Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Austral Ecology (2022) 47, 954–970

Population decline and demography of the world’s largest


breeding population of huemul deer (Hippocamelus bisulcus)
in Bernardo O’Higgins National Park, Chile
SHALYNN M. PACK,1,2* DIANA BERTUOL-GARCIA,1,3 MATTHEW POMILIA,1
1 4
ADAM SPENCER AND DENNIS ALDRIDGE
1
Round River Conservation Studies, 925 E 900 S, Ste 210, Salt Lake City, Utah, 84105, USA
(Email: shalynnpack@gmail.com); 2Department of Fisheries, Wildlife, and Conservation Sciences,
Oregon State University, 104E Nash Hall, Corvallis, Oregon, 97331, USA (Email:
shalynnpack@gmail.com); 3School of Environmental Studies, University of Victoria, Victoria, British
Columbia, Canada; and 4Corporacion Nacional Forestal, Region de Aysen, Chile

Abstract The largest remaining populations of the endangered huemul deer (Hippocamelus bisulcus) are believed to be
concentrated in the remote, rugged and largely uninhabited glacial valleys and fjords surrounding the Southern Pata-
gonian Ice Field that lie within Chile’s Bernardo O’Higgins National Park. On the Park’s northwestern edge is Bernardo
Fjord, an area home to the world’s largest known breeding population of huemul. We present a demographic study of this
population and examine its population trends since 2004. We conducted huemul transect surveys on-foot during 16 expe-
ditions to Bernardo Fjord from October 2014 to March 2020, during which we spent a total of 254 days in the field and
documented 284 observations of huemul. This coastal population of huemul exhibited higher female proportions (mean
0.64  SD 0.05), lower fawn:female ratios (mean 0.18  SD 0.23 in March and mean 0.22  SD 0.07 in November),
and similar group sizes (mean 1.77  1.06 SD individuals) to previous studies. Huemul densities in Bernardo Valley
encountered from 2016 to 2019 were 60% lower than densities observed in a previous study from 2004 to 2008. We pro-
duce a 2019 huemul population estimate for Bernardo Fjord of 84 (SD 7) individuals including 44 observed individuals
and 40 (SD 10) individuals estimated to be undetected by our surveys, which comprises 6–8% of the IUCN global
population estimate. Two of the surveyed sectors – Bernardo Valley and Pampa – are of critical conservation importance,
given that the 5 km2 Pampa is home to 42% of the total huemul population and 41% of the fawn population of Bernardo
Fjord. We examine potential drivers of this population decline, which may include poor recruitment, predation and high
incidences of foot disease exacerbated by climatic conditions, and we offer recommendations for the conservation,
monitoring and management of this and other coastal huemul populations.
Las mayores poblaciones remanentes del amenazado Huemul (Hippocamelus bisulcus) se cree que estan concentra-
das en los remotos, accidentados y mayoritariamente inahabitados valles glaciares y fiordos que rodean el Campo de
Hielo Patag onico Sur, la mayor parte del cual esta contenido en el Parque Nacional Bernardo O’Higgins en Chile.
En el borde noroeste esta el Fiordo Bernardo, un area que alberga la poblaci on reproductiva mas grande conocida.
Presentamos un estudio demografico de esta poblaci on y analizamos su tendencia poblacional desde 2004. Recorri-
mos a pie transectos de huemul durante 16 expediciones en el Fiordo Bernardo, desde Octubre de 2014 hasta Marzo
de 2020, durante las cuales permanecimos un total de 254 dıas en terreno y registramos 284 observaciones de hue-
mules. Esta poblaci on costera de huemul mostr o una proporci on mayor de hembras (media 0.64  SD 0.05), una
relacion crıa:hembra menor (media 0.18  SD 0.23 en Marzo y media 0.22  SD 0.07 en Noviembre), y tama~ nos
de grupo similares (media 1.77  SD 1.06 individuos), que en estudios previos. Las densidades de huemules que se
encontraron para el Fiordo Bernardo en el periodo 2016–2019 fueron 60% menores que las densidades observadas
en un estudio previo de 2004–2008. Concluimos para el a~ no 2019 en el Fiordo Bernardo una poblaci on estimada de
huemules de 84 (SD 7) individuos, considerando 44 individuos observados y 40 (SD 10) individuos que se esti-
man no detectados en los muestreos, lo que representa 6–8% de la estimaci on poblacional global de acuerdo con
UICN. Dos de los sectores muestreados – Valle Bernardo y Pampa – poseen una importancia crıtica de conser-
vacion, ya que los 5 km2 de Pampa albergan el 42% de la poblaci on total de huemules y el 41% de la poblacion de
crıas del Fiordo Bernardo. Analizamos las causas potenciales de la declinaci on poblacional, las que pueden incluir
bajo reclutamiento, depredaci on y alta incidencia de enfermedades podales exacerbadas por las condiciones

*Corresponding author.
Accepted for publication April 2022.
[Correction added on 30 May 2022, after first online
publication: The paragraph spacing was inserted to differenti-
ate the Spanish language in the Abstract]

© 2022 Ecological Society of Australia. doi:10.1111/aec.13178


CRITICAL HUEMUL POPULATION IN DECLINE 955

climaticas. Finalmente, entregamos recomendaciones para la conservaci


on, monitoreo y manejo de esta y otras
poblaciones costeras de huemules.

Key words: Chile, conservation, endangered species, Hippocamelus bisulcus, huemul.

INTRODUCTION comm., 2016). Today, huemul in Bernardo Fjord


continue to live in a habitat nearly unaltered by
The huemul (Hippocamelus bisulcus) is an endangered humans. The only natural predator of adult huemul,
deer species endemic to the south of Chile and the puma (Puma concolor) is likely absent from Ber-
Argentina (Black-Decima et al. 2016). Huemul are nardo Fjord (although Wensing (2005) sighted puma
important both culturally as a symbol of Patagonia tracks, no other study including our 5 years of in situ
and ecologically as one of the few remaining mega- and camera trap research has documented any sign).
fauna species in the ecoregion. Due to hunting and Huemul fawns, however, are frequently predated by
habitat loss during the 19th and 20th centuries, today culpeo foxes (Lycalopex culpaeus), which are the most
its population has been reduced to just 1% of histori- important causes of mortality in this age class in
cal levels, with estimates ranging from 1048 to 2500 inland populations (Smith-Flueck 2003; Corti
individuals (Redford & Eisenberg 1992; Vila et al. 2010).
et al. 2006; Black-Decima et al. 2016). Remaining The most recent demographic study of huemul at
populations occupy 50% of the species’ historical dis- Bernardo – the only such study of a coastal huemul
tribution, and survive in isolated and fragmented population to date – summarized an annual popula-
populations mostly in mountainous and uninhabited tion census from 2004 to 2008 and estimated 71
terrain, habitat which is marginal relative to the valley individual huemul inhabiting the 500-ha survey area
floors from which they have largely been extirpated of Bernardo Valley as of 2008 (Brice~ no et al. 2013).
(Vila et al. 2006). Despite its conservation status and When we, however, began monitoring the Bernardo
cultural relevance, there is a profound lack of formal Fjord sector in 2014, the park guards claimed that
information on the huemul in Chile, particularly out- the huemul population had been in steep decline
side of protected areas, which makes it impossible to since the 2008 survey. Our initial surveys also indi-
determine its global population trend (Wittmer cated this population was significantly smaller than
et al. 2010). Moreover, the majority of what has been previously documented. Given the global importance
published on huemul ecology has been generated in of this population of endangered huemul deer – both
inland areas (Povilitis 1998, 2002; Corti et al. 2010), for its density and for being characteristic of a popu-
which are considered to be of a different ecotype with lation relatively undisturbed by human impacts and
significantly different genetics than coastal popula- influence – it is crucial to understand how the hue-
tions. It is suggested that coastal and inland huemul mul population is changing at Bernardo Fjord and
should be managed separately to preserve their surrounding areas.
genetic diversity (Marın et al. 2013). We aimed to fill this knowledge gap by using inno-
The largest remaining populations of huemul are vative methods to provide updated estimates of hue-
believed to survive near the southern Patagonian Ice mul population size and demographic rates in
Field, most of which lies within Chile’s Bernardo Bernardo Fjord, offering a point of reference for
O’Higgins National Park (Povilitis 1986; Frid 1994; managers and researchers to better understand the
Vila et al. 2006; Black-Decima et al. 2016). Near the huemul populations of the southern Patagonian
northern limit of the park lies Bernardo Fjord, an fjords and Ice Field. From 2016 to 2020, in collabo-
area believed to have the highest density of huemul ration with CONAF, we conducted bi-annual, stan-
anywhere in the world (Frid 2001; CONAF 2015). dardized surveys of the huemul population in this
The huemul population in Bernardo Fjord has been remote and rugged area. We conducted surveys in
studied sporadically for decades, starting with the November during the peak birthing period (F. Pare-
first expedition by Chile’s National Forestry service des, pers. comm., 2016), and in March to determine
(Corporaci on Nacional Forestal or CONAF) in the survival of fawns and generate population esti-
1985, during which they sighted 11 individuals mates outside of the breeding season. In this paper,
(CONAF 2015), and continuing with an in situ our goals are to: (i) assess population trends of this
research project on huemul habitat use, social critical population, comparing with estimates from
dynamics and interactions with introduced cattle Brice~ no et al. (2013); (ii) provide demographic rates
(Frid 1994, 1999, 2001). The cattle were mostly for huemul in the fjords ecoregion to compare with
removed by 2005 (Vila et al. 2019) and the last indi- rates from inland habitats; and (iii) estimate the cur-
vidual was removed in 2014 (F. Paredes, pers. rent population size by integrating detection

