Content uploaded by Richard Bischof
Author content
All content in this area was uploaded by Richard Bischof on Dec 21, 2022
Content may be subject to copyright.
Norwegian University of Life Sciences
Faculty of Environmental Sciences and Natural Resource Management
2022
ISSN 2535-2806 MINA fagrapport 79
Estimates of wolverine density, abundance, and
population dynamics in Scandinavia, 2014–2022
Cyril Milleret
Pierre Dupont
Ehsan Moqanaki
Henrik Brøseth
Øystein Flagstad
Oddmund Kleven
Jonas Kindberg
Richard Bischof
Milleret, C., Dupont, P., Moqanaki, E., Brøseth, H., Flagstad, Ø, Kleven, O., Kindberg, J., and
Bischof, R., 2022. Estimates of wolverine density, abundance, and population dynamics in
Scandinavia, 2014–2022 - MINA fagrapport 79. 35 pp.
Ås, December 2022
ISSN: 2535-2806
COPYRIGHT
©Norwegian University of Life Sciences (NMBU)
The publication may be freely cited where the source is acknowledged
AVAILABILITY
Open
PUBLICATION TYPE
Digital document (pdf)
QUALITY CONTROLLED BY
The Research committee (FU), MINA, NMBU
PRINCIPAL
Naturvårdsverket, Ref: NV-04078-22, Contact person: Robert Ekblom
Miljødirektoratet, Ref: 22047026, Contact person: Terje Bø
COVER PICTURE
Wolverine, J. Percy/Shutterstock.
NØKKELORD
Gulo gulo, jerv, tetthet, populasjonsdynamikk, deteksjonssannsynlighet, ikke-invaderende innsamling av genetisk
materiale, åpen populasjon romlig fangst-gjenfangst, rovdyrforvaltning
KEY WORDS
Gulo gulo, wolverine, population density, population dynamics, detection probability, non-invasive genetic sam-
pling, open-population spatial capture-recapture, carnivore management
Cyril Milleret, Faculty of Environmental Sciences and Natural Resource Management, Norwegian University of
Life Sciences, PO Box 5003, NO-1432 Ås, Norway
Pierre Dupont, Faculty of Environmental Sciences and Natural Resource Management, Norwegian University
of Life Sciences, PO Box 5003, NO-1432 Ås, Norway
Ehsan Moqanaki, Faculty of Environmental Sciences and Natural Resource Management, Norwegian Univer-
sity of Life Sciences, PO Box 5003, NO-1432 Ås, Norway
Henrik Brøseth, Norwegian Institute for Nature Research, PO Box 5685, NO-7485 Trondheim, Norway
Øystein Flagstad, Norwegian Institute for Nature Research, PO Box 5685, NO-7485 Trondheim, Norway
Oddmund Kleven, Norwegian Institute for Nature Research, PO Box 5685, NO-7485 Trondheim, Norway
Jonas Kindberg, Norwegian Institute for Nature Research, PO Box 5685, NO-7485 Trondheim, Norway
Richard Bischof (richard.bischof@nmbu.no), Faculty of Environmental Sciences and Natural Resource Man-
agement, Norwegian University of Life Sciences, PO Box 5003, NO-1432 Ås, Norway
Summary
Background The Scandinavian wolverine (Gulo gulo) population is being monitored annually
using non-invasive genetic sampling (NGS) and recovery of dead individuals. DNA extracted
from feces, urine, hair, secretion, and tissue is used to identify the species, sex, and individual
from which each sample originated. These data have been compiled in the Scandinavian large
carnivore database Rovbase 3.0. (www.rovbase.se, www.rovbase.no).
Approach Using the Bayesian open-population spatial capture-recapture (OPSCR) model de-
veloped by RovQuant, we estimated annual density and vital rates of the Scandinavian wolverine
population for nine consecutive seasons from 2014 to 2022.
Results We generated annual density maps and estimated total and jurisdiction-specific pop-
ulation sizes for the wolverine during 2014 to 2022. Based on the OPSCR model, the size of
the Scandinavian wolverine population was likely (95% credible interval) between 980 and 1088
individuals in 2022, with 625 to 709 individuals attributed to Sweden and 349 to 391 to Norway.
In addition to annual density and jurisdiction-specific abundance estimates, we report, for each
sex, annual estimates of cause-specific mortality, recruitment, and detection probability.
3
Sammendrag
Bakgrunn Den skandinaviske bestanden av jerv (Gulo gulo) blir overvåket årlig ved bruk av
ikke-invasiv genetisk prøveinnsamling (NGS) og gjenfunn av døde individer. DNA ekstrahert fra
skit, urin, hår og vev brukes til å identifisere art, kjønn og individ for hver enkelt prøve. Denne
informasjonen samles og ivaretas i den skandinaviske databasen for store rovdyr; Rovbase 3.0
(www.rovbase.se, www.rovbase.no).
Tilnærming Ved bruk av en Bayesiansk åpen romlig fangst-gjenfangst populasjons modell
(OPSCR), utviklet av RovQuant, estimerte vi årlige tettheter og demografiske rater hos den
skandinaviske jervebestanden i ni sesonger fra 2014 til 2022.
Resultater Vi laget årlige kart med tetthet av jerv fra 2014 til 2022, hvor bestandsstørrelsen
både totalt og innenfor ulike administrative enheter kunne avledes. Basert på OPSCR modellen
var den skandinaviske bestanden av jerv mellom 980 og 1088 individer i 2022 (95% kredibelt
intervall), med 625 til 709 individer i Sverige og 349 til 391 individer i Norge. I tillegg til årlige
tettheter og områdespesifikke bestandsestimater, gir rapporten estimater på dødlighetsfaktorer,
rekruttering og oppdagbarhet.
4
Contents
1 Introduction 6
2 Methods 8
2.1 Data........................................... 8
2.2 Open-population spatial capture-recapture model . . . . . . . . . . . . . . . . . . 9
3 Results 13
3.1 Non-invasive genetic samples and dead recoveries . . . . . . . . . . . . . . . . . . 13
3.2 Densityandabundance ................................ 13
3.3 Vitalrates........................................ 15
3.4 Detectionprobability.................................. 17
4 Summary of improvements made 18
5 Suggestions for future improvements 18
6 Other recommendations 18
7 Acknowledgements 19
8 Data availability 19
References 21
Appendices 22
5
1 Introduction
Sweden and Norway monitor large carnivores using non-invasive genetic sampling (NGS)
and dead recoveries. Both countries have collected an extensive individual-based data set for
the wolverine (Gulo gulo), which is stored in the Scandinavian large carnivore database Rovbase
(www.rovbase.se, www.rovbase.no).
Since 2017, project RovQuant has been developing statistical methods that allow a com-
prehensive assessment of the status and dynamics of large carnivore populations using NGS
data and other sources of information stored in Rovbase (Bischof et al., 2019b, 2020b). The
analytical framework developed by RovQuant is based on Bayesian open-population spatial
capture-recapture (OPSCR) models (Ergon and Gardner, 2014; Bischof et al., 2016; Chandler
et al., 2018). These models use the spatial and temporal information contained in the repeated
genetic detections of individuals to estimate various population parameters, including spatially-
explicit abundance (i.e., density) and vital rates (e.g., recruitment and survival). Importantly,
the approach accounts for imperfect detection during sampling (i.e., the fact that some indi-
viduals are not detected at all) and animal movement (i.e., the fact that individuals may use
and be detected in multiple management units or countries). The OPSCR method brings along
several advantages, including the ability to map density, derive jurisdiction-specific abundance,
estimate survival and recruitment (which are needed for making population projections), and
yield tractable measures of uncertainty (Bischof et al., 2019a, 2020b).
RovQuant reported abundance estimates for wolverines and wolves (Canis lupus) on an
annual basis (Bischof et al., 2019a,b, 2020b; Milleret et al., 2021b, 2022b,c; Flagstad et al., 2021).
During these and other analyses (Milleret et al., 2018, 2019; Bischof et al., 2020a; Dupont et al.,
2021; Turek et al., 2021; Dey et al., 2022), RovQuant has continuously improved the performance
of the OPSCR models. In the present report, we summarize the analysis of a 9-year time series
(2014–2022) using the latest available wolverine monitoring data (Kleven et al., 2022b) and the
most recent version of the OPSCR model. We provide the following information:
•Annual and sex-specific estimates of the number of wolverines for Sweden, Norway, and
both countries combined, as well as estimates by county in Sweden and by carnivore
management region in both countries.
•Annual maps of wolverine density throughout the species’ range in Scandinavia.
•Annual and sex-specific estimates of survival, cause-specific mortality, recruitment, and
population growth rate.
•Estimated proportion of individuals detected through non-invasive genetic sampling.
All estimates are accompanied by credible intervals.
6
Box 1: Terms and acronyms used
AC: Activity center. Model-based equivalent of the center of an individual’s home range during
the monitoring period. “AC location” refers to the spatial coordinates of an individual AC in a
given year and “AC movement” to the movement of an individual AC between consecutive years.
CrI: 95% credible interval associated with a posterior sample distribution.
Detectors: Potential detection locations in the spatial capture-recapture framework. These
can refer to fixed locations (e.g., camera-trap locations) or in this report to areas searched (e.g.,
habitat grid cells where searches for genetic samples were conducted). The searched area was
defined as a 90 km buffer around all NGS data collected during the period considered.
Statsforvalteren: Norwegian state’s representative in the county, responsible for following up
decisions, goals, and guidelines from the legislature and the government.
Habitat buffer: Buffer surrounding the searched area that is considered potentially suitable
habitat but was not searched (60km in this report).
Legal culling: Lethal removal of individuals by legal means, including licensed recreational
hunting, management removals, and defense of life and property.
Länsstyrelserna: Swedish County Administrative Boards, in charge of the monitoring of large
carnivores at the county level.
MCMC: Markov chain Monte Carlo.
NGS: Non-invasive genetic sampling.
