Rare and Opportunistic Use of Torpor in Mammals—An Echo from the Past?

Synopsis Torpor was traditionally seen as a winter survival mechanism employed by animals living in cold and highly seasonal habitats. Although we now know that torpor is also used by tropical and subtropical species, and in response to a variety of triggers, torpor is still largely viewed as a highly controlled, seasonal mechanism shown by Northern hemisphere species. To scrutinize this view, we report data from a macroanalysis in which we characterized the type and seasonality of torpor use from mammal species currently known to use torpor. Our findings suggest that predictable, seasonal torpor patterns reported for Northern temperate and polar species are highly derived forms of torpor expression, whereas the more opportunistic and variable forms of torpor that we see in tropical and subtropical species are likely closer to the patterns expressed by ancestral mammals. Our data emphasize that the torpor patterns observed in the tropics and subtropics should be considered the norm and not the exception.


Introduction
Whole-b o dy en doth ermy, th e th erm o regulato ry ab ili ty of ma inta ining a high a nd stable b o dy temp erature, h a s been one of the key steps in the evol u tio n o f m amm al s and birds ( Cro mpto n et al. 1978 ;Bennett and Ruben 1979 ), as it a l lows anima ls to ma inta in a b o dy temp erature ab ove ambien t tem pera ture , t o be independent o f amb ient co ndi tio n s, and to liv e and be activ e in cold hab i t ats.D esp i te t hese advant ages, t h e en doth ermic lifestyle does, how ev er, co me wi th much higher ener g etic costs than ectot her my.These can be somewhat re duce d by energy savings thr ough cir c adi a n cha nges in b o dy temp eratures with higher values during activ it y and lower values during resting ( Aschoff 1963 ;Tattersa l l 2012 ;Le ves que et al. 2023 ).Da ily va riation in b o dy temp erature can b e p ro n oun ced in en doth erms, espe cia l ly wh en in div idu a ls are subj e c t to hig h da ily va riations in ambien t tem pera ture and fo o d avai labi lit y (e.g ., up to 6 • C in n umba ts; Co op er an d With ers 2004 ).How ev er, by far the most effe ct iv e ener gy-savin g st rategy avai lable to m amm al s and bird s i s t he lower ing of met abolic rate v i a the use of torpor ( G eis er 2004 ).
To rpo r is genera l ly viewe d as a highly controlled mech ani sm of sea son al metabolic depression ( Lyman et a l. 1982 ), usua l ly accomp anie d by a drop in b o dy tem pera ture (but see Grimpo et al. 2013 or Re h er an d Dau sm ann 2021 ).The literature usua l ly different iates between daily h e teroth erms an d hibern at o rs .Daily heterot her ms are defined by expressing daily bouts of torpo r wi th a regular bu t moderate red uctio n in energy expen diture an d body tem pera ture .Suc h bou ts o f daily to rpo r are often rest ricte d to the rest ph a se, a nd a nim al s m ay st i l l ma inta in normal f orag ing act iv it y durin g the activ e p erio d ( Ruf and G eis er 2015 ).Hibernatio n, o n t he ot her h and, i s ch aracterize d by extende d p erio d s of in activ it y during which en ergy expen diture is A dvance A ccess pu blication Jun e 16, 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 ( ht t ps://creat ive co mmo 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 .
re duce d to a fraction of that dur ing nor mot her mic rest and b o dy temp erature is often close to ambient temperature ( G eis er 2021 ).During hibern ation, anim al s under g o to rpo r bou ts o f a few days to w eeks, g enera l ly int errupt ed by s h ort, regula r a rous al per iods ( French 1985 ;Ruf et a l. 2021 ).Typica l examples of cl assic al seaso nal No rt her n hemisphere to rpo r use are s h o wn b y the Djunga ria n ha ms ter ( P ho d op us s u n gorus ), which uses bou ts o f daily to rpo r, an d hi bern ating A r ctic gr ound squirre l (Uro citellus p ar ryii) with b o dy temp eratures b elow 0 • C ( Barnes 1989 ).In both species, to rpo r i s u sed on a regular ba si s during winter, is absent during the summ er, an d i s u sua l ly accomp anie d by m orph ological chan g es, such as a pho to p erio d-induced winter phenotype in Djunga ria n ha mst er s ( Hoffmann 1973 ) or p ro n oun ced s eas onal fattening in Arctic ground squirre ls ( Sh eriff et a l.2013 ).E xamples li ke thos e des cribed a bov e hav e init ia l ly le d to th e gen eral ide a t hat t he ability to use to rpo r m ay h av e ev o l ved as a w inter-surv ival mech ani sm in response to cold hab i ta ts ( Twen te and Tw ente 1964 ).How ev er, to rpo r is not only employed by anim al s living in cold and highly s eas o nal hab i tats, bu t also by species living in warm and/or arid c limat es outside of the Holarctic region (e.g ., rev iewed in Geiser and Kört ner 2010 ; McKe chnie and Mzi li kazi 2011 ; Boyles et al . 