1
Supplemen a y da a o Nokelainen e al. 2019 Func ional Ecology
Ti le: Imp o ed camou lage h ough on ogene ic colou change con e s educed
de ec ion isk in sho e c abs
Au ho s:
Ossi Nokelainen1,2*, Ru h Maynes1, Sa a Myno 1, Na asha P ice1 & Ma in S e ens1*
1Cen e o Ecology and Conse a ion, College o Li e and En i onmen al Science,
Uni e si y o Exe e , TR10 9FE Pen yn, UK
2Depa men o Biological and En i onmen al Science, Uni e si y o Jy äskylä, P.O. Box
35, 40014 Uni e si y o Jy äskylä, Finland
* Au ho s o co espondence ([email p o ec ed], ma in.s e ens@exe e .ac.uk)
Copy igh : All au ho s lis ed abo e and Uni e si ies o Exe e (Uni ed Kingdom) and
Jy äskylä (Finland). Published by B i ish Ecological Socie y.
Licence: C ea i e Commons A ibu ion 4.0 In e na ional (CC-BY 4.0)
Con en s: Page
Summa y 2
Me hods 3
Colou change expe imen 3
Pho og aphy and ision modelling 6
Visual p eda ion compu e de ec ion expe imen 8
S a is ical analyses 12
Re e ences 14
2
SUMMARY
Many animals change appea ance wi h age bu he easons why a e a ely es ed.
Common sho e c abs (Ca cinus maenas), o example, a e known o hei abili y o
change colou o e ime. Young c abs show ema kable a ia ion in colo a ion and i has
been sugges ed ha hei a iable appea ance may help hem o hide om p eda o s in he
habi a s hey use. Howe e , as c abs g ow hey become mo e mobile and adul c abs, in
con as , a e known o possess a mo e uni o m colo a ion. This c ea es a p oblem: how o
emain hidden in habi a s ha a e a iable and e y di e en in appea ance?
To answe his, we i s ea ed young sho e c abs o wo shades, pale o da k, on wo
backg ound ypes esembling di e en habi a s o 10 weeks. We p edic ed ha c abs
would adop a colo a ion ha would imp o e hei ma ch o he backg ound, bu ins ead
hey all de eloped a da k g een o b own appea ance. Nex , we unde ook an expe imen
a he Na u al His o y Museum London du ing he Colou and Vision exhibi ion, whe e
isi o s sea ched o c abs ep esen ing na u al colou a ia ion om di e en habi a s.
Rema kably, c abs we e no ha des o ind agains hei o iginal habi a , bu ins ead he
da k g een o ‘mud la c abs’ we e ha des o de ec agains all backg ounds.
These indings sugges ha sho e c abs change in o simila colou wi h age, which
appea s o be a good gene al solu ion o ma ch se e al habi a ypes and he e o e o hide
om p eda o s. We conclude ha he e olu ion o camou lage can be be e unde s ood
by ecognizing ha he op imal appea ance o hide may change o e he li espan o many
animals.
3
METHODS
Colou change expe imen
The expe imen was conduc ed a he Uni e si y o Exe e , Pen yn Campus, Co nwall
be ween Feb ua y and May 2016. Indi idual c abs used o he common ga den
expe imen we e collec ed om he Gyllyng ase beach (coo dina es in decimal deg ees:
50.141888, -5.063811), Co nwall, UK, du ing Feb ua y 2016. Sho e c abs a e loca ed in
a wide ange o habi a and subs a e ypes a ound he sho e, each wi h di e en
appea ances, including es ua ies, mud la s, sandy beaches, shingle, pebbles, mussel beds,
and ocky coas line (Edwa ds 1958, C o he s 1968, B ian e al. 2006, Todd e al. 2006,
2012, S e ens e al. 2014b). The collec ion me hods la gely ollow es ablished p o ocols
(S e ens e al. 2014b, Nokelainen e al. 2017a). B ie ly, he c abs we e collec ed by hand
du ing low ide alongside he beach om app oxima ely 50 me e s leng h and hus ou
sampling included c abs om di e en subs a es. C abs we e anspo ed om nea by
idal pools in o he labo a o y immedia ely a e cap u e. C abs en e ing he expe imen
we e all o simila size, app oxima ely 15 mm ca apace wid h. A e collec ion, c abs
we e pho og aphed and di ided in o expe imen al g oups based on hei ca apace
ligh ness in a andomized block design (i.e. c abs wi h con as ing ligh ness we e equally
ep esen ed in ea men g oups, see u he ). C abs we e pho og aphed once a week and
a e moul ing. Sho e c abs a e no a p o ec ed species and all wo k was conduc ed unde
app o al om he Uni e si y o Exe e Biosciences e hics commi ee (applica ions
2013/75 and 2014/556). The ield loca ions a e publicly accessible; no u he pe mi s
we e needed.
