Unmanned Ae ial Sys ems in
Conse a ion Biology
Ph.D.
Ma ga i a Mule o Pázmány
2015
U ilización de sis emas aé eos no ipulados
en biología de la conse ación
Tesis doc o al
1
Unmanned Ae ial Sys ems in Conse a ion Biology
U ilización de sis emas aé eos no ipulados en biología de la conse ación
Ph.D. / Tesis doc o al
Ma ga i a Mule o Pázmány
Uni e si y o Se ille. 2015
2
3
Supe iso
D . Juan José Neg o Balmaseda
Depa men o E olu iona y Ecology
Doñana Biological S a ion (DBS-CSIC)
Tu o
D . Ca los An onio G anado Lo encio
Depa men o Vege al Biology and Ecology
Uni e si y o Se ille
4

5
Dedica ed o my amily
6
7
Con en s
• Abs ac
9
• In oduc ion
11
• Aims
20
• S uc u e
21
• Chap e 1: En i onmen al impac assessmen o in as uc u es. A
low cos way o assessing bi d isk haza ds in powe lines: Fixed-
wing small unmanned ai c a sys ems.
23
• Chap e 2: Managemen o endange ed species. Remo ely Pilo ed
Ai c a Sys ems as a Rhinoce os An i-Poaching Tool in A ica.
37
• Chap e 3: Conse a ion in a human domina ed landscape. The eye
in he sky: combined used o unmanned ae ial sys ems and GPS
da a-logge s o ecological esea ch and conse a ion o small bi ds.
61
• Chap e 4: Conse a ion in a p o ec ed a ea. Could Unmanned
Ae ial Sys ems eplace biologging as a ool o animal spa ial
dis ibu ion s udies?
79
• Gene al discussion
97
• Conclusions
102
• Li e a u e ci ed
103
• Acknowledgemen s
117
8
Abs ac
Unmanned Ae ial Sys ems (UAS) ha e been used o decades in he mili a y
ield, mainly o dange ous o edious missions whe e i is p e e able o send a ehicle
equipped wi h senso s aking da a au oma ically han o use con en ional manned
ai c a s.
In ecen yea s echnology has ad anced, he ma ke has g own exponen ially,
p ices ha e descended and he use o he sys ems is simple , which has led o he
inco po a ion o he UAS o he ci ilian wo ld. UAS ha e p o en use ul in ecology
ela ed asks, such as animals moni o ing and habi a s cha ac e iza ion and hei
po en ial o ecology has been poin ed ou , bu o da e, he e a e jus a ew s udies
add essing hei use in conse a ion biology.
This Ph.D. a emp s o ill he gap o knowledge in he use o UAS in
conse a ion biology. I desc ibes o he i s ime he use o hese sys ems in an
immedia ely applicable way in impac assessmen o in as uc u es o wildli e and
p o ec ion o endange ed species. Fu he mo e, i p esen s UAS as a ool o ob aining
high- esolu ion spa io empo al in o ma ion which helps o unde s and habi a use in
apidly changing human domina ed a eas and demons a es ha hese sys ems can
p o ide in o ma ion as alid as he ob ained by con en ional echniques on he spa ial
dis ibu ion o species in p o ec ed a eas.
The expe imen s pe o med in he ame o his Ph.D. indica e ha UAS can
p o ide use ul in o ma ion o conse a ion biology, such as high spa io- empo al
esolu ion ae ial images ob ained by emba ked came as ha allow o moni o he
en i onmen a he esea che ’s desi ed equency and e isi ing si es o pe o m
sys ema ic s udies.
The esul s also e ealed ha UAS use in conse a ion biology p esen s some
cons ain s, mainly ela ed wi h he scope o he missions, he limi ing cos s o he
sys ems, ope a ing cons ains associa ed o wea he condi ions, legal limi a ions and he
need o specialized pe sonnel o ope a ing he sys ems, as well as some limi a ions o
da a analysis ela ed wi h image analysis.
O e all, gi en he no el y o he subjec and he impo ance i is expec ed o
ha e in he nea u u e, I conside ha p o iding in o ma ion on he capabili ies and
limi a ions o hese sys ems, based on solid expe imen s in conse a ion biology, is no
15
• Comme cial ae ial su eillance.
• Media indus y: spo s, ilming companies.
• Oil gas and mine al exploi a ion and p oduc ion.
• Disas e elie and medical assis ance.
• A chaeology esea ch.
• Homeland secu i y: coas al pa ol, domes ic police missions, bo de
su eillance, public p o es s moni o ing, d ug plan a ions de ec ion.
• En i onmen al moni o ing: wildli e census, animal acking, and in asi e
plan s assessmen .
• Land managemen : o es i e damage assessmen , o es i e mapping,
o es i e communica ions, e a dan applica ion,.
• Ag icul u e: c ops p oduc i i y assessmen , c ops sp aying, ineya ds
moni o ing (Be ni e al. 2009).
UAS in eg a ion in en i onmen al esea ch
Gi en he ad ances in UAS echnology and i s g owing di usion in he media, i
is no su p ising ha scien is s s a ed o explo e he use o UAS o en i onmen al
moni o ing. The app ochemen be ween UAS and en i onmen al esea ch has aken
place along he las 15 yea s in wo con ex s di e en ia ed by p ojec budge and UAS
access scena ios, which led o he pa allel exis ence o wo lines o wo k a di e en
scales and hema ic.
1) La ge scale p ojec s: mainly conduc ed by NASA and NOAA (wi h some
excep ions o la ge esea ch g oups) using la ge and medium UAS wi h capaci y o
ca y payload o med by ad anced senso s, high ange (>25 km), and au onomy (>4
hou s). UAS including payload p ices a e gene ally o e 100.000 ! and equi e high
ope a ional cos s i.e.: Global Hawk, Man a, Scan Eagle, Al ai , Ae osonde, Ikhana,
SIERRA, R100 Ma ine and Ae ocam. Resea ch opics (conduc ed and planned) a e
mos ly ela ed wi h ea h science: clima e change, a mosphe ic esea ch (me eo ology
and chemis y), soil s udies, la ge scale i e s udies, ege a ion s uc u e, composi ion
and canopy chemis y, glacie and ice shee dynamics, su ace de o ma ion, imaging
spec oscopy, opog aphic mapping, g a i a ional accele a ion measu emen s, An a c ic
and A ic explo a ion su eys, magne ic ields measu emen s, i e discha ge, soil

16
mois u e and eeze, land all and physical oceanog aphy (Williamson 2011; NOAA
2014).
2) Local scale p ojec s: conduc ed by uni e si y o esea ch cen e s’ depa men s
using small UAS equipped wi h basic payloads (gene ally RGB s ill pho o o ideo
came as and he mal enso s and less equen ly: me eo ological senso s, b oadband o
na owband py anome e - ype adiome ic senso s, o ligh weigh minia u ized
hype spec al adiome e s) (Ande son and Gas on 2013), sho ope a ional anges
(<25km) and au onomy (<4 hou s). UAS including payload p ices a e gene ally below
100.000 ! and ope a ions a e mos ly pe o med by he membe s o he esea ch g oup.
Small UAS a e equen ly ully o pa ially sel -made o acqui ed in he ama eu
ma ke , al hough he e a e also some p o essional/ comme cial sys ems, i.e.: UAS
de eloped by Conse a ion D ones, DIY, Ga ewing x100, Ques UAV, SUMO,
MMAV, BAT-3, CSIRO, D oidwo x mul icop e , I is+, Ae o, D agan lye X4, No a2,
Fulma , C yowing, Mik okop e and MLB Foldba (mo e pla o ms wi h ac ual and
possible uses in en i onmen al esea ch a e e iewed in Ande son & Gas on, 2013).
Resea ch opics a e o en a u he ex ension o he subjec he g oup was al eady
s udying by o he means. The main wo ks ha ha e been pe o med, o a e cu en ly
being conduc ed in his ield (excluding he ones p esen ed in his Ph.D.) a e:
1) Wildli e su eys: hese s udies a e mainly ocused on he e alua ion o he
sys ems o he di e en species de ec ion and easibili y o a mo e gene alized use.
Bi ds: wa e bi ds su eys in Flo ida (Jones 2003; F ede ick e al. 2009; Wa s e
al. 2010), black headed gulls in No h Eas o Spain (Sa dà-Palome a e al. 2012), geese
in Canada (Chabo and Bi d 2012), sandhill c anes in Colo ado (Fa ell 2013), s elle 's
sea eagle nes s in Russia (Po apo and U ekhina I.G., McG ady M.J. 2013), gull
colonies in Ge many (G enzdö e 2013) and osp eys in Mon ana (A e e 2014).
Te es ial mammals: oe dee in Ge many (Is ael 2011), hinoce os o an i-
poaching in Sou h A ica (Dewa 2013; WcUAVC 2013) and in Zimbabwe (Oli a es-
Mendez and Bissyand 2013), elephan s in Democ a ic Republic o Congo (Linchan e
al. 2013), Mozambique (Mande 2013), Bu kina Faso: (Ve meulen e al. 2013), Kenya
(Schi man 2014) (and wi h hinoce os and o angu ans) in Indonesia (Geme e al.
2014).
Ma ine mammals and ish: mana ees (and alliga o s) in Flo ida (Jones 2003;
Jones e al. 2006), ma ine mammals in Washing on (Koski e al. 2009), humpback
17
whales and dugongs in Wes e n Aus alia (Pype 2008; Hodgson e al. 2013) and
Chinook salmon in O egon, Washing on and U ah (Ae yonLabs 2014).
2) Habi a cha ac e iza ion: mainly ocused on ege a ion and landscape
cha ac e iza ion, al hough some ha e a mo e ecological o ien a ion, including animal
species habi a selec ion s udies.
Ecological es o a ion in La in Ame ica, (Wa s e al. 2008), Medi e anean
ipa ian o es in Sou he n F ance (Dun o d e al. 2009), coas al esea ch (Pe ei a e al.
2009), angeland moni o ing in New Mexico (Lalibe e e al. 2011), biodi e si y in
o es s in Ge many (Ge zin e al. 2012), opical o es in Indonesia (Koh and Wich
2012), habi a mapping using Te es ial LiDAR and UAV image y in Russia (Olsoy e
al. 2013), we lands in Canada (Chabo and Bi d 2013; Chabo e al. 2014), i e
mapping (loca ion no speci ied) (Room and Ahmad 2014), communi y-based o es
moni o ing in Malaysia, Nepal and Indonesia (Paneque-Gál ez e al. 2014), ma shlands
moni o ing (ABC 2014) and en i onmen al a iables ela ed o epidemiology in
Malaysia and Philippines (Fo nace e al. 2014).
3) Me hodological s udies: ocused on ad ances in echniques o da a
p ocessing o he design o sys ems speci ically o en i onmen al pu poses.
De elopmen o a UAV o wildli e su eillance (Lee 2004), algo i hm o
au oma ic bi d de ec ion (Abd-el ahman e al. 2005), geo- e e encing echniques
(Wilkinson 2007), es ima ing dis ibu ion o hidden objec s (Ma in e al. 2012),
es ima ing he su ace a ea o sampling s ips (Lisein e al. 2013), bio-logged wildli e
acking (So iano e al. 2009; Kö ne e al. 2010; Leona do e al. 2013), emo e wa e
sampling (Schwa zbach e al. 2014).
Finally, he e a e wo in e es ing a icles ha e iew he use, capabili ies and
limi a ions o UAS in Remo e Sensing (Wa s e al. 2012) and spa ial ecology
(Ande son and Gas on 2013).
Why UAS in conse a ion biology?
Conse a ion biology is a mission-o ien ed science ha ocuses on how o
p o ec and es o e biodi e si y, dealing wi h issues whe e quick ac ion is c i ical and i s
goal is o p o ide p inciples and ools o p ese ing biological di e si y. To e ec i ely
in o m policy and managemen au ho i ies, conse a ion esea ch mus add ess he mos
p essing p oblems and he mos h ea ened sys ems and o ganisms (Soulé 2007; SCB
2014).
18
Despi e he explosion o p ojec s wi h UAS o en i onmen al applica ions in
he ecen yea s, he e is s ill a lack o s udies ha i he philosophy o conse a ion
biology, i.e. p o iding solu ions ha a e immedia ely applicable o sol e conc e e and
u gen en i onmen al p oblems. To add ess hese p oblems, he Doñana Biological
S a ion-CSIC in collabo a ion wi h se e al ins i u ions, among which he School o
Enginee ing o he Uni e si y o Se ille is no ewo hy, has pa icipa ed since 2005 in
h ee mul idisciplina y p ojec s ocusing on he de elopmen o sys ems and echniques
o he applica ion o UAS o conse a ion biology, and in which his Ph.D. is amed.
SADCON (Andalusian Go e nmen , P ojec o Excellence, 2005 / TEP-375).
Dis ibu ed au onomous sys ems o en i onmen al conse a ion.
AEROMAB (Andalusian Go e nmen , P ojec o Excellence, 2007, P07-RNM-
03246). Ae ospace echnologies o biodi e si y conse a ion.
PLANET (7 h F amewo k P og am, coope a ion FP7-257649) Pla o m o he
deploymen and ope a ion o he e ogeneous ne wo ked coope a ing objec s.
19
Aims
The o e all objec i e o his Ph.D. is o e alua e he use o UAS in conse a ion
biology, iden i ying hei capaci ies and limi a ions. The speci ic aims a e:
! How can UAS con ibu e o en i onmen al impac assessmen o
in as uc u es?
• How can UAS con ibu e o managemen o endange ed species?
• Conse a ion in a human domina ed landscape: Can UAS cons i u e a use ul
ool o ob aining high- esolu ion spa io empo al in o ma ion on animals habi a use?
• Conse a ion in a p o ec ed a ea: A e UAS capable o p o iding in o ma ion as
alid as he ob ained by con en ional echniques on he spa ial dis ibu ion o species in
p o ec ed a eas?
20
S uc u e
In addi ion o he gene al in oduc ion, his Ph.D. consis s o ou chap e s ha
explo e he speci ic objec i es s a ed abo e examining ou ep esen a i e use cases. To
accomplish he objec i es, we (a mul idisciplina y eam om Doñana Biological
S a ion-CSIC) conduc ed se e al ield campaigns using small UAS along he las i e
yea s. The i s h ee chap e s co espond o published pape s and he las one o a
submi ed manusc ip .
CHAPTER 1: En i onmen al impac assessmen o in as uc u es
We desc ibe he use o low cos small unmanned ai c a sys ems (sUAS)
equipped wi h on-boa d came as o powe line su eillance o subsequen ly assess hei
impac on bi ds in a low cos way. We cha ac e ized ou powe lines, geo- e e enced
e e y pylon in selec ed po ions, and assessed hei haza d o bi ds. We compa e he
e ec i eness o wo a ian s o he sUAS me hod o da a acquisi ion and wo me hods
o plane con ol.
CHAPTER 2: Managemen o endange ed species
We desc ibe he use o a small low cos RPAS equipped wi h h ee di e en
ypes o came as o es hei abili y o suppo hinoce os an i-poaching asks in he
KwaZulu-Na al p o ince o Sou h A ica. We pe o med se e al ligh s in o de o es
he echnical capabili ies o he sys em o de ec hinoce os, o e eal simula ed
poache s and o do ence su eillance. We e alua ed he in luence o ligh al i ude,
ime and habi a ype in he e ec i eness o he sys em. Conside ing he mos common
modus ope andi o poache s, we alsoanalyzed he aspec s ha a ec emo ely pilo ed
ai c a ’s in eg a ion in an i-poaching ope a ions.
CHAPTER 3: Conse a ion in a human domina ed landscape
We desc ibe he combined use o GPS da a logge s and en i onmen al
in o ma ion eco ded by UASs o s udy habi a selec ion o a small bi d species, he
lesse kes el Falco naumanni, li ing in a highly dynamic landscape. A e downloading
he spa io- empo al in o ma ion om he kes els, we p og ammed he UASs o ly and
documen wi h pic u es he pa hs o hose same bi ds sho ly a e hei ligh , ex ac ing
en i onmen al in o ma ion a quasi- eal ime ha we used o s udy he a ailabili y o
di e en habi a ypes along he bi d ligh pa h.
CHAPTER 4: Conse a ion in a p o ec ed a ea

