This is a sel -a chi ed e sion o an o iginal a icle. This e sion
may di e om he o iginal in pagina ion and ypog aphic de ails.
Au ho (s):
Ti le:
Yea :
Ve sion:
Copy igh :
Righ s:
Righ s u l:
Please ci e he o iginal e sion:
CC BY 4.0
h ps://c ea i ecommons.o g/licenses/by/4.0/
Pool choice in a e ical landscape : Tadpole‐ ea ing si e lexibili y in phy o elm‐
b eeding ogs
© 2021 he Au ho s
Published e sion
Fouilloux, Chloe A.; Se ano Rojas, Shi ley Jenni e ; Ca ajal‐Cas o, Juan Da id;
Valkonen, Janne K.; Gauche , Philippe; Fische , Ma ie‐The ese; Pašukonis,
And ius; Rojas, Bibiana
Fouilloux, C. A., Se ano Rojas, S. J., Ca ajal‐Cas o, J. D., Valkonen, J. K., Gauche , P., Fische ,
M., Pašukonis, A., & Rojas, B. (2021). Pool choice in a e ical landscape : Tadpole‐ ea ing si e
lexibili y in phy o elm‐b eeding ogs. Ecology and E olu ion, 11(13), 9021-9038.
h ps://doi.o g/10.1002/ece3.7741
2021
Ecology and E olu ion. 2021;00:1–18.
|
1www.ecole ol.o g
Recei ed: 19 Ma ch 2021
|
Re ised: 6 May 2021
|
Accep ed: 12 May 2021
DOI: 10.1002/ece3.7741
ORIGINAL RESEARCH
Pool choice in a e ical landscape: Tadpole- ea ing si e
lexibili y in phy o elm- b eeding ogs
Chloe A. Fouilloux1 | Shi ley Jenni e Se ano Rojas2 |
Juan Da id Ca ajal- Cas o3,4  | Janne K. Valkonen1 | Philippe Gauche 5 |
Ma ie- The ese Fische 2 | And ius Pašukonis2,6  | Bibiana Rojas1
1Depa men o Biological and En i onmen al Science, Uni e si y o Jy äskylä, Jy äskylä, Finland
2Depa men o Biology, S an o d Uni e si y, S an o d, CA, USA
3Ins i u o de In es igación de Recu sos Biológicos Alexande on Humbold , Bogo á, Colombia
4Depa men o Biological Sciences, S . John’s Uni e si y, Queens, NY, USA
5USR LEEISA— Labo a oi e Ecologie, E olu ion, In e ac ions des Sys èmes Amazoniens, CNRS- Guyane, Cayenne, F ench Guiana
6Cen e d’Ecologie Fonc ionelle e E olu i e, CNRS, Mon pellie Cedex 5, F ance
This is an open access a icle unde he e ms o he C ea i e Commons A ibu ion License, which pe mi s use, dis ibu ion and ep oduc ion in any medium,
p o ided he o iginal wo k is p ope ly ci ed.
© 2021 The Au ho s. Ecology and E olu ion published by John Wiley & Sons L d.
And ius Pašukonis and Bibiana Rojas a e senio au ho s and con ibu ed equally o his wo k.
Co espondence
And ius Pašukonis, Depa men o Biology,
S an o d Uni e si y, 371 Jane S an o d Way,
S an o d, CA 94305, USA.
Email: apasukonis@s an o d.edu
Bibiana Rojas, Depa men o Biology and
En i onmen al Sciences, Uni e si y o
Jy äskylä, PO Box 35, Jy äskylä, FI 40001,
Finland.
Email: bibiana.[email p o ec ed]
Chloe Fouilloux, Depa men o Biology
and En i onmen al Sciences, Uni e si y o
Jy äskylä, PO Box 35, Jy äskylä, FI 40001,
Finland.
Email: chloe.a. ouilloux@jyu. i
Funding in o ma ion
This p ojec was pa ially unded by he
In es issemen d’A eni unds o he
ANR (AnaEE F ance ANR- 11- INBS- 0001;
Labex CEBA ANR- 10- LABX- 25- 01) in he
amewo k o he Nou agues T a el G an
g an ed o BR, AP, SJSR, and JDCC. BR,
JV, and CF a e unded by he Academy o
Finland (Academy Resea ch Fellowship
o BR, P ojec No. 21000042021). AP is
suppo ed by he Eu opean Union's Ho izon
2020 esea ch and inno a ion p og am
unde he Ma ie Sklodowska- Cu ie g an
ag eemen no. 835530. AP, SJSR, and MTF
we e also suppo ed by Lau en A. O’Connell
wi h S an o d Uni e si y and he Na ional
Science Founda ion (IOS- 1845651) unds.
Abs ac
Many species o Neo opical ogs ha e e ol ed o deposi hei adpoles in small
wa e bodies inside plan s uc u es called phy o elma a. These pools a e small
enough o exclude la ge p eda o s bu ha e limi ed nu ien s and high desicca-
ion isk. He e, we explo e phy o elm use by h ee common Neo opical species:
Os eocephalus oophagus, an a bo eal og ha pe iodically eeds eggs o i s adpoles;
Dend oba es inc o ius, a adpole- anspo ing poison og wi h cannibalis ic adpoles;
and Alloba es emo alis, a e es ial adpole- anspo ing poison og wi h omni o-
ous adpoles. We ound ha D. inc o ius occupies pools ac oss he chemical and
e ical g adien , whe eas A. emo alis and O. oophagus appea o ha e na owe dep-
osi ion op ions ha a e es ic ed p ima ily by pool heigh , wa e capaci y, alkalini y,
and salini y. Dend oba es inc o ius adpoles a e pa icula ly lexible and can su i e
in a wide ange o chemical, physical, and biological condi ions, whe eas O. oophagus
seems o p e e small, clea pools and A. emo alis occupies medium- sized pools wi h
abundan lea li e and low salini y. Toge he , hese esul s show he possible niche
pa i ioning o phy o elma a among ogs and p o ide insigh in o s esso s and esil-
ience o phy o elm b eede s.
KEYWORDS
compe i ion, niche pa i ioning, pa en al ca e, phy o elma a, poison ogs, adpoles
2
|
FOUILLOUX e aL.
