2018 status muskoxen, Maniitsoq & Sisimiut, West Greenland. Technical Report No. 119

1
2018 S a us Muskoxen,
Manii soq & Sisimiu ,
Wes G eenland
Technical Repo No. 119, 2022
Pinngo i ale i ik – G eenland Ins i u e o Na u al Resou ces
2
Ti le: 2018 s a us muskoxen, Manii soq & Sisimiu , Wes G eenland
Au ho s: Ch is ine Cuyle 1, Tiago A. Ma ques2, Iú i J.F. Co eia3, Aslak Jensen4,
Pe e Hegelund1 and Jukka Wagnhol 5
1 Pinngo i ale i ik – G eenland Ins i u e o Na u al Resou ces, P.O.
Box 570, 3900 Nuuk, G eenland
2 CREEM Uni e si y o S And ews, School o Ma hema ics and
S a is ics, Sco land
3 Uni e si y o Lisbon, Facul y o Sciences, Po ugal
4 Sol iaq 15, 3900 Nuuk, G eenland
5 Tusass, P.O. Box 1002, 3900 Nuuk, G eenland
Se ies: Technical Repo No. 119, 2022
Da e o publica ion: 16 Feb ua y 2022
Publishe : Pinngo i ale i ik – G eenland Ins i u e o Na u al Resou ces
Financial suppo : Go e nmen o G eenland and Pinngo i ale i ik – G eenland
Ins i u e o Na u al Resou ces
Co e pho o: Ch is ine Cuyle : G oup o 23 muskoxen in he Sisimiu sub-a ea,
be ween own o Kange lussuaq and he Iso oq i e .
ISBN: 9788797297735
ISSN: 1397-3657
EAN: 9788797297728
Ci ed as: Cuyle , C., Ma ques, T.A., Co eia, I.J.F., Jensen, A., Hegelund, P. &
Wagnhol , J. 2022. 2018 s a us muskoxen, Manii soq & Sisimiu ,
Wes G eenland. Pinngo i ale i ik – G eenland Ins i u e o Na u al
Resou ces. Technical Repo No. 119. 113 pp.
Con ac add ess: The epo is only a ailable in elec onic o ma . PDF- ile copies can
be downloaded a his homepage:
h ps://na u .gl/ o skning/ appo e /
Pinngo i ale i ik – G eenland Ins i u e o Na u al Resou ces
P.O. Box 570
3900 Nuuk
G eenland
Phone: +299 36 12 00
E-mail: in o@na u .gl
www.na u .gl
3
2018 s a us muskoxen,
Manii soq & Sisimiu ,
Wes G eenland
By
Ch is ine Cuyle 1, Tiago A. Ma ques2, Iú i J.F. Co eia3,
Aslak Jensen4, Pe e Hegelund1 and Jukka Wagnhol 5
1 Pinngo i ale i ik – G eenland Ins i u e o Na u al Resou ces, P.O. Box 570, 3900 Nuuk, G eenland
2CREEM Uni e si y o S And ews, School o Ma hema ics and S a is ics, Sco land
3CEAUL, Uni e si y o Lisbon, Facul y o Sciences, Po ugal
4Sol iaq 15, 3900 Nuuk, G eenland
5 Tusass, P.O. Box 1002, 3900 Nuuk, G eenland
Technical Repo No. 119, 2022
Pinngo i ale i ik – G eenland Ins i u e o Na u al Resou ces
4
[Emp y page]
5
Table o Con en s
Summa y (English) .................................................................... 8
Eqikkaaneq (kalaallisu ) ....................................................... 90
Resumé (dansk) ........................................................................ 13
In oduc ion ........................................................................... 136
Me hods .................................................................................... 19
Resul s ..................................................................................... 266
Discussion .............................................................................. 439
Acknowledgemen s ............................................................... 566
Li e a u e ci ed ...................................................................... 566
Figu es
1.
No h egion su eyed by helicop e in 2018, …
Page 16
2.
Repo ed ha es , comme cial, ec ea ional, and combined, o he …
Page 17
3.
A ea co e ed by 2018 ca ibou su ey o he No h egion (23,303 km2) …
Page 22
4.
The 19 line ansec s used in he 2018 ca ibou su ey o he No h egion…
Page 23
5.
Flow diag am o he modelling p ocess du ing he analysis, om DS…
Page 25
6.
Muskox su ey 2018: explo a o y analysis plo s…
Page 30
7.
Muskox su ey 2018: explo a o y analysis: g oup size dis ibu ion…
Page 31
8.
Muskox su ey 2018: explo a o y analysis o non- unca ed…
Page 32
9.
Obse e e ec : his og ams: de ec ed dis ances o he wo obse e s
Page 33
10.
His og am o he di e en binning op ions o he muskox dis ance da a…
Page 33
11.
The de ec ed dis ances wi h he es ima ed de ec ion unc ion o e laid…
Page 35
12.
Es ima ed p obabili ies o de ec ion, gi en non- unca ed and unca ed …
Page 36
13.
Muskox encoun e a e (g oups pe km) pe line ansec , …
Page 36
14.
Rela i e dis ibu ion o muskox numbe s along he line ansec s…
Page 37
15.
Dis ance sampling design-based muskox densi y es ima es wi h CIs…
Page 39
16.
Fi ed DSM plo s co esponding o he i ed smoo h o ele a ion…
Page 40
17.
P edic ed muskox densi y ac oss silhoue e map o en i e No h egion.
Page 42
18.
Fi ed DSM a iabili y silhoue e map o en i e No h egion. Unce ain y …
Page 42
19.
Vi ually snow- ee e ain o he inne Unii ii Fjo d, Angujaa o iup …
Page 44
20.
Vi ually snow- ee lowland ( o eg ound <200 m) o Angujaa o iup …
Page 45
21.
La ge expanse o i ually snow- ee e ain in Angujaa o iup sub-a ea …
Page 45
22.
Index o Manii soq muskox abundance in he Angujaa o iup sub-a ea …
Page 53

6
23.
Plo sampling g id example o o al a ea A di ided in o smalle plo s…
Page 63
24.
Example o a pa ch o und a wi h he ansec in he middle…
Page 66
25.
Hal -no mal ( op ow) and haza d- a e (bo om ow) de ec ion unc ions…
Page 68
26.
Possible shapes o he de ec ion unc ion when cosine adjus men s a e….
Page 69
27.
A good model o he de ec ion unc ion should ha e a shoulde …
Page 72
28.
G oup o 30 muskoxen, which may ha e included 2-3 cal es…
Page 79
29.
G oup o 38 muskoxen, which may ha e included 3 cal es…
Page 79
30.
G oup o 23 muskoxen, which may ha e included 1-2 cal es…
Page 80
31.
G oup o 38 muskoxen, which may ha e included 7-8 cal es…
Page 80
32.
Map illus a ing ele a ion in he su eyed a ea. Legend colou codes…
Page 81
33.
Map illus a ing aspec in he su eyed a ea. Legend colou codes…
Page 81
34.
Map illus a ing slope in he su eyed a ea. Legend colou codes…
Page 82
35.
View om su ace o lake ice du ing a pause be ween su ey lines...
Page 83
36.
View as helicop e lew o e su ace o lake ice, illus a ing he ypical…
Page 83
37.
View while lying line ansec 17, illus a ing spa se snow-co e …
Page 84
38
O e iew o he no h side o Tase suaq Lake (la ge la su ace on le …
Page 84
39.
O e iews illus a ing la -ligh (no shadows) and he windswep e ain…
Page 85
40.
Minimal Ma ch snow co e ypical o lowland alleys in Angujaa o iup…
Page 86
41.
Obse a ion pla o m o ae ial su ey: AS350 helicop e …
Page 86
42.
Map o he 2010-2019 pe iod illus a ing he en i e Manii soq muskox …
Page 88
43.
Map illus a ing he en i e Manii soq muskox ha es managemen a ea …
Page 89
Tables
1.
Cu en naming o egion, municipali y, muskox popula ion, ha es …
Page 19
2.
Summa y o unp ocessed esul s: Su ey o he No h egion, …
Page 26
3.
Beha iou al eac ion o muskoxen o helicop e ly-by, as pe …
Page 27
4.
App oxima e ele a ions o muskox g oups obse ed …
Page 29
5.
Summa y o he coe icien cha ac e is ics o he GLM …
Page 34
6.
Model compa ison ac oss h ee Con en ional Dis ance Sampling models…
Page 38
7
Encoun e a e es ima es pe sub-a ea o muskox g oups …
Page 38
8.
Ma ch 2018, he Dis ance Sampling design-based muskox abundance …
Page 38
9.
Ma ch 2018, he Dis ance Sampling design-based muskox densi y …
Page 38
10.
GAM model summa y able ela i e o smoo h e ms o co a ia es…Tweedie
Page 39
11.
GAM model summa y able ela i e o smoo h e ms o co a ia es…NegBin
Page 40
7
12.
Compa ison o DS design-based and GAM/DSM analyses, o en i e No h …
Page 41
13.
Manii soq and Sisimiu muskoxen 2000-2014: win e minimum coun s, ...
Page 61
14.
Rough densi ies o Manii soq muskoxen in he Angujaa o iup sub-a ea.
Page 62
15.
Commonly used key unc ions & se ies expansions o he de ec ion unc ion…
Page 67
16.
Muskox hun ing seasons and quo as o Sisimiu and Manii soq …
Page 87
17.
Repo ed numbe o muskoxen ha es ed by comme cial and ec ea ional …
Page 90
Appendices
1.
Minimum coun s 2000-2014 o Manii soq and Sisimiu muskoxen
Page 51
2.
S a is ical me hods behind Dis ance Sampling
Page 63
3.
S a is ical me hods behind GAM and DSM modelling
Page 75
4.
Dis ance Sampling Assump ions – sho summa y
Page 78
5.
Sisimiu sub-a ea inland, wi hin 40 km o he Ice Cap, Ma ch 2018: …
Page 79
6.
Maps illus a ing ele a ion, aspec , and slope in he No h egion
Page 81
7.
Angujaa o iup sub-a ea, Ma ch 2018: Pho og aphs su ey condi ions …
Page 83
8.
Hun ing seasons, quo as & epo ed ha es s, p ima ily o Manii soq …
Page 87
9.
2020 Supplemen a y ma e ials o biological ad ice …
Page 91
10.
Recommenda ions o imp o ing u u e su eys o muskoxen
Page 112
Raw da a may be accessed by con ac ing he Pinngo i ale i ik – G eenland Ins i u e o
Na u al Resou ces, Depa men o Mammals and Bi ds, Nuuk, G eenland.
8
Summa y
This epo p esen s esul s om he i s sys ema ic ae ial su ey o muskoxen in
sou hwes G eenland. The su ey in ol ed he No h egion (66°-68°N), which was
di ided in o h ee sub-a eas (Angujaa o iup, Sisimiu , Sisimiu Sou h). The su ey
occu ed ea ly Ma ch 2018 and p o ides he i s es ima e o muskox abundance in
wo muskox ha es managemen a eas, Manii soq (66°-67°N) and Sisimiu (67°-68°N).
The beha io o mos muskox g oups was una ec ed by he helicop e ly-by a 40 m
al i ude. Rega dless o g oup size, 77% o g oups simply s ood s ill. De ec ing
s a iona y g oups is clea ly essen ial o accu a e popula ion es ima es.
Sisimiu muskoxen we e mos o en obse ed a ele a ions o ca. 200 m, which is
ypical o his species as hey p e e o o age in lowland ele a ions e en in win e . In
sha p con as , Manii soq muskoxen used 700-800 m ele a ions, despi e hei
documen ed yea - ound p e e ence o lowlands <400 m. Fu he , a he ime o he
su ey, Manii soq muskoxen we e clumped in o wo ‘ho spo s’ almos inaccessible by
mo o ehicle and ela i ely a om human habi a ion. Since 84% o all muskox
ha es (comme cial and ec ea ional), as well as mos ophy hun ing and qi iu
(muskox inne wool) p oduc ion in G eenland a e aken om he Manii soq muskox
popula ion, hese ac i i ies may ha e a ole in he dis up ion o no mal lowland
dis ibu ion o he Manii soq muskoxen in win e .
F om a subse o he Manii soq da a, he cal (age <1-yea ) pe cen age was asce ained
ca. 18% o Manii soq muskoxen. Conside ing he absence o la ge p eda o s, he
cu en alue is conside ed low. Fac o s in ol ed may include densi y-dependen
issues associa ed wi h he Manii soq muskoxen now o aging on high ele a ion
subop imal habi a in win e . Since his coincides wi h la e ges a ion o muskoxen, cal
p oduc ion could be nega i ely a ec ed. Whe he cu en cal pe cen age is su icien o
suppo popula ion size s abili y o g ow h is deba able.
Popula ion size & densi y es ima es
Con en ional Dis ance Sampling (DS) design-based me hods and analyses, as well as
Gene alized Addi i e Models/Densi y Su ace Modelling (GAM/DSM) based analyses
we e applied o he da ase o ob ain es ima es o muskox popula ion size and densi y.
Mos muskoxen occu ed in he Manii soq muskox ha es managemen a ea
(su eyed Angujaa o iup sub-a ea). Fa ewe muskoxen inhabi ed he Sisimiu
muskox ha es managemen a ea (su eyed Sisimiu sub-a ea), and ze o muskoxen
we e obse ed in he su eyed Sisimiu Sou h sub-a ea. Densi ies om he GAM/DSM
model-based analysis suppo ed ha in ea ly Ma ch, muskoxen s ongly p e e ed
9
Sou hwes acing slopes and speci ically o Manii soq muskoxen he highes densi ies
coincided wi h high ele a ions.
Fo he en i e No h egion, he DS design-based Ma ch 2018 muskox popula ion
abundance was es ima ed a 21,746 muskoxen (95% CI: 11,061–42,751; CV = 28.5%; SE:
6,194), wi h a densi y o 1.1 muskoxen/km2 (95% CI: 0.559–2.160; CV = 28.5%; SE =
0.313). Al e na ely o he en i e No h egion, he GAM/DSM model-based Ma ch
2018 muskox popula ion abundance was es ima ed 23,256 muskoxen (95% CI: 18,102–
29,877; CV = 11.36%), wi h mean densi y o 3.69 muskoxen/km2 (95% CI: 2.87-4.74).
DS es ima es as pe speci ic sub-a ea:
Manii soq muskoxen (Angujaa o iup sub-a ea)
:
The DS design-based es ima e
was ca. 18,906 muskoxen (95% CI: 8,726–40,960; CV = 0.315; SE = 5,948), wi h a
densi y o ca. 2.6 muskoxen/km2 (95% CI: 1.223–5.742; CV = 0.315; SE = 0.834).
Sisimiu muskoxen (Sisimiu sub-a ea)
:
The DS design-based es ima e was ca.
2,840 muskoxen (95% CI: 662–12,178; CV = 0.568; SE = 1,613), wi h a densi y o
ca. 0.22 muskoxen/km
2
(95% CI: 0.052–0.962; CV = 0.568; SE = 0.127).
The popula ion size es ima es om he wo app oaches we e simila since he 95%CIs
o e lap. Despi e he good su ey co e age (10.6%), he high a iabili y wi hin he
da ase was esponsible o subs an ial unce ain y in he DS es ima es and less so in
he GAM/DSM. Rega dless, calcula ing he CV o p obabili y o de ec ion o
muskoxen pe mi ed compa ison o he wo app oaches. The DS p obabili y o muskox
de ec ion had a CV o 5.98%, which was be e han he CV o 11.36% o he
GAM/DSM. Thus, o he 2018 su ey o muskoxen, we ecommend using he DS
design-based abundance and densi y es ima es when making managemen decisions.
Alone, he 2018 es ima e canno indica e popula ion end. Tha equi es a leas wo
addi ional ae ial su ey es ima e poin s using simila me hods. Meanwhile, pas
coun s, densi ies, ha es s, and cal pe cen ages do no sugges ecen popula ion
g ow h bu possibly a decline p io o he 2018 su ey. Rega dless, speci ically he 2018
popula ion size es ima e o Manii soq muskoxen is la ge han p e ious es ima es o
popula ions anywhe e in G eenland. Also, Manii soq muskox densi y is much highe
han elsewhe e in he A c ic. This could inc ease exposu e o indi iduals o in ec ious
pa hogens. Gi en possible densi y-dependen in luences ac ing a cu en popula ion
size, popula ion g ow h may no be ad isable. Whe he he 2018 popula ion size and
densi y o he Manii soq muskox ha es managemen a ea a e wi hin he cu en
he bi o e ca ying-capaci y o he pas u e/ ange emains o be seen.
16
In oduc ion
Muskoxen (O ibos moscha us Zimme mann) a e na i e only o he no h and no heas
pa o G eenland. Ne e heless, he e a e cu en ly se e al muskox popula ions in
wes and no hwes G eenland. These esul ed om ansloca ions, excep ing one
(Ingle ield Land), which is a combina ion o na i e and ansloca ed muskoxen.
T ansloca ions began in he 1960s when 27 muskoxen li e-cap u ed in Eas G eenland
we e anspo ed and in oduced o Kange lussuaq (Sønd e S øm jo d) in Wes
G eenland (Fig. 1). These animals became i mly es ablished and un il 2015
Pinngo i ale i ik – G eenland Ins i u e o Na u al Resou ces (GINR) publica ions
e e ed o hese as he Kange lussuaq muskox popula ion. Today, he Go e nmen o
G eenland designa es hese as he Manii soq popula ion. So named because hey
inhabi wha was once he Manii soq municipali y. F om 1986 o 1991, he new
Manii soq popula ion became he sou ce o se e al u he muskox ansloca ions
along he wes and no hwes coas . Fu he , by he ea ly 2000’s some Manii soq
animals expanded no hwa d in o wha was hen he Sisimiu municipali y. Al hough
no uly ano he popula ion, ha es was managed sepa a ely in acco dance wi h he
hen wo sepa a e municipal ju isdic ions, and i became common o ega d hem as
wo popula ions, Manii soq and Sisimiu . In 2009, hose wo municipali ies me ged in o
one, Qeqqa a Kommunia, howe e , ha es con inues o be managed sepa a ely.
Figu e 1. No h egion su eyed by helicop e in 2018, illus a ing Manii soq and Sisimiu muskox managemen
a eas and he 1960’s elease loca ion o he 27 ansloca ed muskoxen o igina ing om Eas G eenland.

