Improvement of individual camouflage through background choice in ground-nesting birds

Transcription

1 This is he pre-proof acceped version of he manuscrip, he full version can be accessed for free here. Improvemen of individual camouflage hrough background choice in ground-nesing birds Auhors: Marin Sevens*1, Jolyon Troscianko1, Jared K. Wilson-Aggarwal1, and Claire N. Spoiswoode2,3 1 Cenre for Ecology & Conservaion, College of Life & Environmenal Sciences, Universiy of Exeer, Penryn Campus, Penryn, Cornwall, TR10 9FE, UK. 2 Deparmen of Zoology, Universiy of Cambridge, Downing Sree, Cambridge CB2 3EJ, UK 3 DST-NRF Cenre of Excellence a he FizParick Insiue of African Ornihology, Universiy of Cape Town, Rondebosch 7701, Souh Africa. *Corresponding auhor: marin.sevens@exeer.ac.uk Absrac Animal camouflage is a longsanding example of adapaion. Much research has esed how camouflage prevens deecion and recogniion, largely focusing on changes o an animal's own appearance over evoluion. However, animals could also subsanially aler heir camouflage by behaviourally choosing appropriae subsraes. Recen sudies sugges ha individuals from several animal axa could selec backgrounds or posiions o improve concealmen. Here, we es wheher individual wild animals choose backgrounds in complex environmens, and wheher his improves camouflage agains predaor vision. We sudied nes sie selecion by nine species of ground-nesing birds (nighjars, plovers and coursers) in Zambia, and used image analysis and vision modeling o quanify egg and plumage camouflage o predaor vision. Individual birds chose backgrounds ha enhanced heir camouflage, being beer mached o heir chosen backgrounds han o oher poenial backgrounds wih respec o muliple aspecs of camouflage. This occurred a all hree spaial scales esed (a few cm and five meers from he nes, and compared o oher sies chosen by conspecifics), and was he case for he eggs of all bird groups sudied, and for adul nighjar plumage. Thus, individual wild animals improve heir camouflage hrough acive background choice, wih choices highly refined across muliple spaial scales. Keywords: camouflage; animal behaviour; ani-predaor; predaion; animal coloraion; birds 1

2 Animal camouflage has played an imporan role in sudies of naural selecion and adapaion 1-7. Recenly, considerable inerdisciplinary research has esed he mechanisic basis of how differen ypes of visual camouflage work5, 8. This has included field sudies using arificial (human-made) prey, laboraory work and compuer experimens wih birds and human observers, and heoreical models esing how differen ypes of camouflage funcion o preven deecion or recogniion, and esing opimal camouflage sraegies In addiion, oher research has invesigaed camouflage wih regards o colour change and he molecular mechanisms of geneic adapaion 18-20, demonsraing how camouflaged appearance can be uned o differen visual backgrounds. Individuals could also subsanially aler he effeciveness of heir camouflage hrough choosing appropriae subsraes. Such specialized behaviour has long been suspeced: Wallace noed ha leaf-mimicking Kallima buerflies resed in places ha faciliaed heir camouflage 21. Selecion of specific background ypes may be one way of dealing wih visually variable environmens, and differs from sraegies ha srike a compromise beween camouflage o wo or more background ypes while maching none closely12, 22. Early work by Kelewell, alongside his famous peppered moh (Bison beularia) selecion experimens, esed wheher melanic and ypical morphs choose appropriae resing backgrounds based on heir coloraion, showing ha melanic individuals came o res on black sripes, and ypical morphs on whie sripes 23. Furher work confirmed and exended hese findings, demonsraing ha differen morphs of several species of moh selec backgrounds ha are consisen wih heir coloraion The above sudies show ha individuals of a species, or of a discree morph, can selec appropriae backgrounds ha mach heir species- or morph-specific appearance. In conras, many animals show considerable, seemingly coninuous, wihin-species variaion in coloraion. In such cases, we may expec o find individual-level subsrae choice in accordance wih individual phenoype. Consisen wih his, laboraory sudies have shown ha Japanese quail (Cournix japonica) chose o lay heir eggs on background subsraes which provide improved camouflage for heir own individual egg appearance27. Furhermore, a sudy of wall lizards (Podarcis erhardii) across a range of Greek islands showed ha lizards were more likely o be found resing on backgrounds ha improved heir own individual camouflage compared o resing sies chosen by oher lizards on he same island28. This effec was sronger for female lizards (whose coloraion is more uned for camouflage han ha of males 29), and on islands where he risk of predaion is higher. There is also evidence ha subsrae choice for appropriae backgrounds may occur in individuals of species ha can change colour for camouflage, such as fish and prawns30,31. Animals may also adjus heir own alignmen o improve camouflage. Sargen suggesed ha mohs could modify heir resing posiions o coincide wih he exure of he background 32. This idea has been confirmed in recen sudies showing ha mohs released ono rees adjused heir body posiion and orienaion from heir original landing sie and alignmen, which increased heir camouflage33,34. Moreover, individual mohs were more likely o adjus heir resing posiion when heir iniial level of camouflage (judged by human observers) was lower35. The poenial for changes in body posure for camouflage exiss widely in naure, from sick insecs o culefish; he laer, for example can use visual cues o adjus he posure of heir arms o improve heir camouflage36. Alhough he above sudies clearly demonsrae he poenial for individual animals o selec appropriae backgrounds, a number of limiaions and quesions remain. Firs, mos sudies of background choice have judged camouflage based on human observaion or merics of coloraion relevan o human vision (bu see28, 34). Second, curren evidence for individual background selecion is limied o relaively few species and axa, wih he mos comprehensive work based on one laboraory sudy using simple arificial backgrounds, one species of wild lizard, and on conrolled releases of mohs by humans ono naural backgrounds. Third, pas work has no invesigaed subsrae choice a differen spaial scales wihin a habia. To dae, only one sudy (on lizards) has esed he hypohesis ha wild, free-ranging individuals in naural habias selec appropriae backgrounds for concealmen. While his sudy considered predaor vision, i only focused on one species ha moves around an area acively, and whose appearance is also complicaed by he need 2