2013 ;Nowac k et al . 2020 ).Int erestingly, t o rpo r patterns in non-Holarctic species are often more variable t han t h ose comm on ly observe d in Nort her n temperate zon es, an d to rpo r exp ressio n ca n ra n g e from shallow to rpo r ( Sha nka r et al . 2022 ;Nowac k et al . 2023 ) or very s h or t micro-tor p or b outs ( Re h er et al .2018 ) t o exten siv e to rpo r bou ts o f a few mo nths wi thou t regula r a rou sal ph a ses ( Dau sm ann et al. 2004 ;Dau sm ann et a l. 2020 ;Loveg rove et al. 2014 ).Furt her more , t orpo r exp ressio n is o ften mo re flexible than p reviously as s um ed, an d in div idu als of one species or po p u lat ion may differ in their use of t o rpo r and either employ extended p erio d s of in activ it y w ith a series of mu lt iday or w eeklon g torpor b outs ( hiber nation ) , mu lt iday bou ts o f to rpo r lasting a few days ( p ro lo nged t o rpo r ), or s h o rt bou ts o f to rpo r , similar in d uratio n to the bou ts of daily torpor described for daily h eteroth erms (reviewed in Nowack et al. 2020 ; also see Glossa ry f or defini tio ns).Au stralian Ea s tern pygmy pos s ums ( Cercartetu s nanu s ), for example, co mmo nly under g o hibernatio n d uring winter, bu t individ uals vary in their to rpo r pattern, an d som e in div idu al s m ay only u se s h o rt o r p rolo n g ed bouts of torpor and do not under g o hibernation ( Turner et al. 2012 ).Furt her more, male pos s ums m ay al so u se o p portunistic s h o rt bou ts o f to rpo r d uring the s ummer, pos sib l y in response to a negative energy b a lance d ue to p rio ri tizin g searchin g for females over fo o d acquisi tio n ( Turner et al. 2012 ).Impo rt ant ly, such o p portuni stic u se of torpor enables non-Holarctic anim al s to respond to acute emer g encies and h a s been described to be tr ig gered in response to fast-scale natural di sa st er s, suc h as dr oughts, fir es, sto rms, o r flo o ds (re vie wed in Nowack et al. 2017 ) and less predictable c limat es in sub tro pical and tro pical areas also coincide wi th mo re variable d uratio ns o f to rpo r bou ts (re vie wed in G eis er 2020 ; Nowac k et al .2020 ).
The widespre ad occur rence o f to rpo r in a l l thre e m amm ali an subcl asses (m on otrem es, marsu p ials, and placent als) sug gests t h at it i s an evol u tio naril y o ld trait and that endot her my evo l ved v i a a h eteroth ermic ancestor that likely showed high variab ili ty in b o dy temperatures incl uding so me fo rm o f to rpo r use ( Gr ig g et a l.2004 ; Loveg rove 2012 ; Ruf and G eis er 2015 )-a view th at i s n ow wide ly accepted am ong scientists.As ea rly ma mmals evo l ved un der con ditions m ore similar to species in the tropics, this has led to the idea that flexible use of torpor may be the ancest ra l for m t hat would have been used by early mammals at the t ransit ion to th e wh ole-b o dy en doth ermy foun d in exta nt ma mmals and birds ( Gr ig g et a l.2004 ; Loveg rove 2012 ).As glob a l tem pera tures cooled during the Cenozo ic, mo re co nst raine d, pre dict able for ms of torpor w ould hav e been made p ossible ( L o vegro ve 2012 ; Lo vegro ve 2017 ).I f this i s the ca se, o ne would therefo re expect o p portunistic to rpo r use to be widespread and not rest ricte d to spe cies in non-Holar ctic r egions, and s eas ona l, prog rame d torpor to be rest ricte d to tem pera te and polar c limat es.
To evaluate bioge og raphica l and phylogenetic patterns of to rpo r use in mammals, we col late d datasets on m amm al s known to u se to rpo r.We n oted th e types o f to rpo r us ed, the s eas on(s) in which to rpo r h a s been r ecor ded, a s well a s if the anim al s ne e de d sig nificant, i .e .weeks or mo nths' , o f p rep arat ion t ime before entering hib ernation.We hyp oth esized that n on-s eas onal and flexible use o f to rpo r is the ancest ra l co ndi tio n in mamm al s and would ther efor e be found across mu lt iple taxa an d wor ldwide, an d f urt her pre dicte d that st rictly seaso nal to rpo r exp ressio n that r equir es pr ep arat ion wou ld be rest ricte d to cold and/or highly s eas o nal enviro nm ents an d only a few taxa.

