Full Terms & Conditions of access and use can be found at
Emu - Austral Ornithology
ISSN: 0158-4197 (Print) 1448-5540 (Online) Journal homepage: http://www.tandfonline.com/loi/temu20
Relationships among territory size, body size, and
food availability in a specialist river duck
Silvina Ippi, Gerardo Cerón, Leandro M. Alvarez, Rodrigo Aráoz & Pedro G.
To cite this article: Silvina Ippi, Gerardo Cerón, Leandro M. Alvarez, Rodrigo Aráoz & Pedro G.
Blendinger (2018): Relationships among territory size, body size, and food availability in a specialist
river duck, Emu - Austral Ornithology, DOI: 10.1080/01584197.2018.1438848
To link to this article: https://doi.org/10.1080/01584197.2018.1438848
View supplementary material
Published online: 25 Feb 2018.
Submit your article to this journal
View related articles
View Crossmark data
Relationships among territory size, body size, and food availability in a
specialist river duck
, Gerardo Cerón
, Leandro M. Alvarez
, Rodrigo Aráoz
and Pedro G. Blendinger
Departamento de Zoología, CRUB Universidad Nacional del Comahue –CONICET, Bariloche, Argentina;
Asociación para la Conservación de
la Diversidad Biológica Argentina (BIOTA), Mendoza, Argentina;
Centro de Investigaciones y Transferencia de Jujuy (CIT), Universidad
Nacional de Jujuy –CONICET, San Salvador de Jujuy, Argentina;
Instituto de Ecología Regional, Universidad Nacional de Tucumán –
CONICET, Tucumán, Argentina
Models of territorial behaviour predict a reduction in territory size when food availability and
competitive pressure increase. Moreover, body size can play an important role in territorial
defence. The Torrent Duck (Merganetta armata) is a river specialist that exhibits year-round
territorial behaviour and long-term pair bonds. Food availability measured as biomass of inverte-
brates per unit area, territory and body size of Torrent Ducks were studied in the Andes in
Argentina to test predictions of territoriality models. The availability of aquatic invertebrates
decreased with latitude, while territory size increased. As expected, territory size of Torrent Ducks
showed a negative relationship with the availability of aquatic invertebrates, a major food source
for Torrent Ducks. Larger males and females paired together and occupied territories with greater
food availability. Body size may be important for both males and females for the successful
acquisition and defence of territories, especially during the non-breeding season when the
contest winner acquires or maintains the territory and the mate. Our results suggest that
Torrent Duck specialisation on fast-flowing mountain rivers leads to year-round territoriality in
both sexes, a positive correlation between territorial defence and body size, and territory size
proportional to food availability and population density.
Received January 2017
Accepted January 2018
Andes range; food
armata; mountain rivers;
territorial defence; Torrent
Relationships among food availability, foraging effi-
ciency and territory size have been a central issue in
ecology for decades (Brown 1964; Andersson 1978;
Adams 2001; López-Sepulcre and Kokko 2005).
Territory is defined as an area occupied almost exclu-
sively by an animal or group of animals by means of
excluding other members of the same species, directly
via contest competition or indirectly via advertisement
(Wilson 1975). For many species that exhibit food-
based territoriality, where food resources are distribu-
ted homogeneously in space and time, there is consid-
erable disparity in individual territory sizes (Carpenter
et al.1983; Norman and Jones 1984; Mares and Lacher
1987; Marshall and Cooper 2004). For such species,
having larger territories, increase their access to greater
absolute amounts of feeding (i.e. energetic) resources.
However, territory size can reach an optimum beyond
which defence, and movement (e.g. foraging or com-
muting) become overly costly (Andersson 1978).
Optimal foraging models predict a decrease in terri-
tory size with increasing food and competitor density
(Schoener 1983), with these predictions being supported
experimentally in a wide variety of taxa (Adams 2001).
Many studies involving different vertebrate taxa have
assessed the effect of food supply on population dynamics
and found that with the addition of food, territory size
usually decreases and population density increases
(Sinclair 1989;Boutin1990;Grantet al.1998; Newton
1998; Royama 2012). Nevertheless, models of optimal
territory size may not be applicable when animals defend
contiguous territories (Adams 2001), because such mod-
els do not consider interactions among neighbours in
their mathematical definitions. Unlike the non-contigu-
ous situation, where the unoccupied space between terri-
tories allows territory holders to expand their defended
area, contiguous territory holders defend smaller areas
than their non-contiguous optimum, as a result of the
pressure exerted by their neighbours (Hixon 1980;Grant
CONTACT Silvina Ippi email@example.com
Supplemental data for this article can be accessed here.
EMU - AUSTRAL ORNITHOLOGY, 2018
© 2018 BirdLife Australia
Territorial species invest part of their energy and time
defending their territories from conspecifics (Petrie 1984;
Price 1984;BartandEarnst1999; Candolin and Voigt
2001). Energy investment in defence should be related to
food resources and probably territory size. In some terri-
torial species, larger individuals have advantages over
smaller individuals when defending their territories, and
a positive relationship has been described between bird
size and territory size and/or quality (Quinard et al.
2014). Some birds can even adjust their territory sizes as
a function of gained weight (e.g. Carpenter et al.1983).
