The paradox of spring: Thyroid and glucocorticoid responses to cold temperatures and food availability in free living Carneddau ponies

Ho mones and Beha io 161 (2024) 105526 A ailable online 18 Ma ch 2024 0018-506X/© 2024 The Au ho s. Published by Else ie Inc. This is an open access a icle unde he CC BY license (h p://c ea i ecommons.o g/licenses/by/4.0/). The pa adox o sp ing: Thy oid and glucoco icoid esponses o cold empe a u es and ood a ailabili y in ee li ing Ca neddau ponies Jessica G anweile a , * , Ju gi C is ´ obal-Azka a e b , Na han Mo on a , Rupe Palme c , Susanne Shul z a a School o Ea h and En i onmen al Sciences, The Uni e si y o Manches e , Manches e , UK b Depa men o Basic Psychological P ocesses and hei De elopmen , Facul y o Psychology, Uni e si y o he Basque Coun y, Donos ia, Spain c Uni o Physiology, Pa hophysiology and Expe imen al Endoc inology, Depa men o Biomedical Sciences, Uni e si y o Ve e ina y Medicine, Vienna, Aus ia ARTICLE INFO Keywo ds: Co isol Faecal bioma ke Ho ses The mo egula ion Die shi ABSTRACT In seasonal en i onmen s, main aining a cons an body empe a u e poses challenges o endo he ms. Cold win e s a high la i udes, wi h limi ed ood a ailabili y, c ea e opposing demands on me abolism: up egula ion p ese es body empe a u e bu deple es ene gy ese es. Examining endoc ine p o iles, such as hy oid ho mone iiodo hy onine (T3) and glucoco icoids (GCs), p oxies o changes in me abolic a e and acu e s esso s, o e insigh s in o physiological ade-o s. We e alua ed how en i onmen al condi ions and ges a ion impac on aecal ho mone me aboli es ( T3Ms and GCMs) om la e win e o sp ing in a ee-li ing popula ion o Ca - neddau ponies. Faecal T3Ms we e highes in la e Feb ua y and Ma ch, when empe a u es we e lowes . Then, T3Ms concen a ions dec eased h oughou Ap il and we e a he lowes in May be o e inc easing owa ds he end o he s udy. The decline in T3M le els in Ap il and May was associa ed wi h wa me wea he bu poo ood a ailabili y, die di e si y and die composi ion. On he o he hand, GCM le els did no display a clea empo al pa e n bu we e associa ed wi h ep oduc i e s a us, whe e p egnan and lac a ing emales had highe GCM le els as compa ed o adul males and non- ep oduc i e emales. The empo al p o ile o T3Ms le els highligh s me abolic ade-o s in a changing en i onmen . In con as , he epheme al bu synch onous inc ease in GCM concen a ions ac oss he popula ion sugges a sha ed expe ience o acu e s esso s (i.e., wea he , dis u bance o social). This mul i-bioma ke app oach can e alua e he ole o acu e s esso s e sus ene gy budge s in he con ex o in e en ions, ep oduc ion, seasonali y and en i onmen al change, o ac oss mul iple scales om indi iduals o popula ions. 1. In oduc ion Animal popula ions esiding a high la i udes endu e ex eme sea- sonali y. Cold win e s a e challenging o endo he ms as inc easing ene gy is needed o gene a e hea o main aining body empe a u e a he same ime as esou ce a ailabili y declines (Mo ison e al., 2008). This leads o an ene gy con lic be ween me abolic needs and low ood a ailabili y and quali y, as indi iduals canno a o d o con inuously bu n h ough ene gy ese es h oughou win e wi hou being able o eplenish hem (C uz-Ne o and Bozino ic, 2004). Endo he ms mus keep hei co e body empe a u e wi hin speci ic ange and can do so wi hin hei he moneu al zone a low ene ge ic cos s. When ambien em- pe a u es go below he he moneu al zone, e e ed as he lowe c i ical empe a u e, he body mus inc ease me abolic a e o main ain co e body empe a u e and allow no mal body unc ion (Mejdell e al., 2020; Mo ison e al., 2008). Fo he bi o es a high la i udes, he a ailabili y and quali y o ood dec eases h oughou win e , wi h ea ly sp ing (i.e., be o e ege a ion g ow h begins) being he mos challenging pe iod o animals o main ain a posi i e ene gy balance (Kun z e al., 2006; Owen- Smi h, 2008). Low ood a ailabili y s ops he bi o es om being able o up egula e hei me abolic a e, as equi ed o he mo egula ion in win e and ea ly sp ing. A nega i e ene gy budge h ough he win e and in o ea ly sp ing can lead o a decline in a s o es and body con- di ion (Albon e al., 2017). Endo he ms ha e de eloped se e al s a egies o cope wi h he sea- sonal ene gy challenge associa ed wi h high la i ude win e s. Some species mig a e ei he o lowe ele a ion o lowe la i udes o a oid he ha sh condi ions (Hsiung e al., 2018). O he species en e a * Co esponding au ho a : School o Ea h and En i onmen al Sciences, The Uni e si y o Manches e , Manches e , UK. E-mail add ess: [email p o ec ed] (J. G anweile ). Con en s lis s a ailable a ScienceDi ec Ho mones and Beha io jou nal homepage: www.else ie .com/loca e/yhbeh h ps://doi.o g/10.1016/j.yhbeh.2024.105526 Recei ed 2 Augus 2023; Recei ed in e ised o m 20 Feb ua y 2024; Accep ed 29 Feb ua y 2024 Ho mones and Beha io 161 (2024) 105526 2 hypome abolic s a e o o po o hibe na ion/es i a ion o lowe basal me abolism and educe ene gy use (Heldmaie e al., 2004). Howe e , non-mig a o y, non-hibe na ing species who canno escape his ene gy challenge employ ene gy conse a ion s a egies. These include educing expensi e ac i i y such as locomo ion, lowe ing hea a e and basal me abolic a e (A nold, 2020). Changes in ac i i y budge and hea a e ha e been documen ed ac oss a ange o non-mig a o y, non- hibe na ing la ge he bi o es, p o iding suppo o hypome abolism as an adap i e me abolic esponse o win e condi ions (A nold, 2020). These s udies include: ed dee Ce us elaphus (Tu bill e al., 2011) S alba d eindee Rangi e a andus pla y hynchus (T ond ud e al., 2021) alpine ibex Cap a ibex ibex (Signe e al., 2011); moose Alces alces (G æsli e al., 2020); muskoxen O ibos moscha us (Schmid e al., 2020) llamas Lama glama (Riek e al., 2017) P zewalski ho ses Equus e us p zewalskii and She land ponies Equus e us caballus (B inkmann e al., 2014, 2012). In some o hese s udies, hea a e, body empe a u e and ac i i y le els we e on a e age 50 % less in win e han in summe . In addi ion o hea a e and ac i i y, endoc ine p o iles can p o ide mo e di ec measu es o seasonal changes in me abolic a e in esponse o empe a u e and