Methods
We b ase d our dataset o n co mp re h en siv e re vie w articles ( Ruf and G eis er 2015 ;Le ves que et al . 2016 ;Nowac k et al. 2020 ) and ran a systematic literature search to update these with the r ecor ds of new (or previously mi ssed) di scoveries to co mp ile a list o f m amm a l spe cies known to use to rpo r to date (see Supplementary Table S1).To obtain a sta nda r dizable pr oxy for a spe cies' ge og raphica l ran g e, w e added mid-ran g e dist ribut ion (latitude and longitude) for each spe cies b ase d on entries in the m amm al life hi story databa se PanTHERIA ( Jones

Daily heterotherm
Animals that only display bouts of daily torpor lasting less than 24 hours (in contrast to hibernators ).

Daily torpor
Short bouts of torpor of less than 24 hours; emplo y ed by daily heterotherms .

Flexible torpor use
Flexibility in torpor pattern with individuals being able to switch between torpor types or individuals of a population adapting different torpor types, even in response to the same environmental conditions.

Heterothermy
Fluctuations in body temperature and metabolic rate seen in endothermic animals.

Hibernator
Animals that undergo hibernation .

Hibernation
Sequence of multiday torpor bouts interrupted by periodic arousal and short normothermic periods of sev eral hours; r estricted to the hibernation season (sev eral w eeks to months) and without extended activity and foraging.

Opportunistic torpor
Spontaneous and flexible torpor use that can occur at any time of the year in a flexible manner.

Prolonged torpor
Multiday torpor bout after which animals are normothermic and active for days before potentially underg oing ad ditional bouts of short or prolonged torpor; Occurs independent of hibernation.

Seasonal torpor
Torpor use is restricted to a single season, usually the cold or the dry season.

Short torpor
Short bouts of less than 24 hours emplo y ed by animals that can also undergo longer multiday torpor bouts ( prolonged torpor or hibernation ); bouts are occurring at any time of the year and independent of the hibernation season et al. 2009).D at a on t he distr ibu tio n fo r 10 species missin g from PanTHERIA w ere obtained from the Intern ation al Unio n fo r Co nservatio n o f Nature (IUCN) Red List and were down lo ade d a s sh apefiles ( IUCN 2022 ), from which the centroid was ca lcu late d using the "st_cent roid" funct ion in the R package "sf" ( Pe besm a 2018 ).We used the mid-ran g e latitudes to classify each sp ecies as b elon gin g to one of four cl i mat e zo nes ( t ro pical : 0-23.5 • N/S, subt ro pi cal : 23.5-40 • N/S, temp erate : 40-60 • N/S, or p ol ar : 60-90 • N/S).We lacked mid-ran g e coo rdinates fo r the follow ing t wo species: Mu s mu sculus (a commensal species with a worldwide distribution) and one unnamed species Gerbi l lus sp eci es 1.However, t o inc lude these species in the an alysi s, we cla ssified them as sub tro pical : Mu s mu sculu s based on their evol u tio nary o rigin ( Bo urso t et al. 1996 ) a nd Ger bi l lus sp eci es 1 b ase d on the locatio ns o f the study describing to rpo r use for the species ( Gyhrs et al. 2022 ).
We t hen synt hesized t he f ollowing inf o rmatio n fo r each species in our dataset via an exten siv e literatur e sear ch (Supp lementary Tab le S1): maximum torp or b out duration, s eas onality, and torpor pre dictabi li ty.To rpo r bou t d uratio n wa s cla ssified a s eith er d aily t o rpo r (lon g est dura tion < 24 h), pr ol onged torp or (multiday bouts int er sper sed by ext ended activ it y), hibernatio n (seq uence o f mul tiday bou ts), o r da ta deficien t, if n o con clu sive inform ation ab out torp or b out duration wa s available.Sea son ali ty o f to rpo r u se wa s cla ssified as s eas onal if to rpo r u se wa s rest ricte d to one s eas on such as the cold or dry p erio d , non-se asonal if it was seen in different seaso ns, o r d ata d efici ent if this inform ation wa s mi s sing or a s pe cies was on ly invest igate d in one s eas on or under con stant la boratory condition s.
To rpo r p redictab ili ty was defined as p redict ab le when a l l indiv idu al s u sed to rpo r in the same way, u npred ict ab le wh en th ey differ in th eir like lih o o d to enter to rpo r o r t he patter n o f to rpo r they are using, or d ata d efici ent if t his infor mation was missing.We also n oted wh eth er a hib ernating sp ecies was known to pr epar e for hibernation (fo o d hoardin g or fattenin g; y es/n o) an d wh eth er a species that hibernates may be flexible in their use of to rpo r and use sho rt to rpo r bou ts in depen dent of hibern ation.Mi ssing inform ation wa s again li sted a s data d efici ent .
To estimate the ancest ra l c haract er stat e f or va rious pa ra met er s (s eas o nali ty, p redictab ili ty, and to rp or b out duration), we used the "ace" function in the R package "ape" ( Pa radis a nd Schliep 2019 ) with a ph ylogen y t rimme d from the Upham et a l.(2019) m amm a l supert re e .For discret e c haract eris tics, s uch as th e on es u sed in thi s study, the functio n p rovides a maximum-li keli ho o d est imat io n o f t he st a te a t each node following Pagel (1994) .