© 2022 Ecological Society of Australia. doi:10.1111/aec.13178


956 S. M. PACK ET AL.

Fig. 1. Location of the region studied in the Bernardo O’Higgins National Park (left) in Chile and the sectors included in
the biannual census (right). Sectors in the Bernardo Valley Region are in green, the Huemules region in orange and the Ofhi-
dro region in blue. Contour lines represent 100 m of elevation.

probability. Based on our results, we then examine crowberry (Empetrum rubrum) and Gunnera magellanica
potential causes for population trends; and offer (Frid 1994, 2001; Brice~ no et al. 2013).
management recommendations for huemul in this For this study, we divided the Bernardo Fjord area into
and other coastal areas. three regions comprising a total of eight sectors, all of
which lack permanent human colonization and were
selected based on the presence of huemul and a topogra-
phy that allowed field surveys: Bernardo Valley region
METHODS (sectors Pampa, GLOF, Valle Bajo, and Refugio), Hue-
mules region (sectors Valle Huemules, Valle Toros, and
Study area Terrazas) and Ofhidro region (Ofhidro sector; Fig. 1). Of
these sectors, four (Pampa, Valle Bajo, Refugio and Valle
In Bernardo Fjord (Fig. 1), the climate is cold and wet Huemules) were surveyed in 2004–2008 by Brice~ no
throughout the year due to a strong oceanic influence, pre- et al. (2013).
senting an annual mean precipitation of 4000 mm, annual In the Bernardo Valley region, the Pampa sector (5 km2)
temperatures averaging 7–8°C, and snowfall at any time of is a low, flat peninsula covered with scrub that transitions
year (although mainly from June to August) (Carrasco into a wooded hill on its southwest and northeast slopes;
et al. 1998). Soils are rocky, nutrient-poor, and water- this sector has long been known to have a breeding popula-
saturated, due to high rainfall and poor drainage from post- tion of huemul (Frid 1999). The Valle Bajo sector
glacial depressions on the landscape. The topography is (4.6 km2) is a flooded grassland bordered by the fjord and
dominated by steep-walled glacial valleys and retreating gla- dense forest. The GLOF sector (2.95 km2) includes a steep
ciers, with ecosystem types including periglacial grasslands, hillside that transitions from fjord edge into a small grass-
dense thickets of low-lying shrubs including calafate (Ber- land plateau overlooking Bernardo Glacier and its glacial
beris microphylla) and chaura (Gaultheria mucronata), old- lake, which has an annual outburst flood into the fjord (F.
growth forests of Nothofagus spp., and moorlands of red Paredes, pers. comm., 2016). On the west side of the fjord,

doi:10.1111/aec.13178 © 2022 Ecological Society of Australia.


CRITICAL HUEMUL POPULATION IN DECLINE 957

the Refugio sector (4.14 km2) contains a long, narrow area total distance walked by the surveying group and by the
of flat topography that is densely covered in shrubs, and is number of people in the group. The association between
home to the CONAF-Aysen guard station which is occu- both measures of survey effort and the number of individ-
pied sporadically through the year. In the Huemules region, ual huemul seen across sectors and expeditions was tested
the Terrazas sector (6.93 km2) includes a hillside that using a non-parametric Spearman rank correlation, with
climbs out of the Refugio sector, spilling into the wide, flat P-values lower than 0.05 considered to be significant.
grassland valley of the Valle Toros (5.85 km2), aptly named Each time a huemul was sighted, we recorded the num-
for its previous presence of domestic cattle. To its west is ber, age, sex and condition of individuals, and the presence
the Valle Huemules sector (4.83 km2), a wide grassland or absence of an ear tag, date, time, observer UTM coordi-
valley connecting to the Tempano Fjord. Lastly, the Ofhi- nates, distance and angle between observers and individuals
dro region and sector (15.29 km2) lies within a small fjord (calculated with a Nikon Laser 1200 Monarch Gold laser
18 km north of the Bernardo Fjord, and contains narrow rangefinder and compass, respectively), and photo num-
valleys with steep hillsides around the Ofhidro Glacier. bers. We categorized huemul age classes following Corti
Although these sectors represent biogeographical distinc- et al. (2010) as follows: adults (individuals older than
tions, there is certainly some genetic flow of huemul 3 years), juveniles (2–3 years), yearlings (1–2 years), and
between them; for example, we have observed a huemul fawns (<1 year old, usually born in November). In all sur-
swim across the narrow section of Bernardo Fjord from the veys, we reduced the possibility of double-counting by not
Pampa to the Refugio sector. However, given the philopatry counting individuals in an area already surveyed during that
of female huemul, that the area of most sectors is approxi- expedition, by using photo-identification, and by analysing
mately one huemul home range of 3.5–6.5 km2, and that the location and time of each sighting. When mapping our
average daily movements of huemul are <1 km, for this sightings, we used the distance and angle from the individ-
study we assume that huemul populations in the sectors are ual or group sighted to observers to calculate and map the
closed with no emigration or immigration (Fig. 1; Dıaz & true location of the animal.
Smith-Flueck 2000; Gill et al. 2008). We created a database with individuals observed during
surveys in each of the 8 sectors, linked to photo(s) of the
individuals when able. We categorized sightings observed
METHODS outside of our standardized survey methodology – for
example, from a boat-based sighting during travel, or from
Field surveys CONAF data taken prior to 2016 – as “opportunistic” data,
which we used to document the presence of huemul but
From October 2014 to March 2020, huemul deer were sur- excluded from analyses. We tracked all re-sightings of the
veyed biannually in the Bernardo Fjord through a collabo- 10 ear-tagged individuals and also documented any huemul
rative effort between Round River Conservation Studies carcasses, skeletons, or other remains observed during
(RRCS) and CONAF, either in the austral spring (October expeditions. All statistical analyses described were done in
or November) or the austral summer (February or March). the R environment (R Core Team 2021), using the tidy-
Initial surveys in 2014–2016 aimed to determine areas of verse package (Wickham et al. 2019).
huemul habitat and select the study sectors. In August
2016, CONAF visited the Pampa to capture, ear-tag, and
treat nine individuals with health issues, including three Assessing long-term population trends
adult males, five adult females and one juvenile; they trea-
ted seven of these for a foot ailment and one for an abdom- To assess huemul population trends over a longer time
inal tumour. In August 2018 they treated and ear-tagged scale, we compared our results from 2016 to 2020 with
one additional adult male with a foot ailment those from the only previous survey in the region con-
(CONAF 2016). By November 2016, we began surveying ducted from 2004 to 2008 (Brice~ no et al. 2013). We used
routes using the standardized methodology detailed here. the densities estimated as an indicator of changes in abun-
For all sectors except the Pampa, we surveyed on foot in a dance, and we compared our data set to that of Brice~ no
single-file line, one person behind the other, generally in a et al. (2013) by using a matching subset of our larger data
group of 10 people led by a CONAF park guard or RRCS set. Their study was conducted in three valleys, two of
biologist; all surveyors received at least 1 month of previous which overlap with our study: the Bernardo Valley and the
training in the observation and field detection of huemul. Huemules Valley (the same as our sector Valle Huemules).
We elected for this single-line survey method to ensure sur- We assumed that their survey in Bernardo Valley included
veyor safety while maximizing the total area we could sur- our sectors Pampa, Valle Bajo, and Refugio, and so we
vey during the multi-day survey expedition; this method summed our observations from these sectors to compare
still afforded high visual coverage of huemul due to the with their survey results from Bernardo Valley. We repeated
open terrain. For the Pampa sector, however, we deviated their methodology to calculate the area surveyed: a fixed
from this methodology and instead surveyed in a sweeping width transect of 300 m (150 m on each side of the lines
line, with each surveyor walking 250–500 m from the next walked) in Bernardo Valley and full coverage of 3.2 km2 in
observer, following an azimuth of 330°; we elected the Huemules Valley. We excluded from this analysis several of
sweep method on the Pampa because its location, accessi- our surveys in which field conditions prevented a full sur-
bility by boat and small size allowed us to survey it during vey, including Valle Huemules in November 2019 and Ber-
a half-day outing from base camp. Total survey effort var- nardo Valley in February 2016 and March 2020. We then
ied across expeditions and sectors and was measured by the calculated the density of huemul in the region in the same