Worldwide, four species of anatids are river specialists
that inhabit fast-flowing montane rivers: the Blue Duck
(Hymenolaimus malacorhynchos) from New Zealand, the
African Black Duck (Anas sparsa) from sub-Saharan
Africa, Salvadori’sTeal(Salvadorina waigiuensis)from
New Guinea, and the Torrent Duck (Merganetta armata)
from South America (Eldridge 1986a). Although they are
not closely related (Livezey 1986), all river-specialist
ducks are monogamous, exhibiting strong territorial
behaviour and long-term pair bonds. Williams and
Mckinney (1996) suggest that these behaviours may be
adaptations to life in such specialised habitats. The
Torrent Duck lives in fast-flowing rivers, in the Andes
Range from Venezuela to southern Argentina
(Carboneras 1992). Pairs defend river sections consisting
of a mixture of rapids, pools, and waterfalls, generally of
1–2kminlength(Moffett1970), although there is a high
variability in territory sizes. In Colombia, for example,
territories as small as 200 m in length have been recorded
(Cardona and Kattan 2010). Torrent ducks establish ter-
ritory boundaries at specific locations such as emergent
rocks. The locations of territorial boundaries are fixed
year-round, even during the breeding season. At these
boundaries, pairs from contiguous territories perform
territorial displays many times per week when their
neighbours are present (Moffett 1970;Cerón2012). As
has been reported for other avian species with long-term
pair bonds and year-round territorial defence (e.g. Hall
and Peters 2008; Quinard and Cézilly 2012), both males
and females contribute to the defence of territories
Torrent Duck territories are narrow strips along river-
beds, and are defended year-round at only two boundaries
(upstream and downstream) by both members of the pair.
Territories can thus only vary in one dimension, constitut-
ing an interesting model system to assess the relationship
between food abundance, male and female body size, and
territory size. Territorial displays seem energetically expen-
sive (Eldridge 1986b), and the larger the territory, the more
energy is required to defend it. We therefore expect larger
territories to be defended only when smaller ones are
lacking in resources to sustain the pair. In other words,
we predict that territory size will be negatively correlated
with food density, as the optimal models predict.
In order to explore factors affecting the territory size
of Torrent Duck pairs, we conducted a study evaluating
(i) the latitudinal variation in territory size and food
resources over 25 degrees across the southernmost
distribution of the species, and (ii) whether variation
in the territory size of Torrent Ducks is related to food
availability. In addition, we tested the hypothesis that
larger male and female ducks will have territories with
higher food density, assuming that a larger body size is
advantageous for defending territories during intra-
sexual confrontations. If this last hypothesis is true,
there will be a positive correlation between the body
sizes of paired males and females. Understanding the
mechanisms involved in territory delimitation and
defence could provide new information on ecological
traits that lead to territoriality.
Fieldwork was carried out in Torrent Duck territories
located in fast-flowing rivers, spanning almost the entire
latitudinal range of the species in the Argentinian Andes
(24°7′S–49°8′S).Four regions representative of the envir-
onments inhabited by Torrent Ducks in this area were
selected, and a total of 25 territorial pairs were sampled
(Table S1; Figure 1). In the northern region, Torrent
Ducks inhabit rivers in the Andean montane forest
known as southern Yungas, where the mean annual pre-
cipitation is 2000 mm and the average annual tempera-
ture is 13.9°C (Brown et al.2005). In the central-north
region, the mean annual precipitation (mostly snow) is
600 mm, and the mean annual temperature is 14°C
(Abraham and Martínez 2000). In the central-south
region, samples were taken in the temperate austral for-
est. The mean annual precipitation is 1800 mm, concen-
trated mainly as rain and snow in winter, while summers
are dry. The average annual temperature is 14°C
(Mermoz et al.2009). The southern region consists of
cold austral forest, where the average annual precipitation
is 1500 mm, distributed year-round, with a greater con-
centration of rain and snow in the winter and an average
annual temperature of 7.5°C (Cabrera 1976).
Field data were collected during the breeding season (i.e.
austral spring–summer), from September to December
2014–2015, within a month of duckling hatching, and
only in territories where ducklings were present (i.e.
2S. IPPI ET AL.
successful territories). We considered only successful ter-
ritories in order to ensure stability in their size and loca-
tion, since recently established territories could change in
size or be abandoned during the following months
(Cardona and Kattan 2010). During the breeding season,
territorial behaviour reaches its highest level in the year
(Cerón 2012). Hatching takes place in October in the
northern region, November in the central regions and
December in the southern region. A single observer (G.
Cerón) collected all samples and took all measurements,
in order to minimise potential measurer effects.
Torrent Ducks feed on benthic aquatic invertebrates in
fast-flowing rivers (Naranjo and Ávila 2003; Cerón and
Boy 2014), with males and females feeding at the same
microsites. They use two feeding techniques, either
using their beak to scrape the top, side, and down-
stream surfaces of large rocks to remove adhered inver-
tebrates, or searching for invertebrates on the riverbed
at the bottom and between small rocks (hereinafter
‘scraping’and ‘searching’techniques; Cerón and Boy
To estimate food availability, we collected aquatic
invertebrates, endeavouring to imitate Torrent Duck
feeding strategies (see below) according to the propor-
tion of substratum types (which determines the pro-
portion of taxa in each site), and territory size (total
square metres) as in Cerón and Boy (2014).