Global distribution and type of torpor use
The fu l l dataset sp ans 275 spe cies an d in cl udes o ne species o f mo not reme, 57 spe cies o f marsu p ials, and 217 placenta l spe cies (Supplementary Table S1).Species that use to rpo r were found in a l l four climat ic zo nes, wi t h t he va st m ajo ri ty in sub tro pical and tro pical areas ( n = 213 vs. n = 62 for polar and tem pera te).D aily heterot her ms , employing to rpo r bou ts o f < 24 h as the lon g est bouts, made up 42% (117/275) of the species in our dataset.Prolon g ed torpor was the maximum d uratio n o f to rpo r bou ts fo r 8% (21/275) o f th e species, an d 40% (109/275) were found to under g o hiberna tion.F or 10% (28/275) of the species, the actual pattern of to rpo r use remains unclear (see Supplementary Table S1).Hibernation was f ound in ma ny sp ecies in p ola r a nd tem pera te hab i tats but was much less co mmo n in sub tro pical and tro pica l spe cies ( Fig. 1 ).Dai ly to rpo r, o n t he ot h er han d, was mo re co mmo n in warmer c limat e zones and absent in the polar regions ( Fig. 1 ).
All to rpo r types, incl uding p rolo n g ed to rpo r, were foun d across th e m amm al ph ylogen y.Use of a single torp or typ e on ly was found in thre e cl ades: The d asyurid marsu p ials (Or der: Dasyur o mo rpha) an d s hrews (Family Soricidae) have so far been found to only use daily to rpo r, where as t he sciur id subfa mily Ma rmotini (chipmunks and ground squirrels) a l l use hi bernation.Th e p resence o f p rolo n g ed to rpo r an d hi ber nation in ot her marsu p ial sister clades (including other Aust ra lia n ma rsu p ial s) increa sed the like lih o o d of other torpor types at th e an cest ra l marsu p i al node ( Fig . 1 E).Ancest ra l character est imat ion indicate d an almost equal like lih o o d of any of the three to rpo r types being the ancest ra l condit ion (sca le d li keli ho o ds at the root of 0.35 for daily torpo r, 0.33 fo r hibernatio n, and 0.32 fo r p rolo n g ed to rpo r, Fig. 1 E).

Seasonality of torpor use
We could not find info rmatio n o n the s eas o nali ty o f to rpor use for 59% (161/275) of the species.In the 41% (114/275) of species for which information about the s eas o nali ty o f to rpo r u se wa s avai lable, st r ict se a son ality was re lative ly rare (39%; 44/114).To rpo r u se wa s found to be used in several seasons, i .e .non-se asonal in 61% (70/114) of species that were invest igate d in s e veral s easons.
Only ∼10% (22/213) of sub tro pical and tropical species are known for t heir str ict ly se a son al torpor use, in contrast to 36% (22/62) in polar and tem pera te zones regions ( Fig. 2 ).Non-s eas on al u se o f to rpo r was found in 22% of species using daily to rpo r (26/117), 28% of species using hibernation (30/109), and in 48% of species s h owing p rolo nged to rpo r (10/21) and was not rest ricte d to sub tro pical and tro p ical hab i tats ( Fig. 2 ).Only 6% (7/109) o f hibernato rs-a l l spe cies from subt ropica l and t ropica l reg ions-do n ot s h ow s eas onal fattening or fo o d hoarding (Supplementary Table S1).
About 38% (41/109) of species that are known to under g o hibernation w ere found to al so u se o p portunistic sho rt bo uts of t o rpo r or p ro lo nged t o rpo r wi thou t p rio r prep arat ion in response t o acut e n egative en ergy b a lan ce through out th e year.This use of s h ort bouts of torp or by hib ernat or s was found in higher frequency (43%; 26/61) in species in tropical or sub tro p ical hab i tats ( Fig. 2 ) but was also described in 33% (15/46) of temperate zones species (Supplementary Table S1).Short or p rolo n g ed to rpo r bou ts in hibernat or s have not been o bserved in s p ecies from p ol ar regions (0/2; Fig . 2 ).Sea son ality wa s mo re co ncen tra ted in certain clades t han ot h ers an d was th e m ost comm on con dition in t he Car nivora (5/5 species), t he Sciur idae (i .e., marmots an d squirre ls) (16/20), an d th e Primates (7/8), while rare in other no n-sci urid roden ts, ba ts, marsu p ials, and Afrot her ians (i .e., t enrecs, go lden mo les, an d e lephant shrews) ( Fig. 2 ).Bot h se a son ality an d n on-s eas onality were found to be e qua l ly li kely to be the ancest ra l torpo r co ndi tio n (0.499 vs. 0.501, respe ct i vel y, Fig. 2 ).