ene gy a ailabili y. Two endoc ine ma ke s ha can e alua e me abolic a e (e.g., nu i ional o he mal s ess) and en i onmen al s esso s (e.g., social s ess, p eda ion isk o nu i ional s ess) a e iiodo hy onine (T3) and glucoco icoids (GCs), espec i ely (Shul z e al., 2021; Ta low and Blums ein, 2007; Wasse e al., 2010). In addi ion o measu ing ci cu- la ing le els o he ac i e compounds in blood, hey can also be measu ed non-in asi ely in aeces o u ine. Some ho mones a e p i- ma ily exc e ed in hei na i e o m (e.g., T3) and o he ho mones a e hea ily me abolised (e.g., s e oid ho mones including GCs); he e we e e collec i ely o aecally elimina ed compounds as aecal T3 me- aboli es ( T3Ms) and aecal GC me aboli es ( GCMs). Se um GCs and GCMs ha e been employed as bioma ke s o gene al physiological ‘s esso s’. Ele a ed GC and GCM concen a ions ha e been linked o inc eased p eda ion (Boons a e al., 1998; Clinchy e al., 2004), social ins abili y (Edwa ds e al., 2013; Nu˜ nez e al., 2014; Sapolsky, 1992; Van Me e e al., 2009), human dis u bance (Dan ze e al., 2014) o low ood a ailabili y (Foe s e and Mon o , 2010; P ide, 2005). Howe e , a sugges ed link be ween ele a ed GCs, GCMs and ood limi a ion is physiologically no s aigh o wa d. GCs a e in ol ed wi h he mobiliza ion o ene gy ese es in h ee phases: i s by s im- ula ing glucogenesis and inhibi ing glucose up ake by pe iphe al issue ( o a e iew see (Kuo e al., 2015)), second by me abolising a y acids when glucose ese es a e deple ed, and inally by b eaking down p o- eins as he las eso (Wing ield and Rome o, 2015). Ele a ed GCs in- c ease ene gy a ailabili y and a e an adap i e esponse, which can become maladap i e when ene gy ese es a e al eady low o when GC leads o he b eakdown o p o eins (de B uijn and Rome o, 2018). Mo eo e , acu e GC and GCM signals du ing a pe iod o ood es ic ion may indica e an acu e s esso such as social con lic o p eda ion isk a he han a s ess esponse o calo ic es ic ion, o simply be an adap i e esponse. Thus, as GCs and GCMs a e associa ed wi h acu e mobilisa ion o ene gy, hey ha e been used as ma ke s o nu i ional s ess (Foe s e and Mon o , 2010; P ide, 2005), despi e no clea physiological pa hways and no in o ma ion on wha physiological esponse is used by an indi idual o cope wi h nu i ional s ess (i.e., down o up egula ing me abolic a e). T iiodo hy onine (T3) is a mo e di ec bioma ke o ene gy use and oxygen consump ion as i egula es me abolic a e (Beh inge e al., 2018; C is ´ obal-Azka a e e al., 2016; Sil a, 2006). Speci ically, T3 is a key egula o o obliga o y he mogenesis and ene gy sa ing s a egies, wi h high le els linked o inc easing hea p oduc ion (MURAMATSU e al., 1986; O’Malley e al., 1984), and low le els associa ed wi h low ood a ailabili y (Bahnak e al., 1981a; Jeannia d du Do e al., 2009; Bahnak e al., 1981a; Jeannia d du Do e al., 2009; Bahnak e al., 1981b; Jeannia d du Do e al., 2009). The e o e, dec eases in se um T3 and/o T3M le els can indica e pe iods o me abolic o nu i ional s ess and o e a mechanis ic link be ween esou ce a ailabili y and physiological s a e (e.g., hypome abolism). Thy oid ho mones and me- aboli es ha e been used o iden i y he mal s ess (Chen e al., 2021; Hunninck e al., 2020), nu i ional s ess (Dias e al., 2017; Jeannia d du Do e al., 2009) and a combina ion o he wo (C is ´ obal-Azka a e e al., 2016). The e o e, T3Ms may enable us o unde s and wha igge s me abolic esponses such as hypome abolism, which may inc ease su - i al o non-hibe na ing species by allowing indi iduals o conse e ene gy, bu is linked o se ious cos s such as lowe body condi ion (A nold e al., 2006) and po en ial dea h (Schmid e al., 2020). Al hough hypome abolism may be an adap i e esponse o educe ene gy consump ion o e win e mon hs, he ene ge ic equi emen s o p egnan emales can p o ide an opposing p essu e du ing his ime. P egnancy and lac a ion impose high ene ge ic cos s, especially in un- gula es which ca y la ge expensi e young (Owen-Smi h and Ogu u, 2013). In dee and P zewalski’s ho ses, p egnan emales g aze mo e o en, ha e ele a ed me abolic and hea a es (Boyd, 1988; Pekins e al., 1998; Pohlin e al., 2017), and ges a ing howle monkeys ha e ele a ed T3M concen a ions (Dias e al., 2017). These s udies sugges ha p egnan emales ha e o up egula e hei me abolic a e o cope wi h he high ene ge ic demands o ges a ion, which con lic s wi h being in a hypome abolic s a e o conse e ene gy ese es h oughou win e . Fo example, in emale muskoxen hypome abolism was associa ed wi h lowe ep oduc i e ou pu and g ow h, and inc eased abo ion a es (Des o ges e al., 2021). This con lic be ween hypome abolism o win e su i al and hype me abolism imposed by ges a ion could lead o inc eased allos a ic load in p egnan and lac a ing emales, co e- sponding wi h highe GCMs le els seen in many ges a ing mammal species (Edwa ds and Boons a, 2018). Thus, combining he quan i ica ion o GC and T3, o hei me abo- li es (GCMs, T3Ms), can o e an insigh in o how he body egula es ene gy consump ion and ene gy a ailabili y in esponse o en i on- men al challenges, and he ole o physiological pa hways in adap i e esponse o challenges (Cos a-e-Sousa and Hollenbe g, 2012; Sapolsky, 2002; Shul z e al., 2021). A ew s udies suppo T3 and GC in playing an impo an ole egula ing me abolic ade-o s in la ge ungula e species ha unde go hypome abolism in win e (Ho man and Robinson, 1966) ound conside able loss o body weigh , accompanied by his ological e idence o educed hy oidal and ad enal ac i i y du ing la e win e in whi e ailed dee . The au ho s associa ed his o educed a s o es as na u al ood supplies become less a ailable o a e o poo e quali y, and ene ge ic equi emen s o ood inc ease du ing his pe iod. This pa e n in body weigh and hy oidal and ad enal ac i i y is e e ed in Ma ch- Ap il. In line wi h hese esul s, (Bahnak e al., 1981a; Bahnak e al., 1981a; Bahnak e al., 1981b; Ham and Bubenik, 1990) ound ha se um T3 le els dec eased in all, s ayed low du ing win e and inc eased in sp ing, and ha his esponse was media ed by nu i ional s a us. Speci ically, he a i icially eed whi e dee showed highe