In each territory, a flexible 0.09 m
was placed on 15 randomly selected sites of two differ-
ent substrates to sample invertebrate availability (total-
ling an area of 1.35 m
within each pair territory)
according to the following Torrent Duck feeding stra-
tegies: (1) ‘Scraping’: 0.09 m
of large boulders (>40 cm
diameter) were brushed on the top, sides and down-
stream faces. (2) ‘Searching’: 0.09 m
of river bed
(rocks <40 cm diameter) were removed, and the small
rocks were brushed into a Surber net (0.09 m
mesh) held in the downstream current. The number of
aquatic invertebrates per square metre sampled by the
) and ‘searching’(I
techniques was estimated by dividing sampled inverte-
brates (I) by 1.35 m
). Invertebrate availability
per square metre for each territory (AI/m
) was calcu-
ðAI=m2Þ¼ðIsc=m2Þproportion of large rocks
þðIse=m2Þproportion of small rocks:
In each territory, the proportion of submerged small
(<40 cm diameter) and large (>40 cm diameter) rocks
was estimated visually for 100 m sections of the river
until the entire length of the territory was covered, and
the values were averaged.
Invertebrates were identified at family or genus
taxonomic level according to Merritt and Cummins
(1997) and Fernández and Dominguez (2001).
Invertebrates of each taxon from food-availability sam-
ples were counted and their fresh weight determined
(0.01 mg) to estimate biomass per square metre. Items
not found in faeces (see below) or not reported in the
literature as eaten (Naranjo and Ávila 2003; Cerón and
Boy 2014; Vera et al.2014) were excluded from the
calculations to estimate food availability. A reference
collection was made using invertebrates’head capsules,
legs and/or mandibles from food samples to compare
with sclerotised body parts resistant to Torrent Duck
digestion, and this was used to determine diet
Figure 1. Torrent Duck (Merganetta armata) subspecies distri-
bution and sampling areas in the Argentinian Andes: 1 and 2:
north region; 3 and 4: central-north region; 5: central-south
region; and 6: south region.
EMU - AUSTRAL ORNITHOLOGY 3
To determine the prey items eaten by the Torrent
Duck, 10 faeces were collected per territory in rivers
with one territory, and four faeces per territory in
rivers with three territories. These faeces were collected
at the same places and times as the food samples. Fresh
faeces were collected from emergent boulders and
identiﬁed by size (approximately 2.5 cm long) and
content (large proportion of sand and small pebbles
mixed with invertebrates), which differentiates them
from droppings of other birds in the study area
(which are small or composed mostly of uric acid).
Dry faeces were avoided in order to reduce losing
sample material due to weathering. Collected faeces
were preserved in 80% alcohol pending laboratory
analysis. Faeces were disaggregated under a stereo-
scopic microscope and prey items were identiﬁed by
comparison with the reference collection. It was
assumed that faeces were dropped by territorial birds
and not by ﬂoaters, since Torrent Ducks are very
intolerant of intruders, especially during the breeding
season, and quickly evict floaters from their territories.
Territorial individuals seldom leave their territories,
which ensures that they feed in the sampling area
Individual traits and territory size
In each territory, mist nets (black nylon, 100 mm mesh
size, 2 m in height, 9–12 m in width) were placed
across the river, near the location of the territorial
pair. We waited for captures in camouflaged locations
near to the net, quickly removing any captured duck.
Each duck was banded using Darvic rings (size 9 mm)
and identified by a colour code. Once marked, culmen
length was measured from the beginning to the end of
the cornea structure on the upper side of the beak,
using a calliper (accuracy = 0.05 mm), after which the
bird was released. As culmen length is positively related
to other body size measurements in Torrent Ducks
(Gutiérrez-Pinto et al.2014), we used this trait as an
estimator of body size. Body weight can be an inaccu-
rate estimator of body size, because ducks may be
captured wet or dry, with males usually captured dry,
and females, which commonly swim or dive, usually
captured wet (Alza et al.2017; Cerón pers. obs.), which
can result in biased data.
To determine territory boundaries, after ringing the
ducks we watched them from camouflaged spots
located near the suspected territory limits during the
breeding season (based on the presence of different
pairs). Duck pairs perform territorial displays many
times per week at territory boundaries (i.e. emergent
rocks), confronting the contiguous pair. We watched
for these display events until we were able to identify
the precise limits between territories (i.e. displays were
performed more than three times at the same spots,
never trespassing these boundaries). This methodology
was replicated at both upstream and downstream
boundaries of each territory. In addition, some terri-
tories were monitored for several seasons or several
years (as part of other studies), and these boundaries
were maintained even when one member of the pair
was replaced. In the case of territories located in iso-
lated rapids, ducks were similarly watched from spots
located in the transitional section between rapids and
calm waters, to ensure that territorial individuals used
only the rapids section of the river. Torrent Ducks
spend several hours standing on emergent rocks,
where coloured rings are easy to see with binoculars.
Each boundary location was recorded using a GPS
(eTrex Vista®HCx). To estimate the territory size of
Torrent Ducks, the territory length was measured using
Google Earth Pro (126.96.36.19936, 2015). River width was
estimated by averaging 10 randomly selected widths
per territory, measured with a 50 m measuring tape.