© 2022 Ecological Society of Australia. doi:10.1111/aec.13178


958 S. M. PACK ET AL.

way as Brice~ no et al. (2013), by dividing the total number surveyed for each sector and expedition; we restricted our
of individuals seen of all age classes by the area surveyed. estimate of density to yearlings, juveniles and adults
To ensure comparability between our studies, we recalcu- because variation in these age classes is a more meaningful
lated their density values based on their reported number indicator of long-term trends in populations of species with
of observed individuals and distance surveyed for Bernardo high variability in fawn survival (Gaillard et al. 2000; Corti
Valley. Finally, we compared our densities with those et al. 2010; Wittmer et al. 2010). The proportion of females
reported by Brice~ no et al. (2013) by performing a one- was calculated for every sector and expedition by dividing
tailed Wilcoxon Rank Sum test with the hypothesis that the number of females by the total number of individuals
their densities were higher than the ones we encountered; from both sexes. For the female proportion calculation, we
before performing the test, we checked for independence of excluded individuals whose sex was not identified and indi-
observations by calculating the temporal autocorrelation viduals younger than 2 years (Corti et al. 2010; Brice~ no
using the Spearman correlation index between lagged et al. 2013). The fawn:female ratio was calculated for every
observations, and we considered P-values lower than 0.05 sector and expedition by dividing the observed number of
to be significant in both steps. To ensure that our assump- fawns by the observed number of adult females. However,
tion about the area Brice~ no et al. (2013) included in their for November expeditions, this value closely represents the
surveys of the Bernardo Valley did not affect our results, fertility rate, for March expeditions – which occur at least
we also generated different combinations of the sectors 3 months after the birthing period – this value represents a
included as part of the Bernardo Valley and repeated the combination of fertility and fawn survival in the initial
above analysis. months.
To investigate differences in density, female proportion
and fawn:female ratio across regions and sectors, we used
Calculating demographic indices for the region each survey in each sector (e.g. Pampa survey of November
2019) as an observation and performed a Kruskal–Wallis
Pooling every sector and expedition, we calculated key Rank Sum Test. When significant differences across regions
demographic variables from our entire data set to provide a or sectors were encountered, we then performed pairwise
point of comparison with studies of huemul in other Wilcoxon Rank Sum tests with Bonferroni correction for
regions. For these analyses, we calculated the area surveyed the P-value (two-sided) to identify which regions or sectors
using different distances than the ones employed in Brice~no were different from the others. Before performing these
et al. (2013), because we felt this more accurately repre- tests, we checked for independence of observations within
sented our survey conditions. For this, we estimated the each sector by calculating the temporal autocorrelation
total area surveyed in QGIS by using a buffer around the using the Spearman correlation index between lagged
line surveyed and eliminating any overlap; buffers differed observations (Appendix S3). Additionally, we tested
for each sector (Table 1) and were determined based on whether the proportion of huemul in each age class was dif-
our experience of average visibility for each sector. To ferent across regions and sectors by performing Chi-Square
ensure that our estimates of average visibility were reason- tests with each individual huemul as an observation.
able and thus our estimates of survey area were accurate, Finally, we examined patterns in average group size of
we also compared our estimates to the average sighting dis- huemul when females, males and fawns were present, as
tance of all surveys in each sector (Table 1). In the sectors well as by sector, region and season for our standardized
La Pampa and Valle Bajo, due to their flat terrain and low survey data from November 2016 to November 2019, using
vegetation, we assumed our surveys covered the totality of each group sighting as an observation. We performed
the area and, therefore, used a fixed area covered for every Kruskal–Wallis Rank Sum tests to investigate differences
survey in these sectors (5 km2 for La Pampa and 0.46 km2 across regions and sectors, followed by pairwise Wilcoxon
for Valle Bajo), independent of the route surveyed. Rank Sum tests with Bonferroni correction for the P-value
Then, we calculated huemul density by dividing the total (two-sided) if needed to identify which regions or sectors
number of individuals older than 1 year by the area were different from the others. To investigate differences

Table 1. Estimated visibility range (i.e. buffer) for huemul surveyed in each sector, compared with the observed average,
maximum and minimum sighting distance of huemul individuals and/or groups in that sector (surveys between 2014 and
2020)

Buffer Average sighting Minimum sighting Maximal sighting


Region Sector used (m) distance (m) distance (m) distance (m)

Bernardo GLOF 50 47.71 (n = 7) 4 120


Valley Pampa No buffer, see text 82.26 (n = 46) 0 500
Valle Bajo No buffer, see text 40 (n = 2) 20 60
Refugio 100 75.81 (n = 22) 0 400
Huemules Terrazas 200 213.38 (n = 8) 46 700
Valle Huemules 100 40.67 (n = 3) 20 52
Valle Toros 200 163.5 (n = 4) 50 353
Ofhidro Ofhidro 200 166.14 (n = 14) 2 500

doi:10.1111/aec.13178 © 2022 Ecological Society of Australia.


CRITICAL HUEMUL POPULATION IN DECLINE 959

between groups with and without fawns, we performed a literature (Table 2, Appendix S1) to calculate the expected
Wilcoxon Rank Sum test (two-sided). P-values lower than contribution of each individual in 1 year to the population
0.05 were considered to be significant in all tests. in the next year (Caswell 2001). This generated a number
of expected individuals in the next year, which was com-
pared with the number of observed individuals to arrive at
Estimating the current population size by an estimate of undetected individuals for that year. The
including undetected individuals estimated population size for that year was then calculated
as the sum of observed and undetected individuals, and this
To estimate the population size of huemul in the region estimate was used as the base number to calculate the
based on the number of sighted individuals, we needed to expected contribution of each individual for the next year.
account for imperfect detection during our surveys. Avail- Those steps were done first within each sector and then
able methods to calculate detection probabilities are dis- summarized across sectors for a total population estimate
tance sampling (Buckland et al. 2015) or following marked for each year; we started with data from the austral summer
individuals to retroactively estimate detection probability of 2017 and finished with an estimated population size for
(Wittmer et al. 2010). Both of these methods were not fea- the austral summer of 2019. To account for possible varia-
sible in Bernardo Fjord. The number of observations in tion in vital rates and incorporate uncertainty in our esti-
each survey was too small to be able to successfully esti- mate, we conducted 1000 simulations, each time using a
mate detection probability as a function of sighting distance different set of fertility and survival rates. These vital rates
for each sector and survey (Buckland et al. 2015), and visi- were randomly sampled from a normal distribution with a
bility across sectors was too heterogenous to allow pooling mean set to the calculated vital rate average and standard
of all observations and calculate a single detection probabil- deviation set to a value that took into account the average
ity (Appendix S1). Moreover, the remoteness and terrain variability in ungulate vital rates from the literature (Gail-
made following marked individuals nearly impossible. lard et al. 2000; Wittmer et al. 2010) (Table 2). Then, we
Therefore, we estimated the number of undetected individ- used the mean and standard deviation of the 1000 repli-
uals by combining our database of sighted individuals with cates of population size calculated for the austral summer
survival and fertility rates from the literature. To our of 2019 to arrive at our final estimate of the huemul popu-
knowledge, this is an innovative method that can be applied lation size in the Bernardo Fjord region.
in situations in which species ecology, terrain characteristics
and logistical difficulties prevent the use of more estab-
lished methods to account for undetected individuals, as Culpeo fox diet & presence
long as survival and fertility rates are available from another
population. During selected expeditions (November 2016, March 2017,
We estimated the number of undetected individuals for March 2018, November 2018, and March 2019), we col-
every austral summer survey as this provided a larger sam- lected scat piles of culpeo fox encountered during huemul
ple size. We used survival and fertility rates from the surveys, to analyse fox diet and spatial distribution in