3 for socio-sexual signals28, 29. Generally, we sill know comparaively lile abou how animals achieve successful camouflage in complex real environmens. Here, we sudied nine species of ground-nesing birds (nighjars, plovers and coursers) in Zambia (Fig 1), and used image analysis and modelling of predaor vision o es wheher individual appearance (he eggs from all nine species, and he plumage of adul nighjars ha si ighly on heir eggs) beer mached heir chosen backgrounds han oher poenial backgrounds. The differen species all nes in he ho Zambian dry season, siing on heir eggs and fleeing when a predaor approaches (see below). They uilise differen visual backgrounds, ranging from dry leaf lier o scorched bare ground, wih lile o no nes srucure buil. We sudied background (nes sie) selecion a hree spaial scales corresponding o adjacen backgrounds approximaely 5 m from he nes and 5 cm from he nes, and he backgrounds of nes sies chosen by conspecific individuals (Fig 2). The firs (m) and second (cm) spaial scales allow us o es wheher individuals refine heir nes sie selecion owards a given subsrae area or pach, and hen choose a specific fine-scale poin wihin ha pach. The hird spaial scale allows us o deermine wheher individuals choose nes sies ha improve heir own individual camouflage compared o oher poenial nes sies chosen by oher individuals. Our pas work has shown ha he degree of camouflage in hese birds direcly influences he likelihood of nes survival agains wild predaors 37. Specifically, nes survival in nighjars is prediced by adul camouflage (since aduls si ighly on he nes unil a hrea is very close), whereas nes survival in plovers and coursers (which flee early when a hrea arises) depends on egg camouflage. Furhermore, we have also recenly shown ha he birds modify heir escape disance based on heir individual level of camouflage: plovers and coursers show greaer escape disances when heir eggs are a poorer mach o he background (wih good camouflage, aduls can flee lae because he eggs are sill well hidden), and nighjars show greaer escape disances when heir own adul plumage maches he background less effecively 38. This suggess ha, as wih quail 27, individual birds seem o be able o assess heir eggs level of camouflage (or ha of heir own plumage), and use his in heir decision-making. Coupled wih his evidence, our sudy sysem is ideal for invesigaing camouflage and microhabia choice because he background environmen where he birds nes is unambiguous, and because each species ness in areas ha differ wih respec o he range of subsraes encounered and he ype of visual background. Resuls Based on camera-rap recordings of predaion evens a our moniored ness, we modelled camouflage wih respec o he visual sysems of he main predaor groups, specifically erachromaic birds, richromaic primaes, and dichromaic mammals. We chose hese predaor visual sysems based on recordings of egg predaion by diurnal predaors including grey-headed bush-shrikes (Malaconous blanchoi), verve monkeys (Chlorocebus pygeryhrus), and banded mongooses (Mungos mungo)37. Modelling of predaor vision and image analysis allowed us o quanify several merics of camouflage, including colour, luminance (perceived lighness), and paern We used a range of merics ha have previously been shown o predic survival of ness in he field and escape behaviour by aduls, as well as predicing human deecion imes of camouflaged arificial arges (see Mehods). Daa for plovers and coursers were analysed separaely from daa for nighjars because plovers and coursers (Charadriiformes) flee he nes early in response o a poenial hrea, exposing he eggs o poenial deecion for long periods. In conras, adul nighjars (Caprimulgiformes) si igh on heir nes unil a predaor is close, such ha adul plumage camouflage is also crucial for egg survival37. Firs, since here is variaion in egg and plumage appearance among individuals, he hypohesis ha nesing birds choose microhabias ha improve heir camouflage predics ha birds should choose nes sies ha improved heir own specific camouflage compared o sies seleced by conspecific individuals. Second, he hypohesis predics ha background selecion for camouflage should occur a wo furher spaial scales: over several meres (5 m away, and represening nes sie selecion wihin a given area of a habia) and 3

4 wihin a few cm of he chosen nes area (phoographed from direcly above he cluch) represening he specific subsrae pach chosen wihin a sie; see Mehods and Fig 2). Furhermore, based on pas work37, 38 he hypohesis predics ha microhabia choice would be based on egg camouflage in plovers and coursers, and on boh egg coloraion and adul plumage camouflage in nighjars. As prediced, individuals of a given species chose nes sies ha were a beer mach o heir own eggs han o he nes sies of oher individuals of he same species (all resuls summarized in Table 1; Fig 3). Egg paern mach in plovers and coursers (n 92 ness) showed an ineracion beween nes sie and species (in linear mixed-effec models; see Mehods): hree-banded courser, crowned plover, and Temminck's courser eggs mached he paern of heir chosen nes sie beer han he nes sies of conspecifics, whereas hree-banded plovers, bronze-winged coursers and waled plover eggs did no (Table 1, Fig 3). Plover and courser egg luminance and colour maches did no significanly differ beween heir own versus conspecific nes sies. For nighjars (n 105 ness), egg paerns mached heir own nes sie beer han he nes sies of conspecifics. For egg luminance, an ineracion showed ha Mozambique and pennan-winged nighjars mached heir own nes sie luminance beer han he nes sies of conspecifics, whereas fiery-necked nighjars insead mached he luminance of conspecifics nes sies beer han heir own. Nighjar egg colour maches did no differ beween heir own and conspecific nes sies. Nex, we esed wheher individual egg camouflage was a beer mach o i) he chosen background versus comparable areas 5 m away, and ii) he chosen background wihin ca. 5 cm of he cluch versus a comparable area immediaely surrounding he nes. This resuled in four zones, wih wo comparisons planned beween hese zones (i.e. he few cm scale and 5 m scale, Fig 4). For plover and courser eggs (n 92 ness), paern and luminance maching varied significanly beween background zones (F3, , P < and F3, , P < respecively). In all cases bar one, eggs were beer mached o he chosen background han o non-chosen backgrounds a few cm away and 5 m away (Fig 4; see saisical resuls in Table 1). However, eggs mached he luminance of heir chosen background less closely han he background a few cm away. This suggess a poenial rade-off resuling in paern-maching being seleced over luminancemaching. Furher modelling demonsraed a significan negaive correlaion beween paernmaching and luminance-maching in he zones neares he egg ( immediae and near surrounds in Fig. 2), bu his correlaion flaened ou in zones furher from he nes cenre ( disan surrounds, model ineracion beween zone and luminance-maching level: F2, , P < 0.001; comparison beween immediae and disan zones: 6.823, P < 0.001). This finding shows ha paern and luminance maching are independen of each oher in randomly seleced paches of habia. However, he areas chosen as nes sies mached he paerns of he eggs beer han expeced, and mached luminance worse han expeced. This finding is consisen wih he hypohesis ha plovers and coursers selec nes sies ha mach he paerns of heir eggs a he expense of maching he luminance. Egg colour mach did no differ beween chosen and nearby non-chosen backgrounds a eiher spaial scale. Nighjar eggs (n 105 ness) were beer mached wih respec o boh paern (F3, , P < 0.001) and luminance (F3, , P < 0.001) o heir chosen backgrounds han o backgrounds a few cm away and 5 m away (Table 1; Fig 4). An excepion was nighjar egg luminance, which was beer mached o non-chosen backgrounds 5 m away han o heir chosen nes sie. As above, we esed wheher his represens a rade-off beween paern-maching and luminance-maching across he zones, bu here was no significan ineracion beween zone and luminance-difference (F2, , P 0.161). Nighjar egg colour mach was prediced by a significan ineracion beween predaor visual sysem and zone (F4, , P < 0.001): egg colours were beer mached o he chosen nes sie han o backgrounds a few cm away for all visual sysems, bu he effec was mos pronounced for he dichromaic mammalian visual sysem. However, nighjar eggs were less well mached o he chosen background under richromaic and dichromaic mammalian vision, whereas hey were beer mached under avian erachromaic vision. Zone remained a significan predicor in he model on is own (F3, , P < 0.001), showing overall ha eggs were a beer mach o heir chosen background han o he adjacen 4