21
We assess he use ulness o UAS o e ligh s o: i) ge da a o model he
dis ibu ion o ee- anging ca le by compa ing wi h esul s ob ained om GPS-GSM
colla ed ca le, and ii) p edic species densi ies by compa ing wi h ac ual densi y in
Doñana Biological Rese e (Sou h o Spain).
22
“Nobody ealizes ha he e a e some people who spend excessi e
ene gy jus o look no mal”. Albe Camus
Chap e 1
En i onmen al impac assessmen o in as uc u es
[This Chap e is published as: Mule o-Pázmány, M., Neg o, J. J., & Fe e , M. (2014).
A low cos way o assessing bi d isk haza ds in powe lines: Fixed-wing small
unmanned ai c a sys ems. Jou nal o Unmanned Vehicle Sys ems, 2(1), 5–15.
doi:0.1139/ju s-2013-0012]
ABSTRACT
Acciden s on powe lines a e one o he mos impo an causes o man-induced
mo ali y o ap o s and soa ing bi ds. The ac o s ha condi ion he haza d ha e been
ex ensi ely s udied, and cu en ly he e a e a a ie y o echnical solu ions a ailable o
mi iga e he isk. Mos o he esou ces in conse a ion p ojec s o educe a ian
mo ali y now a e in es ed in ieldwo k o moni o he lines, which di e s he esou ces
a ailable o ins all ac ual co ec i e measu es o mi iga e bi d haza d. Li le p og ess has
been achie ed in he me hodology o cha ac e ize line isk, which is an expensi e,
edious and ime-consuming ask. In his wo k we desc ibe he use o low cos small
Unmanned Ai c a Sys ems (sUAS) equipped wi h onboa d came as o powe line
su eillance. As a case s udy, we cha ac e ized ou powe lines, geo- e e enced e e y
pylon in selec ed po ions and assessed hei haza d o bi ds. We compa e he
e ec i eness o wo a ian s o he sUAS me hod o da a acquisi ion and wo ways o
plane con ol. This wo k p o ides e idence o he use ulness o sUAS as a as ,
inexpensi e and p ac ical ool in conse a ion biology, adding o hei al eady known
applica ions in wildli e moni o ing he en i onmen al impac assessmen o
in as uc u es.
INTRODUCTION
Bi d mo ali y on powe lines is an impo an conse a ion issue ecognized
decades ago (Olendo , R.R., Mille , A.D. and Lehman 1981; C i elli, A.J., Je en up,
23
H., Mi che 1988; Fe e e al. 1991). Rap o and la ge bi d species a e especially p one
o elec ocu ion, mos ly on dis ibu ion lines (Neg o, J. J., Fe e 1995), and collision
wi h cables is mo e equen on ansmission lines, a ec ing g ega ious species o bi ds
ha ly a imes wi h educed isibili y (Neg o 1987; Fe e , M., Neg o 1992; Fe e ,
M., Janss 1999).
The dis ibu ion o bi d acciden s on powe lines has a signi ican endency o
accumula e on ce ain pylons o spans (cable leng h be ween wo pylons) (Fe e , M,
Hi aldo 1991; CLAVE S.L. 1992). Thus, e ec i ely co ec ing a small ac ion o all
pylons and/o spans o a gi en line i is possible o educe o al mo ali y d as ically
(Fe e , M, Hi aldo 1991; López-López e al. 2011).
Bi d nes ing on pylons is ano he si ua ion ha may inc ease elec ocu ion isk
and also p oduces damage o he in as uc u es; bo h esul in economic losses and
educe se ice quali y o u ili y companies (Eléc ica 2005; Fe e 2012).
Cu en ly, he bulk o he e o in e ms o ime and cos s o mi iga e he bi d
haza d o powe lines is in es ed in he ieldwo k o he cha ac e iza ion phase o he
s udy. Line moni o ing is no mally done by ca o on oo (Ka asnik e al. 2008),
iden i ying pylon design, eco ding pylon loca ion wi h a GPS, iden i ying bi d
mo ali ies and su eying habi a ypes, all ac o s ha would con ibu e o he
assignmen o isk alues (Fe e , M, Hi aldo 1991).
The e a e o he possibili ies o powe line s udy, such as using con en ional
ai c a wi h au oma ic ideo su eillance sys ems (Whi wo h e al. 2001; Ma and Chen
2004), sa elli e images, o a y-wing unmanned ai c a sys ems (UAS) (Campoy, P.,
Ga cía, P.J., Ba ien os, A., Ce o, J.D., Agui e, I., Roa, A., Ga cía, R., Muñoz 2001;
Peungsungwal, S., Pungsi i, B., Chammong hai, K., Okuda 2001; Ma and Chen 2004;
Jones e al. 2005; Ka asnik e al. 2008; Li and Ruan 2010) and mo e sophis ica ed
solu ions, including climbing- lying obo s (Ka a"nik e al. 2008), bu hey a e oo
expensi e o be applied ou inely in conse a ion biology s udies o hey ha e no been
implemen ed ealis ically in he ield ye .
Fixed-wing small Unmanned Ai c a Sys ems (sUAS) a e unde going
ema kable de elopmen , which has led o a dec ease in p ices and a g ea e a ie y o
equipmen a ailable. Thei use has inc eased conside ably o di e en pu poses in
mili a y and ci il applica ions. SUAS ha e been ecen ly inco po a ed in wildli e
conse a ion, mainly ocusing on ae ial wildli e su eys and habi a s udies (Jones e al.
24
2006; Pe ei a e al. 2009; Wa s e al. 2010; Chabo and Bi d 2012; Rod íguez e al.
2012a; Sa dà-Palome a e al. 2012; Ge zin e al. 2012).
He e we desc ibe he use o ixed-wing sUAS echnology as a ool o
cha ac e ize powe lines o subsequen ly assess hei impac on bi ds in a low cos way.
We also compa e he use ulness o wo di e en ypes o came as o iden i y and geo-
e e ence powe pylons and es as well wo al e na i e a ian s o plane con ol.
MATERIAL AND METHODS
S udy a ea
Fieldwo k was conduc ed in wo loca ions in sou hwes e n Spain: an ag icul u al
a ea in Dos He manas, Se ille (5º56´16.1816´´W, 37º15´22.462´´N) and a p ese ed
a ea wi hin Doñana Na ional Pa k, Huel a (6º31´58.8522´´W, 37º6´53.2887´´N).
Su eys ook place in Ma ch, Ap il and Decembe 2012.
SUAS echnical speci ica ions
We used he adio con olled Easy ly S -330 (S -models, China) p opelled by a
b ushless elec ical mo o . Wingspan is 1.960 m and i has a Maximum Take-O
Weigh (MTOW) o 2 kg wi h a 250 g payload (Figu e 1). I s maximum ange is 10 km,
endu ance 50 minu es and i can ake o and land manually in small pa ches o la and
open e ain.
Ope a ions can be ca ied ou in wo di e en ways and i is possible o swi ch
om one o he o he du ing he ligh .
Au oma ic mode: he plane is con olled and guided by he au opilo sys em. No
in e en ion om he pilo is equi ed du ing he ligh (only aking-o and landing a e
pe o med manually). The au opilo p o ides ligh s abiliza ion and he capabili y o
p og am waypoin s, and i he con ol signal is los , he au opilo ac i a es he “ e u n
home” mode.
Fi s Pe son View sys em (FPV): he pilo con ols he plane in eal ime using
i ual eali y glasses and sees eleme y da a supe imposed on he ideo. The FPV
sys em includes a long- ange adio con ol ecei e .
In bo h con ol modes, On-Sc een Display (OSD) unc ion p o ides eal ime
ligh in o ma ion (cou se, al i ude, speed, waypoin s and a i icial ho izon)
supe imposed on he ideo signal om a came a loca ed on he plane´s nose, which can
31
Aus alia, decla es o spend $80 million a yea on inspec ion (Li and Ruan 2010). In he
Andalusia egion (Spain), app oxima ely 20% o he o al budge spen in e o i ing
dange ous dis ibu ion powe poles o p o ec he endange ed Spanish impe ial eagle
(Aquila adalbe i) was he cos o iden i ica ion o powe pole design, which was
a ound 500,000 ! (López-López e al. 2011). I is impo an o poin ou ha his kind o
su eillance o he poles i is necessa y no only du ing pole cha ac e iza ion p io o
selec which ones mus be modi ied, bu also a pe iodic su ey o he an i-elec ocu ion
de ices is needed. Limi ed li e span o insula ion p o ec i e de ices equi es pe iodic
inspec ions o assu e e ec i e p o ec ion. Simila ly, la ge bi d nes s on powe poles
equi e pe iodic su eys in o de o p e en ou ages. Consequen ly, educ ion in he
o al cos and ime using sUAS would be g ea e .
As e e ence, o he sUAS inspec ion o he 12 km o lines su eyed o his
s udy, 4 ligh s we e needed. On each one o hem, he wo ope a o s in es ed a o al o
2 hou s o he sUAS p epa a ion, ligh and da a p ocessing.
Ou s udy is he i s one demons a ing ha low cos ixed-wing sUAS a e a
use ul ool o powe lines moni o ing and o e ad an ages in cos and ime in es men
e sus o he me hods. Ou sys em, alued a 7,800 !, has been able o geo- e e ence and
cha ac e ize powe lines p o iding he in o ma ion needed o assess bi d elec ocu ion
and collision haza d. Thus, hei use can help o minimize he esou ces in es ed in he
ieldwo k phase o he wo k, o alloca e mos o he unds in o ac ual co ec i e
measu es.
ACKNOWLEDGEMENTS
We hank Es eban Gue e o and Miguel Ángel Aguila , he echnician and pilo ,
espec i ely, esponsible o he UAS. This s udy was conduc ed wi hin he P ojec s:
Ae omab, (Andalusia Go e nmen , P ojec o Excellence, 2007, P07-RNM-03246) and
Plane (Eu opean Commission 7 h FP G an Ag eemen 257649). We also hank
Manuela de Lucas, Ma cello D’Amico and Manuela González o hei e iews, and o
p o iding aluable commen s on his manusc ip .

32
FIGURES
Fig. 1. Desc ip ion o ou small unmanned ae ial sys em: (a) ae ial pla o m, (b)
an ennas, (c) g ound con ol s a ion, (d) wing-moun ed o wa d-poin ing came a, and
(e) wing-moun ed nadi -poin ing came a.
33
Fig. 2. Su eyed powe lines wi h pylons geo- e e enced by h ee di e en me hods: (a)
Dos He manas a ea, (b) Doñana a ea. Ci cle, om GPS a he base o he pylon; squa e,
om sUAS using nadi -poin ing came a; iangle, om sUAS using o wa d-poin ing
came a.
34
Fig. 3. Example o pylon designs eco ded om he UAS (pylon designs classi ied
ollowing Cla e 1992).
35
Fig. 4. Geo- e e encing p ecision o he wo ypes o came as: (a) nadi -poin ing; and
(b) on -poin ing.
Fig. 5. Whi e s o k nes s on he pylons: (a) and (b) eco ded by o wa d-poin ing
came a; (c) and (d) eco ded by nadi -poin ing came a.
36
“A ica, amongs he con inen s, will each i o you ha God and
he De il a e one”. Ka en Blixen
Chap e 2
Managemen o endange ed species
[This Chap e is published as: Mule o-Pázmány, M, S olpe R, an Essen LD, Neg o JJ,
Sassen T. 2014. Remo ely Pilo ed Ai c a Sys ems as a Rhinoce os An i-Poaching Tool
in A ica. PloS ONE 9(1): e83873]
Abs ac : O e he las yea s he e has been a massi e inc ease in hinoce os
poaching inciden s, wi h mo e han wo indi iduals killed pe day in Sou h A ica in he
i s mon hs o 2013. Immedia e ac ions a e needed o p ese e cu en popula ions and
he agen s in ol ed in hei p o ec ion a e demanding new echnologies o inc ease hei
e iciency in he ield. We assessed he use o emo ely pilo ed ai c a sys ems (RPAS)
o moni o o poaching ac i i ies. We pe o med 20 ligh s wi h 3 ypes o came as:
isual pho o, HD ideo and he mal ideo, o es he abili y o he sys ems o de ec (a)
hinoce os, (b) people ac ing as poache s and (c) o do ence su eillance. The s udy
a ea consis ed o se e al la ge game a ms in KwaZulu-Na al p o ince, Sou h A ica.
The a ge s we e be e de ec ed a he lowes al i udes, bu o ope a e he plane sa ely
and in a disc ee way, al i udes be ween 100 and 180 m we e he mos con enien . Open
a eas acili a ed a ge de ec ion, while o es habi a s complica ed i . De ec abili y using
isual came as was highe a mo ning and midday, bu he he mal came a p o ided he
bes images in he mo ning and a nigh . Conside ing no only he echnical capabili ies
o he sys ems bu also he poache s´ modus ope andi and he cu en con ol me hods,
we p opose RPAS usage as a ool o su eillance o sensi i e a eas, o suppo ing ield
an i-poaching ope a ions, as a de e en ool o poache s and as a complemen a y
me hod o hinoce os ecology esea ch. He e, we demons a e ha low cos RPAS can
be use ul o hinoce os s akeholde s o ield con ol p ocedu es. The e a e, howe e ,
impo an p ac ical limi a ions ha should be conside ed o hei success ul and
ealis ic in eg a ion in he an i-poaching ba le.
INTRODUCTION

37
The wo species o A ican hinoce os, he black hinoce os (Dice os bico nis)
and he whi e hinoce os (Ce a o he ium simum) we e d i en o nea ex inc ion in he
1990’s (Emslie and B ooks 1999). Numbe s o bo h species a e aising in A ica since
2007 (Knigh 2011), bu om 2010 he con inued escala ion in popula ion g ow h has
slowed down (Emslie e al. 2013), and he wo species a e s ill ulne able, wi h whi e
hinoce os classi ied as Nea Th ea ened and black hinoce os lis ed as C i ically
Endange ed acco ding o IUCN c i e ia (Emslie 2012).
Sou h A ica holds mo e hinoce os han any o he coun y in he wo ld, wi h
83% o A ica’s indi iduals, and also expe iences he highes absolu e le els o
poaching, which is he main h ea o hei conse a ion (Emslie e al. 2013). O e he
las yea s, and despi e he an i-poaching e o s, he e has been a massi e inc ease in he
numbe o hinoce os poaching inciden s. In 2010 he e was an a e age o 0.9
hinoce os killed pe day; in 2011 i inc eased o 1.2; his numbe escala ed o 1.8 in
2012, ( esul ing in 668 dea hs along he yea ) and i has eached a s agge ing his o ical
eco d o 2.2 pe day in he wo i s mon hs o 2013 (up o Feb ua y 20 h) (Emslie e al.
2013).
The hinoce os poaching is a complex p oblem wi h mul iple causes and
po en ial solu ions (Eus ace 2012). Thei ho n is conside ed o be a adi ional medicine
o a a ie y o ailmen s in Asia (Le e 2004), wi h he highes demands om China,
Hong Kong, Sou h Ko ea and Sou heas Asian coun ies, and i is used o ce emonial
pu poses in Yemen (Loon and Polakow 1997; Milledge 2007). Due o he high demand
and he illegal na u e o he ade, he p ices e ched by he ho n in he black ma ke a e
high. This cons i u es a emp a ion o u al people wi h sca ce esou ces, as he ma ke
alue o one ho n-se may be equal o he sala y o se e al yea s o he poache
(Eus ace 2012).
The e a e a ious long and medium- e m s a egies in p og ess o educe he
illegal ade o hinoce os ho n, and hey emain in cons an discussion: ho n con ol,
legisla ion, coope a ion wi h he ho n pu chasing coun ies, en i onmen al educa ion
and u al de elopmen p ojec s in hinoce os a eas, mos o hem conduc ed by public
ins i u ions o NGOs (Milledge 2007; Knigh 2011). These gene al s a egies a e also
suppo ed by immedia e an i-poaching ac ions in he ield, di ec ed by he managemen
au ho i ies o he landowne s, and ca ied on by ei he pa k ange s o secu i y
companies.
38
In Sou h A ica, a ound a qua e o he o al popula ion o hinoce os li e on
p i a e land (Knigh 2011). The owne s o hese ese es and game a ms a e
inc easingly hi ing specialized companies ha ocus on he p o ec ion o wildli e and
he app ehension o poache s. The se ice o p o ec ing aluable wildli e has led o an
eme gence o his ype o business in ecen yea s. They employ echniques based on
ope a ional me hods o he police and a med o ces. The basis o his s a egy is o
deploy g ound based pa ol uni s ha spend mul iple days acking animals and
poache s, and moni o ing he ence lines o b eaks. While he cos o employing hese
companies is high (a ound 10,800 ! pe yea o main ain 1 gua d pa olling 700-800 ha),
hey a e he mos popula al e na i e o educe he numbe o poaching inciden s in
p i a e land. Bo h p i a e companies and public agen s wo king in hinoce os an i-
poaching a e demanding new echnologies o inc ease hei e iciency o de ec and
in e cep poache s be o e a hinoce os is killed. The need o be mo e e ec i e in
add essing he poaching p oblem was exp essed by he IUCN/SSC A ican Rhinoce os
Specialis G oup (Knigh 2011).
Discussions wi h secu i y companies and conse a ion agencies ha e indica ed
ha ae ial moni o ing may be o assis ance in co e ing mo e g ound, and emo ely
pilo ed ai c a sys ems (RPAS he eina e ) ha e been sugges ed o do his wo k
(Eus ace 2012). Some secu i y i ms al eady pa ol he as a ms by lying wice a day
wi h a mic o ligh ai c a and di ec ing he “boo s on he g ound” o he whe eabou s o
he hinoce os.
Remo ely Pilo ed Ai c a Sys ems (RPAS), some imes also e e ed as
Unmanned Ae ial Vehicles (UAVs), Unmanned Ae ial Sys ems (UASs) o d ones ( he
ones o mili a y pu poses), a e ai c a s ( ixed o o a y wings) ha a e equipped wi h
came as and/o o he senso s and can be sen (using manual, semi-au oma ic o
au oma ic con ol) o a des ina ion o ga he in o ma ion. These ai c a s ac like an “eye
in he sky” (Rod íguez e al. 2012a) wi h he ope a o a he g ound con ol s a ion
ecei ing da a o sending o de s o he ae ial pla o m. RPAS ha e been used o
loca ing “enemies” in mili a y applica ions o he las 20 yea s (Zenko 2013), and mo e
ecen ly hey ha e s a ed o play a ole in many ci ilian asks, including wildli e
moni o ing (Jones e al. 2006; Wa s e al. 2010; Koh and Wich 2012; Rod íguez e al.
2012a; Sa dà-Palome a e al. 2012; Ve meulen e al. 2013).
In his pape , we desc ibe he use o a small low cos RPAS equipped wi h h ee
di e en ypes o came as o es hei abili y o suppo hinoce os an i-poaching asks
39
in coope a ion wi h a specialized secu i y company wo king in he KwaZulu-Na al
p o ince o Sou h A ica. We pe o med se e al ligh s in o de o es he echnical
capabili ies o he sys em o de ec hinoce os, o e eal simula ed poache s and o do
ence su eillance. We e alua ed he e ec i eness o he sys em a di e en al i udes
and imes o he day and nigh , and o e he wo main habi a ypes in he a ea: open
g assland and o es . Conside ing he mos common modus ope andi o poache s, we
analyzed he aspec s ha a ec emo ely pilo ed ai c a ’s in eg a ion in an i-poaching
ope a ions.
MATERIAL AND METHODS
E hics s a emen
A p esen , no egula ions a e in place o he use o RPAS in Sou h A ica.
D a egula ions pe aining o he use o UAVs ha e been published by he Sou h
A ican Ci il A ia ion Au ho i y (SACAA) bu hese ha e no been a i ied o da e. The
Rec ea ional A ia ion Au ho i y o Sou h A ica (RAASA) indica ed ha he ligh s
could be pe o med as long as hey we e conduc ed o e wildli e a eas wi h low
manned ai c a ac i i y and no close o egis e ed ac i e ai ields. The s udy he e o e
complies wi h he cu en Sou h A ican legisla ion in ol ing a ia ion sa e y. The
RPAS ope a o s had he equi ed in e na ional adio ope a o licenses o ope a e in he
equencies used o his wo k.
To ge an insigh in o he poaching p oblem, we me ou people in ol ed in
hinoce os p o ec ion a di e en le els. These in e iews did no con ain pe sonal o
e hically sensi i e in o ma ion, he e o e e hics app o al was deemed unnecessa y by
bo h he E hics Commi ee o Animal Wel a e o Doñana Biological S a ion (CEBA-
EBD) and he Animal E hics Commi ee (AEC - Facul y o Na u al and Ag icul u al
Sciences), a sub-commi ee o he Commi ee o Resea ch E hics and In eg i y o he
Uni e si y o P e o ia. All ou in e iewed people p o ided hei e bal in o med
consen o ake pa in he s udy once in o med abou he na u e and objec i es o he
in es iga ion. The pa icipan s ga e hei implied consen h ough coope a ion and i
was he e o e deemed unnecessa y o ob ain w i en consen . All aspec s o hese
pe sonal communica ions we e w i en down as pa o he da a collec ion p ocess o he
en i e p ojec . E hics commi ee app o al was deemed unnecessa y o app o e his
consen p ocedu e. We hank a m owne s and he secu i y company o p o iding
40
aluable in o ma ion used in his s udy, he lodging and he logis ics o he ield
campaign.
S udy a ea
The s udy a ea comp ised 13 a ms whose a eas anged be ween 1,500 and
25,000 ha, co e ing a o al o 100,000 ha loca ed in KwaZulu-Na al p o ince, Sou h
A ica. The habi a on he a ms is a combina ion o o es pa ches and g assland, and is
u ilized mainly o eco ou ism and hun ing. The hinoce os popula ion (bo h black and
whi e) in he a ea is app oxima ely 500 indi iduals. The ield campaign was pe o med
du ing Augus 2012.
Rhinoce os sa e y equi emen s de ini ion
To de ine poache s’ way o ope a ion and ac ual an i-poaching su eillance
me hods, we sepa a ely me ou people a he onse o he ieldwo k: he secu i y
company manage , he ange s’ coo dina o and wo ange s o he a ms o he s udy
a ea, all o hem esponsible o di e en aspec s o hinoce os sa e y.
Remo ely Pilo ed Ai c a Sys em desc ip ion
-Ai ame
The ixed-wing RPAS is a comme cially a ailable adio con ol plane ai ame
Easy Fly S -330 (S -models, China) modi ied by ou eam. I has a wingspan o 1,960
mm and a maximum ake-o weigh o 2 kg wi h a 350 g payload (Fig. 1). I has a
maximum ange o 10 km; an endu ance o 50 minu es and i is launched by hand and
landed manually in small pa ches o open e ain. I is p opelled by a b ushless elec ical
mo o using a li hium polyme ba e y.
The plane is capable o ope a e in h ee di e en modes, and i is possible o
swi ch om one o he nex du ing he ligh : au oma ic (using he abili ies o he
au opilo ), FPV (“ i s pe son iew mode”) and manually ( adio con ol con en ional
mode, also called “ hi d pe son mode”). I is equipped wi h an onboa d FPV ideo
came a, a GPS (10 Hz, Media ek, model FGPMMOPA6B), a da a-logge wi h a
ba ome ic al i ude senso Eagle ee GPS logge V.4 (Eagle ee sys ems, WA, USA)
and an au opilo (Ika us, Elec onica RC, Spain) which p o ides ligh s abiliza ion and
On Sc een Display (OSD). The OSD p o ides GPS in o ma ion abou he posi ion,
speed, heigh and cou se o he ai c a . The da a combined wi h he FPV ideo signal
om he came a a e sen o he g ound s a ion. Fo noc u nal ligh s we equipped he
47
poache s we e iden i ied in a wide ange o al i udes om 29 o 158 m in g assland and
o es habi a , al hough i was mo e di icul o dis inguish some indi iduals in he
o es , especially ce ain ange s in camou lage clo hing because hey o e ed less
con as wi h he su oundings. Fence su eillance esul s we e accep able a mo ning
and midday hou s, wi h he pic u es p esen ing enough quali y o zoom in and ind
people along i . A he lowes al i ude (40 m) i was also possible o de ec oo p in s in
he sand, bu he quali y was no su icien o check he condi ion o he ence wi es
along he en i e ence ou e. (Fig. 2)
The quali y o he images was bes a midday (80% o he pic u es had high
quali y in his ime pe iod) wi h e ical sunligh , and he esul s we e wo se when he
shades o he ees p oduced da k a eas, which happened in he mo ning (66% high
quali y) and in he e ening, when his e ec is accen ua ed because he ai is less clean
causing a blu y e ec (100% medium quali y pic u es).
Video da a
The HD ideo came a p o ided good esolu ion below 40 m AGL, bu due o he
wide angle o he lens ( o 127º), ligh s abo e 50 m al i ude AGL had no enough
quali y o iden i y people o o su ey he ences. These esul s led us o cancel he
planned ligh s o hinoce os de ec ion, as we conside ed he al i ude had o be so low
o iden i y objec s ha i could be dange ous o ope a ing he ai plane and migh also
dis u b he hinoce os. (Fig. 3 and Video S1 in supplemen a y ma e ial)
The he mal came a p o ided he ines images in he ea ly mo ning, when he
g ound was coldes and he e is mo e con as be ween i and any animal o pe son. We
con i med he p esence o a ge s a al i udes as high as 155 m, bu in gene al, i was
di icul o iden i y hem a he species le el, as hey appea in he ideo as di use
(al hough e y con as ing) whi e spo s. Only 5% o he images aken wi h his came a
p esen ed high quali y, 24% medium and 71% low. A he ea lies hou s o he nigh ,
he esul s ob ained did no allow us o con i m ha any o he spo s we de ec ed when
o e lying a hinoce os was ac ually a hinoce os, and low al i ude was needed o
iden i y he people using de ails such as body shapes. A e hou s o wo king wi h
he mal ideo and “ aining he eyes” we no iced a conside able imp o emen on
de ec ion and shapes iden i ica ion. Resolu ion o e ed by he he mal came a was
enough o ollow ence pos s and o de ec indi iduals, bu ence wi es we e no
dis inguishable a all. (Fig. 4 and Video 2 in supplemen a y ma e ial)