17
Go e nmen egula ed ha es ing o he Manii soq muskox popula ion began in 1988
wi h comme cial ha es , and since 1993 has suppo ed bo h ec ea ional and
comme cial ha es ing. Un il 1998, annual ha es s we e ypically unde 500 muskoxen
(Fig. 2). S a ing in 2002, annual ha es s inc eased sha ply o mo e han 2,500
muskoxen by 2008 and gene ally declined he ea e wi h he 2017-2018 ha es s almos
hal he peak alue. Dec easing ha es s con as wi h s eadily inc easing hun e e o .
Fo example, p io o he win e ha es season 2000 only hun e s om Sisimiu and
Manii soq pa icipa ed. The Sisimiu hun e s used p ima ily dogsleds, while he
Manii soq hun e s used ca. en snowmobiles and in o al he e we e ca. < 25 hun e s.
Since 2000, he numbe o mo o ized ehicles used o anspo a ion o and om he
hun ing-a eas has inc eased s eadily (Hans S. Mølgaa d & Nuka M. Lund pe s. comm.).
Recen ly, in he 2020 win e -hun he e we e 60 hun e s wi h mo o ized ehicles and in
2021 he e we e 70 hun e s wi h mo o ized ehicles (Nuka M. Lund pe s. comm.). The
dec ease in muskoxen ha es ed despi e inc eased hun e e o sugges s ewe
muskoxen a ailable. Causes would include decline in muskox popula ion size o ha
muskoxen a e inc easingly adep a a oiding hun e s.
Figu e 2. Repo ed ha es , comme cial, ec ea ional, and combined, o he Manii soq muskox popula ion om
1962 o 2018. The 1988-1992 pe iod was solely comme cial ha es . T ophy ha es no included. Da a om
Pinia neq eco ds.
Fu he o comme cial and ec ea ional hun ing (Fig. 2), in he ea ly 2000s ophy
hun ing began o Manii soq muskoxen. By he 2011/2012 season 108 muskoxen we e
killed as ophies. In 2012/2013 he e we e 120 ophy muskoxen killed (Cuyle &
Raund up 2014). T ophy hun ing is now well es ablished wi h speci ic hun ing seasons
and a ea concessions (i.e., a ea alloca ed o use by one ophy agen ). As an indus y,
ophy hun ing con inues o g ow (Naalakke suisu 2018). Fo eign ophy hun e s pay
up o Danish k one 50.000,00 (ca. 7,900.00 US Dolla s / 6,700 EUR) pe ophy. In 2016,
0
500
1000
1500
2000
2500
3000
1950 1960 1970 1980 1990 2000 2010 2020 2030
Repo ed muskox ha es
YEAR
To al ha es Comme cial Rec ea ional
18
236 muskoxen we e killed as ophies and p o ided he G eenland go e nmen wi h
e enues o ally Danish k one 312.000,00 (ca. 49,200.00 US Dolla s / 42,000 EUR). In
2017, ophy hun ing ose by 46% wi h 354 ophy muskoxen killed, which p o ided
Danish k one 548.000,00 (ca. 86,500.00 US Dolla s / 73,700 EUR). Addi ionally, he
G eenland qi iu (muskox wool) indus y is ounded p ima ily on he win e ha es o
Manii soq muskoxen, and owing o inc easing demand wo ldwide, o e he pas
decade he qi iu indus y expanded.
All he abo e a es s o a subs an ial economic con ibu ion o local communi ies om
hei use o oday’s Manii soq muskox popula ion, and o a lesse ex en also he
Sisimiu muskoxen. Gi en declining annual ha es s since 2008, asce aining
popula ion abundance and end a e essen ial o app op ia e managemen decisions
o sus ainable ha es ing in he u u e.
In he pas , in equen win e g ound su eys by snowmobile, o e limi ed a eas,
p o ided minimum coun s o he numbe o muskoxen obse ed (Appendix 1). Wi hin
he Manii soq managemen a ea (Fig. 1, Table 1), minimum coun s in he pe iod 2000-
2006 anged om 4186 o 5092 obse ed muskoxen. Applying Bayesian analysis o he
2000-2004 minimum coun s esul ed in a 2004 popula ion es ima e o ca. 7,312 (90%CI:
5538-10202) muskoxen in he Manii soq managemen a ea ha was co e ed by he
win e g ound su eys (Cuyle & Wi ing 2004). G ound su eys o he Sisimiu
managemen a ea we e a ely comple ed. When hese occu ed, ew muskoxen we e
obse ed, and g ound e o was iny ela i e o he managemen a ea.
This epo ocuses on he muskox popula ion size es ima es o Manii soq and Sisimiu
a ained in conjunc ion wi h he 2018 ae ial helicop e su ey o he Kange lussuaq-
Sisimiu (KS) ca ibou popula ion in he No h egion. The name, No h egion,
indica es a ela i ely no he n geog aphical posi ion wi hin he con ex o Wes
G eenland and delinea es KS ca ibou dis ibu ion. The Go e nmen o G eenland’s
muskox ha es managemen a eas, Manii soq and Sisimiu , a e con ained wi hin he
No h egion’s bounda ies (Fig. 1).
P esen su ey
This is he i s Con en ional Dis ance Sampling (DS) ae ial su ey o muskoxen in he
Manii soq and Sisimiu managemen a eas. This epo in es iga es he DS da a se o
muskox obse a ions ob ained o hose a eas du ing GINR’s Ma ch 2018 ca ibou
su ey, which is desc ibed in Cuyle e al. (2021). Ini ially, we use DS analyses o
p esen he i s e e p e-cal ing popula ion es ima es o muskox densi y and
abundance o he Manii soq and Sisimiu muskox ha es managemen a eas. Then,
19
we c ea e a Densi y Su ace Model (DSM) o he muskoxen, whe e densi y can be
spa ially ep esen ed as a unc ion o addi ional co a ia es collec ed du ing su eying.
The DSM p oduces al e na i e es ima es o densi y and abundance o Manii soq and
Sisimiu muskox ha es managemen a eas.
The epo p o ides he i s e e popula ion size es ima es o muskoxen in he No h
egion and hen p esen s sepa a e es ima es o he Manii soq and Sisimiu muskox
managemen a eas. I also p esen s in o ma ion on immedia e muskox eac ion
(mo emen o lack he eo ) o he helicop e ly-by o muskox g oups de ec ed, and an
app oxima e cal pe cen age o he Manii soq muskox managemen a ea.
No e ha an ea lie analysis o 2018 muskox abundance and densi y in he Manii soq
and Sisimiu managemen a eas was un on he same da a (Ma ques 2018), howe e ,
he hen known a eas (km2) we e inco ec . Ma ques’ pape om 2018 is an in e nal
CREEM (Cen e o Resea ch in o Ecological and En i onmen al Modelling (S .
And ews, Sco land)) epo , which is a ailable upon eques .
Table 1. Cu en naming o egion, municipali y, muskox popula ion, ha es managemen - and su eyed a eas o
2018 in Wes G eenland ha a e speci ic o his epo .
Region
Municipali y
G eenland Go e nmen
Muskox ha es managemen a ea
Muskox
Popula ion
Su eyed sub-a ea
2018
No h
Qeqqa a kommunia
Sisimiu managemen a ea
Sisimiu
Sisimiu
No h
Qeqqa a kommunia
Manii soq managemen a ea
Manii soq1
Angujaa o iup
1 P e iously e e ed o as he Kange lussuaq muskox popula ion, in GINR documen s.
Me hods
S udy a ea
The No h egion is wi hin Qeqqa a Kommunia in Wes G eenland. Al hough Qeqqa a
Kommunia has ca. 9,400 inhabi an s (in 2020), no all li e wi hin he bounda ies o he
No h egion. The only la ge se lemen wi hin he egion is he coas al ci y o Sisimiu ,
wi h ca. 5,600 inhabi an s, ollowed by he ca. 500 esiding in he own o
Kange lussuaq. The la e is loca ed on he eas e n inland side o egion nea he
G eenland Ice Cap and is also he si e o G eenland’s p ima y in e na ional ai po .
Toge he , he iny coas al illages o I illeq and Sa anngui con ain a u he 200-300
people.
The No h egion is seasonally ice- ee and co e s an a ea o 23,303 km2, (excluding
lakes, i e s, sand, glacie s, and islands). P e ious su eys epo ed a less p ecise land
a ea o ca. 26,000 km2 (Cuyle e al. 2002, 2005, 2011). Loca ed be ween 66-68° N La , he
20
A c ic Ci cle (66.5° N La ) passes h ough he No h egion. The no he n bo de is
p o ided by he Nassu ooq Fjo d (No d e S øm jo d). The sou he n bo de is o med
by wo ice caps (i.e., Kangaamiu Se mia (Sukke oppen Ice Cap) and Tase siap
Se mia) and he wes e n po ion o he Kange lussuaq Fjo d. Ele a ions each ca. 1700
on he Kangaamiu Se mia and ca. 1800 m on he Tase siap Se mia. The ou e hal o
he Kange lussuaq Fjo d is ice- ee yea - ound and domina ed by cli s o ca. 1000 m.
The wes e n bo de o he egion is he pe manen ly ice- ee seacoas o he Da is
S ai , and eas e n bo de is he G eenland Ice Cap.
The No h egion’s coas al opog aphy is moun ainous wi h peaks whose ele a ion can
be 1000-1800 m and glacie s a e common. The Kange lussuaq Fjo d pene a es he
egion, om SW o NE, ending jus be o e he own o Kange lussuaq, i.e., close o he
G eenland Ice Cap. The Kange lussuaq Fjo d is he p ima y bounda y sepa a ing he
Manii soq and Sisimiu muskox managemen a eas. In he Sisimiu a ea, mo ing
eas wa d, he coas al moun ains g adually gi e way o ugged e ain gene ally
anging 10-900 m ele a ion. In he Manii soq a ea, immedia ely no h o he wo ice
caps, Kangaamiu Se mia and Tase siap Se mia, he e ain is gene ally ba en
highlands wi h ele a ions > 1000 m. Con inuing no hwa d owa ds he own o
Kange lussuaq, ele a ions dec ease, and he e ain includes lowland alleys < 400 m
ele a ion and highlands o gene ally < 1000 m ele a ion.
Common o Wes G eenland, he No h egion exhibi s a clima e g adien on a wes -
eas axis. The wes e n seacoas is we ma i ime; howe e , he clima e becomes d y
con inen al as one mo es eas owa ds he G eenland Ice Cap. Clima e and wea he in
he wes a e unde he ma i ime in luences o he ice- ee Da is S ai and he low-
p essu e oceanic s o m sys ems ha sweep in om he sou hwes . The clima e in he
inland o he No h egion, speci ically he Manii soq a ea, is in luenced by wo ice caps
a i s sou he n bounda y, he Kangaamiu Se mia and Tase siap Se mia. These ha e
ele a ions o 1,700-1800 m espec i ely, and ac as a ba ie o he oceanic s o m
sys ems men ioned abo e, c ea ing a p ecipi a ion shadow on he no he n side, which
in combina ion wi h he domina ing high p essu e o e he G eenland Ice Cap c ea es
he inland’s xe ic con inen al clima e. Loess/sands o ms a e common nea he own o
Kange lussuaq (Cuyle e al. 2005). These a e caused by ka aba ic winds, öhn winds
descending o he G eenland Ice Cap, which a e d y and can ha e speeds o 30-60 m/s
(Pu nins 1970, Rasmussen 1989, Tams o 2004). Be ween Kange lussuaq, and he
G eenland Ice Cap, he e a e wo hea ily b aided i e s, he Akulia usia suup Kuua
and Qinngua a Kuussua. The associa ed alleys exempli y he abo e condi ions, and
hei Danish names, Sand lug dalen and Ø kendalen, ansla e loosely in o ‘Blown
Sand Valley’ and ‘Dese Valley’, espec i ely. Fu he , öhn winds can cause sha p
21
inc eases in ambien empe a u e, which in win e o sp ing can esul in ex ensi e
snowmel (Hansen 1999). A Kange lussuaq win e öhn winds p oduce la ge snow-
ee expanses (F edskild 1996).
In gene al, he ege a ion o he No h egion may be desc ibed as open o alpine
und a. Vege a ion is domina ed by low a c ic species o mainly dwa sh ub hea h,
which changes o p edominan ly s eppe and g assland when mo ing eas owa ds he
G eenland Ice Cap, whe e lichen hea hs a e a e (Tams o e al. 2005). Speci ically, he
Angujaa o iup sub-a ea is a g ass s eppe landscape (Nellemann 1997). Highe
ele a ions a e o en ell ield, ab asion pla eaus and ba e g ound (Tams o e al. 2005).
Aside om he now i mly es ablished ansloca ed popula ion o muskoxen, na i e
wild mammals p esen in he No h egion a e ca ibou (Rangi e a andus g oenlandicus),
a c ic ha e (Lepus a c icus Rhoads) and a c ic ox (Vulpes lagopus Linnaeus). La ge
mammalian p eda o s a e absen .
Field me hods
This s udy was possible owing o he 2018 ae ial su ey o ca ibou. The ae ial su ey
occu ed 01-15 Ma ch 2018 using a helicop e AS350 as he pla o m o obse a ion.
Pe iod and pla o m we e chosen as pe c i e ia o ca ibou (de ails in Cuyle e al.
2021). A cons an al i ude abo e g ound le el was main ained while lying low (40 m)
and slow (ca. 65 km/hou ).
Pa icipan s included h ee obse e s, all wi h p e ious su ey expe ience: GINR’s
senio scien is Ch is ine Cuyle , GINR’s p ojec coo dina o Pe e Hegelund, and
p o essional hun e Aslak Jensen (G eenland Associa ion o P o essional Hun e s
(KNAPK)) om Nuuk. Jensen and Hegelund we e sea ed in he ea o he helicop e
and obse ed animals o all dis ances om he side hey we e si ing, which al e na ed
each ime he helicop e was e ueled, which was usually once daily and some imes
wice. Cuyle always sa in on , obse ed he ack line, including dis ances o ei he
side up o 100 m, and was he da a eco de . Ve bal con ac among he obse e s
pe mi ed he digi al audio eco ding o all obse a ions. Two audio de ices (SONY IC
eco de , ICD-SX712) we e used o eco d sepa a ely he obse a ions speci ic o he le
and igh side o he line ansec . Audio eco ding de ices we e on con inual eco ding
o each line ansec . A he end o each su ey day, audio da a was downloaded o
compu e o s o age and back-up. Obse a ions we e la e pai ed wi h Global
Posi ioning Sys em (GPS) coo dina es o he helicop e a he ime o obse a ion. The
audio eco ding included dis ance o (see below), and size o , each muskox g oup