T3 alues, while ex emely low T3 le els we e ound in malnou ished in- di iduals a he end o win e (Bahnak e al., 1981b). B inkmann e al. (2016) s udied She land ponies kep in semi-ex ensi e condi ions and ound ha compa ed wi h summe , ield me abolic a e and hea a e dec eased subs an ially in win e , and ha his was associa ed o a dec ease in se um T3 le els. Finally, (Hube e al., 2003) epo GCM le els in ed dee o peak du ing Decembe and Janua y o dec ease sha ply in Feb ua y and Ma ch and inc ease again in Ap il. Rela ed o his, he au ho s ound a signi ican nega i e ela ionship be ween GCM le els and minimum ambien empe a u e, he ela ion being bes i ed wi h a non-linea eg ession model. O e all, hese s udies sugges ha by educing me abolic ac i i y ia lowe le els o hy oid ho mones he bi o es e ec i ely educe he need o mobilize ene gy ese es in win e , which should be e lec ed in low GC and GCM le els. Equids a e ound a highe la i udes (Cao e al., 2023), and eadily adap o di e ences in esou ce a ailabili y (Ge sick and Rubens ein, 2017), making hem an ideal s udy g oup o in es iga e physiological ade-o s in esponse o low ood a ailabili y combined o cold clima e. J. G anweile e al. Ho mones and Beha io 161 (2024) 105526 3 The Ca neddau pony is a unique b eed o ho se (Equus caballus) ha is ound in he Ca neddau moun ain ange in Snowdonia, Wales (Win on e al., 2013). They a e ee-li ing, unp o isioned and p eda o - ee popula ion, meaning ha he only a iabili y in hei en i onmen is na u al seasonali y in ood a ailabili y and empe a u es. O en con- on ed o empe a u es ou side hei he moneu al zones (5-25 ◦C – (Mo gan, 1998)) hese ponies exhibi isible signs o win e adap a ion, including a hick win e u coa ha is shed a he end o sp ing, shel- e ing beha iou (i.e., agains a ock o each o he ), sunba hing, and mos bi hs occu ing h oughou sp ing. Addi ionally, he ege a ion ac oss Snowdonia Na ional Pa k is mos ly sh ubs and g asslands, appea ing as a uni o m ood esou ces. The nex esea ch s ep is o quan i y he physiological adap a ions and die shi s o his pony pop- ula ion o su i e he win e . He e we e alua ed changes in T3Ms and GCMs le els measu ed in six g oups o ee-li ing Ca neddau ponies in he Snowdonia Na ional Pa k, Wales, o iden i y hei me abolic esponses o he dual challenges o cold empe a u e and esou ce limi a ion du ing la e win e and ea ly sp ing. We analysed how he le els o hese ho mones change in esponse o: a) die composi ion de e mined ia aecal DNA me a- ba coding, b) abo e g ound ood a ailabili y assessed ia NDVI (No - malised Di e ence Vege a ion Index), c) ambien empe a u e and d) ep oduc i e s a us ( ep oduc i e o non- ep oduc i e ac i e emales o s allions). We p edic ed ha he ponies will inc ease hy oid sec e ion (inc ease me abolism) in esponse o low ambien empe a u e, allowing he ponies o main ain co e body empe a u e as long as body condi ion and ene gy ese es allow. Op imal o aging heo y p edic s ha ponies should expand hei die s o include lowe quali y allback oods du ing pe iods o ood sca ci y (Schoene , 1971) as has been show in moose (Jesme e al., 2020). Thus, a he end o win e and in ea ly sp ing, when ood a ailabili y is lowes , we p edic ed ha indi iduals will di e si y hei die wi h a educ ion in he ela i e abundance o g ass consumed. Low ood a ailabili y and die di e si ica ion should be associa ed wi h dec eases in T3M le els as indi iduals down- egula e me abolic a e (hypome abolism) o minimise ene gy use. In u n, inc eased ood a ailabili y in sp ing should lead o T3M le els being posi i ely associa ed wi h NDVI as a p oxy o ege a ion biomass (Bo owik e al., 2013). On he o he hand, we p edic ed ha GCM le els will be highe du ing pa icula ly challenging pe iods, such as cold spells o du ing low ood a ailabili y. Las ly, we also p edic ed highe me abolic a es and allos a ic load, and hus g ea e T3M and GCM concen a ions, du ing la e ges a ion and lac a ion (Boyd, 1988; Pekins e al., 1998; Pohlin e al., 2017). 2. Me hods 2.1. S udy popula ion The ee-li ing Ca neddau pony popula ion esides in a 51.8 km 2 enclosed a ea in he Ca neddau moun ains, Snowdonia Na ional Pa k, Wales. The ege a ion in he a ea is cha ac e ised by acid g asslands, hea h and bog, including a ic-alpine plan species (Ra cli e, 1959). The popula ion is ee anging bu managed by local a me communi ies and a e accus omed o he p oximi y o humans. The managemen o he ponies does no include ood p o isioning, medical ca e o popula ion con ol (e.g., culling, cas a ion o ansloca ion), bu hey a e ounded up yea ly o assess popula ion size and heal h, wi h he la es es ima e o abou 300 indi iduals (S anley e al., 2018). Pony g oups o m social uni s, gene ally composed o one dominan s allion and associa ed ma es, subadul s and oals (Linkla e e al., 1999). The six ocal g oups a ied in g oup size (9 ±3.94 s.d.) indi iduals (min:5-max:18), and home ange p ope ies (i.e., ele a ion, home ange size, opog aphy he e ogenei y - see supplemen a y able 1 o in o ma ion ega ding composi ion and home ange cha ac e is ics o he s udy g oups). The ep oduc i e s a us o emale ponies was e alua ed based on obse a ion and oaling da es, wi h emales placed in o one o h ee ep oduc i e ca ego ies: p egnan (de ined as s a ing om he day o es ima ed concep ion o he day o pa u i ion), lac a ing, o non- ep oduc i e (nei he p egnan no lac a ing). Visual inspec ion o he T3M and log GCM concen a ions p o iles showed li le di e ence be ween p egnan and lac a ing emales, so we g oup hem in o one ca ego y: ep oduc i e emales. Da a was collec ed non-in asi ely om he Ca neddau pony popu- la ion, and obse a ions we e conduc ed by minimising app oach and dis ess o indi iduals. This s udy complies he U.K. Animals (Scien i ic P ocedu es) Ac , 1986 and was e hically app o ed by he Uni e si y o Manches e (pe mi D060). 