48
DISCUSSION
Rhinoce os poaching is a p essing issue ha needs immedia e solu ions in he
ield. Rhinoce os s akeholde s a e demanding new echnologies (Knigh 2011); social
media ha e al eady sugges ed he use o d ones (Wild 2013) and WWF announced in
2012 ha i sponso s an on-going emo ely ae ial su ey sys em and an i-poaching
p og am in coope a ion wi h Google o p o ec ige s, hinoce os and elephan s (WWF
2013). RPAS ha e al eady p o ed hei e icacy o mili a y and ci il applica ions in
gene al, and wildli e moni o ing in pa icula . Now he ques ion is how o in eg a e
RPAS in hinoce os an i-poaching asks. To answe his ques ion he e a e wo main
aspec s o conside : capabili ies ( echnical and p ac ical) and cu en limi a ions.
Technical conside a ions
The s ill pho o came a p o ided he bes esul s in e ms o image quali y (94%
o he pic u es aken by his came a allowed us o con i m he a ge s) and p ecision in
he loca ion. Tha is why his is he mos a ac i e and cu en ly he me hod o choice in
conse a ion biology s udies (Chabo and Bi d 2012; Ge zin e al. 2012). Howe e , i is
a ela i ely slow p ocedu e, as images mus be downloaded a e RPAS lands and hen
e iewed and pos p ocessed. E en i pic u es we e ansmi ed in eal ime o he
g ound s a ion (which is echnically possible) accele a ing he p ocess, i would s ill
ake ime o e iew hem. The e o e, he use o a s ill pho o came a would no be
sui able o suppo eal ime an i-poaching asks like poache s loca ion du ing a pu sui .
A posi i e aspec is ha s ill pho os would be he bes me hod o p o ide image p oo s
agains poache s because i o e s he bes esolu ion.
Video o e s eal ime da a, so i seems a be e op ion han s ill images o
poaching con ol. I is ecommended o use a ideo came a wi h a na owe iew ield
and zoom capabili ies o iden i y he a ge s a sa e al i udes (o e 100 m AGL) in eal
ime wi h enough magni ica ion. Al hough ideo o e s less p ecision on a ge loca ion,
acco ding o he in e iews wi h he people in ol ed in hinoce os sa e y, accu acy is
no so impo an o an i-poaching pu poses, o a leas i is less impo an han
immediacy.
As a as we know, his s udy o e s he i s noc u nal es s o wildli e
moni o ing using he mal came as onboa d a ixed-wing small RPAS, which is he only
op ion o RPAS nigh ime su eillance. The came a we used p o ided accep able
esul s when lying low, bu he quali y does no gua an ee o iden i y some a ge s and
49
i is possible o miss some, e en one as conspicuous as a hinoce os, when he mal
con as is low o lying a high al i udes. 29% o he he mal images allowed us o
con i m he a ge s, and he es p esen ed low quali y, p ecluding iden i ica ion. I is
impo an o conside ha he las a e s ill use ul, as in a eal an i-poaching si ua ions,
he dubious objec s could be u he inspec ed ei he o e lying lowe he RPAS o by
o he means (as g ound pa ols). Addi ionally, he quali y and esolu ion o he he mal
senso can be imp o ed and he e o e he de ec ion.
As expec ed, habi a ype had an in luence on a ge de ec ion, which is mo e
no iceable when using isual came as, ei he ideo o s ill pho o. Al hough hinoce os
a e la ge enough o be de ec ed om high al i udes wi h s ill pho o came as, people,
especially i wea ing camou lage clo hes o hidden unde a hick ee may no be
de ec able i lying a high al i udes.
Time o day had an in luence on a ge de ec ion. Ou esul s indica ed ha bes
ime o he use o isual came as was om ea ly mo ning o midday, and dec eased
along he e ening. The mal came a p o ided be e esul s when empe a u e con as is
highe (Is ael 2011), mainly a ea ly mo ning and nigh . The de ec abili y limi a ion
linked o he hou ly cycle, which is ela ed o ligh condi ions and ai -g ound he mal
con as , is impo an , as his means ha he use ulness o RPAS as moni o ing ools
does no emain cons an h oughou he day. This e ec would be accen ua ed when he
empe a u es a e highe and humidi y inc eases, as we would expec in he a ea whe e
we pe o med he es s du ing summe , o in places wi h high humidi y le els ( opical
o coas al a eas).
The e is a comp omise in deciding ligh al i ude o an i-poaching. Lowes
al i udes p o ide he bes esul s in e ms o image o ideo esolu ion, bu he su eyed
a ea is smalle . Flying low implies mo e isk o he plane in case o ailu e and easie
de ec ion o he plane om he g ound ( he e o e dis u bing he hinoce os o being
mo e easily de ec ed by poache s). Ou esul s sugges ha an al i ude ange be ween
100 and 180 m AGL is sui able o de ec ing hinoce os o people, and o do ence
su eillance wi h accep able quali y le els, i is a sa e al i ude o he plane and i is no
e y no iceable om he g ound.
P ac ical conside a ions
50
Conside ing poache s modus ope andi and cu en secu i y p ocedu es, he e a e
some limi a ions o he in eg a ion o RPAS in ou ine an i-poaching wo k in a ealis ic
and e icien manne .
Legal aspec s
Sou h A ica, as wi h many o he coun ies in he wo ld, does no ye ha e a
legal amewo k o ope a ing unmanned ae ial sys ems. The absence o egula ion o
lying beyond line o sigh cons ains he ange o wo k o he ai c a s, s ongly
limi ing he ac ual echnological capaci ies o he sys ems o jus sho ange ope a ions
o RPAS ope a ed by manual adio con ol (SAMAA 2001), as he ones we p esen ed in
his pape . Some au ho s al eady add essed his issue a guing ha ope a ions ha do no
pose a sa e y h ea o humans in he ai o on he g ound should be pe mi ed (Ingham
e al. 2006). They sugges ed Ligh UAVs o poaching si e su eillance and p oposed
ideas including UAV co ido s, a oiding inhabi ed a eas and equen ly used ai space,
all in o de o ly hese ai c a s sa ely. We suppo hese p oposals, as hinoce os
dis ibu ion coincides wi h e y low popula ed a eas whe e he isk o hi ing a pe son
o c ashing wi h ano he ai c a o in as uc u e is low, especially lying a al i udes
below 300 m AGL. The Sou h A ican Ci il A ia ion Au ho i y (SACAA) has
published d a UAS egula ions (SACAA 2008; Mamba 2009) ha include excep ional
pe mi s o public in e es uses o UASs (as an i-poaching could be classi ied).
Howe e , o da e he e has been no o icial no ice ha he SACAA has app o ed any
p o ocol o UASs ligh s.
Scale o wo k and ange
Scale o wo k is a limi ing ac o in using RPAS o an i-poaching asks. The
e i o ies hinoce os inhabi a e la ge and popula ion densi y is low (1 hinoce os/200
ha on a e age in ou s udy a ea). We demons a ed ha i is possible o ha e an “eye in
he sky”, bu his eye canno look e e ywhe e all he ime, so ha logis ics ha e o be
e alua ed. How many eyes a e necessa y and how o en do hey ha e o look? The
managemen and applica ion o a RPAS o mul iple RPAS is a key ques ion ha
hinoce os sa e y s akeholde s need o conside and de ine be o e planning RPAS use.
Small low cos RPAS ypically ly o 30-40 min and hei ange is limi ed up o
10-15 km. Roughly conside ing ha a RPAS lying a 150 m AGL could co e 711 ha,
o su ey he 100,000 ha o ou s udy a ea would ake a ound 140 hou s (5.8 days). And
ha excludes he ime o mo e he G ound Con ol S a ion om one poin o ano he ,
aking o and landing, changing and cha ging he ba e ies, da a p ocessing, and
51
assuming 24 hou s pe sonnel a ailabili y. Ob iously, ha ime would be educed i
ha ing mo e RPAS a ailable, bu ha would en ail highe associa ed cos s.
The e is a comp omise be ween he a ea o con ol and he equency o his
con ol. A easonable solu ion would be o ocus RPAS o moni o ing ho spo s: ei he
hinoce os p e e ed loca ions o mos sensi i e poaching a eas, which a e gene ally
known by secu i y companies o pa k ange s, o a eas whe e access by an i-poaching
pa ols and/o ehicles is complica ed by o he ac o s such as di icul e ains e c.
Wea he condi ions
Small RPAS a e sa e o ly up o 15-20 km/h wind speed. They a e no sui able
o ope a e in ainy condi ions because he elec onics can be damaged and he da a
ob ained by he came as in low ligh le els would no be use ul.
Tempe a u e and e ain al i ude a ec ai densi y, which in luences he powe
needed o ly he plane, ai c a ba e y consump ion and consequen ly endu ance and
ange. These a iables also in luence he powe equi ed o akeo , which is highe he
colde i is, o in highe e ains. This can also ansla e in o mo e ailed akeo s. In
expe imen s pe o med o o he pu poses, we ound ha ou sys em los 10 minu es o
endu ance (a ound 30%), when compa ing sea le el in summe in Spain o win e a
2,000 m in Sou h A ica.
RPAS possible nega i e e ec s
Rhinoce os did no show any ala m o discom o eac ions du ing ou ligh s.
Howe e , he e is no p oo ha RPAS could no dis u b hem o o he animals i hei
use is con inuous, so u he in es iga ion o his aspec is needed. Some a ms ha ha e
hinoce os also o e eco ou ism ac i i ies ha b ing impo an income. The e o e,
isi o accep ance o he p esence o RPAS in hose a eas would be impo an .
Choosing he igh RPAS
The ange o RPAS a ailable is ex ensi e and g owing by he day. F om mic o
sys ems ha i in he palm o a hand up o 2 ons ai planes, he e is a huge a ie y in
ma ke o e . Conside ing he scale o wo k, he unding limi a ions and he senso
equi emen s, “close ange” (Blyenbu gh 1999) RPAS seem o be he bes choice o
an i-poaching pu poses.
RPAS’ use s always wan o imp o e sys em pe o mances o maximize
endu ance, ange and senso s capabili ies (da a quali y), and o minimize ano he se o
cha ac e is ics associa ed wi h he RPAS: p ice (o he sys em and spa es), logis ics
(size, anspo a ion, aking o and landing equi emen s), and expe ience le el needed
52
o i s ope a ion. Un o una ely, any imp o emen in he sys em pe o mances en ails an
undesi able e ec in one o mo e o he second se o cha ac e is ics ha would make
RPAS less a o dable o p ac ical. Thus, he mos sui able choice is a balanced
comp omise he use has o accep conside ing all he p os and cons o his speci ic
pu poses.
Cos s and bene i s
The ecommended close ange RPAS a e ypically ligh e han 5 kg, ha e 30-45
minu es endu ance and o e an ope a ional ange be ween 5-20 km. The p ice,
capaci ies and eliabili y a y acco ding o he manu ac u e . In gene al, he e is an
in es men in a whole sys em, composed by he g ound con ol s a ion, an ennas, and
wo o h ee planes ha need o be epai ed o subs i u ed when hey each a ce ain
numbe o ligh s. As a e e ence, he sys em we used has pe o med mo e han 500
ligh s wi h an app oxima e o al in es men o 14,000 ! including he senso s payload
(see Table 1). The e a e mo e a o dable op ions a ailable in he ma ke , bu om ou
expe ience, eliabili y o some e y cheap componen s like se os, ba e ies o e en
ipods is no gua an eed and hei ailu e may cause se ious p oblems a ec ing
expensi e componen s, so i is wo h o ge a leas medium quali y spa es.
The bene i o in eg a ing RPAS in an i-poaching wo k is di icul o e alua e in
economic e ms, as i s calcula ion would in ol e o pu a p ice on he li e o a
hinoce os and o e alua e how many could be sa ed by using RPAS. I has been
poin ed ou (Fe ei a e al. 2012) ha whi e hinoce os ca y wo ypes o alues: a
comme cial alue (li e hinoce os ade and hinoce os hun ing) and a conse a ionis
o aes he ic alue. The i s one could be calcula ed (whi e hinoce os a e age p ice in
2012 was 17,330 !, eco d p ice in 2012 was 53,784 !; black hinoce os eco d p ice in
2012 was 44,969 !) bu he second one is ha dly ansla ed in o numbe s. Cu en ly
he e is no eal wo k using RPAS o be able o es ima e he numbe o hinoce os ha
could be sa ed by RPAS use o o calcula e o he ypes o su eillance cos s ha migh
be educed by using his echnology. As a e e ence, he in es men needed o a small
low cos RPAS (including spa e pla o ms, spa es, ools, e c.) ha could las o abou
wo yea s being used weekly (a ound 30,000 !), plus a ound 6,000 ! o ain ope a o s,
could be assumed by a medium size secu i y company o ins i u ions ha con ol a eas
be ween 50,000-100,000 ha (Secu i y company manage , pe s. comm.). The business o
an i-poaching is g owing, especially in p i a e land, wi h he esul ha RPAS will be

