249
Es ima ing in asi e oden abundance using emo al da a and
hie a chical models
Oli ie Gimenez1
1 CEFE, Uni Mon pellie , CNRS, EPHE, IRD, Mon pellie , F ance
Co esponding au ho : Oli ie Gimenez (oli ie .gimenez@ce e.cn s. )
Copy igh : © Oli ie Gimenez.
This is an open access a icle dis ibu ed unde
e ms o he C ea i e Commons A ibu ion
License (A ibu ion 4.0 In e na ional – CC BY 4.0).
Me hods
Abs ac
In asi e oden s pose signi ican ecological, economic, and public heal h challenges. Robus
me hods a e needed o es ima ing popula ion abundance o guide e ec i e managemen . T a-
di ional me hods such as cap u e- ecap u e a e o en imp ac ical o in asi e species due o e hi-
cal, legal and logis ical cons ain s. He e, he applica ion o hie a chical mul inomial N-mix u e
models o es ima ing he abundance o in asi e oden s using emo al da a is highligh ed. Fi s ly,
a simula ion s udy was pe o med which demons a ed minimal bias, as well as good p ecision
and eliable co e age o con idence in e als ac oss a ange o sampling scena ios. Addi ionally,
he consequences o iola ing he popula ion closu e assump ion we e illus a ed by showing
how be ween-occasion dynamics can bias in e ence. Secondly, emo al da a was analyzed o
wo in asi e oden species, namely coypus (Myocas o coypus) in F ance and musk a s (Onda a
zibe hicus) in he Ne he lands. Using hie a chical mul inomial N-mix u e models, he e ec o
empe a u e on abundance was examined, while accoun ing o impe ec and ime- a ying cap-
u e p obabili ies. Addi ionally, his s udy demons a ed how o accommoda e spa ial a iabili y
using andom e ec s, quan i y unce ain y in pa ame e es ima es, and accoun o iola ions o
closu e by i ing an open-popula ion model o mul i-yea da a. Taken oge he , hese app oaches
demons a e he lexibili y and u ili y o hie a chical models in in asi e species managemen .
Key wo ds: In asi e species, Mul inomial N-mix u e, popula ion size, s a is ical ecology
In oduc ion
In asi e species a e a signi ican global issue, wi h wide- anging impac s on ecosys-
ems, economies, and public heal h (Pyšek e al. 2020; Roy e al. 2024). Among
hese, he inancial, epidemiological, social, and ecological cos s associa ed wi h in-
asi e oden s a e subs an ial, as hey damage in as uc u es, deg ade ag icul u al
sys ems, and ac as ese oi s o zoono ic diseases (Diagne e al. 2023).
E ec i e managemen o in asi e species equi es he es ima ion o popula-
ion abundance o guiding con ol e o s and e alua ing he success o e ad-
ica ion o egula ion p og ams (Williams e al. 2002; Thompson e al. 2021).
Howe e , he challenge in es ima ing animal abundance is ha , because o
impe ec de ec ion, indi iduals a e no always obse ed e en when p esen
(Bo che s e al. 2010; Sebe and Scho ield 2023). Igno ing impe ec de ec ion
leads o biased es ima es o popula ion abundance (Ké y and Schmid 2008). To
accoun o impe ec de ec ion, cap u e- ecap u e me hods a e usually used o
Academic edi o : Sand o Be olino
Recei ed:
4 Janua y 2025
Accep ed:
14 May 2025
Published:
4 No embe 2025
Ci a ion: Gimenez O (2025) Es ima ing
in asi e oden abundance using
emo al da a and hie a chical models.
NeoBio a 103: 249–265. h ps://doi.
o g/10.3897/neobio a.103.145876
NeoBio a 103: 249–265 (2025)
DOI: 10.3897/neobio a.103.145876
Ad ancing esea ch on alien species and biological in asions
A pee - e iewed open-access jou nal
NeoBio a
250
NeoBio a 103: 249–265 (2025), DOI: 10.3897/neobio a.103.145876
Oli ie Gimenez: Roden abundance and emo al da a
co ec obse ed coun s (McC ea and Mo gan 2014). Ye , o in asi e species,
cap u e- ecap u e is o en imp ac ical, as e hical and managemen conce ns yp-
ically p e en he elease o cap u ed animals.