2.2. Faecal sample collec ion Co id a el es ic ions limi ed he sampling pe iod o be ween 18 h o Ma ch and 1s o June 2021. All aecal samples we e collec ed du ing his ime pe iod. All g oups we e ollowed as andomly as possible o ensu e equal and andom sampling. Each g oup was mainly isi ed weekly o a leas once e e y wo weeks, wi h a leas ou days in be- ween sampling da es o he same g oup. This esul ed in six isi s pe g oup, wi h excep ion o one g oup (Ski les) which was isi ed o a o al o eigh isi s. Indi iduals we e obse ed on oo be ween 10 am and 5 pm, a 5-30 m dis ance, and aecal samples om adul indi iduals we e collec ed immedia ely a e de eca ion o ensu e indi idual iden- i ica ion and a oid de e io a ion o samples. A e collec ion o each aecal sample, he da e, ime, g oup and indi idual name. To ensu e homogenisa ion o he sample, 4-5 aecal boluses we e collec ed ac oss he aecal piles and hen mixed in he collec ion bag. The sample bag was hen s o ed in a coole wi h ice packs o he emaining o he ield day and hen ans e ed o a -20 ◦C eeze un il analysis. 2.3. Faecal T3Ms alida ion and analysis Be o e his s udy, he use o T3 me aboli es measu ed in aeces had no been alida ed o he assessmen o changes in me abolic physi- ology o ho ses. Valida ions a e necessa y o ensu e ha he ho mone o in e es (T3Ms) is eliably measu ed in aeces (Touma and Palme, 2005; Wasse e al., 2010). We conduc ed bo h an analy ical and biological alida ion using aecal samples collec ed om wo leisu e ho ses and one leisu e pony ha had been placed on a calo ie es ic ed die o manage weigh and p e-lamini is symp oms. A pa allelism es was un o analy ically alida e he p o ocol, con i m ha he an ibody in he assay binds e ec i ely o T3Ms in aeces and ha he e was minimal ma ix in e e ences. A pool sample was c ea ed om six Ca neddau pony samples ( andomly selec ed ac oss he s udy pe iod and con aining h ee adul s emales and h ee s allions) and hen se ial dilu ed om x2 concen a ion o a 1:16. The slope o he concen a ion s logi 10 binding o he se ial dilu ed pooled samples was compa ed o he slope o he s anda d cu e (Gesquie e e al., 2018). Once pa allelism o bo h lines was isually con i med, he op imal binding concen a ion (50 % binding is he op imal binding poin o assay accu acy (Weiss e al., 2009)) was de e mined as 2×concen a ion and selec ed o all u u e analyses. The biological alida ion was based on he well-es ablished nega i e e ec o die a y es ic ion on T3 sec e ion (Keech e al., 2010; Wasse e al., 2010). The wo leisu e ho ses we e placed on es ic ed g azing paddocks and he leisu e pony was placed on a empo a y die o soaked hay wi h limi ed access o g ass. The eeding es ic ion expe ienced by hese wo leisu e ho ses and one leisu e pony we e implemen ed by hei owne s, and aecal sampling was pe o med oppo unis ically. The samples we e collec ed o e a mon h (May-June 2021), s a ing a one week p e ious, o 3 weeks pos die /enclosu e change (see supplemen- a y esul s and Fig. 1). The change be ween p e/pos die es ic ion T3M le els o samples collec ed om he h ee leisu e animals was analysed using mixed e ec model wi h indi idual as a andom e ec . Faecal T3M le els signi ican ly dec eased a e die es ic ion (linea J. G anweile e al. Ho mones and Beha io 161 (2024) 105526 4 model (see supplemen a y); es ima e =-0.017 ±0.005, 16.308 =-3.330, p =0.004) alida ing ha ou p o ocol o quan i y T3Ms can be used o eliably assess hy oidal unc ion in ho ses. Faecal samples we e p ocessed ollowing a modi ied me hod desc ibed in (Wasse e al., 2010). B ie ly, i s 0.5 g aecal sample was mixed wi h 5 mL o 70 % e hanol, hand- o exed (2 min) and cen i- uged (2500 g o 15 min). Second, 1 mL o he supe na an was aken, d ied down in a Gene ac™ miVac Cen i ugal Concen a o and inally, econs i u ed in 500 μ L o assay bu e . T3M assays we e conduc ed a he Uni e si y o Manches e ’s labo a o y using a comme cial De ec X® T iiodo hy onine (T3) Enzyme Immunoassay ki , and p ocedu e was ollowed acco ding o he manu ac u e ’s ins uc ions. The pla e shaken o i e seconds p io o eading and abso bance ead a 450 nm using a The mo Scien i ic™ Mul iskan™ FC Mic opla e Pho ome e . The op ical densi y measu emen s we e hen uploaded o MyAssays (My Assays L d., 2022), which p oduced he s anda d cu e, he samples concen a ions, and he coe icien o a iabili y (CV) be ween duplica e wells. The in a-assay CV o he T3M ELISA was 8.4 % and he in e -assay CV was 8.6 % (n =2 con ol samples). 2.4. Faecal GCMs analysis To quan i y GCM le els in aeces o ponies a p e iously physiologi- cally and biologically alida ed me hod was used (Hinchcli e e al., 2021; Me l e al., 2000; M¨ os l e al., 1999; Palme, 2019). The ex ac ion p ocess was based on (Me l e al., 2000)’s s udy and is esumed in he ollowing s eps: i s 5 mL o 80 % me hanol was added o 0.5 g o aecal ma e , hand- o exed (2 min) and cen i uged (2500g o 15 min); second, 1 mL o he supe na an was ans e ed and combined wi h 5 mL o die hyle he and 0.25 mL o 5 % NaCO 3 , hand o exed and cen i uged (15 min); hi d, he supe na an was d ied down in a Gen- e ac™ miVac Cen i ugal Concen a o ; and inally edissol ed in 500 μ L o assay bu e . The second s ep o he ex ac ion abo e is necessa y due o low concen a ion o GCMs being na u ally ound in equids aecal ma e (Me l e al., 2000). The GCM analysis was pe o med wi h an 11-oxoae iocholanolone enzyme immunoassay (EIA) p e iously desc ibed in de ail (Palme and Mos l, 1997), alida ed and applied in ho ses (Hinchcli e e al., 2021; M¨ os l e al., 1999). The in a-assay CV o GCM EIAs was 2.1 % and he in e -assay CV was 3.2 % (n =8 con ol samples). 