22
obse ed and name o he obse e . O en, beha io al eac ion/ ligh by muskox
g oups and en i onmen al condi ions we e eco ded.
Figu e 3. A ea co e ed by he 2018 ca ibou su ey o he No h egion (23,303 km2). Th ee di e en colou s
illus a e he h ee sub-a eas, designa ed as Sisimiu (blue), Sisimiu Sou h (o ange) and Angujaa o iup (pu ple).
The e m ’Bye bygde ’ e e s o ci y/ own/ illage iden i ied in Fig. 1.
Su ey design
The su eyed No h egion a ea, 23,303 km2, was di ided in o h ee sub-a eas (s a a),
a bi a ily named Sisimiu (12,658 km2), Sisimiu -Sou h (3,512 km2) and
Angujaa o iup (7,133 km2) (Fig. 3). The sampling design o he 2018 su ey
conside ed 19 sys ema ic pa allel line ansec s o a iable leng h sepa a ed by 15 km
and placed o e he h ee sub-a eas (Fig. 4). Those ansec s p o ide he maximum a ea
co e age possible gi en he inancial esou ces a ailable. An ini ial line ansec was
compu e gene a ed a andom, and he o he s ollowed 15 km apa . Aligning line
ansec s pe pendicula o known g adien s wi hin he su eyed a ea can maximize
p ecision o he esul ing es ima e by lowe ing he encoun e a e a iance (Buckland e
al. 2001). Thus, he ansec axis di ec ion was chosen as pe pendicula o p e iously
known animal dis ibu ion g adien s in Ma ch. Lines 1 o 13 ollowed a wes -eas axis,
which also e lec s he clima e g adien om we ma i ime o d y con inen al. Line
ansec s 14 o 19 ollowed a no h-sou h axis, e lec ing animal, clima e, and
opological g adien s ha exis be ween he combined ice caps (Kangaamiu Se mia
and Tase siap Se mia) and he own o Kange lussuaq.
23
Figu e 4. The 19 line ansec s used in he 2018 su ey o he No h egion, employing he same h ee colou s as
applied o he h ee sub-a eas in abo e igu e 3: Sisimiu (blue), Sisimiu Sou h (o ange) and Angujaa o iup
(pu ple).
Dis ance collec ed was he pe pendicula dis ance om he helicop e ’s lown ack line
o a muskox g oup (objec -o -in e es ), i.e., one o mo e animals. Tigh ly cohesi e
beha io iden i ied g oups o mul iple indi iduals. Dis ance eco ded was he
obse e ’s ins an aneous and subjec i e dis ance o he app oxima e cen e o he
muskox g oup om he ack line be o e any mo emen by he g oup occu ed. Exac
dis ance measu emen s we e no possible p ima ily because o p ac ical conside a ions
o his su ey (de ails in Cuyle e al. 2021). Ins ead, dis ance measu emen used he
ollowing dis ance bins: 0, 50-, 100-, 200-, 300-, 400-, 500-, 750- and 1500-me e s
pe pendicula o he line ansec . These alues co espond o he uppe limi o a
speci ic bin ha he muskox obse a ion was included in. Fo analysis, hese we e
ecoded o he mid dis ance o a speci ic bin. No e, binning accu acy elies hea ily on
obse e abili y o co ec ly es ima e dis ance o he obse ed animals. Thus, be o e
s a ing he su ey he helicop e ho e ed a he 40m al i ude used du ing line
ansec s, while each obse e used a Leica lase ange inde 1600 o gauge dis ances
ac oss he e ain. Then hey ma ked hei window wi h masking ape delinea ing he
app oxima e dis ances o each bin. When possible while lying line ansec s, he lase
ange inde s we e used o double-check epo ed bin dis ances o de ec ed g oups. The
eco ded dis ances a e used o es ima e a de ec ion unc ion, hen es ima e he
de ec ion p obabili y and inally o es ima e he densi y o he muskoxen wi hin he
24
su eyed a ea (Buckland e al. 2001). The de ec ion unc ion, 𝑔(𝑦), desc ibes he
p obabili y o de ec ing an objec -o -in e es gi en ha is a a dis ance 𝑦, om he
cen e line (0-line), hus being a non-inc easing unc ion o 𝑦 (Buckland e al. 2015). Fo
line ansec s, 𝑦 is he pe pendicula dis ance om he 0-line o he de ec ed objec .
Wi hin DS me hods, he p obabili y o de ec ion is explained ecu ing o hese
obse ed dis ances (Buckland e al. 2001).
Dis ance Sampling
The muskox g oup was he selec ed sample uni o he DS analysis o he 2018 su ey.
Nei he he indi idual muskoxen wi hin a g oup, no indi idual line ansec s we e
conside ed as he sample uni .
The eco ded dis ances o he obse ed muskox g oups we e used o es ima e a
de ec ion unc ion. Wi h his, bo h he muskox de ec ion p obabili y and densi y wi hin
he su eyed a ea could be es ima ed (Buckland e al. 2001). The de ec ion unc ion
desc ibes he p obabili y o de ec ing an objec -o -in e es (muskox g oup) ha is a a
dis ance 𝑦, om he cen e line ( ack line), hus being a non-inc easing unc ion o 𝑦
(Buckland e al. 2015). Fo line ansec s, 𝑦 is he pe pendicula dis ance om he ack
line o he de ec ed objec . Wi hin DS me hods, he p obabili y o de ec ion is explained
ecu ing o hese obse ed dis ances (Buckland e al. 2001).
P io o DS analysis, he aw da a was i s p ocessed o inconsis encies, e.g., species o
obse e names w i en sligh ly di e en ly had o be s anda dized be o e analyses o
a oid being assigned a di e en ca ego y, and one obse a ion lacked dis ance
( eplaced by he a e age obse ed dis ance). Then ex ensi e explo a o y da a analysis
was comple ed, including e alua ion o obse ed dis ances, be o e p oceeding o
de e mining he de ec ion unc ion h ough model i ing and selec ion (Buckland e al.
2001; Ma ques e al. 2011; Thomas e al.
2010).
To de e mine he de ec ion unc ion,
se e al models we e conside ed (Thomas e al. 2010). The model p esen ing he lowes
AIC alue was chosen. The subsequen analysis was based on Ma ques (2018). De ails
ega ding DS heo y, me hods and analysis a e a ailable in Buckland e al. (1993, 2001,
2015), and a b ie e summa y p o ided in Appendix 2. Fo analysis, we used R
S a is ical So wa e (h ps://www. -p ojec .o g/).
Gene alized Addi i e Model (GAM)
Gene alized Addi i e Models (GAM) a e an ex ension o Linea Models (LM) and
Gene alized Linea Models (GLM) whe e non-linea esponses wi h smoo hing
unc ions can be i ed o he da a. De ails ega ding Gene alized Addi i e Models,
25
me hods and analysis a e a ailable in Wood (2017), and a b ie e summa y is p o ided
in Appendix 3.
Densi y Su ace Model (DSM)
Con en ional DS me hods p o ide a e age es ima es o abundance o e a egion bu
no in o ma ion abou he dis ibu ion o he objec s o in e es wi hin he su ey egion.
An efficien op ion is o build a spa ial model ha inco po a es spa ially e e enced
en i onmen al co a ia es. Densi y su ace modelling uses he GAM amewo k (Wood,
2017) o build models o abundance/densi y as a unc ion o en i onmen al co a ia es,
ypically as pa o a wo-s age me hod (Fig. 5). In he i s s age, he de ec abili y ia
DS is modelled and in he second s age he coun s, co ec ed o de ec abili y, a e
modelled o e space. De ails ega ding
DSM
, me hods and analysis a e a ailable in
Ka sane akis (2007) and Mille e al. (2013). B ie e summa y is in Appendix 4.
Figu e 5. Flow diag am o he modelling p ocess du ing he analysis, om he Dis ance Sampling analysis h ough
o he Densi y Su ace Modelling.
32
Rega ding g oups o muskoxen, he numbe o de ec ions pe uni ansec leng h is he
encoun e a e. Conside ing he en i e No h egion al oge he , non- unca ed da a had
a mean encoun e a e o 0.016 muskox g oups pe kilome e . Speci ically, he Sisimiu
sub-a ea had a mean encoun e a e o 0.003 muskox g oups pe kilome e , while he
Angujaa o iup sub-a ea had a mean o 0.026. No muskoxen we e encoun e ed in he
Sisimiu Sou h sub-a ea. Wi hin he sub-a ea Sisimiu , encoun e a es we e ze o o
h ee line ansec s and spa se o he emaining i e. Fo he Angujaa o iup sub-a ea,
he encoun e a es we e high in wo possible ho spo s, he eas and wes sides o he
sub-a ea, and we e leas in he middle (Fig. 8).
His og ams examining obse e e ec s (Fig. 9) we e somewha dissimila and he e o e
his co a ia e will be in es iga ed as o whe he i in luenced esul s o de ec abili y.
O he po en ial co a ia es, like sun gla e o snow co e ing, we e a ailable bu he e
we e oo many missing obse a ions and/o inconsis ency when e e ing o he
ca ego ies o hese o be used in he analysis.
The p elimina y analysis p o ides he expec a ion o less p ecision in u he analyses
o de ec ions wi hin he Sisimiu sub-a ea, because o da a a iabili y. P ecision is
expec ed o be g ea e o Angujaa o iup, as all line ansec s con ained muskoxen.
Ne e heless, he in o ma ion ag eed well wi h an icipa ed a p io i, e.g.,
Angujaa o iup sub-a ea would ha e mo e muskoxen han he o he wo sub-a eas.
Figu e 8. Muskox su ey 2018: explo a o y analysis o non- unca ed da a o muskox encoun e a e (g oups pe
km) pe line ansec using and illus a ing sub-a ea: Sisimiu (blue, line ansec s 1-8), Sisimiu Sou h (o ange,
line ansec s 9-13) and Angujaa o iup (pu ple, line ansec s 14-19).
Line ansec numbe

33
Dis ance Sampling analysis
Be o e conduc ing any modelling, an analysis o he obse ed dis ances was made o
e alua e whe he any majo assump ion iola ion occu ed o o he da a- ela ed issue,
Figu e 9. Obse e e ec : his og ams illus a ing de ec ed dis ances o he wo obse e s (a co a ia e wi h wo
le els). Densi y, y-axis, e e s o he densi y o obse a ions.
Figu e 10. His og am o he di e en binning op ions o he muskox dis ance da a. Le : he o iginal bins as
collec ed on he su ey. Righ : an al e na i e binning o educe he e ec o heaping. The a ea o he ec angles is
p opo ional o he numbe o poin s wi hin each bin.
34
as s a ed in p e ious sec ions. These analyses a e om Ma ques (2018). The his og am
o obse ed dis ances wi h no de ined unca ion dis ance (non- unca ed da a) is like
ypical Dis ance Sampling da a, pe haps showing some o e -dispe sion, wi h no -
equally-spaced bins (Fig. 10). Gi en he his og am o binned dis ances, a s ip hal -
wid h o
𝑤 =
0.75 km was selec ed (i.e., all obse a ions a dis ances beyond 750 me e s
we e disca ded). This unca ion educed he sample size om 256 o 210 muskox
g oups o he Dis ance Sampling analysis. Da a unca ion is a common p ocedu e
because o he wise ex a adjus men e ms may be needed o i he long ail o he
de ec ion unc ion. Fu he , li le in o ma ion is los by unca ion, since da a
obse a ions loca ed mo e han 0.75 km om each side o he line make a minimal
con ibu ion o he abundance es ima e.
T unca ed da a
Wi h he o iginal binning Op ion 1 (Fig. 10), he e seem o be less han expec ed
obse a ions on he 0.05-0.10 km and 0.30-0.40 km in e als, when compa ed o he
0.10-0.30 km and he 0.40-0.50 km bins. This sugges s heaping, a phenomenon ha
occu s when obse e s end o eco d some p e e ed alues o e o he s (Buckland e
al. 2001) e.g., a ound dis ances ha a e easily chosen in he absence o a igo ous
dis ance measu ing me hod, as o en occu ed in his s udy when ugged e ain o ced
obse e s o assign dis ance bins subjec i ely. The al e na i e binning Op ion 2, wi h
less bins, educes he in luence o po en ial measu emen e o s in he obse ed
dis ances. This al e na i e binning op ion includes bin cu poin s o 0, 0.10, 0.30, 0.50
and 0.75 km (Fig. 10).
Bo h binning op ions we e conside ed in model i ing, albei only Op ion 2 minimizes
he e ec o measu emen e o induced by heaping. Since binning Op ion 1 was no
sui able o g ouping, only he analyses whose i ed models conside he second
binning op ion a e illus a ed below. The ad an age o choosing he second binning
op ion is ha i esul s in mo e eliable de ec ion unc ions. Howe e , owing o ewe
deg ees o eedom, he small numbe o bins a ec s he 𝜒2 Goodness-o -Fi es s
ollowing model i ing. The eg ession analysis sugges s ha dis ance is no a
s a is ically signi ican a iable explaining g oup size (Table 5).
Table 5. Summa y o he coe icien cha ac e is ics o he GLM be ween obse ed unca ed dis ance and he
muskox g oup size while conside ing a Poisson dis ibu ion.
Pa ame e
Es ima e
S anda d
E o
z- alue
p- alue
In e cep
1.752
0.050
35.18
0.00000
Dis ance
-0.091
0.155
-0.58
0.55926
No e: AIC = 1564.2, Null De iance = 879.5, Residual De iance = 879.2.
35
De ec ion unc ion models i ed wi h he i s binning op ion did show poo i ing,
including he bes i wi hin his g oup, since hese p esen ed se e al adjus men e ms,
due o he heaping phenomena. Below, he de ec ion unc ions a e i ed o unca ed
da a and conside ing he second binning op ion (Fig. 10), which imp o es he models.
T unca ed da a – Dis ance Sampling Models
Fo hese models, e e y combina ion o key unc ion and adjus men e ms was es ed.
The only addi ional co a ia es assessed we e obse e and g oup size, conside ing
𝑤 =
0.75
km. A summa y o he in o ma ion om each model i ed o he da a (Table 6)
p o ides a simple o e iew o se e al models, and includes he espec i e key
unc ions, adjus men e ms, model o mula
,
𝜒
2
Goodness-o -Fi es
p
- alue,
es ima es
o he de ec ion p obabili y, espec i e s anda d e o
(
se (𝑃
a)
),
and
Δ𝐴𝐼
𝐶
compa ison
be ween each model and he model wi h he lowes AIC
.
The bes model i ed o he
da a possesses he lowes change in AIC alue (Δ𝐴𝐼𝐶 = 0). Fo he 2018 muskox su ey
da a, he Hal -No mal key unc ion was selec ed because i was he mos lexible key
(AIC = 550.65). The second-bes model was Uni o m wi h cosine adjus men e ms o
o de 1 (AIC = 550.692, i.e., Δ𝐴𝐼𝐶 = 0.042). The bes i ed de ec ion unc ion pa ame e s
supe imposed wi h he obse ed dis ances’ his og am, indica es ha nei he g oup size
no obse e we e ele an co a ia es in de ec abili y o he unca ed da a. (Table 6).
The es ima ed a e aged p obabili y o de ec ion o he No h egion was 𝑃
a = 0.557 (se
= 0.033, Table 6).
Figu e 11. The de ec ed dis ances unca ed da a wi h he es ima ed de ec ion unc ion o e laid, conside ing he
binning op ion 2 ha educes he e ec o heaping.
36
Each g oup size has i s sepa a e de ec ion unc ion, co esponding o di e en
es ima es o he p obabili y o de ec ion (Fig. 12). Fo non- unca ed da ase , a g oup
size o 2 muskoxen p esen s an es ima ed p obabili y o de ec ion o 0.273, a g oup size
o 10 has an es ima e o 0.45, a g oup size o 20 has an es ima e o 0.74, while a g oup
size o 30 has an es ima e o 0.96 (Fig. 12). Wi h inc easing g oup size, he p obabili y o
de ec ion also inc eases. Fo he unca ed da ase , he p obabili y o de ec ion (0.557)
was he same ega dless o muskox g oup size. The Kolmogo o -Smi no and C amé -
on Mises es s (Appendix 2) canno be applied since he dis ances we e ep esen ed as
a disc e e a iable.
Figu e 12. Es ima ed p obabili ies o de ec ion, gi en non- unca ed and unca ed da ase , o each obse ed g oup
size (muskoxen) ob ained wi h he i ed model. Illus a es g oup size no impo an in unca ed da ase .
Figu e 13. Muskox encoun e a e (g oups pe km) pe line ansec , su ey 2018: explo a o y analysis o unca ed
da a, illus a ing sub-a ea: Sisimiu (blue), Sisimiu Sou h (o ange) and Angujaa o iup (pu ple).
37
Encoun e a e es ima es sugges he Angujaa o iup sub-a ea has he mos muskoxen,
since i s es ima e is la ge han he o he sub-a eas (Table 7, Fig. 13). Visualiza ion o
he de ec ed muskox dis ibu ion shows his was somewha con inuous only o line
ansec s in he Angujaa o iup sub-a ea, wi h wo ‘ho ’ spo s, i.e., eas , and wes e n
sides o he sub-a ea (Fig. 14). Muskox dis ibu ion in Sisimiu was spo adic and a e,
and in Sisimiu Sou h nonexis en . The DS design-based es ima es o muskox
abundance and densi y also e eal ha Angujaa o iup is he sub-a ea wi h he mos
muskoxen (Tables 8, 9, Fig. 15).
Figu e 14. ‘Hea ’ map illus a ing ela i e dis ibu ion o muskoxen obse a ions along he line ansec s in he
No h egion, Wes G eenland. Fo in e p e a ion o unde lying map see Fig. 1.
In Ma ch 2018, he DS design-based popula ion es ima e o he combined
Angujaa o iup and Sisimiu sub-a eas was app oxima ely 21,746 muskoxen (95% CI:
11,061 – 42,751), wi h a CV o 28.5% (Table 8). The DS densi y es ima e was 1.1
muskoxen pe km2, wi h 95% CI: 0.56 – 2.16 (Table 9, Fig. 15).
Sepa a ely, he Manii soq muskox ha es managemen a ea (Angujaa o iup sub-
a ea) had a DS design-based popula ion es ima e o 18,906 muskoxen (95% CI: 8,726 –
40,960), wi h a CV o 31.5%. Simila ly, Sisimiu muskox ha es managemen a ea had
a popula ion es ima e o 2,840 muskoxen (95% CI: 662 – 12,178), wi h a CV o 56.8%.