1 | INTRODUCTION
The su i al o young o en hinges on he quali y o he ea ing en i-
onmen s c ea ed o chosen by hei pa en s. Whe he i is by build-
ing nes s (bi ds: B own & B own, 1991; mice: Bul & Lynch, 1997,
Zhao e al., 2016), digging bu ows ( oden s: Ebenspe ge e al., 2014;
S endsen, 1976), o deposi ing clu ches/la ae (e.g., salamande s:
Ruano- Faja do e al., 2014, ogs: Pe i e al., 2018), he ecology o
ea ing si es is undamen al in shaping o sp ing success. Fo animals
wi h ex e nal e iliza ion, b eeding si e choice can be especially
impo an , as op imal condi ions o egg clu ches may di e om
he op imal en i onmen o ha chlings and adul s ( ish: O esen &
Bolla, 1998, Mikhee e al., 2001; salamande s: Nussbaum, 1987,
Sih & Moo e, 1993; ogs: Vági e al., 2019). Many o hese animals
assess and p e e bio ic and abio ic p ope ies o b eeding si es
ha can enhance o sp ing su i al (B own & Shine, 2005; Ma sh
& Bo ell, 2001; Mokany & Shine, 2003; Touchon & Wo ley, 2015).
Thus, cha ac e izing he nu se ies whe e o sp ing occu and whe e
hey do no can p o ide in o ma ion on he quali ies pa en s assess
when making hese c i ical ep oduc i e decisions.
The challenge o inding an op imal ea ing si e becomes espe-
cially appa en in e es ial o a bo eal b eeding animals, whose
la al o ms a e aqua ic. Fo example, some ee ogs lay clu ches
o e hanging wa e bodies. The placemen o clu ches is essen ial,
as adpoles om poo ly placed clu ches isk ha ching and alling
on o he g ound (Wa ken in, 2011; Wells, 2007). One ema kable
amphibian s a egy adap ed o changing habi a s be ween egg and
la al s ages in ol es pa en s ha physically anspo ecen ly
ha ched adpoles om e es ial o iposi ion si es o small wa e -
holding plan s uc u es (phy o elma a), ponds, o s eams (Schul e
e al., 2020; Summe s & Tumul y, 2014). Unlike o he e es ial
b eeding amphibians, he physical anspo o young allows pa -
en s o selec he ideal en i onmen o hei o sp ing o de elop.
Al hough i is di icul o ex ensi ely cha ac e ize s eams and ponds
due o hei la ge size and in e connec edness wi h o he wa e bod-
ies, mic ohabi a s like phy o elma a p o ide a unique oppo uni y
o ully measu e he biological, chemical, and physical aspec s o a
nu se y, c ea ing an oppo uni y o in e p e deposi ion choices wi h
a dep h o ecological in o ma ion ha is a ely a ailable o o he
ea ing si es. He e, we in es iga e he chemical and physical p op-
e ies o aqua ic nu se ies ha p edic he p esence o Neo opical
adpoles in phy o elm- b eeding ogs.
The use o phy o elma a as adpole nu se ies can seem coun-
e in ui i e, as hei small olume makes hem p one o desicca-
ion and limi ed in ood (Summe s & McKeon, 2004; Summe s
& Tumul y, 2014). Howe e , hei small size p o ides p o ec ion
om la ge p eda o s and o e all educed in e speci ic compe i-
ion (Ki ching, 2001; Summe s & Tumul y, 2014). Va ious species
ha e e ol ed di e en s a egies o hei o sp ing o succeed in
hese pools (subs a e specializa ion: on May e al., 2009; Pe i
e al., 2018; ophic egg eeding: B own e al., 2010; Weygold , 1980;
la al agg ession/cannibalism: G ay e al., 2009; Poelman &
Dicke, 2007; Rojas, 2014; pool choice based on speci ic physical
o chemical cues: Lin e al., 2008; Schul e e al., 2011). Despi e he
widesp ead use o phy o elma a (Leh inen, 2021), and he non-
andom si e selec ion shown by many og pa en s, ew s udies go
beyond quan i ying basic pool dimensions and pool occupa ion o
unde s and adpole deposi ion decisions. Fu he , he bulk o phy-
o elm s udies a e ocused only on b omeliads (Mageski e al., 2016;
Pe i e al., 2018; Ruano- Faja do e al., 2014), while wo k explo -
ing po en ial ade- o s associa ed be ween di e en phy o elma a
(i.e., physical and chemical p ope ies as well as ood- and p eda o -
ela ed p essu es), and how hese change ac oss a e ical g adien ,
has gone la gely o e looked (bu see B own e al., 2008a).
To unde s and wha a iables d i e phy o elm selec ion, we
compa ed pool occupa ion by h ee Neo opical ogs (Dend oba es
inc o ius (Dend oba idae), Alloba es emo alis (A omoba idae), and
Os eocephalus oophagus (Hylidae)) ha we e mos equen ly de-
ec ed in phy o elma a h oughou ield su eys a ou s udy si e
in F ench Guiana. Following b oad species- wide compa isons, we
ocus on a mo e de ailed analysis o pool choice in D. inc o ius, a
phy o elm specialis wi h p eda o y and cannibalis ic adpoles which
a e deposi ed in a ange o phy o elm ypes (e.g., palm b ac s, ee
holes, allen ees; Figu es 1 and 2) ha occu om he o es loo
o mo e han 20 m in e ical heigh (Gauche , 2002; Rojas, 2014,
2015). The use o he high canopy pools is pe plexing because D.
inc o ius is commonly success ul in e es ial pools (Rojas, 2014).
I is known ha pool chemis y can change d as ically depending
on subs a e (“dead” o “li e”; see Figu e 1), heigh , and communi y
composi ion (Pe i e al., 2018; Ruano- Faja do e al., 2014). Thus,
be e unde s anding he ecology o high a bo eal pools and cha ac-
e izing phy o elma a ac oss he e ical g adien could help explain
bo h he appa en success o D. inc o ius in a wide ange o pools
and why pa en s some imes ca y hei o sp ing o such heigh s.
To ou knowledge, his is he i s s udy p o iding de ailed bio ic,
physical, and chemical compa isons o phy o elm choice be ween
Neo opical species.