5 background a few cm away, bu here was no significan difference in mach o he chosen background versus oher poenial backgrounds 5 m away. Finally, since nighjars flee from he nes only when a predaor is nearby (our daa show a mean flush disance across all hree nighjar species of 1.9 m ± 1.3 sandard deviaion 38), we expeced nes survival o be more srongly affeced by parenal camouflage han egg camouflage 37. As prediced, nighjar plumage (n 98 aduls) mached he paern, luminance and colours of he individuals chosen backgrounds beer han hose of heir conspecifics backgrounds (Table 1). When comparing camouflage a a scale of cm and m (n 98 ness), he paern model reained an ineracion beween zone and species (zone x species: F6, , P 0.002; zone F3, , P < 0.001; luminance: F3, , P < 0.001; colour: F3, , P < 0.001; Fig 4; Table 1). The plumage paern of Mozambique and pennan-winged nighjars mached heir chosen nes backgrounds beer han he adjacen background a few cm away, whereas fiery-necked nighjars did no; individuals of all nighjar species mached heir chosen background for paern beer han oher poenial backgrounds 5 m away. For luminance, nighjars were beer mached o chosen backgrounds han o adjacen backgrounds boh a few cm away and 5 m away. Plumage colour was a beer mach o chosen nes sies a he mere scale, bu no he cm scale. Discussion Our sudy demonsraes ha individual camouflage in muliple species of free-ranging wild animals can be improved no jus by changes in an animal s own appearance, bu also by behavioural selecion of appropriae resing backgrounds as viewed hrough heir predaors eyes. Nine species of ground-nesing bird (based on over 90 individual plover/courser ness and over 100 nighjar ness) were capable of highly nuanced nes sie selecion ha improved heir egg and plumage camouflage a several spaial scales. Firs, hey were able o choose a suiable nesing pach in he general habia a a scale of approximaely five meres, and second, hey refined heir nes sie selecion wihin ha pach a a very fine scale (wihin a few cm). Moreover, because females seleced sies ha mached heir own eggs and plumage beer han hose of heir conspecifics, decisions were made wih reference o heir own individual phenoype raher han following a general species-wide sraegy. Our findings are consisen wih hose of a recen laboraory sudy on subsrae selecion in nesing quail27, and also ie in wih our recen sudy of escape disances in he presen sudy sysem. The laer showed ha individuals modulae escape behaviour based on heir level of camouflage, providing furher evidence ha nesing birds can modify heir behaviour in response o perceived levels of concealmen38. These findings also concur wih a sudy of island populaions of wall lizards, which found ha individuals were more likely o be found siing on backgrounds ha provide beer camouflage han on oher poenial sies 28. Our work here shows ha he benefis of microhabia choice and behavioural changes based on assessmen of individual camouflage exend across a wide range of avian species, several spaial scales, and wo life hisory sages (aduls and eggs). The above resuls demonsrae ha behavioural choice of subsraes and backgrounds may offer a major roue o enhancing camouflage, and sugges ha sudies ha simply compare he camouflage of individuals agains random background samples may someimes yield inaccurae findings if individuals vary boh in appearance and subsrae preference. Many camouflaged species show eiher discree polymorphisms or high levels of coninuous phenoypic variaion 3, 9, 19, 42, 43. In such cases i may be paricularly beneficial for individuals o have corresponding subsrae preferences, boh o improve heir level of camouflage and perhaps o increase he range of microhabias exploied. Microhabia choice may also have imporan broader evoluionary consequences. For example, some insec species comprise several morphs ha occur on differen hos plans, and disrupive selecion agains inermediaes may poenially drive speciaion hrough reproducive isolaion of populaions in sympary42. We canno enirely rule ou he possibiliy ha our resuls could be parially explained by predaion having eliminaed poorly-camouflaged ness from our daase before we could record 5

6 hem. While his would iself be an imporan finding (given ha individual variaion in camouflage maching in wild animals has rarely been direcly demonsraed o affec predaion risk), we feel ha his explanaion is unlikely o fully explain our resuls. Firs, our daase includes sufficien variaion in camouflage among individuals o predic predaion risk (shown recenly using he same sample of ness37), which demonsraes ha we did no exclusively sudy individuals wih good camouflage. Second, a proximae mechanism of background selecion based on individual egg coloraion has already been esablished in conrolled laboraory experimens wih ground-nesing birds, implying ha his is a more parsimonious explanaion for our resuls 27. They do no preclude oher facors, beyond he scope of his sudy, influencing nes sie selecion in birds; hese may include habia visibiliy for deecing predaor approaches, hermal consideraions, and viciniy o oher nesing birds. In addiion, some of he plover and courser species in our sudy make modes modificaions ( scrapes ) o heir nes area, poenially influencing camouflage mach by modifying heir environmen via an exended phenoype44, 45. This is, however, highly unlikely o have affeced he majoriy of our findings (see Supplemenary Informaion). These facors may add o a rich complexiy of facors influencing background selecion in birds and oher animals. Beyond he quesion of how widespread background choice for camouflage migh be, here is much o be gained from rying o disenangle he mechanisms involved. In birds, egg coloraion appears o be srongly heriable, wih relaively lile environmenal influence 46. Exacly how birds make appropriae decisions is no ye clear bu we sugges i could arise hrough wo (no muually exclusive) mechanisms. Firs, if background choice is also heriable, a geneic correlaion could allow individuals wih a given egg phenoype o also inheri he appropriae subsrae preference. Alernaively, behavioural preferences could develop wih experience as birds learn wha heir eggs look like, and so o make appropriae decisions. The laer mechanism seems more likely because inheried behavioural choice would offer lile flexibiliy, and also because here is good evidence ha birds learn heir egg appearances in oher conexs. For example, hoss of brood parasies appear o learn wha heir own eggs look like in iniial breeding aemps, and hen rejec any subsequen parasiic eggs ha deviae from his emplae of appearance 47. Our resuls sugges ha oher birds likely have mechanisms o know no only wha heir own eggs look like, bu also heir own plumage, and use his informaion o make adapive decisions. A furher and relaed poenial mechanism could involve chicks imprining on specific backgrounds afer haching, and basing nes sie choice on his when hey laer become breeders. Fuure work could es hese poenial mechanisms by addressing wheher and how subsrae choice differs beween naïve firs-ime breeders and more experienced individuals. Conclusions Overall, camouflage can be enhanced no only hrough geneic or developmenal changes in individual appearance, bu also hrough individual behavioural choices. Thus, in many species he value and uning of animal camouflage may resul from a complex mixure of morphology, behaviour, and environmen. More broadly, our sudy underlines ha animals possess sensory and poenially cogniive mechanisms ha allow hem o improve he adapive value of heir own individual phenoype by choosing appropriae backgrounds. We should furher look for individual background choice in he many oher conexs where signalling success is affeced by aspecs of he environmen, such as conspicuous warning coloraion and sexual signals Mehods The sudy sysem, general mehods, and quanificaion of camouflage closely followed our pas work (including ha demonsraing how our camouflage merics predic survival of he ness of he birds we sudy here) and a range of pas and recen mehodological approaches 37, 38, 51. Our daase here overlaps wih our previous work wih respec o he individual birds recorded and measured, and o some of he images of he naural backgrounds used o assess camouflage, wih he addiion of furher comparison background images aken a 5 m scales used only for his sudy. 6