This study is structured at three levels of analysis: (i)
individual level (including inter-sex variation), (ii) ter-
ritory level, and (iii) the relationship between indivi-
dual and territory characteristics. We explored
different relationships analytically, using linear mixed-
effect models (LMM; Bolker et al.2009) because of the
hierarchical structure of our data (i.e. one pair per
territory, territories nested in rivers, and rivers nested
To describe body size and the differences between
sexes, we compared culmen length in males and
females including sex as categorical explanatory vari-
able, culmen length as response variable, and terri-
tories, rivers and regions as random factors. In
addition, we calculated the coefficient of variation
(CV in percentage) of culmen length for males and
We correlated the culmen length between paired
male and female ducks. Since residuals were non-
homogeneously distributed, we used a generalised
least squares (GLS) model which allows heterogeneity
by including a weight function (Zuur et al.2009).
Different weight functions of variance were included
in models, one at a time, maintaining constant
response and explanatory variables. We selected as
4S. IPPI ET AL.
the best model the one with the lowest value of Akaike
information criterion corrected by the small sample
size (AICc; Zuur et al.2009). The second-best model
had delta AICc > 2, so we did not consider it in our
results (Burnham and Anderson 2004). In the selected
model, the variance was modelled as a power of the
male culmen length (weights function varPower),
because residual values decreased with increasing
male culmen length. This function allowed us to solve
the problem of variance heterogeneity for a LMM
(Zuur et al.2009).
In the first model for this section we explored how the
biomass of invertebrates per unit area (i.e. response
variable) varied with latitude, including latitude as
explanatory variable, and rivers as random factor
(regions were excluded here in testing the effect of
latitude). To test the hypothesis that territory size is
affected by food availability, we constructed a second
model, including the territory size as response variable
and the biomass of invertebrates per unit area as expla-
natory variable, including rivers in regions as random
factors. As biomass and latitude were strongly corre-
lated (see results from the first model), latitude was
excluded from this model in order to avoid multicolli-
nearity effects (Zuur et al.2009).
We explored whether territory size varied as a func-
tion of latitude through a model including territory size
as response variable, latitude as explanatory variable,
and river as random factor. Because normality and
homogeneity of variance assumptions were not met,
we modelled variance using different weights function
of variance (Zuur et al.2009), and compared the dif-
ferent models via the AICc criterion. The selected
model included the function varFixed. The second-
best model had a delta AICc > 2, so we did not include
it in our results (Burnham and Anderson 2004). The
selected model does not calculate a new parameter, but
assumes a positive relationship between variance of
territory size and latitude.
Relationship between individual and territory
To test the prediction that larger ducks can obtain and
maintain larger territories, we conducted an analysis
including culmen length as explanatory variable and
territory size as response variable. This analysis was
performed for both sexes separately (to avoid collinear-
ity between culmen length and sex; see results below),
and included rivers and regions as random factors. In
order to analyse the variation in duck sizes in response
to food availability, the variation in culmen length was
modelled including the biomass of available food per
unit area and sex as explanatory variables. In this
model, territories were nested in rivers, and rivers in
regions to account for the non-independence of the
individuals from the same territory (breeding pair),
from the same river, and from the same region.
In all analyses, biomass per unit area of available
food was transformed via natural logarithm to achieve
normality and homogeneity of the residuals; interpre-
tation and figures are based on the transformed vari-
able. In LMM analysis, the proportion of variance
explained by explanatory variables (marginal R
by both explanatory and random effects (conditional
) was calculated, following Nakagawa and Schielzeth
(2013). Data analyses were performed with R 3.4.0 (R
Core Team 2016), using nlme, bbmle and MuMIn
packages. Data are provided as mean ± standard devia-
Twenty-five adult male and 24 female ducks from 25
territories in 11 rivers were captured in two consecu-
tive breeding seasons during 2014–2015 (in one terri-
tory we failed to capture the female).
The culmen length of Torrent Ducks was larger in
males (36.03 ± 0.99 mm; range: 34.60–38.40 mm)
than in females (33.97 ± 1.48 mm; range: 30.70–
36.60 mm; F
= 175.23; p< 0.0001). Male culmen
length was on average 6.34% longer than female cul-
men length. Females had higher variability than males
in culmen length (CV = 4.4 vs. CV = 2.7). The culmen
length of paired males and females was positively cor-
related (t= 11.30, p< 0.0001, df = 22. Pearson correla-
tion = 0.88) (Figure 2).
Torrent Duck territories were highly variable in size,
food availability and proportions of substratum type
(supplementary material Table S2). The mean territory
size of a duck pair was 7173.96 ± 3425.66 m
(range = 2992–15 120 m
). The mean estimated total
food biomass was 36.22 ± 54.51 kg per territory
(range = 0.12–226.15 kg). The available invertebrate
biomass per unit area averaged 6.03 ± 9.12 g/m
territory (range = 0.01–39.25 g/m
). The mean percen-
tage of small rocks present in the river substratum was
EMU - AUSTRAL ORNITHOLOGY 5
45.6 ± 20.0 (range = 0–95), and the mean percentage of
large rocks was 54.4 ± 20.0 (range = 5–100).