38
Table 6. Model compa ison ac oss he h ee Con en ional Dis ance Sampling models and models conside ing
muskox g oup size and obse e as co a ia es.
Key unc ion
Fo mula
𝒙𝟐
p- alue
𝑷
a
se
(𝑷
a)
∆
AIC
Hal -no mal
(same esul o Cosine/Simple Polynomial/He mi e)
1
0.932
0.557
0.033
0.000
Uni o m wi h cosine adjus men e ms o o de 1
NA
0.913
0.565
0.025
0.042
Uni o m wi h He mi e polynomial adjus men e m o o de 4
NA
0.448
0.604
0.024
1.509
Hal -no mal
G oup size
0.717
0.557
0.033
1.673
Hal -no mal
Obse e
0.709
0.557
0.033
1.953
Haza d- a e (same esul o Cosine/Simple Polynomial/He mi e)
1
0.616
0.584
0.054
2.111
Uni o m wi h simple polynomial adjus men e ms o o de 2,4,6
NA
0.498
0.577
0.047
2.325
Haza d- a e
G oup size
NA
0.596
0.052
3.575
Haza d- a e
Obse e
NA
0.589
0.053
4.004
No e: unde Fo mula explana o y a iables a e as ollows: G oup size = g oup size as a iable, 1 = o Uni o m key, Obse e = obse e as
a iable, NA = no explana o y a iables/co a ia es. The e we e no enough deg ees o eedom o he 𝜒2 Goodness-o -Fi es , hus he ‘NA’
alues. Fo each key unc ion, all h ee se ies expansions (Cosine, Simple Polynomial, He mi e (Appendix 2, Table 15) we e applied.
Table 7. Encoun e a e (ER) es ima es pe sub-a ea o muskox g oups conside ing unca ed da a, h ee sub-a eas
(s a a), ou bins, and a Hal -No mal de ec ion unc ion i ed wi hou co a ia es.
Su eyed
Sub-a ea*
Muskox ha es
Managemen a ea
Encoun e
a e
S anda d E o
(se)
Coe icien o Va iance
(c )
Sisimiu
Sisimiu
0.020
0.007
0.360
Angujaa o iup
Manii soq
0.414
0.092
0.222
TOTAL
Sisimiu + Manii soq
0.173
0.067
0.387
*Thi d sub-a ea, Sisimiu -Sou h, does no appea in he able because he e we e no obse a ions o muskoxen.
Table 8. Ma ch 2018, he Dis ance Sampling design-based muskox abundance es ima es pe s a um (sub-a ea) in
he No h egion, conside ing unca ed da a, h ee sub-a eas (s a a), ou bins and a Hal -no mal de ec ion
unc ion wi h no co a ia es.
Sub-a ea*
Muskox ha es
Managemen a ea
Popula ion Size
Es ima e
SE
CV
95% Con idence In e al
Lowe
Uppe
Sisimiu
Sisimiu
2,840
1,613
0.568
662
12,178
Angujaa o iup
Manii soq
18,906
5,948
0.315
8,726
40,960
TOTAL
Sisimiu + Manii soq
21,746
6,194
0.285
11,061
42,751
No e: SE = S anda d E o , CV = Coe icien o Va iance.
*Thi d sub-a ea, Sisimiu -Sou h, does no appea in he able because he e we e no obse a ions o muskoxen.
Table 9. Ma ch 2018, he Dis ance Sampling design-based muskox densi y es ima es pe s a um (sub-a ea) in he
No h egion, conside ing unca ed da a, h ee sub-a eas (s a a), ou bins and a Hal -no mal de ec ion unc ion
wi h no co a ia es.
Sub-a ea*
Muskox ha es
Managemen a ea
Densi y
Es ima e
SE
CV
95% Con idence In e al
Lowe
Uppe
Sisimiu
Sisimiu
0.224
0.127
0.568
0.052
0.962
Angujaa o iup
Manii soq
2.650
0.834
0.315
1.223
5.742
TOTAL
Sisimiu + Manii soq
1.099
0.313
0.285
0.559
2.160
No e: SE = S anda d E o , CV = Coe icien o Va iance.
*Thi d sub-a ea, Sisimiu -Sou h, does no appea in he able because he e we e no obse a ions o muskoxen.
39
Figu e 15. Dis ance Sampling design-based muskox densi y es ima es wi h co esponding con idence in e als o
he wo sub-a eas wi h muskoxen, Sisimiu and Angujaa o iup, and inally o he o al No h egion.
GAM & DSM
A e spli ing he ansec s in o segmen s, hei cen oids we e de e mined and
in e sec ed wi h he shape iles conce ning he spa ial a iables. These a iables we e
GPS coo dina es, ege a ion, ele a ion, slope, and he aspec (geog aphical compass
di ec ion, which ela es o sun exposu e and empe a u e o a pa icula loca ion)
(Appendix 6). The ege a ion co a ia e was excluded om he analysis since he
cu en in o ma ion a ailable lacks he necessa y esolu ion. Fu he mo e, a p edic ion
da a se was gene a ed o he whole s udy egion by con e ing he egion in o small
1.5 km sided squa e cells (i.e., 2.25 km2, no e 𝑤 = 0.75 km).
Once he dis ance model was i ed, 𝑃
a could be de e mined o each g oup size and
hus 𝑛𝑖, he numbe o muskoxen de ec ed in g oup i, can be co ec ed as 𝑛i = 𝑛𝑖
𝑃
a. These
p edic ed alues o coun s we e hen modelled using a GAM i ed o he spa ial
co a ia es along longi ude (lon) and la i ude (la ) conside ed join ly. Wi hin he model,
he p e iously es ima e p obabili y o de ec ion was de ined as he o se e m along
wi h he cell a ea o 2.25 km2. The dis ibu ion conside ed in model i ing we e
Tweedie and Nega i e Binomial. These p o ide a lexible al e na i e o he quasi-
Poisson dis ibu ion, which does no cap u e he esponse o e dispe sion.
Table 10. GAM model summa y able ela i e o smoo h e ms o co a ia es and conside ing Tweedie dis ibu ion.
E ec i e Deg ees o F eedom
Chi-squa e
p- alue
s(lon, la )
19.235
9.021
0.0000
s(ele a ion)
4.024
4.238
0.0009
No e: 𝑅𝑎𝑑𝑗
2 = 0.378, De iance explained = 63.6%.
40
Table 11. GAM model summa y able ela i e o smoo h e ms o he co a ia es and conside ing a Nega i e
Binomial dis ibu ion.
E ec i e Deg ees o F eedom
Chi-squa e
p- alue
s(lon, la )
19.667
159.429
0.0000
s(aspec )
1.000
4.215
0.0401
s(ele a ion)
3.692
15.404
0.0123
No e: 𝑅𝑎𝑑𝑗
2 = 0.091, De iance explained = 67.6%.
Fi ing GAM/DSM
The wo dis ibu ion models, Tweedie, and Nega i e Binomial (Tables 10, 11) we e
i ed and he bes model wi hin each was selec ed o u he wo k. The Tweedie
dis ibu ion used longi ude, la i ude, and ele a ion. The Nega i e Binomial dis ibu ion
(NegBin) used longi ude, la i ude, aspec , and ele a ion. The hi d bes NegBin model
was simila o he i s (i.e., Table 11) bu lacked ele a ion and he ∆AIC was low,
0.8799. The second bes , which included all co a ia es, had a ∆AIC di e ence o 0.61.
Thus, hese models we e almos iden ical.
Figu e 16. Fi ed DSM plo co esponding o he i ed smoo h o he ele a ion co a ia e (Y-axis is he i ed
smoo hing pa ame e ). G een shading illus a es he s anda d e o o he es ima es, while he espec i e
obse a ions a e ep esen ed on he ho izon al axis.
Despi e he simila i y among esul s, he model conside ing he Tweedie dis ibu ion
was chosen owing o i s lowe AIC alue (AICTweedie = 1350.1 and AICNegBinom = 1461.5).
A la ge E ec i e Deg ees o F eedom (EDF) esul ed o he bi a ia e smoo h
associa ed wi h he pai ed longi ude and la i ude ela i e o he o he en i onmen al
co a ia es, since mo e basis unc ions a e equi ed o i a su ace han a line. Each o
hese en i onmen al co a ia es does no appea o be linea ly ela ed wi h he
esponse, since he espec i e EDF is la ge han 1. Rega ding he smoo h unc ions, he
ela ionship be ween he esponse a iable and each explana o y a iable appea s o be
Ele a ion (m)
41
non-linea o he ele a ions below 1000m (Fig. 16). Speci ically, ele a ions om 300m
o 700m seem o be p e e ed by he muskoxen in he ea ly Ma ch pe iod o he su ey
and densi y was posi i ely co ela ed wi h ele a ion un il a ound ca. 400 m when i
began o all sligh ly wi h ele a ion (Fig. 16). A high ele a ion, e.g., abo e 900m,
a iabili y was g ea e , and densi y ell. Muskoxen appea o p e e sou h- acing slopes
(90°-270°), since hese in ol ed mos obse a ions and he e was he leas a iabili y in
he es ima es. Fu he mo e, he esul s sugges ed ha among sou h acing slopes,
sou hwes aspec s we e mos p e e ed by muskoxen.
Spa ial p edic ion
The numbe o muskoxen wi hin each cell was hen p edic ed using he GAM model
(𝑛i) and spa ially ep esen ed. The hea map p esen ing he p edic ion o muskox
abundance in he No h egion indica es ha muskoxen a e sca ce in mos o he No h
egion, excep ing he Angujaa o iup sub-a ea, which clea ly illus a ed he muskoxen
dis ibu ed in o wo clus e s (ho spo s) a he ime o he su ey (Fig. 17). The
GAM/DSM model-based app oach p oduced a p edic ed mean muskox densi y alue
o he en i e No h egion o 3.69 muskoxen/km2 (95% CI: 2,87 – 4,74). The p edic ed
ange was om 0 o 125 muskoxen/km2 (Fig. 17). The maximum alue occu ed a ely
and only on a emo e sec ion o line ansec 14, which was on he a wes side o he
Angujaa o iup sub-a ea. This was one o he wo ho spo s o clumped dis ibu ion
whe e mos o he da ke cells we e in he ange o 25-30 muskoxen/km2. The
clumping coincided wi h unusually high ele a ion.
The GAM/DSM model-based app oach p oduced a popula ion es ima e o 23,256 (95%
CI: 18,102 – 29,877) muskoxen o he en i e No h egion. Addi ionally, a a iabili y
map was p oduced wi h he CV o each es ima e in he su ey egion (Fig. 18), he CV
es ima e was 11.36%.
Table 12. Compa ison o Dis ance Sampling design-based and GAM/DSM analyses, o en i e No h egion.
Analysis
Popula ion Size
CV o
p obabili y o
de ec ion
Es ima e
SE
CV
95% Con idence In e al
Lowe
Uppe
Dis ance Sampling
21,746
6,194
0.285
11,061
42,751
5.98%
GAM / DSM
23,256
0.114
18,102
29,877
11.36%
No e: SE = S anda d E o , CV = Coe icien o Va iance.
To compa e he model-based GAM/DSM o he design-based DS app oach we used
he CV o p obabili y o de ec ion o he muskoxen (i.e., no o abundance o densi y).
The GAM/DSM had a p obabili y o de ec ion CV o 11.36% (like he es ima e’s CV),
while he DS had a p obabili y o de ec ion CV o 5.98% (Table 12). Fo he o e all
analyses, i.e., DS and GAM/DSM, he p obabili y o de ec ion CV was 12.84%.
48
season and du ing he Ma ch ophy season Manii soq muskoxen inhabi ed unusually
high ele a ions, >700 m, we sugges ha cu en an h opogenic dis u bance associa ed
wi h hun ing may play a majo ole in causing muskox use o high ele a ions.
Consequences o win e o aging in sub-op imal habi a
Rega dless o he cause(s), in 2018 mos Manii soq muskoxen in he Angujaa o iup
sub-a ea o aged a ele a ions o 700-800 m in win e . This is subop imal habi a . Gi en
he la i ude (ca. 67°N), win e o age a ailable a 700-800 m ele a ion is in e io and less
plen i ul han ha ound in lowland alleys (Kö ne 2007), and speci ically o he
Angujaa o iup sub-a ea high ele a ions a e o en ell ield, ab asion pla eaus and ba e
g ound (Tams o e al. 2005). Fo aging in poo habi a a high ele a ion could
nega i ely a ec win e body condi ion and ul ima ely su i al o Manii soq muskoxen
(see Appendix 9 o de ails). Fu he , i is common knowledge ha among la ge
he bi o es mos e al g ow h occu s in la e ges a ion and ha his g ow h subs an ially
inc eases ma e nal ene gy equi emen s. Muskox cal es a e usually bo n mid-Ap il
h ough June, wi h bi hing da es as ea ly as 05 Ap il and as la e as 19 June (Len 1988)
as he e is no b eeding synch ony as seen in ca ibou (Len 1966). Depending on he da e
o pa u i ion, la e ges a ion o muskoxen anges om Feb ua y h ough May. Thus,
Manii soq cows o aging a high ele a ion poo pas u e could become nu i ionally
de icien in la e ges a ion and bi h weake cal es o low weigh (Olesen e al. 1994).
Cal pe cen age
Fo age quali y a ies widely seasonally, wi h win e associa ed wi h lowes o age
quali y (De mane e al. 2015). The Angujaa o iup sub-a ea s addles he A c ic Ci cle
(66°33’ N). A hese la i udes, o age dec eases apidly wi h ising ele a ion (Tams o
e al. 2005, Kö ne 2007). Muskoxen p e e ing o o age in lowlands yea - ound is
he e o e no su p ise. We suspec ha obse ed win e use o ege a ion poo high
ele a ions, 700-800 m, by muskoxen in Angujaa o iup sub-a ea, could comp omise
hei body condi ion, wi h nega i e consequences o body condi ion and inally win e
su i al and cal p oduc ion. Since muskoxen a e capi al b eede s (Des o ges e al.
2019), cows ely on body ese es o pa u i ion and ea ly lac a ion. Thus, muskox
cows en e ing pa u i ion wi h low body condi ion may bi h cal es bu lack su icien
ese es o milk p oduc ion, which may educe cal su i al. Thus, he win e hun ing
season coinciding wi h la e ges a ion is no expec ed o be compa ible wi h a high
pe cen age o cal es. The gene al decline o cal pe cen age in he 2000-2010 pe iod and
addi ionally he Ma ch 2018 cal (age <1-yea ) pe cen age o ca. 18.4%, appea s o
suppo his. The Ma ch 2018 alue is a he low end o he ange 17-24% epo ed o
expanding Alaskan and Canadian muskox popula ions (Jing o s & Klein 1982, Gunn e
al. 1984). Fu he , i is well below he excellen 32% ha cha ac e ized he Manii soq

49
muskox popula ion in he 1970’s and in o he la e 1980’s (Olesen 1993, Cuyle &
Wi ing 2004, Appendix 9).
In addi ion o he possible ole o he win e hun ing season in educing cal
pe cen age, oo many animals in an a ea ela i e o he a ailable quali y and quan i y
o o age can b ing o h densi y-dependen ac o s ha may educe cal p oduc ion in
la ge he bi o es (Kie & Whi e 1985, Ouelle e al. 1997). Densi y-dependen ac o s a e
suspec ed o playing a ole in he s eady decline o Manii soq muskox cal pe cen age
in he 2000–2004 pe iod (Cuyle & Wi ing 2004). A ha ime, obse ed densi ies on
p e e ed lowland win e o age a e aged om 4 o 16 muskoxen/km2 (Cuyle &
Wi ing 2004), wi h maximums o 21-29 muskoxen/km2 in alleys nea he own o
Kange lussuaq (Cuyle e al. 2001). The minimum coun s o 2000, 2001, 2002 and 2004
obse ed simila numbe s o muskoxen (Cuyle & Wi ing 2004). Ini ial cal pe cen age
was 26%, and d opped wi h each coun , 25%, 21% and 18%, espec i ely. The la e wo
we e also below hose p edic ed by Bayesian modelling (Cuyle & Wi ing 2004).
Meanwhile, he ough densi y o muskoxen in lowland ele a ions appea ed o ise,
being ca. 2.5/km2 in 2000 and 3.1/km2 in 2004 (Cuyle & Wi ing 2004). The high
densi ies we e obse ed p io o he win e hun ing season and made densi y-
dependen ac o s a plausible cause behind he declining cal pe cen ages in 2000-2004.
The obse ed dec easing win e cow ump a dep h and p egnancy a es o he same
pe iod suppo his, since no due o empo a y ac o s, e.g., wea he e en s, since
simila wea he condi ions applied ac oss yea s (Cuyle & Wi ing 2004). Fu he ,
win e ha es ing may also ha e played a ole in he alling cal pe cen ages. Al eady
om win e 2000, once snowmobile use commenced o win e hun ing hen mos
muskoxen le he alleys and mo ed in o highe ele a ions (Cuyle unpublished).
Gi en high ele a ion pas u e is poo ela i e o lowlands and p e ious muskox
densi ies in lowlands, his mo emen o muskoxen o highe ele a ions may ha e
exace ba ed he densi y-dependen ac o s suspec ed o al eady ope a ing in he
lowlands.
In win e 2018 mos o he es ima ed popula ion o 18,906 Manii soq muskoxen we e
obse ed a ele a ions abo e 700 m, which p o ide sub-op imal habi a o o aging.
Since cal pe cen age was only app oxima ely 18% cal es, popula ion g ow h is
possible (albei likely slow) bu no ce ain. Unde cu en ha es egimes ha cal
pe cen age may be insu icien o popula ion size s abili y. Wi h almos 19,000
Manii soq muskoxen on poo high ele a ion pas u e in win e , densi y-dependen
e ec s may be expec ed. Cal pe cen age migh imp o e i Manii soq muskox cows
we e able o o age undis u bed in lowlands du ing la e ges a ion.
50
Muskox popula ion size & densi y
Su ey co e age o he No h egion (23,303 km2) was 10.6%, which is usually su icien
o acili a e eliable es ima es. Bo h DS design- and GAM/DSM model-based
app oaches we e applied o he da ase o ob ain compa able es ima es o muskox
abundance and densi y. DS design-based es ima es a e based on he selec ed ansec s,
which may o may no adequa ely ep esen e e y ea u e wi hin he s udy egion, as
some ea u es may be o e - ep esen ed, while o he s unde - ep esen ed. Meanwhile,
GAM modelling conside s he en i onmen al co a ia es o he whole egion, allowing
he spa ial ep esen a ion o he es ima es ob ained and a isualiza ion o pa e ns in
abundance.
Mos muskoxen we e obse ed in he Angujaa o iup sub-a ea (Manii soq muskox
ha es managemen a ea), ew in he Sisimiu sub-a ea (Sisimiu muskox ha es
managemen a ea), and none in he Sisimiu Sou h sub-a ea.
Conside ing he combined Angujaa o iup and Sisimiu sub-a eas, he DS analysis
es ima ed muskox abundance a ca. 21,746 muskoxen (Table 12). Tha es ima e,
howe e , had a wide 95% CI (11,061 – 42,751), which indica es he ange o possible
abundance is b oad. Fu he he CV was la ge, 28.5%, illus a ing he magni ude o
imp ecision on he es ima e. The GAM/DSM analyses p oduced a sligh ly la ge
es ima e o 23,256 muskoxen, which had a igh e 95% CI (18,102 – 29,877) and a lowe
CV o 11.4%. The wo abundance es ima es may be conside ed simila because hey
di e by only 7-8% and 95%CIs o e lap.
The DS popula ion es ima e was no p ecise as e idenced by he la ge CV. The main
ac o con ibu ing o imp ecision was he high a iabili y wi hin he da ase . Fo
example, mos ansec s lown had ze o muskox obse a ions while a ew ansec s had
many. Fu he , he numbe o muskox g oup obse a ions was low (n = 210, unca ed
da ase ) combined wi h a la ge ange in he numbe o muskoxen indi iduals wi hin
g oups. Speci ic o he Angujaa o iup sub-a ea (Manii soq muskox ha es
managemen a ea) was he unusual, clumped dis ibu ion o he muskoxen in o wo
ho spo s and ew muskoxen elsewhe e. Speci ic o he Sisimiu sub-a ea (Sisimiu
muskox ha es managemen a ea) was he eno mous a ea su eyed and he
dominance o ze o obse a ions.
Conside ing he sub-a eas su eyed, he DS es ima e o he Angujaa o iup sub-a ea
was ca. 18,906 Manii soq muskoxen (95% CI: 8,726–40,960; CV = 31.5%) (Table 8).
Simila ly, he Sisimiu sub-a ea was ca. 2,840 Sisimiu muskoxen (95% CI: 662–12,178;
51
CV = 56.8%). Fo each sub-a ea, like wi h he o e all es ima e, he DS p o ided
imp ecise es ima es wi h a b oad ange o possible popula ion size.
A i s glance, he GAM/DSM es ima e o he en i e No h egion appea s bes ,
howe e , o pe mi compa ison o he DS and GAM/DSM app oaches we used he CV
o p obabili y o de ec ion o muskoxen wi hin he en i e egion. The esul was ha
DS was be e han he GAM/DSM, wi h CVs o 5.98% and 11.36%, espec i ely (Table
12). This indica es ha despi e unce ain y he design-based DS app oach p o ided a
mo e eliable popula ion es ima e han he model-based GAM/DSM o his da ase .
Rega ding ha es managemen decisions, we sugges using he design-based DS
es ima es o abundance, albei ully awa e o la ge imp ecision. The DS also p o ides
es ima es speci ic o each popula ion, i.e., Manii soq and Sisimiu , which can be
help ul o applying popula ion speci ic ha es managemen . Gi en he unce ain y in
he es ima es, e ing owa ds cau ion is sugges ed ega ding abundance, i.e., conside
alues in he lowe ange o he 95% CIs.
Rega ding muskox densi y, he design-based DS densi y es ima es a e assumed mo e
immedia ely usable han hose om he GAM/DSM. The design-based DS es ima e o
he combined Angujaa o iup and Sisimiu sub-a eas was 1.1/km2 (95% CI: 0.56 –
2.16). Speci ically, he Angujaa o iup sub-a ea densi y was ca. 2.65 muskoxen/km2
and he Sisimiu sub-a ea ca. 0.22 muskoxen/km2. These muskox densi ies in ui i ely
ma ch obse e expe ience. Howe e , densi y es ima es om he model-based
GAM/DSM app oach did no . The GAM/DSM densi y es ima e o he en i e No h
egion was 3.69 muskoxen/km2 (95% CI: 2,87 – 4,74), which al hough no ealis ic is
easonable gi en he high concen a ion o muskoxen a wo ho spo s. Simila ly
imp obable, es ima ed GAM/DSM densi ies had a maximum o 125 muskoxen/km2
(Fig. 17) o a e high ele a ion loca ions in he Angujaa o iup sub-a ea. We do no
sugges a li e al in e p e a ion. Ins ead, he maximum GAM/DSM alue e lec s he
ex eme deg ee muskox dis ibu ion was clumped in he Angujaa o iup sub-a ea
du ing he su ey pe iod. I is inhe en ly ha de o es ima e abundance and densi y
when clumping is se e e. This da ase had eno mous a iabili y, which in addi ion o
clumping also included highly a iable g oup size. Addi ionally, i is impo an o
conside he a iabili y in he densi y da a associa ed wi h he co a ia es (lon, la , and
ele a ion). The ho spo wi h he highes densi ies was associa ed wi h ele a ions ha
ypically anged om 300 m o 500 m. Howe e , he second ho spo was associa ed
wi h lowe ele a ions, which made he densi y es ima es lowe , albei he densi ies
we e s ill la ge (e.g., 70 muskox/km2). Fu he , he Sisimiu sub-a ea has ex ensi e
e ain wi h ele a ions om 300 m o 500 m (Appendix 6 - Fig. 32), bu no ho spo
52
occu ed. Ins ead, muskox de ec ions we e spa se in Sisimiu , and he absence o an
es ima ed ho spo is based on he muskox obse a ions. Finally, he la ges es ima ed
GAM/DSM densi ies in he ho spo s a e associa ed wi h he highes unce ain y (Figs.
17, 18). The abo e makes es ima ing abundance and densi y di icul . Ne e heless, he
GAM/DSM densi y pa e n (Fig. 17) clea ly illus a es he clumping o Manii soq
muskoxen in o a eas emo e om human in luence.
As he e is less p ecision on he GAM/DSM densi y and he CIs om he wo
app oaches do no o e lap, we sugges gi ing he GAM/DSM densi y less ‘weigh ’,
al hough alues in he lowe CI ange may be ele an a speci ic loca ions. We expec
he design-based DS densi ies a e close o ac ual densi ies and ecommend hei use in
managemen decisions.
In 2018 he Manii soq muskox popula ion in he Angujaa o iup sub-a ea had a
densi y, 2.65/km2. Elsewhe e in he A c ic (Canada’s No hwes Te i o ies and
Nuna u , and No heas G eenland), muskox densi ies a e ypically much lowe ,
ha ing a ange om 0.02 o 1.8/km2 and a e aging ca. 0.86 muskoxen/km2 (Cuyle e
al. 2001). In he pe iod 2000-2004 Manii soq muskoxen densi y in lowland ele a ions
<400 m was ca. 4-5/km2 (Cuyle e al. 2001, Cuyle and Wi ing 2004). These alues,
howe e , applied only o he no he n po ion o he Angujaa o iup sub-a ea.
Rega dless, bo h p esen and pas Manii soq muskox densi ies a e ela i ely high,
which a es s o he habi a sui abili y o he Angujaa o iup sub-a ea o muskoxen.
This is u he illumina ed by he ex eme densi ies encoun e ed in he 2000-2004
pe iod o muskoxen o aging in lowland ele a ions <400 m. Examples om co e
win e g azing a eas included, 21.6 muskox/km2 in he alley, Ø kendalen (n=589
muskoxen in a ea o 20.25 km2), and 29.1 muskox/km2 in he lowlands and i e
alleys su ounding he Ammalo oq Lake (n = 876 muskoxen in a ea o 40.5 km2)
(Cuyle & Wi ing 2004). The alling cal pe cen ages o he 2000-2004 pe iod, sugges
ha , a hose densi ies, densi y-dependen ac o s may ha e begun nega i ely a ec ing
he popula ion.
A s udy o examine he cu en he bi o e ca ying-capaci y o he Angujaa o iup sub-
a ea’s lowland g ass/sedge Kob esia s eppe is wa an ed, as is a simila s udy o
ele a ions o 700-800 m. Speci ically, examining he e ec o o aging and ampling on
ege a ion a cu en muskox densi y, while conside ing ha öhn winds c ea e
expanses o snow- ee win e pas u es whe e he ege a ion is exposed o ha sh
condi ions including empe a u es well below ze o and powe ul winds.
53
Muskox popula ion end
F om he 2018 su ey, he DS and GAM/DSM analyses p o ided popula ion size
es ima es o muskoxen in he No h egion. These a e he i s popula ion size
es ima es based on ae ial su ey da a. Alone, he 2018 es ima es canno indica e cu en
popula ion end, because his would equi e a leas wo addi ional ae ial su ey
es ima e poin s using simila me hods. Addi ional u u e su eys will be needed. In he
pas , Manii soq muskoxen (Angujaa o iup sub-a ea), ecei ed g ound-based
minimum coun s (Appendix 1: Table 13, Appendix 9). Howe e , hese canno be used
o compa ison o he 2018 ae ial es ima e, p ima ily because hose coun s epo only
he numbe o animals obse ed and do no es ima e popula ion size. Rega dless, he
2018 es ima es o muskox popula ion size clea ly illus a e an immense popula ion
g ow h since 27 indi iduals we e ansloca ed o he a ea o e 50 yea s ago.
Fo he Manii soq muskoxen, popula ion g ow h appea s o ha e been s eepes in he
la e 1980’s and ea ly ‘90’s (Appendix 9). In he 2000-2020 pe iod, g ound-based win e
minimum coun s obse ed ha o al muskox numbe and numbe o g oups became
ewe , while simul aneously g oup size, cal pe cen age and cal ec ui men dec eased
(Appendix 9). Combined, hese indica e a declining Manii soq popula ion since he
mid-2000s, as does he plo o obse ed muskoxen om minimum coun s o he
Angujaa o iup sub-a ea (Fig 22).
Figu e 22. Index o Manii soq muskox abundance in he Angujaa o iup sub-a ea since ansloca ion o he egion
in 1963-65. Da a a e minimum coun s o only hose indi iduals obse ed and canno be in e p e ed as es ima es o
popula ion size. All bu 2018 a e g ound coun s. All a e o a ying iming, e o , and a ea co e age. Beginning
2000, all a e p e-cal ing coun s om ei he win e o la e win e . Those om 2000 o 2010 we e p e-ha es , while
hose 2014-2020 we e pos -ha es . In he la e muskoxen we e a oiding p e e ed lowlands. The ac ual numbe o
indi idual muskoxen obse ed du ing he Ma ch 2018 helicop e su ey () has been included as a ype o
minimum coun , which e o and co e age we e he mos comp ehensi e ela i e o any o he minimum coun
p esen ed. (F om Cuyle 2020).
0
1000
2000
3000
4000
5000
6000
1950 1960 1970 1980 1990 2000 2010 2020 2030
Numbe muskoxen obse ed
Yea