OPSCR: Open-population spatial capture-recapture
p0:Baseline detection probability; probability of detecting an individual at a given detector, if
the individual’s AC is located exactly at the detector location.
σ: Scale parameter of the detection function; related to the size of the circular home-range.
SCR: Spatial capture-recapture.
SNO: Statens naturoppsyn (Norwegian Nature Inspectorate) is the operative field branch of the
Norwegian Environment Directorate (Miljødirektoratet).
RovQuant: Research group at the Norwegian University of Life Sciences (Ås, Norway) that
develops and applies OPSCR models.
7
2 Methods
2.1 Data
We included data from multiple sources, the primary one being the Scandinavian large carnivore
database Rovbase 3.0 (rovbase.se and rovbase.no; last extraction: 2022-10-28). This database
is used jointly by Norway and Sweden to record detailed information associated with large car-
nivore monitoring, including, but not limited to, NGS data, dead recoveries, and GPS search
tracks. In the following sections, we describe the various types of data used in the analysis. We
used data collected during nine consecutive monitoring seasons from 2014 to 2022.
Non-invasive genetic sampling In Norway, the collection of wolverine scat, urine, glandular
secretion, and hair is managed at the level of counties by SNO. Sample collection is conducted
by SNO field officers, wardens at Statskog Fjelltjenesten (statskog.no), wardens at Fjellstyrene
(fjellstyrene.no), local predator contacts, hunters and other members of the public. Rovdata
(rovdata.no), a unit within the Norwegian Institute for Nature Research, has responsibility
for the Norwegian large carnivore monitoring program. In Sweden, the collection of scat and
hair is managed by Länsstyrelserna at the regional level and carried out by field officers from
Länsstyrelserna. NGS collection was conducted primarily between December 1 and June 30 each
year. NGS data collected late in the monitoring season and suspected to be from cubs were not
included. This means that we only retained samples from individuals that were one year or
older. DNA was isolated with an extraction robot (Maxwell 16, KingFisher or QIAsymphony
instrument) and the samples were genotyped using 96 SNPs (Single Nucleotide Polymorphism)
on a Fluidigm platform for sex determination and individual identification. For further details
on the DNA analysis procedure see Flagstad et al. (2004), Flagstad et al. (2021), Kleven et al.
(2022a), and Kleven et al. (2022b).
Dead recoveries In Scandinavia, all large carnivores killed legally (e.g., legal hunting, manage-
ment kills, defense of life and property) have to be reported to the state authorities (Fylkesman-
nen or SNO in Norway and Länsstyrelserna or the police in Sweden). All wolverines found dead
due to other reasons (e.g., natural deaths, vehicle and train collisions, illegal hunting) also have
to be reported, but an unknown proportion remains undetected. Tissue is collected from all
reported dead carnivores for DNA extraction and analysis.
GPS search tracks Government employees involved in systematic searches for wolverine DNA
along roads and following wolverine tracks (via snowmobiles, skis, snowshoes, etc.) documented
their effort with GPS track logs, which were registered in Rovbase 3.0. GPS search tracks were
included in the OPSCR model to account for spatial and temporal variation in search effort
during NGS.
Observation reports in Skandobs We used all observation records in the Skandobs database
that were recorded during the wolverine monitoring seasons since 2012 (skandobs.se, skan-
dobs.no; last extraction: 2022-06-09). Skandobs is a web application that allows anyone to
anonymously register observations (visual, tracks, feces, etc.) of bears (Ursus arctos), lynx
(Lynx lynx), wolves, and wolverines in Scandinavia. This data currently consists of more than
20 000 records of possible large carnivore observations. Although most observations are not ver-
ified, they offer the best available proxy for spatio-temporal variation in opportunistic effort at
this time.
8
2.2 Open-population spatial capture-recapture model
We analysed the data using a Bayesian open-population spatial capture-recapture (OPSCR)
model (Bischof et al., 2019b), which addresses three challenges associated with population-level
wildlife inventories:
1. Detection is imperfect and sampling effort is heterogeneous in space and time: not all
individuals present in the study area are detected (Kéry and Schaub, 2012).
2. Individuals that reside primarily outside the surveyed area may be detected within it.
Without an explicit link between the population size parameter and the geographic area
the population occupies, density cannot be estimated and population size is ill-defined
(Efford, 2004).
3. Non-spatial population dynamic models usually estimate “apparent” survival and recruit-
ment, as these parameters include the probability of permanent emigration and immigra-
tion, respectively. By explicitly modelling movement of individuals between years, the
OPSCR model can help return unbiased estimates of demographic parameters (Ergon and
Gardner, 2014; Schaub and Royle, 2014; Gardner et al., 2018).
The OPSCR model is composed of three sub-models:
1. A model for population dynamics and population size.
2. A model for density and individual movement.
3. A model for detections during DNA searches.
Population dynamics and population size sub-model We used a multi-state formulation
(Lebreton and Pradel, 2002), where each individual’s life history is represented by a succession
of up to 3 discrete states zi,t: (1) “unborn” if the individual has not yet been recruited into
the population (state “unborn” is required for the data augmentation procedure, see below); (2)
“alive” if it is alive; (3) “dead” if it is dead. We then modelled the transition from one state to
another between consecutive monitoring seasons (tto t+ 1) to estimate vital rates (recruitment
and mortality). More details are available in Bischof et al. (2019b) and Bischof et al. (2020b).
This formulation of the population dynamic model means that, contrary to previous analyses
(Bischof et al., 2019b, 2020b; Flagstad et al., 2021; Milleret et al., 2022b), we did not use dead
recoveries or model cause-specific mortality directly in the OPSCR model. Cause-specific mor-
tality was instead derived after model fitting (see section "Other derived parameters"). We used
data augmentation (Royle and Dorazio, 2012), whereby additional, undetected individuals are
available for inclusion in the population at each time step.
Density and movement sub-model We used a Bernoulli point process to model the distribu-
tion of individual ACs (Zhang et al., 2022). In the first year, individuals were located according
to an intensity surface, which was a function of the locations of known dens at time t−1(see
Bischof et al., 2019b and Bischof et al., 2020b for more details). For all subsequent years (t > 1),
the location of individual ACs was a function of the distance from previous locations of ACs
(at time t−1) and the locations of known wolverine dens (at time t−1). Similar to the wolf
abundance estimation by Milleret et al. (2022c), we used an exponential model to describe the
movement of individuals between years, as it better accommodates distributions with long tails
(i.e., a few individuals that make exceptionally long dispersal movements).
Detection sub-model SCR models take into account the spatial variation in individual de-
tection probability based on the distance between AC locations (estimated by the density sub-
model) and a given detector. A half-normal function was used to express the declining proba-
bility of detection with increasing distance between the AC and the detector (Royle et al., 2013).
9
In Scandinavia, DNA material from live wolverines is collected following two main processes.
First, authorities collect genetic samples and record the corresponding search effort during of-
ficial searches ("structured sampling" thereafter). Second, DNA material can be collected by
any member of the public (e.g., hunters) or by the authorities in a more or less opportunistic
manner, which means that search effort is not directly available ("unstructured sampling" there-
after). Currently, it is not possible to unambiguously distinguish between samples collected
by the authorities during the structured or unstructured sampling in Rovbase. We therefore
assigned each sample to structured or unstructured sampling based on whether a given sample
matched in time and space with recorded search tracks: a sample was assigned to the "struc-
tured" sampling if it was collected by the authorities (marked as collected by "Statsforvalteren"
or "SNO" in Rovbase) and located within 500 m from a GPS search track recorded the same
day. All remaining samples were assigned to the unstructured sampling.
We assumed that both sampling processes could in theory occur within the entire study area
and therefore used the same 10 ×10 km detector grid for both observation processes. Samples
were then assigned to the closest detector (see details in Bischof et al., 2019b, and Bischof et al.,
2020b). However, spatial and temporal variation in the probability to detect a sample during
structured or unstructured sampling were assumed to be driven by different processes.
We accounted for spatial, temporal and individual heterogeneity in detectability during struc-
tured sampling using:
•Spatio-temporal variation in search effort represented by the length of GPS search tracks
in each detector grid cell.
•Spatio-temporal variation in snow cover during the monitoring period calculated as the av-
erage percentage of snow cover in each detector grid cell (MODIS at 0.1 degrees resolution,
https://cmr.earthdata.nasa.gov, accessed 2022-09-29).
•Spatio-temporal variation in monitoring regimes between jurisdictions (groups of counties
in Sweden, carnivore management regions in Norway, Figure A.6).
•Individual variation linked with a detection during the previous occasion that could be
expected to influence the probability of being detected at the next occasion.
We accounted for spatial, temporal, and individual heterogeneity in detectability during un-
structured sampling using:
•Spatio-temporal variation in unstructured sampling (Figure A.1). For each detector grid
cell and during each monitoring season (Dec 1 - Jun 30), we identified whether a) any car-
nivore sample had been registered in Rovbase (excluding successfully genotyped wolverine
samples already used in the OPSCR analysis) or b) any observation of carnivores had been
registered in Skandobs. Roughly, this binary variable distinguishes areas with very low
detection probability from those with a higher probability that carnivore DNA samples, if
present in a detector grid cell, could have been detected and submitted for genetic analysis
(Figure A.1).
•Spatio-temporal variation in snow cover during the monitoring period calculated as the av-
erage percentage of snow cover in each detector grid cell (MODIS at 0.1 degrees resolution,
https://cmr.earthdata.nasa.gov, accessed 2022-09-29).
•Spatial variation in accessibility measured as the average distance to the nearest road.
•Spatio-temporal variation between countries Figure A.7).
10
•Individual and temporal variation linked with a previous detection that could influence
the probability of being detected at subsequent occasions.