Invertebrate availability decreased as latitude increased
=13.99;p=0.005;Figure 3). The marginal R
0.53, while the conditional R
was 0.90. Territory size
showed a negative relationship with invertebrate avail-
=9.67;p=0.008;Figure 4(B)). The marginal
for this model was 0.38, while the conditional R
0.75. The territory size of Torrent Ducks increased with
increasing latitude (F
The marginal R
was 0.48, and the conditional R
Relationship between individual and territorial
Culmen length was not related to territory size for either
= 0.58) or females (F
p= 0.941; marginal R
< 0.01, conditional R
Culmen length was larger in territories with higher
invertebrate availability (F
= 7.80; p= 0.015;
Figure 6). There was no significant interaction between
sex and invertebrate availability on the culmen length
= 3.41; p= 0.078; marginal R
= 0.48, conditional
Torrent Duck territory size varied considerably,
increasing with latitude, over the 25 degrees of latitude
that we studied. The availability of aquatic prey species
decreased with latitude, as expected for invertebrate
biomass and densities in river systems (Cummins
1974; Rosenzweig 1995). The inverse relationship
observed between territory size and biomass of the
main food source for Torrent Ducks suggests that the
availability of aquatic invertebrates is a relevant factor
influencing the maintenance and defence of territories
through optimal foraging behaviours. Finally, as
expected, in our study area, larger males were paired
with larger females, occupying territories with higher
food density, i.e. better quality territories.
Territory is usually defended by male birds, particu-
larly in species in which males are larger than females
(e.g. Price 1984; Hyman et al.2004). However,
although male Torrent Ducks are approximately 6%
larger than females (Carboneras 1992), both sexes
engage in territorial defence (Cerón 2012). Positive
correlations between body size and territory size
(Adams 2001; Woodward et al.2005), or between
male body size and territory quality have been recorded
in several bird species (e.g. Bart and Earnst 1999).
However, we found no relationship between body size
and territory size for the Torrent Duck, despite the
positive relationship between body size and food
During the breeding season, male and female
Torrent Ducks defend their territories together, regard-
less of the intruder’s sex. However, during the non-
breeding season, males fight only with males and
females only with females, without the support of
their mates, which will remain in the territory with
the contest winner (Cerón 2012). Body size can play
an important role in territorial encounters in birds (e.g.
Petrie 1984; Price 1984; Bart and Earnst 1999), and,
although we did not conduct experiments to assess
directly whether body size is decisive in defining the
winner of each encounter, we found that larger ducks
defended territories with higher invertebrate availabil-
ity. If body size is advantageous in territorial encoun-
ters for both males and females, and since re-pairing
mostly involves intra-sexual confrontations (Cerón
Figure 2. Relationship between culmen length of paired males
and females in Torrent Ducks: correlation equation: female
culmen length = −9.16 + 1.20*male culmen length; Pearson
R= 0.88. See supplementary material S3 for parameter
Figure 3. Latitudinal variation in biomass of available food (inver-
tebrates) per unit area (log (kg/m
) in the four sampling regions in
the Argentinian Andes: density of food = 7.15–0.20*latitude;
= 0.53, conditional R
= 0.90. See supplementary
material S3 for other parameter estimates in this model.
6S. IPPI ET AL.
2012), a positive correlation between male and female
body size should occur. Larger body sizes would thus
allow for better quality territory, with access to a better
quality mate (Erikstad et al.1993; Hepp et al.1993). In
addition, beaks play a significant role during Torrent
Duck territorial displays (Moffett 1970; Úbeda et al.
2007), suggesting that beak length could be an honest
signal of individual quality (Zahavi 1975; Quinard et al.
2014). Therefore, we suggest that intra-sexual territor-
ial confrontations expressed by males and females dur-
ing the non-breeding season (Cerón 2012), with
advantages for larger body sizes, could explain why
larger male and female Torrent Ducks were paired
The size of territories defended by Torrent Ducks
was highly variable and negatively related to food avail-
ability, as expected in animals that display food-based
territoriality (Carpenter et al.1983; Norman and Jones
1984; Mares and Lacher 1987; Marshall and Cooper
2004), and is predicted by the optimal foraging models.
According to these models, territory size is also
expected to decrease with increasing competitor
density (Schoener 1983; Adams 2001). In the Andean
rivers of northern Argentina, Torrent Duck densities
ranged from 1.7 to 4.2 ducks/km (Sardina Aragón et al.
2011) and from 0.7 to 3.5 ducks/km in the central-
north region (Álvarez et al.2014). Small populations
have been recorded in the central-south region, but
without density estimations (Cerón and Trejo 2012),
and density was estimated at 0.5 ducks/km in the south
region (Vila and Aprile 2005). The latitudinal increase
in territory size can therefore be explained both by the
steep decline in food availability and by the decrease in
competitor densities. The river specialist Blue Duck
from New Zealand can also markedly reduce territory
size in response to increasing population densities
(Williams 1991). However, to fully understand the
possible regulatory mechanisms of intraspecific inter-
actions on the territory size of anatids specialised in
fast flooded rivers, variation in duck density and terri-
tory size should also be studied at the population level.
Our results showed that territory size was negatively
and significantly related to food density. However,
much of the variability in territory size and food
Figure 4. Variation in territory size of Torrent Duck breeding pairs in the Argentinian Andes in response to (A) biomass of available
food (invertebrates) per unit area (log (kg/m
): territory size = 7768.64 −1023.79*biomass per unit area (log); marginal R
= 0.75; and (B) latitude: territory size = 429.76 + 198.87*latitude; marginal R
= 0.48, conditional R
= 0.98. See
supplementary material S3 for other parameter estimates in this model.