53
no only app ecia ed o hei eal use ulness, bu also as a compe i i e asse o hose
companies ha include hem in hei su eillance p og ams.
RPAS in eg a ion in an i-poaching asks
Conside ing bo h he echnical and p ac ical aspec s we p opose h ee
al e na i es o RPAS in eg a ion in o an i-poaching wo k:
1-As a sec e ool o su eillance. Secu i y companies and public en i ies could
use RPAS as a “hidden” ool o moni o sys ema ically poaching ho spo s o sensi i e
a eas in o de o ge da a, de ec in ude s, check hinoce os p esence and sa e y, as well
as p o ide e idence ha could be used on cou agains poache s. In his case, RPAS
mus be as disc e e as possible. This would en ail minimize he noise and camou lage
he plane i sel and o p e en locals o know abou i s use.
2-As a suppo ing ool du ing poaching inciden s. The ole o RPAS could be o
suppo g ound pa ols du ing he pu sui o poache s, p o iding eal ime in o ma ion
abou suspec numbe s, loca ions and mo emen s. Images aken may be used as
e idence in cou i needed. RPAS equi e less logis ics han con en ional ai c a , bu
hey s ill do equi e some. Fo his ype o e y immedia e use, echnical e o s should
be concen a ed on de eloping mobile uni s in eg a ed in small aile s o 4x4 ehicles
ha could pe mi a as deploymen .
3-As a de e en ool. Secu i y company manage s sugges ed ha by making
widely known ha he a ea is unde cons an igilance by RPAS, i would discou age
locals o poach. Tha would include pe o ming demons a ions o he local
communi ies and appea ing in media wi h awa eness campaigns, which could make
hem a aid and awa e ha hey can be de ec ed e en wi hou no ice. In his case, i
would be con enien o ocus he e o wi h RPAS on a m pe ime e s su eillance and
o ge p oo o i egula use o he a ea, gi ing media co e age o hem.
The h ee al e na i es may be combined in di e en imes o a eas o op imize
he use o he sys em. Fo example - keep RPAS use sec e un il hey con ibu e o ca ch
a poache and hen publicize i widely in he local a ea.
The e is also a ou h use o RPAS, no ela ed o poaching bu also in ol ing
hinoce os conse a ion. RPAS can p o ide quasi- eal ime in o ma ion o habi a
changes a ec ing species mo emen beha io (Rod íguez e al. 2012a). Thus,
combining high- esolu ion images o he a eas wi h indi iduals’ loca ions, RPAS can
con ibu e o answe ecological ques ions ha ha e been iden i ied as key conse a ion
ac o s, such as popula ion densi y, nu i ion and die (Knigh 2011).
54
We also o esee a p omising ield o wo k using o he senso s (like s a ic
su eillance came as and mo emen de ec o s) ha could wo k oge he wi h RPAS
o ming an he e ogeneous coope a ing objec s ne wo k o sensi i e a eas su eillance.
CONCLUSSIONS-MANAGEMENT IMPLICATIONS
Ou s udy is he i s app oach using emo ely pilo ed ai c a sys ems o an i-
poaching asks and i can be expanded o o he a eas o species ha su e om he
same p oblem. Some o he A ican and Asia ic coun ies ha e hinoce os poaching
p oblems oo, (Milledge 2007; Ma in and Ma in 2010) and la ge mammals such as
elephan s also su e om illegal hun ing (Dublin 2011). We ha e demons a ed ha
cu en low cos RPAS p esen enough echnical capabili ies o p o ide use ul da a, bu
he e a e also impo an p ac ical and echnical limi a ions ha mus be conside ed,
e alua ed and sol ed by use s and au ho i ies be o e hese sys ems can be deployed in a
ealis ic way (see Table 3 o a summa y o he bes and wo s scena ios). The ole
RPAS can play in an i-poaching should no be o e es ima ed and in es men in his
echnology should be p opo ional o he esul s ob ained because he esou ces o
hinoce os conse a ion a e limi ed.
ACKNOWLEDGEMENTS
We hank a m owne s and he secu i y company (who p e e ed o emain
anonymous due o comme cial easons and hei hinoce os’ sa e y) o p o iding
aluable in o ma ion used in his s udy, he lodging and he logis ics o he ield
campaign. We also hank he logis ical suppo p o ided by he Cen e o Wildli e
Managemen , Uni e si y o P e o ia and CSIR du ing he campaign. We a e g a e ul o
Es eban Gue e o and Miguel Ángel Aguila , he echnicians o he Ae omab and Plane
p ojec s, who pilo ed he ai c a and wo ked in da a p ocessing. Addi ionally, we wish
o hank Ai am Rod íguez, Ma cello D’Amico and Manuela González o hei e iews
and o p o iding aluable commen s on his manusc ip .
55
TABLES
Table 1. Cos o he RPAS equipmen (Ma e ial bough in Spain in June 2012)
Componen
P ice (!)
Ai ame wi h he elec onic sys em
1,000
G ound con ol s a ion (an ennas included)
6,000
S ills Pho o Came a
450
HD Video came a
300
The mal came a
6,000
To al
13,750
56
Table 2. Fligh s esul s.
Came a
Time
pe iod
Time
s a
Time
end
Ta ge
Habi a
Resul
Al i ude (m)
(Min-Max)
S ill pho o
Mo ning
09:03
09:26
People
G assland, Mixed
Con i med
32-149
Fences
Mixed
Con i med
40-175
09:05
09:38
Rhinoce os
Fo es
Con i med
57
09:42
10:02
People
Mixed, Fo es ,
g assland
Con i med
29-82
Fences
Mixed
Con i med
42-72
09:52
10:12
Rhinoce os
Fo es
Con i med
31-137
Midday
10:16
10:39
Fence
Mixed
Con i med
50-175
People
G assland
Con i med
123-158
11:22
11:43
Rhinoce os
G assland, Fo es
Con i med
38-239
13:14
13:56
People
Fo es
No con i med
E ening
17:19
17:38
Rhinoce os
Fo es
Con i med
82
People
G assland, Fo es
No con i med
Fences
Mixed
No con i med
The mal
ideo
Mo ning
07:51
08:11
Fence
Mixed, G assland
Con i med
27-155
People
Mixed, G assland
Con i med
31-100
08:21
08:55
Fence
Mixed
Con i med
37-98
People
Mixed
No con i med
08:27
08:56
Fence
Mixed
No con i med
09:25
10:03
Fence
Mixed
Con i med
48-54
People
Mixed
No con i med
Midday
10:27
10:46
Rhinoce os
Fo es
No con i med
10:40
11:07
Rhinoce os
Fo es
No con i med
12:32
13:04
Rhinoce os
Fo es , G assland
No con i med
Nigh
18:19
19:02
People
G assland, Fo es
Con i med
12-125
Fences
Mixed
No con i med
18:41
19:00
Rhinoce os
Fo es
No con i med
19:17
19:40
Fence
Mixed
No con i med
People
G assland
Con i med
36
19:27
19:45
Rhinoce os
G assland
No con i med
Visual
ideo
Midday
11:08
11:27
Fences
Mixed, Fo es ,
G assland
Con i med
10-17
People
Mixed, Fo es ,
G assland
Con i med
10-35
We p o ide he minimal and maximum al i ude a which a a ge was con i med in each
ligh . When only one alue is p esen ed i means ha he a ge was loca ed jus once.
63
cen u y. Howe e , he popula ion has been conside ed s able o he las wo decades,
and consequen ly, i has been ecen ly downlis ed om ‘Vulne able’ o ‘Leas Conce n’
acco ding o IUCN c i e ia (Bi dLi e In e na ional 2011). P esumably, he main cause
o he decline o he lesse kes el in wes e n Eu ope was habi a loss and deg ada ion as
a esul o ag icul u e in ensi ica ion (Bi dLi e In e na ional 2011). Du ing he chick
ea ing pe iod, lesse kes els selec ield ma gins and ce eal ield as o aging a eas
(Tella e al. 1998; F anco e al. 2004). In addi ion, kes els associa e wi h g ain
ha es e s o ca ch he a h opods lushed by hese machines. One o he mos impo an
s uc u al changes associa ed wi h ag icul u e in ensi ica ion is ield enla gemen , and
consequen ly, he educ ion o ield ma gins (Rod íguez and Wiegand 2009). Likewise,
he use o machines o ha es ce eal ields has educed he ime o ha es ing a a
locali y o jus some weeks o days. So, bo h ac o s a e concu en ly limi ing kes el
o aging oppo uni ies.
S udy a ea
Due o he lesse kes el decline and also o esea ch pu poses, se e al b eeding
p og ams ha e been pu in place in Spain in ecen yea s (Poma ol 1993; Neg o e al.
2007; Alcaide e al. 2010). One o hese ein oduc ions was ca ied ou in he oo o
ou own ins i u e (Doñana Biological S a ion, Se ille, Spain), whe e we conduc ed his
s udy. In 2008, a hacking p og am was s a ed eleasing o he wild a o al o 149
nes lings (51, 58 and 40 in 2008, 2009 and 2010, espec i ely) o igina ing om a
cap i e b eeding p og am (DEMA, Almend alejo, Spain, www.demap imilla.o g). In
addi ion, inju ed adul bi ds (1–4 indi iduals) we e main ained du ing ou b eeding
seasons (2008– 2011) a an ex e nal cage (6x2x2 m) o acili a e conspeci ic a ac ion
a he colony. B eeding pai s es ablished hemsel es a he colony a e he second yea
(one, h ee, six and h ee b eeding pai s in 2009, 2010, 2011, 2012, espec i ely). The
colony is o med by wo elonga ed cons uc ions on he oo o a i e- loo building.
Fo y wooden nes boxes wi h sliding doo s o cap u e he bi ds a he nes s om inside
he building a e open o he no h wall (see Figu e S1). Al hough he colony is loca ed
wi hin he u ban a ea o Se ille, i is in he no he nmos edge o he ci y acing
ag icul u al ields and he communica ion ing o he ci y (highways, ail oads, and a
high densi y o powe line co ido s). Ag icul u al ields ex end owa d he no hwes ,
he nea es ones being no mo e han 500 m away om he colony.