For years and areas without comprehensive sampling effort (i.e., Norrbotten county in Swe-
den in all years except 2017, 2018, and 2019), we removed all samples collected within the
county and fixed detection probability to 0 for both structured and unstructured sampling. The
different model components and data sources for covariates are described in detail in Bischof
et al. (2019a), Bischof et al. (2019b), and Bischof et al. (2020b).
Model fitting We fitted sex-specific Bayesian OPSCR models using MCMC simulation with
NIMBLE version 0.12.2 (de Valpine et al., 2017; Turek et al., 2021; de Valpine et al., 2022) and
RovQuant’s R package nimbleSCR version 0.2.0 (Bischof et al., 2021) in R version 4.1.0 (R Core
Team, 2021). We ran 4 chains each with 25 000 iterations, including a 10 000-iterations burn-in
period. Due to the computing challenge associated with post-processing large amounts of data,
we thinned chains by a factor of 10 before deriving abundance estimates. We considered models
as converged when the Gelman-Rubin diagnostics (Rhat, Gelman and Rubin, 1992) was ≤1.1
for all parameters and when mixing between chains was satisfactory based on visual inspection
of trace plots.
Abundance estimates To obtain an estimate of abundance for any given area, we summed
the number of predicted AC locations (individuals detected during sampling or predicted to be
alive by the model) that fell within that area for each iteration of the MCMC chains. This
produced a posterior distribution of abundance for that area. From such posteriors, abundance
estimates and the associated uncertainty can be extracted for any spatial unit, including coun-
tries, counties or management regions (Figure A.2). Individuals detected near a border can have
their model-predicted AC placed on different sides of that border in different model iterations
(even if detections are only made on one side of the border). As a result, the probability of des-
ignating such individuals to either side of the border can be integrated into jurisdiction-specific
abundance estimates. This is especially relevant for wolverines detected along the Swedish and
Norwegian border ("cross-boundary wolverines", Wabakken et al., 2022) as individual wolverines
can be partially designated to both countries (Bischof, 2015; Bischof et al., 2016).
To ensure that abundance estimates for spatial sub-units (jurisdictions) add up to the overall
abundance estimate, we used the mean and associated 95% credible interval limits to summarize
posterior distributions of abundance. Combined (female and male) parameter estimates were
obtained by merging posterior samples from the sex-specific models.
Density maps We used both the distribution of model-estimated AC positions and the scale
parameter (σ) of the detection function to construct density maps based on individual utilization
distributions. These maps are not only based on the position of the activity center of an indi-
vidual, but also take into account the area over which that individual’s activity is spread, i.e.,
its space use (Bischof et al., 2020b). To do so, we constructed raster maps (5 km resolution) of
individual utilization distributions, scaled values in each raster to sum to one, and then summed
rasters across individuals to create a single population-level raster map for each iteration. An
overall density map was derived by calculating the mean across iterations in each cell (Bischof
et al., 2020b). Note that this approach assumes circular home ranges of average size for all
individuals of a given sex and does not take into account individual variation in home-range size
and shape.
Other derived parameters We did not use dead recoveries and did not model cause-specific
mortalities explicitly. Instead, we used the posterior distribution of the number of individuals
that died between consecutive occasions, as estimated by the OPSCR model, and the recorded
11
number of legally culled individuals to derive cause-specific mortality estimates indirectly.
The average proportion of individuals detected and the associated uncertainty were obtained
by dividing the number of individuals detected through NGS each year (Table A.2) by corre-
sponding the abundance estimates and their associated credible intervals, respectively.
We derived the proportion of females in the population and the associated uncertainty by
dividing the estimated number of females by the total abundance for each iteration, thus gener-
ating a posterior distribution of the proportion of females from which the median and credible
interval could be derived (Table A.2). Yearly population growth rates (λ; Table A.5) were cal-
culated similarly as λt=Nt+1/Ntfor each iteration of the MCMC chains.
Focus on uncertainty Although we reported median (or mean for abundance; see above) es-
timates for all parameters in the tables, we intentionally focused the main results of our report
on the 95% credible interval limits of the estimates. We did so with the aim of drawing the
reader’s attention to the uncertainty around population size estimates, rather than a single point
estimate (Milleret et al., 2022b).
12
3 Results
3.1 Non-invasive genetic samples and dead recoveries
A total of 18 745 (8 418 female; 10 327 male) genotyped wolverine genetic samples were included
in the analysis, of which 43% originated from Sweden. These samples were associated with
2 550 (1360 female; 1190 male) individuals. We did not include individuals with unknown sex in
this analysis. Among all genotyped samples, 12 021 (5298 female; 6723 male) were assigned to
structured sampling and 6 724 (3120 female; 3604 male) to unstructured sampling. Annual total
and country-specific tallies of detections and associated individuals, as well as dead recoveries
are provided in the appendices (NGS samples: Table A.1, NGS individuals: Table A.2, dead
recoveries: Table A.3)
3.2 Density and abundance
Wolverine abundance for the entire study area (612 350 km2, excluding the buffer area) was likely
(95% credible interval) between 980 and 1088 individuals in 2022 (Table 1, Figure 1). Estimates
refer to the status of the population at the start of the annual sampling period (December 1).
The proportion of females in the Scandinavian wolverine population was likely between 59% and
63% in 2022. Based on the model-predicted location of ACs, we estimated that in 2022, between
625 and 709 individuals could be attributed to Sweden and 349 to 391 to Norway (Table 1). See
Table 1 for total and sex-specific estimates for each country and carnivore management region.
See Table A.4 for annual estimates for all of Scandinavia and by region between 2014 and 2022.
Note that estimates for different years (Figure A.3) shown here differ slightly from those pro-
vided in Bischof et al. (2020b) and Milleret et al. (2021a). This is due to the use of an updated
OPSCR model and the inclusion of an additional year of data. The analysis yielded annual
density maps, which illustrate changes in the distribution of wolverines over time (Figure A.4).
13
Table 1: Wolverine population size estimates in 2022 by sex at several spatial scales: the entire study area, by
country, by management unit (carnivore management regions in Norway and "Rovdjursförvaltningsområden" in
Sweden), and by counties (“Län” in Sweden); see Figure A.2 for a labelled map. Only counties and management
units that are within or that intersect the study area are included in the table. The percentage of the total
area of each unit included in the analysis is provided in the column "% Area". Readers should focus on the 95%
credible interval provided in parentheses, as these - unlike mean values - convey uncertainty inherent in abundance
estimates. Numbers are based on estimated AC locations of wolverines. Combined female-male estimates were
obtained by joining sex-specific posterior distributions. Rounding may result in small deviations between total
estimates and the sum of the estimates for constituent regions. Estimates for Norrbotten county in years without
comprehensive non-invasive genetic sampling were derived solely using the prediction from the OPSCR model
(shown in grey and marked with *). Total estimates in Sweden and for the entire study area that includes
estimates from Norrbotten without comprehensive NGS are shown in grey and marked with **.
Females Males Total % Area
TOTAL 632.9 (585-680)** 402.3 (380-427)** 1035.2 (980-1088)** 82
NORWAY 229.6 (212-251) 137.7 (130-148) 367.2 (349-391) 93
Region 1 6 (2-11) 4.5 (2-8) 10.4 (6-16) 84
Region 2 3 (0-7) 2.1 (0-5) 5.1 (1-9) 73
Region 3 16.9 (13-22) 10.1 (7-13) 27 (21-33) 100
Region 4 0.7 (0-3) 0.5 (0-2) 1.2 (0-4) 75
Region 5 41.2 (36-48) 39.8 (36-44) 81 (75-89) 100
Region 6 58.5 (52-66) 33.2 (29-37) 91.7 (84-100) 100
Region 7 44.4 (39-51) 22.3 (20-25) 66.7 (61-74) 100
Region 8 58.8 (51-70) 25.3 (21-30) 84.1 (75-96) 100
SWEDEN 403.3 (367-441)** 264.6 (245-286)** 667.9 (625-709)** 74
Norra 349 (315-383) 208.7 (188-231) 557.6 (517-598) 100
Jämtland 122.4 (109-136) 71.8 (65-79) 194.1 (180-211) 100
Norrbotten* 114.3 (92-136)* 73 (56-92)* 187.3 (158-216)* 97
Västerbotten 77.9 (67-90) 39.1 (34-45) 117 (104-129) 96
Västernorrland 34.4 (28-41) 24.8 (20-29) 59.2 (51-67) 100
Mellersta 54.1 (45-63) 55.8 (50-62) 109.9 (99-122) 68
Dalarna 28.2 (22-34) 26.8 (23-31) 54.9 (48-62) 100
Gävleborg 14.2 (10-18) 15.9 (13-20) 30.1 (25-36) 100
Örebro 1.7 (0-4) 1.2 (0-4) 2.8 (0-7) 100
Uppsala 0.1 (0-1) 0 (0-1) 0.1 (0-1) 5
Värmland 8.9 (6-13) 11.2 (8-15) 20.1 (16-25) 100
Västmanland 0.6 (0-2) 0.4 (0-2) 0.9 (0-3) 62
VästraGötaland 0.5 (0-2) 0.4 (0-2) 0.9 (0-3) 10
Södra 0.3 (0-2) 0.2 (0-1) 0.4 (0-2) 1
Östergötland 0.1 (0-1) 0.1 (0-1) 0.2 (0-2) 4
Södermanland 0.1 (0-1) 0.1 (0-1) 0.2 (0-1) 7
14
2022
1000 km
0
0.5
0.9
1.4
Individuals/100 km2
Figure 1: Wolverine density based on individual utilization distributions in Scandinavia in 2022. Note that
no comprehensive NGS was conducted in Norrbotten (polygon outlined in black) in 2022, which means that the
results are solely based on the OPSCR model prediction and assumption. The grey area represents areas that
were considered not searched and therefore were not included in the analysis. This map is freely available as a
geo-referenced raster file at https://github.com/richbi/RovQuantPublic
3.3 Vital rates
The OPSCR model produced annual estimates of mortality and per capita recruitment rates
(Figure 2; Table A.6). Mortality rates varied between years, with the risk of mortality from
other causes than culling generally higher than the risk of mortality from legal culling. Overall,
males had a higher mortality probability than females (Figure A.5).