2.5. Die me aba coding Die composi ion was assessed om aecal samples h ough aecal DNA me aba coding. This echnique p oduces an accu a e and de ailed snapsho o wha he animal ecen ly consumed, and i coupled wi h hy oid ho mone concen a ions, can in o m on he nu i ional quali y o ce ain ood ypes o link sub-op imal o aging wi h nu i ional s ess (Jeannia d du Do e al., 2009; Shul z e al., 2021). Fi s , plan DNA was ex ac ed om he aecal samples using QIA- GEN QIAamp DNA mini-s ool ki (Qiagen, Ge man) - he manu ac u e ’s ins uc ions we e ollowed using 0.2 g o each aecal sample. Secondly, quan i ica ion o he amoun o DNA ex ac ed pe sample was pe - o med using he In i ogen Qubi ki and In i ogen™ Qubi ™ 4 Fluo ome e . Thi d, plan DNA was ampli ied o he P6 loop o he n-L (UAA) chlo oplas gene egion (Tabe le e al., 2007). Las ly, Sequencing was ca ied ou by Ilumina MiSeq. Sequencing used pai ed- end eads (2x150bp) and a MiSeq V2 Reagen ki . The inal lib a y loading concen a ion was 10pM and a 10 % spike-in o 4pM PhiX con ol V3 was included. Fu he de ails on he cycle imes and PCR p ocess a e ound in he supplemen a y ma e ial. In o de o analyse he DNA eadings om he sequencing p ocess, he o wa d and e e se adap e s p esen in he amplicon sequences we e emo ed using Cu adap 2.1 pipeline (Ma in, 2011). The sequencing gene a ed 3,258,818 aw eads, which we e il e ed and me ge using he DADA2 ( 1.18.0) (Callahan e al., 2016) package in R S udio (RS udio, 2020). Any samples o <5000 eads (n =10) we e emo ed (A amenko e al., 2015). The emaining eads we e compa ed o a sequence e e ence lib a y, combining lib a ies om Gill e al. (2019); and Kowalczyk e al. (2019). Taxonomy was assigned o amily le el o 44 % o amplicon sequence a ian s’ (ASV) and o genus le el Fig. 1. Sca e plo o NDVI and a e age, minimum and maximum 24 h s ai empe a u e ( ◦C) a he s udy loca ion o June 2020 o July 2021. The blue da a poin s and solid line ep esen he a e age 24 h s empe a u e (◦C), he ligh and da k blue do ed line a e he maximum and minimum ( espec i ely) ai empe a u e (◦C) o e a 24 h s pe iod, and he ligh -g een da a poin s and line ep esen he NDVI alues. J. G anweile e al. Ho mones and Beha io 161 (2024) 105526 5 o 42 %. A inal phylogene ic ee o he eads was gene a ed by he DECIPHER package (W igh , 2016) in R S udio and cons uc ing a maximum likelihood ee in he phango n (Schliep, 2011) and phy ools (Re ell, 2012) packages in R S udio. The Phyloseq package (McMu die and Holmes, 2013) was used o compu e alpha die di e si y and ela i e ead abundancy (RRA) o he main ege a ion amilies ound in he aecal samples. Alpha die di- e si y es ima es die ichness using he Shannon-Wea e index. A low die a y alpha di e si y means ha he aecal sample con ain a low numbe o di e en ege a ion amilies. Measu es o RRA co ela e wi h biomass pe ege a ion amily consumed, and he e o e is used as an indica o o p opo ion o each axa consumed (Wille sle e al., 2014). G ass being he majo ood sou ce o equids, only he RRA o Poaceae was conside ed in he analyses o he bes p edic o s o T3M and GCM le els. 2.6. Food a ailabili y and clima e a iables To assess changes in ood a ailabili y, NDVI alues we e ex ac ed om NDVI maps gene a ed by Global Land Se ice o Cope nicus (300 m esolu ion e e y 10 days). NDVI is an es ima e o ege a ion g eenness, a p oxy o ege a ion biomass and ood a ailabili y o he bi o es, whe e high NDVI alues ep esen ‘g eene ’ ege a ion (Bo owik e al., 2013). Fo his, he as e calcula o ool in QGIS (QGIS De elopmen Team, 2016) was used o sub ac alues o he ed band om he nea -in a ed band and di ided by he sum o he ed and nea -in a ed bands. To in es iga e he in luence o ambien empe a u e on T3M and GCM concen a ions, da a o daily minimum, maximum and a e age empe a u e was calcula ed based on eigh daily aw empe a u e measu es om he nea by Capel Cu ig wea he s a ion ex ac ed om www. imeandda e.com. Re ospec i e-in eg a ed ambien empe a u e da a ha e been shown o co ela e be e wi h T3Ms le el changes in u ine han wi h he empe a u e on he day o u ine collec ion (Hassi e al., 2001). To accoun o his, in addi ion o daily minimum, maximum and a e age empe a u e alues on he day o aecal collec- ion, he a e age minimum, a e age and maximum empe a u e ac oss 48 hou s and se en days be o e he day he aecal sample we e also calcula ed. In o al six empe a u e a iables we e gene a ed: he min- imum, maximum and a e age empe a u e a 48 hou s and 7 day imescale. Clima e a iables ini ially included NDVI, empe a u e (minimum, a e age and maximum a a 48 hou s and 7 day window) and ain all (minimum, a e age and maximum a a 48 hou s and 7 day window). A e a PCA analysis, only NDVI and empe a u e a iables we e kep ( u he de ailed in supplemen a y ma e ial; Supplemen a y Figs. 2 and 3). NDVI and a e age empe a u e alues we e calcula ed o he whole 2021 yea (Fig. 1) o pu he ambien empe a u e and ood a ailabili y alues du ing he s udy pe iod in a b oade con ex o be e in e p e- a ion o he esul s. 2.7. S a is ical analyses We in es iga ed he e ec o in insic ( ep oduc i e s a e) and ex insic (ambien empe a u e, ood a ailabili y and die composi ion) on he aecal T3M concen a ions in ee-li ing Ca neddau ponies (Table 1). The global models o bo h T3M and GCM concen a ions included ep oduc i e s a e [ ep oduc i e emales (p egnan o lac a ing), non- ep oduc i e emale and s allion], NDVI (quad a ic e m), alpha die di e si y, RRA o Poaceae (logged o no mali y) and a empe a u e a iable (◦C) (quad a ic e m) as ixed e ec s, and indi- idual ID nes ed wi h g oup as andom e ec (Table 1). Due o high co ela ion be ween he empe a u e a iables [a e age co ela ion co- e icien =0.638 ±0.226 ( min: 0.146, max: 0.960)], only one could be kep o u he modelling (Do mann e al., 2013). Mo eo e , PCA analysis indica ed ha all empe a u e a iables loaded in a simila di ec ion (supplemen a y Fig. 3), he e o e indi idual models was a - ou ed o e inclusion o a PC1 a iable o de ec di e ences in ime window sensi i i y be ween T3M and GCM concen a ions. We he e- o e an se en global models o T3M and se en global models o GCMs, each a ying in he empe a u e a iable (null model and six empe a u e a iables). AiCc based model selec ion was pe o med o selec he bes global model ou o he se en empe a u e global models o bo h he T3M and GCM analyses. The a e age empe a u e ac oss he pas 7 days was selec ed as he empe a u e a iable bes explaining T3M a ia ion and a e age empe a u e 48 hou s p io o aecal sample collec ion be e explained he a ia ion in GCM concen a ions. We only included he bes global model in ou esul s, he o he global models ou pu s and model selec ion can be ound in he supplemen a y ma e ial (supplemen a y able 6-33). To es ou p edic ions mixed-e ec linea models we e conduc ed using he lme4 package (Ba es e al., 2015). Model selec ion and a e - aging was conduc ed wi h he d edge unc ion o he package MuMin, which gene a es all possible model combina ions and e alua