Supplemen a y Ma e ials - Phillips e al., 2025
Table o con en s
Supplemen a y me hods - page numbe
1. Da a acquisi ion me hods - 2
Dissec ions - 2
Con as enhanced mic oCT - 2
Ai b ea hing as e idence o lungs - 2
Li e a u e sea ch - 3
Assignmen o ecomo phological ca ego ies - 3
2. Phylogeny me hods - 5
Figu e S1 - 6
3. Phylogene ic compa a i e me hods and implemen a ion
a. BayesT ai s ex ended me hods - 5
b. Full lis o BayesT ai s models es ed - 8
c. Bayes Fac o es ima ion and in e p e a ion - 11
d. Co HMM me hods - 11
e. S ochas ic cha ac e mapping and ances al s a e es ima ion - 12
. Phylogene ic gene alized linea models - 12
Supplemen a y Resul s - 13
1. Supplemen a y Table 1 (Table S1) Legend - 13
2. Full BayesT ai s model compa isons - 14
Table S2: summa y o all BayesT ai s models - 14
3. Visualiza ions o di e en e olu iona y his o ies o di e en BayesT ai s
models - 15 o 19
Figs. S2, S3, S4, S5, and S6
4. Full Co HMM ou pu s - 20 o 24
Table S3: summa y o Co HMM models
5. Phylogene ic gene alized linea model esul s - 25 o 26
Table S4: summa y o phyloglm models
Re e ences - 27 on
2
Supplemen a y Me hods
1 - Da a Acquisi ion Me hods
Dissec ions
Specimens o 315 species we e dissec ed unde a dissec ing scope while pinned o a silicone ma a he
bo om o a plas ic con aine ha was illed wi h 70% e hanol, 10% o malin, o wa e , depending on
cu a o ial p e e ences. To access he lungs, we cu in o he en al skin and pulled back he en al
muscula u e o e eal he spi al in es ine. I he lungs a e la ge, hey can some imes be isualized
immedia ely wi h no u he damage o he specimen (e.g., Fig. 1A). In o he cases, small lungs and
lung buds could only be isualized by emo ing he in es ine (e.g., Fig. 1F) and he smalles lung buds
equi ed ca e ul examina ion o he mos an e io sec ion o he gu ube (e.g., Fig. 1G). In a e cases (N
= 3), lung buds could no be eliably iden i ied a all. Some specimens examined by PHD we e
p e iously s ained wi h alcian blue solu ion o musculo-skele al s udies, wi h no damage o he lungs.
We sco ed adpole lungs as “absen ” i lungs could no be ound, i he lung buds we e solid and
ob iously non- unc ional, o i hey we e less han 5% he leng h o he abdominal ca i y (Figu e 1).
Con as enhanced mic o-compu ed omog aphy (CEμCT) scanning
We also ob ained lung p esence/absence da a om CEμCT scanning o 23 species. Specimens o hese
species we e s ained wi h ei he phospho ungs ic acid in 70% E hanol o an aqueous Lugol’s iodine
s ain and hen econs uc ed in o oxels o a a ie y o o he p ojec s. I lung issue could be clea ly
iden i ied in he scan, hen a judgmen o p esen (i.e., unc ional) s. absen (i.e., non unc ional) was
made. Howe e , i lung issue could no be iden i ied o he lungs appea ed ambiguous, we pe o med a
dissec ion o con i m lung s a us. Those species a e coun ed as “dissec ion” in Supplemen al Table 1,
which shows he e idence used o each axon included in he analyses.
Ai b ea hing as e idence o lungs
We also kep ack o ai b ea hing obse a ions as a way o es o lung p esence, which we
obse ed in 27 species. Howe e , in e e y case bu one (C uziohyla sil iae adpoles obse ed
3
b ea hing nea Gamboa, Panama by JRP), lung s a us was con i med ei he h ough dissec ions,
CEμCT scan, o li e a u e epo s.
Li e a u e sea ch e c.