15
0.0 0.2 0.4 0.6 0.8
Years
Mortality
0.0 0.2 0.4 0.6 0.8
2014 to
2015 2015 to
2016 2016 to
2017 2017 to
2018 2018 to
2019 2019 to
2020 2020 to
2021 2021 to
2022
0.0 0.2 0.4 0.6 0.8
Years
Mortality
0.0 0.2 0.4 0.6 0.8
2014 to
2015 2015 to
2016 2016 to
2017 2017 to
2018 2018 to
2019 2019 to
2020 2020 to
2021 2021 to
2022
Index
1
Index
1
Female
Index
1
Male
Index
1
Legal
culling
Other
mortality
Figure 2: Mortality probabilities due to legal culling (light green) and all other causes (dark green) for female
and male wolverines. Darker and lighter bars show the 50% and 95% credible intervals, respectively. Shown
are overall estimates throughout the study area. Estimates refer to deaths occurring between the start of one
sampling season and the start of the next.
0 50 100 150 200 250 300
Years
Number of recruits
2014 to
2015 2015 to
2016 2016 to
2017 2017 to
2018 2018 to
2019 2019 to
2020 2020 to
2021 2021 to
2022
0 50 100 150 200 250 300
Females
Males
Figure 3: Estimated annual number of male and female recruits in the Scandinavian wolverine population
between the start of one sampling season and the start of the next. Recruitment represents the number of new
individuals present in the population on Dec 1 (i.e., individuals that were born between the two consecutive
monitoring seasons and survived to Dec 1 or that immigrated into the study area). Darker and lighter bars show
the 50% and 95% credible intervals, respectively.
16
3.4 Detection probability
The overall proportion of detected individuals in the population was likely between 63% and
70% in 2022, and overall, larger in Norway than in Sweden (Table A.10). The baseline detec-
tion probability for the structured and unstructured sampling varied both in time and space
(Figure A.6 and Figure A.7). More specifically, the length of recorded search tracks positively
affected detection probability during structured sampling (2022; males: β= 0.45, CrI: 0.37 -
0.52; females: β= 0.42, CrI: 0.35 - 0.49; Table A.8). Detection of an individual during the
previous year and the average proportion of snow cover had no significant effect on detection
probability during structured sampling (Table A.8). The proxy for search effort during un-
structured searches, derived using the observation data in Skandobs and Rovbase, had a strong
positive effect on detection probability during unstructured sampling (2021/2022; males: β=
0.48, CrI: 0.22 - 0.76; females: β= 0.61, CrI: 0.29 - 0.94; Table A.9). Detection of an individual
during the previous year also tended to have a positive effect on detection probability during
unstructured sampling but the pattern was not consistent across years (Table A.9).
17
4 Summary of improvements made
The analysis described in this report includes the following adjustments compared with previous
analyses of wolverine density in Scandinavia by RovQuant (Bischof et al., 2020b; Flagstad et al.,
2021; Milleret et al., 2021b):
1. Addition of data from the 2022 monitoring season. We also excluded the 2013 winter
monitoring season, as this enhanced model performance.
2. Due to the uncertainty about the dead recovery process, we excluded dead recoveries from
this analysis.
3. Replaced the half-normal movement model with the exponential for modelling inter-annual
movement. We found in an ongoing study that the latter is better suited for capturing
the inter-annual movement patterns in populations that include long-distance dispersing
individuals.
4. Used separate detection sub-models for structured and unstructured sampling to account
for fundamental differences in how samples accumulate and better represent variation in
search effort.
5. Allowed for annual variation in all covariates in the detection sub-model (i.e., covariates
on p0and σare time-dependent).
6. Used carnivore observation reports in Skandobs and ancillary carnivore samples recorded
in Rovbase to generate a new detection covariate that serves as a proxy for sampling effort
(spatially and temporally varying) during unstructured sampling.
5 Suggestions for future improvements
RovQuant continues to work on improving the functionality and efficiency of OPSCR models.
We plan to test and potentially implement the following developments in future analyses of the
Scandinavian wolverine monitoring data:
1. Review and adjust spatial covariates on density. This may involve the addition of relevant
spatial variables (Moqanaki et al., 2022).
2. Account for individual heterogeneity in detection for example by using a finite-mixture
approach (Cubaynes et al., 2010).
3. Consider alternative detection models that do not assume a half-normal shape and/or
circular home ranges (Sutherland et al., 2015; Dey et al., 2022).
4. Account for spatial variation in survival (Milleret et al., 2022a).
6 Other recommendations
In addition, we suggest the following:
1. Indicate sample association with the search tracks in Rovbase (if any) to unambiguously
identify samples arising from structured vs unstructured sampling.
2. Consider full-coverage NGS in all regions for which estimates are desired (e.g., Norbotten
or reindeer (Rangifer tarandus) herding areas).
18
3. Report information about how samples are selected for DNA analysis.
4. Record coarse measures of search conditions at the search track level (e.g., presence/absence
of snow, days since last snowfall, experience level of searchers).
5. Unambiguously and consistently indicate the species targeted during searches when record-
ing GPS search tracks.
6. Clearly identify and delineate areas excluded from structured and unstructured sampling
and indicate the reason for exclusion (e.g., unable to search the area or low priority due
to assumed lack of presence of the target species).
7. Explore the feasibility of using station-based detectors (e.g., hair snares or similar) for
better control over the observation process.
7 Acknowledgements
This work was made possible by the large carnivore monitoring programs and the extensive
monitoring data collected by Swedish (Länstyrelsena) and Norwegian (SNO) wildlife manage-
ment authorities, as well as the public in both countries. Our analyses relied on genetic analy-
ses conducted by the laboratory personnel at the DNA laboratories at the Swedish University
of Agricultural Sciences, the Norwegian Institute for Nature Research, and Uppsala Univer-
sity. We also thank Swedish and Norwegian wildlife managers for feedback provided during
project RovQuant and the Research Council of Norway for partial funding (NFR 286886; project
WildMap). Computation was performed on resources provided by NMBU’s computing cluster
Orion, administered by the Centre for Integrative Genetics and by UNINETT Sigma2 - the
National Infrastructure for High Performance Computing and Data Storage in Norway. We are
grateful to P. de Valpine and D. Turek for help with the formulation of the OPSCR model in
Nimble. J. Vermaat provided helpful comments on a draft of this report.
8 Data availability
Data, R code to reproduce the analysis, as well as figures, tables, and raster maps (Figure A.4)
are available at https://github.com/richbi/RovQuantPublic.
19
References
Bischof, R. (2015). Approaches for assessing illegal hunting of brown bears and other large carnivores in Sweden.
Report 2015:5 from the Scandinavian Brown Bear Research Project to the Swedish Environmental Protection
Agency. Norwegian University of Life Sciences, Ås, Norway.
Bischof, R., Brøseth, H., and Gimenez, O. (2016). Wildlife in a Politically Divided World: Insularism Inflates
Estimates of Brown Bear Abundance. Conservation Letters, 9(2):122–130.
Bischof, R., Dupont, P., Milleret, C., Chipperfield, J., and Royle, J. A. (2020a). Consequences of ignoring group
association in spatial capture–recapture analysis. Wildlife Biology, 2020(1):1–10.
Bischof, R., Milleret, C., Dupont, P., Chipperfield, J., Åkesson, M., Brøseth, H., and Kindberg, J. (2019a).
Estimating the size of the Scandinavian wolf population with spatial capture recapture and conversion factors.
MINAfagrapport 57, 80pp.
Bischof, R., Milleret, C., Dupont, P., Chipperfield, J., Brøseth, H., and Kindberg, J. (2019b). Estimating density,
abundance and population dynamics of bears, wolverines, and wolves in Scandinavia. MINAfagrapport 63,
79pp.
Bischof, R., Milleret, C., Dupont, P., Chipperfield, J., Tourani, M., Ordiz, A., de Valpine, P., Turek, D., Royle,
J. A., Gimenez, O., Flagstad, , Åkesson, M., Svensson, L., Brøseth, H., and Kindberg, J. (2020b). Esti-
mating and forecasting spatial population dynamics of apex predators using transnational genetic monitoring.
Proceedings of the National Academy of Sciences, 117(48):30531–30538.
Bischof, R., Turek, D., Milleret, C., Ergon, T., Dupont, P., Dey, S., and de Valpine, P. (2021). nimbleSCR: Spatial
Capture-Recapture (SCR) Methods Using ’nimble’. R package version 0.1.2.
Chandler, R. B., Hepinstall-Cymerman, J., Merker, S., Abernathy-Conners, H., and Cooper, R. J. (2018). Char-
acterizing spatio-temporal variation in survival and recruitment with integrated population models. The Auk,
135(3):409–426.
Cubaynes, S., Pradel, R., Choquet, R., Duchamp, C., Gaillard, J. M., Lebreton, J. D., Marboutin, E., Miquel,
C., Reboulet, A. M., Poillot, C., Taberlet, P., and Gimenez, O. (2010). Importance of accounting for detection
heterogeneity when estimating abundance: the case of French wolves. Conservation Biology, 24(2):621–626.
de Valpine, P., Paciorek, C., Turek, D., Michaud, N., Anderson-Bergman, C., Obermeyer, F., Wehrhahn Cortes,
C., Rodrìguez, A., Temple Lang, D., and Paganin, S. (2022). NIMBLE User Manual. R package manual version
0.12.2.
de Valpine, P., Turek, D., Paciorek, C. J., Anderson-Bergman, C., Lang, D. T., and Bodik, R. (2017). Pro-
gramming with models: writing statistical algorithms for general model structures with NIMBLE. Journal of
Computational and Graphical Statistics, 26(2):403–413.