es model i based on AICc in o ma ion c i e ia (Ba on, 2020), using a di e ence be ween AICc (ΔAICc) o less han wo as a h eshold o op model selec ion and model a e aging (Symonds and Moussalli, 2011). Ele a- ion was kep in p elimina y models bu soon emo ed due o he lack o pa e ns in he da a (see Supplemen a y sec ion I and supplemen a y able 2 o mo e de ail). A e inspec ion o he sp ead o he aw da a, bo h NDVI and he empe a u e a iables we e included as a quad a ic e m in he ollowing models. Bo h T3M and GCM da a was checked o no mal dis ibu ion, bu only GCM le els we e skewed and logged o es o e no mali y. The global model o T3M concen a ions included ep oduc i e s a e [ ep oduc i e emales (p egnan o lac a ing), non- ep oduc i e emale and s allion], NDVI (quad a ic e m), alpha die di e si y, RRA o Poaceae (logged o no mali y) and 7 day a e age Table 1 Va iables used in he gene al linea mixed e ec models. Va iable Desc ip ion/Le els Dependen a iables Le els o T3Ms* (ng/g) Con inuous Va iable (CV) ( ange =0.279–1.703; mean ±SD =0.941 ±0.347) Le els o GCMs † (ng/g) CV ( ange =0.950–8.847; mean ±SD =3.373 ± 1.432) Independen a iables ( ixed e ec s) Ambien empe a u e 7 day a e age empe a u e (◦C) A e age empe a u e o 7 days p io o he aecal sample collec ion. CV ( ange =3.321–14.250; mean ± SD =7.177 ±2.106) A e age 48 hou s empe a u e (◦C) A e aged empe a u e 48 hou s p io o he aecal sample collec ion. CV ( ange =2.688–15.938; mean ± SD =7.520 ±2.388) NDVI No malised Di e ence Vege a ion Index. CV ( ange = 0.395–0.631; mean ±SD =0.488 ±0.074) Die Alpha die di e si y Die ichness es ima ed using Shannon-Wea e index. CV ( ange =1.465–3.344; mean ±SD =2.730 ± 0.327) RRA o Poaceae Rela i e Read Abundancy (RRA) o he Poacea (g ass) amily. CV ( ange =0.042–0.855; mean ±SD =0.393 ±0.274) Ele a ion (m a.s.l. ₸ ) CV ( ange =254 - 661; mean ±SD =415.700 ± 108.195) Sex / Rep oduc i e ca ego y Fac o a iable: 3 le els (s allions, ep oduc i e emales (p egnan o lac a ing), non- ep oduc i e emale) Independen a iables ( andom e ec s) Indi idual Fac o a iable: 61 le els (61 indi iduals: 55 adul emales and 6 s allions) * Faecal T3 me aboli es. † Faecal glucoco icoid me aboli es. ₸ Me e s abo e sea le el. J. G anweile e al. Ho mones and Beha io 161 (2024) 105526 6 empe a u e (◦C) (quad a ic e m) as ixed e ec s, and indi idual ID nes ed wi h g oup as andom e ec (Table 1). The global model explo ing a ia ion in log GCM concen a ions (logged o no mali y) con ained he same a iables as he global model o T3Ms, o he excep ion ha he empe a u e a iable was eplaced o a e age 48 h empe a u e (◦C) (Table 1). Mo eo e , da e o sample collec ion was no included in he global model because he empo al pa e n o sampling was unbalanced ac oss ca ego ies and indi iduals (i.e., some da es did no include samples o all indi iduals o ca ego ies). Ins ead, a singula model including da e was pe o med o in es iga e he p esence o acu e empo al spikes in he GCM da a. In all models, con inuous a iables we e cen ed a ound he mean, and scaled esiduals inspec ed o ou - lie s and model i (Ha ig, 2020). Mul icollinea i y o he a iables in he global models was lowe han 1.821. Pos -hoc pai wise compa isons we e conduc ed wi h he emmeans package (Len h e al., 2018). Sum- ma y s a is ics epo he mean ±s anda d e o o he mean unless no ed. The epo ed impo ance (Tables 2 and 3) is he sum o Akaike weigh s o e all models including he explana o y a iable. The e ec size was calcula ed by di iding he absolu e alue o a p edic o ’s co- e icien by he s anda d de ia ion o he p edic o . A Bon e oni co ec ion was applied o he p- alue o he empe a u e a iables in ou global model, o accoun o he alse posi i es and ampli ied p- alues om mul iple compa ison es ing (Cabin and Mi chell, 2000). All s a is ical analyses we e pe o med on R e sion 4.1.0 (R Co e Team, 2020). 3. Resul s 3.1. Clima e and ege a ion pa e ns in 2021 NDVI and a e age 24 hou s empe a u e (◦C) we e calcula ed o a 12 mon h pe iod p io o he end o da a collec ion (June 2020-July 2021). Bo h NDVI and a e age 24 hou s empe a u es we e highes in summe 2020, dec easing h oughou he all o 2020 o each he lowes alues in win e 2020 (Decembe – Ma ch). NDVI and a e age 24 hou s empe a u e hen inc eased again om Ma ch 2021 un il summe 2021. Ou s udy pe iod, highligh ed below in o ange, includes bo h he end o he low NDVI and a e age 24 hou s empe a u e and he inc ease in sp ing (Fig. 1). 3.2. Die composi ion The mos abundan amilies consumed by he s udy pony popula ion we e iden i ied as Poaceae (38 %, g asses), E icaceae (11 %, hea he and bilbe y), Juncaceae (10 %, ushes), Rosaceae (5 %, e.g., haw ho n, owan om he ose amily) and o he amilies comp omised ≤3 % each (Fig. 2). Th oughou he s udy pe iod, he ponies’ die was ini ially domina ed by Poaceae in Ma ch o mid-Ap il, o hen di e si ied include o he amilies such as E icaceae and Juncaceae in mid-Ap il o ea ly May and Rosaceae in mid o la e May (Fig. 2). 3.3. Faecal T3M concen a ions Be ween Ma ch o June 2021, a o al o 100 aecal samples we e collec ed (Adul emales n =64, S allions n =36) ac oss 61 indi iduals (Adul emales n =55, S allions n =6) (Supplemen a y Table 4 o de ails). Faecal T3M concen a ions showed a sha p 45 % decline be- ween Ma ch o mid-May (1.450 s 0.646 ng/g), emained low du ing mos o May and hen inc eased a he end o he s udy pe iod (Fig. 3). This pa e n ma ched NDVI and a e age weekly empe a u es ends, wi h a posi i e quad a ic ela ionship be ween T3M le els and NDVI (Table 1, Fig. 3A) and wa me a e age ambien empe a u es (Table 1, Fig. 3B). Mo eo e , inc eased T3M concen a ions we e signi ican ly linked o a dec ease in alpha die di e si y and inc ease in RRA o Poaceae (Table 1, Fig. 3C and D). 