Table 2. Rates used in the estimation of undetected individuals

Values used
Rate (mean  SD) Source

Fertility 0.72  0.11 Mean value from Corti et al. (2010) and standard deviation
considering an average variability of 15% from Wittmer
et al. (2010)
Fawn survival (November–March) 0.50  0.10 Estimated using an exponential decay function fitted with mean values
from Corti et al. (2010) and considering an average variability of
40% from Wittmer et al. (2010)
Fawn survival (March–March) 0.65  0.15 Estimated using an exponential decay function fitted with mean values
from Corti et al. (2010) and considering an average variability of
40% for fawns and 15% for yearlings from Wittmer et al. (2010)
Yearling survival (March–March) 0.89  0.13 Estimated assuming constant survival throughout the year with data
from Corti et al. (2010) and an average variability rate of 15%
from Wittmer et al. (2010)
Juvenile survival (March–March) 0.91  0.09 Estimated assuming constant survival throughout the year with data
Females 0.93  0.09 from Corti et al. (2010) and an average variability rate of 10%
Males 0.86  0.09 from Wittmer et al. (2010)
Adult survival (March–March) 0.91  0.09 Average adult survival from Wittmer et al. (2010), sex-specific adult
Females 0.94  0.09 survival from Corti et al. (2010) and average variability rates of
Males 0.83  0.08 10% from Wittmer et al. (2010)
Female proportion 0.64  0.05 This study

Details of how these rates were calculated, their differences, and how they were used to estimate the expected number of
individuals from 1 year to the next, can be found in Appendix S1.

© 2022 Ecological Society of Australia. doi:10.1111/aec.13178


960 S. M. PACK ET AL.

Bernardo Fjord. For each collected sample, we recorded not correlated with the total number of individuals
the date, UTM coordinates and sector, and later processed seen, both when considering the total distance
unique sample piles by placing into a fine sieve and rinsing walked (r = 0.21, P = 0.07) and the number of peo-
to remove digested material, drying over low heat (200°F) ple in the group (r = 0.14, P = 0.15).
and examining the remaining contents under a microscope.
From October 2014 to March 2020, we docu-
In each scat pile, we noted the presence or absence of hue-
mented a total of 284 observations involving 548
mul hair, rodent hair or bones, insects, birds, berries and
fish. We referenced Chehebar and Martın (1989) to differ- individual huemul including 70 fawns
entiate huemul from fox and rodent hairs by their length, (Appendix S2); these initial figures include double
colour and shape. We calculated the frequency of occur- counts between seasons and are not a population
rence of each dietary component by dividing the number of estimate. Of fawns, 87.1% (n = 61) were observed in
scat samples containing each component by the total num- the Bernardo Valley region, the majority of these
ber of scat samples, and mapped the spatial distribution of (n = 42) in the Pampa sector, and the remaining
huemul remains and collected fox scat, noting which scat 8.6% (n = 6) in Ofhidro and 4.3% (n = 3) in Hue-
piles contained huemul hair. We tested whether the propor- mules Valley regions. We generated photo records
tion of fox scat containing huemul hair was different across for 76 observations and compiled 158 photographs
survey sectors and seasons by performing Chi-square tests
into a photo-database (Appendix S2). We observed
using each scat sample as an observation, with P-values
below 0.05 considered to be significant, and repeated this
huemul in all sectors surveyed and our opportunistic
test for berries. sightings demonstrated huemul presence in addi-
tional locations (Fig. 2).
Excluding any potential double counts of individu-
RESULTS als within the same expedition, the Pampa sector
consistently held the most individuals, reaching 35
individuals observed in one expedition compared to a
Field surveys maximum of 12 in other sectors (Fig. 3).

We compiled huemul sightings data from 16 expedi-


tions to Bernardo Fjord from 9 October 2014 to 14 Assessing long-term population trends
March 2020, during which we spent a total of
254 days in the field. Of these, we utilized standard- Huemul densities (including all age classes) in Ber-
ized survey methodologies in seven expeditions from nardo Fjord encountered from 2016 to 2019 were
11 November 2016 to 14 March 2020, which totalled smaller than those estimated for the period 2004–
to 69 days spent surveying for huemul. Each expedi- 2008 by Brice~ no et al. (2013) (W6,6 = 32.5,
tion lasted an average of 9 days and involved an aver- P = 0.01), and the same trend was found for the
age of nine observers (Table 3). Survey effort was Huemules Valley (W6,4 = 21, P = 0.03; Fig. 4).

Table 3. Expeditions for huemul surveys conducted by Round River Conservation Studies and CONAF from 2014–2020

Expedition dates Expedition lead # Of surveyors† Km surveyed

November 27–30, 2014 CONAF 2 NR


October 8–29, 2014 RRCS 13 24.5
January 18–26, 2015 CONAF 2 NR
February 2–19, 2015 RRCS 16 24.2
July 15–October 25, 2015 CONAF 2 NR
January 29–February 23, 2016 RRCS 12 29.9
December 12, 2016 CONAF 2 NR
November 11–22, 2016‡ RRCS 12 70.5
March 8–21, 2017‡ RRCS 14 61.6
July 4, 2018 CONAF 2 NR
March 9–19, 2018‡ RRCS 15 112.2
October 23, 2018 CONAF 2 NR
November 12–21, 2018‡ RRCS 11 55.8
March 22–April 1, 2019‡ RRCS 15 114.4
November 10–17, 2019‡ RRCS 10 97.9
March 12–14, 2020‡ RRCS 6 9.3

NR, not recorded.



Of these surveyors, on average two were from CONAF and the remainder were from RRCS.

These are standardized surveys used in the demography analysis of this paper.

doi:10.1111/aec.13178 © 2022 Ecological Society of Australia.


CRITICAL HUEMUL POPULATION IN DECLINE 961

(a) (b)

Fig. 2. Location of huemul sightings from opportunistic and survey expeditions from 2014 to 2020 in the surroundings of
Bernardo Fjord (a) and Ofhidro Fjord (b).

Considering the Refugio, Valle Bajo and Pampa sec- more than double the density of most other expedi-
tors as part of the Bernardo Valley, the average hue- tions (Table 4).
mul density in Bernardo Valley was 60% lower in Densities were different across regions (H = 10.6,
2016–2019, at 3.48 (SD 1.66) ind km 2 than in d.f. = 2, P = 0.005) and sectors (H = 25.8, d.f. = 7,
2004–2008, at 8.47 (SD 3.45) ind km 2. In Hue- P < 0.001), with Valley Bernardo and Ofhidro
mules valley, the density was 68% lower in 2016– regions presenting higher densities than the Hue-
2019 at 0.52 (SD 0.61) ind km 2 than in 2004– mules region (Huemules vs. Bernardo: W = 268,
2008, at 1.62 (SD 0.73) ind km 2. These trends Nhuemules = 23, Nbernardo = 16, P = 0.04, two-sided;
were the same regardless of the sectors considered as Huemules vs. Ofhidro: W = 0, Nhuemules = 16,
part of Bernardo Valley (Appendix S3). Nofhidro = 4, P = 0.007, two-sided). Although pair-
wise Wilcoxon Rank Sum tests did not indicate
which sectors presented statistically different densi-
Estimating demographic indices ties, we can note that the highest densities of huemul
were encountered in the Pampa, Refugio and Ofhi-
From November 2016 to November 2019, we docu- dro sectors (Table 5).
mented 108 observations of huemul in our standard- Adult individuals comprised the majority of the
ized survey routes, totalling between 17 and 50 observed population, which also had on average
unique individuals observed per seasonal expedition more females than males. The overall mean female
(Table 4). Across all areas surveyed in each expedi- proportion, including adults and juveniles, was
tion, we documented an average density of individu- skewed towards females (mean 0.64  SD 0.05)
als older than 1 year of 2.06 ind km 2 (SD 1.2) (Table 4), and this trend was slightly more marked
(Table 4). There were marked variations in the num- when considering just adults (mean 0.66  SD
ber of sighted individuals and huemul density across 0.05). The female proportion did not differ signifi-
expeditions; for example, density in March 2017 was cantly across survey regions (H = 5.27, d.f. = 2,

© 2022 Ecological Society of Australia. doi:10.1111/aec.13178


962 S. M. PACK ET AL.

Fig. 3. Number of huemul individuals observed within all RRCS survey sectors from 2014 to 2020. Double counts have
been eliminated so these data represent unique individuals within each expedition. Note that the y-axis scale is different for
the sector Pampa. NS, not surveyed; V, Valle.

doi:10.1111/aec.13178 © 2022 Ecological Society of Australia.