EMU - AUSTRAL ORNITHOLOGY 7
availability resided in random effects, i.e. in the natural
structure of our study system (note the differences
between marginal R
and conditional R
and Schielzeth 2013). For instance, food availability
explained 38% of the total variance in territory size,
while the combined effects of food availability and the
nested structure of rivers in the regions explained 75%
of the total variance. The tendency to a more strongly
negative relationship between food availability and ter-
ritory size in the southern region than in the others
Figure 5. Variation in territory size in relation to culmen length of (A) female (territory size = 9246.51 −29.44*female culmen); and
(B) male (territory size = 684.84 + 208.27*male culmen) Torrent Ducks in the four sampling regions. See supplementary material S3
for parameter estimates.
Figure 6. Relationships between culmen length of male (black symbols) and female (grey symbols) Torrent Ducks and aquatic
invertebrate biomass per unit area in four regions of the Argentinian Andes: (A) north; (B) central-north; (C) central-south; and
(D) south regions. Different predicted lines with different intercepts for each region are shown. See supplementary material S3
for parameters estimates.
8S. IPPI ET AL.
(Figure 4(A)) suggests that territory size is more
strongly related to duck densities and food availability
where food availability is limiting. The reasons that
caused the slight differences in the relationship
between food availability and territory size in the dif-
ferent regions (Figure 4(A)) require further studies
including, for example, flood events. Flood events
have a negative effect on Torrent Duck populations
(Pernollet 2010), probably because they reduce inverte-
brate densities, as has been described for the Blue Duck
(Veltman et al.1991; Godfrey et al.2003). Therefore,
although territory size seems to be influenced by food
availability and duck density, it could also be expected
to be influenced by other characteristics of territory
quality not evaluated in our study, such as the avail-
ability of shelters and nesting sites, and the stability of
the territory when floods occur. The relative impor-
tance of these variables could be larger in areas less
limited by food, such as the northern and central-north
regions in the Argentinian Andes.
For habitat-specialist species, habitat availability may be
a limiting resource which affects population size because
competition for space may restrict the breeding popula-
tion, favouring the most aggressive and/or larger birds
(Brown 1964). High territoriality in the Torrent Duck
supports the idea that territories are a valuable and scarce
or limiting resource (Cerón 2012). Moreover, the best
territories in terms of food availability are owned by the
pairs of ducks with the largest body sizes. However, the
spatial structure of territory quality highlights the impor-
tance of using empirical data across the species distribution
range to assess models of bird territoriality. We suggest
that the strong selective pressure imposed by a changing
environment such as mountain rivers (Benistom 2003)
may have led the Torrent Duck to develop territorial
behaviour involving (i) year-round territoriality in both
sexes (Eldridge 1986b), (ii) territory defence positively
related to body size, and (iii) territory sizes that depend
on food availability and population densities.
We are grateful to Daniela Zaffignani, Jorge Gomez, Michelle
Delaloye, Nicolás Ferreiro, Rodolfo Freire, Sebastián Otta,
Silvina Sturzenbaum, Soledad Hourmilougue, and Soledad
Ovando for their field support and technical assistance dur-
ing this study. Sebastian Abades kindly helped us with sta-
tistical analyses and Nataly Dudinszky with the English
Abraham, M., and Martínez, F. R. (2000). ‘Recursos y
Problemas Ambientales de Zona Árida. Primera Parte:
Provincias de Mendoza, San Juan y La Rioja.
Caracterización Ambiental.’(IADIZA: Mendoza,
Adams, E. S. (2001). Approaches to the study of territory size
and shape. Annual Review of Ecology, Evolution, and
Systematics 32, 277–303. doi:10.1146/annurev.
Álvarez, L. M., Astié, A. A., Debandi, G. O., and Scheibler, E.
E. (2014). Effect of food availability and habitat character-
istics on the abundance of Torrent Ducks during the early
breeding season in the central Andes, Argentina. The
Wilson Journal of Ornithology 126, 525–533. doi:10.1676/
Alza, L., Bautista, E., Smith, M., Gutiérrez-Pinto, N., Astie,
A., and Mccracken, K. G. (2017). Capture efficiency of
torrent ducks by the active mist-net method. Wildlife
Society Bulletin. doi:10.1002/wsb.757
Andersson, M. (1978). Optimal foraging area: Size and allo-
cation of search effort. Theoretical Population Biology 13,
Bart, J., and Earnst, S. L. (1999). Relative importance of male
and territory quality in pairing success of male rock ptar-
migan (Lagopus mutus). Behavioral Ecology and
Sociobiology 45, 355–359. doi:10.1007/s002650050571
Benistom, M. (2003). Climatic change in mountain regions:
A review of possible impacts. Climatic Change 59,5–31.
Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W.,
Poulsen, J. R., Stevens, H. H., and White, J.-S.-S. (2009).
Generalized linear mixed models: A practical guide for
ecology and evolution. Trends in Ecology & Evolution 24,
Boutin, S. (1990). Food supplementation experiments with
terrestrial vertebrates patterns problems and the future.
Canadian Journal of Zoology 68, 203–220. doi:10.1139/
Brown, A., Martinez Ortiz, U., Acerbi, M., and Corcuera, J.
(2005). ‘La Situación Ambiental Argentina.’(Fundación
Vida Silvestre Argentina: Buenos Aires, Argentina.)