Predictability of torpor use
We found that 41% (111/275) of a l l spe cies display some deg re e of flexi bili ty in to rpo r use , while pre di ctabl e torpor use (i .e., a l l indiv idu als of one species use torpor always in the same way) was only r ecor ded for 8% (23/275) of a l l spe cies [data deficient: 51% (141/275)].Flexible to rpo r u se wa s highest in t ropica l an d su bt ropica l spe cies (44%; 93/213), and a l though i t wa s al so found in species in tem pera te a nd pola r regions (29%; 18/62), many Nort her n hemis phere s pe cies have high ly p redictable to rpo r use where a l l indiv idu als of a popu lat ion use torpor in the same way (24% (15/62) vs. 4% (8/213) in t ropica l an d su b tro p ical areas; Su pplem entary Fig. S1).Phylogen et ic p atterns in pre dictabi lity mir ror t h ose foun d in s eas o nali t y w it h t he major ity of pre dictable spe cies consist ing of the Carnivora, the Primates, an d th e Sciuridae (i .e., ma rmots a n d squirre ls).There was also a much higher li keli ho o d that the ancest ra l co ndi tio n wa s flexible, th at i s, no n-p redictable (0.70), than predictable (0.30, Supplementary Fig. S1).

Discussion
Our co mp re h en siv e re vie w of m amm alian to rpo r use s h ows that o p po rtunistic, no n-s eas o nal to rpo r is more co mmo n in trop ical and sub tro pical than in polar and tem pera te zone m amm al s, but it i s not limit ed t o these climatic zones and can also be found in species living in s eas o nal No rt her n h emisph ere hab i tats wi th cold wint er s.Thus, the bioge og raphic dist ribut ion data su ppo rt our hypothesis th at non-sea son al and flexible use of torpor is likely the ancest ra l co ndi tio n in m amm al s and that predictable s eas o nal to rpo r u se i s t he der ived for m.However, t h e an cest ra l state re const ruct ion ana lyses wer e mor e amb iguous.Al though we found a 70% li keli ho o d for no n-p redictab le, flexib le to rpo r use to be the ancest ra l co ndi tio n, there was an e qua l li keli ho o d for s eas onal and non-s eas onal torpor.Similar ly, th e an cest ra l state ana lysis revea le d an a lmost e qua l li keli ho o d of any of the three to rpo r types being th e an cest ra l co ndi tio n.Furth erm o re, our p re dict ion t hat se aso nal to rpo r exp ressio n requirin g exten siv e