3.4. Faecal GCM concen a ions Faecal concen a ions o T3 and GC me aboli es we e no co ela ed ( 99 =0.076, p- alue =0.448,N =100). Log GCM le els showed a dec easing end ac oss he s udy pe iod (Fig. 4A), wi h a nega i e quad a ic ela ionship be ween a e age empe a u e o e he pas 48 hou s and log GCM concen a ions (Table 2). Log GCM le els we e highe a bo h ex eme o he p e ious 48 hou s empe a u e ange (Fig. 4B). Addi ionally, p egnan , o lac a ing emales had highe log GCM le els han non-p egnan o non-lac a ing emales and s allions (Fig. 5). Da e o sample collec ion was no included in he global model because he empo al pa e n o sampling was unbalanced ac oss ca e- go ies and indi iduals (i.e., some da es did no include samples o all indi iduals o ca ego ies). Ins ead, a singula model including da e was pe o med o in es iga e he p esence o acu e empo al spikes in he Table 2 Full model-a e aged es ima es o ou en i onmen al a iables p edic ing aecal T3M concen a ions (ng/g) in he Ca neddau ee-li ing pony popula ion. Alpha die di e si y, ela i e ead abundance (RRA) o Poaceae, 7 day a e age empe a u e (◦C) and NDVI we e cen ed and s anda dised indi idually, p io o compu a ion o he model and model a e aging. 7 day a e age empe a u e (◦C) and NDVI we e all included as a polynomial ac o (deg ee 2). RRA o Poaceae in aecal samples was logged o no mali y. Only he a iables kep a e model selec ion (ΔAICc <2) and a e aging we e p esen ed in his able. # Pos Bon e oni co ec ion he p- alue = 0.021. Dependen a iable: T3Ms Independen a iables: Rep oduc i e s a e, NDVI 2 , 7 day a e age empe a u e 2 , Die a y alpha di e si y and log RRA o Poacea. Random e ec : Indi idual ID nes ed in g oup name. Pa ame e Es ima e Uncondi ional SE Con idence in e al E ec size Z alue p- alue Rela i e impo ance (In e cep ) 1.132 0.052 1.029, 1.235 21.560 Die a y alpha di e si y −0.107 0.028 −0.162, −0.051 0.382 3.779 <0.001 1.00 Log (RRA Poaceae (g ass)) 0.154 0.036 0.082, 0.226 0.428 4.183 <0.001 1.00 7 day a e age empe a u e (◦C) Linea e m 0.234 0.309 −0.381, 0.848 0.076 0.745 0.456 Quad a ic e m −0.848 0.285 −1.413, −0.282 0.298 2.937 0.003 # 1.00 NDVI Linea e m 0.136 0.307 −0.475, 0.746 0.044 0.435 0.663 Quad a ic e m 1.395 0.294 0.812, 1.979 0.475 4.688 <0.001 1.00 J. G anweile e al. Ho mones and Beha io 161 (2024) 105526 7 GCM da a. The da e o sample collec ion (as a ac o ) also had a s ong e ec on log GCM concen a ions (F (1,18) =3.408, p- alue <0.001), sugges ing a s ong day o day a ia ion in glucoco icoid le els. Faecal samples collec ed on he 14 h Ap il 2021, in pa icula , had a highe log GCM concen a ions han he o he collec ion da es (p- alue anged om 0.012 o <0.001; see Supplemen a y Table 5). 4. Discussion Ca neddau ponies in Wales espond o he dual ene ge ic challenge o low ambien empe a u e and educed ood a ailabili y o win e wi h T3M le els dec easing om high concen a ions in la e win e o low le els in ea ly sp ing. Bo h a e age weekly empe a u e and ood a ailabili y (NDVI) dec eased h oughou he i s mon hs o he s udy yea in Snowdonia Na ional Pa k, and hen inc eased om he second hal o Ap il. In line wi h ou p edic ion, aecal T3Ms acked NDVI and Table 3 Full model-a e aged es ima es o h ee en i onmen al a iables p edic ing log aecal GCM concen a ions (ng/g) in he Ca neddau ee-li ing pony popula ion. A e age 48 hou s empe a u e and log (RRA Poaceae) we e cen ed and s anda dised indi idually. A pos -hoc emmeans analysis was un o e alua e he pai wise compa ison be ween ep oduc i e s a es. *Z- alue was epo ed o he a iables wi h he global model, whe eas - alue was epo ed o he pos -hoc es o signi icance be ween he le els o ep oduc i e s a e (Rep oduc i e emales =P egnan /Lac a ing emales; Non- ep oduc i e =Non-p egnan /Non-lac a ing emales; S allions). Only he a iables kep a e model selec ion (ΔAICc < 2) and a e aging we e p esen ed in his able. # Pos Bon e oni co ec ion he p- alues a e 0.028 and 0.042 espec i ely. Dependen a iable: log GCMs Independen a aiables: Rep oduc i e s a e, NDVI 2 , a e age 48 h s empe a u e 2 , Die a y alpha di e si y and log RRA o Poacea. Random e ec : Indi idual ID nes ed in g oup name. Pa ame e Es ima e Uncondi ional SE Con idence in e al Z alue* E ec size p- alue Impo ance In e cep 1.362 0.068 1.228, 1.496 19.901 Log (RRA Poaceae (g ass) 0.033 0.045 -0.017, 0.150 0.738 0.073 0.461 0.5 Rep oduc i e S a e 18.738 <0.001 1.00 P egnan /lac a ing ♀ VS Non- ep oduc i e ♀ 0.316 0.095 0.085, 0.547 3.329 0.333 0.005 P egnan /lac a ing ♀ VS S allions 0.286 0.085 0.066, 0.506 3.367 0.336 0.011 Non- ep oduc i e ♀ VS S allions −0.030 0.094 −0.281, 0.221 −0.316 0.032 0.947 A e age 48 hou s empe a u e (◦C) Linea -0.964 0.329 -1.617, -0.312 2.898 0.293 0.004 # Polynomial 0.921 0.330 0.266, 1.576 2.756 0.279 0.006 # 1.00 Fig. 2. Desc ip i e ba -plo o die composi ion o he Ca neddau ponies ac oss he s udy pe iod. Die composi ion was es ablished om DNA me aba coding o plan ma e ound in aecal samples. Week a e calenda weeks wi hin a yea . RRA is ela i e ead abundancy (RRA) o each plan amily – illus a ing he p opo ion o each amily wi h he die . J. G anweile e al. Ho mones and Beha io 161 (2024) 105526 8 empe a u e, sha ply declining om Ma ch un il end o Ap il (45 % decline) – sugges i e o hypome abolism, and hen inc easing om mid- May wi h inc easing ood a ailabili y. This seasonal pa e n has also been obse ed in whi e- ailed dee , a species ha unde goes hypo- me abolism in win e (A nold, 2020), wi h dec eased ci cula ing se um T3 le els in ee anging (Ham and Bubenik, 1990) and penned semi- s a ed indi iduals (Bahnak e al., 1981a; Bahnak e al., 1981a; Bah- nak e al., 1981b). Simila ly, (B inkmann e al., 2016) ound ha win e ci cula ing le els o T3 o She land ponies in Ge many declined 61 % om No embe o Ma ch, his educ ion in hy oidal ac i i y being accompanied by a 32 % educ ion in ield me abolic a e. Alongside NDVI da a, die DNA me aba coding enabled us o quan- i y changes in die composi ion which we e associa ed wi h he dec ease in T3M le els, possible s a o hypome abolism, in he Ca - neddau ponies. Speci ically, he ponies a e ela i ely less g ass and hei die had a highe alpha di e si y in ea ly sp ing, sugges ing ha ponies we e swi ching om a p ima ily g ass-domina ed die o ei he allback oods o seasonally a ailable lea shoo s. This same pe iod was associ- a ed wi h lowe T3M le els. G azing species in o he seasonal habi a s simila ly espond o a educ ion in he a ailabili y o he baceous ege- a ion by shi ing o a mo e di e se non-g ass die (Ka zinel e