2 | MATERIALS AND METHODS
The s udy was ca ied ou in he p ima y lowland e a- i me o es
nea he Camp Pa a é a he CNRS Nou agues Ecological Resea ch
S a ion in he Na u e Rese e Les Nou agues, F ench Guiana
(4°02′N, 52°41′W) o e wo ield seasons: 1s Feb ua y o 20 h
Ma ch 2019, and 30 h Janua y o 26 h Feb ua y 2020. The s udy
a ea (app oxima ely 0.2 km2) was chosen speci ically because o
he high abundance o D. inc o ius (Rojas & Pašukonis, 2019). Pools
we e ound wi h a combina ion o ield me hods. We oppo unis-
ically sea ched o pools a ge ing sui able mic ohabi a s such as
allen ees, ees wi h bu esses, and palm ees. In addi ion, pools
we e disco e ed by using acking o ollow D. inc o ius du ing p e-
ious s udies (Pašukonis e al., 2019). We also used expe imen ally
induced adpole anspo in combina ion wi h acking (Pašukonis
|
3
FOUILLOUX e aL.
e al., 2017) o ind addi ional pools used by A. emo alis. T ees wi h
high a bo eal pools we e disco e ed by loca ing calls p oduced by
he ee hole- b eeding ogs T achycephalus esini ic ix and T. had o-
ceps du ing nigh su eys.
2.1 | S udy species
Th oughou he cou se o his wo k, h ee species o med he co e
o ou da a. D. inc o ius and A. emo alis a e bo h small poison ogs
belonging o he supe amily Dend oba oidea. A. emo alis is a e es-
ial og whose adul males agg essi ely de end e i o ies du ing he
ainy season (Na ins e al., 2003; Roi hmai , 1992), om which hey
ca y ecen ly ha ched adpoles o a a ie y o e es ial pools in-
cluding phy o elma a close o he g ound (Ringle e al.,2009, 2013).
Tadpoles o his species a e omni o ous (McKeon & Summe s, 2013),
bu no cannibalis ic (Summe s & McKeon, 2004). Simila ly o A. em-
o alis, D. inc o ius males ca e o hei o sp ing by ca ying hem o
pools o wa e . Males o his species a e adep climbe s (deposi ing
hei adpoles om he g ound o mo e han 20 m in e ical heigh ;
Gauche , 2002; Rojas, 2014, 2015), and hei adpoles a e agg essi e
cannibals (Rojas, 2014; Rojas & Pašukonis, 2019).
Os eocephalus oophagus is a hylid ee og wi h bipa en al ca e
and obliga ely oophagous adpoles (Jung e e al., 2000; Jung e &
Weygold , 1999). As in ou ield si e, adul s ha e been ound o call
and b eed in b omeliads, ee holes, and palm axils close o he o es
loo (Jung e & Weygold , 1999). Tadpoles o his species de elop in
he same pool in which he eggs a e deposi ed.
2.2 | Sampled pools
We exclusi ely conside ed phy o elma a h oughou his s udy. Pools
could be classi ied in o wo ca ego ies: dead subs a es, which in-
cluded holes in dead b anches, allen ees, and allen Oenoca pus
palm b ac s, and li e subs a es, which included li e ee unks,
b anches, oo s, and bu esses. We did no sample b omeliads and
nonphy o elm pools as hese pools a e no used by D. inc o ius. Based
on he pools’ heigh and accessibili y o di e en og species, we
e med he pools as “g ound access,” “low a bo eal,” o “high a bo eal”
(Figu es 1 and 2). G ound access pools did no equi e e ical climb-
ing abili y o each and included dead allen s uc u es as well as pools
in li e oo s o low bu esses. Low a bo eal pools we e inside e ical
s uc u es low on he unk o on high bu esses. High a bo eal pools
we e high on he unk o in canopy b anches and we e accessed o
sampling using ope- based canopy access me hods. The e was a clea
e ical sepa a ion be ween g ound access and low a bo eal pools,
which we e all unde 212 cm in heigh and be ween hose and high
a bo eal pools, which we e all abo e se en me e s in heigh . In o al,
we sampled 84 unique pools ac oss he 2019 and 2020 ield seasons.
Se e al unique pools we e some imes ound and sampled in he
same ee. Fo all pools, we eco ded he pool ype, loca ion (la i ude/
FIGURE 1 Visual o e iew o sampled pool di e si y. Pho og aphs show he di e si y o pools ac oss he e ical g adien . Phy o elma a
used by ogs include “li e” subs a e pools such as ee holes (a), high a bo eal pools (b), and bu esses (c). The e we e also commonly
occupied “dead” subs a e pools such as allen ees (d, e) and palm b ac s ( )
4
|
FOUILLOUX e aL.
longi ude), heigh om he g ound o he pool edge, la ges wid h and
leng h pa allel o he wa e su ace, and he pool dep h (maximum
possible wa e dep h o he phy o elma a) om he solid sedimen
bo om o he maximum wa e le el line. Based on hese measu e-
men s, we es ima ed he maximum wa e - holding capaci y o each
pool using he olume o mula o a semi- ellipsoid as in Rojas (2014).
O he sampling me hods di e ed be ween he wo ield seasons.
2.3 | 2019 ield season sampling
In 2019, we quan i ied physical measu es (heigh , pool dimensions,
lea li e olume), bio ic measu es (amphibian and in e eb a e
coun s and di e si y), and chemical measu emen s (see Appendix 2
o desc ip ion o all a iables measu ed). Fo pools accessible om
he g ound and smalle a bo eal pools, we a emp ed o sample all
FIGURE 2 Ve ical pa i ioning o he phy o elm- b eeding anu an communi y in he Nou agues Na u e Rese e, F ench Guiana.
Numbe s indica e se en species de ec ed du ing his s udy: 1. Dend oba es inc o ius, 2. Alloba es emo alis 3. Rhinella cas aneo ica, 4.
Rani omeya amazonica, 5. Os eocephalus oophagus, 6. T achycephalus esini ic ix, and 7. T. had oceps. Le e s indica e species wi h (a)
e es ial o (b) aqua ic eggs. La in nume als indica e commonly used pool ypes: I. ee holes a a ious heigh s, II. allen palm b ac s, III.
g ound puddles, and IV. b omeliads. *Mos commonly encoun e ed species included in he analysis and ** pool ypes no sampled in his
s udy (see me hods)

|
5
FOUILLOUX e aL.
adpoles and Odona a la ae (p eda o s o adpoles; Caldwell, 1993;
Fincke, 1999; Summe s & McKeon, 2004) in each pool. Ini ially, we
ca e ully obse ed he undis u bed pool and a emp ed o ca ch
all adpoles and Odona a la ae using a a ie y o ools. We hen
syphoned he en i e olume o he wa e and sedimen om he
pool, emp ied he lea li e , and sea ched o adpoles and Odona a
la ae. The olumes o wa e , sedimen , and lea li e we e meas-
u ed. Fo deep a bo eal pools, we epea edly ne ed and obse ed
he pool un il no mo e adpoles we e caugh du ing i e minu es
o con inuous ne ing. We ca e ully sc aped he inne walls o he
pools and caugh as many Odona a la ae as possible. The lea li e
olume could no be accu a ely measu ed o some deep a bo eal
pools, bu hey ypically we e p o ec ed om alling lea es and had
li le lea li e in hem.