An al e na i e app oach in ol es he use o emo al me hods (Rod iguez de
Ri e a and McC ea 2021) in which indi iduals a e cap u ed and pe manen ly
emo ed om he s udy a ea du ing successi e sampling occasions. This p ocess
leads o a dec ease in he expec ed numbe o cap u es by a consis en p opo ion
o e ime ( a he han by a ixed amoun decline), which in o ms on he o al
abundance as he ini ial popula ion de e mines how quickly he numbe o indi-
iduals a ailable o cap u e diminishes.
While s anda d emo al me hods a e well-es ablished (Mo an 1951; Zippin
1956; 1958, Rod iguez de Ri e a and McC ea 2021) ecen ad ances in popula-
ion ecology emain unde u ilized in he con ex o in asi e species. Hie a chical
models, in pa icula , ha e gained ac ion (Royle and Do azio 2008; Ké y and
Royle 2015) due o hei abili y o: (i) explici ly sepa a e biological p ocesses o
in e es (e.g., popula ion dynamics) om obse a ion p ocesses (e.g., impe ec
de ec ion), hus enabling mo e accu a e modeling; (ii) inco po a e en i onmen al,
spa ial, o empo al co a ia es a mul iple le els, allowing explo a ion o how a i-
ous ac o s in luence ecological p ocesses; and (iii) sha e in o ma ion ac oss g oups
(e.g., yea s) by modeling pa ame e s hie a chically wi h andom e ec s, which im-
p o es es ima es o g oups wi h ewe da a.
In his pape , I showcase he applica ion o a hie a chical o mula ion o
emo al models, he mul inomial N-mix u e model (Do azio e al. 2005), o
es ima e he abundance o oden s in Eu ope. In his s udy, I ocus on he
coypu (Myocas o coypus) in F ance and he musk a (Onda a zibe hicus) in he
Ne he lands. Bo h species a e semi-aqua ic oden s in oduced o Eu ope in
he ea ly 20 h cen u y ollowing escapes o eleases om u a ms. The coypu,
na i e o Sou h Ame ica, has o med widesp ead in asi e popula ions in F ance
(Bonne e al. 2023), whe e i causes signi ican damage o in as uc u e and
c ops. Addi ionally, i se es as a heal hy ca ie o lep ospi osis, a zoono ic
disease wi h po en ially se ious consequences. Simila ly, he musk a , na i e o
No h Ame ica, has es ablished ex ensi e popula ions in he Ne he lands. By
bu owing in o i e banks, dykes, and dams, musk a s comp omise he in eg-
i y o hese s uc u es, posing a h ea o public sa e y (Van Loon e al. 2017).
Bo h species a e also widesp ead in o he Eu opean coun ies; upda ed dis i-
bu ion maps a e a ailable ia he Eu opean Alien Species In o ma ion Ne wo k
(EASIN) pla o m (h ps://easin.j c.ec.eu opa.eu/spexplo e /sea ch/).
Using emo al da a, I demons a e he applica ion o he mul inomial
N-mix u e model o es ima e he abundance o oden popula ions. Fi s , I
conduc a simula ion s udy o e alua e he model’s pe o mance unde a ying
numbe s o sampling si es and sampling occasions. Second, I p esen a case
s udy on a coypu popula ion in F ance o illus a e he hie a chical s uc u e
o he mul inomial N-mix u e model, demons a ing how co a ia es can be
inco po a ed o accoun o a ia ions in abundance and cap u e p obabili ies.
Thi d, I use a case s udy on musk a s in he Ne he lands o illus a e he in e-
g a ion o andom e ec s wi hin he model and demons a e how o elax he
closu e assump ion. To acili a e ep oducibili y, I p o ide he accompanying
code and da a, aiming o p omo e he b oade adop ion o emo al models in
he s udy o biological in asions.