Dey, S., Bischof, R., Dupont, P. P. A., and Milleret, C. (2022). Does the punishment fit the crime? consequences
and diagnosis of misspecified detection functions in Bayesian spatial capture–recapture modeling. Ecology and
Evolution, 12(2):e8600.
Dupont, P., Milleret, C., Tourani, M., Brøseth, H., and Bischof, R. (2021). Integrating dead recoveries in open-
population spatial capture–recapture models. Ecosphere, 12(7):e03571.
Efford, M. (2004). Density estimation in live-trapping studies. Oikos, 106(3):598–610.
Ergon, T. and Gardner, B. (2014). Separating mortality and emigration: modelling space use, dispersal and
survival with robust-design spatial capture–recapture data. Methods in Ecology and Evolution, 5(12):1327–
1336.
Flagstad, , Hedmark, E., Landa, A., Brøseth, H., Persson, J., Andersen, R., SegerstrOm, P., and Ellegren, H.
(2004). Colonization history and noninvasive monitoring of a reestablished wolverine population. Conservation
Biology, 18(3):676–688.
Flagstad, , Kleven, O., Brandsegg, H., Spets, M., Eriksen, L., Andersskog, I., Johansson, M., Ekblom, R., Ellegren,
H., and Brøseth, H. (2021). DNA-basert overvåking av den skandinaviske jervebestanden 2020. Norsk institutt
for naturforskning(NINA), Trondheim, NINA Rapport nr 1956.
Gardner, B., Sollmann, R., Kumar, N. S., Jathanna, D., and Karanth, K. U. (2018). State space and movement
specification in open population spatial capture–recapture models. Ecology and Evolution, 8(20):10336–10344.
Gelman, A. and Rubin, D. (1992). Inference from iterative simulation using multiple sequences. Statistical Science,
7:457–511. http://www.stat.columbia.edu/~gelman/research/published/itsim.pdf.
Kéry, M. and Schaub, M. (2012). Bayesian population analysis using WinBUGS: a hierarchical perspective.
Academic Press, Waltham, MA.
Kleven, O., Andersskog, I., Brandsegg, H., Eriksen, L., Spets, M., Königsson, H., Spong, a., Milleret, C., Dupont,
P., Bischof, R., Flagstad, , and Brøseth, H. (2022a). DNA-basert overvåking av den skandinaviske jervebe-
standen 2021. Norsk institutt for naturforskning(NINA),Trondheim, NINA Rapport nr 2111.
Kleven, O., Forfang, K., Opsahl, N. N., Königsson, H., Spong, G., Milleret, C., Dupont, P., Bischof, R., Flagstad,
, and Brøseth, H. (2022b). DNA-based monitoring of the scandinavian wolverine population 2022. Norsk
institutt for naturforskning(NINA), Trondheim, NINA Rapport nr 2236.
20
Lebreton, J.-D. and Pradel, R. (2002). Multistate recapture models: modelling incomplete individual histories.
Journal of Applied Statistics, 29:353–369.
Milleret, C., Bischof, R., Dupont, P., Brøeth, H., Odden, J., and Mattisson, J. (2021a). GPS collars have an
apparent positive effect on the survival of a large carnivore. Biology Letters, 17(6):20210128.
Milleret, C., Dey, S., Dupont, P., Brøseth, H., Turek, D., de Valpine, P., and Bischof, R. (2022a). Estimating
spatially variable and density-dependent survival using open-population spatial capture–recapture models.
Ecology, n/a(n/a):e3934. e3934 ECY22-0819.
Milleret, C., Dupont, P., Bonenfant, C., Brøseth, H., Flagstad, O., Sutherland, C., and Bischof, R. (2019). A local
evaluation of the individual state-space to scale up Bayesian spatial capture recapture. Ecology and Evolution,
9(1):352–363.
Milleret, C., Dupont, P., Brøseth, H., Flagstad, O., Kindberg, J., and Bischof, R. (2022b). Estimates of wolverine
density, abundance, and population dynamics in Scandinavia, 2013–2021. MINAfagrapport 74, 30pp.
Milleret, C., Dupont, P., Brøseth, H., Kindberg, J., Royle, J. A., and Bischof, R. (2018). Using partial aggregation
in spatial capture recapture. Methods in Ecology and Evolution, 9(8):1896–1907.
Milleret, C., Dupont, P., Åkesson, M., Brøseth, H., Kindberg, J., and Bischof, R. (2021b). Estimates of wolf
density, abundance, and population dynamics in Scandinavia, 2012 - 2021. MINAfagrapport 72, 30pp.
Milleret, C., Dupont, P., Åkesson, M., Svensson, L., Brøseth, H., Kindberg, J., and Bischof, R. (2022c). Estimates
of wolf density, abundance, and population dynamics in Scandinavia, 2013–2022. MINAfagrapport 77, 35pp.
Moqanaki, E., Milleret, C., Dupont, P., Brøseth, H., and Bischof, R. (2022). Wolverine density distribution
reflects past persecution and current management in Scandinavia. bioRxiv.
R Core Team (2021). R: A Language and Environment for Statistical Computing. R Foundation for Statistical
Computing, Vienna, Austria.
Royle, J. A., Chandler, R. B., Sollmann, R., and Gardner, B. (2013). Spatial capture-recapture. Elsevier/Academic
Press, Amsterdam.
Royle, J. A. and Dorazio, R. M. (2012). Parameter-expanded data augmentation for Bayesian analysis of capture–
recapture models. Journal of Ornithology, 152(2):521–537.
Schaub, M. and Royle, J. A. (2014). Estimating true instead of apparent survival using spatial Cor-
mack–Jolly–Seber models. Methods in Ecology and Evolution, 5(12):1316–1326.
Sutherland, C., Fuller, A. K., and Royle, J. A. (2015). Modelling non-Euclidean movement and landscape
connectivity in highly structured ecological networks. Methods in Ecology and Evolution, 6(2):169–177.
Turek, D., Milleret, C., Ergon, T., Brøseth, H., Dupont, P., Bischof, R., and de Valpine, P. (2021). Efficient
estimation of large-scale spatial capture–recapture models. Ecosphere, 12(2):e03385.
Wabakken, P., Svensson, L., Maartmann, E., Nordli, K., Flagstad, O., and Åkesson, M. (2022). Bestandsovervåk-
ing av ulv vinteren 2021-2022. inventering av varg vintern 2021-2022. Bestandsstatus for store rovdyr i Skan-
dinavia. Beståndsstatus för stora rovdjur i Skandinavien 1-2022, 1(59s).
Zhang, W., Chipperfield, J. D., Illian, J. B., Dupont, P., Milleret, C., de Valpine, P., and Bischof, R. (2022).
A flexible and efficient bayesian implementation of point process models for spatial capture-recapture data.
Ecology, page e3887. e3887.
21
Appendices
22
2014 2015 2016 2017 2018
2019 2020 2021 2022
Figure A.1: Covariate used to account for spatio-temporal variation in unstructured sampling in the study area.
Green cells (10 ×10 km) represent areas with at least one carnivore record from Rovbase (rovbase.no, rovbase.se,
excluding the wolverine samples used in the OPSCR model) or an observation record from Skandobs (skandobs.se,
skandobs.no) during a given monitoring season (Dec 1 – Jun 30).
23
Figure A.2: Management jurisdictions in Norway and Sweden. Polygons represent carnivore management
regions in Norway and counties in Sweden. Thick outlines delineate Swedish carnivore management regions
("Rovdjursförvaltningsområden") encompassing multiple counties.
24
0
200
400
600
800
1000
1200
Estimated number of wolverines
2014 2015 2016 2017 2018 2019 2020 2021 2022
Norway Sweden Total
Figure A.3: Total (black) and country-specific (blue: Sweden, red: Norway) annual wolverine population size
estimates in Scandinavia between 2014 and 2022. Darker and lighter bars show the 50% and 95% credible intervals,
respectively. Credible intervals indicate uncertainty in estimates given the model and data used to generate the
estimates. Changes in the model and data can result in different estimates and associated uncertainty compared
with estimates provided in previous reports by RovQuant.
25
2014 2015 2016 2017 2018
2019 2020 2021 2022
1000 km
0
0.5
0.9
1.4
Individuals/100 km2
Figure A.4: Wolverine density based on individual utilization distributions in Scandinavia between 2014 and 2022. Note that no comprehensive NGS was conducted in
Norrbotten (polygon outlined in black) from 2014-2016 and from 2020-2022, which means that the results are solely based on the OPSCR model prediction and assumption.
The grey area represents areas that were considered not searched and therefore were not included in the analysis. These maps are freely available as geo-referenced raster files
at https://github.com/richbi/RovQuantPublic.
26
0.0 0.2 0.4 0.6 0.8 1.0
Years
Survival
0.0 0.2 0.4 0.6 0.8 1.0
2014 to
2015 2015 to
2016 2016 to
2017 2017 to
2018 2018 to
2019 2019 to
2020 2020 to
2021 2021 to
2022
Index
1
Females
Males
Figure A.5: Annual survival probabilities for female and male wolverines. Darker and lighter bars show the
50% and 95% credible intervals, respectively. Shown are overall estimates for the entire study area between 2014
and 2022.