In addi ion o ou p ima y da ase o dissec ions and CEμCT Scanning (338 species), lung
p esence/absence da a was ob ained om p e iously published s udies, books, and pho og aphs as well
as co espondences wi h expe s in speci ic axa. The p esence/absence o lungs was ou igh men ioned
in some cases, while in o he s he p esence o lungs was in e ed by he epo ed occu ence o
ai -b ea hing beha io . O he s udies made no ema k on lungs, bu p esen ed pho og aphs ha showed
he lungs ei he in dissec ion o in li e, as many adpoles a e highly anspa en , and he lungs can be
isualized om he ou side. We also used 25 pho og aphs o adpoles ha clea ly showed in la ed lungs
h ough he body wall as e idence o lung p esence. We we e able o con i m epo ed lung s a us in he
li e a u e wi h dissec ions o 23 axa and an addi ional species ia CEμCT.. Supplemen al able 1
includes he sou ces, e idence ype used, and sample sizes o all axa examined.
We sampled as widely as possible, p io i izing he inclusion o as many amilies and gene a as
possible. In o al, we we e able o sample 252 o 376 gene a wi h ee-li ing la ae and 44 o 47
possible amilies. The og amilies B achycephalidae, B e icipi idae, Caligoph ynidae,
Ce a oba achidae, Ceu homan idae, C augas o idae, Eleu he odac ylidae, Neblinaph ynidae, and
S aboman idae a e so-called “di ec de elope s” and lack ee-li ing la ae, meaning hey could no be
included in his p ojec . Supplemen al Fig. 1 shows he dis ibu ion o sampled axa wi hin a sample o
5,242 anu ans wi h non- iable axa (gene a wi h no known ee-li ing s age) shown in o ange.
Assignmen o ecomo phological ca ego ies
We assigned adpoles o a se o o e lapping ca ego ies based on he gene al ecomo phological guilds
o (Al ig and Johns on 1986). We i s spli adpoles in o aqua ic and e es ial. Mos ee-li ing
4
adpoles a e aqua ic, bu some a e ei he ully e es ial o “semi- e es ial”. Fully e es ial adpoles
a e o en endo ophic (no ee- eeding), being pa en -associa ed and/o li ing in nes s (e.g., Sooglossus
and Alloba es nidicola). Semi- e es ial adpoles a e conside ed a lo ic guild and consis o adpoles
ha li e p ima ily on we ock aces alongside s eams (56). We cha ac e ized all o hese adpoles as
“ e es ial”, and all o he adpole o ms as “aqua ic”.
We nex spli adpoles in o len ic and lo ic ca ego ies. We sco ed b eeding loca ion as len ic
when he desc ip ion e e ed o a pond, pool, puddle o simila wa e body, and as lo ic when he
desc ip ion e e ed o a s eam, o en , apid, o ocky su ace associa ed wi h such en i onmen s. We
sco ed species ha we e epo ed o b eed in bo h habi a ypes (N=5) as len ic, because hei adpoles
a e gene ally mo e simila o len ic adpoles (Al ig and Johns on 1989). “Lo ic, semi- e es ial”
adpoles we e coun ed as lo ic o ou -s a e analyses and as e es ial in eigh -s a e analyses.
Wi hin lo ic adpoles, we u he ca ego ized adpoles as ei he “specialized” o “o he lo ic”. To
cla i y, specialized is a sho hand o wo speci ic ypes o specializa ion, he e a e many highly
specialized, lo ic adpoles ha would be included as “o he ” in ou analysis. We included adpoles as
specialized in cases whe e hey ei he had some o m o suc ion appa a us (suc o ial) o we e
specialized o bu owing ( osso ial). Suc o ial adpoles use suc ion o a ach o ocks in as - lowing
o en s, including he p e iously de ined guilds: lo ic, suc o ial (mou h-sucke s) and he biza e lo ic,
gas omyzopho ous guild, which use a belly-sucke (78, 79). Fosso ial adpoles a e long and wo m-like,
li ing unde and wi hin dense subs a es like sand, g a el, and lea ma s (80, 81). We included he
ecomo phological guilds “lo ic, osso ial” and “lo ic, psammonic” (sand- eeding) unde he single
umb ella o “ osso ial” as hese g oups all li e bu ied unde solid subs a a. We conside ed adpoles
specialis s i hey had been placed in any o he abo e guilds by (23) o o he s. In a e cases whe e a
adpole had no been placed in any guild, we assigned i ou sel es (suppl. Table 1), based on ecological
and mo phological e idence. All o he lo ic guilds we e sco ed as “o he ”.