7 Sudy sysem The sudy sie comprised c ha around Musumanene and Semahwa Farms (cenred on 16 46'S, 26 54'E) and c. 400 ha on Muckleneuk farm (cenred on 16 39'S, 27 00'E) in souhern Zambia (Choma Disric), during Sepember November Fieldwork was underaken during he ho dry season, when an open undersorey affords nesing habia for he ground-nesing bird species we sudied. The field sies are in an agriculural region, bu he culivaed areas (primarily maize and obacco crops) are comparaively small and occur wihin a greaer area of naural habia (deciduous miombo woodland and grassland). As such, predaor communiies should no differ grealy from condiions occurring before han human impac on he region. We sudied hree nighjar species (fiery-necked nighjar Caprimulgus pecoralis, Mozambique nighjar Caprimulgus fossii and pennan-winged nighjar Macrodiperyx vexillaria), hree plover species (crowned plover Vanellus coronaus, waled plover Vanellus senegallus and hree-banded plover Charadrius ricollaris), and hree courser species (bronze-winged courser Rhinopilus chalcoperus, Temminck s courser Cursorius emminckii and hree-banded courser Rhinopilus cincus). Mos ness were found by local farm workers, deeced when he birds flushed on approach, or hrough nocurnal eye-shine from orchligh. Our sample of ness may lack he exremes of camouflage maching if we were unable o find he mos camouflaged ness, and if some of he leas camouflaged ness were aacked by predaors firs. However, our resuling sample should remain ecologically represenaive of he surviving ness, and indeed here was considerable variaion in survival and camouflage among hem37. All work was approved by he Universiy of Exeer Animal Ehics Commiee (applicaion number 2013/282) and conduced under licence from he Zambia Wildlife Auhoriy. The field locaions are privae land accessed wih he landowners permission, and no furher licenses or permis were needed. Phoography and vision modelling We ook digial images wih Nikon D7000 cameras, fied wih 105 mm Micro-Nikkor lenses, which ransmi ulraviole (UV) ligh. The cameras had undergone a quarz conversion (Advanced Camera Services Limied, Norfolk, UK) o allow sensiiviy o boh human-visible and ulraviole wavelenghs, involving replacing he UV and IR blocking filer wih a quarz shee o allow visual analysis hroughou he avian-visible specrum 51, 52. For phoographs in he human-visible par of he specrum, he lens was fied wih a Baader UV-IR blocking filer (ransmiing 420 o 680 nm). UV phoographs were aken using a Baader UV pass filer (ransmiing 320 o 380 nm). All images were aken a f/8, ISO400, in RAW forma, avoiding a ime period wihin wo hours of sunrise or sunse, and were only aken in direc sunligh as his corresponded mos closely wih weaher and ligh condiions during in he Zambian dry season and wih he diurnal predaion evens we recorded. During he brief crepuscular periods a our sudy sie, here would be changes in ambien ligh specra, background conras, and shadows, bu we canno es hose effecs wih our curren daase. To quanify adul nighjar camouflage, we closely followed pas work on he same sysem 37, 38. Images of nighjars siing on heir ness were aken from a sanding posiion from 5 m disance and he flank leas obscured by vegeaion. If boh sides were clearly visible, images were aken so as o avoid direcly facing he sun. Acquiring images of adul plovers and coursers was no possible because hese birds frequenly flush a long disances. Afer he adul nighjar was flushed from is nes, a 40% Specralon grey sandard (Labsphere, Congleon, UK) was placed beside he eggs and phoographed from 2 m using he same camera seings o hose for he aduls. This enabled us o conrol for lighing condiions in he adul bird images wihou he sandard needing o be in he same phoograph (he sequenial mehod53). Images of plover, courser, and nighjar eggs were acquired in siu from 1.25 m direcly overhead, as well as under more conrolled lighing; shaded from sunligh and agains a whie background wih he eggs nex o he grey sandard. We chose 5 m as he phoography disance for he meer scale because adul nighjars could reliably be phoographed a his disance wihou fleeing heir ness, and because conrol phoographs aken 5 m on eiher side of he nes did no overlap wih one anoher. For he fine cm scale, we 7

8 phoographed from direcly above he cluch because his included boh he larges cluches and surrounding nes sie area. To calibrae he images, all phoos were linearized o conrol for he non-linear response of he camera o ligh inensiy, and hen sandardized agains he grey sandard o remove effecs of he ligh condiions39. For he linearizaion and mapping o predaor vision, we measured he image values of eigh Specralon reflecance sandards wih values ranging from 2 99% (Labsphere) and produced polynomial linearizaion curves; all channels having R2 values Visible and UV phoographs were aligned and scaled using an auomaed scrip, minimizing he absolue spaial difference beween pixels. This accouned for focal lengh changes when re-focusing in UV, and minor shifs in camera posiion. Aligned images were saved as 16-bi TIFFs for he visible LW ( red ), MW ( green ) and SW ( blue ) channels, and UV. To model predaor vision, we chose appropriae visual sysems based on recorded predaion evens a a subse of ness from cusom buil moion-riggered cameras37. The diurnal predaors we recorded included animals wih hree differen visual sysems: dichromas (banded mongoose Mungos mungo), richromas (verve monkey Chlorocebus pygeryhrus and human), and erachromas (grey-headed bushshrike Malaconous blanchoi). We used he ferre Musela puorius as he closes available visual sysem for modelling banded mongoose vision. Ferre cone sensiiviies (absorpance daa) were obained from elecroreinogram flicker phoomery-based daa54 and used o model visual pigmen absorbance55, correced for ligh ransmission hrough he ocular media 56. Verve monkey cone sensiiviies are very similar o hose of humans57, and so we used human vision models here58. For birds, he greyheaded bushshrike likely has a viole sensiive (VS) visual sysem 59, and so we used represenaive peafowl Pavo crisaus sensiiviy daa for his visual sysem60. For each vision model, prediced cone cach values were obained by ransforming he images from camera o animal colour space wih a widely used mapping echnique 39, 51, 61. We used a daase of 3139 naural reflecance specra o model prediced camera and visual sysem responses. This comprised 2361 reflecance specra from he Floral Reflecance Daabase 62 and he remainder from specra of bird eggs, plumage, insecs, minerals, ree bark, and vegeaion colleced by us using an Ocean Opics USB2000 specromeer wih an Ocean Opics PX2 pulsed xenon lamp 51. All calculaions were based on daa from 300 o 700 nm in 1 nm incremens, under D65 illuminaion. Models were calculaed using cusom code in ImageJ 63 and R64. As has been demonsraed in muliple previous sudies, mapping from camera o animal colour space is highly accurae and wih very low error raes compared o modelling of phoon cach daa wih specromerye.g. 15, 51, 52. In fac, images much more accuraely accoun for illuminaing condiions and angles, and measure larger areas of he focal objec or scene, han is possible wih specromery, meaning ha using image analysis is likely even more accurae han purely based on is high correspondence wih specromery daa. The R2 values of he models convering from image values o animal cone cach values were all Cone cach images for each visual sysem were used for all subsequen image processing in 32-bis/channel floaing poin operaions, ensuring ha no daa were los due o sauraion ha can occur a he op of he 16-bi range. Image processing and analysis Our aim was o compare how closely he eggs of all species and adul plumage of nighjars mached heir chosen nes sie compared o oher poenial sies a differen spaial scales. We herefore compared egg/adul camouflage o he nes locaions ha individuals had chosen hemselves, relaive o nes locaions chosen by oher individuals of he same species. The conspecific comparison revealed wheher individual birds chose sies based on heir own specific egg/adul appearance. In addiion, we also compared how closely egg and adul birds mached chosen nes sies and poenial oher sies in he same area a a very fine (< 5 cm) and larger (5 m) scale. For he analyses of egg coloraion, our overhead phoographs were segregaed auomaically ino wo zones corresponding o he cluches' immediae surroundings (an area wih a radius of 400 8

