an d coo rdinatio n) fro m stride cy cles ex e cute d by AGA piglets in the 0 to 10 hours pp p erio d (cf.Mielke et al. 2023 ).Next, the m a ss or size of i magi nary piglets are pre dicte d u sing kinem atics and coo rdinatio n ext racte d from SGA-st ride cycles (exe cute d 0-10 h ours pp ).Th ese pre dict ions are then comp are d wit h t he m a sses an d sizes of th e real SGApiglets that exe cute d the st rides that served as m ode l in put.Systema t ica l ly, th e m ode l predicts th e im agin ary piglets as h eavier an d lar g er than the real SGA piglets (i .e., im agin ary piglets are "AGA-li ke"; se e Figs. 9 A, B).In other words, the co mpu tatio n al cla ssifier does not find anything spe cia l about t he SGA-dat a, and classifies them "a s norm al ," whic h can be int erpret ed as evidence for neurome chanica l simi larity to matching AGA.

Stagnation in neuro-motor performance due to energy-deficits
Alt hough t he a bov e findin gs seem to s ugges t that SGApiglets are just only smaller t han t heir AGA-litter mates, it rem ains th at they a re f ound to be less mobile, less vita l, less compet it ive, more vu ln erable, an d even often die before day 3 pp ( F eldpa usch et al. 2019 ;Quiniou et al. 2002 ).This mo rb idi ty (an d m o rtali ty) is p robab l y refle cte d in the ou tco me o f an oth er Bayesian predictive m ode l, similar ly train ed on AGA-kin ematics and coo rdinatio n, p re dict in g ag e pp (in stead of size or m a ss) of i magi nary SGA-piglets from stride cycle da ta of r eal SGA-indiv idu a ls (exe cute d 0-10 hours pp ).Fo r cycles fro m SGA-p iglets older t han t h e 4 h ours p p bounda ry (i .e., the ag e a bov e which no f urt her "size-free" ma tura tion differences could be det ect ed between SGA-an d AGA-piglets), th e age is underest imate d (with 2 to 5 hours pp ) in 75% of a l l cases (see Fig. 9 C; cf.Mielke et al. 2023 ).In other words, m ost of th e str ides of t he older (re al) SGA-indiv idu als (5 to 10 hours pp ) seem to have ret ained t h e n eurom otor leve l oth erwi se ch aracteri stic for the y oun g er AGA-piglets.
This s ta g nat ion in n euro-m otor per for ma nce ca n probab l y be explained in terms of energy deficits.lar g er size and higher vi tali ty give the AGA-p iglets mo re co mpeti tive access to the sow's teats to fe e d.More over, gly cog en reserv es (in the liv er) at b irth are co nsiderab l y higher in AGA-than in SGA-piglets ( Fig. 10 B; in m a ss %), and these reserves are mobi lize d from birth onward in the AGA-piglets only ( Fig. 10 B; purple versus g re en arrows).App arent ly, dur ing t he first hours of deve lopm ent, it becom es in creasing ly diffic ul t fo r the SGA-piglets to keep up with their AGA-litt ermat es because of energy-rather than n eurom e chanica l deficits.

General conclusions
The in tegra tio n o f a l l resu l ts p resented in our fo rmer studies on the locomotion of term AGA-and SGApiglets ( Vanden Hole et al. 2017, 2018a, 2018b, 2019a, 2019b ; Miel ke et a l.2023 ) wit h t h e n ew data on b o dy p ropo rtio ns, jo in t kinema tics, and single-limb dynamics as given in the present synthesi s, en ables u s to conclude the following: -Observed n euro-(loco)m otor differen ces between AGA -and SGA -p iglets d ur ing e ar ly deve lopm ent se em to boi l down to a n effect of size .Hence , size does matter!-Observed differences in vi tali ty an d m ob ili ty between AGA-and SGA-p iglets d ur ing e ar ly deve lopm ent are probab l y not caused by differences in n eurom echanical ma tura tio n bu t resul t fro m differences in the gestat iona l-and ear ly en ergy uptake and metabolism.-W hen SGA-p iglets are u sed a s a (loco)m otor deve lopm ental m ode l f or huma n IUGR infa nts, the very s h ort tim e win dow for n eurom e chanica l ma tura tion ( < 4h), at least in loco motio n, should be considered.-St i l l, so me cau tio n may be ne e de d when interpreting the above resu lts.On ly SGA piglets that did wa l k fro m b irth o nward (yet, classified as "SGA/low vitality;" cf."Methods" se ct ion a nd Va nden Hole et al. 2017 and 2018a fo r cri teria) were used.Although inevitable in a study on the early development of loco motio n, this could h ave bia sed th e con c lusions t o some extent.