5
2 - Phylogeny Me hods
We p io i ized including as many sampled axa in ou phylogene ic analyses as possible (Suppl. Fig. 1).
In cases whe e we sampled axa o lung da a and he axon was no p esen in (Po ik e al. 2023), we
a emp ed o place hose axa in he ee by subs i u ing o a p esen axon we did no sample. We
accep ed a subs i u ion o a sampled axon o one p esen in he ee unde h ee scena ios. Fi s , i he
sampled axon ha is missing in he ee is he only membe o i s genus sampled, we assumed gene ic
monophyly and subs i u ed he sampled axon o a andom membe o he genus p esen in he ee.
Second, i he sampled axon has been analyzed phylogene ically by ano he s udy such ha i could be
aligned wi h a axon p esen in he ee lacking lung da a, we subs i u ed he sampled species o he
species in he ee. Thi d, i we sampled wo membe s o a Genus, and only one was p esen in he ee,
we andomly chose a hi d, p esen species in he genus o subs i u e o he sampled species. This hi d
jus i ica ion assumes gene ic monophyly, and p oduces ees wi h easonable bi u ca ing ela ionships,
bu po en ially al e s “ ue” b anch leng hs. We belie e he alue o adding addi ional species o he
analysis ou weighs he po en ial nega i e e ec s o inco ec b anch leng hs, especially because bo h
ecology and lung s a us a e o en uni o m wi hin gene a, and he e is al eady unce ain y in b anch
leng h es ima es. Be o e making any subs i u ions, we i s checked whe he each genus was
monophyle ic in he published ee, and i no , we did no assume monophyly. See supplemen a y Table
1 o a ull lis o which axa we e subs i u ed o each ee, wi h ci a ions.
3 - Phylogene ic Compa a i e Me hods implemen a ion
A - BayesT ai s Ex ended Me hods
We used he p og am BayesT ai s (BT) (V.5.0 windows e sion) (Pagel 1994; Pagel and Meade 2006,
n.d.) o es ima e and es Mk models. We used a cus om-buil R amewo k o analyze and isualize
BayesT ai s ou pu s. Code used o p epa e BayesT ai s da ase s and analyze he ou pu s a e a ailable a
h ps://gi hub.com/phillipsja /Phillips_e al_2025_ adpole_lung_loss/ and BayesT ai s command iles
and ac ual ou pu s a e a ailable as supplemen al iles he e: h ps://doi.o g/10.5061/d yad.3 x95x6sk. Q
ma ices a e included below.
6
Fo all analyses, we used he Bayesian mode wi h a jMCMC sample wi h a uni o m p io
om 0 o 100 o all po en ially ee pa ame e s. Following he BT manual, we scaled ees down
because he b anch leng hs in (Po ik e al. 2023) co espond o millions o yea s a he han base pai
subs i u ions, which is he assumed scale o BT. We scaled ees by unning se s o ull models wi hou
a scaling ac o , and hen scaling he ee by h ee ac o s o en (.1, .01, .001). We chose he scale ha
esul ed in he highes pa ame e s being es ima ed abo e 10 and no leading o a unca ion a he uppe
end o ou p io s (100). Scaling p opo ionally inc eases mean pa ame e es ima es, p e en ing BT om
ailing o dis inguish low a es wi h ze o. The BT manual (p. 10) ecommends scaling b anch leng hs o
a mean o .1 o a oid his p oblem, bu we p e e ed o scale in en ionally, such ha a es can easily be
7
back- ans o med la e on. We scaled he 4-s a e and 8-s a e models independen ly, bu any models
compa ed wi h 2logBayesFac o s we e scaled o he same amoun . A e es ing di e en scales, we
chose o scale he ou -s a e models by a ac o o 0.001 and he eigh -s a e models by a ac o o 0.01.