CRITICAL HUEMUL POPULATION IN DECLINE 963

P = 0.04). Although pairwise Wilcoxon Rank Sum


tests did not indicate which regions presented statisti-
cally different group sizes, larger groups of huemul
were encountered in the Ofhidro region, with average
group size of 2.23 (SD 0.93) compared to 1.74
(SD 1.09) in Bernardo Valley and 1.53 (SD 0.92)
in Valle Huemules (Table 6).
During all expeditions, we tracked the survival of
the 10 adult individuals that were marked by
CONAF with coloured ear tags in August 2016 and
2018. The number of days since capture that a
tagged individual was last observed averaged
432 days, ranging from a minimum of 94 days to a
maximum of 1190 days or 3.3 years. The foot injury
treatment that CONAF applied during capture
appears to have been largely successful: of the eight
individuals that were treated for a foot ailment upon
their capture in 2016 or 2018, seven did not present
a limp or foot injury at their last sighting. Tagged
individuals were not seen on most surveys of their
tagging site: although we conducted six surveys of
the Pampa, on average each tagged individual was
only seen on two of the six surveys. When re-sighted,
Fig. 4. Density of huemul of all age classes encountered
however, individuals were always located in the same
in standardized surveys conducted in a previous study
(Brice~
no et al. 2013) and in this study in the Bernardo and sector where they were captured.
Huemules Valley of the Bernardo Fjord. Black lines repre-
sent mean density for each study and grey shadings repre-
sent standard deviations. Estimating the current population size by
including undetected individuals
P = 0.07) or sectors (H = 8.11, d.f. = 6, P = 0.23)
(Table 5). The average fawn:female ratio for Novem- We estimate the huemul population size for the Ber-
ber expeditions was 0.22  SD 0.07 and 0.18  SD nardo Fjord region in the austral summer of 2019 to
0.23 for March expeditions (Table 4), and did not be 84 (SD 8) individuals. Because we directly
differ significantly across regions (H = 0.48, d.f. = 2, observed 44 individuals during this 2019 survey, we
P = 0.78) or sectors (H = 7.03, d.f. = 6, P = 0.32) estimate that 40 (SD 7) individuals were unde-
(Table 5). The highest number of fawns were tected by our surveys, arriving at a detection proba-
observed in the Pampa, Refugio and Ofhidro sectors bility of 52%. We estimate 69% of all individuals to
(Fig. 3), although age structure did not vary signifi- be located in the Bernardo Valley, especially in the
cantly across survey regions (v2 = 8.43, d.f. = 6, Pampa sector which holds an estimated 50% of the
P = 0.21) or sectors (v2 = 19.01, d.f. = 18, Bernardo Fjord huemul population. The Pampa was
P = 0.39). also the sector with the highest proportion of fawns
Of our 108 observations during standardized sur- with 46.7% of the total estimated fawns, followed by
veys, group size varied from 1 to 6 individuals, with a the Ofhidro sector with 26.7% (Table 7).
mean of 1.77 (SD 1.06) individuals. The majority
of observations were of a solitary huemul (51.4%) or
a group of two individuals (31.2%). The average Culpeo fox diet & presence
group size remained around 2 individuals, regardless
of whether a male (adult or juvenile; We collected and identified the contents of 101 piles
mean = 2.01  SD 1.21) or a female (adult or juve- of culpeo fox scat over five expeditions, huemul hair
nile; mean = 1.97  SD 1.12) was present in the in 15% of scats, rodent in 88%, berries/fruits (B. mi-
group. However, when a fawn was present in the crophylla, E. rubrum, G. mucronata) in 62%, birds in
group, the mean group size was 2.83 (SD 1.25), 41%, insects in 32% and fish scats in 0.9%. The pro-
significantly larger than when fawns were absent portion of scats containing huemul hair differed by
(W = 288, Nfawn = 18, Nnofawn = 90, P < 0.001, two- sector (v2 = 20.9, d.f. = 5, P < 0.001), with the high-
sided). Group size did not vary across sectors est rates in GLOF (60%, N = 10), Pampa (20%,
(H = 11.7, d.f. = 6, P = 0.07) but significantly dif- N = 15), Refugio (12.5%, N = 40) and Valle Toros
fered between survey regions (H = 6.64, d.f. = 2, (4%, N = 24). Some of the samples containing
© 2022 Ecological Society of Australia. doi:10.1111/aec.13178
964

Table 4. Observed number of huemul individuals by sex and age class and demographic variables for each survey expedition in the Bernardo Fjord (M = male, F = female,

doi:10.1111/aec.13178
U = unknown sex; V=Valle)

Observed
S. M. PACK ET AL.

Adults Juveniles Yearlings


Area surveyed Female Fawn: female Density
Year Month Sectors surveyed (km2) F M U F M U F M U Fawns Total proportion ratio (ind km 2)

2016 Nov Pampa, V Bajo, Refugio, 18.04 16 8 0 0 0 0 1 0 0 4 29 0.67 0.25 1.39


Terrazas, V Huemules
and V Toros
2017 Mar GLOF, Pampa, V Bajo, 14.32 31 13 0 2 0 0 0 0 3 1 50 0.72 0.03 3.42
Refugio, Terrazas, V
Huemules and V Toros
2018 Mar GLOF, Pampa, V Bajo, 20.03 18 7 0 0 2 0 0 0 0 3 30 0.67 0.17 1.35
Refugio, Terrazas, V
Huemules, V Toros and
Ofhidro
2018 Nov GLOF, Pampa, V Bajo, 11.68 11 6 0 0 1 0 0 0 0 3 21 0.61 0.27 1.54
Refugio, Terrazas and
Ofhidro
2019 Mar GLOF, Pampa, V Bajo, 19.54 20 10 0 1 1 0 1 1 0 10 44 0.66 0.50 1.74
Refugio, Terrazas, V
Huemules, V Toros and
Ofhidro
2019 Nov Pampa, V Bajo, Refugio, 17.61 7 5 1 1 1 0 1 0 0 1 17 0.57 0.14 0.91
Terrazas, V Huemules, V
Toros and Ofhidro
2020 Mar Refugio 1.22 3 2 0 0 0 0 0 0 0 0 5 0.60 0.00 4.10
Mean  SD 0.64  0.05 Nov: 0.22  0.07 2.06  1.20
Mar: 0.18  0.23

© 2022 Ecological Society of Australia.