54
Repo ed ha es s also sugges declining muskox numbe s. Ha es inc eased a e
2002, he peak was in 2008 and gene ally has declined e e since (Fig. 2, Appendix 8:
Table 17). The declining ha es s ollowing 2008 (Fig. 2) a e e lec ed in he declining
index o muskox abundance s a ing om abou he mid-2000s (Fig. 22). Bayesian
analyses o he 2000-2004 minimum coun s in he Angujaa o iup sub-a ea concluded
ha ha es ing mo e han ca. 1600 muskoxen annually would likely educe Manii soq
muskox popula ion size and ha es s ha la ge would no be sus ainable o e he long
e m (Cuyle & Wi ing 2004). Ne e heless, epo ed annual ha es om 2004 o 2016
always exceeded 1600 muskoxen (Appendix 8: Table 17). Se en o hose yea s exceeded
2000 muskoxen annually. Then, he ha es s in 2017 and 2018 we e he lowes in o e a
decade, alling below 1600 muskoxen. Since hun e e o o he win e ha es has
emained high and e en inc eased while epo ed ha es s diminished, his suppo s
he 2004 p edic ion ha ha es s o e 1600 animals would no be sus ainable, likely
owing o popula ion decline. Al e na ely, wi h mos muskoxen in high inaccessible
e ain in win e , as in Ma ch 2018, ha would no acili a e hun e success ei he .
Meanwhile, sex-biased ha es ing may ha e occu ed. Fo he pe iod 2011-2013 he
comme cial and ec ea ional ha es s as pe sæ meldingsskemae ne (special epo ing
o ms, which in ol e only a po ion o he Pinia neq epo ed ha es ) we e sex-biased
owa ds cows, i.e., 54.5% cows and 44.2% bulls (Cuyle & Raund up 2014). Whe he
his cow-biased ha es ing among comme cial and ec ea ional hun e s con inued, o
changed has no been in es iga ed, no whe he he sex-biased esul s om he
sæ meldingsskemae ne can be applied o he en i e ha es , i.e., Appendix 8: Table 17.
Ob iously, i annual ha es s epea edly ook mo e cows han bulls hen popula ion
decline could be expec ed.
This cen u y has seen a apid expansion o local and in e na ional demand o qi iu
and qi iu p oduc s. This mo i a ion likely played a ole in aising numbe s ha es ed
since 2000, speci ically because p ices paid o aw skins ha e sky ocke ed and a e
highe han e e be o e. By 2017, he inc easing wo ldwide demand o qi iu esul ed
in in e na ional buye s willing o pay G eenland hun e s up o DKK 9.500,00 ($1,500 US
/ 1,280 EUR) o one win e skin (Fe nando Al a ez pe s. comm.). Since 2014, in ac
skinless ca casses ha e been disco e ed despi e a emp s o hide hem (Cuyle
unpublished 2014 ield epo ). Illegal skin-only ha es is an addi ional mo ali y o
unknown magni ude, which migh con ibu e o possible popula ion decline.
Rega ding Manii soq muskox cal pe cen age, he 2018 alue o 18.4% (albei a ough
app oxima ion) means hey a e less p oduc i e han be o e 2000, when cal pe cen age
ypically was abo e 25% (Roby 1978, Thing e al. 1984, Olesen 1993, Cuyle & Wi ing
55
2004, Appendix 9). Al eady in he 2000-2004 pe iod, cal pe cen age dec eased om ca.
26% o ca. 18% (Appendix 1, Table 13). In ha pe iod, muskox densi ies we e o en
exceedingly high in he lowland ele a ions, which sugges s nega i e densi y-
dependen ac o s could ha e been in ol ed. I may be coincidence ha o e he same
pe iod epo ed ha es mo e han doubled om 716 o 1614 muskoxen annually
(Appendix 8, Table 17) and muskoxen we e no ed o mo e in o inaccessible high
ele a ions once he win e hun ing season began. The ea e , excep ing 2005, cal
pe cen ages con inued o decline, 15.5% in 2006, 11.0% in 2009 and ca. 12% in 2010.
Again, ha es s we e hea y, bu now muskox densi ies had allen i.e., densi y-
dependen ac o s would ha e had less a ec . Cal pe cen ages om 14.6% o 16.5% in
No h Ame ican ca ibou popula ions do no pe mi popula ion size inc ease because
ha le el o cal ec ui men equals adul mo ali y, while below hose decline is
ine i able (Be ge ud e al. 2008). Meanwhile o muskoxen, cal pe cen ages o 17-24%
can acili a e popula ion g ow h (Jing o s & Klein 1982, Gunn e al. 1984). The 2018
Manii soq cal pe cen age o 18.4% is in he low end o ha ange, speci ically gi en he
absence o la ge p eda o s. P eda ion is no p essing cal p oduc ion down o
Manii soq muskoxen. Unde cu en condi ions, in e p e ing whe he a alue o 18.4%
cal es is su icien o popula ion s abili y o g ow h is di icul . Fu u e popula ion
end is no ob ious.
We suspec ha a ac o behind low cal p oduc ion is poo cow body condi ion.
Muskoxen a e gene ally no as p oduc i e as ca ibou, and he impo ance o cow body
condi ion begins al eady p io o he la e summe u . Muskox cows equi e 22% body
a o ha e a 50% p obabili y o p egnancy, while ca ibou need only 7% body a (C ê e
e al. 1993, Adamczewski e al. 1998, Pachkowski e al. 2013). This is impo an because
he p obabili y o success ul b eeding du ing he u inc eases wi h he body mass o
cows (Rowell e al. 1997, Whi e e al. 1997). Thus, muskox cow p egnancy a es a e
sensi i e o nu i ional in luences, which i poo lead o ep oduc ion declines
(Adamczewski & Flood 1997, Whi e e al. 1997, Adamczewski e al. 1998). E e y hing
ha makes ege a ion/ o age inaccessible, ul ima ely educes cow body ese es and
hus cal p oduc ion. Fo example, win e hun ing ac i i ies in he ege a ion ich
lowlands, appea o cause muskoxen o o age a high ele a ion whe e ege a ion
quan i y and quali y a e low. Coinciding wi h la e ges a ion, cow body condi ion is
likely nega i ely a ec ed and may be e lec ed in lowe cal su i al. Wi h ac o s(s)
causing cu en 18.4% cal es un esol ed, he chance o ha alue o acili a e s abili y
o g ow h o he Manii soq muskox popula ion is deba able.
Ci cums ances o win e ha es ing in he pas decade may ha e esul ed in a hun ing
p essu e ha was no compa ible wi h sus ainable use o his enewable esou ce. The
56
combined esul s om pas coun s, densi ies, ha es s, and cal pe cen ages p esen he
possibili y o some muskox popula ion decline al eady ha ing occu ed. I ue, he
Manii soq muskox popula ion may in he pas ha e been la ge han he cu en 2018
es ima es. The e is no indica ion ha abundance g ew since 2010. We sugges u u e
popula ion decline is possible and may al eady be in p og ess. Gi en he abo e, a
u u e s able o g owing Manii soq muskox popula ion is no ce ain.
This epo ’s popula ion size es ima es, speci ically o Manii soq muskoxen, a e la ge
han any p e ious es ima e o muskox popula ions anywhe e in G eenland, while he
Manii soq muskox DS densi y is much highe han elsewhe e in he A c ic. I
conside ing only ele a ions <400 m ha densi y becomes e en g ea e . I is well known
ha high animal densi y can inc ease exposu e o indi iduals o in ec ious pa hogens
(diseases, pa asi es). Thus, popula ion g ow h is pe haps no o be ecommended gi en
possible densi y-dependen in luences a cu en popula ion size. Whe he he 2018
Manii soq muskox popula ion size (ca. 19.000) and densi y (ca. 2.65/km2) o he
Manii soq muskox ha es managemen a ea (Angujaa o iup sub-a ea) a e wi hin he
cu en he bi o e ca ying-capaci y o he pas u e/ ange emains o be seen.
Acknowledgemen s
This p ojec was inanced p ima ily by he G eenland Go e nmen and o he wise by
he G eenland Ins i u e o Na u al Resou ces, Nuuk G eenland. G a e ul hanks go o
Ai G eenland Cha e and hei helicop e pilo Kå e Be li o his sa e lying. Thanks
also o he G eenland Associa ion o P o essional Hun e s (KNAPK) o p o iding an
expe ienced obse e , excellen a spo ing animals despi e poo de ec ion condi ions.
We hank Ka ine Raund up and Rikke Guldbo g Hansen o e iew o he manusc ip .
Li e a u e ci ed
Ande son M. & Fe gusen M. 2016. Mo emen s and habi a use o muskoxen (O ibos moscha us) on
Ba hu s , Co nwallis and De on Islands, 2003.2006. S a us Repo 2016-08, Nuna u Depa men
o En i onmen , Wildli e Resea ch Sec ion, Igloolik, NU. 112 pp.
Adamczewski J.Z. & Flood P.F. 1997. Seasonal pa e ns in body composi ion and ep oduc ion o emale
muskoxen (O ibos moscha us). J. Zool. Lond. 242: 245-269.
Adamczewski J.Z., Fa gey P.J., Laa eld B., Gunn A. & Flood P.F. 1998. The in luence o a ness on he
likelihood o ea ly-win e p egnancy in muskoxen (O ibos moscha us). The iogenology. 50: 605-614.
Be ge ud A.T., Lu ich S.N. & Camps L. 2008. The e u n o ca ibou o Unga a. McGill-Queen’s
Uni e si y P ess, Mon eal & Kings on / London / I hica. 586 pp.
57
Beume L.T, an Bees F.M., S el ig M. & Schmid N.M. 2019. Spa io empo al dynamics in habi a
sui abili y o a la ge A c ic he bi o e: En i onmen al he e ogenei y is key o a seden a y li es yle.
Global Ecology and Conse a ion 18 (2019) e00647. h p://doi.o g/10.1016/j.gecco.2019.e00647
B ewe M.J., Bu le A. & Cooksley S.L. 2016. The ela i e pe pe o mance AIC, AICc and BIC in he
p esence o unobse ed he e ogenei y. Me hods in Ecology and E olu ion 7(6): 679–692.
Buckland S.T. 1992. Fi ing densi y unc ions wi h polynomials. Applied S a is ics 41: 63-76.
Buckland S.T., Ande son D.R., Bu nham K.P. & Laake J.L. 1993. Dis ance Sampling: Es ima ing
Abundance o Biological Popula ions. Sp inge .
Buckland S.T., Ande son D.R., Bu nham K.P., Laake J.L., Bo che s D.L. & Thomas L. 2001. In oduc ion
o Dis ance Sampling. Ox o d: Ox o d Uni e si y P ess.
Buckland S.T., Ande son D.R., Bu nham K.P., Laake J.L.,
Bo che s D.L. & Thomas
L. 2004. Ad anced
Dis ance Sampling: Es ima ing abundance o biological popula ions. Ox o d Uni e si y P ess.
Buckland S.T., Rexs ad E.A., Ma ques T.A. & Oedeko en C.S. 2015. Dis ance Sampling: Me hods and
Applica ions. Sp inge .
Co eia I.J.F. 2020. Es ima ing ca ibou abundance in Wes G eenland using dis ance sampling me hods.
MSc. Thesis. Uni e si y o Lisbon, Po ugal. 63 pp.
Cou u ie S., Dale A., Wood B. & Snook J. 2018. Resul s o a Sp ing 2017 ae ial su ey o he To nga
Moun ains Ca ibou He d. Technical epo , To nga Wildli e, Plan s and Fishe ies Sec e a ia
.
C ê e M., Huo J., Naul R. & Pa enaude R. 1993. Rep oduc ion, g ow h and body composi ion o Ri iè e
Geo ge ca ibou in cap i i y. A c ic 46(3): 189-196. DOI: 10.14430/a c ic1343
Cuyle C. 2020. Biologisk ådgi ning o ensdy - og moskusokse angs i 2020. Ad iso y documen
p epa ed o he Di ec o a e o Fishing, Hun ing and Ag icul u e, 15 May 2020. Pinngo i ale i ik –
G eenland Ins i u e o Na u al Resou ces, Nuuk, J. No. GN 40-59/40-00-02-42. 4 pp (Supplemen a y
ma e ials 26 pp).
Cuyle C., Landa A., Wi ing, L., Rosing M., Linnell, J. & Loison, A. 2001. Rådgi ning o moskusokse-
bes anden i Kange lussuaq, egion No d/A annaa i 2002, 2003 og 2004. Rappo il Di ek o a e o
Miljø og Na u . 13 No embe 2001. G eenland Ins i u e o Na u al Resou ces. J. No. 28.21.21./b e
n . 01168. 14 pp.
Cuyle C., Ma ques T.A., Co eia I.J.F. A onso B.C., Jensen A., Hegelund P. & Wagnhol J. 2021. 2018
s a us Kange lussuaq-Sisimiu ca ibou, Wes G eenland. Pinngo i ale i ik – G eenland Ins i u e o
Na u al Resou ces. Technical Repo No. 117. 79 pp.
Cuyle C., Nymand J., Jensen A. & Mølgaa d H.S. 2016. 2012 s a us o wo Wes G eenland ca ibou
popula ions, 1) Ame alik, 2) Qeqe a sua siaa . G eenland Ins i u e o Na u al Resou ces Technical
Repo No. 98, 179 pp.
Cuyle C. & Raund up K. 2014. S a på spø gsmål angående ensdy - og moskusokse angs 2014/2015.
Ad iso y documen p epa ed o he Di ec o a e o Fishing, Hun ing and Ag icul u e. G eenland
Ins i u e o Na u al Resou ces, Nuuk. 02 May, 2014. J. N . 40.00.01.42/14, 3 pp.
Cuyle L.C., Rosing M., Egede J., Hein ich R. & Mølgaa d H. 2005. S a us o wo Wes G eenland ca ibou
popula ions; 1) Akia-Manii soq, 2) Kange lussuaq-Sisimiu . Pinngo i ale i ik – G eenland Ins i u e
o Na u al Resou ces. Technical Repo No. 61. Pa I-II, 64+44 pp.
Cuyle C., Rosing M., Linnell J.D.C., Loison A., Inge sle T. & Landa A. 2002. S a us o he
Kange lussuaq-Sisimiu ca ibou popula ion (Rangi e a andus g oenlandicus) in 2000, Wes
64
𝑃
desc ibed abo e assumes ha e e y indi idual o in e es is de ec ed (Mille e al.
2016). F equen ly, his assump ion canno be me , speci ically i among he indi iduals
o in e es he e a e animals impossible o obse e owing o low sigh abili y. Se e al
ac o s cause low sigh abili y, including opog aphical ba ie s, wea he condi ions,
g ound su ace condi ions and many o he s ela ed o obse e aining and su ey
design. The p opo ion o indi iduals ha we e no de ec ed can be es ima ed using he
de ec ion unc ion i ed o he obse ed dis ances (Thomas e al. 2002). Once his
p opo ion is es ima ed, i can be conside ed o ob ain mo e accu a e es ima es and
hen, an ex apola ion o a wide egion can be done simila ly as shown in Equa ion
(2).
In Dis ance Sampling, his p opo ion o de ec ed objec s in he a ea 𝑎 is de ined as he
p obabili y o de ec ion, 𝑃𝑎. The e o e, a densi y es ima e can be ob ained as pe
Equa ion (1) by adjus ing 𝑛𝑝𝑙𝑜𝑡 by 𝑃𝑎, i.e., by co ec ing he de ec ions o hose ha we e
missed. Since he la e canno be known, in gene al, an es ima e mus be also ob ained,
hus
Equa ion
(3)
whe e
𝑃
󰆹
a
is an es ima e o
𝑃
𝑎
ob ained om he dis ance da a, and
𝑎
is he a ea o he
sampled egion. Usually 𝑎 = 2𝑤𝐿, wi h 𝑤 as he unca ion dis ance, o bo h sides o
he cen eline, and he o al ansec leng h
𝐿 = ∑𝑙
𝑗
𝑘
𝑗=1
, whe e l is he leng h o ansec
j. Abundance can be de e mined using a easoning analogous o ha abo e (Equa ion
2). The unca ion dis ance is de ined as he dis ance beyond which dis ances a e no
eco ded. This can be de ined in he ield o a he analysis s age.
The coe icien o a ia ion o 𝐷
,
c
𝑣
(𝐷
), is ela ed wi h wo andom componen s
e e ed abo e, encoun e a e
(
𝑛
𝑝𝑙𝑜𝑡
/𝐿
),
and
𝑃
󰆹
a
, plus a hi d one ha is he es ima e o
he expec ed size o de ec ed clus e s
(
𝐸
󰆹
(
𝑠)
).
Assuming independence be ween he e,
he o me is gi en by
Equa ion
(4)
An app oxima ion o he s anda d e o o 𝐷
,
𝑠𝑒(
𝐷