27
0.00 0.01 0.02 0.03 0.04
Years
Detection probability
0.00 0.01 0.02 0.03 0.04
2014 2015 2016 2017 2018 2019 2020 2021 2022
NO2
0.00 0.01 0.02 0.03 0.04
Years
Detection probability
0.00 0.01 0.02 0.03 0.04
2014 2015 2016 2017 2018 2019 2020 2021 2022
SE1
0.00 0.01 0.02 0.03 0.04
Years
Detection probability
0.00 0.01 0.02 0.03 0.04
2014 2015 2016 2017 2018 2019 2020 2021 2022
NO3
0.00 0.01 0.02 0.03 0.04
Years
Detection probability
0.00 0.01 0.02 0.03 0.04
2014 2015 2016 2017 2018 2019 2020 2021 2022
SE2
0.00 0.01 0.02 0.03 0.04
Years
Detection probability
0.00 0.01 0.02 0.03 0.04
2014 2015 2016 2017 2018 2019 2020 2021 2022
SE3
0.00 0.01 0.02 0.03 0.04
Years
Detection probability
0.00 0.01 0.02 0.03 0.04
2014 2015 2016 2017 2018 2019 2020 2021 2022
Female
Male
NO1
0.00 0.01 0.02 0.03 0.04
Years
Detection probability
0.00 0.01 0.02 0.03 0.04
2014 2015 2016 2017 2018 2019 2020 2021 2022
NO4
0.00 0.01 0.02 0.03 0.04
Years
Detection probability
0.00 0.01 0.02 0.03 0.04
2014 2015 2016 2017 2018 2019 2020 2021 2022
NO5
Figure A.6: Sex-specific baseline detection probabilities (p0structured ) for the different Scandinavian jurisdictions
during structured sampling as estimated by the open-population spatial capture-recapture (OPSCR) model. Bars
represent 95% credible intervals. Results are separated into panels based on regions. Estimates are shown for
the mean values of the detection covariates. Note that baseline detection probability (p0) is a theoretical value of
detection probability when a detector coincides with the location of an individual’s AC; it is not to be confused
with detectability, i.e, the overall probability of detecting an individual.
28
0.000 0.005 0.010 0.015
Years
Detection probability
0.000 0.005 0.010 0.015
2014 2015 2016 2017 2018 2019 2020 2021 2022
Female
Male
NOR
0.000 0.005 0.010 0.015
Years
Detection probability
0.000 0.005 0.010 0.015
2014 2015 2016 2017 2018 2019 2020 2021 2022
SWE
Figure A.7: Sex- and country-specific baseline detection probabilities (p0unstructur ed) of wolverines during
unstructured sampling as estimated by the open-population spatial capture-recapture (OPSCR) model. Bars
represent 95% credible intervals. Estimates are shown for the mean values of the detection covariates. Note that
baseline detection probability (p0) is a theoretical value of detection probability when a detector coincides with
the location of an individual’s activity center; it is not to be confused with detectability, i.e, the overall probability
of detecting an individual.
29
2014
2015
2016
2017
2018
2019
2020
2021
2022
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
Norw
a
y
521
578
412
445
468
579
606
670
462
747
592
731
575
684
639
739
568
747
Sw
eden
186
154
228
221
236
275
487
552
636
839
504
618
355
538
460
562
483
648
T
otal
707
732
640
666
704
854
1093
1222
1098
1586
1096
1349
930
1222
1099
1301
1051
1395
Table A.1: Annual number of wolverine non-invasive genetic samples included in the analysis. Numbers are reported by country, for females (F) and males (M), and for each
type of sampling (structured and unstructured). We included only samples collected within the study area during the primary monitoring period (Dec 1 - June 30) between
2014 and 2022.
2014 2015 2016 2017 2018 2019 2020 2021 2022
F M F M F M F M F M F M F M F M F M
Norway Structured 341 375 244 247 304 378 377 423 277 471 377 506 379 453 390 480 411 520
Unstructured 180 203 168 198 164 201 229 247 185 276 215 225 196 231 249 259 157 227
Sweden Structured 87 102 131 128 131 158 276 319 343 509 367 477 250 351 286 351 327 475
Unstructured 99 52 97 93 105 117 211 233 293 330 137 141 105 187 174 211 156 173
Total Structured 428 477 375 375 435 536 653 742 620 980 744 983 629 804 676 831 738 995
Unstructured 279 255 265 291 269 318 440 480 478 606 352 366 301 418 423 470 313 400
Table A.2: Annual number of individual wolverines detected via non-invasive genetic sampling and included in the analysis. Numbers are reported by country, for females
(F) and males (M), and for each type of sampling (structured and unstructured). We included only individuals associated with samples collected within the study area during
the primary monitoring period (Dec 1 - Jun 30) between 2013/2014 and 2021/2022. Some individuals were detected in both countries during the same year, hence the sum of
the national counts can exceed the total number of unique individuals detected in Scandinavia.
2014 2015 2016 2017 2018 2019 2020 2021 2022
F M F M F M F M F M F M F M F M F M
Norway Structured 141 109 111 90 133 96 143 100 127 105 147 124 154 121 149 112 161 108
Unstructured 102 91 93 78 106 91 109 100 103 87 108 93 107 97 102 91 95 93
Sweden Structured 56 50 81 61 74 80 156 131 165 172 187 176 131 130 155 134 154 157
Unstructured 45 33 57 52 49 60 115 119 143 128 89 84 72 88 98 91 72 92
Total Structured 196 158 188 148 206 175 287 217 286 264 327 291 282 248 301 241 314 257
Unstructured 145 124 149 126 155 148 220 208 243 207 196 176 178 182 200 179 167 184
30
Table A.3: Number of cause-specific dead recoveries of wolverines in Scandinavia between 2014 and 2022. Numbers are reported by country, for females (F) and males (M).
Dead recovery data was only used to derive cause-specific mortality (Figure 2).
Country 2014 2015 2016 2017 2018 2019 2020 2021 2022
F M F M F M F M F M F M F M F M F M
Other Norway150231327113334410
Sweden131202010111011121
Legal culling Norway 42 34 69 52 42 43 47 61 27 31 57 37 49 63 44 44 28 35
Sweden 13 8 22 18 6 9 5 2 2 5 6 3 7 19 3 10 5 1
Total Total 57 50 92 74 51 55 55 66 36 38 65 44 59 86 52 59 36 37
31
Table A.4: Annual abundance estimates for wolverine at several spatial scales: the entire study area, by country, by management unit (carnivore management regions in
Norway and "Rovdjursförvaltningsområden" in Sweden), and counties (“Län” in Sweden); see also Figure A.2. Only counties and management units that are within or that
intersect the study area are included in the table. Estimates are based on model-estimated activity center locations. Credible intervals (95%) are shown in parentheses. Small
deviations between the total estimate and the sum of abundance estimates from the constituent sub-regions may arise due to rounding. Note that the numbers reported here
are predictions from a statistical model which always represents an oversimplification of reality and is based on available data (NGS). As a consequence, especially at the local
scale, the model-estimated number of wolverines based on DNA sampling can deviate from the number of wolverines inferred from ancillary observations (e.g., camera traps).
Estimates for Norrbotten county in years without comprehensive NGS were derived solely using the prediction from the OPSCR model (shown in grey and marked with *).
Total estimates in Sweden and for the entire study area that includes estimates from Norrbotten without comprehensive NGS are marked with **.
2014 2015 2016 2017 2018 2019 2020 2021 2022
TOTAL 970.3 (887-1065)** 827.6 (777-886)** 924.9 (876-975)** 905.6 (877-941) 952.6 (925-983) 1063 (1027-1101) 1077.6 (1033-1124)** 1068.4 (1028-1115)** 1035.2 (980-1088)**