251
NeoBio a 103: 249–265 (2025), DOI: 10.3897/neobio a.103.145876
Oli ie Gimenez: Roden abundance and emo al da a
Me hods
Mul inomial N-mix u e model
Think o a dice wi h six sides. The dice has a 1 in 6 chance o landing on ace 1, he
same o ace 2, and so on. I I oll he dice 30 imes, I would expec , on a e age
o e many epe i ions o his expe imen , o ge ace 1 i e imes, ace 2 i e imes,
and so on. You can es his in p og am R by unning he command “ mul9nom”
(wi h a gumen s n = 1, size = 30 and p ob = c(1/6, 1/6, 1/6, 1/6, 1/6, 1/6)) epea -
edly. In his expe imen , y1, he numbe o 1s, y2, he numbe o 2s, ..., and y6, he
numbe o 6s, ollows a mul inomial dis ibu ion wi h pa ame e s he numbe o
olls (30) and p obabili ies (1/6, 1/6, ..., 1/6).
Now hink o a emo al campaign conduc ed o e 3 mon hs. We eco d he
numbe o oden s y1 cap u ed in mon h 1, y2 in mon h 2, y3 in mon h 3, and
le y4 ep esen he numbe o oden s ne e cap u ed. Le p be he p obabili y o
cap u ing a oden in a gi en mon h. The p obabili y o cap u ing a oden in he
i s mon h is π1 = p. The p obabili y o cap u ing a oden in he second mon h is
π2 = (1-p)p he p obabili y o no cap u ing i in he i s mon h (1 - p) mul iplied
by he p obabili y o cap u ing i in he second mon h p. The p obabili y o cap u -
ing a oden in he hi d mon h is π3 = (1-p)(1-p)p, he p obabili y o no cap u ing
i in he i s and second mon hs, (1 - p)(1 - p), mul iplied by he p obabili y o
cap u ing i in he hi d mon h, p. Finally, he p obabili y o ne e being cap u ed
is π4 = 1 - (π1 + π2 + π3) he complemen o he p obabili y o being cap u ed in he
i s , second, o hi d mon h. I we assume ha N ep esen s he abundance, hen
we ha e ha he ec o o coun s (y1, y2, y3, y4) ollows a mul inomial dis ibu ion
wi h pa ame e s N and p obabili ies (π1, π2, π3, π4). This is he obse a ion p ocess.
In gene al, we assume ha N ollows a Poisson dis ibu ion wi h pa ame e he
expec ed numbe o oden s deno ed λ. This is he s a e o ecological p ocess. And
he e you ha e i , he mul inomial N-mix u e model o a emo al expe imen ,
which is simila o h owing a dice N imes and he π’s gi e he p obabili ies ha
I ge a gi en ace o ha dice. Unlike a ai die, howe e , he p obabili ies in a
emo al expe imen a e no equal; hey e lec a ying de ec ion p obabili ies o e
ime, which depend on ac o s like e o , animal beha io , o en i onmen al con-
di ions. Also, in gene al, we moni o oden s in se e al popula ions i = 1,...,S and
we need o es ima e local abundance Ni. To do so, Do azio e al. (2005) ex ended
mul inomial N-mix u e models o accoun o spa ial a ia ion in abundance and/
o cap u e, and showed ha abundance es ima es had simila o be e p ecision
han hose ob ained om analyzing emo al da a o each popula ion sepa a ely.
Pa ame e s N, p, and λ a e unknown and need o be es ima ed. In a equen is
amewo k, ma ginaliza ion is pe o med by summing o e all possible alues o N
(Do azio e al. 2005). In a Bayesian amewo k, all hese pa ame e s a e es ima ed
di ec ly, which simpli ies he p ocess (Royle and Do azio 2006). Bo h pa ame e s,
λ and p, can be modeled as unc ions o explana o y spa ial and empo al a iables,
in he spi i o gene alized linea models, and Poisson (wi h a log link unc ion) o
logis ic eg essions (wi h a logi link unc ion) o example.