We used isually appa en mo phological ai s o iden i y ad-
poles, excep o Alloba es emo alis, A. g an i, and Amee ega hahneli,
which we could no eliably di e en ia e in he ield. Because
Alloba es emo alis was mo e common in ou s udy a ea han A. g an i
and Am. hahneli and we ne e obse ed A. g an i and Am. hahneli di-
ec ly a he pools, we classi ied all A. emo alis- like adpoles as such.
I is impo an o no e ha some A. g an i and Am. hahneli adpoles
may ha e been misclassi ied as A. emo alis. Howe e , his does no
a ec he in e p e a ion o ou esul s as all h ee species a e c yp ic
e es ial poison ogs simila in appea ance, ecology, and beha io .
We also oppo unis ically eco ded all species o adul ogs hea d
o seen a each pool h oughou he sampling pe iod. This was used
as an amphibian di e si y index be ween 0 and 8 species obse ed
a each pool. Tadpoles o only h ee ou o se en eco ded species,
namely D. inc o ius, O. oophagus, and A. emo alis, we e de ec ed in
pools wi h su icien equency o u he analysis (N = 34 (2019),
N = 7, and N = 10 pools, espec i ely).
Sampled in e eb a es we e coun ed, pho og aphed, and clas-
si ied only o a g oup le el (usually o de o class) appa en in he
ield. To es ima e he p eda ion p essu e on adpoles, we used he
o al coun and a e age size o all Odona a la ae de ec ed in he
p ocedu e desc ibed abo e. To es ima e densi y and di e si y o
aqua ic in e eb a es, we ca e ully sea ched and coun ed in e e-
b a es in a sample o up o 10 li e s o wa e and up o one li e
o sedimen in p opo ion o he o al es ima ed pool olume. Fo
each li e o he wa e olume sampled, we sampled ~100 ml o sed-
imen om he bo om o he pool. When he wa e olume was
less han one li e o he amoun o sedimen was less han 100 ml,
we sampled he en i e pool and eco ded he exac olumes. In
he inal analysis, we used he in e eb a e densi y (coun di ided
by he olume sampled) and he di e si y index co esponding o
ou classi ica ion (be ween 0 and 12). The ollowing 12 ca ego ies
we e used o quan i y in e eb a e di e si y: Odona a Zygop e a
la ae, Odona a Anisop e a la ae, su ace Coleop e a adul s, di -
ing Coleop e a adul s, Coleop e a Sci idae la ae, T ichop e a la -
ae, Dip e a Culicidae la ae, Dip e a Chi onomidae la ae, Dip e a
Tipulidae la ae, o he Dip e a la ae, small ed Annelida, and o he
uniden i ied la ae. All wa e , sedimen , adpoles, and in e eb a es
we e eleased back in o he pool a e sampling.
We measu ed wa e conduc i i y, salini y, o al dissol ed solids
(TDS), dissol ed oxygen, and empe a u e wi h elec onic senso s
(EZDO 7200 and pHenomenal OX4110H). Wa e chemis y (KH
(also known as alkalini y), ha dness, and NO3) was eco ded using
aqua ium wa e es ing s ips (JBL EasyTes ). All measu es we e
aken om he undis u bed su ace wa e o he pool.
2.4 | 2020 ield season sampling
The 2020 da ase ocused solely on D. inc o ius adpole coun s and
pH measu emen s o weekly esampled g ound access phy o elma a
(N = 26) o e he ime pe iod o a mon h (Feb ua y 2020). Rain all
da a we e p o ided by he Nou agues Ecological Resea ch S a ion
om an abo e- canopy wea he s a ion in he s udy a ea. High a bo-
eal pools (N = 8, 2020) we e only measu ed once. pH was eco ded
using a pH me e (AMTAST Wa e p oo pH Me e ). The pH me e
was calib a ed once pe day, p io o pool sampling, using bo h acidic
(pH = 4) and neu al (pH = 7) calib a ion solu ions. The pH o g ound
access pools was aken di ec ly by subme ging he pH p obe in o he
pool, and he measu emen was eco ded once ead- ou s abilized.
Fo a bo eal pools, a sample o wa e was collec ed using a sy inge,
which was hen sealed a bo h ends. Once on he g ound, one end o
he sy inge was opened, and he pH was measu ed by subme ging
he pH p obe in o he sy inge. Sy inges we e ne e eused. Be ween
pool sampling, he pH p obe was wiped wi h a clean clo h and insed
wi h aqui e wa e .
2.5 | S a is ical analyses
The p esence o D. inc o ius in pools was analyzed using 2019 ield
da a. As a esul o he high collinea i y be ween a iables in he
2019 da ase (see Figu e S1), we used a p incipal componen eg es-
sion o analyze phy o elm ecology da a. We i s checked da a o
a non andom s uc u e ollowing Bjö klund (2019); hen, we es ab-
lished ha he co ela ion ma ices we e signi ican ly di e en om
andom (ψ = 10.22, p = 0;
𝜙
= 0.238, p <.001) o ensu e hey we e
sui able o a PCA. Based on each PC di e ence om andom ma-
ices, we selec ed he i s h ee p incipal componen s as p edic-
o s o p obabili y o D. inc o ius adpole p esence as a binomial
esponse in he p incipal componen eg ession (PC1- 3 explained
abou 53% o he a iabili y o he da a (whe e PC1 = 0.24 ± 0.48,
PC2 = 0.17 ± 0.40, PC3 = 0.11 ± 0.33 ( a iance explained ± SE)). We
e alua ed he i o nega i e binomial GLMs based on second- o de
AIC anks (AICc) using he package AICcmoda g (Maze olle, 2020)
which a e specialized o smalle sample sizes (Akaike, 1974; see
Table S1). Models wi hin wo AIC sco es o each o he we e u he
e alua ed by assessing he signi icance o in e ac ions be ween
model e ms.
To be e unde s and which a iables con ibu ed signi ican ly o
each p incipal componen , we calcula ed which a iables had index
loadings la ge han andom da a. Following he me hods ou lined by
6
|
FOUILLOUX e aL.
Bjö klund (2019) and Viei a (2012), we andomized he da a and cal-
cula ed new co ela ion ma ices which we pe mu ed 1,000 imes.
We hen compa ed he index o loadings (
IL
ij =u
2
ij
×𝜆
2
i
, see Viei a
(2012) o de ails) be ween each obse ed PC and he andomly gen-
e a ed da a o see which a iables con ibu ed signi ican ly o each
p incipal componen .