Fi s we s udy i ju enile sho e c abs adjus hei appea ance (i.e. including bo h
colou and pa e n) wi hin and o e successi e moul s in o de o inc ease hei
4
esemblance o he e ogeneous subs a es (unlike ou p e ious wo k, which has ended o
ocus on mo e simpli ied uni o m backg ounds; S e ens e al. 2014a, S e ens 2016).
Expe imen al animals we e di ided in o ou ea men g oups using a 2 x 2 ac o ial se
up wi h c abs o wo shades (pale, da k) on wo na u alis ic backg ound ypes (i.e. ock
pool and mud la – Fig. 1). Ca apace b igh ness was used o di ide c abs in wo dis inc
g oups. G oup disc e eness was u he alida ed based on he came a-ob ained spec al
da a (see below; ANOVA o ca apace b igh ness be ween da k and pale ea men
g oups, N = 60, F = 34.15, d =1, p < 0.001). Beginning wi h wo unambiguous g oups
allowed us o con ol o he ex ensi e pheno ypic a ia ion o ju enile c abs.
We chose backg ound ypes in which o ea c abs ha ep esen wo common
na u al ex emes: ela i ely homogeneous mud la and, mo e he e ogeneous ock pool
backg ounds. We eplica ed hese backg ounds using s anda d aqua ium g a el
(UNIPAC) a e subjec i e e alua ion o hei gene al p ope ies o colou and pa e n
om pho og aphs. ‘Mud la ’ backg ound was a mix u e o b own and g een (i.e.
ep esen ing b own mud and g een algae) aqua ium g a el (1:1 a io), whe eas ‘ ock
pool’ backg ound was a mix u e o black, g ey, whi e and pu ple aqua ium g a el (wi h
equal a ios). We delibe a ely chose no o use ac ual na u al subs a es as his may
con ain chemical cues o p eda o s o o he s imuli ha may in luence c ab de elopmen
and ha may also di e in ex u e / size as well as colou pa e n, he eby hinde ing ull
con ol o e he expe imen . Using a i icial g a el also enabled g ea e s anda disa ion o
backg ound samples among indi iduals. We compa ed he ma ch o ou a i icial
backg ounds o na u al ones using calib a ed pho og aphic da a (see below). Simila i y o
he backg ounds in a ich oma ic RGB colou space was calcula ed based on e lec ance
da a o b igh ness (i.e. a e age e lec ance ac oss all colou channels; R+G+B / 3) and
5
hue (i.e. ed di ided by blue channel). A i icial backg ounds ep esen ed simila albei
no pe ec ly ma ching na u al a ia ion o colou ul idal en i onmen s (Fig. S1). In
pa icula , he a i icial backg ounds mos e ec i ely ma ched he b igh ness o hei
na u al coun e pa s. In na u e, ock pools ha bou a g ea ange o ch oma ic a iabili y,
bo h wi hin and among pa ches, including pink-colou ed elemen s such as ed enc us ing
co alline algae and also ha e blue-colou ed elemen s such as mussels. Mud la s ins ead
a e cha ac e ised by b own ones o we soil and g a el and ge mixed by g een b own
and ed algae. The e o e, al hough ou a i icial subs a es a e no a pe ec ma ch o he
na u al subs a es, hey a e b oadly ep esen a i e, and c ucially, he appea ance o he
mud la and ock pool ea men s is e y di e en .
Al oge he , we ea ed 60 c abs (17 in ‘da k-mud’ ea men , 16 in ‘da k- ock’
ea men , 13 in ‘pale-mud’ ea men and 14 in ‘pale- ock’ ea men ) in cus omized
aqua ium anks (90 x 45 cm in a ea) o 10 weeks. Each ank was di ided in o 24 simila
sec ions (11 x 15 cm). The sec ion walls we e glued using adhesi e silicon glue and walls
con ained a mesh-co e ed hole ensu ing wa e ci cula ion h ough he sys em. Tanks we e
illed wi h dechlo ina ed ap wa e mixed wi h a i icial sea sal (Aqua ium Sys ems
Ins an Ocean Sal , Swell UK L d., UK) o simula e na u al seawa e , which was es ed
wi h a e ac ome e (D&D's Re ac ome e , Swell UK L d., UK) o ensu e salini y o 30
pp . The wa e was passed h ough a il a ion sys em (Eheim classic 350 EHEIM GmBH
& Co. KG, Deizisau, Ge many) and coole (D&D DC300 aqua ium coole 300w cooling
powe , Swell UK L d., UK), keeping he wa e bo h clean and a a cons an empe a u e.