al., 2015). He e we demons a e die shi s associa ed wi h educed hy oid sec e ion. Speci ically, he ea ly sp ing decline in T3M concen a ions, sugges ing hypome abolism in ou s udy popula ion, was mos likely a esponse o poo nu ien in ake and declining ene gy ese es. This being said, he sp ing ‘ lush’ associa ed wi h inc eased NDVI was no associa ed wi h an immedia e swi ch o highe consump ion o Poaceae, meaning ha al hough ood was mo e plen i ul (highe NDVI) he ponies s ill main ained a di e se die . One explana ion may be ha ponies inc ease consump ion o all oods as new ege a ion g ow h in- c eases. As RRA is an index o ela i e eads i canno accoun o changes in o al amoun consumed. NDVI is a p oxy o ege a ion biomass as i measu e o ege a ion g eenness (Bo owik e al., 2013) and inc easing NDVI may be associa ed wi h ponies inc ease consump ion o g ass and o he i ems. Due o he seasonal lush, g ass and o he plan amilies will ha e nu ien ich new shoo s, which he ponies migh indisc imina ely eed on (Kams a, 1973). Including a measu e o o e all biomass inges ed could complemen die composi ion o be e indica e changes in calo ic in ake. In con as o T3Ms, GCM le els did no a y seasonally and we e no ela ed o esou ce a ailabili y o die composi ion. Declining esou ce a ailabili y associa ed wi h inc eased GCs in an numbe o mammalian species (Foe s e and Mon o , 2010; Jeannia d du Do e al., 2009; P ide, 2005) has been in e p e ed as e idence o Fig. 3. T3M concen a ions (ng/g) measu ed in aeces o Ca neddau ee-li ing ponies and 4 en i onmen al a iable: (A) no malised di e ence ege a ion index (NDVI) (g een line), (B) 7 day a e age empe a u e ( ◦C) (blue line), (C) die a y die di e si y (b own line) and (D) ela i e eads abundance (RRA) o Poacea (yellow line), ac oss he s udy pe iod. Poin s display aw T3Ms da a wi h a quad a ic line i ed h ough he aw da a (black line). The shaded black a ea ep esen s he s anda d e o a ound he quad a ic line. (Fo in e p e a ion o he e e ences o colo in his igu e legend, he eade is e e ed o he web e sion o his a icle.) J. G anweile e al. Ho mones and Beha io 161 (2024) 105526 9 indi iduals mobilizing ene gy ese es in he absence o adequa e ood (Kuo e al., 2015). The lack o e idence o such a esponse in ou s udy popula ion sugges s ha Ca neddau ponies educe hy oid ac i i y o adap a i ely manage ene gy ese es a he end o win e , wi hou equi ing he egula o y ac ion o GCs. Faecal GCM concen a ions we e nega i ely co ela ed wi h a e age empe a u e o e p e ious 48 hou s, wi h he highes GCM concen a ions associa ed wi h he cold spells. GCs can p omo e se e al adap i e he mo egula o y esponses o cold empe a u es, including asodila ion and hea conduc ion (de B uijn and Rome o, 2018; Gu h ie and Lund, 1998) and so a e po en ially an adap i e homeos a ic esponse o main ain co e body empe a u e. Ele a ed aecal GCM concen a ions a colde empe a u es ha e also been epo ed in geladas The opi hecus gelada (Beehne and McCann, 2008), ed dee (Co la i e al., 2011; Hube e al., 2003) and a c ic g ound squi els U oci ellus pa yii (Zhang e al., 2020). Hube e al. (2003) also obse ed his in ed dee and no ed ha he ela ionship was mos p onounced when empe a u e we e <10 ◦C. In endo he ms, body empe a u e egula ion is leas cos ly wi hin hei he mal neu al zone (Benne , 1988; Mo ison e al., 2008), which in domes ic ho ses is be- ween 5 ◦C - 25 ◦C (Mo gan, 1998; Snoeks e al., 2015). This is consis en wi h he ela ionship we documen ed, wi h a s ong nega i e co ela ion be ween GCM concen a ions and empe a u e a low empe a u es, bu a less clea ela ionship a highe empe a u es. Ele a ed GCM concen a ions, bu no T3M concen a ions, we e ound in p egnan and lac a ing emales compa ed o non- ep oduc i e emales and s allions. Des o ges e al. (2021) epo ed ha p egnan muskoxen emale emained in a hypome abolic s a e, by main aining low body empe a u e, o su i e win e condi ions. Hypome abolism expe ienced by his pony popula ion, including he p egnan and lac a ing emales, may masks any expec ed T3Ms signal a ound pa u- i ion and lac a ion. The obse ed highe GCM concen a ions in ep oduc i ely ac i e emales (i.e., p egnan o lac a ing emales) is due o he elease o oes ogen and co isol om he placen a (McLean and Smi h, 1999). Highe co isol le els du ing la e ges a ion has been e- po ed in mammals ( e iewed in (Edwa ds and Boons a (2018)). Howe e , he e is con lic ing e idence in ungula es, wi h no e ec o p egnancy/lac a ion on co isol le els in some s udies [domes ic ho ses (Au ich e al., 2015; Yo k and Schul e, 2014); sheep (Win ou e al., 1976); cow (Comline e al., 1974); ed dee (Hube e al., 2003; Ven- ella e al., 2020)], and a posi i e associa ion be ween p egnancy and ele a ed co isol in o he s [ho ses (Nagel e al., 2012); sheep (B une and Sebas ian, 1991); cow (B aun e al., 2017; Pa el e al., 1996); ed dee (Pa i e al., 2016)]. Mos o hese s udies we e on domes ica ed popula ions, which a e ca ed o o mi iga e he ex a needs o p eg- nancy – condi ions ha a e no a ailable o wild popula ions. In wild ungula es, e idence o ele a ed GCMs we e linked o pe iod o heigh en bu den du ing p egnancy, such as he las weeks o p egnancy (B aun e al., 2017; B une and Sebas ian, 1991; Nagel e al., 2012; Pa el e al., 1996), age o he p egnan mo he (Pa i e al., 2016), ea ly lac a ion (Lang e al., 2012) o sex o he o sp ing (Pa i e al., 2016). He e, GCM le els o ep oduc i ely ac i e emales we e only measu ed a he end o hei p egnancy o he s a o lac a ion, bo h pe iods o high me abolic cos s. Such condi ions may inc ease allos a ic load and igge a s ess esponse (i.e., GC sec e ion) allowing hese emales o empo- a ily mobilise ene gy ese es o acili a e he high ene gy demands o la e p egnancy and lac a ion. This adap i e esponse could be u he Fig. 4. (A) Log GCM concen a ions (ng/g) measu ed in aeces o Ca neddau ee-li ing ponies ac oss he s udy pe iod, and (B) wi h a e age 48 hou s empe a u e. Poin s display log GCMs da a wi h a local polynomial eg ession i (loess) line i ed h ough he aw log da a (black line). The shaded g ey a ea ep esen s he s anda d e o a ound he loess line. J. G anweile e al.