(E)
The phylogenetic relationships of species with known torpor use type.The tip labels r epr esent the longest torpor type observ ed in the species, using the same color scheme as in (D).Pie charts at the nodes w er e estimated using ancestral state estimation, and the proportion of each color r epr esents the likelihood that the shared ancestor used that particular torpor type.prep arat ion wou ld be rest ricte d to cold, high ly s eas onal environm ents an d only a few taxa was n o t sup port ed .
These ambiguous results may be lar g ely due to sampling bia s, a s much work h a s focu sed o n studying to rpo r in species living in harsh and s eas o nal enviro nments an d se ldom in cl udes milder hab i tats.Ge og raphica l and phylogenetic biases in physiolog ica l datasets are unfortunate ly comm on an d can s e verely limit bo th o ur ab ili ty t o under st and cur ren t pa ttern s as w el l as to pre dict future ou tco mes ( W hi t e et al .2021 ).F or exam ple, str ict ly s eas o nal to rpo r an d hi ber nation t hat r equir es pr eparation was mostly found in tem pera te a nd pola r hab i tats (an d th ere m ostly con cen tra ted in the Marmo tini) b ut wa s al s o obs erv ed in a bou t 10% o f sub tro pical and tropica l spe cies.How ev er, these w ere m ainly clu stered in one family of Malagasy lemurs, the Cheirogalidae (reviewed in Dau sm a nn a nd Wa rneck e 2016 ; Nowack et al . 2020 ), whic h are known to use hibernation, live in the highly s eas onal dry forest or m ontan e regions of Mad agasc a r.Compa rable data are lacking from closely relate d spe cies in less s eas o nal enviro nments such as t he e aster n ra inf o rests o f Mad agasc a r.We also f ound th at for m any species, inform ation wa s mi ssin g a bout th e possi ble occurren ce o f to rpo r ou tside o f the dry and/or cold s eas on.Thus, non-s eas on al u se of torpor may be more widespre ad t han captured in our current dataset.
In contrast to o ur hypo t hesis, str ict ly se aso nal to rpor and hibernation that r equir ed pr eparation was not found to be rest ricte d to highly s eas onal tem pera te and polar zo ne hab i tats bu t also found in sub tro pical and t ropica l spe cies.Desp i te str ict ly se a son al u se by some sp ecies, torp or use can often st i l l be re lative ly flexi ble, as to rpo r use may differ between individuals within a single po p u lat ion (re vie we d in McKe chnie and Mzi li kazi 2011 ; Nowack et al. 2020 ).In the Malagasy mouse lem urs Micr ocebu s gri seorufu s and M. mu ri nus , for example, a l l indiv idu als ent er t orpor during winter, but the extent varies depending on body co ndi tio n, and indiv idu als of one po p u lat ion can use s h or t tor por, prolon g ed torpor, or hibernation in response to the same enviro nmental co ndi tio ns ( Ko b be et al. 2011 ; Vuarin et a l.2013 ).Unpre dictable use o f to rpo r h a s al so been foun d in North ern h emisph ere s pecies.In many s pecies, such as the li ttle b rown bat Myo tis luci fug us , t he extent o f to rpo r use varies in re lation to th e in div idu al's r epr od uctive status, wi th p regnant individ uals typ ica l ly using s h orter an d s ha l lower to rpo r bou ts ( Dzal and Brigham 2013 ), and to rpo r use in Mo n g olia n ha mst er s Meriones u n gu i cul atus in the labo rato ry was only found in females, b ut no t in males ( Watanabe et al. 2016 ).In contrast to sub tro pical species, a high percentage of Northern h emisph ere sp ecies do es, how ev er, use to rpo r in a high ly pre dictable ma nner, a nd a l l indiv idu als of a pop-u lat io n use to rpo r o n t he s ame days and in the same way.O ur ancest ra l state ana lysis on the pre dictabi lity o f to rpo r use s ugges ts t hat predict able to rpo r u se m ay be a derived co ndi tio n in response to highly s eas onal hab i tats.
It is im portan t t o not e t hat t here m ay be a di screpan cy between th e spe cies' cap acity to use no n-seaso nal to rpo r an d th eir actual implem entatio n o f i t, as the latter wi l l refle ct environmenta l co ndi tio ns an d n ot physiolog ica l cap abi lit ies ( Landry-Cuer r ier et al. 2008 ).Labo rato ry experiments have, f or exa mple, s h own that even species known for their highly s eas o nal to rpo r use , suc h as thirte en-line d g r ound squirr els Ict idomys t rideceml i neat us , ca n, if k ept under a rt ificia l warm roo m co ndi tio n s, chan g e their to rpo r pattern fro m hibernatio n to regular s h o rt to rpo r bou ts employed d uring winter, sprin g, and ev en summ er ( MacCann e ll an d Staples 2021 ).Furt her mo re, al t hough hiber natio n typ ica l ly requires exten siv e prep arat ion, the examples of the few ex clusiv e ly su b tro pical and tropical species that do not s h ow s eas o nal fattening o r fo o d hoardin g sugg est that under certain enviro nmental co ndi tio ns, a mo re flexi ble, n on-s eas on al u se of hibern ation m ay b e p ossible.Neverth e less, our m acroan alysi s s h ows that oppo rtunistic no n-s eas o nal to rpo r use was found globa l ly, and spe cies kn own for th eir us e of s eas onal hibernation in winter can under g o shorter o p portunistic bouts in s ummer, s uch as the European co mmo n gar den dormice Muscard i nus av ell anari us ( Pr etzlaff et a l.2014 ), e dible dormice Gl is gl is ( Wilz and Heldmaier 2000 ;Hoelzl et al . 2015 ), hispid poc ket mice Ch a e todipu s (Perognatu s) hi spidu s ( Wang and Hudson 1970 ), and a number of bat spe cies ( Fj el lda l et al. 2022 ).F or exam ple, although brown long-eared bats Plecot us aurit us u se cla ssical hibern atio n wi th lo ng periods o f to rpo r thro ugho ut winter, the y als o display var iable patter ns of torpor thro ugho ut the rest of the y ear, includin g short and p rolo n g ed to rpo r ( Fjelldal et al. 2022 ).Furt her, alt hough li sted a s tem pera te zone species here due to their mid-ran g e dist ribut ion, the study o n lo ng-eare d b ats was underta ken in the northernmost part of the species' ran g e in the sub arct ic polar r egion, r ev ealin g that ev en in highly s eas onal hab i tats, o p po rtunistic to rpo r use thro ugho ut the year is wides pread.Mos t hibernat or s were found t o under g o significant ph en otypic chan g es necessary to allow survival during long, cold, dry, and unproductive p erio ds.