Tempe a u e was se o 16°C o mimic local sea empe a u e a he ime o collec ion.
Two sec ions we e no used o accommoda e c abs, bu ins ead housed he inpu s and
ou pu s o he il a ion sys em o allow o maximum wa e low h ough each sec ion o

6
he ank. An ai s one (Aqua line High Ou pu Ai Comp esso , 2880 Li e/Hou ) was
accompanied wi h he il e ou pu sec ion o allow as much oxygen o low h ough he
ank as possible. We used wo dayligh lamps and one nea UV lamp (G obeam600
Ul ima and AquaBeam 600 Ul ima MW, T opical Ma ine Cen e UK) o simula e na u al
ligh condi ions, which we e con olled by a ime o es ablish a cons an ligh cycle
(12:12 L/D-cycle). C abs we e ed daily wi h s anda d ma ine c us acean aqua ium ood.
Wa e was changed, il e s checked and anks cleaned weekly o main ain li ing
condi ions o c abs. Some c abs did no su i e h ough 10-week-expe imen . Howe e ,
mo ali y was no signi ican ly di e en wi h ega ds o backg ound ype o c ab ini ial
shade, no he e was di e ence in moul ing a es be ween he ea men s.
Pho og aphy and ision modelling
Pho og aphy, ini ial image calib a ion and analysis b oadly ollowed p e iously used
me hods (S e ens e al. 2014a). Full de ails a e gi en in supplemen a y ma e ial (Table
S1). B ie ly, imaging was unde aken wi h a Samsung NX1000 digi al came a con e ed
o ull spec um wi h no qua z il e o enable UV sensi i i y, and i ed wi h a Nikon EL
80 mm lens. Fo he human isible pho os, we placed a UV and in a ed (IR) blocking
il e in on o he lens, which ansmi s wa eleng hs only be ween 400 – 680 nm
(Baade UV/IR Cu Fil e ). Fo he UV images, a UV pass and IR blocking il e was
used (Baade U il e ), which ansmi s be ween 320-380 nm. G ey e lec ance s anda ds,
which e lec ligh equally a 7% and 93% be ween 300 and 750 nm, we e used.
Fo each image we measu ed he en i e do sal side o he c ab ca apace o ob ain
colou and pa e n in o ma ion. We analysed he da a bo h wi h no malised came a
esponses and ish ision modelled da a (see below). Fo e lec ance da a (i.e. colou ), we
7
used no malised came a esponses o b igh ness, ed, g een, blue and UV channel. The
pa e n analysis echnique (a ‘g anula i y’ analysis) in ol ed decomposing an image in o
a se ies o di e en spa ial equencies (‘g anula i y bands’) using Fou ie analysis and
band pass il e ing, ollowed by de e mining he ela i e con ibu ion o di e en ma king
sizes o he o e all pa e n (Ba bosa e al. 2008, Hanlon e al. 2009, S odda d and S e ens
2010). Fo he pa e n da a (see u he de ails in supplemen a y Table S1), we used
maximum powe (i.e. pa e n dominance – he ene gy a he spa ial equency wi h he
highes pixel ene gy), p opo ional powe (i.e. pa e n di e si y – maximum o peak
ene gy alue di ided by he summed ene gy), o al powe (i.e. o e all con as o
ampli ude – he ene gy summed ac oss all scales) and mean powe (i.e. a e age con as
ac oss he spec um). Pa e n analysis was conduc ed in cus om iles o Image J
(T oscianko and S e ens 2015).
To examine he le el o backg ound ma ch, we calcula ed how changes in he c ab
ca apace in luenced hei le el o ma ch o he expe imen al backg ounds. To do so, we
used a ecep o noise limi ed isual disc imina ion model (Vo obye e al. 1998), which is
based on di e ences in colou o luminance based on pho on ca ch alues. Fo
calcula ions, all c abs we e pho og aphed weekly o e he cou se o he expe imen . Also,
he backg ounds (i.e. aqua ium g a el mix u es om he slo s indi idual c abs we e kep
on) we e pho og aphed. Thus, di e ence me ics (see below) we e calcula ed be ween
c ab ca apace and he e y backg ound each c ab was ea ed on ma ching he size o he
en i e slo (c. 10 cm in diame e ). We used a ish ision model based on he longwa e
(LW) and sho wa e (SW) isual sensi i i y o he pollack (Pollachius pollachius)
(Shand e al. 1988). A Webe ac ion alue o 0.05 was used o he mos abundan cone
ype wi h ecep o cone a ios o SW 168 and LW 339 (Go a do skii e al. 2000). The
8
ecep o noise model yield alues in ‘jus no iceable di e ences’ (JNDs), whe eby
di e ences be ween 1 and 3 a e in e p e ed ha wo s imuli a e unlikely o be
disc imina ed by an obse e (and hence indica e a good backg ound ma ch). La ge
alues han his a e inc easingly likely o be disc iminable, whe eas alues lowe han
his (<1 JND) should be i ually indis inguishable (Kelbe e al. 2003, Siddiqi e al.