CRITICAL HUEMUL POPULATION IN DECLINE 965

Table 5. Demographic variables of huemul by region and sector in the Bernardo Fjord

Expeditions Density
to the Female proportion Fawn:female ratio (ind. km 2,
Region and sector Dates surveyed sector (mean  SD) (mean  SD) mean  SD)

Bernardo Valley 0.61  0.12 0.20  0.30 2.44  2.46


region
† †
GLOF Mar 2017, Mar 2018, Nov 4 00
2018, and Mar 2019
Pampa Nov 2016, Mar 2017, Mar 6 0.66  0.06 0.25  0.23 3.13  1.88
2018, Nov 2018, Mar
2019, and Nov 2019,
Valle Bajo Nov 2016, Mar 2017, Mar 6 0.58  0.12 00 1.81  2.89
2018, Nov 2018, Mar
2019, and Nov 2019
Refugio Nov 2016, Mar 2017, Mar 7 0.57  0.15 0.21  0.39 3.78  2.3
2018, Nov 2018, Mar
2019, Nov 2019, and
Mar 2020
Huemules region 0.82  0.22 0.26  0.43 0.4  0.42
Terrazas Nov 2016, Mar 2017, Mar 6 0.79  0.25 00 0.42  0.39
2018, Nov 2018, Mar
2019, and Nov 2019,
Valle Huemules Nov 2016, Mar 2017, Mar 5 10 0.50  0.71 0.3  0.42
2018, Mar 2019, and
Nov 2019,
Valle Toros Nov 2016, Mar 2017, Mar 5 0.75  0.25 0.44  0.51 0.47  0.53
2018, Mar 2019, and
Nov 2019
Ofhidro region 0.60  0.08 0.46  0.63 3.22  1.1
Ofhidro Mar 2018, Nov 2018, Mar 4 0.60  0.08 0.46  0.63 3.22  1.1
2019, and Nov 2019

No huemul was seen during a standardized survey in the GLOF sector, so female proportion and fawn:female ratio do not
apply.

Table 6. Huemul group size by region and sector in the containing huemul hair did not vary across survey
Bernardo Fjord
seasons (v2 = 0.0712, d.f. = 1, P = 0.79), whereas
Region and sector N Group size (mean  SD) the proportion of scats containing berries did differ
significantly by season (v2 = 20.7, d.f. = 1,
Bernardo Valley region 80 1.74  1.09 P < 0.001) with scats found in the austral summer

GLOF 0 containing a higher proportion of berries. Addition-
Pampa 57 1.84  1.21 ally, we observed many culpeo foxes during each
V Bajo 4 1.25  0.50
expedition, and the individuals showed signs of
Refugio 19 1.53  0.70
Huemules region 15 1.53  0.92 excellent health in their size, body mass and coat
Terrazas 7 1.29  0.50 sheen (Appendix S2).
V Huemules 3 1  0 Additionally, we documented the presence of 18
V Toros 5 2.2  1.30 unique huemul remains, including four fresh car-
Ofhidro region 13 2.23  0.93 casses, seven older skeletons and seven craniums
Ofhidro 13 2.23  0.93 (Fig. 5). Six of the 18 remains were of fawns, 11

No huemul was seen during a standardized survey in the
were of adults, and 1 was of unknown age. Two of
GLOF sector, so female proportion and fawn:female ratio the remains indicated a possible cause of death,
do not apply. including one adult male carcass with a dislocated
back leg joint and one fawn that likely drowned in
the rising tide. At least five remains – two adults and
huemul hair also included huemul skin, large pieces three fawns – showed signs of fox scavenging, includ-
of fatty tissue (1 9 1 cm) and a strong smell of car- ing body and joint distortion, torn flesh and the pres-
rion after washing, which was not present in other ence of nearby fox scat and hair. The remains were
scat samples. The proportion of scat samples concentrated on the Pampa sector (n = 9), followed

© 2022 Ecological Society of Australia. doi:10.1111/aec.13178


966 S. M. PACK ET AL.

by GLOF (n = 4), Refugio (n = 3), Terrazas (n = 1)


Table 7. Number (mean  SD) of estimated huemul individuals present in the study regions and sectors of Bernardo Fjord in the austral summer of 2019, divided by sex

13  1 (15.5%)

13  1 (15.5%)
58  6 (69%)
and Ofhidro (n = 1) (Fig. 5).

Total

42  5

11  0

13  1
84  8
00

51

50
20
61
DISCUSSION

Using the same methodology as previous studies, we


observed significantly lower huemul densities in
1 (100%)
Unknown sex

2016–2019 than those reported in 2004–2008 by


0 (0%)

0 (0%)
Brice~no et al. (2013) or by Wensing (2005), pointing
to a decline in the huemul population in the Ber-
0
1
0
0

0
0
0

0
1












nardo region over the past 15 years. This is in con-
3
0
3
0
0
0
0
0
0
0
0
3
trast to previous studies that indicated an increasing
population (Frid 1999; Wensing 2005; Brice~ no
et al. 2013). Despite the decline, huemul densities in
12  1 (70.6%)

3  0 (17.6%)

2  0 (11.8%)

Bernardo Valley (3.48 huemul km 2) are almost


Male
Adults

twice as large as those observed in inland locations


(1.79  0.33 huemul km 2) (Corti et al. 2010). Our
17  1
00
71
10
40

20
00
10

20

estimate of 84  8 individuals inhabiting the region


accounts for 6 to 8% of the estimated global popula-
tion of huemul, a disproportionately high number
considering that 101 subpopulations are thought to
26  3 (72.2%)

6  0 (16.7%)

4  0 (11.1%)

exist (Vila et al. 2006; Black-Decima et al. 2016).


Female

While the cause of the apparent decline is unclear,


we have identified potential factors that could be
19  3

36  3
00

20
50

20
10
30

40

driving higher mortality and or lower recruitment


among the Bernardo population.
0 (33.3%)

0 (66.6%)

Possible drivers of the population decline


Numbers in parenthesis are the percentage of the total for each sex and age class.
Juveniles

0 (0%)

One factor that could explain a decline in the huemul


0
0
0
0

0
0
0

0
0

population is poor recruitment. We reported lower















fawn:female ratios (0.22  SD 0.07 for November


1
0
1
0
0
0
0
0
0
2
2
3

and 0.18  SD 0.23 for March) than those reported


in inland habitats (0.72  0.20; Corti et al. 2010).
7  2 (70%)

2  0 (20%)

1  0 (10%)

However, this comparison should be made with cau-


Yearlings

tion as the latter estimate is from a study where all


10  2

huemul individuals were radio-collared and moni-


00
52
10
10

00
10
10

10

tored to assess reproduction and survival (Corti


et al. 2010), a method that certainly resulted in a
higher fawn detection rate than our methods and
study site allowed. Nonetheless, taking into account
2  0 (13.3%)

4  0 (26.7%)
9  2 (60%)

Corti’s fertility rate of 0.72 and 50% survival over


Fawns

summer, the numbers we encountered in March,


15  2

when fawns are more visible, are still lower than


00
72
10
10

10
00
10

40

those observed in inland populations.


Although predation by culpeo is the leading cause
of huemul fawn mortality in many populations (Corti
et al. 2010), our data did not present decisive evi-
Bernardo Valley region

dence pointing to fox as a main cause of huemul


Valle Huemules

population decline since 2008. Although we docu-


Huemules region
and age classes

Ofhidro region

mented that huemul are being consumed by culpeo


Valle Toros
Valle Bajo

fox, with 15% of scat piles containing huemul hair,


Terrazas

Ofhidro
Refugio
Pampa
GLOF

culpeo are long-time residents (Frid 1999) and there


Sector

Total

is limited evidence to support the theory that huemul


possess inadequate predator responses due to
doi:10.1111/aec.13178 © 2022 Ecological Society of Australia.
CRITICAL HUEMUL POPULATION IN DECLINE 967

(a) (b)

Fig. 5. Spatial distribution of collected culpeo fox scat and encountered huemul remains from 2014 to 2020 in the sur-
roundings of Bernardo Fjord (a) and Ofhidro Fjord (b).

evolutionary absence of predators (Flueck & Smith- foot disease, which would have been introduced to
Flueck 2012). Rather, if Bernardo culpeo are heavily Bernardo by domestic cattle, is leading to elevated
reliant on huemul prey, we would expect the fox huemul mortality. In a study of foot lesions in 24 hue-
population to eventually decline in response to a mul from Bernardo O’Higgins National Park between
decline in huemul, allowing the population to 2005 and 2010, Vila et al. (2019) found that males
recover. That is, unless there are additional, and nov- were 2.59 more likely to be infected with parapoxvirus
el, species interactions influencing the system. A the- and 2.49 more likely to display lesions. Animals with
ory held by a long-term park guard (Paredes, F., foot disease show signs of intense pain and reduced
pers. comm., 11/21/19) is that since the previous mobility followed by loss of body condition and
study occurred, the abundance of non-native salmon recumbency, which often precedes death. Morbidity
in Bernardo Fjord has increased, providing a winter and mortality in that study reached 80% and 40%
food source for foxes that reduces winter mortality, respectively (Vila et al. 2019).
thus increasing predation pressure on huemul fawns. The higher male mortality due to parapoxvirus
Although this theory was not directly supported by could also potentially explain the skewed female pro-
our data as we only found fish scales in one fox scat, portion encountered in this study (0.64  SD 0.05),
ecosystem impacts of non-native salmonids have which is outside the range of 0.44–0.63 (mean of
been noted at other sites (Novaro et al. 2000; Zava- 0.51) estimated for the inland huemul population at
leta et al. 2001) and could warrant further research Lago Cochrane National Reserve (Wittmer
as a potential ecosystem change influencing huemul et al. 2010) and higher than ratios reported previ-
in Bernardo. ously for the same population (0.48 for Huemules
Alternatively, another compelling explanation of the Valley and 0.50 for Bernardo Valley; Brice~ no
apparent decrease in population size is that putative et al. 2013). This statement needs to be made with
parapoxvirus (Chordopoxvirinae subfamily)-associated caution, because it is common for dimorphic species