The 2020 da ase consis ed o weekly esampled pools h ough-
ou Feb ua y 2020. Models ook epea ed measu es o pool ID in o
accoun as a andom e ec . Bo h he p esence o D. inc o ius ad-
poles (coun ; nega i e binomial amily) and pH (Gaussian amily) om
esampled pools we e modeled using a mixed e ec s gene alized
linea model in he package “glmmTMB” (Magnusson e al., 2020).
P edic o s uc u e o bo h pH and D. inc o ius models was buil
based on biologically ele an a iables (pool subs a e, ime,
D. inc o ius adpole coun ( o pH model), wa e capaci y, su ace
a ea:dep h a io). Using hese a iables, models we e i s i wi h el-
e an in e ac ions (see Tables S2 and S3), which we e hen emo ed
i hey did no con ibu e signi ican ly o he model using single e m
dele ions (using base R unc ion, d op1; Zuu e al., 2009). Residuals
we e diagnosed using he package “DHARMa” (Ha ig, 2020). Final
models we e hen checked o o e dispe sion and ze o- in la ion
(using DHARMa); none o he inal models equi ed co ec ion. All
code was done in R (R Co e Team, 2015).
3 | RESULTS
3.1 | Species- wide ends
We ound 7 di e en species o ogs (ei he adpoles o adul s) om
4 amilies in he phy o elma a we su eyed o b eeding (Figu e 2):
Dend oba idae: Dend oba es inc o ius and Rani omeya amazonica;
A omoba idae: Alloba es emo alis; Hylidae: Os eocephalus oophagus,
T achycephalus esini ic ix, and T. ha d oceps; and Bu onidae: Rhinella
cas aneo ica. The adpoles o only h ee species (D. inc o ius,
O. oophagus, and A. emo alis, p esen in N = 34, N = 7, and N = 10
pools, espec i ely) we e de ec ed equen ly enough o u he
analysis. The species- wide da ase is based on he sampling o 70
unique pools in 2019.
Di e ences in pool accessibili y a e highligh ed in Figu e 3.
Compa ed o A. emo alis and O. oophagus, one o he mos s iking
aspec s o D. inc o ius ecology is i s lexibili y wi h espec o si e
FIGURE 3 Tadpole p esence ac oss he e ical landscape. Panel (a) shows all sampled pools. Fo he “All” ca ego y, colo ed/emp y
iangles ep esen p esence/absence da a o a leas one o he h ee species in he pool. Dashed line is d awn a 220 cm; pools abo e his
limi a e classi ied as high a bo eal pools. Panel (b) highligh s occupied pools below 220 cm. Dend oba es inc o ius (N = 34) adpoles occu
in pools ac oss he e ical landscape. Dis ibu ion o O. oophagus (N = 7) and A. emo alis (N = 10) adpoles shows possible e ical niche
pa i ioning. Boxplo whiske s ex end 1.5 * in e qua ile ange. Violin plo s ep esen densi y dis ibu ion o species occu ence. Da a a e
om he 2019 ield season
|
7
FOUILLOUX e aL.
choice on a e ical axis. Dend oba es inc o ius adpoles we e ound
in pools om he o es loo o mo e han 15 me e s in he canopy.
Fo O. oophagus, a s ic ly a bo eal og in i s adul s age, adpoles
we e de ec ed only in low a bo eal pools whe e climbing is equi ed
o access, anging om 20 cm o 1.7 m in heigh . In A. emo alis,
adpoles we e only ound in g ound access pools whe e no e ical
climbing is equi ed and occu ed a a maximum heigh o 71 cm.
Despi e small sample sizes, we ound clea ends: O. oophagus
adpoles a e hea ily biased owa d small, clea pools and A. emo -
alis is p esen in medium and la ge pools wi h la ge amoun s o lea
li e , whe eas D. inc o ius occu s h oughou he sampled ange
(Figu e 4). Wi h espec o densi y, we ound ha A. emo alis and
O. oophagus adpoles occu in highe numbe s in pools (median = 7.5
and 10, espec i ely) compa ed o D. inc o ius adpoles (median = 2),
hough a la ge ange o adpole densi ies was ound o all species
(A. emo alis: 1– 51 adpoles, O. oophagus: 1– 50 adpoles, D. inc o-
ius: 1– 43 adpoles).
As opposed o A. emo alis and O. oophagus, D. inc o ius can
occupy chemically di e se pools, showing ema kable lexibili y
wi h espec o KH, salini y, and ha dness ha appea s o limi
he o he species. Alloba es emo alis and O. oophagus appea o
exis in simila KH anges (KH = 3– 6), while D. inc o ius appea s
mo e ole an o ex eme alues (KH = 3 – 2 0 ) . Alloba es emo alis
adpoles occu ed in pools wi h a salini y ange om 5 o 37 ppm,
while O. oophagus adpoles occupied pools wi h a ange om 48 o
225 ppm (Figu e 5, Panel C). Dend oba es inc o ius again appea s
o ha e no unc ional limi a ion, occupying pools wi h salini y om
11 ppm up o 955 ppm.
3.2 | Deposi ion si e decisions:
Dend oba es inc o ius
Because we de ec ed D. inc o ius adpoles much mo e equen ly,
we we e able o conduc a mo e ho ough analysis o he a iables
p edic ing adpole p esence in his species (see Appendix 2). We
used p incipal componen s as p edic o s o D. inc o ius p esence.
Based on an AIC model compa ison, we did no de ec any signi i-
can in e ac ions be ween componen s (Table S1). A nega i e bino-
mial GLM only de ec ed PC1 o play a signi ican ole in p edic ing
FIGURE 4 Pool occupancy based on wa e capaci y and lea li e olume o phy o elma a. All da a a e subse ed o low a bo eal and
g ound access pools (<220 cm). Panel (a) is he p obabili y o pool occupancy (binomial, 0/1) based on wa e capaci y; da a a e ace ed based
on ela i e pool size (small = <1,000 ml, medium = <5,000 ml, and la ge = >5,000 ml). Poin s a e plo ed wi h a small amoun o andom
noise on he y- axis o acili a e isualiza ion o o e lapping da a. Panel (b) illus a es he co ela ion be ween lea li e and heigh , ace ed by
he same pool ca ego ies as Panel (a). Poin s a e colo ed by species p esence. Dashed line indica es he e ical limi o A. emo alis (<75 cm).