Fo each un o he 4-s a e models, we used 2,500,000 i e a ions pe un wi h a bu nin o 500,000 and
h ee uns pe model. Fo each un o he eigh -s a e models, we used 110,000,000 i e a ions and a
bu nin o 20,000,000 i e a ions. Eigh -s a e models we e much mo e complex, and so slowe o un and
con e ge, so we g ea ly inc eased bu nin and i e a ions.
We assessed wi hin- un con e gence by examining ace plo s, and ac oss- un con e gence by
compa ing es ima ed ma ginal likelihoods and isually compa ing iolin plo s o pos e io da a. We
selec ed models o isualiza ion ha con e ged bo h wi hin and ac oss uns, p esen ing he bes
indi idual un (as de e mined by he highes es ima ed log ma ginal likelihood) in Fig. 4. We calcula ed
highes densi y in e als as c edible in e als o all es ima ed pa ame e s using he R package
HDIn e al, .0.2.4 (Me edi h and K uschke 2020).
The mo e complica ed, eigh -s a e models o cha ac e e olu ion ha we used equi ed some
ossiliza ions o eliably p oduce easonable es ima es o cha ac e e olu ion. We ound ha when
un es ic ed, he eigh -s a e models o en p oduced solu ions ha econs uc ed he ances o s o mos
clades as being e es ial. Because bo h specialized lo ic and e es ial adpoles a e associa ed wi h
speci ic la al apomo phies no ound ac oss la ge pa s o he anu an ee, we uled his solu ion o no
be biologically easible, and so es ic ed se e al key nodes o he ee o be ei he len ic o gene alized
lo ic, wi h equal p obabili ies o being lunged o lungless in ei he s a e. We did his o he mos ecen
common ances o (MRCA) o all ogs, he MRCA o Neoba achia, he MRCA o he Dend oba idae
amily, he MRCA o he Lep odac ylidae amily, and he MRCA o he Cyclo hamphidae and
Alsodidae. The only e ec o hese ossiliza ions was o pola ize ansi ions om aqua ic adpoles o
e es ial adpoles, an assump ion s ongly suppo ed by biological in e ence.
8
B - Full lis o models es ed and Q ma ices
We compa ed a se o po en ially easonable e olu iona y models o lung e olu ion in BT o bo h he
ou -s a e and eigh -s a e analyses. Among he ou -s a e analyses, we began wi h a ull dependen
model, which included eigh po en ially unique pa ame e s (lung loss and lung gain in bo h len ic and
lo ic habi a s, and ansi ions be ween len ic and lo ic habi a s in bo h he lunged and lungless
condi ions). We also es ed a se ies o nes ed, mo e es ic ed e sions his ull model, including a
model in which len ic lung loss was es ic ed o ze o, lung egains in len ic and lo ic habi a s we e se
equal, lung egains we e se o ze o, and bo h lung egains and len ic lung loss we e se o ze o. We also
es ed an “independen ” model in which lung e olu ion in any di ec ion was se equal o bo h habi a
ypes and habi a e olu ion in ei he di ec ion was se equal ega dless o lung s a us, leading o only
ou po en ially unique pa ame e s.
Among he eigh -s a e models, we again es ed a “ ull” model, which included up o 28 unique
pa ame e s. This model included ou a es o lung loss ( o each habi a classi ica ion: len ic, lo ic
non-specialized, lo ic specialized, and e es ial), an addi ional ou a es o lung egain in each habi a ,
and hen a se ies o po en ial ansi ions be ween habi a ypes o bo h he lunged and lungless s a uses.