64
Unmanned Ae ial Sys ems (UASs)
The ae ial pla o m was buil in o a ST-model Easy Fly plane (S -models, China)
wi h a wingspan o 1.96 m and a weigh o abou 2,000 g (Figu e S2). I is p opelled
using a b ushless elec ical engine (li hium polyme ba e y). The UAS was con olled
om a g ound s a ion using a long- ange adio con ol sys em. I ca ied an onboa d
ideo came a, a GPS (10 Hz, Media ek, model FGPMMOPA6B), a da a logge wi h a
ba ome ic al i ude senso Eagle ee GPS logge V.4 (Eagle ee sys ems, WA, USA), an
Ika us au opilo (Elec onica RC, Spain), which p o ided ligh s abiliza ion and On
Sc een Display (OSD), and a Panasonic Lumix LX-3 digi al pho o came a 11MP
(Osaka, Japan). The came a was in eg a ed in he plane wing aimed o he g ound, and
was ac i a ed using a mechanical se o, se in speed p io i y mode and in i s wides
zoom posi ion. The Ika us OSD p o ided GPS in o ma ion abou he posi ion, speed,
heigh and cou se o he ai c a . These da a we e combined wi h he ideo signal om
he came a and sen o he g ound s a ion in 2,4 GHz. The au opilo p o ides
s abiliza ion o he ai c a , waypoin ollowing capabili y (including al i ude) and an
‘‘eme gency e u n home’’ unc ion. The ake-o and landing o he plane is by manual
con ol. The g ound s a ion is composed by a moni o , a DVD eco de , he ideo
ecei e and he con ol signal ansmi e wi h hei associa ed an ennas. I also
includes a Lap op PC o p og am he au opilo , o s o e he pic u es and da a logs, and o
decode in- ligh eleme y allowing o ack he posi ion o he UAS in eal ime on a
Mic oso map (Redmond, WA, USA).
Expe imen al p ocedu es
Du ing he 2011 nes ling pe iod (June–July), we i ed 5 g GPS da a logge s o
bo h membe s o wo b eeding pai s o kes els using Te lon ibbon backpack ha nesses
(Mic o size, T ackPack, Ma shall Radio Teleme y, No h Sal Lake, U ah, USA). Two
GiPSy2 GPS da a logge s (2361566 mm, 1.8 g plus 3.2 g ba e y, Technosma , I aly)
we e p og ammed in con inuous mode (1 ix/ sec) o a ou s hou pe iod. To a oid
moni o ing abno mal beha io due o cap u e s ess and ha ness i ing, bi ds we e i s
cap u ed and i ed wi h a ha ness and a 5 g dummy GPS da a logge . One week la e
bi ds we e ecap u ed and he dummy subs i u ed by a eal GPS da a logge
p og ammed o s a eco ding da a he nex day a e ecap u e. To download he da a
om he da a logge s, bi ds we e ecap u ed a hei nes boxes when hey we e
deli e ing ood o hei nes lings, a e ba e ies we e exhaus ed one day la e .
65
A e he download o he bi d acks, six ligh s we e made by he UAS. Th ee
o hem wi h he aim o epea ing he ligh s made by he lesse kes els om hei nes s
o hei o aging a eas, and h ee addi ional ligh s ollowing andom ansec s o e he
ag icul u al ields. Random ligh s connec ed loca ions andomly selec ed in a s aigh
line. Pic u es o he a ea o e lown we e aken using he onboa d pho o came a ha was
shoo ing con inuously while he ai c a was ollowing he ou es.
Da a analysis
Gi en ha he accu acy on al i ude measu emen s o he GPS used o
na iga ion is ela i ely low, o geo e e ence he pic u es aken by he came a onboa d
we used in o ma ion p o ided by an Eagle ee GPS logge V.4 (Eagle ee sys ems, WA,
USA) ha includes a ba ome ic al i ude senso . The pic u es we e geo e e enced using
a cus omized ex ension o ENVI so wa e ha used Eagle ee da a o gene a e GeoTIFF
iles.
Images aken om he UAS le us clea ly iden i y six ypes o ield c ops (o
land uses): ha es ed ce eal, ully g own ce eal (unha es ed), oli e ees, sun lowe s,
allow land and ‘o he s’ (e.g., a m houses, ba ns, oads, s eams). Using A cGIS .10
(ESRI, Redlands, CA, USA), we measu ed he pe cen age o o al dis ance o e lown
by he UAS o e each ield ype, as well as he numbe o ield ma gins c ossed by he
UAS. To e alua e he capaci y o UAS o ollow kes els’ ou es, we used he ool
‘NEAR’ implemen ed in A cGIS o calcula e he dis ance be ween each kes el ix o
he nea es UAS ix. Fo his analysis, we dele ed he pa o kes el acks ela ed o
ac i e hun ing ac i i ies and dis inc om displacemen ligh s be ween he colony and
he ac ual o aging g ounds (easily ecognizable by changes in ele a ion, di ec ion and
speed be ween consecu i e ixes a he dis al pa o acks; see Figu e 1).
RESULTS
We ob ained 4,460 high esolu ion images along six di e en ligh s ( h ee
ollowing he kes els plus h ee andom ansec s), bu he e was a high deg ee o
o e lap, and we inally selec ed 466 o hem o build he pho o-mosaics. The kes el
ac ual ligh s eco ded by he bi d da a logge s we e always included in he image y
aken by he UAS (Figu e 1). UASs ollowed he kes el acks wi h high p ecision, wi h
he majo i y o eco ded dis ances be ween kes el and UAS ixes lowe han 50 m. The
75 h and 90 h pe cen iles we e 85.9 and 128.9 m, espec i ely (Figu e 2). Spa ial
66
esolu ion o image y depends on he al i ude a which images a e aken (Figu e S3).
Ou UAS lew a a mean al i ude o 184 m, and hus, he mean spa ial esolu ion o
image y was 7.7 cm.
The a ea o e lown by kes els is in ensi ely cul i a ed, being di ided in o small
plo s o sun lowe , ce eal (mainly whea ), oli e g o es, and o he mino cul i a ions.
P opo ions o o e lown ield ypes did no show signi ican di e ences be ween
ligh s (i.e. go, e u n and andom ansec ligh s; Table 1), so ha kes els lew hem in
p opo ion o hei a ailabili y. Addi ionally, go and e u n ligh s did no di e om
he andom ligh s pe o med by he UASs in ela ion o he p opo ion o habi a ypes.
This sugges s ha he kes els did no ollow speci ic p ospec ing s a egies when
ge ing o he o aging a eas o lea ing hem. Howe e , local en i onmen al condi ions
a ec ing kes el ligh decisions a a mic oscale, such as wind gus s, could no be
eco ded in ou ae ial pho og aphs.
DISCUSSION
The lesse kes el is one o he smalles ap o s in Eu asia and i s size, and
pa icula ly body mass, poses a se ious limi o he weigh o bio eleme y de ices o
logge s ha can be a ached (abou 5– 6 g maximum, depending on he indi idual) o
eco d spa ial posi ion o beha io al ac i i y. Du ing he cou se o ou in es iga ions on
he lesse kes el, ha began in 1988 (Neg o 1997), we ha e always pu sued o ge an
accu a e knowledge o hei daily mo emen s a hei b eeding g ounds. Applying adio
ansmi e s and di ec beha io al obse a ions o unma ked indi iduals we ha e been
able o de e mine o aging habi a p e e ences (Donáza e al. 1993; U sua e al. 2005;
Ribei o 2007) bu soon ealized ha we los ack o he bi ds mo e o en han we
loca ed hem, biasing ou s udies o loca ions nea he b eeding colony. La e on,
geoloca o s ha e pe mi ed us o de e mine ha kes els om sou he n Spain win e ed
in he Sahel a ea o wes e n A ica (Rod íguez e al. 2009). While his was a
b eak h ough wi h conse a ion implica ions, due o he low spa ial p ecision o he
echnology, i was useless o moni o mo emen s a he b eeding g ounds. I was no
un il ecen ly ha p og ammable GPS da a logge s small enough o be i ed in a lesse
kes el became a ailable.
This echnology has e ealed ha indi idual kes el some imes o age 15–20 km
away in s aigh line om he b eeding colony (da a no shown). A ques ion eme ged as
wha ype o habi a s he kes els we e selec ing ou o he a ailable ones. Lesse
67
kes els a e colonial bi ds ha exploi sudden ou bu s o in e eb a e p ey (C amp and
Simmons 1980). They de end no o aging g ounds and locks o se e al bi ds may be
sigh ed ho e ing and di ing a imes on g ound-based o low lying po en ial p ey
(C amp and Simmons 1980). Al hough in o ma ion on c op ypes may be ob ained om
sa elli e images, kes els a e known o espond o apid s uc u al changes o ege a ion
in hei en i onmen (Ribei o 2007). A lock o kes els may hun on a pa icula
ha es ed ield o one o wo days and ne e be back. Keeping his in mind, we used
he UAS, as i could be deployed immedia ely a e we downloaded GPS da a om
indi idual kes els.
The esul s p esen ed he e a e mean as a demons a ion o he capabili ies o he
UAS o ob ain a mosaic o images co espond- ing o he ac ual ull o aging ips o
ee- anging small bi ds. The UAS ligh pa hs ep oduced he kes el ligh s eliably, as
indica ed by he ac ha hei ajec o ies ended o be less han 100 m apa (see Figu e
2). The p ecision i o he UAS au opilo depends on he numbe o waypoin s included
in he se ings (no e ha ou Ika us au opilo admi s 32 waypoin s), as well as he
me eo ological condi ions, so we o esee p ecision will be imp o ed using be e
au opilo s. In addi ion, images aken by he came a ins alled in he UAS lying a
a e age al i ude o 184 m abo e sea le el co e ed an a ea on he g ound ha always
con ained he bi d ack p ojec ion (Figu e S4). Pos -p ocessing o he pic u es esul ed
in a mosaic o geo e e enced images allowing an e alua ion o habi a ypes as well as
plo sizes and o he landscape ea u es, such as g assy ield ma gins, oads, powe lines,
o e en he p esence o ha es e s in he ields (da a no shown; bu see Figu e 1 o
examples o ield ma gins and oads). In ac , UAS images aken om a mean al i ude
o 184 m showed a highe esolu ion (7.7 cm) han eely a ailable sa elli e images (e.g.
hose coming om MODIS, 250 m, o Landsa TM o ETM+, 30 m), unde eques
comme cial sa elli e images (e.g. Digi alGlobe, Colo ado, USA, 30–65 cm) o
o hopho og aphies (e.g. Jun a de Andalucía, Spain, 1–1.5 m).
To ob ain habi a in o ma ion, he e a e o he al e na i e (o complemen a y)
op ions (see Table S2). The mos basic would be o ge o he s udy a ea and su ey i
by oo o using a g ound ehicle. This is ime consuming, i has logis ical
complica ions and some landscape a iables (a la ge scales) may no be easily
quan i ied. S a iona y came as o senso s sca e ed in he landscape can p o ide
in e es ing in o ma ion abou en i onmen al changes, bu hey in ol e a huge economic
in es men and p e ious knowledge o animal mo emen s, long pos -p ocessing o he
68
da a, and i is always isky o he equipmen , especially in open a eas whe e hey can
be damaged o s olen. Sa elli e images a e e y use ul o spa ial s udies, bu hei
spa ial and empo al esolu ion may no sui esea ch objec i es. In ou s udy case,
eely a ailable sa elli e images do no each he necessa y spa ial and empo al
esolu ion o dis inguish changes in he highly dynamic habi a (e.g. ha es ed s. non-
ha es ed ields). Fo example: NASA’s Ea h Obse ing Sys em Da a and In o ma ion
Sys em (EOSDIS) can p o ide only 250-m esolu ion images om MODIS senso
wice a day o Spain; bu hey a e a ec ed by clouds and ha e a spa ial esolu ion oo
low o ou aims. Comme cial sa elli e images wi h he app op ia e spa ial esolu ion
could be a ailable, bu a a high cos and he e is g ea e delay in da a acquisi ion
compa ed wi h UAS. Ae ial pho og aphs can be o de ed om specialized i ms, bu a
mosaic o geo e e enced images o he landscape would be qui e expensi e, and i
would be logis ically p oblema ic o ob ain he pic u es when needed, i.e. a he desi ed
empo al esolu ion.
In he case o small bi ds, he ec ea ion o ligh pa hs o bi ds has been
achie ed using adio- acking de ices and minia u ized ideo came as (Ru z e al. 2007;
Millspaugh e al. 2008; Ru z and Blu 2008; Blu and Ru z 2008; Sakamo o e al.
2009; G émille e al. 2010). Howe e , i home ange is la ge enough o lose he adio
signal o he e is no p e ious in o ma ion on whe e he bi ds a e mo ing, his
me hodology may bias he esul s (see (Millspaugh e al. 2008)). In la ge bi ds,
came as ha e been a ached on hem (e.g. seabi ds (Sakamo o e al. 2009; G émille e
al. 2010)), bu in a non-sys ema ic way and wi h no possibili y o ge zeni hal images o
enough high quali y ha could be p ocessed in a s a is ical manne . In ou case, he e is
admi edly a delay o se e al hou s be ween he ligh o he bi d and ha o he UAS,
bu his is o li le ele ance o answe ing mos o ou ecological ques ions.
In ou s udy, GPS da a o bi d posi ions was ob ained a a equency o one ix-
pe -second. In he ade-o among ix equency s. leng h o he egis a ion pe iod,
we a o ed he o me o imp o ed spa io- empo al accu acy. Ou decision es ed on
wo ac s: one, his con igu a ion le us o dis inguish among soa ing, gliding and
hun ing ligh s (i.e. ho e ing and s ikes) acco ding o ele a ion, di ec ion and speed o
ixes; and wo, he kes els we we e acking, e en i ee- anging, we e easily cap u ed
in he colony si ua ed on he oo o ou headqua e s. This condi ion, he easy o
e ie ing he GPS da a logge o download da a, is no me in a majo i y o
in es iga ions on wild bi ds (Millspaugh e al. 2008). The e o e, u u e echnological

69
ad ances o inely ack a wide ange o small sized species should include emo e
wi eless downloading o he GPS in o ma ion by GSM, Blue oo h o adio. Fo he
momen , his echnology has only been inco po a ed o ela i ely la ge de ices ha can
only be moun ed on co espondingly la ge bi d species (see www.cell ack ech.com,
www. echnosma .eu, ( an Die men e al. 2009)). In addi ion, UASs can be con igu ed
o ca y on boa d addi ional senso s, such as ba ome e s, he mome e s o ideo
came as. These capabili ies o he UAS as a non- in usi e ool o ecological esea ch
can also be en isaged as ex emely use ul in s udies o ligh dynamics (e.g. eco ding
a mosphe ic pa ame e s such as empe a u e, wind di ec ion and s eng h, o ba ome ic
p essu e (Shepa d e al. 2011)), p eda o -p ey in e ac ions (e.g. eco ding UV ligh om
p ey u ine acks which may a ac o p eda o s (Vii ala e al. 1995)), social dynamics
(e.g. moni o ing bi ds o di e en species du ing mig a ion (Chabo and Bi d 2012)) o
beha io al decisions ela ed o he conse a ion o species (e.g. eco ding wha
shea wa e ledglings would see when hey a e a ally a ac ed o a i icial ligh s du ing
hei i s ligh s om nes -bu ows o sea (Rod íguez and Rod íguez 2009; Rod íguez
e al. 2012b)). As a u u e e inemen , UASs may also be used o loca e and ack a a
sa e dis ance animals equipped hemsel es wi h adio ansmi e s o o he loca ing
de ices. All he hea y equipmen , such as ideo o s ill came as, would go in he UAS
and he animal would jus ca y a ligh weigh loca ion de ice.
Ou UAS lew p og ammed ou es, p o iding geo e e enced images o he a ea
o e lown by kes els. The combina ion o he GPS posi ion p o ided by he da a
logge s and he images p o ided by he UAS ec ea e he ajec o y o a bi d ca ying a
came a. I imp o es, howe e , he pe o mance o he o he echniques a ailable o da e
o s udy he en i onmen as con en ional ieldwo k, sa elli e image y, ae ial pic u es o
s a iona y came as.
ACKNOWLEDGEMENTS
We hank Manuel Baena o i ing a ideo came a o moni o he nes s, Ca los
Ma il and Miguel Fe nández o hei help du ing he cap u es o lesse kes el in he
colony, and Da id A agonés and Isabel A án o hei ad ice wi h GIS wo k a LAST-
EBD. Es eban Gue e o and Miguel Ángel Aguila , o he Ae omab p ojec eam,
manned he UAS and p epa ed he image mosaic. Ra a Sil a and Pepe Ojeda buil he
simula ion ligh o he kes el and he UAS. Todd Ka zne , Pascual López-López,
Ka he ine Ren on and an anonymous e iewe made use ul commen s on ea ly d a s o
70
his manusc ip . We hank Jun a de Andalucía and UE o g an ing he esea ch p ojec s
(i.e., AEROMAB, HORUS and PLANET)
TABLES
Table S1 Budge a y cos o he equipmen used in his s udy.
71
Table S2 P os and cons o commonly used echniques o eco ding en i onmen al
in o ma ion. This able is based on ou s udy case, i.e. an ac ual case o s udy he habi a
selec ion o Lesse Kes el using he kes el ligh acks. No e ha
ad an ages/disad an ages may change acco ding o he aims o he s udies.
72
FIGURES
Fig. 1. T ack o a lesse kes el o aging ligh o e he images ob ained by an
unmanned ae ial sys em. A Whi e and black acks co espond o unmanned ae ial
sys em and lesse kes el ligh s, espec i ely. The ci cle indica es he hun ing a ea. The
ec angle indica es he enla ged a ea in B. B High esolu ion images showing
sun lowe s, oli e ees, oad and ha es ed ce eal ields.
79
a non-in asi e me hodology able o p o ide accu a e spa ial da a use ul o ecological
esea ch, wildli e managemen and angeland planning.
INTRODUCTION
Assessing he dis ibu ion o animal species among a ailable en i onmen s and
he easons behind hose pa e ns a e ecu en ecological ques ions ha may also a ec
human ac i i ies and conse a ion e o s (Mo ison e al. 2006). Resou ce u iliza ion,
wildli e managemen , conse a ion planning, ecological es o a ion and p edic ion o
possible u u e impac s o land use o clima e changes a e all applied a eas ha bene i
om spa ial dis ibu ion models o indi iduals, popula ions, species and communi ies
(Collinge 2010; Qama e al. 2011).
Nume ous me hodologies a e a ailable o collec spa ial da a o animals. Di ec
me hods include obse a ion, cap u e, bio eleme y, ada , lase and came as, whe eas
indi ec me hods a e dependen on some e idence o animal ac i i y in an a ea o
speci ic si e (e.g. bed si es, eces, nes s o acks) (Mcdonald e al. 2012). Bio-logging
consis s in he emo e da a collec ion om ee- anging animals using a ached
elec onic de ices (Cooke e al. 2004). This is an inc easingly popula op ion among
ecologis s because i p o ides aluable in o ma ion on he animals’ mo emen s and
habi a use. This me hod has expe ienced a ema kable de elopmen hanks o he
con inuous echnological ad ances, especially hose ega ding ags minia u iza ion in
ecen yea s. Ne e heless, bio-logging echniques p esen some cons ain s, including
logis ical challenges, possible undesi able e ec s on he animals du ing he cap u e,
handling and along he pe iod on which he indi iduals a e agged (see Mu ay & Fulle
2000 o a e iew) and he limi a ion in he numbe o animals ha can be s udied,
cons ained by he numbe o ags deployed, which a e o en expensi e (Ru z and Hays
2009).
Reliable es ima es o species abundance a la ge spa ial scales a e highly
demanded in o de o es ablish bases on which managemen schemes can be sus ained.
I is well known ha wildli e popula ion abundance is no easily es ima ed and lo s o
me hods a e desc ibed o his pu pose in he scien i ic li e a u e (e.g. Mo elle e al.
2010). Fo a gi en species, he e o equi ed o apply each me hod is highly a iable
and i de e mines hei applicabili y o be used, mainly a la ge spa ial scales.
Ob iously, he e o s equi ed o de e mine he abundance o a species a la ge spa io-
empo al scales exclusi ely om ieldwo k a e unwo kable o mos o he s udies.