© 2022 Ecological Society of Australia. doi:10.1111/aec.13178


968 S. M. PACK ET AL.

with opportunistically polygynous mating systems, inland habitats where domestic dogs and habitat loss
such as deer, to exhibit a female-biased sex ratio present the main risks (Black-Decima et al. 2016);
(Mysterud et al. 2000). However, the differences in however, based on our findings, both inland and
female proportions when compared to inland popula- coastal populations may share the impacts of diseases
tions and from surveys of this population from previ- transmitted from domestic cattle. Active management
ous years suggest additional factors might be playing may improve conservation outcomes in these remote
a role in driving the current higher proportion of areas. For example, during our study, seven of the
females in Bernardo Fjord region. eight huemul treated for their foot injury by CONAF
Moreover, habitat and weather at Bernardo, in 2016 did not present with an injury when last re-
including extremely wet conditions and rugged, rocky sighted over a year later. Therefore, periodic sanitary
terrain, may create additional complications for hue- interventions could be an effective strategy for reduc-
mul affected with foot disease (Vila et al. 2006; ing huemul mortality from parapoxvirus-associated
CONAF 2015), and nutritional deficiencies linked to foot injuries in the future. In order to determine
marginal habitat conditions may confer additional whether parapoxvirus is indeed driving the observed
disease susceptibility (CONAF 2015). Huemul con- population decline, future studies might generate
fined to mountainous areas are known to have diets comparative demographic data, specifically female
deficient in trace minerals like Selenium, impairing proportion and fawn:female ratios, in coastal popula-
bone metabolism and resulting in hoof ailments, tions of Bernardo O’Higgins National Park without
lameness, and bone deformities, especially in males historic presence of cattle. Alternatively, should more
(Flueck et al. 2014; Flueck 2015, 2020; Flueck & research capacity become available in the Park, an
Smith-Flueck 2017). Our experience and those of experimental approach to management could be
local park guards and managers indeed indicates that explored, which would test the efficacy of various
small, deformed antler are the norm at Bernardo treatments addressing different hypothesized causes
(CONAF 2015), as well as are foot problems result- of the decline. These treatments, each applied to dif-
ing in limping, swollen joints and mortality. If para- ferent sectors, could involve treating parapoxvirus
poxvirus is indeed driving the decline of this lesions, providing mineral supplements and control-
population, directly or indirectly through changes in ling fox populations.
adult sex ratio, it would dispel the claim that “no evi- Beyond reporting on the population trend of hue-
dence supports. . .disease from livestock as having mul in Bernardo Fjord, we have used an innovative
caused declines” in huemul (Flueck & Smith- method to account for undetected individuals that
Flueck 2012). can be applied in other studies that aim to estimate
population size of low-density, difficult-to-observe
species, especially those that display strong trends of
Management implications philopatry. However, this method requires detailed
data on fertility and survival rates, which in our case
The Bernardo Fjord region is currently zoned as an were only available for inland areas. Hence, the accu-
“Untouchable Zone” – a zoning similar to that of a racy of our population estimates is dependent on the
“wilderness” category – under the 2018 Bernardo similarity between demographic rates of fjords and
O’Higgins National Park Public Use Plan for the inland areas. This is an important topic for huemul
Aysen region (CONAF 2018). This zoning prohibits research and conservation, as our results suggest that
development and tourism and allows only visitation population dynamics differ between inland and
and use related to scientific study and management. coastal regions. For example, lower survival rates for
However, in our years at the site, we witnessed huemul in the fjords might explain some of the
groups ranging from tourists, adventure groups and encountered differences in female proportion or
fishermen visiting the Pampa and Refugio sectors. fawn:female ratio. If this is the case, our estimates
Given that it contains the world’s largest breeding might be overestimating the size of the huemul popu-
population of endangered huemul deer, one that lation in the Bernardo Fjord.
appears to be declining due to unclear causes, we Nonetheless, in the absence of local demographic
support a precautionary approach that maintains this parameters, estimating undetected individuals with
zoning and its active enforcement. the best available information is a step forward from
We have observed that this population, within a previous surveys in the region in which population
habitat that has experienced relatively little human estimates were based only on detected individuals
impact, has shown variation in density and demo- (Brice~no et al. 2013). To avoid overestimating the
graphic rates in recent years. Given its conservation huemul population, we were conservative in our
relevance, increased monitoring efforts are needed to methods by considering individuals of same sex and
better understand the possible causes of these trends. age observed in different years in the same sector as
Threats to huemul in the fjords likely differ from the same individual. Overall, despite its limitations,
doi:10.1111/aec.13178 © 2022 Ecological Society of Australia.
CRITICAL HUEMUL POPULATION IN DECLINE 969

we believe this to be a promising approach in the resources (equal); software (lead); supervision
cases were traditional methods such as distance sam- (equal); validation (lead); visualization (lead); writing
pling are not viable. For future improvement, we rec- – original draft (equal); writing – review and editing
ommend repeated surveys of the same area or the (equal). Matthew Pomilia: Conceptualization
permanent presence of park rangers to be able to (equal); data curation (supporting); formal analysis
compare and verify the number of estimated and (supporting); funding acquisition (lead); investigation
observed individuals each year. Despite the time and (supporting); methodology (lead); project administra-
effort associated with this approach, we believe the tion (lead); resources (lead); supervision (lead); vali-
conservation importance of the huemul population in dation (supporting); visualization (supporting);
the Bernardo Region, together with evidence pointing writing – review and editing (supporting). Adam
towards a recent decline, warrant continued monitor- Spencer: Conceptualization (supporting); data cura-
ing and active management of this population, tion (equal); investigation (lead); methodology (lead);
enhanced conservation protections and zoning project administration (equal); resources
enforcement, and investigation of possible drivers of (supporting); writing – original draft (supporting);
huemul population dynamics in the Chilean fjords. writing – review and editing (supporting). Dennis
Aldridge: Conceptualization (equal); data curation
(supporting); formal analysis (supporting); funding
ACKNOWLEDGEMENTS acquisition (equal); investigation (supporting);
methodology (equal); project administration (equal);
This project would not have been possible without sup- resources (equal); supervision (supporting); writing –
port from CONAF, especially the guardaparques Feli- review and editing (equal).
dor Paredes, Raul Pereda, Orlando Beltran, Fabio
Paredes, and Luis Paredes, who have accompanied us
with expertise and great ability in all our years working CONFLICT OF INTEREST
in Bernardo O’Higgins National Park. We thank Piero
Caviglia of CONAF Aysen for his leadership and sup- The authors declare that they have no relevant or
port of our expeditions, and Carolina Panichine Faun- material financial interests that relate to the research
des for her support in photo identification and ageing. described in this paper, and that they have no indus-
Last but not least, we thank the leaders of Round River trial links or affiliations that could create a conflict of
Conservation Studies including Fernando Iglesias interest with this research.
Letelier, Elizabeth Brunner, Scott Braddock, Valeria
Briones, Samara Moreira, Kyla Zaret, Gabriel Kayano,
Violeta Martinez Amigo and Jessica Hightower, and all REFERENCES
of the students who engaged in fieldwork through even
the most challenging weather from the Southern Patag- Black-Decima P. A., Corti P., Dıaz N. et al. (2016)
onia Icefield. The authors declare that they have no rel- Hippocamelus bisulcus. The IUCN Red List of Threatened
Species 2016: e.T10054A22158895. IUCN, Zurich,
evant or material financial interests that relate to the
Switzerland.
research described in this paper, that they have no Brice~
no C., Knapp L. A., Silva A. et al. (2013) Detecting an
industrial links or affiliations that could create a conflict increase in an Endangered huemul Hippocamelus bisulcus
of interest with this research, and that they received no population following removal of cattle and cessation of
financial support for the authorship or publication of poaching in coastal Patagonia, Chile. Oryx 47, 273–9.
this article. The author(s) received no financial support Buckland S. T., Rexstad E. A., Marques T. A. & Oedekoven
for the research, authorship, and/or publication of this C. S. (2015) Distance Sampling: Methods and Applications.
Springer International Publishing, Cham.
article.
Carrasco J. F., Casassa R. G. & Rivera A. (1998) Climatologıa
actual del Campo de Hielo Sur y posibles cambios por el
incremento del efecto invernadero. An. Inst. Patagon. 26,
AUTHOR CONTRIBUTIONS 119–28.
Caswell H. (2001) Matrix Population Models: Construction,
Shalynn McKenzie Pack: Conceptualization (lead); Analysis, and Interpretation. Sinauer Associates, Inc.,
data curation (lead); formal analysis (supporting); Sunderland.
Chehebar C. & Martın S. (1989) Guıa para el reconocimiento
investigation (equal); methodology (equal); project
microscopico de los pelos de los mamıferos de la
administration (lead); resources (equal); supervision Patagonia. Do~nana Acta Vertebr. 16, 247–91.
(equal); writing – original draft (lead); writing – CONAF (2015) Huemules en el Parque Nacional Bernardo
review and editing (lead). Diana Bertuol Garcia: O’Higgins: Aportes al conocimiento de su estado, distribucion y
Conceptualization (supporting); data curation (sup- abundancia en el sector norte del Parque, Comuna de Tortel,
porting); formal analysis (lead); investigation (equal); Region de Aysen. Corporaci on Nacional Forestal,
Coyhaique, Chile.
methodology (equal); project administration (equal);
© 2022 Ecological Society of Australia. doi:10.1111/aec.13178
970 S. M. PACK ET AL.