We es ic ed habi a ansi ions be ween len ic and specialized lo ic, belie ing a mo e ealis ic scena io
equi es an e olu iona y ansi ion in o lo ic non-specialized. Howe e , we did allow ansi ions om all
len ic and lo ic habi a s di ec ly o e es ial. We also es ed a se o mo e es ic ed eigh -s a e models,
including a model in which lo ic, non-specialized loss was se o ze o, a model o which len ic lung
loss was se o ze o, one model wi h all lung egains we e se o ze o, ano he wi h lung egains all se o
a single pa ame e alue, ano he wi h len ic lung egains se o ze o, and a inal model wi h all egains
se o ze o and non-specialized lo ic lung loss se o ze o.
9
Q ma ices
Fou s a e:
Independen Model
1 - lungless lo ic
2 - lunged lo ic
3 - lungless len ic
4 - lunged len ic
1
-
𝛼2
(gain lungs)
𝛼1
(lo ic o len ic)
-
2
𝛽2
(lose lungs)
-
-
𝛼1
(lo ic o len ic)
3
𝛽1
(len ic o lo ic)
-
-
𝛼2
(gain lungs)
4
-
𝛽1
(len ic o lo ic)
𝛽2
(lose lungs)
-
Dependen Model
1 - lungless lo ic
2 - lunged lo ic
3 - lungless len ic
4 - lunged len ic
1
-
q12: gain lungs
(lo ic)
q13: lo ic o len ic
(lungs)
-
2
q21: lose lungs
in lo ic
-
-
q24: lo ic o len ic
(no lungs)
3
q31: len ic o
lo ic (no lungs)
-
-
q34: gain lungs
(len ic)
4
-
q42: len ic o lo ic
(lungs)
q43: lose lungs
(len ic)
-
Fou -s a e model es ic ions
Dependen : no es ic ions
Regains se equal: q12 = q34
Regains p e en ed: q12 = q34 = 0
No len ic loss: q43 = 0
No egains and no len ic loss q12 = q34 = q43 = 0
Independen q12 = q34 ⇒ 𝛼1; q21 = q43 ⇒ 𝛽2; q13 = q24 ⇒ 𝛼1; q31 = q42 = 𝛽1
16
Supplemen al Figu e 3: 100 blended s ochas ic cha ac e maps o he ull ou -s a e dependen
e olu iona y model wi h ances al s a e es ima es, whe e a es o lung and habi a e olu ion a y
depending on he s a e o he o he ai . No e ci cled nodes, which di e be ween independen and
dependen models. Q ma ix om median alues o pa ame e es ima es in he bes indi idual un o he
ull dependen model ( un 2 in Supp. Table 2) and ee is a p une maximum likelihood phylogeny om
(Po ik e al. 2023).
17
Supplemen al Figu e 4: 100 blended s ochas ic cha ac e maps o he bes ou -s a e dependen
e olu iona y model wi h ances al s a e es ima es, whe e a es o lung and habi a e olu ion a y
depending on he s a e o he o he ai and a es o egain and lung loss in len ic habi a s a e se o
ze o. No e ci cled nodes, which di e be ween independen and dependen models. Q ma ix om
median alues o pa ame e es ima es in he bes indi idual un o he bes dependen model ( un 1 in
Supp. Table 2) and ee is a p une maximum likelihood phylogeny om (Po ik e al. 2023).
18
Supplemen al Figu e 5: 100 blended s ochas ic cha ac e maps o he ull eigh -s a e e olu iona y
model (wi h ances al s a e es ima es), which includes e es ial adpoles as well as dis inguishes
be ween lo ic adpoles specialized o suc o iali y and osso iali y om “o he ” lo ic adpoles. Q ma ix
om median alues o pa ame e es ima es in he bes un o he ull eigh s a e model ( un 2 in Supp.
Table 2) and ee is a p une maximum likelihood phylogeny om (Po ik e al. 2023).
19
Supplemen al Figu e 6: 100 blended s ochas ic cha ac e maps o he bes eigh -s a e e olu iona y
model wi h ances al s a e es ima es, whe e lung egains a e impossible and lung loss canno occu in
lo ic adpoles no specialized o suc o iali y o osso iali y. Q ma ix om median alues o pa ame e
es ima es in he bes un o he ull eigh s a e model ( un 1 in Supp. Table 2) and ee is a p une
maximum likelihood phylogeny om (Po ik e al. 2023).