9 pixels, or approximaely 4.4 cm, from he edge of he cluch) and neighbouring surroundings occupying he remainder of he image beyond 400 pixels from he arge. We chose an area of 400 pixels because his reliably encompassed he nes area (including any poenially modified subsraes around he nes) and cenre of he image. This comparison of he immediae chosen nes area o ha of an adjacen near pach enabled us o es nes sie selecion a a very fine scale. Nex, he amalgamaion of hese immediae and near zones, called he chosen 'local' zone was analysed and compared o he wo 'disan' conrol phoographs 5 m away. We used an amalgamaed zone raher han jus he immediae nes area so ha similarly sized areas could be compared. The second comparison of he chosen local pach wih disan paches enabled us o es microhabia choice a a larger scale wihin he general nes area. For he conspecific nes comparisons we compared he camouflage of he birds' chosen local zones o hose of oher individuals. Eggs were seleced using an egg-shape selecion ool65. For he adul nighjars, we also followed he above area selecions and comparisons, wih individuals seleced using he freehand selecion ool in ImageJ. Egg/bird edges were generally clearly visible, bu any obsrucing objecs were avoided o preven ambiguous areas of he arge or background from being measured (alhough his process was no underaken blind o our hypoheses, we minimized any subconscious bias by analysing nes phoos separaely from he 5 m conrol phoos, precluding any possibiliy of direc comparison). Adul nighjars could no be caugh and phoographed under conrolled (diffuse) lighing condiions. As such, colour and paern merics for adul nighjars were based on in siu images, and any local lighing effecs (such as dappled shadows) would have been cas on boh he adul bird and is surrounds. Alhough he normalisaion would conrol for overall colour and luminance differences, paern could have been affeced by dappled shadows. This effec should be mos pronounced in fiery-necked nighjars, which ofen nesed under dappled ligh, whereas Mozambique and pennan-winged nighjars ended o nes in open sies. However, in our previous sudy37, we found ha he paern mach beween adul nighjars and heir surroundings was he bes predicor of nes survival, suggesing ha our in siu measures of paern were a leas ecologically relevan. Eggs were phoographed under conrolled lighing, seleced from hese images and resized o mach he pixels/mm scale of he in siu surrounds in he background images (excluding he in siu eggs), using bilinear inerpolaion image reducion. We calculaed hree merics for camouflage maching: he level of colour, luminance (perceived lighness), and paern mach by he bird or egg o he relevan background sample. Our merics have been used in several pas sudies and, crucially, hey predic boh he survival of wild birds/ness by predaors in he field37, and he deecion imes of humans when searching for arificial arges on compuer screens66. They also relae o oher aspecs of behaviour linked o camouflage, such as escape behaviour by incubaing aduls38. Therefore, we can be highly confiden ha he merics used here provide ecologically relevan measuremens of camouflage. Paern and luminance merics used he luminance-channel image (following pas work40) because paern is widely hough o be encoded principally by achromaic vision67. Ferre luminance was based on he LW cone sensiiviy since hese are more abundan han he SW cones by 14:1 54. Human luminance was derived as (LM)/261, and he double cones were used o calculae peafowl luminance60. Naural backgrounds have luminance levels ha are spaially correlaed due o a mix of ligh and dark objecs, wih a roughly log-normal disribuion of inensiies. However, while animal paerns also demonsrae his spaial non-independence of inensiies, hey ofen have wo or more main levels (such as dark spos on a pale background in eggs). Parameric approaches are herefore no suiable for analysing hese muli-modal disribuions. As wih our pas work 37, 38, 66, luminance disribuion differences (Luminancediff) were calculaed as absolue differences in couns of he numbers of pixels in each arge (plover egg or adul nighjar plumage) o he relevan background a 32 linear levels of luminance spanning 0% o 100%37, 38: Luminancediff values qunaify o wha exen he egg or nighjar luminance values, o each visual sysem, mach he values of heir surrounds 37, 51. This meric overcomes he problems of relying on 9

10 mean values in daa ha are no normally disribued, and significanly predics human deecion imes of camouflaged objecs66. Paern differences were generaed based on Fourier analysis and bandpass filering via a granulariy specrum, following a wide range of pas sudies ha have used his approach o quanify animal paerns in quesions spanning avian brood parasiism and egg mimicry o camouflage in culefishe.g.40, 68, and deecion imes of humans searching for hidden camouflaged arges66. Paern differences were generaed wih Fas Fourier Transform bandpass filers a 17 levels (from 2 pixels, increasing exponenially wih he square roo of 2, up o 512 pixels), using he sandard deviaion of he luminance values a each spaial scale o derive he 'energy' a ha spaial scale. Fourier analysis and bandpass filering has been used in various previous sudies o analyse animal markings in erms of a granulariy specrum 40, 68. Nex, we calculaed overall energy differences across all spaial frequencies (Paerndiff) in a similar manner o Luminancediff, by summing he absolue differences in energy beween he arge and he background a each spaial scale s37, 38, 66: Differences in paern energy beween he samples over he spaial scales resuls in increased Paerndiff values. As such, Paerndiff quanifies how closely egg and plumage paerns mach he size and conras of hose background feaures37, 51. This meric should provide several advanages over pas approaches ha derive muliple descripive saisics from granulariy specra40, 68. For example, granulariy specra are ofen muli-peaked, such ha selecing only he main peak in he specrum discards poenially imporan informaion a oher scales. As wih luminance inensiies (above), he paern energy in adjacen scale bands is correlaed, resuling in smooh energy specra. Uilising jus one meric of paern mach also simplifies he saisical analysis and inerpreaion, and is well suppored by behavioural evidence as i prediced he likelihood of nighjar nes predaion in he same sudy sysem37 and human deecion imes of hidden arges 66. Our approach here is mos relevan o he concep of background-maching, where we expec he irregular paerns of he arge o mach he size and conras of hose in he irregular background across a range of spaial scales. I does no es a masquerade hypohesis where one would predic ha he sample should be recognized as a differen class of background objec (i.e. an objec recogniion ask); his would require he developmen of mehods o analyse and inerpre he phase componens of he Fourier specra, or anoher ype of analysis alogeher. Colour analysis was based on a log form of he Vorobyev-Osorio recepor noise model of colour discriminaion41 for esimaing jus noiceable differences (JNDs). Versions of his model are commonplace in sudies of animal coloraion (reviewed by 69). As is convenion, JNDs describe colour differences beween wo objecs in prediced discriminaion values, whereby values less han 1.00 can be inerpreed ha wo objecs are indiscriminable, and increasing values above his likely o resul in a greaer likelihood of deecion. Colour differences for boh adul nighjars and he eggs of all bird groups was he mean difference (in JNDs) beween he mos abundan colour in he camouflaged objec and all he colours found in is surrounds, weighed by coverage, as in our previous work37, 51. This approach is in principle, herefore, fairly sraighforward: firs we find he dominan colour in he prey, and hen we es how close his was on average o is background colours wih a weighing for he proporion of he background composed of ha colour (so if he colour of he prey was a good mach o a very small pach of he background, his would no coun as much as being a good mach o a colour covering a larger proporion of he background). This mehod is able o downplay he influence of small objecs in he background images ha were highly differen colours o he prey (such as he occasional green leaf), focussing on he maching of he larger background secions. We have used his approach previously in a sudy of nes covering behaviour in plovers, and shown i o be one of he bes predicors of nes maerial selecion in hose birds45. Overall, his approach allowed us o compare objec and background colours, using some of he mos advanced models of visual discriminaion currenly available. 10