Considerations about the evolution of endothermy
Im portan tly, our da ta provide further empirical su ppo rt that flexi ble an d o p po rtunistic to rpo r u se m ay be the ancest ra l form that would have been used by early mamm al s during the t ransit ion to whole-b o dy en doth ermy ( Gr ig g et a l.2004 ; Loveg rove 2012 ).Ea rly ma mmals were li kely sma l l, nocturna l, and inse ct ivo rous, wi th low t her mogenic c apacit y and high rates of heat loss ( Cro mpto n et al. 1978 ;O' Le ary et al. 2013 ).Although th e climatic con ditions at this time would have been war mer t h an today, anim al s would still have to deal with co ndi tio ns similar t o t oday's t ropica l and sub tro pical hab i tats, which can also see p erio ds of low nighttime tem pera tures and varying fo o d avai labi lity ( G eis er et al. 2017 ).Our cur rent underst anding of the evol u tio n o f en doth ermy i s th at early m amm al s and bird s evo l ved en doth ermy v i a a stepw i se increa se in metabolic rate that would have at least a l lowe d indiv idu als to decouple daytime activ it y from environmental conditions unt i l insu lat ion cap acit ies increase d ( Cro mpto n et al. 1978 ;Ruben 1995 ;G eis er et al. 2017 ), and it h a s been suggested t hat t he p rocess was facili t ated by t he use of o p po rtunistic to rpo r d uring colder days ( G eis er et al. 2017 ).More derived forms of t orpor, inc luding the highly s eas onal and p rolo n g ed form s, w ould hav e then evo l ved dur ing t he per iods of glob a l coo ling fo llowing the E ocene C limat ic O pt imum ( ∼40 Mya) concurrent wit h t he exten siv e m amm ali an radi at ions fol lowing the K-Pg m a ss ext inct ion ev ent ( Lov egrov e 2012 ; Lo vegro ve 2017 ).
Interestin gly, no sin gle to rpo r use t ype c ame out as m ost like ly to be th e an cest ra l st ate wit h daily to rpo r having only a slig htly hig h er like lih o o d than b oth prolon g ed torpor an d hi bernation.Ear lier a ttem pts to identify th e an cestral to rpo r type incl ude d invest igat io ns o f to rpo r use in clades that split from the rest of the mammal ph ylogen y at a n ea rlier dat e , suc h as t he Pat agonian opos s um ( Les t o d el p hys hall i , Or der: Dide lphim orphia), which was found to employ both s h ort an d deep p rolo n g ed bou ts o f to rpo r ( G eis er a nd Ma rtin 2013 ) .Thi s m ay s ugges t that s uch flexible u se m ay be the torpo r exp ressio n fro m which the more derived and seaso nal fo rms o f daily to rpo r an d hi bernatio n o rig inate d in marsu p ia ls.A simi lar ly flexi ble to rpo r use wi th s h ort and p rolo n g ed bou ts o f to rpo r and high s eas o nali ty has a lso be en foun d in th e m o ni to de l m on te ( Dr omi ci ops gl i roi d es , Or der: Micr obiot her i a), a cl ade that sits between t he Sout h Amer ica n a nd Aust ra l asi a n ma rsu p ial s ( Mejía s et al. 2022 ).How ev er, it needs to be noted t hat bot h species s h ow tail fatt ening, whic h is likely a der ived adapt ation to se a son al h abit ats.In tot al, at le ast 38% o f hibernato rs were found to be able to use s h ort and/o r p rolo n g ed to rpo r bou ts. Thi s percentage i s, however, lik ely a n underest imat io n, d ue to the high number o f species fo r which d ata are l ac king or whic h were only observ ed durin g one s eas on, an d th e numb er of sp ecies em ploying m u lt iple torpor use types is likely to increase wit h f urt her study.
Interestin gly, ev en species such as Djunga ria n ha mst er s tha t en ter regul ar d aily bou ts o f spo ntaneous to rpor in wint er aft er under g oin g pho to p erio d-induced m orph olog ica l chan g es can under g o o p po rtunistic to rpor in response to fasting thro ugho ut the year

Conclusion
Desp i te th e in creasing number of species found to use to rpo r in regio ns ou tside o f t he cold, se a son al tem pera te, a nd pola r zones a n d th e ackn owledgem ent that to rpo r u se i s not prim aril y emp loyed for winter survival , the c lassica l Holarct ic to rpo r pattern is st i l l widely seen as a ma mmalia n norm.This is perhaps partly due to a lack of con sen sus in the definitions of different patterns of to rpo r use, s ugges ting the ne e d fo r mo re universa l l y app lie d definit ions of torpor such as those presented in our gloss ary.Our dat as et re ve aled t hat t here are a lar g e number of species known to use to rpo r bu t a re lacking deta il s on m any a spects o f to rpo r use, including s eas ona lity, pre dictabi lity, an d flexi bility.Furt her more, our analyses were rest ricte d to spe cies that have been found to use to rpo r, which is in turn biased by ge og raphica l locat io n, level o f seaso nali ty, and taxo no my.The vast majo ri ty o f studies have examined to rp or in sp ecies s us pe cte d to u se it.Ba sed on our findings as well as a recen t syn thesis of the lit erature , it is m ore like ly that h om eoth ermy is the exception and heterot her my (inc luding t orpor use) is th e n orm, at least in sma l l m amm al s.We predict th at a s mor e data ar e obtained from fre e-rang ing spe cies, we are li kely to see m ore instan ces of o p por tunistic tor por as well as potent ia l ly more types of torpor.In addition to hig hlig hting th e n e e d to underta ke mor e r esear ch on t he se a son ality an d flexi b ili ty o f to rpo r use, our da ta em ph a size th at the to rpo r pattern s observ ed in the tropics and sub tro pics can no lon g er be considered exceptions to the hibernat ion ru les derive d fr om the Holar ct ic.O ur findings furt her sug gest t hat t h e n on-s eas onal, o p portunistic, and flexible use of torpor is like ly th e an cest ra l state and that t he se a son al, inflexible u se of torpor is a derived form of h eteroth ermy.