), is de ined as
Equa ion
(5)

65
Once hese a e ob ained, an app oxima e 100(1 − 𝛼)% con idence in e al (CI) can
be de e mined by
Equa ion
(6)
Whe e is he quan ile o he N(0,1) dis ibu ion
1.96 o a 95% con idence in e al). Howe e , he dis ibu ion o he 𝐷
is posi i ely
skewed, hus an in e al assuming ha 𝐷
 is log-no mally dis ibu ed has be e
co e age. Acco ding wi h Buckland e al. (2015), a 100(1-alpha)% con idence in e al
can be gi en by
Equa ion
(7)
whe e
Equa ion
(8)
and
Equa ion
(9)
Fo u he de ails see Buckland e al. (2001) and Buckland e al. (2015).
P obabili y o de ec ion
Gi en he abo e, he p obabili y o de ec ing an objec , gi ing ha i is wi hin he
a ea
co
e ed
b
y
he
ansec s,
𝑃

a
,
needs
o
b
e
es ima ed.
F
o
his
p o
jec ,
he
ob
jec
o
in
e es
consis s in muskox g oups.
To illus a e he impo ance o his p obabili y, conside ha an obse e walks ac oss a
la ge pa ch o und a and de ec s 8 muskoxen (Fig. 24). While discussing wi h he local
biologis , and conside ing he biologis ’s expe ience, he/she will s a e ha , on a e age, only
one hi d o all muskoxen p esen a e de ec ed (i.e.
,
𝑃

a
= 1/3) meaning ha p obably he e
we e a ound 24 muskoxen wi hin ha pa ch
o und a and 16 ha e been missed. Tha is
whe e Dis ance Sampling is use ul, since i allows a igo ous amewo k o he
es ima ion o
P
a
and hen an es ima e o abundance can be ob ained as shown in
Equa ion (3).
Dis ance Sampling me hods
The de ec ion unc ion, 𝑔(𝑦), desc ibes he p obabili y o de ec ing an objec o in e es
gi en ha i is a a dis ance 𝑦, om he cen eline (also known as 0-line), hus being a
non-inc easing unc ion o 𝑦 (Buckland e al. 2015).
66
Figu e 24.
Example o a pa ch o und a wi h he ansec in he middle. Blue do s ep esen eigh obse ed muskoxen,
while o ange do s ep esen he 16 unde ec ed ones. The lines pe pendicula
o he ansec ep esen he eco ded
dis ances.
Fo line ansec s, 𝑦 is he pe pendicula dis ance om he 0-line o he de ec ed objec .
Wi hin Dis ance Sampling me hods, he p obabili y o de ec ion is explained ecu ing
o hese obse ed dis ances (Buckland e al. 2001). Some imes co a ia es may be added
o explain hei ela ionship wi h he de ec ion p obabili y. In his si ua ion, we a e
wi hin he Mul iple Co a ia e Dis ance Sampling (MCDS) amewo k (Buckland e al.
2001).
Con en ional Dis ance Sampling
Con en ional Dis ance Sampling (CDS) occu s when no addi ional co a ia es a e
added o he model. Once he de ec ion unc ion is es ima ed
,
𝑃

a
can be ob ained ia he
ollowing equa ion
Equa ion (10)
whe e 𝜋(𝑦)= 1
𝜔 and, he e o e, used o es ima e densi y using Equa ion (3). Fo 𝑔(𝑦) i
is also speci ied a lexible semi-pa ame ic model, composed by a key unc ion and
some addi ional se ies expansions, known as adjus men e ms, and hei pa ame e s
a e es ima ed (Ma ques e al. 2007).
67
To ob ain obus es ima es o densi y, lexible models o
𝑔(𝑦)
a e needed wi h he o m
(Buckland e al. 2001)
Equa ion (11)
whe e
𝑘(𝑦)
is he pa ame ic key unc ion and
𝑠(𝑦)
ep esen s he addi ional
adjus men e ms (Table 15).
Table 15. Commonly used key unc ions and se ies expansions o he de ec ion unc ion. Adap ed om Buckland
(2001).
The uni o m key unc ion has no pa ame e s, while he hal -no mal and he haza d- a e
unc ions include a scale pa ame e , 𝜎, which de e mines he a e a which he unc ion
dec eases wi h inc easing dis ance (Fig. 25). Fu he mo e, he haza d- a e unc ion also
includes a shape pa ame e
,
𝑏
,
ha p o ides g ea e lexibili y o his unc ion
compa ing o he o he s (Buckland e al. 2001).
I is no always necessa y o include adjus men e ms, and in such cases, hese models
a e e e ed o as “key only” models. When he key unc ions a e no enough o i ing
𝑔(𝑦)
,
some se ies expansions e ms may be added o modi y i s shape (Fig. 26). These
e ms can be ei he cosine, simple polynomial o He mi e polynomial (Table 15).
I is impo an o no e ha hese adjus men e ms do no depend di ec ly on 𝑦 bu on
𝑦𝑠 which is a scaled alue o
𝑦
,
whe e
𝑦
𝑠
=
𝑦
𝜔
wi h
𝜔
being he unca ion dis ance. This
allows independence be ween he shape o he se ies expansion and he uni s used o
𝑦
(Ma ques e al. 2007).
68
Figu e 25. Hal -no mal ( op ow) and haza d- a e (bo om ow) de ec ion unc ions wi hou adjus men s, a ying scale (σ) and, only o haza d- a e, shape (b) pa ame e s.
Values es ed a e p esen ed abo e he plo s. On he op ow om le o igh , he s udy species becomes mo e de ec able (highe p obabili y o de ec ion a la ge dis ances).
The bo om ows show he haza d- a e model’s mo e p onounced shoulde . Adap ed om Buckland e al. (2001).
69
Figu e 26. Possible shapes o he de ec ion unc ion when cosine adjus men s a e included o hal -no mal and haza d- a e models. Adap ed om Buckland e al. (2001).

70
Righ unca ion o he da a, o he emo al o he la ges dis ances, is a common
p ocedu e ha aids model i ing. Some p ecision migh be los wi h unca ion;
howe e , i is usually sligh . On he o he hand, p ecision is inc eased since he da a is
easie o model and, consequen ly, ewe pa ame e s and adjus men e ms a e
equi ed o model he de ec ion unc ion (Cou u ie e al. 2018).
Mul iple Co a ia e Dis ance Sampling
CDS me hods can be ex ended o MCDS, so ha
𝑔(𝑦)
is modelled as a unc ion no only o
dis ance, bu also o a ec o o
𝐽
addi ional co a ia es o each o he 𝑛 objec s o
in e es ,
z
i
=
z
𝑖1
, ...,
z
𝑖𝐽
,
𝑖 = 1, ..., 𝑛
. Acco dingly, he unc ion ha
desc ibes
he
p obabili y
o
de ec ion a a gi en dis ance, is ep esen ed by
𝑔(𝑦,
z
)
. These
addi ional
co a ia es
can
ei he be disc e e o con inuous, such as obse e and g oup size, and a e assumed o a ec
only he scale, 𝜎, o he de ec ion unc ion (Ma ques e al. 2007; Mille e al. 2016). Fo
line ansec s,
𝑃 (
z
i)
, i.e., he p obabili y o de ec ing he
𝑖
- h objec o in e es gi en i s
espec i e ec o o co a ia es
z
i
can be es ima ed using he o mula p esen ed in Equa ion
(12).
Equa ion (12)
wi h 𝜋(𝑦)= 1
𝜔. Conside ing he h ee key unc ions p e iously p esen ed, only he
uni o m key is excluded om MCDS since i does no ha e a scale pa ame e . Hal -
no mal and haza d- a e unc ions can ha e hei scale pa ame e w i en as a unc ion
o he co a ia e alues as
Equa ion (13)
Whe e 𝛽0and all he 𝛽𝑗’s a e he J + 1 coe icien s o be es ima ed wi h J being he o al
numbe o co a ia es. The es ima ion o he pa ame e s o bo h CDS and MCDS is
ypically done ia maximum likelihood (Ma ques e al. 2007).
Once he de ec ion unc ion is es ima ed, acco ding wi h (Buckland e al. 2004), densi y
can be es ima ed as
Equa ion (14)
whe e
𝑎
is he o al a ea su eyed
,
𝑃
(
z
i
)
is he es ima ed p obabili y o de ec ing he
𝑖
- h
objec o in e es gi en i s espec i e ec o o co a ia es zi.
71
Finally, Ma ques e al. (2007) s a es ha MCDS me hods po en ially o e imp o ed
in e ence in ou si ua ions, when compa ing o CDS me hods:
1.
when a subse o da a is used o es ima e densi y, e.g., by s a a, whe e his
in o ma ion can be in oduced as a ac o co a ia e. In CDS, he s a egy is mo e
complex, ei he o es ima e
𝑃
𝑎
o each s a um and hus, s a um-le el es ima es
o densi y o o use a global es ima e o he p obabili y o de ec ion, bu his
second in oduces bias, o example, i one s a um a ou s he animals when
compa ed o o he s a a which uses ewe pa ame e s han a ully s a i ied
de ec ion unc ion model;
2.
whe e pooling obus ness does no hold o CDS analyses, e.g., when su ey
in ensi y a ies acco ding wi h p e-de ined s a a o inc ease efficiency, o when
he de ec ion p obabili y aces ex eme he e ogenei y due o di e en objec
habi a s o beha io s, o example, showy males con as ing wi h c yp ic emales
in animal su eys;
3.
educes he a iance o densi y es ima es by modelling he he e ogenei y in he
de ec ion unc ion;
4.
i he e a e co a ia es o in e es o be included in he model.
Model selec ion
Since he es ima o o densi y is closely linked o he de ec ion unc ion, i is o c i ical
impo ance o selec models o he de ec ion unc ion ca e ully. Th ee p ope ies
desi ed o a model o 𝑔(𝑦) a e, in o de o impo ance, model obus ness, a shape
c i e ion and es ima o efficiency (Buckland e al. 2001, 2015; Mille e al. 2016).
The mos impo an p ope y o a model o he de ec ion unc ion is model obus ness.
Acco ding wi h Buckland e al. (2001, 2015), his means ha he model is a gene al,
lexible unc ion ha can ake a a ie y o plausible shapes o he de ec ion unc ion.
The concep o pooling obus ness is also included he e. Models o
𝑔(𝑦)
a e pooling
obus i he da a can be pooled o e many ac o s ha a ec de ec ion p obabili y and
s ill yield a eliable es ima e o densi y. A model is pooling obus i , o example, a
s a i ied es ima ion o densi y,
𝐷

s
,
and a pooled es ima ion o densi y
,
𝐷

p, a e
app oxima ely he same. In he i s scena io, he da a is s a i ied by ac o s, such as
obse e o habi a ype, and an es ima e o densi y in each s a um is made. Then
hese es ima es a e combined in o
𝐷

a
, an a e age densi y es ima e. In he second
scena io, all da a could be pooled, ega dless o any s a i ica ion, and a single es ima e
compu ed,
𝐷

p. A model is pooling obus i
𝐷

a
≈
𝐷
p.
72
Acco ding o Buckland e al. (2001), he shape c i e ion consis s in he ac ha he
de ec ion unc ion should ha e a ‘shoulde ’ nea he line (Fig. 27), i.e., de ec ion
emains nea ly ce ain a small dis ances om he sampling uni ’s ack line
(𝑔
′
(0)
=
0)
.
This allows he eliable es ima ion o objec densi y (Thomas e al. 2002). Gene ally,
good models o
𝑔(𝑦)
will sa is y he shape c i e ion nea he ze o-dis ance ack line,
which is especially impo an in he analysis o da a whe e some heaping a ze o
dis ance is suspec ed.
Figu e 27.
A good model o he de ec ion unc ion should ha e a shoulde , wi h p obabili y
o de ec ion s aying a o
close o one a sho dis ances om he cen eline o poin . A la ge
dis ances, i should all away smoo hly. The
unca ion dis ance
𝜔
co esponds o he s ip
hal -wid h ( o Line T ansec Dis ance Sampling). Adap ed om
Buckland e al. (2001).
Es ima o efficiency is he hi d mos impo an p ope y (Buckland e al. 2001), which
means ha i is desi able o selec a model ha p o ides es ima es ha a e ela i ely
p ecise, i.e., ha ha e small a iance. This p ope y is o bene i only o models ha
a e model obus and ha e a shoulde nea ze o dis ance, o he wise he es ima ion
migh be p ecise bu biased.
Besides hese h ee c i e ia, he model should be a mono onic unc ion o dis ance om
he line, ha is, he p obabili y o de ec ion a a gi en dis ance canno be g ea e han
he p obabili y o de ec ion a any smalle dis ance (Fig. 27) (Buckland e al. 2001).
The e is no ixed s anda d me hod o selec he bes i ing model, i.e., choosing he
mos app op ia e key unc ion and se ies expansion (Ma ques e al. 2007). I is usually
done by applying he Akaike’s In o ma ion C i e ion (AIC), Kolmogo o -Smi no es ,
C amé - on Mises es and he 𝜒2 Goodness-o -Fi es (GOF es ). The likelihood a io
es can also be used bu , since i is only applicable o nes ed models, AIC is he
ecommended me hod (Ma ques e al. 2007). A p ope model should be simple wi h an
adequa e i wi hou o e i ing he da a.
73
Akaike In o ma ion C i e ion
The ela i e i o al e na i e models may be e alua ed ecu ing o AIC, o AICc, in
case o small samples, p o iding a small sample bias co ec ion (Buckland e al. 2001).
These c i e ia can be de e mined as ollows
Equa ion (15)
Equa ion (16)
whe e
ℒ
is he likelihood unc ion,
𝑞
is he numbe o es ima ed pa ame e s in he
model, and
𝑛
is he sample size. This measu e p o ides a ade-o be ween bias and
a iance. AIC includes wo e ms, one ela ed wi h he i ed model, and he o he
wo king as a penal y conside ing he excess o pa ame e s in he model (B ewe e al.
2016).
Kolmogo o -Smi no es
The Kolmogo o –Smi no es is one o he es s ha can be applied o he de ec ion
unc ion o assess model i (Buckland e al. 2004). This es is only applicable o
con inuous da a, being p e e able o he 𝜒2 GOF es o MCDS me hods.
Conside ing he cumula i e dis ibu ion unc ion (c.d. .)
𝐹 (𝑥) = 𝑃 (𝑋 ≤ 𝑥)
and he
empi ical c.d. . (e.d. .)
𝑆(𝑥)
,
he null hypo hesis o be es ed is
𝐻
0
∶ 𝐹 (𝑥) = 𝐹
0
(𝑥), ∀𝑥
.
The
al e na i e hypo hesis s a es ha bo h unc ions di e o a leas some alue o
𝑥
.
In
p ac ice,
𝐹 (𝑥)
is eplaced by i s es ima e, and 𝐻0 s a es ha he assumed model is he
ue model o he da a (Buckland e al. 2004). The la ges absolu e di e ence be ween
𝐹
󰆹
(𝑥)
and
𝑆
(𝑥)
,
deno ed
𝐷𝑛, is he es s a is ic (Gibbons & Chak abo i 2011). The
co esponding 𝑝- alue can be app oxima ed by
Equa ion (17)
C amé - on Mises es
Simila o he Kolmogo o -Smi no es , he C amé - on Mises es sha es he same
null hypo hesis and basis on di e ences be ween c.d. . and e.d. . Howe e , ins ead o
conside ing only he la ges di e ence be ween he wo unc ions, his es is based on
hei en i e ange (Buckland e al. 2004). The es s a is ic can be gi en by
Equa ion (18)
80
Figu e 30. G oup o 23 muskoxen, which may ha e included 1-2 cal es (age ca. 10-mon hs).
Figu e 31. G oup o 38 muskoxen, which may ha e included 7-8 cal es (age ca. 10-mon hs). Eigh
adul s kep hemsel es somewha sepa a ed om he main g oup.