11 Conrolling for Effecs of Nes Scrapes Nighjars and Temminck's coursers lay heir eggs direcly on he subsrae, making no modificaion o he surrounds. Plovers and bronze-winged coursers ofen make a shallow scrape, someimes conaining maerial from he immediae surrounds. Three-banded coursers someimes parially bury heir eggs during early incubaion. While he subjecive visual impression of he modified nes area in hese laer species is minimal, is modificaion could poenially affec our habia selecion resuls. However, his is highly unlikely o have affeced he majoriy of our findings, for wo reasons. Firs, he nighjars and Temminck's coursers do no underake any nes modificaion, meaning ha his could no have conribued o hose resuls a any spaial scale for hese species. Second, we re-ran our analysis for he conspecific comparisons and for he inermediae scale (pach choice a a scale of several meers) for plovers and coursers while excluding daa from he immediae nes area. To do so, we compared egg camouflage for he conspecific comparisons and comparisons a he m scale afer excluding he area poenially conaining any scrape (based on he larges scrape area we could deec from our daase), and he conclusions were unchanged. Therefore, a mos, nes modificaion could only have influenced he resuls for cerain plover and courser species a he fines (cm) scale (see Supplemenary Informaion). Saisical Mehods All saisical ess were performed in R v Due o repeaed measures using hree predaor visual sysems, linear mixed-effec models were run using he lme4 package v wih a Gaussian error srucure and fied wih resriced maximum likelihood 70. Species ideniy was included in each model as a fixed effec in order o saisically eliminae any beween-species differences and deec camouflage effecs shared beween all species, raher han explicily esing for beween-species differences (which would require phylogeneically conrolled saisics, and es differen hypoheses o hose ha we invesigae in his sudy). Full ineracion models were specified wih he above image merics as response variables, and hen simplified using he filmer funcion of he LMERConvenienceFuncions v2.5 using AIC o backwards-fi he fixed effecs on maximum likelihood models and forward-fi he random effecs. Year and nes ID were specified as random facors, alhough year was removed because i explained lile variance and did no improve he model fi. Comparisons of camouflage maching o he nes sies of conspecifics were based on large marices in which each cluch or adul was compared o each oher nes s surroundings, under all hree predaor visual sysems. We conrolled for his pseudoreplicaion in models by specifying boh he adul/cluch ideniy, and each nes surrounding s ideniy as random facors, and predaor visual sysem as a fixed effec. A variable denoing wheher he comparison was beween an individual s phenoype and is own nes surroundings, or a differen individual s nes surroundings, was hen used as a fixed effec for he ype of comparison ( sameordifferennes in he following example). An example of he full mixed model srucure was herefore: lmer(paerndifference ~ nighjarspecies * predaorvisualsysemspecies * sameordifferennes (1 nesid) (1 nessurroundid)). Model residuals were checked o verify assumpions of homogeneiy of variance and a normal error srucure, and variables were logransformed o mee hese assumpions. Conservaive degrees of freedom were used o calculae Pvalues from maximum likelihood models70. To ensure he srucure of hese models did no creae a lack of independence, causing ype I or II errors, we fied he models o randomly generaed dependen variables. The p-values for he sameordifferennes variable when repeaed on normally disribued random daa 1000 imes demonsraed a fla disribuion, implying model srucure did no affec our resuls. In ligh of he evidence suggesing a poenial rade-off beween egg paern maching and luminance maching in plover/courser eggs and nighjar eggs, we ran wo addiional linear mixed models o es wheher here was a negaive correlaion beween paernmaching and luminance-maching merics in he paches seleced for nesing. The following model srucure was specified: lmer(logpaerndifference ~ logluminancedifference * Zone predaorvisualsysemspecies (1 nesid)). 11

12 Daa availabiliy All daa generaed or analysed during his sudy are included in his published aricle (and is supplemenary informaion). Acknowledgemens J.T., J.W-A. and M.S. were funded by a Bioechnology and Biological Sciences Research Council (BBSRC) gran BB/J018309/1 o M.S., and a BBSRC David Phillips Research Fellowship (BB/G022887/1) o M.S., and C.N.S was funded by a Royal Sociey Dorohy Hodgkin Fellowship, a BBSRC David Phillips Fellowship (BB/J014109/1) and he DST-NRF Cenre of Excellence a he FizParick Insiue. In Zambia we hank he Bruce-Miller, Ducke and Nicolle families, Collins Moya and numerous oher nes-finding assisans and land owners, Lackson Chama, and he Zambia Wildlife Auhoriy. We also hank Ron Douglas for supplying lens ransmission daa for he ferre, and hree referees for consrucive commens. Auhor Conribuions All auhors designed and conceived he sudy. Fieldwork was conduced by J.T., J.K.W-A., and C.N.S. a a sudy sie se up by C.N.S.. Image analysis and vision modelling by J.T., J.K.W-A., and M.S., and he saisical analysis primarily by J.T. M.S. wroe he iniial manuscrip, which was reviewed and approved by all auhors prior o submission. Addiional Informaion Correspondence and requess for maerials should be addressed o M.S. Compeing ineress The auhors declare no compeing ineress. References Co, H. B. Adapive Coloraion in Animals. London: Mehuen & Co. Ld (1940). Diamond, J. & Bond, A. B. Concealing Coloraion in Animals. Harvard: Harvard Universiy Press (2013). Kelewell, H. B. D. Selecion experimens on indusrial melanism in he Lepidopera. Herediy. 9, (1955). Sevens, M. Cheas and Deceis: How Animals and Plans Exploi and Mislead. Oxford: Oxford Universiy Press (2016). Sevens, M. & Merilaia, S. Inroducion. Animal camouflage: curren issues and new perspecives. Phil. Trans. R. Soc. B. 364, (2009). Thayer, G.. H. Concealing-Coloraion in he Animal Kingdom: An Exposiion of he Laws of Disguise Through Color and Paern: Being a Summary of Abbo H. Thayer's Discoveries. New York: Macmillan (1909). Wallace, A. R. Darwinism. An Exposiion of he Theory of Naural Selecion Wih Some of is Applicaions. London: Macmillan & Co (1889). Sevens, M. & Merilaia, S. Animal Camouflage: From Mechanisms o Funcion. Cambridge: Cambridge Universiy Press (2011). Bond, A. B. & Kamil, A. C. Visual predaors selec for crypiciy and polymorphism in virual prey. Naure 415, (2002). Cuhill, I. C., Sevens, M., Sheppard, J., Maddocks, T., Párraga, C. A. & Troscianko, T.S. Disrupive coloraion and background paern maching. Naure 434, (2005). Merilaia, S. & Lind, J. Background-maching and disrupive coloraion, and he evoluion of crypic coloraion. Proc. R. Soc. B. 272, (2005). 12