To e alua e model adequacy, I used s anda d goodness-o - i p ocedu es adap ed
o bo h equen is and Bayesian amewo ks. In he equen is amewo k, we ap-
ply a pa ame ic boo s ap app oach: we gene a e a la ge numbe o eplica e da a-
se s om he maximum likelihood es ima es, e i he model o each eplica e, and
compu e diagnos ic s a is ics such as he F eeman–Tukey s a is ic. I he esul ing
252
NeoBio a 103: 249–265 (2025), DOI: 10.3897/neobio a.103.145876
Oli ie Gimenez: Roden abundance and emo al da a
boo s ap p- alues all wi hin a non-ex eme ange, his indica es no e idence o
lack o i . In he Bayesian amewo k, I assessed model adequacy using pos e io
p edic i e checks based on Bayesian p- alues. A each MCMC i e a ion, a eplica e
da ase is d awn om he join pos e io dis ibu ion, and he F eeman–Tukey dis-
c epancy is compu ed o bo h he obse ed and eplica e da a. Gi en he condi-
ional mul inomial s uc u e o he model, which sepa a es he obse a ion p ocess
om he abundance p ocess, I calcula ed wo disc epancy measu es: one o he
de ec ion his o ies and ano he o he o al coun s obse ed. Bayesian p- alues
nea 0.5 (and away om 0 o 1) indica e no e idence o sys ema ic lack o i .
Fo a de ailed desc ip ion o he mul inomial mix u e model, I wa mly ecom-
mend chap e 7 in Ké y and Royle (2015).
Simula ion s udy
I conduc ed a simula ion s udy o e alua e he model’s pe o mance by examining
pa ame e bias unde a ying numbe s o sampling si es and sampling occasions. I
simula ed emo al da a o e 1, 5, 10 and 50 si es using a Poisson dis ibu ion wi h
expec ed numbe o animals λ be ween 10 and 100 (20 alues) o he ecological
p ocess. I simula ed he obse a ion p ocess wi h a cap u e p obabili y p a ying
be ween 0.3 and 0.9 (20 alues) ac oss 3, 5 and 10 occasions pe si e. In o al, I
conside ed 4800 scena ios. I i ed he mul inomial N-mix u e model o he sim-
ula ed da a wi hin he equen is amewo k using unc ion “mul inomPois()” in
he R package “unma ked” (Kellne e al. 2023), and I epea ed his p ocedu e 100
imes. E en ually, I calcula ed he ela i e bias, oo mean squa e e o (RMSE),
and co e age o he 95% con idence in e al o each pa ame e .
To assess he e ec o iola ing he closu e assump ion, I implemen ed an addi-
ional se o simula ions in which he popula ion could change be ween sampling oc-
casions. Speci ically, indi iduals s aying in he popula ion wi h p obabili y 0.8, and
new indi iduals a i e acco ding o a Poisson p ocess wi h mean 1. Apa om hese
be ween-occasion dynamics, all o he aspec s o he simula ion se up emained he
same. This se up b eaks he closu e assump ion in wo ways. Some indi iduals lea e
he popula ion be ween sampling occasions, iola ing he assump ion ha declines
in abundance a e due o emo al alone; his can bias de ec ion p obabili y and abun-
dance es ima es. New indi iduals en e be ween sampling occasions, in la ing he
pool o animals a ailable o de ec ion and leading o o e es ima ion o abundance.
Since I delibe a ely i a closed model o da a om an open p ocess, any esul ing
bias di ec ly e lec s he impac o iola ing closu e. While his simula ion ocuses
on geog aphic closu e, he same logic applies o demog aphic closu e, whe e he s ay
and a i als pa ame e s co espond o su i al and ec ui men p ocesses.
No e ha I used a equen is implemen a ion o he simula ion s udy o e-
duce compu a ion ime gi en he la ge numbe o scena ios. The model s uc u e
emains hie a chical, as in he Bayesian case s udies, and bo h in e en ial app oach-
es would yield simila esul s. The aim was o assess model pe o mance ac oss
ecological and sampling condi ions, no o compa e s a is ical pa adigms.
Case s udies
In his sec ion, I analyzed emo al da a om wo oden species: coypus in F ance and
musk a s in he Ne he lands. Wi h hese case s udies, I aimed a illus a ing speci ic ea-
253
NeoBio a 103: 249–265 (2025), DOI: 10.3897/neobio a.103.145876
Oli ie Gimenez: Roden abundance and emo al da a
u es o hie a chical mul inomial N-mix u e models. Fo bo h species, I explo ed he
po en ial e ec o empe a u e on abundance (e.g., Gosling 1981; Simpson and Bou-
in 1993). A comp ehensi e analysis o he ecological ac o s in luencing popula ion
dynamics was beyond he scope o his wo k and will be add essed in u u e s udies.