NORWAY 427.7 (390-467) 351 (328-376) 379.7 (357-403) 355 (340-373) 360 (344-376) 411.8 (394-430) 400.6 (383-420) 368.8 (352-388) 367.2 (349-391)
Region 1 19.5 (12-27) 11.2 (7-16) 9.2 (5-14) 7.3 (4-12) 6.1 (3-10) 5.9 (2-10) 5.3 (2-10) 6.3 (2-11) 10.4 (6-16)
Region 2 8 (3-14) 3 (0-7) 3.1 (0-6) 2.7 (1-6) 2 (0-5) 2.2 (0-6) 2.3 (0-6) 4.3 (1-8) 5.1 (1-9)
Region 3 35.8 (29-43) 26.8 (22-32) 26.6 (22-32) 23.4 (19-28) 22.5 (18-27) 27.9 (23-33) 30.4 (26-35) 32.7 (28-38) 27 (21-33)
Region 4 2.6 (0-6) 1.1 (0-3) 0.8 (0-3) 0.7 (0-3) 0.5 (0-2) 0.6 (0-2) 0.9 (0-3) 0.8 (0-3) 1.2 (0-4)
Region 5 62.9 (55-72) 53.9 (47-62) 63.4 (57-70) 63.3 (57-70) 76.5 (70-84) 91.7 (85-98) 88 (81-96) 82.4 (75-90) 81 (75-89)
Region 6 82.9 (71-96) 70.5 (61-81) 80.7 (72-91) 79.8 (73-88) 90.6 (83-99) 102.3 (94-111) 85.4 (77-95) 79.2 (71-87) 91.7 (84-100)
Region 7 78.5 (69-89) 71.2 (64-79) 82.3 (75-90) 70.1 (64-76) 66.9 (62-74) 78.5 (72-86) 88.2 (81-95) 69.4 (64-75) 66.7 (61-74)
Region 8 137.5 (123-152) 113.3 (103-125) 113.5 (103-124) 107.7 (101-115) 94.9 (87-103) 102.6 (94-112) 100 (93-109) 93.7 (85-103) 84.1 (75-96)
SWEDEN 542.6 (484-608)** 476.7 (440-521)** 545.2 (506-581)** 550.5 (525-579) 592.6 (571-616) 651.2 (620-683) 677 (635-719)** 699.6 (663-740)** 667.9 (625-709)**
Norra 476.2 (427-536) 425 (391-464) 484.2 (448-519) 481.6 (460-505) 509.5 (490-530) 551.8 (523-580) 572.1 (536-610) 594.2 (559-630) 557.6 (517-598)
Jämtland 147.8 (126-171) 139.4 (125-155) 168.3 (155-184) 172.1 (159-187) 191.9 (180-205) 211.7 (198-227) 215.1 (200-230) 220.6 (207-233) 194.1 (180-211)
Norrbotten 181.6 (153-213)* 158 (134-182)* 173 (151-196)* 170.5 (159-183) 173.8 (163-185) 171 (154-188) 181.6 (158-206)* 188.8 (162-214)* 187.3 (158-216)*
Västerbotten 126.3 (112-144) 109.5 (100-121) 122.8 (111-135) 113.9 (104-124) 109 (99-119) 128.9 (118-141) 130.9 (120-143) 135.5 (124-146) 117 (104-129)
Västernorrland 20.6 (13-30) 18.1 (12-25) 20.1 (14-27) 25.1 (19-32) 34.8 (29-41) 40.1 (33-47) 44.5 (37-53) 49.3 (42-57) 59.2 (51-67)
Mellersta 65.8 (51-82) 51.2 (40-64) 60.6 (50-71) 68.6 (58-78) 82.8 (74-92) 99.1 (90-109) 104.6 (94-117) 105 (93-116) 109.9 (99-122)
Dalarna 30.4 (22-41) 26.3 (20-33) 30.8 (24-38) 35.4 (29-42) 46.4 (41-52) 48.4 (42-56) 50.1 (42-58) 53.2 (47-61) 54.9 (48-62)
Gävleborg 14.6 (7-22) 12.4 (7-18) 15 (10-21) 17.4 (12-24) 21.5 (17-27) 35.2 (30-41) 35.2 (30-42) 32.1 (27-38) 30.1 (25-36)
Örebro 4.9 (1-10) 2.5 (0-6) 2.1 (0-5) 1.7 (0-5) 2 (0-5) 1.5 (0-4) 2.5 (0-6) 2.7 (0-6) 2.8 (0-7)
Uppsala 0.3 (0-2) 0.1 (0-1) 0.2 (0-1) 0.1 (0-1) 0.1 (0-1) 0.2 (0-1) 0.1 (0-1) 0.1 (0-1) 0.1 (0-1)
Värmland 12 (7-19) 7.8 (4-13) 10.7 (7-15) 12.5 (9-17) 11.5 (8-15) 12.4 (9-17) 14.8 (10-20) 15.2 (11-20) 20.1 (16-25)
Västmanland 1.9 (0-5) 1.1 (0-3) 0.9 (0-3) 0.7 (0-3) 0.6 (0-3) 0.6 (0-2) 0.9 (0-3) 0.9 (0-3) 0.9 (0-3)
VästraGötaland 1.7 (0-5) 1 (0-3) 0.9 (0-3) 0.8 (0-3) 0.6 (0-2) 0.7 (0-3) 0.9 (0-3) 0.8 (0-3) 0.9 (0-3)
Södra 0.6 (0-2) 0.4 (0-2) 0.4 (0-2) 0.4 (0-2) 0.3 (0-2) 0.4 (0-2) 0.4 (0-2) 0.4 (0-2) 0.4 (0-2)
Östergötland 0.3 (0-2) 0.2 (0-1) 0.2 (0-1) 0.2 (0-1) 0.1 (0-1) 0.2 (0-1) 0.2 (0-1) 0.2 (0-1) 0.2 (0-2)
Södermanland 0.3 (0-2) 0.2 (0-1) 0.2 (0-1) 0.2 (0-1) 0.1 (0-1) 0.2 (0-1) 0.2 (0-1) 0.2 (0-2) 0.2 (0-1)
32
Table A.5: Annual population growth rate estimates for the wolverine population in Scandinavia ("Total") and separately for Norway and Sweden. Estimates were derived
using the posterior distributions of annual abundance estimates (Table A.4). Credible intervals (95%) are shown in parentheses.
2014-2015 2015-2016 2016-2017 2017-2018 2018-2019 2019-2020 2020-2021 2021-2022
Norway 0.82 (0.75-0.90) 1.08 (0.99-1.16) 0.94 (0.88-1.00) 1.01 (0.95-1.08) 1.14 (1.07-1.22) 0.97 (0.92-1.03) 0.92 (0.87-0.98) 1.00 (0.93-1.07)
Sweden 0.88 (0.79-0.98) 1.15 (1.05-1.26) 1.01 (0.94-1.10) 1.08 (1.02-1.14) 1.10 (1.04-1.17) 1.04 (0.98-1.12) 1.03 (0.97-1.10) 0.96 (0.89-1.02)
Total 0.85 (0.78-0.94) 1.12 (1.04-1.20) 0.98 (0.92-1.04) 1.05 (1.01-1.10) 1.12 (1.06-1.17) 1.01 (0.96-1.07) 0.99 (0.95-1.04) 0.97 (0.91-1.03)
2014-2015
2015-2016
2016-2017
2017-2018
M
F
M
F
M
F
M
F
ρ
0.26
(0.17-0.37)
0.15
(0.06-0.26)
0.44
(0.33-0.54)
0.35
(0.25-0.47)
0.35
(0.27-0.43)
0.19
(0.11-0.26)
0.41
(0.35-0.46)
0.27
(0.21-0.32)
ϕ
0.56
(0.49-0.64)
0.71
(0.64-0.77)
0.64
(0.57-0.71)
0.78
(0.72-0.83)
0.65
(0.59-0.71)
0.77
(0.71-0.82)
0.66
(0.60-0.71)
0.76
(0.72-0.81)
h
0.06
(0.06-0.07)
0.06
(0.06-0.07)
0.15
(0.14-0.17)
0.10
(0.09-0.11)
0.11
(0.10-0.12)
0.06
(0.
05-
0.
06)
0.12
(0.11-0.13)
0.06
(0.06-0.06)
w
0.37
(0.31-0.43)
0.23
(0.17-0.28)
0.20
(0.15-0.26)
0.12
(0.07-0.17)
0.23
(0.19-0.28)
0.18
(0.13-0.22)
0.22
(0.19-0.25)
0.17
(0.14-
0.
21)
Table A.6: Estimates of the demographic parameters obtained from the sex-specific wolverine open-population spatial capture-recapture (OPSCR) models. Parameters
represent transition rates from Dec 1 to Nov 30 in the following year. Median estimates and 95% credible intervals (in parentheses) for per capita recruitment rate (ρ), survival
(ϕ), mortality due to legal culling (h) and mortality due to other causes (w) are presented for males (M) and females (F). Note that mortality due to legal culling was not
estimated in the model, but derived from the recorded number of dead recoveries.
2014-2015 2015-2016 2016-2017 2017-2018
M F M F M F M F
ρ0.26 (0.17-0.37) 0.15 (0.06-0.26) 0.44 (0.33-0.54) 0.35 (0.25-0.47) 0.35 (0.27-0.43) 0.19 (0.11-0.26) 0.41 (0.35-0.46) 0.27 (0.21-0.32)
ϕ0.56 (0.49-0.64) 0.71 (0.64-0.77) 0.64 (0.57-0.71) 0.78 (0.72-0.83) 0.65 (0.59-0.71) 0.77 (0.71-0.82) 0.66 (0.60-0.71) 0.76 (0.72-0.81)
h 0.06 (0.06-0.07) 0.06 (0.06-0.07) 0.15 (0.14-0.17) 0.10 (0.09-0.11) 0.11 (0.10-0.12) 0.06 (0.05-0.06) 0.12 (0.11-0.13) 0.06 (0.06-0.06)
w 0.37 (0.31-0.43) 0.23 (0.17-0.28) 0.20 (0.15-0.26) 0.12 (0.07-0.17) 0.23 (0.19-0.28) 0.18 (0.13-0.22) 0.22 (0.19-0.25) 0.17 (0.14-0.21)
2018-2019 2019-2020 2020-2021 2021-2022
M F M F M F M F
ρ0.41 (0.36-0.48) 0.29 (0.24-0.36) 0.38 (0.32-0.44) 0.25 (0.19-0.32) 0.35 (0.30-0.41) 0.24 (0.18-0.30) 0.37 (0.32-0.44) 0.20 (0.15-0.25)
ϕ0.69 (0.64-0.75) 0.82 (0.77-0.87) 0.62 (0.57-0.68) 0.76 (0.70-0.81) 0.61 (0.56-0.67) 0.76 (0.70-0.82) 0.61 (0.55-0.67) 0.75 (0.68-0.83)
h 0.08 (0.07-0.08) 0.04 (0.04-0.04) 0.08 (0.07-0.08) 0.07 (0.06-0.07) 0.15 (0.14-0.16) 0.06 (0.05-0.06) 0.09 (0.09-0.10) 0.05 (0.05-0.05)
w 0.23 (0.20-0.26) 0.14 (0.10-0.18) 0.30 (0.27-0.34) 0.18 (0.13-0.22) 0.23 (0.19-0.27) 0.18 (0.13-0.23) 0.29 (0.26-0.33) 0.20 (0.13-0.26)
33
Table A.7: Estimates of the density and movement process parameters obtained from the sex-specific wolverine open-population spatial capture-recapture (OPSCR) models.
βdens represents the effect of the number of known wolverine dens on activity center locations (Bischof et al., 2020b). The scale parameter σof the detection function is
expressed in kilometers and estimated separately for each year. λ(in km) represents the mean of the exponential movement parameter, describing individual movement
distances between years. Credible intervals (95%) are shown in parentheses. Parameters that were not estimated separately each year are marked with ∗.