20
4 - Full Co HMM Ou pu s om R
Ra es a e unscaled, so based on he scale o he ee’s b anch leng hs, which a e in millions o yea s.
The e o e, a es e e o expec ed numbe o changes pe million yea s.
Example Q ma ices wi h explana o y a es:
Non-Hidden Ma ko Model (only 1 se o a es ac oss he en i e ee):
lungless len ic
lunged len ic
lungless lo ic
lunged lo ic
(1,R1)
(2,R1)
(3,R1)
(4,R1)
1,R1
-
gain lungs
(len ic)
len ic o lo ic
(no lungs)
-
2,R1
lose lungs
(len ic)
-
-
len ic o lo ic
(lungs)
3,R1
lo ic o len ic
(no lungs)
-
-
gain lungs
(lo ic)
4,R1
-
lo ic o len ic
(lungs)
lose lungs
(lo ic)
-
Hidden Ma ko Model (2 possible a e ma ices ac oss he ee, each se o ansi ion a es colo ed):
lungless
len ic
lunged
len ic
lungless
lo ic
lunged lo ic
lungless
len ic
lunged
len ic
lungless
lo ic
lunged lo ic
(1,R1)
(2,R1)
(3,R1)
(4,R1)
(1,R2)
(2,R2)
(3,R2)
(4,R2)
1,R1
-
gain lungs
(len ic)
len ic o lo ic
(no lungs)
-
Ra e 1 o
Ra e 2
-
-
-
2,R1
lose lungs
(len ic)
-
-
len ic o
lo ic (lungs)
-
Ra e 1 o
Ra e 2
-
-
3,R1
lo ic o len ic
(no lungs)
-
-
gain lungs
(lo ic)
-
-
Ra e 1 o
Ra e 2
-
4,R1
-
lo ic o len ic
(lungs)
lose lungs
(lo ic)
-
-
-
-
Ra e 1 o
Ra e 2
1,R2
Ra e 2 o
Ra e 1
-
-
-
-
gain lungs
(len ic)
len ic o lo ic
(no lungs)
-
2,R2
-
Ra e 2 o
Ra e 1
-
-
lose
lungs
(len ic)
-
-
len ic o lo ic
(lungs)
3,R2
-
-
Ra e 2 o
Ra e 1
-
lo ic o
len ic (no
lungs)
-
-
gain lungs
(lo ic)
4,R2
-
-
-
Ra e 2 o
Ra e 1
-
lo ic o
len ic
(lungs)
lose lungs
(lo ic)
-
21
Indi idual ou pu s o each es ed model, wi h es ima es ounded o ou decimal places.
simpli ied_independen _model_ i
Fi
-lnL AIC AICc Ra e.ca n ax
-367.8601 739.7201 739.7444 1 497
Ra es
lungless len ic
lunged len ic
lungless lo ic
lunged lo ic
(1,R1)
(2,R1)
(3,R1)
(4,R1)
1,R1
-
0.0066
0.0066
-
2,R1
0.0024
-
-
0.0066
3,R1
0.0024
-
-
0.0066
4,R1
-
0.0024
0.0024
-
simpli ied_hidden_Ma ko _independen _model_ i
Fi
-lnL AIC AICc Ra e.ca n ax
-352.5233 717.0465 717.218 2 497
Ra e
s
lungless
len ic
lunged
len ic
lungless
lo ic
lunged
lo ic
lungless
len ic
lunged
len ic
lungless
lo ic
lunged
lo ic
(1,R1)
(2,R1)
(3,R1)
(4,R1)
(1,R2)
(2,R2)
(3,R2)
(4,R2)
1,R1
-
0.0431
0.0431
-
0.0357
-
-
-
2,R1
0.0109
-
-
0.0431
-
0.0357
-
-
3,R1
0.0109
-
-
0.0431
-
-
0.0357
-
4,R1
-
0.0109
0.0109
-
-
-
-
0.0357
1,R2
0.0093
-
-
-
-
0.0014
0.0014
-
2,R2
-
0.0093