Fig. 1
Fig. 1 Global distribution of torpor patterns, based on the maximum duration of bouts.(A) Daily torpor ( n = 117); (B) Prolonged torpor ( n = 21); and (C) Hibernation ( n = 109).(D) Frequency distribution of torpor patterns (maximum duration) in relation to the four climate zones (tr opical: 0-23.5 • N/S, subtr opical: 23.5-40 • N/S, temperate: 40-60 • N/S, or polar: 60-90 • N/S).Black: data deficient ( n = 28); dark blue: hibernation; middle blue: prolonged torpor; and light blue: daily torpor.Sample sizes for the four climate zones are: polar ( n = 2), temperate ( n = 60), subtropical ( n = 118), and tropical ( n = 95).(E)Thephylogenetic relationships of species with known torpor use type.The tip labels r epr esent the longest torpor type observ ed in the species, using the same color scheme as in(D).Pie charts at the nodes w er e estimated using ancestral state estimation, and the proportion of each color r epr esents the likelihood that the shared ancestor used that particular torpor type.

Fig. 2
Fig. 2 (A) Global distribution of seasonality in torpor use and whether torpor is used seasonal (blue, n = 44/275) or non-seasonal (red, n = 70/275); data deficient is not depicted ( n = 161/275).(B) Frequency distribution of seasonality of torpor use based on climate zone.Blue: strictly seasonal use; red: torpor use not restricted to one season; black: data deficient.Sample sizes for the four climate zones (tropical: 0-23.5 • N/S, subtropical: 23.5-40 • N/S, temperate: 40-60 • N/S, or polar: 60-90 • N/S) are polar (2), temperate (60), subtropical (118), and tropical (95).(C) The phylogenetic relationships of species with data on torpor use over multiple seasons.Pie charts at the nodes w er e estimated using ancestral state estimation, and the proportion of each color r epr esents the likelihood that the shar ed ancestor used non-seasonal (red) or seasonal torpor (blue).
( Diedrich and Stein le chn er 2012 ; Przybyls ka-Piech an d Jefim ow 2022 ).Such induced to rpo r bou ts have been found to be s h orter an d less deep than s eas o nal to rpo r in wint er ( Diedric h and St ein le chner 2012 ) and are thus seen as a form of emer g ency respon se due t o deplet ed ener gy reserv es.In recent y ea rs, more a n d m ore in cidents o f to rpo r u se a s a fo rm o f emer g ency shutdown have b een rep ort ed , a nd in ma ny o f these species, to rpor is only used by a few indiv idu als of the po p ulatio n ( Jo n es an d G eis er 1992 ; C hrist ia n a nd G eis er 2007 ; Ba rk er et al .2012 ; Nowac k et al .2013 ; Nowac k et al .2015 ; Dau sm ann et al .2022 ; Nowac k and Turbi l l 2022 ).So me o f these species also s h o w surprisingly lo w rewa rming rates a n d re lative ly s ha l low de creases in b o dy tem pera ture ( Ba rk er et al .2012 ; Nowac k et al .2013 ; Nowack and Turbi l l 2022 ), which does no t sup port the view o f to rpo r as a highly co ntrolled mech ani sm where b o dy temp erature is re duce d con sidera b l y and entry a nd rewa rming fro m to rpo r are fast.The low rewarming rates have been att ribute d to deplete d ener gy reserv es ( Nowack et al. 2013 ; Nowack and Turbi l l 2022 ).These less cont rol le d to rpo r bou ts in respo nse to acu te energet ic bottlene cks may be very simi lar t o t o rpo r patterns in ea rly ma mmalia f o rms wi th individ ua ls li kely having relie d on b asking to rewarm fro m to rpo r as their thermogenic cap acit ies wou ld have be en low ( G eis er et al. 2017 ).Simi lar p attern s are observ ed durin g the dev elop ment o f endot her my in juvenile m amm al s.At first, they are unable to ma inta in a high b o dy temp erature dur ing night l y co ld exposure, bu t then grad ually incre ase t her ma l cap acity, which enables them to maintain h om eoth ermy for th e first half of the night before they a l low their b o dy temp eratur e to dr o p, b ut init ia l ly can only rewarm with th e h e l p o f an exter nal he at source ( Hi l l 1976 ; Wacker et a l .2017 ).Basking t o rewarm fro m to rpo r i s al so co mmo nly found in extant ad ul t m amm al s (e.g., Gei ser et a l.2002 ; Mzi li kazi et al. 2002 ; G eis er and Pavey 2007 ; Warnecke et al. 2008 ; Wa rneck e a nd G eis er 2010 ; G eis er et al .2016 ; Wac ker et al. 2017 ), and anim al s are able to crawl into the sun with b o dy temp eratures as low as 14.6 • C deg re es ( Wa rneck e et al. 2008 ).