Fig. 1
Fig.1Conceptual framework.Control components of the locomotor system interact with internal and external mechanics (symbolized by the spring-dampers) and inertia (symbolized by the weights) thus providing the collective output from the integrated neuromechanical system (i.e., spatiotemporal gait data [r epr esented by a theoretical footfall pattern], joint kinematics, and limb dynamics).See text for more explanations.

Fig. 2
Fig.2Spatiotemporal gait data (box and whiskers plots) for SGA-(purple) and AGA-(green) piglets when walking at self-selected voluntary speed during early development.When piglet categories differ, this is indicated above the graphs (i.e., in case of speed, stride length, and step length).Asterisks refer to significant differences with the variable considered as measured at the reference age of 96 hours pp.

Fig. 3
Fig. 3 (A) BM as a function of pp age.Color codes and indication of significant differences are as in Fig. 2 .(B) Masses of the front (circles; dashed curve; 0.41BM 1.02 ) and hind (squares; solid curve; 0.38BM 1.16 ) parts of the piglets as a function of BM (see text for more explanation).(C) Functional hip height as a function of pp age .C (double headed ar ro w) ref ers to the inter piglet categ ory difference, P (double headed arrows) to the developmental time dependent postural change in AGA and SGA (see text for more explanations).Color codes in panels A-C and indication of significant differences in panels A and C are as in Fig. 2 .

Fig. 4
Fig. 4 Dimensionless spatiotemporal gait data (normalized according to dynamic similarity; see text) for SGA-and AGA-piglets when walking at self-selected voluntary speed during early development.Color codes and indications of significant differences are as in Fig. 2 .

Fig. 5
Fig. 5 limb joint angle kinematics (in radians; π = completely extended joint, 0 = completely flexed joint) of SGA-and AGA-piglets (measured as show in panel G; adapted from Mielke et al. 2023 ) when walking at self-selected voluntary speed.Relative 0 and 1 ( x -axis) refer to consecutive touchdowns of the limb considered (frontlimb = panels A-C; hindlimb = panels D-F).Color codes are as in Fig. 2 .Dashed straight horizontal lines indicate the temporal average of the joint angle profiles (AGA: green, SGA: purple).The dashed curve is the average SGA-joint profile ( + range in purple), the solid curve the average AGA-joint profile ( = range in green).

Fig. 7
Fig.7Single-limb vertical, and fore-aft ground reaction forces (averages + ranges; expressed in BWs as a function of relative stance time) of the frontlimb (A and C) and the hindlimb (B and D) of SGA-and AGA-piglets walking at self-selected voluntary speeds.Color codes are as in Fig.2.The number of cycles (# meas.) as obtained from the number of individuals (# ind.) are indicated.Yellow shaded areas r epr esent significant differences between the SGA-and AGA-profiles for the profiles (vertical or fore-aft) marked by the arrow heads.

Fig. 8
Fig. 8 Force-generating capacity (in BW-s) for front (A) and hind (B) limbs of SGA-and AGA-piglets as a function of pp age.(C) r epr esents the fibre composition (% fast fibres) for the front limb (hind limb %-s are identical for ages and piglet categories).Significant differences between piglet categories are shown above the gra phs.Indiff er ences ar e indicated with the letter-labelling (same letter = indifferent).

Fig. 9
Fig. 9 Differences between the real mass (A; kg), linear size (B; arbitrary measures), and age (C; hours pp) between values as predicted by Bay esian pr edictiv e modeling and the actual masses, sizes and ages of the SGA-and AGA-piglets.Color codes as in insets.See text for more explanation.

Fig. 10
Fig. 10 Blood glucose levels (A) and liver glycogen (B) in SGA-and AGA-piglets during early development.Color codes are as in Fig. 2 .Identical letter-labels show indifferences within the piglet categories.Arrows indicate qualitatively the major differences in the developmental changes of the concentrations (no change = horizontal) between SGA-and AGA-piglets.See text for more explanations.
Fig. 10 is b ase d on the results presented by Vanden Hole et al. (2019b) .At birth, blo o d glucose levels (in mg/dl) are ident ica l for SGA-and AGA-piglets ( Fig. 10 A).This picture, how ev er, chan g es v ery rapidly.At 4 hours pp (the ma tura tion b oundary cf.ab ove), this level h a s more than doubled in AGA-p iglets, bu t not in the SGA-piglets where blo o d gl ucose remained at b irth level ( Fig. 10 A; g re en versus purple arrow).The explanation for this diff erence is twof old.The uptak e of colost rum imme diat ely aft er b irth is very impo rtant in early development ( Dec lerc k et al .2016 ; Deviller s et al .2011 ), an d th eir