-
-
0.0002
-
-
0.0014
3,R2
-
-
0.0093
-
0.0002
-
-
0.0014
4,R2
-
-
-
0.0093
-
0.0002
0.0002
-
simpli ied_co ela ed_model_ i
Fi
-lnL AIC AICc Ra e.ca n ax
-338.0255 684.0511 684.1324 1 497
Ra es
lungless len ic
lunged len ic
lungless lo ic
lunged lo ic
(1,R1)
(2,R1)
(3,R1)
(4,R1)
1,R1
-
0.0001
0.0112
-
2,R1
0.0001
-
-
0.0077
3,R1
0.0112
-
-
0.0069
4,R1
-
0.0077
0.0069
-
22
simpli ied_hidden_Ma ko _co ela ed_model_ i - bes model (lo ic lung loss bolded, len ic lung loss in i alics)
Fi
-lnL AIC AICc Ra e.ca n ax
-323.3033 666.6067 667.0594 2 497
Ra e
s
lungless
len ic
lunged
len ic
lungless
lo ic
lunged
lo ic
lungless
len ic
lunged
len ic
lungless
lo ic
lunged
lo ic
(1,R1)
(2,R1)
(3,R1)
(4,R1)
(1,R2)
(2,R2)
(3,R2)
(4,R2)
1,R1
-
1.00E-09
0.0278
-
0.0376
-
-
-
2,R1
1.00E-09
-
-
13.7856
-
0.0376
-
-
3,R1
0.0278
-
-
0.0041
-
-
0.0376
-
4,R1
-
13.7856
0.0041
-
-
-
-
0.0376
1,R2
0.0058
-
-
-
-
1.00E-09
0.0032
-
2,R2
-
0.0058
-
-
1.00E-09
-
-
1.00E-09
3,R2
-
-
0.0058
-
0.0032
-
-
0.0087
4,R2
-
-
-
0.0058
-
1.00E-09
0.0087
-
independen _model_ i
Fi
-lnL AIC AICc Ra e.ca n ax
-353.0246 714.0492 714.1305 1 497
Ra es
lungless len ic
lunged len ic
lungless lo ic
lunged lo ic
(1,R1)
(2,R1)
(3,R1)
(4,R1)
1,R1
-
0.0007
0.0079
-
2,R1
0.0017
-
-
0.0079
3,R1
0.0118
-
-
0.0007
4,R1
-
0.0118
0.0017
-
23
hidden_Ma ko _independen _model_ i
Fi
-lnL AIC AICc Ra e.ca n ax
-326.5059 673.0119 673.4645 2 497
Ra e
s
lungless
len ic
lunged
len ic
lungless
lo ic
lunged
lo ic
lungless
len ic
lunged
len ic
lungless
lo ic
lunged
lo ic
(1,R1)
(2,R1)
(3,R1)
(4,R1)
(1,R2)
(2,R2)
(3,R2)
(4,R2)
1,R1
-
0.0012
0.1349
-
0.0234
-
-
-
2,R1
0.008
-
-
0.1349
-
0.0234
-
-
3,R1
1.00E-09
-
-
0.0012
-
-
0.0234
-
4,R1
-
1.00E-09
0.008
-
-
-
-
0.0234
1,R2
0.006
-
-
-
-
0.0005
1.00E-09
-
2,R2
-
0.006
-
-
1.00E-09
-
-
1.00E-09
3,R2
-
-
0.006
-
0.0526
-
-
0.0005
4,R2
-
-
-
0.006
-
0.0526
1.00E-09
-
co ela ed_model_ i
Fi
-lnL AIC AICc Ra e.ca n ax
-331.4721 678.9441 679.2392 1 497
Ra es
lungless len ic
lunged len ic
lungless lo ic
lunged lo ic
(1,R1)
(2,R1)
(3,R1)
(4,R1)
1,R1
-
1.00E-09
0.0127
-
2,R1
0.0004
-
-
0.0068
3,R1
1.00E-09
-
-
0.0022
4,R1
-
0.0158
0.0056
-
24
hidden_Ma ko _co ela ed_model_ i
Fi
-lnL AIC AICc Ra e.ca n ax
-322.9828 681.9656 683.3966 2 497
Ra e
s
lungless
len ic
lunged
len ic
lungless
lo ic
lunged