Coypus in F ance
Remo al da a on coypus we e collec ed om annual con ol ope a ions conduc ed
since 2015 in se e al ci ies wi hin he Hé aul depa men , loca ed in he Occi anie
egion o sou he n F ance. These ope a ions a e ca ied ou yea - ound, wi h he ex-
cep ion o July and Augus . Coypus a e apped using cages by a ne wo k o olun-
ee s coo dina ed by he Syndica Mix e du Bassin de l’O and he Fédé a ion Dépa -
emen ale des Chasseu s de l’Hé aul (h ps://e ang-de-l-o .com/lu e- agondins/).
Fo his s udy, I ocus on da a om 2022, speci ically om sampling occasions in Feb-
ua y, Ma ch, and Ap il. The da a, co e ing S = 6 ci ies, a e summa ized in Table 1.
I i ed a model whe e he expec ed numbe o coypus was modeled as a unc ion
o empe a u e, while he cap u e p obabili y was allowed o a y by mon h. A key
assump ion o he mul inomial N-mix u e model is ha abundance ollows a Poisson
dis ibu ion, which implies equal mean and a iance. When his assump ion is io-
la ed - i.e., in he p esence o o e dispe sion - a common and e ec i e solu ion is o
eplace he Poisson wi h a nega i e binomial dis ibu ion. I illus a e how o i such
an o e -dispe sed model using he coypu da ase . No e ha a si e andom e ec was
no included he e, as he spa ial scale o he coypu da ase was limi ed. Howe e , such
e ec s may be impo an o conside in b oade -scale p og ams whe e unobse ed
spa ial he e ogenei y is likely o be mo e p onounced, as in he musk a case s udy.
Musk a s in he Ne he lands
Remo al da a on musk a s in he Ne he lands we e collec ed by p o essional ap-
pe s. The da a we e egis e ed in a las blocks (5 × 5 km) pe pe iods o ou weeks.
Fo his s udy, I ocus on da a om 2014, speci ically om sampling occasions in
Janua y, Feb ua y, and Ma ch. The da a we e made a ailable h ough he LIFE
MICA p ojec (Ca uy els e al. 2024) and can be eely downloaded om h ps://
www.gbi .o g/da ase /7d75109d-a6cb-4e90-89d0-79d08577c580 (Moe kens e
al. 2025). The da a, co e ing S = 215 ci ies (ou o he 342 ci ies in he Ne h-
e lands), a e p esen ed in Fig. 1. I i ed he same model as o he coypus da a,
excep ha I added a si e andom e ec on abundance o accommoda e he spa ial
a ia ion ha was no explained by empe a u e.
Table 1. Numbe o in asi e coypus emo ed mon hly and he a e age 3-mon h empe a u e ac oss
se e al ci ies in he He aul depa men , F ance, in 2022.
Ci y Remo ed in
Feb ua y
Remo ed in
Ma ch
Remo ed in
Ap il
A e aged
empe a u e
Candilla gues 18 12 38 9.5
Lansa gues 15 17 75 8.8
Mauguio 20 9 6 9.2
Sain -Nazai e-de-Pezan 169 41 15 9.3
Sain -Jus 85 61 77 9.2
Vale gues 0 1 3 9.4
254
NeoBio a 103: 249–265 (2025), DOI: 10.3897/neobio a.103.145876
Oli ie Gimenez: Roden abundance and emo al da a
Figu e 1. To al numbe o in asi e musk a s emo ed o e he pe iod Janua y-Feb ua y-Ma ch ( op panel), and he a e age 3-mon h
empe a u e (bo om panel) ac oss he Ne he lands in 2014.