80
Thus, su eying a numbe o ep esen a i e popula ions, on which he ela ionships
be ween species abundance and he en i onmen al condi ions can be de e mined, is a
way o o ecas he abundance in unsampled e i o ies, by gene alizing he adjus ed
species abundance–en i onmen al g adien s ela ionships (e.g. E he ing on e al. 2009;
Ace edo e al. 2014). A his ega d, o eco d eliable in o ma ion o species
abundance is one o he challenges o wildli e managemen .
Unmanned Ae ial Sys ems (UAS he eina e ) ha e p o en use ul o add ess
a ious ecological challenges in ol ing animal moni o ing (Jones 2003; Wa s e al.
2010; Sa dà-Palome a e al. 2012; Ve meulen e al. 2013) and habi a cha ac e iza ion
(Koh and Wich 2012; Ge zin e al. 2012). The po en ial alue o UAS o spa ial
ecology is eno mous (Ande son and Gas on 2013) bu o da e, he e a e jus a ew
s udies ha ha e explo ed hei possibili ies (Rod íguez e al. 2012a; Chabo e al.
2014).
The aims o his wo k a e o es he sui abili y o ae ial images ob ained om
UAS ligh s o i) modeling spa ial dis ibu ion pa e ns o animals as compa ed agains
a widely used me hod (bio-logging using GPS-GSM colla s), and ii) p edic ing species
abundance by compa ing wi h eal abundance in he s udy a ea. We use as model
species ee- ange ca le Bos au us inhabi ing Doñana Na u e Rese e (Sou hwes o
Spain) unde a adi ional husband y sys em. Ca le a e la ge mammals ha o e
logis ical ad an ages o bio-logging deploymen , a e easily de ec able in UAS images
and ac ual abundance da a a e a ailable. In addi ion, he knowledge o he spa ial
dis ibu ion o hese la ge he bi o es is c i ical o ecosys em managemen (Lazo 1995;
Bailey e al. 1996). Resea che s and pa k manage s a e specially in e es ed because
ca le p esence and hei o aging impac in he p o ec ed a ea is a con o e sial issue
(Espacio Na u al Doñana 2000). Heal h issues a e also a s ake, as ca le sha e habi a ,
esou ces and diseases such as ube culosis wi h wild ungula es (see Go áza e al.
2008).
MATERIAL AND METHODS
S udy si e and species
Doñana Na u e Rese e (DNR he eina e , 37°0' N, 6°30' W) is loca ed in he
igh bank o he Guadalqui i i e es ua y in he A lan ic Coas (Andalusia, Sou hwes
o Spain). DNR co e s 1,008 km2 and hos s a unique biodi e si y and ecosys ems
including ma shlands, lagoons, sc ub woodland, o es s and sand dunes ha led o i s
81
decla a ion as a Wo ld He i age Si e and Biosphe e Rese e (UNESCO 2014). The a ea
has a Medi e anean clima e classi ied as d y sub-humid wi h ma ked seasons. We
pe o med he ield wo k du ing he d y season, when s udy a ea includes he ollowing
main habi a s: (LT1) dense sc ub domina ed by E ica scopa ia and Pis acia len iscus,
(LT2) low-clea sh ub land, mainly o Halimium halimi olium, Ulex mino and Ulex
aus alis (LT3) he baceous g assland, (LT4) Eucalip us sp. and Pinus sp. woodlands,
(LT5) ba e lands, sandy dunes and beaches, (LT6) wa e bodies and ege a ion
associa ed wi h wa e cou ses co e ed mainly by Juncus sp. pa ches (Fig 1). A no h-
sou h o ien ed longi udinal humid eco one can be iden i ied be ween he sc ublands and
he edge o he d y ma shlands, domina ed by Sci pus ma i imus and Galio palus is
wi h Juncus ma i imus associa ions. The s udy a ea in DNR is di ided in ou
Managemen A eas (MA he eina e ) om Sou h o No h named espec i ely:
Ma ismillas (MA1), Pun al (MA2), Biological Rese e (MA3), and So os (MA4).
Ou model species is ee- anging ca le Bos au us ha occupy di e en MA
and a e cap u ed jus once pe yea o sani a y handling. Since 2000, ca le a e
managed acco ding o he Ca le Use Plan (Espacio Na u al Doñana 2000) which
de e mines he numbe o indi iduals allowed on each MA (MA 1=318, MA 2=152,
MA 3=168, MA 4=350). Doñana ca le is an au och honous b eed, named Mos enca,
al hough some c oss-b eeds exis in some he ds.
Unmanned Ae ial Sys ems (UAS) me hodology
We comple ed a o al o 192 km o UAS diu nal ae ial acks o wo ypes (eas -
wes and no h-sou h o ien ed ansec s) on each ca le managemen a ea wi h six
eplica es (Fig 1). UAS su eys ook place du ing Augus and Sep embe 2011, he end
o he d y season and a ime when ood esou ces become mo e limi ing o he bi o es
in DNR in e ms o wa e and o age a ailabili y (see Bugalho & Milne 2003) be ween
3 p.m. and 8 p.m. (local ime). The acks we e pe o med a an a e age speed o 40
km/h a 100 m al i ude abo e g ound le el. The co e ed s ips we e app oxima ely 4 km
long and 100 m wide (Fig 1).
The ligh s we e pe o med wi h a small UAS (1.96 m wingspan; see Fig 2)
assembled a Doñana Biological S a ion using a oam uselage o an Easy Fly plane (S -
models, China) p opelled by an elec ical engine. I is equipped wi h an Ika us au opilo
(Elec onica RC, Spain), which p o ides waypoin ollowing capabili y and an
Eagle ee GPS logge V.4 (Eagle ee sys ems, WA, USA) wi h a ba ome ic al i ude
82
senso . The digi al pho o came a Panasonic Lumix LX-3 11MP (Osaka, Japan) is
in eg a ed in he plane wing nadi poin ing and he shoo is ac i a ed by a mechanical
se o. The images we e aken in speed p io i y mode and in i s wides zoom posi ion
wi h con inuous shoo ing. To al p ice o he sys em was a ound 5,700 ! as o June 2012.
We geo- e e enced he images using he in o ma ion p o ided by he UAS wi h
a cus omized ex ension o ENVI so wa e using Eagle ee da a o p oduce GeoTIFF
iles. Accu acy o ou UAS loca ions is es ima ed in he ange o 10-50 m (Mule o-
Pázmány e al. 2014, au ho s' unpublished da a) be o e pos -p ocessing, and was
imp o ed up o 1-3 m a e GIS co ec ions (supe imposing he image on o hopho os
and manually co ec ing i by using e e ence poin s). We aced he animals in he
images and p ocessed hem o e a 1 ha app oxima ed pa ch size (g id) as p oposed in
de ailed s udies on ungula e beha io (Gibson and Guinness 1980).
GPS-GSM me hodology
Twel e Mos enca b eed ca le we e equipped wi h GPS-GSM colla s along July
2011 in he Biological Rese e (MA3) (Fig 2), du ing ou ine e e ina y inspec ions
wi h he animals es ained in a ca le chu e. The colla s included a sa elli e posi ion
cap u e sys em (GPS) and a Global Sys em o Mobile communica ions (GSM)
(Mic osenso y Sys em, Spain) (Cano e al. 2007). The p ice pe colla is 2,750! plus
sms se ice, co e ed by he manu ac u e s in ou case. The colla s we e p og ammed o
ake a GPS loca ion e e y hou , sending encoded packe s wi h 20 posi ions o he
cen al s a ion when mobile phone co e age allowed. Da a collec ed included da e,
ime, geog aphic coo dina es and Loca ion Acquisi ion Time (LAT he eina e ,
p ecision measu e o ob ain a ix; ange om 0 o 160 sec). We sc eened ou da a using
LAT # 154 sec o de ec anomalous ixes (manu ac u e 's echnical da a; Mic osenso y
Sys em, Spain). We ob ained a ix- a e o 93.95%, which is accep able conside ing ha
ix a e success o < 90% can cause habi a -induced bias in esou ce selec ion s udies
(F ai e al. 2004). Posi ional e o associa ed wi h GPS loca ions was 26.64 m on
a e age, SD= 23.5 m, acco ding o s a iona y es s ca ied ou in he cen e o ou s udy
a ea.
Da a analysis
Landscape co a ia es.
83
En i onmen al a iables we e es ima ed om hema ic ca og aphy 1:10,000
scale (Conseje ía de Medio Ambien e y O denación del Te i o io. 2013) using
Quan um GIS e sion 1.8.0 Lisboa (QGIS De elopmen Team 2012) and we e
de e mined ollowing he in o ma ion o he ac o s po en ially egula ing ungula es
spa ial abundance in he s udy a ea (B aza and Al a ez 1987). Fo each 1 ha g id o he
s udy a ea ( o al = 29,532 g ids, including he 10.1% co esponding o UAS ack g ids;
n=2,983) and o each 26 m adius bu e (acco ding o GPS posi ional e o ) a ound
each GPS ca le (used) and andom (a ailable) loca ions (Je de and Vissche 2005), we
calcula ed: dis ance o nea es a i icial wa e hole (DW); dis ance o nea es ma sh-
sh ub eco one (DE); exac g id a ea (GA) o con ol he a ia ion in UAS image a eas in
he case o UAS ack g ids, and p opo ion o he di e en land co e ypes (LT1-
LT6). Dis ances, a eas and land co e ype p opo ions we e ea ed as con inuous
a iables (Table S1) and ca le managemen a ea (MA) as a ca ego ical a iable.
Dis ance a iables we e ob ained as he sho es dis ance om each g id and bu e
cen oid o he nea es en i onmen al ea u e.
To co ec isibili y educ ion p oduced by ege a ion co e o ca le de ec ion
in UAS images, we calcula ed de ec ion coe icien s o LT1 and LT4 land co e ypes.
We es ima ed he de ec ion p opo ion o 100 andom ci cle poin s (1 m2 size) c ea ed in
QGIS om en di e en habi a images (1 ha) o each ca le managemen a ea and land
co e ype (80 images analyzed). De ec ion coe icien s used in s a is ical analysis we e
0.544 o LT1, and 0.360 o LT4 espec i ely. Colinea i y be ween explana o y
a iables was es ed wi h Spea man’s pai wise co ela ion coe icien s > |0.5| (Hosme
and Lemeshow 2000).
Ca le dis ibu ion modeling
We es ed he ac o s a ec ing he spa ial dis ibu ion o ca le by (i) using UAS
images as a i s app oach; and (ii) using GPS-GSM colla loca ions as a second
app oach, by means o Gene alized Linea Models (GLM).
Fo he UAS model, we only included he eas -wes UAS ack da a, because
no h-sou h UAS acks showed low habi a ea u e a ia ion ( hese da a we e la e used
o model alida ion). The esponse a iable was he numbe o de ec ed animals pe
UAS g id and was modeled wi h a nega i e binomial dis ibu ion and loga i hmic link
unc ion. The inal UAS model was ob ained using a backwa d s epwise p ocedu e
based on Akaike’s in o ma ion c i e ion (AIC) (Akaike 1974).
84
Fo he GPS model, we used Resou ce Selec ion Func ion (RSF) logis ic
eg ession (Manly 2002) whe e used loca ions (only conside ing he ones ob ained
du ing he same pe iod-hou s o UAS ligh s) we e coded as 1, and andom loca ions
(a ailable, en pe used GPS loca ion), inside he indi idual Fixed Ke nel (95%
u iliza ion dis ibu ion) home anges, as 0. The esponse a iable is p esence/absence o
ca le in he g id, and he model included he a iables selec ed o UAS app oach
excep he MA ca ego ical ac o (since he colla ed animals we e es ic ed in MA3).
Valida ion and compa ison be ween he wo me hods
UAS model alida ion was pe o med by mean o Pea son’ co ela ions wi h
independen (20%) da a o he eas -wes acks and all in o ma ion in no h-sou h UAS
ack da ase . GPS model alida ion was pe o med by assessing he p edic i e capaci y
o each model wi h he a ea unde a ela i e ope a ing cha ac e is ic (ROC) cu e
(AUC), o a e he p obabili y ha he models co ec ly disc imina ed be ween used and
andom loca ions. The AUC anges om 0.5 o models wi h no disc imina ion abili y
o 1 o models wi h pe ec disc imina ion (Pea ce and Fe ie 2000).
Spa ial p edic ions o bo h inal models we e ans e ed o MA3 a ea whe e
isual and quan i a i e compa isons we e conduc ed o e i y co espondence be ween
p edic ions o UAS and GPS app oaches by Spea man’s pai wise co ela ion. All
s a is ics we e pe o med in R e sion 3.0.1 (R Founda ion o S a is ical Compu ing,
Vienna 2013).
We also compa ed he densi ies (numbe o animals / su ace) p edic ed by he
UAS model wi h he cu en densi y in he di e en MAs (da a p o ided by Doñana
Biological Rese e and Doñana Na ional Pa k au ho i ies) e alua ing ca le agg ega ion
in he g ids by a iance o mean a io (Ellio 1977).
RESULTS
A o al o 358 indi idual ca le we e iden i ied and loca ed on he UAS ack
images along DNR (Fig 2). We did no obse e any dis u bance eac ions o he UAS
du ing he o e ligh s om he ca le no om o he ungula es p esen in he a ea.
O e all, he GPS colla s ixed 1,752 loca ions o he 12 ma ked animals du ing he
same pe iod o UAS ligh s. Table S1 illus a es he desc ip i e s a is ics o he
analyzed con inuous landscape co a ia es in he UAS ack g ids, GPS (used and
a ailable) loca ion bu e s and o al MA3 and DNR g ids.

85
Resul s o he a iables included in he spa ial dis ibu ion models selec ed by
he s epwise p ocedu e ($AIC), es ima ed coe icien s, s anda d e o s and signi icance
a e summa ized in Table 1 o each app oach. The bes UAS i ing model (AIC = 397,
$AIC om sa u a ed model = -32) ound ha he en i onmen al co a ia es in luencing
ca le dis ibu ion a e mainly ela ed o landco e ypes, wi h a posi i e e ec o
g asslands on he ungula es dis ibu ion and a nega i e e ec o he dis ance o he
eco one and o sh ubs. UAS bes i ing model also e ealed a signi ican e ec o he
managemen a ea on ca le abundance. GPS me hod iden i ied all he included a iables
as signi ican and showed he same e ec o hem o e ca le p esence.
Valida ion o he model p edic i e pe o mance on independen UAS ack
da ase s showed ha he selec ed bes spa ial dis ibu ion model pe o med adequa ely
wi h signi ican Pea son’s ank co ela ions (eas -wes da a: = 0.30, p < 0.001, n =
258; and no h-sou h da a: = 0.32, p < 0.001, n = 852). The assessmen pe o med o
he GPS loca ion model showed a high p edic i e capaci y (AUC = 0.945). These
alida ion esul s pe mi ed he ans e ence o he models o he MA3 by using o al 1
ha g ids (Fig 2).
The map ep esen ing p edic ed spa ial dis ibu ion o ca le shows common
dis ibu ion pa e ns h oughou MA3 be ween UAS and GPS app oaches. High
ela ions we e ound be ween he p edic ed alues o UAS and GPS me hods in he
MA3 by Spea man’s ank co ela ion: = 0.716, p < 0.001, n=6,501.
The mean o p edic ed densi ies calcula ed by he UAS app oach o each MA
we e highe han he densi ies p o ided by DNR au ho i ies, showing di e ences
be ween he ou MA o DNR, wi h mo e o e es ima ed alues in he MA wi h highe
agg ega ion coe icien s (Table 2).
DISCUSSION
In an e o o assess he abili y o UAS o con ibu e o spa ial ecology s udies,
we compa ed he p edic ed spa ial pa e ns o ee- anging ca le in Doñana Biological
Rese e ob ained by using animal loca ions om UAS o e ligh s images agains
loca ions om bio-logged ca le (GPS-GSM colla s). Bo h models, using he same
en i onmen al co a ia es, pe o med well and p o ided simila spa ial dis ibu ions o
ca le a a e y ine scale (1 ha g ids).
Models esul s
86
The en i onmen al a iables selec ed by he UAS model o explain he
abundance o ca le a e hose expec ed o be mo e impo an om an ecological
pe spec i e. The posi i e in luence o he baceous g asslands on ungula es dis ibu ion
e lec ed by ou models has been p e iously iden i ied by o he au ho s (Bailey e al.
1996) indica ing he need o o age on g een pas u es du ing he d y season. The
eco one be ween he sh ub lands and he ma shlands is he iches a ea o DNR, keeping
mo e humidi y han o he a eas and o e ing no only g asslands bu also shade and
e uge which a e aluable o ungula es in he d y season (see B aza & Al a ez 1987).
Models also showed a nega i e e ec o dense and low-clea sh ub on ca le p esence,
ha end o a oid hose land ypes in a o o he open g assland a eas (Casasús e al.
2012). Howe e , his wo k is limi ed o da a ob ained a a speci ic ime o he day, as
ou main goal is o compa e wo me hods in he same condi ions, and he e o e gene al
habi a use by ca le should be add essed in a mo e comple e s udy pe o med all day
ound.
Al hough he UAS me hod wo ked success ully o p edic ing ca le spa ial
pa e n, i o e es ima ed ca le densi y in all he managemen a eas (Table 2). This
disc epancy may be explained because he ligh loca ions we e biased owa ds he
a eas whe e ca le is mo e concen a ed, a p oblem which could be sol ed by
pe o ming s a i ied su eys in he di e en habi a s. Also, he o e es ima ion is no
homogeneous along DNR, bu highe in hose a eas wi h a mo e agg ega ed
dis ibu ion. This ac has been p o en ele an o animal su eys in gene al and
manned ae ial censuses, mo e ela ed wi h UAS, in pa icula (Telle ía 1986; Fleming
and T acey 2008). The e a e a ious p o ocols o assess his e ec (Red e n e al. 2002;
T acey e al. 2008) and echniques o co ec i (Bayliss and Yeomans 1989; Fleming
and T acey 2008) ha should be conside ed i he esea che main objec i e was
es ima ing abundance, o ins ance inc easing sampling e o as ca le spa ial
agg ega ion does.
Me hods compa ison
Al hough bio-logging and UAS app oaches p o ed o be use ul in ou s udy,
he e a e se e al ac o s ha condi ion hei gene al applicabili y in spa ial ecology. On
he basis ha he mos desi able in spa ial ecology s udies is o maximize sampling size,
da a accu acy, di e si y and equency o bo h he animals and he habi a , while
87
minimizing impac , cos , logis ic and da a p ocessing e o , we p o ide below an
analysis o he p os and cons o each me hod.
Sampling size
Sampling size o bio-logging is limi ed by inancial cons ains and/o apping
success (Cooke e al. 2004; Ru z and Hays 2009). This in ol es a isk o da a bias
caused by he selec ion o animals o be i ed wi h ags, including ha p oduced by he
non- andom selec ion in ela ion o age, sex and geog aphic loca ion, which inc eases i
he apping me hod is no selec i e. Deployed ags can ail because hey may s op
sending da a o become los u he educing he sample size, a ac ha can lead o
biased in e ences by ocusing on he space use o a ew indi iduals while igno ing he
posi ion o non- agged animals (con- o he e o-speci ics).
Sampling size o UAS moni o ing depends on he a ea he sys em is able o
co e du ing he ligh s (which depends on i s ange and au onomy) and hei de ec ion
capaci y. Fleming & T acey (2008) analyzed e icacy o manned ae ial su eys, also
applicable o UAS, iden i ying he size, shape, colo , shadow and con as agains
backg ound o he animals, as well as hei esponse o he ai c a as ele an ac o s
o de ec ion. UAS ligh al i ude mus be a comp omise be ween ob aining adequa e
esolu ion o dis inguish he species unde in es iga ion and he size o he a ea o co e .
Ca le and o he smalle ungula e species we e easily spo ed in ou images ob ained
wi h an emba ked 11 MP comme cial came a a 100 m al i ude abo e g ound le el, in
con as wi h o he s udies (Ve meulen e al. 2013), whe e animals smalle han
elephan s could no be easily iden i ied lying wi h a 10 MP came a a he same al i ude,
maybe because hey jus made a apid naked-eye image analysis.
Species beha io and habi a cha ac e is ics also a ec de ec abili y by means o
UAS. Bayliss & Yeomans (1989) no ed ha he main sou ce o ae ial su ey bias o
e al li es ock is obs uc i e ege a ion co e . We add essed his p oblem in ou s udy
by using co ec ion isibili y ac o s adequa e o he p esen land co e s. This ac o ,
es ima ed om andom loca ion o poin s, assumes ha animals a e also andomly
dis ibu ed wi h espec o ee co e , bu i he animals selec o co e , hen he UAS
densi y es ima es could be biased low, o he opposi e i indi iduals selec ed o he wise.
Besides, selec ion o co e may a y by species, season and ime o day (in ou case all
he ligh s we e pe o med in he la e a e noon and in summe ). Equipping UAS wi h
he mal came as allows dis inguishing animals in dense ege a ion a eas o a nigh , bu
i has been p o en ha de ec abili y wi h he mal came as is low o dayligh condi ions
88
and in dense ege a ion habi a s (Mule o-Pázmány e al. 2014b). Admi edly, beha io al
esponses and habi a cha ac e is ics a e less c i ical when da a a e ob ained h ough
bio-logging. Assuming a sui able de ec ion a e o UAS, one o he main ad an ages o
his me hod e sus bio-logging is ha i p o ides he esea che wi h an image o he
animals ha a e p esen in he a ea, pe mi ing o include g oup in luence o
in e speci ic agg ega ion as a iables o he ecological s udies.
Da a accu acy, di e si y and equency
Spa ial accu acy o he animal loca ions ob ained by UAS a e p ocessing is
es ima ed be ween 1-3 m. This cons i u es a majo ad an age o UAS in spa ial
dis ibu ion s udies agains bio-logging ha p o ides less accu acy (e.g. 26 m o he
GPS colla s we used).
The use o speci ic senso s in bio-logging ags is de eloping as , allowing o
measu e indi idual pa ame e s (e.g. physiological, beha io al, mo emen speed and
ange), which is in o ma ion ha could no be ob ained wi h UAS app oach. On he
o he hand, UAS ha e he capaci y o p o ide eal ime in o ma ion on habi a
cha ac e is ics, which is especially in e es ing in highly dynamic landscapes (Rod íguez
e al. 2012a), whe e sho e m changes a ec ing animals’ mo emen s (i.e. p oduced by
i es, human in e en ions, ain alls) may no be e lec ed on sa elli e o GIS esou ces
a ailable wi h p ope spa ial- empo al esolu ion. This empo al accu acy is impo an ,
as ob aining animal in o ma ion and en i onmen al a iables a he same le el o de ail
and eliabili y would signi ican ly imp o e ecology s udies (Gailla d e al. 2010).
While apping animals may be complex, once he animals a e bio-logged hey
p oduce eno mous olumes o da a o a long pe iod o ime. Long- e m da a wi h UAS
equi es addi ional ligh ield campaigns and i is di icul o associa e da a o speci ic
indi iduals. UAS ligh s a e subjec ed o a o able me eo ological condi ions ha also
cons ain he pe iod on which da a collec ion is possible.
Impac
Bio-logging equi es cap u e and handling o he animals ha migh a ec hei
beha io and su i al (Sil y e al. 2012) and ob aining pe mi s a e ac o s ha can
complica e he use o bio-logging (Cooke e al. 2004). A poin in a o o he use o
UAS is hei low impac on he su eyed animals. Due o he small size and he educed
noise UAS p oduce, animal esponse is e y low (a leas no isually no iceable in ou
case) so ha he me hod does no dis u b he s udy subjec s. Elec ic UAS a e also ze o-
emission ehicles and his is an aspec pa icula ly impo an when su eying na u e
95
SUPPLEMENTARY MATERIAL
Table S1. En i onmen al co a ia es, desc ip ions, mean alues (X) and s anda d
de ia ions (SD) o UAS ack g ids and GPS loca ions bu e s e sus MA3 and o al
s udy a ea g ids used in he analysis o ca le spa ial abundance pa e ns in Doñana
Na u e Rese e (DNR).
Code
Va iable
UAS g id
(X±SD)
GPS loca ion
bu e used
and a ailable
(X±SD)
To al MA3
(X±SD)
To al s udy
a ea (X±SD)
DW
Dis ance o
nea es wa e
poin (km)
0.33±0.19
0.55±0.38
0.94±0.79
0.97±0.73
DE
Dis ance o
nea es ma sh-
sh ub eco one
(km)
1.22±0.90
1.50±1.49
4.06±2.76
2.47±2.17
GA
Exac ed UAS g id
a ea (ha)
1.25±0.85
0.21±0
1±0
1±0
LT1
Dense sc ub (%)
24.66±35.20
20.95±36.1
18.50±31.54
11.09±26.28
LT2
Low-clea sh ub
(%)
27.54±34.37
31.14±42.15
48.59±41.54
32.04±39.89
LT3
He baceous
g assland (%)
14.25±26.79
28.67±40.94
10.92±26.23
12.34±26.82
LT4
Woodland (%)
18.40±34.24
4.16±18.27
11.14±26.28
19.84±34.77
LT5
Ba e land (%)
8.82±24.41
2.55±12.93
4.58±15.71
11.30±26.88
LT6
Wa e cou se
ege a ion (%)
6.33±19.95
12.28±30.03
4.50±17.61
11.28±28.24
MA
Ca le
managemen a ea
(ca ego ical 1-5)
-
-
-
-