81
Appendix 6
Maps illus a ing ele a ion, aspec , and slope in he No h egion
Ele a ion
Figu e 32. Map illus a ing ele a ion in he su eyed a ea. Legend colou codes he
ele a ion.
Aspec
Figu e 33. Map illus a ing aspec in he su eyed a ea. Legend colou codes he
compass headings o aspec .
82
Slope
Figu e 34. Map illus a ing slope in he su eyed a ea. Legend colou codes he angle
o slope.
83
Appendix 7
Angujaa o iup sub-a ea, Ma ch 2018: Pho os o su ey condi ions and
snow co e obse ed. Pho os by C. Cuyle
Figu e 35. View om su ace o lake ice du ing a pause be ween su ey lines, illus a ing lack o snow common
in he Angujaa o iup sub-a ea. Also illus a ing la -ligh (no shadows).
Figu e 36. View as helicop e lew o e su ace o lake ice, illus a ing he ypical spa se snow co e on he
g ound ahead, line ansec 18 Angujaa o iup sub-a ea. Also illus a ing la -ligh (no shadows).
84
Figu e 37. View while lying line ansec 17, illus a ing spa se snow-co e common o he Angujaa o iup
sub-a ea. Also illus a ing la -ligh (no shadows).
Figu e 38. O e iew o he no h side o Tase suaq Lake (la ge la whi e su ace on le : lake ele a ion 312 m),
illus a ing la -ligh (no shadows) and spa se snow-co e common o he Angujaa o iup sub-a ea.
85
Figu e 39. O e iews illus a ing la -ligh (no shadows) and he windswep e ain wi h spa se snow-co e
common o he Angujaa o iup sub-a ea. Ho izon al whi e su aces a e lake o jo d ice.

86
Figu e 40. Minimal Ma ch snow co e ypical o lowland alleys in Angujaa o iup sub-a ea. View NW in o
he Manii su A annguisa Kuussua i e alley om line ansec 19. Valley bo om ele a ion <100 m.
Figu e 41. Obse a ion pla o m o ae ial su ey: AS350 helicop e on e minal mo aine wi h G eenland Ice
Cap in backg ound, illus a ing windswep ocky e ain wi h spa se snow-co e .
87
Appendix 8
Hun ing seasons, quo as and epo ed ha es s, p ima ily o
he Manii soq muskox managemen a ea.
Table 16. Muskox hun ing seasons and quo as o he Sisimiu and Manii soq muskox managemen a eas in he
2015-2020 pe iod (comme cial and ec ea ional combined) (Nuka M. Lund pe s comm (Minis y o Fishe ies &
Hun ing: APN). T ophy hun ing seasons and quo as no included. Manii soq muskox managemen a ea
co esponds o he Angujaa o iup sub-a ea su eyed by helicop e in Ma ch 2018, and he Sisimiu
managemen a ea co esponds o Sisimiu sub-a ea su eyed.
Muskox
managemen
a ea
Hun ing
a ea1
WINTER HUNT
AUTUMN HUNT
Season
Quo a
Season
Quo a
Manii soq
2015
1
10 Janua y – 10 Ma ch
Open
1 Augus – 15 Oc obe
Open2
2
Closed
0
Closed
0
3
10 Janua y – 10 Ma ch
Open
1 Augus – 15 Oc obe
Open
2016
1
10 Janua y – 10 Ma ch
Open
1 Augus – 15 Oc obe
Open
2
Closed
0
Closed
0
3
10 Janua y – 10 Ma ch
Open
1 Augus – 15 Oc obe
Open
2017
1
10 Janua y – 10 Ma ch
Open
1 Augus – 30 Sep embe
Open
2
22 Janua y – 31 Janua y
400
Closed
0
3
10 Janua y – 10 Ma ch
800
1 Augus – 30 Sep embe
Open
2018
1
10 Janua y – 15 Feb ua y
Open
1 Augus – 15 Oc obe
Open
2
Closed
0
Closed
0
3
10 Janua y – 15 Feb ua y
800
1 Augus – 15 Oc obe
Open
2019
1
25 Janua y – 15 Feb ua y
Open
1 Augus – 15 Oc obe
Open
2
Closed
0
Closed
0
3
25 Janua y – 15 Feb ua y
Open
1 Augus – 15 Oc obe
Open
4
25 Janua y – 15 Feb ua y
Open
1 Augus – 15 Oc obe
Open
2020
1
25 Janua y – 14 Feb ua y
100
1 Augus – 15 Oc obe
Open
2
Closed
0
1 Augus – 15 Oc obe
Open
3
25 Janua y – 14 Feb ua y
1200
1 Augus – 15 Oc obe
Open
4
25 Janua y – 14 Feb ua y
1 Augus – 15 Oc obe
Open
Sisimiu
2015
---
10 Janua y – 10 Ma ch
520
1 Augus – 15 Oc obe
400
2016
---
10 Janua y – 10 Ma ch
520
1 Augus – 15 Oc obe
400
2017
---
Closed
0
1 Augus – 15 Oc obe
400
2018
---
Closed
0
1 Augus – 15 Oc obe
400
2019
---
Closed
0
1 Augus – 15 Oc obe
400
2020
---
Closed
0
1 Augus – 15 Oc obe
400
1 See Figu es 40-41 o he G eenland go e nmen ’s muskox managemen hun ing a eas o Manii soq (Angujaa o iup sub-a ea su eyed).
2 Open = No limi on he numbe o muskoxen ha may be ha es ed.
88
Figu e 42. Map o he 2010-2019 pe iod illus a ing he Go e nmen o G eenland’s Manii soq muskox
ha es managemen a ea wi h speci ic hun ing a eas o egula ing ha es o he Manii soq muskox
popula ion. Hun ing a eas a e sepa a ed by s ippled lines and labelled wi h “Pinia ik/Jag om åde” and a
numbe . A eas whe e hun ing is p ohibi ed a e ma ked wi h diagonal lines and/o he label,
“Pinia igeqqusaangi soq/Jag i om åde”. The la ges encompasses he ai po own o Kange lussuaq and
nea icini y including Lake Fe guson, he beginnings o Ø kendalen and he Sand lug dalen i e alley up o
and including he Inland Ice Cap. The smalle and iangula a ea encompasses he inne po ion o he alley,
A nange nup Kuua (Pa adisdalen) and i s side alleys (Naalakke suisu 2019).
89
Figu e 43. Map illus a ing he en i e Manii soq muskox ha es managemen a ea wi h he cu en hun ing
sub-a eas o egula ing ha es o he Manii soq muskox popula ion in he Qeqqa a municipali y. The hun ing
sub-a eas a e de ined by he Go e nmen o G eenland. Hun ing a eas a e labelled wi h “Pinia ik” and a
numbe . Hun ing a ea 4, es ablished in 2018, was i s implemen ed om summe 2019 (Naalakke suisu 2019,
bilag 4). A eas whe e hun ing is p ohibi ed a e ei he ma ked wi h diagonal lines and he label,
“Pinia igeqqusaangi soq”, o a e a solid ed colou . Since 1984, hun ing has been p ohibi ed in he inne
po ion o he alley, A nange nup Kuua (Pa adisdalen), which is indica ed by he solid ed wi hin Pinia ik 4.
(Naalakke suisu 2020).
96
Figu e 3. Muskox popula ions in G eenland. Na i e popula ions (blue) a e in No h and No heas G eenland.
Ingle ield Land muskoxen a e a mix o na i e and ansloca ed s ock. In wes e n G eenland, speci ically om
Cape A holl o Nano alik, all popula ions a e ei he ansloca ed (o ange) o he esul o expansion (g een)
om a neighbou ing ansloca ion. Popula ion bo de s a e app oxima ions and size does no e lec animal
abundance, e.g., G eenland’s la ges muskox popula ion, Manii soq, inhabi s a ela i ely small a ea. Bounda y
o he Na ional Pa k in no heas G eenland is indica ed. Cu en ly, egula ed ha es is pe mi ed on mos
popula ions, howe e hun ing is p ohibi ed in Washing on Land, Na ional Pa k, Nano alik and Nuuk.

97
Manii soq muskox popula ion
In 1963 and 1965, 27 ju enile muskoxen we e ansloca ed om Eas G eenland o he a ea
sou h o he Kange lussuaq (Sønd e S øm jo d) In e na ional ai po in Wes G eenland.
P e iously, his popula ion was commonly called he Kange lussuaq muskoxen (Cuyle &
Wi ing 2004), bu he ea e will be e e ed o as he Manii soq muskoxen as pe APNN
(2019a). All hun ing was p ohibi ed o 25 yea s, i s ha es was 1988, and by 1990 he
muskox popula ion was conside ed well es ablished (Olesen 1993).
Al eady in he 1980’s, he muskoxen clea ly selec ed lowland ele a ions unde 400 m. A ha
ime, his included he lowlands su ounding he Kange lussuaq (Sønd e S øm jo d)
in e na ional ai po and se e al la ge alleys, i.e., Ø kendalen (Bioshy e), Ammalo oq
Lake, and Pa adisedalen (A nanga nup Qoo ua) (Olesen 1993). The o al a ea a ailable o
Manii soq muskoxen is 7,853 km2, (co ec ed a ea om his s udy is 7133 km2), howe e
only ca. 950 km2 is habi a unde 400 m ele a ion (Olesen 1993). Cal p oduc ion indica ed
ha lowland habi a wi hin he 7,853 km2 egion was well sui ed o muskoxen. Al hough
nei he is no mal o muskox popula ions elsewhe e, in he 1970-80’s, cal es accoun ed o
24-28% o he Manii soq muskox popula ion, while many cows p oduced a cal e e y yea
(Roby 1978, Thing e al. 1984, 1987, Olesen 1993).
To manage he Manii soq muskox ha es , APNN subdi ided he egion in o ou hun ing
managemen a eas. Ini ially (2010) he e we e h ee hun ing managemen a eas (Pinia ik 1,
2, 3) bu a ou h was added in win e 2018 and i s implemen ed in summe 2019 (Figu e 4).
Figu e 4. APNN’s ou hun ing a eas (Pinia ik 1, 2, 3, 4) o ha es managemen o he Manii soq muskox
popula ion in Qeqqa a municipali y (APNN 2019b). Top igh and indica ed by ed diagonal lines is he hun ing-
p ohibi ed a ea (Pinia igeqqusaanngi soq) ha su ounds he Kange lussuaq (Sønd e S øm jo d) in e na ional
ai po and includes Sand lug dalen Valley and Isunngua a eas. Hun ing has always been p ohibi ed in he inne
Pa adise Valley, indica ed by solid ed.
98
Cu en popula ion end, Manii soq muskoxen
A 2018 helicop e su ey p o ided a popula ion size es ima e o ca. 20,334 muskoxen (95%
CI 9,386 – 44,055; SE 6397; CV 0.31) o he Manii soq muskox popula ion (Ma ques 2018).
This numbe canno be compa ed wi h p e ious g ound-based minimum coun s, because i is
he i s and only es ima ion o o al abundance o muskoxen in he en i e a ea sou h o he
Kange lussuaq in e na ional ai po . The high 2018 es ima e p o ides e idence o he
explosi e inc easing end o abundance ha occu ed since 27 indi iduals we e ansloca ed
o he a ea in he 1960’s. The s eep g ow h cu e in he 1980’s and ‘90’s (Figu e 5) suppo s
his. Rega dless, alone he 2018 es ima e canno indica e cu en popula ion end. Ins ead,
pas , ecen , and cu en ends a e indica ed by he minimum coun and cal pe cen age da a
ga he ed o e a 40-yea pe iod (Figu es 5, 6). These sugges he cu en end is s eady
decline in abundance since 2010. The mos ecen minimum coun , in 2020, also suppo s ha
popula ion size and p oduc ion ha e dec eased. The abundance in he 2000-2010 pe iod
migh ha e been g ea e han he 2018 ae ial su ey es ima e o ca. 20,334 muskoxen. Below
ollows a b ie discussion o he minimum coun s, cal p oduc ion, 2018 ae ial su ey,
muskox densi y, and dis u bance.
Minimum coun s 1963 – 2020, Manii soq muskoxen
Since he ini ial elease o 27 muskoxen in 1963-65, g ound-based minimum coun s om he
1980’s o 2020 p o ide indices o abundance. Al hough no popula ion size es ima es, hey
illus a e gene al popula ion ends (Figu e 5). Minimum coun s in his a ea a e usually done
by wo obse e s a elling by skidoo in he Janua y-Ap il pe iod and using binocula s and
elescopes eco ding all muskoxen seen. The 2018 minimum coun was he ac ual numbe o
muskoxen seen du ing a helicop e su ey o ca ibou. Ce ainly, he inconsis en iming,
e o and co e age o he di e se minimum coun s makes compa isons ac oss ime less
eliable. Fo example, he 1977-1995 minimum coun s we e ca ied ou be o e he hun ing
season and co e ed hun ing a eas 1 and 2 (see Figu e 4). Minimum coun s in 2000-2010
occu ed also p io o win e ha es ing, bu ypically co e ed mos o hun ing a eas 1, 2, 4
and no he n po ion o a ea 3. In 2010, lack o snow and sea ice limi ed he minimum coun
o hun ing a eas 1 and 2 and a small po ion o 3, while 4 was impossible o co e .
Meanwhile, win e ha es in luenced minimum coun esul s pos -2014 because coun s we e
du ing o a e he win e hun ing season. Du ing hese coun s, animals had p obably mo ed
o less accessible e ain, which p e en ed hei de ec ion. Fu he mo e, 2014-2020 a ea
co e age was highly a iable and o en much educed. The excep ion being he 2018
minimum coun om helicop e , which had he g ea es e o and co e age o any coun o
da e. Owing o di e en iming and e o , muskox abundance ends should be conside ed
wi h cau ion. Ne e heless, gene al ends a e appa en .
The index o muskox abundance illus a es a gene al apid expansion in he 1980’s and ea ly
1990’s, a possible peak om he mid 1990’s un il 2010, and a decline he ea e (Figu e 5).
Nei he he magni ude no slope o he decline a e likely as s eep as hey appea , since
99
minimum coun s a e 2014 occu ed a e hun ing had al e ed he dis ibu ion and
de ec abili y o he animals. Rega dless, he sugges ed declining end in popula ion size since
2010 is suppo ed by simul aneous declining numbe o muskox g oups obse ed, maximum
g oup size, as well as a ca. 37% educ ion in cal ec ui men (Appendix 2). Fu he mo e,
minimum coun da a om hun ing a ea 2 (Appendix 3), which ecei ed consis en e o and
co e age on all g ound-based coun s, al hough iming a ied, simila ly suppo s an
abundance decline in he Manii soq popula ion since 2010.
Figu e 5. Index o Manii soq muskox abundance since ansloca ion o he egion in 1963-65. Da a a e
minimum coun s o only hose indi iduals obse ed and canno be in e p e ed as es ima es o popula ion size.
All bu 2018 a e g ound coun s. All a e o a ying iming, e o , and a ea co e age. Beginning 2000, all a e
p e-cal ing coun s om ei he win e o la e win e . Those om 2000 o 2010 we e p e-ha es , while hose
2014-2020 we e pos -ha es . In he la e muskoxen a e a oiding lowlands owing o ecen hun ing. The ac ual
numbe o indi idual muskoxen obse ed du ing he Ma ch 2018 helicop e su ey () has been included as a
ype o minimum coun , which e o and co e age we e he mos comp ehensi e ela i e o any o he minimum
coun . (Da a om Roby 1978, Thing e al. 1984, Olesen 1993 and Cuyle unpublished).
Cal p oduc ion 1977-2020, Manii soq muskoxen
Un il a leas 1981 cal pe cen age o Manii soq muskoxen was excellen (Figu e 6) esul ing
in he highes eco ded popula ion g ow h a e, 35%, o any muskox popula ion globally
(Thing 1984, Olesen 1993). Popula ion g ow h o Manii soq muskoxen was abo e he 17-
24% epo ed o a simila ly ansloca ed popula ion in Alaska (Jing o s & Klein 1982) and
he 21% om Queen Maud Gul , Canada (Gunn e al. 1984). Manii soq’s popula ion
inc emen , ca. 30%, con inued almos unaba ed un il 1990 (Olesen 1993), and e en in o
2000-2001 (Cuyle e al. 2001). Beginning in 2002 howe e , cal pe cen ages declined and
ha e no eco e ed o p e-2000 alues (Figu e 6). Cal pe cen age seemingly imp o ed since
he 2010 minimum coun , howe e , he 2014-2020 pe iod in ol ed only pos -ha es coun s.
Resul ing cal pe cen ages may jus be an a i ac o hun e s shoo ing all excep cal es,
he eby educing he o al numbe while cal numbe emained he same, which a i icially
0
1000
2000
3000
4000
5000
6000
1950 1960 1970 1980 1990 2000 2010 2020 2030
Numbe muskoxen obse ed
Yea
100
aised cal pe cen age in he coun s. Speci ically, he 2014 alue is likely un eliable and
a i icially high because he coun was pos -ha es , no-hun ing zones did no ye exis and
he small size o su eyed a ea, which was easily accessible o hun e s. In he 2016-2020
pe iod, a no-hun ing zone was es ablished, and minimum coun e o was concen a ed he e.
Thus, he ca. 15% cal es o his pe iod p o ides he bes cu en expec ed cal pe cen age,
which is ca. 40% below he p e-2000 alues and suppo ed by he ca. 37% known decline in
cal ec ui men (cal es pe 100 cows; Appendix 2).
Figu e 6. Cal pe cen ages in he Manii soq muskox popula ion om 1977 o 2020. All esul s a e om g ound
coun s. Beginning 2000, all esul s a e om win e (p e-cal ing) coun s, 2000-2010 p e-ha es and 2014-2020
pos -ha es . The Ap il 2014 ou lie alue () is likely un ealis ic and a i icially ele a ed, p ima ily owing o he
iming o he coun , which was pos -ha es , and coun e o in ol ed a limi ed a ea easily accessible o
hun e s. Thus, while o al muskox numbe was educed cal numbe emained he same (no sho ), which
a i icially aised cal pe cen age. (Da a om Roby 1978, Thing e al. 1984, Olesen 1993 and Cuyle
unpublished).
The 1980’s and 1990’s we e associa ed wi h apid popula ion g ow h, when he pe cen age o
cal es was usually ca. 25% (Figu e 6) o he o al numbe o muskoxen obse ed. In 1990,
nea Kange lussuaq (Sønd e S øm jo d) in e na ional ai po , Ø kendalen Valley
(Bioshy e/Bios cabin) and Ammalo oq Lake, densi y was almos 2 muskoxen/km2 a
ele a ions unde 400 m, howe e , densi y was g ea e in he mos commonly used alley
a eas, and ye he e we e no obse a ions o densi y-dependen mo ali y i.e. no inc eased cal
mo ali y had been obse ed e en in he a eas o highes densi ies (Olesen 1993). This
sugges s ha o Manii soq muskoxen, a densi y o ca. 2 muskoxen/km2 in lowland habi a
was no impedimen o cal p oduc ion. By win e 2001, densi y g ew o a leas 4-5
muskoxen/km2 in he same a ea (Cuyle e al. 2001). Again, because win e g azing was o en
concen a ed, some alley loca ions had much highe densi ies, e.g., ca. 29 muskoxen/km2
0
5
10
15
20
25
30
1960 1970 1980 1990 2000 2010 2020 2030
Cal pe cen age (%)
Yea
101
nea Bios cabin Ø kendalen (589 muskoxen in 20.25 km2) and ca. 22 muskoxen/km2 a ound
Ammalo oq Lake (876 muskoxen in 40.5 km2) (Cuyle e al. 2001). E en a hose densi ies,
cal pe cen age was ca. 25% and annual popula ion g ow h ca. 30% (Cuyle e al. 2001),
which indica es ha he ca ying capaci y o ha lowland habi a was excellen . Howe e ,
some hing changed in he pe iod 2001-2004, because cal pe cen age dec eased each yea and
his end con inued un il 2010 (Figu e 6).
Typical causes o poo cal p oduc ion include p eda ion, disease, ad e se wea he , and
ma e nal body condi ion. Elsewhe e, la ge p eda o s (Reynolds e al. 2002, Ma qua d-
Pe e sen 1998) and a al disease ou b eaks (Y ehus e al. 2008) can nega i ely a ec cal
pe cen ages. Howe e , o Manii soq muskoxen, la ge p eda o s a e absen and a al disease
has no been obse ed. This lea es poo cow body condi ion and ad e se wea he as p ima y
ac o s. Gi en he ela i e s abili y o he xe ic con inen al clima e o he egion, wea he
condi ions o e en s a e no expec ed o explain he obse ed con inuous decline in cal
p oduc ion in he 2000-2010 pe iod, o he ela i ely s able bu low alues he ea e . This
makes cow body condi ion he likely p ima y ac o . Causes o poo cow body condi ion
would include muskox densi ies oo high o he o age quan i y, quali y, and a ailabili y in
he egion, and excessi e dis u bance p e en ing cows om egaining body ese es in
summe o main aining hose ese es h ough win e .
Simul aneously wi h he declining end o cal pe cen ages, in he pe iod 2001-2004 he e
occu ed a signi ican (p<0.05) educ ion in cow ump a dep h, and p egnancy a es (cows
bea ing a oe us) ell om 74.6% o 61.5% (Cuyle & Wi ing 2004). Poo cow body
condi ion ( a ese es) sugges s he possibili y o densi y-dependen e ec s and/o excessi e
dis u bance. Rega ding he o me , he high densi ies obse ed in 2000 and 2001 did no
appea o change cal pe cen age. Rega ding dis u bance, a leas o he 2001-2004 pe iod
we know ha es s doubled (Cuyle & Wi ing 2004). I may only ha e been coincidence ha
cal % d opped jus as ha es ing inc eased, speci ically win e ha es ing. On he o he hand,
he inc eased dis u bance occasioned by win e hun ing ac i i ies may ha e nega i ely
in luenced cow body condi ion and ul ima ely cal p oduc ion.
Win e hun ing began in 1994 and ini ially ha es dis u bance was minimal. Fo example,
he i s six yea s win e quo as we e small, a e aging 227 muskoxen, o abou hal he
annual quo a. The ha es s in ol ed a small numbe o comme cial hun e s om only he
owns o Sisimiu and Manii soq, hun ing was only in wha a e now e e ed o as hun ing
a eas 1 and 2, and Sisimiu hun e s used almos exclusi ely dog sleds o anspo . Manii soq
hun e s had a go e nmen g an ed dispensa ion o use skidoos. In he pe iod 2000-2010
se e al hings occu ed. The Manii soq win e ha es quo as began a 500 muskoxen and
ose quickly he ea e . The ha es con inued o in ol e a g owing bu s ill limi ed numbe
o comme cial hun e s, s ill mos ly om Sisimiu and Manii soq. Many Sisimiu hun e s
eplaced dog- eams wi h mo o ized ehicles. Spo hun e s we e admi ed. Finally, by he end
o he pe iod a hi d hun ing a ea was added, which mean win e hun ing now occu ed o e
he en i e egion, excep ing Pa adise Valley and he nea ai po a ea. In sha p con as o he
ea ly days, since ca. 2010 and con inuing oday, win e ha es in ol es la ge numbe s o