255
NeoBio a 103: 249–265 (2025), DOI: 10.3897/neobio a.103.145876
Oli ie Gimenez: Roden abundance and emo al da a
A key assump ion unde lying he p ope use o mul inomial N-mix u e models
is ha o popula ion closu e, which assumes no bi hs, dea hs, immig a ion, o
emig a ion occu du ing he apping pe iod. A s aigh o wa d app oach o elax
his assump ion is o i mul iple yea s o da a (a.k.a. s acking he da a) in o a
s anda d mul inomial N-mix u e model. In his app oach, yea -si e combina ions
a e ea ed as sepa a e si es, and yea is included as a si e co a ia e o andom e ec
in he model. I used his me hod o e alua e a empo al e ec on he ela ionship
be ween empe a u e and abundance. Assuming an inc ease in empe a u e o e
ime, one migh p edic a decoupling o weakening o he ela ionship be ween
abundance and empe a u e. To es his, I conduc ed an addi ional analysis span-
ning he 1987–2014 pe iod, modeling he slope o he empe a u e-abundance
ela ionship as a linea unc ion o ime.
Implemen a ion
Fo all analyses, I used he s a is ical language R (R Co e Team 2024). I used he
“ idy e se” (Wickham e al. 2019) sui e o packages o da a manipula ion and
isualiza ion, “s ” (Pebesma and Bi and 2023) o dealing wi h spa ial da a and
“k igR” (Kusch and Da y 2022) o ge empe a u e da a. Fo he simula ions, I
used he R package “unma ked” (Kellne e al. 2023), see he “Simula ion s udy”
sec ion. Fo he wo case s udies, I i ed models wi hin a Bayesian amewo k
using Ma ko chain Mon e Ca lo (MCMC) algo i hms. I used bo h he “NIM-
BLE” (de Valpine e al. 2017) and he “ubms” (Kellne e al. 2022) packages. The
o me o e s high lexibili y, enabling use s o de ine cus om likelihoods, hough
i equi es manual coding, while he la e ea u es simple syn ax wi h p e-buil
mul inomial N-mix u e models, albei limi ed o a Poisson dis ibu ion o abun-
dance. I speci ied weakly in o ma i e p io s o all pa ame e s, speci ically no mal
dis ibu ions wi h mean 0 and s anda d de ia ion 1.5 o eg ession pa ame e s,
and a uni o m dis ibu ion o he s anda d de ia ion o he andom e ec s. I an
wo chains o a o al o 200,000 i e a ions wi h a bu n-in o 20,000 i e a ions. I
summa ized pos e io dis ibu ions wi h pos e io mean and 95% c edible in e -
als. I assessed con e gence using s anda d Bayesian diagnos ics: he R-ha s a is ic
( alues close o 1 indica e con e gence), e ec i e sample size (which e lec s he
amoun o independen in o ma ion in he pos e io sample, should be > 100),
and isual inspec ion o ace plo s (which should show good mixing and s a ion-
a i y o he chains).
Resul s and discussion
The esul s o he simula ion s udy a e p esen ed in Figs 2, 3. O e all, he analysis
e ealed minimal bias, good p ecision and sa is ying co e age, wi h he excep ion
o one si e ha showed a no able de ia ion (Fig. 2).
Inc easing he numbe o si es o 10 signi ican ly educed his bias, and no bias
was obse ed wi h 50 si es, suppo ing he ecommenda ion by (Do azio e al.
2005) o analyze da a join ly a he han sepa a ely.
When he closu e assump ion was no me , he analysis e ealed ha bo h bias
and p ecision me ics we e highly sensi i e o he in oduc ion o be ween-occa-
sion popula ion dynamics (Fig. 3). Speci ically, ela i e bias inc eased and co e age
d opped in many scena ios, pa icula ly when de ec ion p obabili y was low o he
256
NeoBio a 103: 249–265 (2025), DOI: 10.3897/neobio a.103.145876
Oli ie Gimenez: Roden abundance and emo al da a
Figu e 2. Rela i e bias ( op panel), oo mean squa e e o (RMSE; middle panel) and co e age o he 95% con idence in e al (bo om
panel) o abundance es ima es om a mul inomial N-mix u e model wi h cons an pa ame e s. Cap u e p obabili ies (X-axis) ange om
0.3 o 0.9, while abundance (Y-axis) a ies be ween 10 and 100 indi iduals. Scena ios conside 3, 5, and 10 cap u e occasions (columns)
and 1, 5, 10, and 50 si es ( ows). Resul s a e based on 100 simula ions.