Brown, J. L. (1964). The evolution of diversity in avian
territorial systems. The Wilson Bulletin 76, 160–169.
Burnham, K. P., and Anderson, D. R. (2004). Multimodel
inference: Understanding AIC and BIC in model selection.
Sociological Methods & Research 33, 261–304. doi:10.1177/
Cabrera, A. L. (1976). ‘Regiones Fitogeográficas Argentinas.
Enciclopedia Argentina de Agricultura y Ganadería.’
(Primera imprenta: Buenos Aires, Argentina.)
Candolin, U., and Voigt, H. R. (2001). Correlation between
male size and territory quality: Consequence of male com-
petition or predation susceptibility? Oikos 95, 225–230.
Carboneras, C. (1992). Family Anatidae (ducks. geese and
swans). In ‘Handbook of the Birds of the World. Vol. 1’.
(Eds J. del Hoyo, A. Elliott, and J. Sargatal.) pp. 536–628.
(Lynx Edicions: Barcelona.)
Cardona, W., and Kattan, G. (2010). Comportamiento terri-
torial y reproductivo del pato de torrentes (Merganetta
armata) en la Cordillera Central de Colombia.
Ornitología Colombiana 9,38–45.
Carpenter, F. L., Paton, D. C., and Hixon, M. A. (1983).
Weight gain and adjustment of feeding territory size in
EMU - AUSTRAL ORNITHOLOGY 9
migrant hummingbirds. Proceedings of the National
Academy of Sciences 80, 7259–7263. doi:10.1073/
Cerón, G. (2012). Uso de hábitat y tendencias poblacionales
del pato de los torrentes (Merganetta armata armata)enel
Parque Nacional Nahuel Huapi. Ph.D. Thesis, Universidad
Nacional del Comahue, Río Negro, Argentina.
Cerón, G., and Boy, C. C. (2014). Prey selection and energy
value of main food items of the torrent duck (Merganetta
armata) in Northwestern Patagonia. Argentina.
Waterbirds 37, 153–161. doi:10.1675/063.037.0204
Cerón, G., and Trejo, A. (2012). Torrent Duck (Merganetta
armata) population trend in northwestern Patagonia,
Argentina. Ornitologia Neotropical 23, 407–415.
Cummins, K. W. (1974). Structure and function of stream
ecosystem. BioScience 24, 631–640. doi:10.2307/1296676
Eldridge, J. L. (1986a). Territoriality in a river specialist: The
Blue Duck. Wildfowl 37, 123–135.
Eldridge, J. L. (1986b). Observations on a pair of Torrent
Ducks. Wildfowl 37, 113–122.
Erikstad, K. E., Bustnes, J. O., and Moum, T. (1993). Clutch-
size determination in precocial birds: A study of the
Common eider. The Auk 110, 623–628. doi:10.2307/
Fernández, H. R., and Dominguez, E. (2001). ‘Guía para la
determinación de los artrópodos bentónicos sudamerica-
nos.’(Tucumán University Press: Tucumán, Argentina.)
Godfrey, J. D., Bryant, D. M., and Williams, M. (2003).
Energetics of blue ducks in rivers of differing physical
and biological characteristics. Science for Conservation
Grant, J. W. A. (1997). Territoriality. In ‘Behavioural Ecology
of Teleost Fishes’. (Ed J.-G. J. Godin.) pp. 81–103. (Oxford
University Press: Oxford.)
Grant, J. W. A., Steingríson, S. Ó., Keeley, E. R., and Cunjak,
R. A. (1998). Implications of territory size for the measure-
ment and prediction of salmonid abundance in streams.
Canadian Journal of Fisheries and Aquatic Sciences 55,
Gutiérrez-Pinto, N., McCracken, K. G., Alza, L., Tubaro, P.,
Kopuchian, C., Astie, A., and Cadena, C. D. (2014). The
validity of ecogeographical rules is context-dependent:
Testing for Bergmann’s and Allen’s rules by latitude and
elevation in a widespread Andean duck. Biological Journal
of the Linnean Society 111, 850–862. doi:10.1111/
Hall, M. L., and Peters, A. (2008). Coordination between the
sexes for territorial defence in a duetting fairy-wren.
Animal Behaviour 76,65–73. doi:10.1016/j.
Hepp, G. R., Stangohr, D. J., Baker, L. A., and Kennamer, R.
A. (1993). Factors affecting variation in the egg and duck-
ling components of Wood ducks. The Auk 104, 435–443.
Hixon, M. A. (1980). Food production and competitor den-
sity as the determinants of feeding territory size. The
American Naturalist 115,10–530. doi:10.1086/283577
Hyman, J., Hughes, M., Searcy, W. A., and Nowicki, S.
(2004). Individual variation in the strength of territory
defense in male song sparrows: Correlates of age, territory
tenure, and neighbor aggressiveness. Behaviour 141,15–
Livezey, B. C. (1986). A phylogenetic analysis of recent
Anseriform genera using morphological characters. Auk
López-Sepulcre, A., and Kokko, H. (2005). Territorial
defense, territory size, and population regulation. The
American Naturalist 166, 317–325. doi:10.1086/432560
Mares, M. A., and Lacher, T. E. Jr. (1987). Social spacing in
small mammals: Patterns of individual variation.