102
comme cial and spo hun e s, using many and di e se mo o ized ehicles, which pene a e
deeply in o he egion and speci ically all lowlands. Meanwhile, ophy ha es ing has always
been pe mi ed in o Ap il, ending jus as muskox cal ing begins. In he beginning he ophy
ha es in ol ed ew agen s and hun e s. Like he egula ha es ing, howe e , oday he
ophy ha es indus y is g ea ly expanded in ol ing mul iple ac o s and concession dis ic s
co e ing much o he egion.
Muskox cal ing no mally begins in mid-Ap il and las s o pas mid-June (Len 1988).
Dis u bing pa u ien cows in la e ges a ion is no expec ed o be compa ible wi h good cal
p oduc ion. The cu en Janua y-Feb ua y ha es likely dis u bs no mal win e dis ibu ion,
as well as o aging and umina ion ime o muskoxen. Likely his also applies o he Ma ch-
Ap il ophy ha es . Each win e since 2000, hun ing appea s o cause Manii soq muskoxen
o a oid lowlands and seek ehicle-inaccessible e ain a ela i ely high ele a ions (Cuyle
unpublished). A oiding a eas accessible o hun e s was appa en du ing he Ma ch 2018
ae ial su ey (Figu e 7). Al hough 2 weeks a e he end o he egula 2018 win e hun ing
season, mos muskoxen had no e u ned o lowlands, and he hun ing o ophy bulls was s ill
on-going a he ime. A oiding essen ial lowland habi a in win e likely has a nega i e e ec
on he win e body condi ion o p egnan cows. Thus, win e ha es may ha e in luenced he
s eep decline in cal pe cen age in he 2000-2010 pe iod (Figu e 6).
Finally, he 2020 minimum coun in ol ed only hun ing a eas 1 and 2, which con ain he 950
km2 unde 400 m ele a ion desc ibed in Olesen (1993). The 2020 pos -ha es coun yielded
16.9% cal es, and a densi y o 0.9 muskoxen/km2 in Olesen’s lowlands (Cuyle & Mølgaa d
unpublished). Bo h a e well below he 2000 alues: 26% and 4-5 muskoxen/km2 espec i ely
(Cuyle e al. 2001).
Al hough elsewhe e in he A c ic muskox densi ies a e ypically lowe , o Manii soq
muskoxen he ela i ely high 2000-2001 densi ies o 4-5 muskoxen/km2 in lowland habi a
(unde 400 m ele a ion) did no appea o in luence cal pe cen age, which emained high.
This sugges s ha oday’s 2020 pos -ha es densi y o 0.9 muskoxen/km2 is unlikely o be a
limi ing ac o o popula ion g ow h, p e equisi e on ha cows can u ilize lowland habi a as
hey did p io o and including 2000-2001. Ins ead, ha es may be he p ima y limi ing
ac o , speci ically i s associa ed dis u bance, which causes muskoxen o a oid lowland
habi a , al hough o al ca ch is likely also impo an .
Imp o ing cow body condi ion would p omo e cal p oduc ion. S a egies would include
educing he du a ion o human dis u bance. Sho e hun ing season(s) could inc ease
undis u bed o aging in op imal habi a . A sho e summe hun ing season, e.g., beginning
a e mid-Augus , would likely acili a e ebuilding cow body condi ion los du ing he win e
and due o cal ing and lac a ion, esul ing in mo e cows achie ing he p e equisi e 22% a
o o ula ion du ing he u . Rega ding he win e ha es , a sho e win e season ending well
be o e pa u i ion would likely acili a e success ul ges a ion and bi hing. Fu he mo e, he
magni ude o dis u bance associa ed wi h an o he wise sho e seasons migh be minimized i
ewe pe sons and mo o ized ehicles we e in ol ed han is cu en ly no mal.
103
2018 su ey Manii soq muskoxen
Ma ch 2018 was he i s e e sys ema ic ansec dis ance sampling ae ial su ey o
Manii soq muskox abundance and dis ibu ion in he egion sou h o he Kange lussuaq
in e na ional ai po in Wes G eenland. The 2018 popula ion size es ima e o Manii soq
muskoxen was ca. 20,334 muskoxen (95% CI: 9,386 – 44,055; SE = 6,397; CV = 0.31). The
la ge Coe icien o Va iance (CV) e lec s poo accu acy on he popula ion es ima e. This
es ima e canno p o ide popula ion end, because i was he i s using his me hod, and he e
a e no o he es ima es o compa ison.
The Ma ch 2018 ae ial su ey occu ed wo weeks pos -ha es (albei ophy ha es
con inued) and muskoxen we e s ill a oiding lowlands and o he a eas easily accessible o
hun e s. This is likely he p ima y ac o causing he nonuni o m dis ibu ion obse ed ac oss
he egion (Figu e 7). Few muskoxen (n=8, 0.5%) we e obse ed in he sou he nmos po ion
o he egion nea he Sukke oppen Ice Cap, likely because ha a ea is cha ac e ized by
ba en high ele a ions o a ound 1000 m o mo e. Despi e he la ge a ea o lowlands in he
no h only a ew muskoxen (n=81, 5.5%) u ilized ha a ea. This is ema kable, gi en ha
lowlands a e he p e e ed habi a o muskoxen. This a ea is, howe e , wi hin easy each o
hun e s coming om he Kange lussuaq in e na ional ai po own. Many muskoxen (n=200,
13.6%) we e in an a ea hal way be ween he ai po and he Sukke oppen Ice Cap.
Meanwhile, he majo i y concen a ed in wo suba eas on ei he side o he egion. Mos
(n=632, 43.1%) we e in he eas nea he G eenland Ice Cap and he es (n= 546, 37.2%)
we e in he wes . This was no su p ising because in he win e 2018, muskox hun ing was
p ohibi ed in hun ing a eas 2 and 4, while mo o ehicles we e p ohibi ed in hun ing a ea 1.
These es ic ions coincide wi h he wo g ea es concen a ions o muskoxen pos -ha es .
Rega dless o whe e hey we e obse ed, ew muskoxen we e u ilizing lowlands.
Figu e 7. Ma ch 2018 dis ibu ion o he 1,467 muskoxen obse ed wo weeks pos -ha es (albei s ill ophy-
bull season) du ing he ae ial su ey o he Manii soq muskox popula ion. Bounda ies a e app oxima e. (Cuyle
unpublished).
104
Supplemen a y Ma e ials Cuyle (2020: Appendix 2)
Manii soq muskox popula ion
Popula ion ends, 2000 – 2020
Da a om g ound-based minimum coun s, using skidoo and ATV. Da a om 2000-2010
pe iod is p e-ha es , while ha om 2014-2020 is pos -ha es .
Figu e 9. Numbe o g oups o Manii soq muskoxen obse ed du ing minimum coun s in he pe iod 2000-2020.
Figu e 10. Numbe o g oups o Manii soq muskoxen obse ed du ing minimum coun s in he pe iod 2000-
2020.
y = -27,769x + 56154
R² = 0,5817
0
100
200
300
400
500
600
700
800
900
1000
1995 2000 2005 2010 2015 2020 2025
Numbe o g oups obse ed
Yea
y = -0,1115x + 232,48
R² = 0,239
0
2
4
6
8
10
12
14
1995 2000 2005 2010 2015 2020 2025
A e age g oup size o muskoxen
Yea
105
Figu e 11. Maximum Manii soq muskox g oup size obse ed du ing minimum coun s in he pe iod 2000-2020.
Figu e 12. Win e cal ec ui men as ep esen ed by cal es (age ca. 10 mon hs) pe 100 cows, obse ed du ing
ou demog aphic coun s comple ed in he pe iod 2006-2020 o Manii soq muskoxen.
The ec ui men alues in 2017 and 2019 a e lowe han hose ob ained in 2006 and 2010
(Figu e 11). Gi en he o me a e pos -ha es alues, al hough ela i ely low, hese a e
p obably a i icially high since cows a e commonly ha es ed bu cal es a e no . Ac ual cal
ec ui men in 2017 and 2019 was likely lowe han shown.
y = -1,9725x + 4018,5
R² = 0,465
0
10
20
30
40
50
60
70
80
90
100
1995 2000 2005 2010 2015 2020 2025
Maximum muskox g oup size obse ed
Yea
0
5
10
15
20
25
30
35
40
45
50
2004 2006 2008 2010 2012 2014 2016 2018 2020
Cal es pe 100 Cows
Yea
112
Appendix 10
Recommenda ions o imp o ing u u e su eys o muskoxen.
Ae ial su ey me hods & design:
The app oxima ely 11% su ey co e age in 2018
p omo es accu acy o abundance es ima es and should be con inued in u u e o acili a e
popula ion end. Gi en muskoxen a e s a iona y and he win e hun ing season appea s o
dis u b hei dis ibu ion, in u u e he bes op ion may be helicop e su eys speci ic o
muskoxen and lown be o e he win e hun ing season begins. This is because hun ing
seems o dis u b no mal dis ibu ion, which esul ed in ‘ho -spo s’ o high concen a ions o
muskoxen ha made o high a iance among line ansec s. To ob ain g ea e accu acy and
p ecision o es ima es o muskox abundance and densi y, hen he p oblem o une en
dis ibu ion and lack o p esence in lowlands mus be ec i ied. Al e na i es migh include a
la e su ey pe iod han cu en ly used e.g., la e Ma ch o ea ly Ap il. Albei , all win e
su eys should end be o e he onse o cal ing, which o muskoxen can begin by mid-Ap il.
Ano he al e na i e would be o p ohibi hun ing and mo o ehicle ac i i y be o e (>2
weeks) and du ing he su ey pe iod. A one-mon h pe iod, o mo e, wi hou hun ing o
o he human dis u bance may be necessa y be o e muskox igilance elaxes, as exhibi ed by
o aging in he p e e ed lowland ele a ions and mo ing in o known co e win e ange in
hun ing a eas 1 and 2.
Muskox Demog aphics:
Ob aining accu a e demog aphics was no possible du ing he
Dis ance Sampling su ey. E en a he slow helicop e speed lown while lying line
ansec s, when a g oup o muskoxen was de ec ed, he e was usually only su icien ime o
ob ain o al g oup size, dis ance om he 0-line and some imes he g oup’s beha io al
eac ion o he ly-by. Accu a e iden i ica ion o he ela i ely small-bodied cal es (age 9-10
mon hs) was di icul , owing o animals o en milling abou and cal es ypically emaining
hidden among o behind la ge membe s o he g oup ega dless o hei dis ance om he
ack line. Muskox he d s uc u e da a mus be collec ed in a speci ic e o sepa a e om
lying he line ansec s o Dis ance Sampling da a. Fu u e budge s mus pe mi a minimum
3 o 6 hou s helicop e ime de o ed exclusi ely o muskox demog aphic da a collec ion.
Al e na ely, a sepa a e g ound e o using ae ial d ones could ob ain accu a e
demog aphics i used p io o hun ing season(s) and on co e ange.
Helicop e Logis ics:
Always check whe he o he helicop e op ions a e a ailable. To
da e, he smalles helicop e a ailable is he AS350 and only om Ai Cha e (Ai
G eenland). The AS350 pe mi s limi ed ision o ea obse e s, owing o he small window
size con aining se e al ba /s u s, and which unde cold ambien empe a u es always og
wi h ice- os . These ac o s educe isibili y o e ain.

113
[Emp y page]