96
Gene al discussion
The ecen ema kable de elopmen o UAS has led o a dec ease in p ices and a
la ge a ie y o equipmen in he ma ke , which has a o ed he inco po a ion o hese
sys ems o en i onmen al esea ch. The no el y o he echnology explains ha he e is
almos no scien i ic li e a u e up o only en yea s ago, bu he explosion o published
pape s and news in he las h ee yea s con i ms ha he use o UAS is being explo ed
by nume ous esea ch eams wo ldwide.
This Ph.D. a emp s o ill he gap o knowledge in he use o UAS in
conse a ion biology. I desc ibes o he i s ime he use o hese sys ems in an
immedia ely applicable way in impac assessmen o in as uc u es o wildli e and
p o ec ion o endange ed species. Fu he mo e, i p esen s UAS as a ool o ob aining
high- esolu ion spa io empo al in o ma ion which helps o unde s and habi a use in
apidly changing human domina ed a eas and demons a es ha hese sys ems can
p o ide in o ma ion as alid as he ob ained by con en ional echniques on he spa ial
dis ibu ion o species in p o ec ed a eas.
Wha do UAS b ing in conse a ion biology?
Ae ial pe spec i e o e ed by UAS o e s ce ain ad an ages o e da a collec ion
obse e s om he g ound (as a as a ge a e de ec able by he sys em) as i allows
co e ing la ge a eas, which sa es ime and e o on asks ha can be edious by o he
means. E en low cos UAS equipped wi h a basic payload ha e p o ed use ul o
ou ine asks such as powe lines cha ac e iza ion (Chap e 1) whe e he sa ings in
logis ics allows di e ing esou ces owa ds he ins alla ion o mi iga ion measu es,
which is ul ima ely he goal o he impac assessmen s udy. In his line, UAS can also
con ibu e o he impac assessmen o o he in as uc u es ha p esen p oblems in
e ms o conse a ion, such as oads, wa e channels o wind a ms (C ock o d 1992;
Fo man and Alexande 1998; Kings o d 2000). Ae ial pe spec i e allows s udying he
su oundings o hese in as uc u es: dea h eco ds, dis ibu ion o wildli e, ege a ion
g adien s o he appea ance o in asi e species ha use hem o expand, p o iding
use ul in o ma ion o conse a ion and managemen .
The mos in e es ing UAS capabili y o en i onmen al applica ions is he high
spa ial and empo al esolu ion in o ma ion hey can p o ide by means o he emba ked
97
senso s (Jones e al. 2006; Ande son and Gas on 2013). Spa ial esolu ion depends on
he quali y o onboa d senso s and ligh heigh , bu e en low cos UAS equipped wi h
comme cial came as al eady o e su icien quali y o iden i y designs o elec ic poles
(Chap e 1), loca ing animals (chap e 2, (Wa s e al. 2010; Is ael 2011; Sa dà-
Palome a e al. 2012; Ve meulen e al. 2013)) and habi a cha ac e iza ion (Chap e 3
and 4, (Koh and Wich 2012; Ge zin e al. 2012; Chabo e al. 2014) ) wi h a esolu ion
o cen ime e s, much highe and a a lowe cos han he achie able wi h o he a ailable
echnologies (sa elli e, manned ai c a ). In chap e 4 we demons a e ha UAS a e able
o p o ide dis ibu ion models ha show equi alen pa e ns han wi h biologging.
The e o e, i he species is de ec able, UAS may cons i u e an al e na i e o mo e
in asi e (biologging) o expensi e (manned ligh ) echniques. Besides, wo king a a
high esolu ion scale pe mi s he esea che o ga he in o ma ion om all he animals
de ec able in he su eyed a eas, allowing o include in e -indi idual o in e speci ic
ela ions in he s udy and o compa e he animals dis ibu ions wi h he en i onmen al
a iables, which a e ecu en ecological ques ions ha may also a ec human ac i i ies
and conse a ion e o s (Mo ison e al. 2006).
Tempo al accu acy is impo an , as ob aining animal in o ma ion and
en i onmen al a iables a he same le el o de ail and eliabili y would signi ican ly
imp o e ecology s udies (Gailla d e al. 2010). Mic oelec onics e olu ion has allowed
ga he ing high accu acy and equency animals posi ion da a animals, bu sample he
en i onmen in which hey mo e wi h he desi ed equency o he esea che is no
possible wi h con en ional echnology. This is especially impo an when he
en i onmen changes apidly, especially i he changes ha occu a ec sho - e m
mo emen s o he species s udied, as demons a ed in chap e 3 and may be ex ended o
o he si ua ions (i.e. human in e en ions, i es, loods). Mo eo e , hanks o hei easy
deploy and o au opilo s capabili y UAS ligh s a e easily epe eable which allows o
e isi si es and o pe o m sys ema ic s udies (Wa s e al. 2008; Ande son and Gas on
2013). We o esee ha he ole o UAS in his aspec will be e olu iona y, as i can
help us o unde s and habi a selec ion a a ine scale and con ibu e o mo emen
ecology.
Mos small elec ic UAS like he ones used in he majo i y o en i onmen al
esea ch p oduce educed noise and a e isually disc e e. We did no see any nega i e
eac ions on he auna du ing any o he pe o med ligh s (no a e eco ded in he
scien i ic li e a u e), which in combina ion wi h being ze o-emission make UAS a low
98
impac me hod. This cons i u es a majo ad an age o hei use in auna su eys, as a
basic p inciple o any esea che is no o in e e e in he objec o s udy, bu i is also
undamen al in econnaissance ope a ions whe e he desi able is no o be de ec ed by
he “in ude ”. This makes UAS a e y con enien ool o endange ed species o
p o ec ed a eas moni o ing (i.e. an i-poaching, see chap e 2, (Lisein e al. 2013;
Ve meulen e al. 2013)).
Cu en UAS limi a ions in conse a ion biology
Al hough UAS ha e p o ed o be use ul o pe o m ae ial su eys (Wa s e al.
2010; Chabo and Bi d 2012; Sa dà-Palome a e al. 2012; Ve meulen e al. 2013), o o
do su eillance asks (chap e 2) he sys ems ha a e a o dable o a common esea ch
g oup (<100,000 !) ha e a s anda d ange o 10-30 km, limi ing he cu en scope o
UAS in ield biology o missions wi hin ha ange. The e o e i is no ealis ic wi hin
he cu en ma ke scena io o conside subs i u ing a manned ai c a o a UAS in
ypical wildli e su ey campaigns (i.e. pe iodic ae ial censuses in la ge p o ec ed a eas)
o o moni o la ge a eas, bu o complemen hem.
The dec ease in UAS p ices and payload senso s is one o he a o able
ci cums ances o hei in eg a ion in ecological esea ch. Pa icula ly in he las wo
yea s he e is an eme gence o inexpensi e (< 3,000 !) UAS and expe ienced model
ai c a hobbyis s may be able o buil hei own sys ems o e en less han 1,000 !.
Ve y low cos sys ems may be capable o pe o m basic missions, bu om ou
expe ience, eliabili y o cheap componen s is low and hei ailu e may cause se ious
p oblems p oducing dange ous si ua ions o diminishing da a quali y. Usabili y and
quali y ha e a p ice (Wa s e al. 2010) and esea che s ha wan o pe o m scien i ic
UAS campaigns need o in es a conside able budge o acqui e unc ional and eliable
sys ems, which may be a limi ing ac o .
UAS p esen some ope a ing cons ain s (Jones e al. 2006; Ande son and
Gas on 2013): i) UAS usabili y is cons ained by a o able wea he condi ions (less
han 15– 20 km/h wind speed o he pla o m sa e y and p e e ably clea days o good
quali y images), ii) The legal s a us o UAS is complica ed, a con o e sial opic in
which di e en agencies (mainly FAA in US and EUROCONTROL in Eu ope) a e
wo king o a long ime. Cu en ly, he si ua ion a ies be ween coun ies om a clea
p ohibi ion, he possibili y o au ho ized ligh s wi h al i ude and ange es ic ions up o
a o al absence o egula ion, a unclea si ua ion ha poses signi ican limi a ions o
99
en y o scien i ic use s, iii) he e is a need o specialized pe sonnel o ope a e he
sys ems, some knowledge o main enance (Jones e al. 2006) and some expe ise o
da a p ocessing.
As s a ed by Ellwood e al. 2007 who analyzed echnology in eg a ion in
conse a ion biology “ ailu e o unde s and echnology limi a ions can ha e se ious
consequences”. Mos o ecology s udies using UAS a e based on image analysis,
he e o e, expe imen al design mus be ca e ully ma ched o a oid “d owning” in o
housands o images and so ha esea che s unde s and he signi icance o missing da a.
Animal de ec abili y depends on hei size, shape, colo , shadow, con as agains
backg ound, hei esponse o he ai c a (lis ed by Fleming & T acey 2008 o manned
ae ial su eys), bu also on he UAS came a cha ac e is ics, ligh al i ude and s abili y
o he ai c a and he en i onmen al condi ions. Habi a cha ac e is ics such as
ege a ion co e (Bayliss and Yeomans 1989) also a ec animal’s de ec abili y, bu
besides, o selec o a oid o co e may a y by species, season and ime o day. The
inclusion o de ec abili y coe icien s (chap e 4) o using au oma ic pa e n ecogni ion
echniques in o da a-p ocessing p ocedu es may imp o e he p ocess (Abd-el ahman e
al. 2005) bu s ill, a wo d o cau ion should be aised ega ding he cos /bene i a io o
some o hese sys ems and hei applicabili y.
Fu u e p ospec s o UAS in conse a ion biology
The p esen Ph.D. ied o con ibu e o analyze UAS in eg a ion in conse a ion
biology, howe e all he expe imen s we e pe o med wi h small UAS ha , as
men ioned abo e, ha e ange and au onomy limi a ions ha cons ain he scope o he
esea ch. To gain access o la ge UAS, which is cu en ly limi ed o a ew in e na ional
agencies, would allow o add ess issues o g ea impo ance in conse a ion in a global
con ex , such as he s udy o clima e change, de o es a ion and habi a agmen a ion,
which a e majo causes o biodi e si y loss. Al hough his possibili y seems
complica ed in he sho e m, i may be easible h ough ag eemen s be ween esea ch
and mili a y agencies, a leas o sha ing some da a o o emba k scien i ic equipmen
in he aining missions o la ge mili a y UAS.
One o he UAS capabili ies ha has ha dly been exploi ed in small UAS is he
inclusion o o he senso s di e en om came as as payload. Me eo ological senso s,
wa e p obes and sampling de ices can p o ide ano he insigh on e ical habi a
100
cha ac e is ics and con ibu e o unde s and animal mo emen pa e ns, such as ligh
dynamics and mig a ion.
As poin ed ou in chap e 3 and 4, UAS can be combined wi h o he a ailable
echnologies such as biologging, and as sugges ed in chap e 2 i is possible o achie e
be e esul s combining UAS wi h s a ic su eillance came as and mo emen de ec o s.
A u he s ep would be o communica e o he senso s (s a ic o mobile) ha could wo k
oge he wi h UAS o ming an he e ogeneous coope a ing objec s ne wo k o sensi i e
a eas su eillance.