257
NeoBio a 103: 249–265 (2025), DOI: 10.3897/neobio a.103.145876
Oli ie Gimenez: Roden abundance and emo al da a
Figu e 3. Rela i e bias ( op panel), oo mean squa e e o (RMSE; middle panel) and co e age o he 95% con idence in e al (bo om
panel) o abundance es ima es om a mul inomial N-mix u e model wi h cons an pa ame e s, i ed o da a whe e he closu e assump-
ion was delibe a ely iola ed. Be ween cap u e occasions, indi iduals emained in he popula ion wi h p obabili y 0.8, and new indi idu-
als a i ed acco ding o a Poisson p ocess wi h mean 1, in oducing bo h emig a ion and immig a ion be ween sampling e en s. Cap u e
p obabili ies (X-axis) ange om 0.3 o 0.9, while abundance (Y-axis) a ies be ween 10 and 100 indi iduals. Scena ios conside 3, 5, and
10 cap u e occasions (columns) and 1, 5, 10, and 50 si es ( ows). Resul s a e based on 100 simula ions.
264
NeoBio a 103: 249–265 (2025), DOI: 10.3897/neobio a.103.145876
Oli ie Gimenez: Roden abundance and emo al da a
Shea K, Possingham HP, Mu doch WW, Roush R (2002) Ac i e Adap i e Managemen in Insec Pes
and Weed Con ol: In e en ion wi h a Plan o Lea ning. Ecological Applica ions 12: 927–936.
h ps://doi.o g/10.1890/1051-0761(2002)012[0927:AAMIIP]2.0.CO;2
Simpson MR, Bou in S (1993) Musk a li e his o y: A compa ison o a no he n and sou he n pop-
ula ion. Ecog aphy 16: 5–10. h ps://doi.o g/10.1111/j.1600-0587.1993. b00052.x
Thompson BK, Olden JD, Con e se SJ (2021) Mechanis ic in asi e species managemen models
and hei applica ion in conse a ion. Conse a ion Science and P ac ice 3: e533. h ps://doi.
o g/10.1111/csp2.533
Tibe i R, Buchaca T, Boiano D, Knapp RA, Pou Ro i a Q, Ta ecchia G, Ven u a M, Tenan S (2021)
Alien ish e adica ion om high moun ain lakes by mul iple emo al me hods: Es ima ing esid-
ual abundance and e adica ion p obabili y in open popula ions. Jou nal o Applied Ecology 58:
1055–1068. h ps://doi.o g/10.1111/1365-2664.13857
Van Loon EE, Bos D, Van Hellenbe g Huba CJ, Ydenbe g RC (2017) A his o ical pe spec i e on he
e ec s o apping and con olling he musk a (Onda a zibe hicus) in he Ne he lands: His o y
o musk a apping in he Ne he lands. Pes Managemen Science 73: 305–312. h ps://doi.
o g/10.1002/ps.4270
Wickham H, A e ick M, B yan J, Chang W, McGowan LD, F ançois R, G olemund G, Hayes
A, Hen y L, Hes e J, Kuhn M, Pede sen TL, Mille E, Bache SM, Mülle K, Ooms J,
Robinson D, Seidel DP, Spinu V, Takahashi K, Vaughan D, Wilke C, Woo K, Yu ani H
(2019) Welcome o he Tidy e se. Jou nal o Open Sou ce So wa e 4: 1686. h ps://doi.
o g/10.21105/joss.01686
Williams BK, Nichols JD, Con oy MJ (2002) Analysis and managemen o animal popula ions:
modeling, es ima ion, and decision making. Academic P ess, San Diego, 817 pp.
Womack-Bulline KM, Amelon SK, Thompson FR, Leb un JJ (2019) Pe o mance o Hie a chical
Abundance Models on Simula ed Ba Cap u e Da a. Ac a Chi op e ologica 20: 465–474. h ps://
doi.o g/10.3161/15081109ACC2018.20.2.016
Yackel Adams AA, Hos e e NJ, Link WA, Con e se SJ (2024) Iden i ying Pa e o-e icien e ad-
ica ion s a egies o in asi e popula ions. Conse a ion Le e s 17: e13051. h ps://doi.