2004, Olsson e al. 2015). Cau ion mus be used in in e p e a ion o JNDs, because he
me hod is sensi i e o es ima es o ecep o noise, ligh condi ions and animal cogni ion.
As such, we ollow pas wo k and use a sligh ly b oade egion o unce ain y in
disc imina ion h esholds (1-3 JNDs), bu ul ima ely he key conside a ion is ha smalle
JND alues should equa e o be e camou lage ma ch.
Visual p eda ion compu e de ec ion expe imen
To es camou lage e icacy o di e en c ab pheno ypes in a ied backg ounds, we made
a p eda ion game whe e human pa icipan s sea ched o c abs o a ious sizes p esen ed
on a ouch sc een. Ou main ques ions we e: does he isual complexi y o he
backg ound make i ha de o ind he p ey, and a e c abs ha des o ind agains hei
local habi a ype (i.e. consis en wi h a backg ound-speci ic camou lage hypo hesis)?
To ob ain c ab and backg ound images o he game, we sampled c abs om nine
loca ions a ound Co nwall in he sou hwes UK and pho og aphed hem. These in e idal
si es ep esen backg ounds o di e en isual complexi y (wi h highe complexi y
in ol ing subs a es o many ex u es, con as s, colou s, shapes, and di e en -sized
g anules). He e, ock pools ep esen subjec i ely he mos isually complex (A-C),
mussel beds medium (D-F), and mud la s he simples (G-I) si es. Si es we e: A)
Falmou h (all coo dina es in decimal deg ees, 50.141888, -5.063811) on he sou h coas ,
9
comp ising a s e ch o sho eline collec i ely encompassing Cas le and Gyllyng ase
beaches. Si es hold ock pools wi h ocky c e ices wi h s ony o g a el subs a es in he
pools and, lowe down on he sho e, inc easing abundance o seaweed. B) Summe s
beach a S . Mawes (50.157095, -5.017370), on he sou h coas comp ising ock pools,
g a el, and some low seaweed co e adjacen o a pebbled beach. C) Flushing
(50.162191, -5.066843), on he sou h coas comp ising ock pools, g a el, and seaweed
co e . D) God e y Poin (50.249499, -5.320966) on he no h coas , which p ima ily
consis s o exposed ocky ou c ops wi h mussel beds. E) Polzea h (50.576169, -
4.920206), on he no h coas o Co nwall, comp ising mos ly mussel bed co e adjacen
o a beach. F) Mawgan-Po h (50.466705, -5.041101), on he no h coas o Co nwall,
comp ising mos ly mussel bed co e and pools adjacen o a beach. G) Hel o d Passage
(50.098763, -5.132556), an es ua ine loca ion on he sou h coas has a la ge mud la a ea
as well as ie ed c aggy ock pools. H) Pen yn (50.166956, -5.082634), mos ly mud la s
wi h a co e ing o g een algae. I) Hayle (50.188010, -5.428120), on he no h coas o
Co nwall, an es ua ine loca ion has a la ge mud la a ea.
Fo he game, c abs as well as he na u al backg ounds om he ield si es we e
pho og aphed using he me hods desc ibed abo e. B ie ly, we used calib a ed Samsung
NX1000 equipped wi h Nikon EL-80 mm Nikko and Nikon D7000 came a wi h a 60
mm Coas al Op ics lens. The c abs we e de ached om he backg ound using GIMP2
image manipula ion so wa e and he backg ound images we e c opped o 16:9 aspec
a ios o he ouch sc een game. C abs we e scaled in o he same pixel/mm aspec a io
o show c abs agains he backg ound images in na u al size wi h espec o he
backg ound scale. Due o he numbe o c ab images needed, cus om so wa e was
designed (called 'au oc ab') o au oma e he p ocess o backg ound sub ac ion. This