2014 2015 2016 2017 2018
M F M F M F M F M F
β∗
dens 0.44 (0.40-0.47) 0.48 (0.45-0.52) 0.44 (0.40-0.47) 0.48 (0.45-0.52) 0.44 (0.40-0.47) 0.48 (0.45-0.52) 0.44 (0.40-0.47) 0.48 (0.45-0.52) 0.44 (0.40-0.47) 0.48 (0.45-0.52)
σ8.33 (7.91-8.79) 5.86 (5.52-6.25) 8.48 (8.05-8.93) 5.63 (5.32-5.98) 8.23 (7.86-8.63) 6.09 (5.77-6.45) 8.39 (8.08-8.73) 6.54 (6.27-6.82) 8.00 (7.74-8.26) 6.24 (5.98-6.51)
λ∗14.63 (14.02-15.35) 7.99 (7.69-8.31) 14.63 (14.02-15.35) 7.99 (7.69-8.31) 14.63 (14.02-15.35) 7.99 (7.69-8.31) 14.63 (14.02-15.35) 7.99 (7.69-8.31) 14.63 (14.02-15.35) 7.99 (7.69-8.31)
2019 2020 2021 2022
M F M F M F M F
β∗
dens 0.44 (0.40-0.47) 0.48 (0.45-0.52) 0.44 (0.40-0.47) 0.48 (0.45-0.52) 0.44 (0.40-0.47) 0.48 (0.45-0.52) 0.44 (0.40-0.47) 0.48 (0.45-0.52)
σ7.73 (7.45-8.03) 5.63 (5.39-5.88) 7.98 (7.68-8.31) 6.25 (5.96-6.55) 8.09 (7.80-8.40) 6.30 (6.02-6.60) 8.07 (7.79-8.38) 5.70 (5.43-5.99)
λ∗14.63 (14.02-15.35) 7.99 (7.69-8.31) 14.63 (14.02-15.35) 7.99 (7.69-8.31) 14.63 (14.02-15.35) 7.99 (7.69-8.31) 14.63 (14.02-15.35) 7.99 (7.69-8.31)
Table A.8: Estimates of the detection process parameters for the structured sampling. β1Structured corresponds to the effect of previous detection of an individual, β2Str uctured
to the effect of search-effort (track length), and β3Structured to the effect of average snow cover during the monitoring period on baseline detection probability (p0Structured ).
Coefficients are associated with scaled covariates. Credible intervals (95%) are shown in parentheses.
2014 2015 2016 2017 2018
M F M F M F M F M F
β1Structured 0.91 (0.66-1.16) 0.61 ( 0.35-0.87) 0.50 (0.23-0.77) 0.03 (-0.24-0.31) 0.53 (0.31-0.76) 0.27 ( 0.01-0.52) 0.45 (0.26-0.64) 0.04 (-0.16-0.23) 0.54 (0.37-0.71) 0.11 (-0.09-0.31)
β2Structured 0.48 (0.40-0.56) 0.39 (0.32-0.47) 0.44 (0.35-0.54) 0.43 (0.34-0.53) 0.43 (0.36-0.51) 0.49 (0.41-0.59) 0.44 (0.37-0.50) 0.39 (0.32-0.45) 0.42 (0.36-0.48) 0.41 (0.34-0.48)
β3Structured 0.43 ( 0.13-0.75) 0.50 ( 0.19-0.83) -0.02 (-0.23-0.19) 0.20 (-0.02-0.42) 0.22 ( 0.00-0.43) 0.10 (-0.13-0.33) 0.32 ( 0.14-0.50) 0.20 ( 0.02-0.37) 0.12 ( 0.00-0.24) 0.04 (-0.09-0.18)
2019 2020 2021 2022
M F M F M F M F
β1Structured 0.57 (0.39-0.75) 0.14 (-0.06-0.33) 0.17 (0.00-0.36) -0.01 (-0.21-0.19) 0.48 (0.29-0.67) -0.15 (-0.36-0.06) 0.42 (0.23-0.60) 0.06 (-0.15-0.28)
β2Structured 0.47 (0.40-0.55) 0.49 (0.41-0.57) 0.43 (0.37-0.49) 0.46 (0.39-0.53) 0.50 (0.43-0.58) 0.55 (0.48-0.63) 0.44 (0.37-0.52) 0.42 (0.35-0.49)
β3Structured -0.09 (-0.24-0.06) -0.06 (-0.22-0.12) 0.34 ( 0.16-0.52) 0.33 ( 0.10-0.56) 0.13 (-0.01-0.27) 0.24 ( 0.08-0.40) 0.08 (-0.07-0.22) 0.08 (-0.08-0.24)
34
Table A.9: Estimates of the detection process parameters for the unstructured sampling. β1U nstructur ed corresponds to the effect of previous detection, β2U nstructured to
the effect of distance to the nearest road, β3Unstr uctured to the effect of average snow cover during the monitoring period, and β4Unstructur ed to the effect of spatio-temporal
heterogeneity in unstructured sampling derived using the observation data in Skandobs and Rovbase on baseline detection probablity (p0Unstructur ed). Coefficients are
associated with scaled covariates. Credible intervals (95%) are shown in parentheses.
2014 2015 2016 2017 2018
M F M F M F M F M F
β1Unstr uctured 0.73 ( 0.43-1.03) 0.41 ( 0.10-0.73) 0.16 (-0.13-0.45) 0.13 (-0.19-0.43) 0.56 ( 0.29-0.83) 0.37 ( 0.06-0.67) 0.25 ( 0.03-0.46) -0.16 (-0.39-0.07) 0.85 ( 0.64-1.06) 0.22 ( 0.00-0.45)
β2Unstr uctured 0.56 ( 0.25-0.88) 0.61 ( 0.31-0.94) 0.29 ( 0.09-0.51) 0.44 ( 0.22-0.67) -0.03 (-0.22-0.17) 0.30 ( 0.08-0.52) 0.42 ( 0.23-0.61) 0.36 ( 0.17-0.56) 0.18 ( 0.04-0.32) 0.14 (-0.01-0.30)
β3Unstr uctured 0.09 (-0.08-0.26) 0.18 ( 0.02-0.35) 0.02 (-0.17-0.21) 0.13 (-0.05-0.31) 0.04 (-0.15-0.22) -0.12 (-0.32-0.06) 0.13 ( 0.04-0.21) 0.15 ( 0.06-0.23) 0.02 (-0.07-0.10) 0.04 (-0.04-0.13)
β4Unstr uctured 0.74 (0.44-1.04) 0.53 (0.24-0.83) 0.48 (0.20-0.76) 0.73 (0.41-1.04) 0.77 (0.50-1.05) 0.30 (0.01-0.59) 0.58 (0.36-0.80) 0.55 (0.32-0.78) 0.61 (0.40-0.83) 0.82 (0.58-1.07)
2019 2020 2021 2022
M F M F M F M F
β1Unstr uctured 0.55 ( 0.29-0.81) 0.08 (-0.17-0.35) 0.32 ( 0.09-0.56) -0.05 (-0.32-0.22) 0.63 ( 0.40-0.86) 0.40 ( 0.14-0.66) 0.88 ( 0.63-1.15) 0.14 (-0.14-0.43)
β2Unstr uctured 0.06 (-0.14-0.26) 0.20 (-0.01-0.41) 0.29 ( 0.09-0.50) 0.61 ( 0.32-0.93) 0.18 ( 0.02-0.34) 0.01 (-0.17-0.20) 0.19 ( 0.01-0.37) 0.04 (-0.17-0.24)
β3Unstr uctured 0.02 (-0.11-0.15) -0.02 (-0.15-0.10) -0.11 (-0.30-0.06) -0.18 (-0.38-0.00) -0.12 (-0.29-0.05) -0.06 (-0.24-0.10) -0.14 (-0.34-0.04) -0.12 (-0.33-0.08)
β4Unstr uctured 0.77 (0.50-1.06) 0.88 (0.59-1.18) 0.36 (0.12-0.61) 0.36 (0.09-0.64) 0.57 (0.33-0.84) 0.80 (0.51-1.11) 0.48 (0.21-0.76) 0.61 (0.29-0.95)
Table A.10: Average proportion of individuals detected via non-invasive genetic sampling (NGS) in Sweden and Norway for males (M) and females (F). Values were
calculated as the number of individuals detected with NGS (Table A.2) divided by the total and sex-specific abundance estimates obtained from the open-population spatial
capture-recapture (OPSCR) models (Table A.4). Credible intervals (95%) are shown in parentheses. Note that in some years in Norway, male wolverines detected exceeded
the estimated number of wolverines. This is possible when wolverine detection probability was very high and wolverines with activity centers in Sweden were detected on the
Norwegian side of the border.
2014 2015 2016 2017 2018
F M F M F M F M F M
Norway 0.70 (0.62-0.79) 0.86 (0.78-0.96) 0.69 (0.63-0.75) 0.92 (0.85-0.99) 0.72 (0.67-0.78) 0.96 (0.89-1.02) 0.79 (0.74-0.83) 1.07 (1.01-1.14) 0.73 (0.69-0.77) 0.99 (0.94-1.04)
Sweden 0.26 (0.23-0.30) 0.32 (0.27-0.38) 0.37 (0.33-0.41) 0.46 (0.40-0.51) 0.32 (0.29-0.36) 0.50 (0.45-0.55) 0.64 (0.60-0.69) 0.80 (0.75-0.84) 0.67 (0.63-0.71) 0.86 (0.81-0.90)
Total 0.46 (0.41-0.51) 0.54 (0.48-0.60) 0.50 (0.45-0.54) 0.62 (0.57-0.67) 0.49 (0.45-0.53) 0.66 (0.62-0.71) 0.67 (0.64-0.71) 0.83 (0.80-0.87) 0.67 (0.64-0.70) 0.85 (0.82-0.88)
2019 2020 2021 2022
F M F M F M F M
Norway 0.74 (0.69-0.79) 0.99 (0.93-1.04) 0.78 (0.73-0.83) 0.99 (0.93-1.06) 0.81 (0.76-0.87) 0.99 (0.93-1.05) 0.84 (0.77-0.91) 1.00 (0.93-1.05)
Sweden 0.58 (0.54-0.61) 0.75 (0.70-0.80) 0.42 (0.38-0.45) 0.59 (0.55-0.64) 0.48 (0.45-0.52) 0.62 (0.58-0.67) 0.48 (0.44-0.52) 0.67 (0.62-0.72)
Total 0.63 (0.60-0.66) 0.81 (0.77-0.84) 0.55 (0.51-0.58) 0.72 (0.69-0.76) 0.59 (0.55-0.62) 0.73 (0.69-0.77) 0.61 (0.56-0.66) 0.76 (0.71-0.80)
35