o g/10.1111/conl.13051
Zhou M, McC ea RS, Ma echou E, Cole DJ, G i i hs RA (2019) Remo al models accoun ing o
empo a y emig a ion. Biome ics 75: 24–35. h ps://doi.o g/10.1111/biom.12961
Zippin C (1956) An E alua ion o he Remo al Me hod o Es ima ing Animal Popula ions. Biome -
ics 12: 163–189. h ps://doi.o g/10.2307/3001759
Zippin C (1958) The Remo al Me hod o Popula ion Es ima ion. The Jou nal o Wildli e Man-
agemen 22: 82–90. h ps://doi.o g/10.2307/3797301 h ps://doi.o g/10.1201/9780429459016
Supplemen a y ma e ial 1
Assessmen o abundance es ima o p ope ies in emo al models
Au ho : Oli ie Gimenez
Da a ype: pd
Explana ion no e: Simula ions o compu e bias, RMSE and co e age in abundance as es ima ed in a
emo al model wi h cons an pa ame e s.
Copy igh no ice: This da ase is made a ailable unde he Open Da abase License (h p://openda a-
commons.o g/licenses/odbl/1.0/). The Open Da abase License (ODbL) is a license ag eemen
in ended o allow use s o eely sha e, modi y, and use his Da ase while main aining his same
eedom o o he s, p o ided ha he o iginal sou ce and au ho (s) a e c edi ed.
Link: h ps://doi.o g/10.3897/neobio a.103.145876.suppl1
265
NeoBio a 103: 249–265 (2025), DOI: 10.3897/neobio a.103.145876
Oli ie Gimenez: Roden abundance and emo al da a
Supplemen a y ma e ial 2
Assessmen o abundance es ima o p ope ies in emo al models when he closu e
assump ion is no me
Au ho : Oli ie Gimenez
Da a ype: pd
Explana ion no e: Simula ions o compu e bias, RMSE and co e age in abundance as es ima ed
in a emo al model wi h cons an pa ame e s, i ed o da a whe e he closu e assump ion was
delibe a ely iola ed.
Copy igh no ice: This da ase is made a ailable unde he Open Da abase License (h p://openda a-
commons.o g/licenses/odbl/1.0/). The Open Da abase License (ODbL) is a license ag eemen
in ended o allow use s o eely sha e, modi y, and use his Da ase while main aining his same
eedom o o he s, p o ided ha he o iginal sou ce and au ho (s) a e c edi ed.
Link: h ps://doi.o g/10.3897/neobio a.103.145876.suppl2
Supplemen a y ma e ial 3
Es ima ing coypus abundance
Au ho : Oli ie Gimenez
Da a ype: pd
Explana ion no e: Applica ion o hie a chical mul inomial models o coypus emo al da a. I illus a e
he hie a chical s uc u e o he mul inomial N-mix u e model, and demons a e how o use co-
a ia es on he abundance and cap u e p obabili ies. I also illus a e how o deal wi h o e dispe -
sion by ing a model wi h a nega i e binomial (ins ead o a Poisson) dis ibu ion o abundance.
I use emo al da a on coypus in F ance.
Copy igh no ice: This da ase is made a ailable unde he Open Da abase License (h p://openda a-
commons.o g/licenses/odbl/1.0/). The Open Da abase License (ODbL) is a license ag eemen
in ended o allow use s o eely sha e, modi y, and use his Da ase while main aining his same
eedom o o he s, p o ided ha he o iginal sou ce and au ho (s) a e c edi ed.
Link: h ps://doi.o g/10.3897/neobio a.103.145876.suppl3
Supplemen a y ma e ial 4
Es ima ing musk a s abundance
Au ho : Oli ie Gimenez
Da a ype: pd
Explana ion no e: Applica ion o hie a chical mul inomial models o musk a s emo al da a. I illus-
a e he use o andom eec si e on abundance. I use emo al da a on musk a s in he Ne he lands.
Copy igh no ice: This da ase is made a ailable unde he Open Da abase License (h p://openda a-
commons.o g/licenses/odbl/1.0/). The Open Da abase License (ODbL) is a license ag eemen
in ended o allow use s o eely sha e, modi y, and use his Da ase while main aining his same
eedom o o he s, p o ided ha he o iginal sou ce and au ho (s) a e c edi ed.
Link: h ps://doi.o g/10.3897/neobio a.103.145876.suppl4