Ou o he 62 g ound access and low a bo eal pools obse ed, D. inc o ius co- occu ed once wi h A. emo alis and once wi h O. oophagus; O.
oophagus and A. emo alis adpoles we e ne e ound in he same pool
8
|
FOUILLOUX e aL.
adpole p esence (Table 1, CI: 0.08– 0.42, p = .003), whe e an in-
c ease in componen alue inc eased he p obabili y o de ec ing
adpoles.
Following Bjö klund (2019), we ound ha , when compa ed o
andomly gene a ed ma ices, i e ou o he o iginal 14 ai s (see
Appendix 2 o ai de ini ions) con ibu ed signi ican ly o he i s
p incipal componen . The signi ican ai s can be b oadly ca ego ized
using h ee desc ip o s: (a) chemical (KH, p < .001; IL = 1.50, ha dness,
p = .001, IL = 1.30; salini y, p < .001, IL = 1.62); (b) physical (heigh ,
p =.013, IL = 1.06); and (c) biological (in e eb a e di e si y, p < .001,
IL = 1.20) (see Figu e 6). Al oge he , hese esul s show ha D. inc-
o ius adpoles we e ound signi ican ly mo e equen ly in pools wi h
highe le els o ha dness, KH, and salini y; highe in he e ical g adi-
en ; and wi h mo e di e se in e eb a e communi ies (Figu e 6).
3.3 | Dend oba es inc o ius ac oss empo al scales
Using bo h 2019 and 2020 da ase s, we we e able o ollow phy o el-
ma a ac oss mul iple imescales: 13 weekly esampled g ound access
and low a bo eal pools, 13 annually esampled g ound access and
low a bo eal pools, and 7 annually esampled high a bo eal pools.
O e all, we ound ha pools can pe sis o e mul iple sampling sea-
sons. High a bo eal pools appea o be he mos s able wi h espec
o bo h adpole coun and adpole densi y compa ed o low a bo eal
and g ound access pools sampled bo h yea s (
x
High (2019) = 13.14 ad-
poles,
x
High (2020) = 10 adpoles e sus
x
Low (2019) = 0.92 adpoles,
x
Low (2020) = 1.31 adpoles). High a bo eal pools also had he highes
a e age pH and KH (pHHigh = 6.73, KHHigh = 15.14) compa ed o a -
e ages o o he pool subs a es (pH(Low)Li e = 4.35, pH(Low)Dead = 5.68;
KH(Low)Li e = 5.69, KH(Low)Dead = 5.88). Due o di icul accessibili y,
high a bo eal pools we e sampled only once pe yea and hus we e
excluded om u he analysis in ol ing epea ed sampling.
FIGURE 5 Chemical and physical p edic o s o adpole p esence in Neo opical adpoles. We ind ha D. inc o ius adpoles a e ole an
o a wide ange o KH, ha dness, heigh , and salini y alues, bu appea o be limi ed wi h espec o high lea li e olumes. Plo s a e based
on a iables wi h a ied species limi s (dashed lines). Colo ed poin s ep esen species p esence. Black lines a e i wi h a GLM smoo he ,
wi h 95% CI highligh ed in ligh g ay
TABLE 1 P incipal componen eg ession o adpole p esence
in phy o elm pools. Using a nega i e binomial GLM, we ound
ha only he i s p incipal componen is signi ican in p edic ing
adpole p esence
P edic o s
Tadpole p esence (Y/N)
Es ima es CI p
(In e cep ) −0.85 −1.28 o −0.49 <.001
PC1 0.25 0.07 o 0.42 .003
PC2 0.09 −0.19 o 0.40 .582
PC3 0.21 −0.07 o 0.48 .144
The bold ex was o emphasize column names in he o iginal ables.
|
15
FOUILLOUX e aL.
Aus al Ecology, 28, 33– 37. h ps://doi.o g/10.1046/j.1442- 9993.
2003.01239.x
Na ins, P. M. e al. (2003). Bimodal signal equisi e o agonis ic beha io in
a da - poison og, Epipedoba es emo alis. P oceedings o he Na ional
Academy o Sciences o he Uni ed S a es o Ame ica, 100, 577– 580.
Nussbaum, R. A. (1987). Pa en al ca e and EGG size in salamande s: An
examina ion o he sa e ha bo hypo hesis. Resea ches on Popula ion
Ecology, 29, 27– 44. h ps://doi.o g/10.1007/BF025 15423
O esen, O. H., & Bolla, S. (1998). Combined e ec s o empe a u e
and salini y on de elopmen and su i al o A lan ic halibu la ae.
Aquacul u e In e na ional, 6, 103– 120.
Pašukonis, A. e al. (2017). Induced pa en al ca e in a poison og: A ad-
pole c oss- os e ing expe imen . Jou nal o Expe imen al Biology, 220,
3949– 3954. h ps://doi.o g/10.1242/jeb.165126
Pašukonis, A. e al. (2019). How a do adpoles a el in he ain o es ?
Pa en - assis ed dispe sal in poison ogs. E olu iona y Ecology, 33,
6 1 3 – 6 2 3 . h p s : / / d o i . o g / 1 0 . 1 0 0 7 / s 1 0 6 8 2 - 0 1 9 - 0 9 9 9 4 - z
Pe i , B. A. e al. (2018). P edic o s and bene i s o mic ohabi a selec-
ion o o sp ing deposi ion in golden ocke ogs. Bio opica, 50,
919– 928. h ps://doi.o g/10.1111/b p.12609
Poelman, E. H. e al. (2013). Amazon poison ogs (Rani omeya amazonica)
use di e en phy o elm cha ac e is ics o de e mine hei sui abili y
o egg and adpole deposi ion. E olu iona y Ecology, 27, 661– 674.
h p s : / / d o i . o g / 1 0 . 1 0 0 7 / s 1 0 6 8 2 - 0 1 3 - 9 6 3 3 - 3
Poelman, E. H., & Dicke, M. (2007). O e ing o sp ing as ood o canni-
bals: O iposi ion s a egies o Amazonian poison ogs (Dend oba es
en imacula us). E olu iona y Ecology, 21, 215– 227. h ps://doi.
o g / 1 0 . 1 0 0 7 / s 1 0 6 8 2 - 0 0 6 - 9 0 0 0 - 8
R Co e Team (2015). R: A language and en i onmen o s a is ical
compu ing.
Ramos, G. J. P. e al. (2017). Algae in phy o elma a om Caa inga: Fi s
eco d o he genus Rhopalosolen Fo (Chlo ophy a) o B azil.
Check Lis , 13, 403– 410.