13 Merilaia, S., Tuomi, J. & Jormalainen, V. Opimizaion of crypic coloraion in heerogeneous habias. Biol. J. Linn. Soc. 67, (1999). Rowland, H. M., Cuhill, I. C., Harvey, I. F., Speed, M. P. & Ruxon, G. D. Can' ell he caerpillars from he rees: counershading enhances survival in a woodland. Proc. R. Soc. B. 275, (2008). Schaefer, M. H. & Sobbe, N. Disrupive coloraion provides camouflage independen of background maching. Proc. R. Soc. B. 273, (2006). Sevens, M. & Cuhill, I. C. Disrupive coloraion, crypsis and edge deecion in early visual processing. Proc. R. Soc. B. 273, (2006). Troscianko, J., Lown, A. E., Hughes, A. E. & Sevens, M. Defeaing crypsis: deecion and learning of camouflage sraegies. PLoS ONE. 8, e73733 (2013). Webser, R. J., Hassall, C., Herdman, C. M. & Sherra, T. N. Disrupive camouflage impairs objec recogniion. Biol. Le. 9, (2013). Duare, R. C., Flores, A. A. V. & Sevens, M. Camouflage hrough colour change: mechanisms, adapive value, and ecological significance. Phil. Trans. R. Soc. B. 372, (2017) Nachman, M. W., Hoeksra, H. E. & D'Agosino, S. L. The geneic basis of adapive melanism in pocke mice. PNAS. 100, (2003). Rosenblum, E. B. Convergen evoluion and divergen selecion: Lizards a he Whie Sands ecoone. Am. Na. 167, 1-15 (2006). Wallace, A. R. Mimicry and oher proecive resemblances among animals. Wesminser Rev. 1 July, 1-43 (1867). Merilaia, S., Lyyinen, A. & Mappes, J. Selecion for crypic coloraion in a visually heerogeneous habia. Proc. R. Soc. B 268, (2001). Kelewell, H. B. D. Recogniion of appropriae backgrounds by he pale and black phases of Lepidopera. Naure. 175, (1955). Endler, J. A. Progressive background maching in mohs, and a quaniaive measure of crypsis. Biol. J. Linn. Soc. 22, (1984). Kelewell, H. B. D. & Conn, D. L. T. Furher background-choice experimens on crypic Lepidopera. J. Zool. 181, (1977). Sargen, T. D. Background selecions of geomerid and nocuid mohs. Science 154, (1966). Lovell, P. G., Ruxon, G. D., Langridge, K. V. & Spencer, K. A. Individual quail selec egglaying subsrae providing opimal camouflage for heir egg phenoype. Curr. Biol. 23, (2013). Marshall, K. L. A., Philpo, K. E. & Sevens, M. Microhabia choice in island lizards enhances camouflage agains avian predaors. Sci. Rep. 6, (2016). Marshall, K. L. A. & Sevens, M. Wall lizards display conspicuous signals o conspecifics and reduce deecion by avian predaors. Behav. Ecol. 25, (2014). Duare, R. C., Sevens, M. & Flores, A. A. V. Shape, colour plasiciy, and habia use indicae morph-specific camouflage sraegies in a marine shrimp. BMC Evol. Biol. 16, 218 (2016). Gilby, B. L., Mari, R. A., Bell, E. G., Crawford, E. W., Jun, D., Lederer, B. I., Tibbes, I. R. & Burfeind, D. D. Colour change in a filefish (Monacanhus chinensis) faced wih he challenge of changing backgrounds. Environ. Biol. Fish. 98, (2015). Sargen, T. D. Behavioural adapaions of crypic mohs III: Resing aiues of wo barklike species, Melanolophia canadaria and Caocala ulronia. Anim. Behav. 17, (1969). Kang, C. K., Moon, J. Y., Lee, S. I. & Jablonski, P. G. Camouflage hrough an acive choice of a resing spo and body orienaion in mohs. J. Evol. Biol. 25, (2012). 13

14 Kang, C. K., Sevens, M., Moon, J. Y., Lee, S. I. & Jablonski, P. G. Camouflage hrough behavior in mohs: he role of background maching and disrupive coloraion. Behav. Ecol. 26, (2015). Kang, C. K., Moon, J. Y., Lee, S. I. & Jablonski, P. G. Mohs on ree runks seek ou more crypic posiions when heir curren crypiciy is low. Anim. Behav. 86, (2013). Barbosa, A., Allen, J. J., Mähger, L. M. & Hanlon, R. T.. Culefish use visual cues o deermine arm posures for camouflage. Proc. R. Soc. B. 279, (2012). Troscianko, J., Wilson-Aggarwal, J., Sevens, M. & Spoiswoode, C. N. Camouflage predics survival in ground-nesing birds. Sci. Rep. 6, (2016). Wilson-Aggarwal, J., Troscianko, J., Sevens, M. & Spoiswoode, C. N. Escape disance in ground-nesing birds differs wih individual level of camouflage. Am. Na. 188, (2016). Sevens, M., Párraga, C. A., Cuhill, I. C., Parridge, J. C. & Troscianko, T. S. Using digial phoography o sudy animal coloraion. Biol. J. Linn. Soc. 90, (2007). Soddard, M. C. & Sevens, M. Paern mimicry of hos eggs by he common cuckoo, as seen hrough a bird's eye. Proc. R. Soc. B. 277, (2010). Vorobyev, M. & Osorio, D. Recepor noise as a deerminan of colour hresholds. Proc. R. Soc. B. 265, (1998). Nosil, P. & Crespi, B. J. Experimenal evidence ha predaion promoes divergence in adapive radiaion. PNAS. 103, (2006). Sevens, M., Lown, A. E. & Wood, L. E. Camouflage and individual variaion in shore crabs (Carcinus maenas) from differen habias. PLoS One. 9, e (2014). Dawkins, R. The Exended Phenoype. Oxford: Oxford Universiy Press (1989). Troscianko, J., Wilson-Aggarwal, J., Spoiswoode, C. N. & Sevens, M. Nes covering in plovers: How modifying he visual environmen influences egg camouflage. Ecol. Evol. 6, (2016). Gosler, A. G., Barne, P. R. & Reynolds, S. J. Inheriance and variaion in eggshell paerning in he grea i Parus major. Proc. R. Soc. B. 267, (2000). Rohsein, S. I. Mechanisms of avian egg-recogniion: do birds know heir own eggs. Anim. Behav. 23, (1975). Douce, S. M., Mennill, D. J. & Hill, G. E. The evoluion of signal design in manakin plumage ornamens. Am. Na. 169, S62-S80 (2007). Marchei, K. Dark habias and brigh birds illusrae he role of he environmen in species divergence. Naure. 362, (1993). Sevens, M. & Ruxon, G. D. Linking he evoluion and form of warning coloraion in naure. Proc. R. Soc. B. 279, (2012). Troscianko, J. & Sevens, M. Image calibraion and analysis oolbox a free sofware suie for objecively measuring reflecance, colour and paern. Meh. Ecol. Evol. 6, (2015). Sevens, M., Lown, A. E. & Wood, L. E. Colour change and camouflage in juvenile shore crabs Carcinus maenas. Fron. Ecol. Evol. 2, 14 (2014). Sevens, M., Soddard, M. C. & Higham, J. P. Sudying primae color: owards visual sysem dependen mehods. In. J. Primaol. 30, (2009). Calderone, J. B. & Jacobs, G. H. Specral properies and reinal disribuion of ferre cones. Vis. Neurosci. 20, (2003). Govardovskii, V. I., Fyhrquis, N., Reuer, T., Kuzmin, D. G. & Donner, K. In search of he visual pigmen emplae. Vis. Neurosci. 17, (2000). Douglas, R. H. & Jeffery, G. The specral ransmission of ocular media suggess ulraviole sensiiviy is widespread among mammals. Proc. R. Soc. B. 281, (2014). Jacobs, G. H., Neiz, J., Crognale, M. A. & Brammer, G. L. Specral sensiiviy of verve monkeys (Cercopihecus aehiops sabaeus) and he issue of caarrhine richromacy. Am. J. Primaol. 23, (1991). 14