lo ic
lungless
len ic
lunged
len ic
lungless
lo ic
lunged
lo ic
(1,R1)
(2,R1)
(3,R1)
(4,R1)
(1,R2)
(2,R2)
(3,R2)
(4,R2)
1,R1
-
1.00E-09
0.0141
-
0.0147
-
-
-
2,R1
1.00E-09
-
-
0.0154
-
0.0147
-
-
3,R1
1.00E-09
-
-
0.0126
-
-
0.0147
-
4,R1
-
0.0455
0.0034
-
-
-
-
0.0147
1,R2
0.0046
-
-
-
-
1.00E-09
0.0151
-
2,R2
-
0.0046
-
-
1.00E-09
-
-
0.0009
3,R2
-
-
0.0046
-
0.0155
-
-
1.00E-09
4,R2
-
-
-
0.0046
-
1.00E-09
0.0096
-
Supplemen a y Table 3: summa y o all Co HMM models
Co HMM Model
AIC
del aAIC
numbe o
a es
Independen
714.05
47.44
4
indep_HMM
673.01
6.40
10
Dependen
678.94
12.33
8
dep_HMM
681.97
15.36
18
indep_simpli ied
739.72
73.11
2
indep_HMM_simpli ied
717.05
50.44
6
dep_simpli ied
684.05
17.44
4
dep_HMM_simpli ied
666.61
0
10
25
5 - Phylogene ic gene alized linea model esul s
Full ou pu s om R, wi h coe icien s combined in o a single able below
Model 1: co ela ion o b eeding habi a (len ic/lo ic) wi h lung p esence and absence
phyloglm( o mula = lung ~ ecology, da a = da a_aqua, phy = ull_ ee,
me hod = c("logis ic_MPLE"), s a .be a = NULL, s a .alpha = NULL,
boo = 1000, ull.ma ix = TRUE)
AIC logLik Pen.logLik
204.75 -99.37 -97.86
Me hod: logis ic_MPLE
Mean ip heigh : 179.3411
Pa ame e es ima e(s):
alpha: 0.001150445
boo s ap mean: 0.0008360564 (on log scale, hen back ans o med)
so possible downwa d bias.
boo s ap 95% CI: (0.0001061738,0.006334944)
Model 2: co ela ion o ecomo phological guild (specialized lo ic guilds/o he ypes) wi h
lung p esence and absence
phyloglm( o mula = lung ~ Spec_lo ic, da a = da a_aqua, phy = ull_ ee,
me hod = c("logis ic_MPLE"), s a .be a = NULL, s a .alpha = NULL,
boo = 1000, ull.ma ix = TRUE)
AIC logLik Pen.logLik
145.54 -69.77 -69.08
Me hod: logis ic_MPLE
Mean ip heigh : 179.3411
Pa ame e es ima e(s):
alpha: 0.0002628877
boo s ap mean: 0.0002568441 (on log scale, hen back ans o med)
so possible downwa d bias.
boo s ap 95% CI: (0.000102581,0.002927915)
Model 3: ull model es ing he co ela ion o b eeding habi a , ecomo phological guild, and
e es iali y wi h lung p esence and absence
phyloglm( o mula = lung ~ ecology + Spec_lo ic + e es ial,
da a = da a_ ull, phy = ull_ ee, me hod = c("logis ic_MPLE"),
s a .be a = NULL, s a .alpha = NULL, boo = 1000, ull.ma ix = TRUE)
AIC logLik Pen.logLik
160.46 -75.23 -71.99
Me hod: logis ic_MPLE