101
Conclusions
This Ph.D. desc ibes o he i s ime he use o UAS o impac assessmen o
in as uc u es, p o ec ion o endange ed species, habi a selec ion and o de e mine
animal spa ial dis ibu ion pa e ns, demons a ing ha hese sys ems can p o ide
in o ma ion in conse a ion biology.
The main bene i s ha UAS b ing o conse a ion biology a e:
1) Ae ial pe spec i e ha o e s ad an ages o e da a collec ion
obse e s om he g ound a lowe cos s han o he a ailable echnologies.
2) High spa ial esolu ion images ha allow o assess en i onmen al
impac o in as uc u es, wildli e and habi a cha ac e iza ion.
3) High empo al esolu ion in o ma ion ha pe mi s o moni o he
en i onmen a he esea che ’s desi ed equency ha allows e isi ing si es o
pe o m sys ema ic s udies.
The main limi a ions o UAS in eg a ion in conse a ion biology a e:
1) Scope o he missions is limi ed by ange and au onomy o he
a o dable UAS o sho ange (<30 km adio)
2) Al hough UAS p ices a e dec easing, o sa ely pe o m
p o essional missions budge is s ill limi ing ac o .
3) Ope a ing cons ains: wea he condi ions, legal limi a ions and
specialized pe sonnel o ope a ing he sys ems.
4) Expe imen al design mus be ca e ully add essed o a oid
“d owning” in o housands o images and o conside he signi icance o missing
da a (mainly ela ed wi h de ec abili y p oblems).
102
Li e a u e ci ed
ABC (2014) UAS o Moni o Ma shlands in New Sou h Wales. UAS news 1–4.
Abd-el ahman A, Pea ls ine L, Pe ci al F (2005) De elopmen o Pa e n Recogni ion
Algo i hm o Au oma ic Bi d De ec ion om Unmanned Ae ial Vehicle Image y.
Su L In Sci 65:37–45.
Ace edo P, Qui ós-Fe nández F, Casal J, Vicen e J (2014) Spa ial dis ibu ion o wild
boa popula ion abundance: Basic in o ma ion o spa ial epidemiology and
wildli e managemen . Ecol Indic 36:594–600. doi: 10.1016/j.ecolind.2013.09.019
Ae yonLabs (2014) Moni o ing sensi i e salmon popula ions on Snake Ri e is sa e
wi h Ae yon Scou . In: Appl. Case S ud. h p://www.ae yon.com/applica ions/case-
s udies/588-moni o ing-sensi i e-salmon-popula ions-on-snake- i e -is-sa e -wi h-
ae yon-scou .h ml. Accessed 27 No 2014
Akaike H (1974) A new look a he s a is ical model iden i ica ion. IEEE T ans Au om.
Con AC. pp 19:716–723
Alcaide M, Neg o JJ, Se ano D, e al. (2010) Cap i e b eeding and ein oduc ion o
he lesse kes el Falco naumanni: a gene ic analysis using mic osa elli es. Conse
Gene 11:331–338. doi: 10.1007/s10592-009-9810-7
Ande son K, Gas on KJ (2013) Ligh weigh unmanned ae ial ehicles will e olu ionize
spa ial ecology. F on Ecol En i on 11:138–146. doi: 10.1890/120150
A jomandi M (2007) Classi ica ion o Unmanned Ae ial Vehicles. Mechanical
Enginee ing. 3016. Ae onau Eng 49.
A e e N (2014) D ones Take O as Wildli e Conse a ion Tool. Audubon Mag 1–8.
Bailey DW, G oss JE, Laca EA, e al. (1996) Mechanisms ha esul in la ge he bi o e
g azing dis ibu ion pa e ns. J Range Manag 49:386–400.
Bayliss P, Yeomans K (1989) Co ec ing bias in ae ial su ey popula ion es ima es o
e al li es ock in no he n Aus alia using he double-coun echnique. J Appl Ecol
26:925–933.
Be ni JAJ, Za co-Tejada P, Suá ez L, Fe e es E (2009) The mal and na owband
mul ispec al emo e sensing o ege a ion moni o ing om an unmanned ae ial
ehicle. IEEE T ans Geosci Remo e Sens 47:722–738.
Bi dLi e In e na ional (2011) Species ac shee : Falco naumanni.
Blackbu n T, Gas on K (1994) The dis ibu ion o body sizes o he wo lds bi d species.
Oikos 127–130.
103
Blu L a, Ru z C (2008) A quick guide o ideo- acking bi ds. Biol Le 4:319–22. doi:
10.1098/ sbl.2008.0075
Blyenbu gh P Van (1999) UAVs - Cu en si ua ion and conside a ions o he way
o wa d. De . Ope . UAVs Mil. Ci . Appl.
Bologna F, Maha ho N, Hoch D (2002) In a- ed and ul a- iole imaging echniques
applied o he inspec ion o ou doo ansmission ol age insula o s. IEEE A icon.
Elec o ech. Se . A ica. IEEE, 345 E 47 Th S . NY 10017 USA, P e o ia, Sou h
A ica, pp 593–598
B aza F, Al a ez F (1987) Habi a use by ed dee and allow dee in Doñana Na ional
Pa k. Miscellània Zoològica 11:363–367.
Bugalho MN, Milne J a (2003) The composi ion o he die o ed dee (Ce us elaphus)
in a Medi e anean en i onmen : a case o summe nu i ional cons ain ? Fo Ecol
Manage 181:23–29. doi: 10.1016/S0378-1127(03)00125-7
Campoy, P., Ga cía, P.J., Ba ien os, A., Ce o, J.D., Agui e, I., Roa, A., Ga cía, R.,
Muñoz JM (2001) An s e eoscopic ision sys em guiding an au onomous
helicop e o o e head powe cable inspec ion. In . Wo k. Rob. Vis. Auckland,
New Zealand, p pp 115–124
Cano M, G anados J, Cas illo A, e al. (2007) Nue as ecnologías aplicadas al
seguimien o de ungulados sil es es en Sie a Ne ada: Colla es GPS-GSM. In:
SGHG (ed) Biodi e s. y Conse . auna y lo a en Ambien . medi e áneos.
G anada, pp 691–705
Casasús I, Riedel JL, Blanco M, Be nués A (2012) Ex ensi e li es ock p oduc ion
sys ems and he en i onmen . In Animal a ming and en i onmen al in e ac ions in
he Medi e anean egion. 81–88.
Chabo D, Bi d DM (2012) E alua ion o an o - he-shel Unmanned Ai c a Sys em
o Su eying Flocks o Geese. Wa e bi ds 35:170–174.
Chabo D, Bi d DM (2013) Small unmanned ai c a : p ecise and con enien new ools
o su eying we lands. Jounal Unmanned Veh Sys 24:15–24.
Chabo D, Ca ignan V, Bi d DM (2014) Measu ing Habi a Quali y o Leas Bi e ns in
a C ea ed We land wi h Use o a Small Unmanned Ai c a . We lands 34:527–533.
doi: 10.1007/s13157-014-0518-1
Chapman A (2014) I ’ s okay o call hem d ones. J Unmanned Veh Sys 2:1–3.
CLAVE S.L. (1992) Análisis de impac os de líneas eléc icas sob e la a i auna de
espacios na u ales p o egidos. Se ille, Spain
Collinge S (2010) Spa ial Ecology and Conse a ion. Na Educ Knowl 3:69.
104
Conseje ía de Medio Ambien e y O denación del Te i o io. (2013) Ca og a ía y
e aluación de la ege ación y lo a de los ecosis emas o es ales de Andalucía a
escala de de alle (1:10.000). Spain
Cooke S, Hinch S, Wikelski M, e al. (2004) Bio eleme y: a mechanis ic app oach o
ecology. T ends Ecol E ol 19:334– 343.
Coulombe ML, Massé A, Cô é SD (2006) Teleme y Quan i ica ion and Accu acy o
Ac i i y Da a Measu ed wi h VHF and GPS Teleme y. Wildl Soc Bull 34:81–92.
Cox TH, Nagy CJ, Skoog MA, Some s IA (2004) Ci il UAV capabili y assessmen .
103.
C amp S, Simmons K (1980) Hawk o Bus a ds. Handb. Bi ds Eu . Middle Eas No h
A ica.Vol 2.
C i elli, A.J., Je en up, H., Mi che T (1988) Elec ic powe lines: a cause o
mo ali y in Pelecanus c ispus B uch, a wo ld endange ed bi d species. Wa e bi ds
11:301–305.
C ock o d NJ (1992) A e iew o he possible impac s o wind a ms on bi ds and o he
wildli e. Pe e bo ough (Uni ed Kingdom)
Dewa H (2013) Te apin 1 d one soa s, p o ec ing hinos om poache s. In: News.
h ps://cmns.umd.edu/news-e en s/ ea u es/1101. Accessed 24 No 2014
Van Die men J, an Manen W, Edwin B (2009) Te eingeb uik en ac i i ei spa oon
an Wespendie en Pe nis api o us op de Veluwe. Takkeling 109–133.
Donáza J, Neg o J, Hi aldo F (1993) Fo aging habi a selec ion, land-use changes and
popula ion decline in he lesse kes el Falco naumanni. J Appl Ecol 30:515–522.
Dublin H (2011) A ican Elephan Specialis G oup Repo . Pachyde m 49:1–5.
Due AE, Mille T a, Lanzone M, e al. (2012) Tes ing an eme ging pa adigm in
mig a ion ecology shows su p ising di e ences in e iciency be ween ligh modes.
PLoS One 7:e35548. doi: 10.1371/jou nal.pone.0035548
Dun o d R, Michel K, Gagnage M, e al. (2009) Po en ial and cons ain s o Unmanned
Ae ial Vehicle echnology o he cha ac e iza ion o Medi e anean ipa ian o es .
In J Remo e Sens 30:4915–4935. doi: 10.1080/01431160903023025
Eléc ica R (2005) Red Eléc ica y la a i auna: 15 años de in es igación aplicada. 38.
Ellio JM (1977) S a is ical analysis o samples o ben hic in e eb a es. Ambleside
Ellwood SA, Wilson RP, Addison AC (2007) Technology in conse a ion: a boon bu
wi h small p in . In: Macdonals DW, Se ice K (eds) Key Top. Conse . Biol.
Blackwell Publishing, Pads ow, UK, p 329
111
Poma ol M (1993) Lesse Kes el eco e y p ojec in Ca alonia. In: Nicholls M, Cla ke
R (eds) Biol. Conse . small alcons P oc. 1991 Hawk Owl T us Con . The Hawk
and Owl T us , London, pp 24–28
Po apo , U ekhina I.G., McG ady M.J. RD (2013) Usage o UAV o Su eying S elle
’ s Sea Eagle Nes s. Rap o s Conse 27:253–260.
Pype W (2008) Popula ion su ey pilo s an unmanned ai c a . ECOS 17.
Qama FM, Ali H, Ash a S, e al. (2011) Dis ibu ion and habi a sele ion mapping o
key auna species in selec ed a eas o Wes e n Himalaya, Pakis an. J Anim Plan
Sci 21:396–399.
QGIS De elopmen Team (2012) QGIS Geog aphic In o ma ion Sys em Open Sou ce
Geospa ial Founda ion P ojec .
R Founda ion o S a is ical Compu ing, Vienna A (2013) R: A Language and
En i onmen o S a is ical Compu ing.
Red e n J V., Viljoen PCP, K uge JMM, Ge z W. M (2002) Biases in es ima ing
popula ion size om an ae ial census: a case s udy in he K uge Na ional Pa k,
Sou h A ica. S A J Sci 98:455–461.
Ribei o E (2007) Seasonal a ia ion in o aging habi a p e e ence in Lesse Kes el
Falco naumanni. 30.
Rod íguez A, Neg o JJ, Bus aman e J, e al. (2009) Geoloca o s map he win e ing
g ounds o h ea ened Lesse Kes els in A ica. Di e s Dis ib 15:1010–1016. doi:
10.1111/j.1472-4642.2009.00600.x
Rod íguez A, Neg o JJ, Mule o M, e al. (2012a) The eye in he sky: combined use o
unmanned ae ial sys ems and GPS da a logge s o ecological esea ch and
conse a ion o small bi ds. PLoS One 7:e50336. doi:
10.1371/jou nal.pone.0050336
Rod íguez A, Rod íguez B (2009) A ac ion o pe els o a i icial ligh s in he Cana y
Islands: e ec s o he moon phase and age class. Ibis (Lond 1859) 151:299–310.
doi: 10.1111/j.1474-919X.2009.00925.x
Rod íguez A, Rod íguez B, Cu belo ÁJ, e al. (2012b) Fac o s a ec ing mo ali y o
shea wa e s s anded by ligh pollu ion. Anim Conse 15:519–526. doi:
10.1111/j.1469-1795.2012.00544.x
Rod íguez C, Tapia L, Kieny F, Bus aman e J (2010) Tempo al changes in lesse kes el
(Falco naumanni) die du ing he b eeding season in sou he n Spain. J Rap o Res
120–128.
Rod íguez C, Wiegand K (2009) E alua ing he ade-o be ween machine y e iciency
and loss o biodi e si y- iendly habi a s in a able landscapes: The ole o ield
size. Ag ic Ecosys En i on 129:361–366. doi: 10.1016/j.agee.2008.10.010

112
Room MHM, Ahmad a (2014) Mapping o a i e using close ange pho og amme y
echnique and unmanned ae ial ehicle sys em. IOP Con Se Ea h En i on Sci
18:012061. doi: 10.1088/1755-1315/18/1/012061
Rope -Coude Y, Wilson R (2005) T ends and pe spec i es in animal- a ached emo e
sensing. F on Ecol En 3:437–444.
Ru z C, Blu L (2008) Animal bo ne imaging akes wing, o he dawn o “wildli e
ideo- acking.” T ends Ecol E ol 292–294.
Ru z C, Blu L, Wei A, Kacelnik A (2007) Video came as on wild bi ds. Science (80-
) 765.
Ru z C, Hays GC (2009) New on ie s in biologging science. Biol Le 5:289–92. doi:
10.1098/ sbl.2009.0089
SACAA (2008) In e im policy o ci il Unmanned Ai c a Sys ems in Sou h A ica.
Second d a . Sou h A ica
Sakamo o KQ, Takahashi A, Iwa a T, T a han PN (2009) F om he eye o he
alba osses: a bi d-bo ne came a shows an associa ion be ween alba osses and a
kille whale in he Sou he n Ocean. PLoS One 4:e7322. doi:
10.1371/jou nal.pone.0007322
SAMAA (2001) Manual o p ocedu e o model ai c a and emo e-pilo ed ai ehicle
ope a ion. Sou h A ica
Sa dà-Palome a F, Bo a G, Viñolo C, e al. (2012) Fine-scale bi d moni o ing om ligh
unmanned ai c a sys ems. Ibis (Lond 1859) 154:177–183. doi: 10.1111/j.1474-
919X.2011.01177.x
SCB (2014) Socie y o Conse a ion Biology. h p://www.conbio.o g. Accessed 25
No 2014
Schi man R (2014) D ones Flying High as New Tool Fo Field Biologis s. Science (80-
) 344:2014.
Schwa zbach M, Laiacke M, Mule o-Pazmany M, Kondak K (2014) Remo e wa e
sampling using lying obo s. 2014 In Con Unmanned Ai c Sys 72–76. doi:
10.1109/ICUAS.2014.6842240
Shepa d E, Lambe ucci S, Vallmi jana D, Wilson R (2011) Ene gy Beyond Food:
Fo aging Theo y In o ms Time Spen in The mals by a La ge Soa ing Bi d. PLoS
One e27375.
Sil y NJ, Lopez RR, Pe e son MJ (2012) Wildli e Ma king Techniques. In: N.J.Sil y
(ed) Wildl. Tech. Manual., The Johns . Bal imo e, MD, USA., pp 230–257
113
So iano P, Caballe o F, Olle o A (2009) RF-based Pa icle Fil e localiza ion o
Wildli e T acking by using an UAV. 40 h In . Symp. Robo . Ba celona, España.,
pp 239–244
Soulé ME (2007) Wha is Conse a ion Biology? Bioscience 35:727–734.
S olpe R, Ha J, Maha ho N (2009) The design and e alua ion o a mul i-spec al
imaging came a o he inspec ion o ansmission lines and subs a ion equipmen .
Insul News Mag Rep 1–14.
Teal G oup (2014) Teal G oup P edic s Wo ldwide UAV Ma ke Will To al $91 Billion
in I s 2014 UAV Ma ke P o ile and Fo ecas . Pa is, F ance
Tella J, Fo e o M, Hi aldo F, Donáza J (1998) Con lic s be ween Lesse Kes el
conse a ion and Eu opean Ag icul u al Policies as iden i ied by habi a analyses.
Cons Biol 593–604.
Telle ía JL (1986) Manual pa a el censo de los e eb ados e es es. Fac Biol Uni …
1–32.
Tesla N (1898) Me hod o and appa a us o con olling mechanism o mo ing essels
o ehicles.
T acey JP, Fleming PJS, Mel ille GJ (2008) Accu acy o some ae ial su ey es ima o s:
con as s wi h known numbe s. Wildl Res 35:377. doi: 10.1071/WR07105
U.S. A my (2010) Eyes o he A my – Unmanned Ai c a Sys ems. Roadmap 2010–
2035. 140.
UK Minis y o De ence (2010) Unmanned Ai c a Sys ems: Te minology, De ini ions
and Classi ica ion. Join Doc ine No e 3/10. 21.
UNESCO (2014) Wo ld he i age lis . h p://whc.unesco.o g/en/ lis /685. Accessed 5 Feb
2014
U sua E, Se ano D, Tella J (2005) Does land i iga ion ac ually educe o aging habi a
o b eeding lesse kes els? The ole o c op ypes. Biol Conse 122:643–648.
doi: 10.1016/j.biocon.2004.10.002
US Depa men o De ense (2014) US DoD. In: UAV Unmanned Ae . Veh.
h p://www.de ense.go /specials/ua 2002/. Accessed 23 No 1BC
Vala anis KP (2008) Ad ances in Unmanned Ae ial Vehicles: S a e o he A and he
Road o Au onomy.Volume 33 o In elligen Sys ems, Con ol and Au oma ion:
Science and Enginee ing. 568.
Ve meulen C, Lejeune P, Lisein J, e al. (2013) Unmanned Ae ial Su ey o Elephan s.
PLoS One 8:e54700. doi: 10.1371/Ci a ion
114
Vii ala J, Ko plmäki E, Palokangas P, Koi ula M (1995) A ac ion o kes els o ole
scen ma ks isible in ul a iole ligh . Na u e 373:425–427. doi:
10.1038/373425a0
Wa s AC, Amb osia VG, Hinkley E a. (2012) Unmanned Ai c a Sys ems in Remo e
Sensing and Scien i ic Resea ch: Classi ica ion and Conside a ions o Use. Remo e
Sens 4:1671–1692. doi: 10.3390/ s4061671
Wa s AC, Bowman WS, Abd-El ahman AH, e al. (2008) Unmanned Ai c a Sys ems
(UASs) o ecological esea ch and na u al- esou ce moni o ing (Flo ida). Ecol
Res o 26:10–15.
Wa s AC, Pe y JH, Smi h SE, e al. (2010) Small Unmanned Ai c a Sys ems o
Low-Al i ude Ae ial Su eys. J Wildl Manage 74:1614–1619. doi: 10.2193/2009-
425
WcUAVC (2013) Wildli e Conse a ion UAV Challenge. h p://www.wcua c.com/.
Accessed 20 Dec 2013
Whi wo h C, Dulle A, Jones D, Ea p G (2001) Ae ial ideo inspec ion o o e head
powe lines. Powe Eng J 15:25–32.
Wild S (2013) D ones now pa o an i-poaching a senal. In: Bussines day.
h p://www.bdli e.co.za/na ional/science/2013/02/08/d ones-now-pa -o -an i-
poaching-a senal. Accessed 10 Ma 2013
Wilkinson BE (2007) The design o he geo e e encing echniques o unmanned
au onomous ae ial ehicle ideo o use wi h wildli e in en o y su eys: a case
s udy o he Na ional Bison Range, Mon ana. 95.
Williamson SP (2011) U iliza ion o Unmanned Ai c a Sys ems o En i onmen al
Moni o ing. 36.
WWF (2013) Google Helps WWF S op Wildli e C ime.
h p://wo ldwildli e.o g/s o ies/google-helps-ww -s op-wildli e-c ime. Accessed 3
May 2013
Za alaga CB, Dell’Omo G, Becciu P, Yoda K (2011) Pa e ns o GPS acks sugges
noc u nal o aging by incuba ing Pe u ian pelicans (Pelecanus hagus). PLoS One
6:e19966. doi: 10.1371/jou nal.pone.0019966
Zenko M (2013) Re o ming U . S . D one S ike Policies. Council Special Repo No.
65. Washing on, USA
115
116
Acknowledgemen s

The o e all objec i e o his Ph.D. is o e alua e he use o
UAS in conse a ion biology, iden i ying hei capaci ies
and limi a ions in he ollowing applica ions
 )PXDBO6"4DPOUSJCVUFUPFOWJSPONFOUBMJNQBDU
assessmen o in as uc u es?
 )PXDBO6"4DPOUSJCVUFUPNBOBHFNFOUPGFOEBOHFSFE
species?
 $POTFSWBUJPOJOBIVNBOEPNJOBUFEMBOETDBQF$BO6"4
cons i u e a use ul ool o ob aining high- esolu ion
spa io empo al in o ma ion on animals habi a use?
 $POTFSWBUJPOJO B QSPUFDUFEBSFB"SF6"4DBQBCMF
o p o iding in o ma ion as alid as he ob ained by
con en ional echniques on he spa ial dis ibu ion o
species in p o ec ed a eas?