15 Sockman, A. & Sharpe, L. T. The specral sensiiviies of he middle-and long-wavelenghsensiive cones derived from measuremens in observers of known genoype. Vis. Res. 40, (2000). Ödeen, A. & Håsad, O. & Alsröm, P. Evoluion of ulraviole vision in he larges avian radiaion-he passerines. BMC Evol. Biol. 11, 313 (2011). Har, N. S. Vision in he peafowl (Aves: Pavo crisaus). J. Exp. Biol. 205, (2002). Lovell, P. G., Tolhurs, D. J., Párraga, C. A., Baddeley, R., Leonards, U., Troscianko, J. & Troscianko, T. Sabiliy of he color-opponen signals under changes of illuminan in naural scenes. J. Op. Soc. Am. 22, (2005). Arnold, S. E., Faruq, S., Savolainen, V., McOwan, P. W. & Chika, L. FReD: he floral reflecance daabase a web poral for analyses of flower colour. PLoS One. 5, e14287 (2010). Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image o ImageJ: 25 years of image analysis. Na. Meh. 9, (2012). R Core Team: R: A Language and Environmen for Saisical Compuing (Vienna, Ausria: R Foundaion for Saisical Compuing) (2016). Troscianko, J. A simple ool for calculaing egg shape, volume and surface area from digial images. Ibis 156, (2014). Troscianko, J., Skelhorn, J. & Sevens, M. Quanifying camouflage: how o predic deecabiliy from appearance. BMC Evol. Biol. 17, 7 (2017). Osorio, D. & Vorobyev, M. Phoorecepor specral sensiiviies in erresrial animals: adapaions for luminance and colour vision. Proc. R. Soc. B. 272, (2005). Chiao, C-C., Chubb, C., Buresch, K. C., Siemann, L. & Hanlon, R. T. The scaling effecs of subsrae exure on camouflage paerning in culefish. Vis. Res. 49, (2009). Renoul, J. P., Kelber, A. & Schaefer, H. M. Colour spaces in ecology and evoluionary biology. Biol. Rev. (2015) DOI: /brv Baes, D., Maechler, M., Bolker, B. & Walker, S. lme4: Linear mixed-effecs models using Eigen and S4. R package version 11-7 (2014). Table 1: Summary of resuls comparing camouflage mach o he chosen versus non-chosen locaions for all hree merics. The code corresponds o when camouflage was a beer mach o he chosen background han oher locaions, - corresponds o a beer mach o he non-chosen background, and O corresponds o no significan difference. The majoriy of resuls show suppor for microhabia selecion, especially for paern. Where a model did no reain zone as a predicor, Chi-square model comparison resuls are shown. F-saisics are shown for conspecific comparisons as he variable has wo levels (same or differen nes). T-values for planned conrass are shown for fine scale and local scale resuls (Tukey Pos-hoc es). Group Plover/courser eggs Scale Conspecific Local scale (5 m) Camouflage Meric Paern F5,7854 O 4.77, Model comparison Chi-square p , 15 Luminance 4.90, Colour O Model comparison Chisquare p O Model comparison Chisquare p 0.795

16 Fine scale (< 5 cm) 5.73, Nighjar eggs Conspecific Local scale (5 m) Fine scale (< 5 cm) Nighjar adul Conspecific Local scale (5 m) Fine scale (< 5 cm) O -3.81, Model comparison Chisquare p O Model comparison Chisquare p F1, , (X for one species) F2, , p , O p , 3.14, p F1, , F1, , F1, , p , (O for one species) p< , O p , -1.72, 3.60, 7.49, 4.18, 1.00, Figure Legends Figure 1a: Examples showing camouflage of eggs and aduls of six of he species of nighjar and plover/courser included in he sudy (individuals/cluches are in he cenre of each image). This shows he variaion in camouflage mach among species, and some of he differen colours and markings found in he eggs and aduls used for concealmen. 1b: Images of one adul nighjar, one nighjar cluch, and one courser cluch o predaor vision. These images (see Mehods) correspond o a dichromaic mammal, which sees colours ha o humans are yellows and blues, a richromaic primae wih equivalen colour percepion o humans, and a erachromaic bird. The laer has colour vision involving four cone ypes, including ulraviole, and because here is no sandard way o illusrae he range of colours birds may see (convenional images being resriced o hree channels), we here presen separae richromaic images (based on he avain LW, MW, and SW cones), and greyscale UV (viole; VS) cone images (whereby brigher pixels correspond o greaer UV informaion). 16

17 Figure 2: Analysis of egg camouflage beween chosen nes sies and poenial oher sies. We compared he camouflage of eggs or adul nighjars a hree spaial scales. A he fines cm scale (or zone), we compared egg or plumage camouflage o he chosen ( immediae comparison wihin a few cm) background surrounding he eggs (400 pixels, or approximaely 4.4 cm radius) o a ( near ) background immediaely adjacen o his. Nex, we compared arge camouflage o he chosen nes background ( local, comprising he combinaion of he immediae and near areas) o poenial backgrounds in he same area 5 m away ( disan comparison wihin 5 m). Finally, we compared he camouflage of arges o he chosen nes background of each individual o he sies chosen by oher conspecifics. 17

18 Figure 3: Model esimaes of planned comparisons comparing he mach beween each subjec (egg or adul bird) and is chosen nes background versus is mach o backgrounds chosen by is conspecifics. Error bars show 95% confidence inervals. Esimaes above zero indicae ha he chosen background is a beer fi han non-chosen conspecific backgrounds, supporing he hypohesis ha females selec backgrounds ha mach heir own camouflage, raher han a speciesspecific habia preference. The y-axis (model esimaes) shows he model esimae muliplier for paern, luminance, or colour maching. For example, his shows ha on average a nighjar s chosen nes has a (logged) JND value 0.16 imes lower han ha of a conspecific s. 18

19 Figure 4: Model esimaes of planned comparisons in paern, luminance, and colour camouflage maching beween zones wih 95% confidence inervals. Esimaes greaer han zero show ha he zone nearer he cluch maches he eggs or adul beer han he zone furher away. Esimaes below zero show he opposie effec, maching he more disan zone beer han he nearer zone. Colour maching for plover and courser eggs did no differ beween zones, so is no included. 19