Ringle , E. e al. (2013). Tadpole anspo logis ics in a Neo opical poi-
son og: Indica ions o s a egic planning and adap i e plas ici y
in anu an pa en al ca e. F on ie s in Zoology, 10, 67. h ps://doi.
o g / 1 0 . 1 1 8 6 / 1 7 4 2 - 9 9 9 4 - 1 0 - 6 7
Ringle , E. e al. (2018). Hie a chical decision- making balances cu en
and u u e ep oduc i e success. Molecula Ecology, 27, 2289– 2301.
h ps://doi.o g/10.1111/mec.14583
Ringle , M. e al. (2015). Popula ions, pools, and pecca ies: Simula ing
he impac o ecosys em enginee s on ain o es ogs. Beha io al
Ecology, 26, 340– 349.
Ringle , M., U sp ung, E., & Hödl, W. (2009). Si e ideli y and pa e ns o
sho - and long- e m mo emen in he b illian - highed poison og
Alloba es emo alis (A omoba idae). Beha io al Ecology and Sociobiology,
63( 9 ) , 1 2 8 1 – 1 2 9 3 . h p s : / / d o i . o g / 1 0 . 1 0 0 7 / s 0 0 2 6 5 - 0 0 9 - 0 7 9 3 - 7
Roache, M. C. e al. (2006). E ec s o salini y on he decay o he esh-
wa e mac ophy e, T iglochin p oce um. Aqua ic Bo any, 84, 45– 52.
Roi hmai , M. E. (1992). Te i o iali y and male ma ing success in he da -
poison og, Epipedoba es emo alis (dend oba idae, Anu a). E hology,
92, 331– 343. h ps://doi.o g/10.1111/j.1439- 0310.1992. b009 70.x
Rojas, B. (2014). S ange pa en al decisions: Fa he s o he dyeing poi-
son og deposi hei adpoles in pools occupied by la ge canni-
bals. Beha io al Ecology and Sociobiology, 68, 551– 559. h ps://doi.
o g / 1 0 . 1 0 0 7 / s 0 0 2 6 5 - 0 1 3 - 1 6 7 0 - y
Rojas, B. (2015). Mind he gap: T ee alls as d i e s o pa en al ade-
o s. Ecology and E olu ion, 5, 4028– 4036. h ps://doi.o g/10.1002/
ece3.1648
Rojas, B., & Pašukonis, A. (2019). F om habi a use o social beha io :
Na u al his o y o a oiceless poison og, Dend oba es inc o ius.
Pee J, 7, e7648.
Rosenblum, E. B. e al. (2010). The deadly chy id ungus: A s o y o
an eme ging pa hogen. PLoS Pa hogens, 6, e1000550. h ps://doi.
o g/10.1371/jou n al.ppa .1000550
Ruano- Faja do, G. e al. (2014). B omeliad selec ion by wo salamande
species in a ha sh en i onmen . PLoS One, 9, e98474. h ps://doi.
o g/10.1371/jou n al.pone.0098474
Rudol , V. H. W., & Rödel, M.- O. (2005). O iposi ion si e selec ion in a
complex and a iable en i onmen : The ole o habi a quali y and
conspeci ic cues. Oecologia, 142, 316– 325. h ps://doi.o g/10.1007/
s 0 0 4 4 2 - 0 0 4 - 1 6 6 8 - 2
Sawidis, T. e al. (2011). T ees as bioindica o o hea y me al pollu ion
in h ee Eu opean ci ies. En i onmen al Pollu ion, 159, 3560– 3570.
h ps://doi.o g/10.1016/j.en pol.2011.08.008
Schul e, L. M. e al. (2011). The smell o success: Choice o la al ea -
ing si es by means o chemical cues in a Pe u ian poison og.
Animal Beha io , 81, 1147– 1154. h ps://doi.o g/10.1016/j.anbeh
a .2011.02.019
Schul e, L. M. e al. (2020). De elopmen s in amphibian pa en al ca e
esea ch: His o y, p esen ad ances, and u u e pe spec i es.
He pe ological Monog aphs, 34, 71– 97.
Sih, A., & Moo e, R. D. (1993). Delayed ha ching o salamande eggs in
esponse o enhanced la al p eda ion isk. Ame ican Na u alis , 142,
947– 960. h ps://doi.o g/10.1086/285583
Sil es ini, R. A. e al. (2011). Simula ing i e egimes in he Amazon in e-
sponse o clima e change and de o es a ion. Ecological Applica ions,
21, 1573– 1590. h ps://doi.o g/10.1890/10- 0827.1
Summe s, K. (1990). Pa e nal ca e and he cos o polygyny in he g een
da - poison og. Beha io al Ecology and Sociobiology, 27, 307– 313.
h ps://doi.o g/10.1007/BF001 64001
Summe s, K., & McKeon, C. S. (2004). The e olu iona y ecology o phy-
o elma a use in Neo opical poison ogs. Miscellaneous Publica ions,
Museum o Zoology, Uni e si y o Michigan, 193, 55– 73.
Summe s, K., & Tumul y, J. (2014). Chap e 11 – Pa en al ca e, sexual
selec ion, and ma ing sys ems in neo opical poison ogs. In R.
H. Macedo, & G. Machado (Eds.), Sexual selec ion (pp. 289– 320).
Academic P ess.
S endsen, G. E. (1976). S uc u e and loca ion o bu ows o yellow-
bellied ma mo . The Sou hwes e n Na u alis , 20, 487– 494. h ps://
doi.o g/10.2307/3669865
Touchon, J. C., & Wo ley, J. L. (2015). O iposi ion si e choice unde con-
lic ing isks demons a es ha aqua ic p eda o s d i e e es ial
egg- laying. P oceedings o he Royal Socie y B: Biological Sciences, 282,
20150376. h ps://doi.o g/10.1098/ spb.2015.0376
Vági, B. e al. (2019). Pa en al ca e and he e olu ion o e es iali y
in ogs. P oceedings o he Royal Socie y B: Biological Sciences, 286,
20182737. h ps://doi.o g/10.1098/ spb.2018.2737
Viei a, V. (2012). Pe mu a ion es s o es ima e signi icances on P incipal
Componen s Analysis. Compu a ional Ecology and So wa e, 2,
103– 123.
on May, R. e al. (2009). B eeding- si e selec ion by he poison og
Rani omeya biola in Amazonian bamboo o es s: An expe imen al
app oach. Canadian Jou nal o Zoology, 87, 453– 464. h ps://doi.
o g/10.1139/Z09- 026
Wa ken in, K. M. (2011). En i onmen ally cued ha ching ac oss axa:
Emb yos espond o isk and oppo uni y. In eg a i e and Compa a i e
Biology, 51, 14– 25. h ps://doi.o g/10.1093/icb/ic 017
Wells, K. D. (2007). The ecology and beha io o amphibians. Uni e si y o
Chicago P ess.