CONAF (2016) Captura de huemules en sector norte del Parque density in red deer: the importance of spatial scale.
Nacional Bernardo O’Higgins. Corporaci on Nacional J. Anim. Ecol. 69(6), 959–74.
Forestal, Coyhaique, Chile. Novaro A. J., Funes M. C. & Susan W. R. (2000) Ecological
CONAF (2018) Plan de Uso P ublico del Parque Nacional extinction of native prey of a carnivore assemblage in
Bernardo O’Higgins: sectores de la Region de Aysen. Argentine Patagonia. Biol. Conserv. 92, 25–33.
Corporaci on Nacional Forestal, Coyhaique, Chile. Povilitis A. (1986) Huemuls in areas adjacent to glaciers in
Corti P., Wittmer H. U. & Festa-Bianchet M. (2010) southern Chile. Mt. Res. Dev. 6, 273–5.
Dynamics of a small population of endangered huemul Povilitis A. (1998) Characteristics and conservation of a
deer (Hippocamelus bisulcus) in Chilean Patagonia. fragmented population of huemul Hippocamelus bisulcus in
J. Mammal. 91, 690–7. central Chile. Biol. Conserv. 86, 97–104.
Dıaz N. I. & Smith-Flueck J. A. (2000) El huemul patagonico: Povilitis A. (2002) El estado actual del Huemul (Hippocamelus
un misterioso cervido al borde de la extincion. LOLA, Buenos bisulcus) en Chile central. Gayana Concepc. 66, 59–68.
Aires. R Core Team (2021) R: A Language and Environment for
Flueck W. & Smith-Flueck J. A. (2012) Huemul heresies: Statistical Computing. R Foundation for Statistical
Beliefs in search of supporting data. 2. Biological and Computing, Vienna.
ecological considerations. Anim. Prod. Sci. 52, 694– Redford K. H. & Eisenberg J. F. (1992) Mammals of the
706. Neotropics. Vol. 2 – The Southern Cone. University of
Flueck W. T. (2015) Osteopathology and selenium deficiency Chicago Press, Argentina, Uruguay.
co-occurring in a population of endangered Patagonian Smith-Flueck J. M. (2003) La ecologıa del huemul
huemul (Hippocamelus bisulcus). BMC. Res. Notes 8, 330. (Hippocamelus bisulcus) en la Patagonia Andina de
Flueck W. T. (2020) Nutrition as an etiological factor causing Argentina y consideraciones sobre su conservaci on. PhD
diseases in endangered huemul deer. BMC. Res. Notes 13, Thesis. Universidad Nacional del Comahue, Bariloche.
276. Vila A. R., Brice~ no C., McAloose D. et al. (2019) Putative
Flueck W. T. & Smith-Flueck J. A. M. (2017) Troubling parapoxvirus-associated foot disease in the endangered
disease syndrome in endangered live Patagonian huemul huemul deer (Hippocamelus bisulcus) in Bernardo O’Higgins
deer (Hippocamelus bisulcus) from the Protected Park National Park, Chile. PLoS One 14, e0213667.
Shoonem: unusually high prevalence of osteopathology. Vila A. R., L opez R., Pastore H., Fa undez R. & Serret A.
BMC. Res. Notes 10, 739. (2006) Current distribution and conservation of the
Flueck W. T., Smith-Flueck J. A. M., Mincher B. J. & Winkel Huemul (Hippocamelus bisulcus) in Argentina and Chile.
L. H. E. (2014) An alternative interpretation of plasma Mastozool. Neotropical 13, 263–369.
selenium data from endangered Patagonian Huemul Deer Wensing D. A. (2005) Conservation study of the huemul
(Hippocamelus bisulcus). J. Wildl. Dis. 50, 1003–4. (Hippocamelus bisulcus) within the Bernardo O’Higgins
Frid A. (1994) Observations on habitat use and social National Park, Chile. MSc Thesis. Utrecht University,
organization of a Huemul Hippocamulus bisulcus coastal Netherlands.
population in Chile. Biol. Conserv. 67, 13–9. Wickham H., Averick M., Bryan J. et al. (2019) Welcome to
Frid A. (1999) Huemul (Hippocamelus bisulcus) sociality at a the tidyverse. J. Open Source Softw. 4, 1686.
periglacial site: sexual aggregation and habitat effects on Wittmer H. U., Corti P., Saucedo C. & Galaz J. L. (2010)
group size. Can. J. Zool. 77, 1083–91. Learning to count: adapting population monitoring for
Frid A. (2001) Habitat use by endangered huemul Endangered huemul deer Hippocamelus bisulcus to meet
(Hippocamelus bisulcus): cattle, snow, and the problem of conservation objectives. Oryx 44, 516–22.
multiple causes. Biol. Conserv. 100, 261–7. Zavaleta E. S., Hobbs R. J. & Mooney H. A. (2001) Viewing
Gaillard J. M., Festa-Bianchet M., Yoccoz N. G., Loison A. & invasive species removal in a whole-ecosystem context.
To€ıgo C. (2000) Temporal variation in fitness Trends Ecol. Evol. 16, 454–9.
Components and population dynamics of large herbivores.
Annu. Rev. Ecol. Syst. 31, 367–93.
Gill R., Saucedo G. C., Aldridge D. & Morgan G. (2008)
Ranging behaviour of huemul in relation to habitat and SUPPORTING INFORMATION
landscape. J. Zool. 274, 254–60.
Marın J. C., Varas V., Vila A. R., L opez R., Orozco-ter W. P.
Additional supporting information may/can be found
& Corti P. (2013) Refugia in Patagonian fjords and the
eastern Andes during the Last Glacial Maximum revealed online in the supporting information tab for this article.
by huemul (Hippocamelus bisulcus) phylogeographical Appendix S1. Additional methods.
patterns and genetic diversity (ed B. Riddle). J. Biogeogr.
40, 2285–98.
Appendix S2. Dataset and photo database online
Mysterud A., Yoccoz N. G., Stenseth N. C. & Langvatn R. repository.
(2000) Relationships between sex ratio, climate and Appendix S3. Additional results.

doi:10.1111/aec.13178 © 2022 Ecological Society of Australia.

You might also like