Weygold , P. (1980). Complex b ood ca e and ep oduc i e beha iou
in cap i e poison- a ow ogs, Dend oba es pumilio O. Schmid .
Beha io al Ecology and Sociobiology, 7, 329– 332.
Williams, B. K. e al. (2008). Lea li e inpu media es adpole pe o -
mance ac oss o es canopy ea men s. Oecologia, 155, 377– 384.
h p s : / / d o i . o g / 1 0 . 1 0 0 7 / s 0 0 4 4 2 - 0 0 7 - 0 9 2 0 - y
Yang, S. F. e al. (2008). Fo ma ion and cha ac e isa ion o ungal and
bac e ial g anules unde di e en eeding alkalini y and pH condi-
ions. P ocess Biochemis y, 43, 8– 14. h ps://doi.o g/10.1016/j.p ocb
io.2007.10.008

16
|
FOUILLOUX e aL.
Zhao, Q.- S. e al. (2016). Nes si e choice: A po en ial pa hway linking
pe sonali y and ep oduc i e success. Animal Beha io , 118, 97– 103.
h ps://doi.o g/10.1016/j.anbeh a .2016.05.017
Zuu , A. F., Ieno, E. N., Walke , J. N., Sa elie , A. A., & Smi h, G. M. (2009).
Mixed e ec s models and ex ensions in ecology wi h R. Sp inge Science.
SUPPORTING INFORMATION
Addi ional suppo ing in o ma ion may be ound online in he
Suppo ing In o ma ion sec ion.
How o ci e his a icle: Fouilloux CA, Se ano Rojas SJ,
Ca ajal- Cas o JD, e al. Pool choice in a e ical landscape:
Tadpole- ea ing si e lexibili y in phy o elm- b eeding ogs.
Ecol E ol. 2021;00:1– 18. h ps://doi.o g/10.1002/ece3.7741
FIGURE A1 Hypo he ical succession
o ela i e cue impo ance in phy o elma a
ac oss ime
APPENDIX 1
SPECULATIONS AND ALTERNATIVE VIEWPOINTS
Dend oba es inc o ius males ypically a he egg clu ches o 2– 5 ad-
poles pe clu ch and b eed yea - ound (Rojas & Pašukonis, 2019). As
a esul o he p esumed high ene ge ic expense om ca ying each
adpole om each clu ch singly, we hypo hesize ha adpoles ans-
po ed la e may be subjec o be - hedging by a he s.
Combined wi h he impo an chemical aspec s o pools shown
om 2019 da a, i seems ha a he s can cue on ei he chemical
(KH, salini y, ha dness) o biological componen s ( he p esence o
conspeci ics) as in o ma ion abou pool s abili y. We specula e ha
he impo ance o chemical and physical cues changes wi h espec
o pool age. Fo example, in new unin aded pools, chemical cues o
he pool may be mo e impo an (le side o igu e), while in olde ,
es ablished pools ha a e mo e densely occupied, densi y se es as
a main cue o anspo ing a he s ( igh side o igu e). Finally, he
alue o hese cues may a y wi h he amoun o o sp ing a he s
a e ca ing o . Hypo he ically, a he s who mus anspo mo e o -
sp ing a e less disce ning o whe e hey anspo la e adpoles
because hey can a o d o make less “ideal” deposi ion decisions
because o hei la ge ep oduc i e ou pu .
|
17
FOUILLOUX e aL.
APPENDIX 2
DEFINITION OF VARIABLES (TRAITS) CONSIDERED IN THE PRINCIPAL COMPONENT ANALYSIS
Va iable Ca ego y Desc ip ion
Salini y Chemical Quan i ica ion o sal in solu ion. Range = 0– 10,000 ppm.
KH (alkalini y) Chemical Quan i ica ion o pool bica bona e/ca bona e in solu ion. Range = 0– 25 dKH
NO−
3Chemical Quan i ica ion o ni a e in solu ion. Range = 0– 160 ppm.
Ha dness Chemical Quan i ica ion o ions in solu ion (e.g., calcium). Range = 0– 425 ppm.
Heigh Physical Ve ical heigh om he g ound o he pool en ance. Measu ed in cm.
Wa e capaci y Physical Wa e - holding capaci y o he pool. Es ima ed om pool wid h, leng h, and dep h using a
semi- ellipsoid o mula.
Su ace a ea o dep h a io Physical Su ace a ea o dep h a io. Su ace a ea calcula ed om semi- ellipsoid o mula.
Lea li e olume Physical The measu e o lea li e olume in each pool.
Amphibian di e si y Biological The sum o all species obse ed using each pool including adul s, calling, dead adpoles, and
oppo unis ic obse a ions a e he sampling.
In e eb a e densi y Biological Sum o all in e eb a e densi ies (coun s di ided by sampling olume)
In e eb a e di e si y Biological Numbe o dis inc in e eb a e ca ego ies obse ed in each pool (be ween 0 and 12)
P eda o coun Biological Numbe o Odona a la ae in each pool.
A e age p eda o size Biological The a e age size o Odona a la ae in each pool. Size is calcula ed by di iding
(p ed_size_sum)/(p ed_coun )
To al o he Biological Sum o O. oophagus and A. emo alis adpoles co- occu ing in he pool.
Colo s we e jus o highligh he change in ca ego ies (chemical, physical, ec .).
APPENDIX 3
Rela ionship be ween lea li e olume and salini y. Dashed line is a
40 ppm which is he limi whe e we de ec ed A. emo alis adpoles.
Below his le el, i appea s ha lea li e and salini y ha e a sligh ly
posi i e ela ionship, hough in e p e a ion is limi ed by sample size
(blue poin s, NFemo alis = 10). Da a a e subse ed o g ound access
pools, and salini y uppe bound was limi ed o 70 ppm o emphasize
po en ial lea li e e ec .
18
|
FOUILLOUX e aL.
APPENDIX 4
Rela ionship be ween heigh and salini y, in e eb a e di e si y, and
ha dness. The e is a posi i e ela ionship be ween salini y and ha d-
ness wi h heigh . We see ha in e eb a e di e si y does no mean-
ing ully change wi h heigh . Dashed line ep esen s 95% CI. GLM
line i ed wi h a y ~ x o mula.