American Zoologist 27, 293–306. doi:10.1093/icb/27.2.293
Marshall, M. R., and Cooper, R. J. (2004). Territory size of a
migratory songbird in response to caterpillar density and
foliage structure. Ecology 85,432–445. doi:10.1890/02-0548
Mermoz, M., Úbeda, C., Grigera, D., Brion, C., Martín, C.,
Bianchi, E., and Planas, H. (2009). ‘El Parque Nacional
Nahuel Huapi. Sus características ecológicas y estado de
conservación.’(APN Press: Bariloche, Argentina.)
Merritt, R. W., and Cummins, K. W. (Eds.) (1997). ‘An
Introduction to the Aquatic Insects of North America.’
(Kendall/Hunt Publishers: Dubuque, Iowa.)
Moffett, G. M. (1970). A study of nesting Torrent Ducks in
the Andes. Living Bird 9,5–27.
Nakagawa, S., and Schielzeth, H. (2013). A general and
simple method for obtaining R
from generalized linear
mixed-effects models. Methods in Ecology and Evolution 4,
Naranjo, L. G., and Ávila, V. J. (2003). Distribución habita-
cional y dieta del pato de torrentes (Merganetta armata)
en el Parque Regional Natural Ucumari en la Cordillera
central de Colombia. Ornitología Colombiana 1,22–28.
Newton, I. (1998). ‘Population Limitation in Birds.’
(Academic Press: New York.)
Norman, M. D., and Jones, G. P. (1984). Determinants of
territory size in the pomacentrid reef fish. Parma
Victoriae. Oecologia 61,60–69. doi:10.1007/BF00379090
Pernollet, C. (2010). Selección de hábitat y efectos de las
crecidas en el pato cortacorrientes (Merganetta armata
armata) en dos ríos intervenidos de la Región de
O’Higgins (Chile Central): implicancias para su
conservación. Master Thesis, Chilean University,
Santiago de Chile.
Petrie, M. (1984). Territory size in the moorhen (Gallinula
chloropus): An outcome of RHP asymmetry between
neighbours. Animal Behaviour 32, 861–870. doi:10.1016/
Price, T. (1984). Sexual selection on body size, territory, and
plumage variables in a population of Darwin’s finches.
Evolution 38, 327–341. doi:10.1111/evo.1984.38.issue-2
Quinard, A., and Cézilly, F. (2012). Sex roles during conspe-
cific territorial defence in the Zenaida dove, Zenaida
Aurita.Animal Behaviour 83,47–54. doi:10.1016/j.
Quinard, A., Dechaume-Moncharmont, F.-X., and Cézilly, F.
(2014). Pairing patterns in relation to body size, genetic
similarity and multilocus heterozygosity in a tropical
monogamous bird species. Behavioral Ecology Sociobiol
68, 1723–1731. doi:10.1007/s00265-014-1780-1
R Core Team (2016). R: A language and environment for
statistical computing. (Foundation for Statistical
Computing, V., Austria, Ed.) Available at http://www.R-
Rosenzweig, M. L. (1995). ‘Species Diversity in Space and
Time.’(Cambridge University Press: Cambridge.)
10 S. IPPI ET AL.
Royama, T. (2012). ‘Analytical Population Dynamics. Vol.
10.’(Springer Science & Business Media: New York.)
Sardina Aragón, P. N., Rivera, L., and Politi, N. (2011).
Variación de la abundancia del pato de torrente
(Merganetta armata) y características del hábitat en dos
ríos de montaña de la provincia de Jujuy, Argentina.
Ornitología Neotropical 22, 589–599.
Schoener, T. W. (1983). Simple models of optimal-feeding
territory size: A reconciliation. The American Naturalist
121, 608–629. doi:10.1086/284090
Sinclair, A. R. E. (1989). The regulation of animal popula-
tions. In ‘Ecological Concepts: The Contribution of
Ecology to an Understanding of the Natural World’. (Ed
J. M. Cherrett.) pp. 197–241. (Blackwell: Oxford.)
Úbeda, C., Cerón, G., and Trejo, A. (2007). Descripción de
un nuevo comportamiento en hembra de pato cortacor-
rientes (Merganetta armata, Anatidae). Boletín Chileno de
Veltman, C. J., Triggs, S., Williams, M., Collier, K. J., McNab, B.
K., Newton, L., Haskell, M., and Henderson, I. M. (1991).
The blue duck mating system are river specialists any differ-
ent? In ‘Acta XX Congressus Internationalis Ornithologici’.
(Ed B. D. Bell.) pp. 860–867. (New Zealand Ornithological
Congress Trust Board: Wellington, NZ.)
Vera, J., Ríos Zapata, A., and Cerón, G. (2014). Selección de
alimento del pato de los torrentes (Merganetta armata)en
la cuenca alta del río Quindío, Colombia. Ornitología
Neotropical 25, 145–157.
Vila, A. R., and Aprile, G. 2005. Línea de base ‘Pato de los
torrentes’(Merganetta armata). Estancia ‘Los Huemules’–
El Chaltén, Santa Cruz, Argentina. Univ. Nacional de la
Patagonia Austral –Cielos Patagónicos S. A, Caleta Olivia,
Williams, M. (1991). Social and demographic characteristics
of blue duck Hymenolaimus malacorhynchos. Wildfowl
Williams, M., and Mckinney, F. (1996). Long-term mono-
gamy in a river specialist- the Blue Duck. In 'Partnerships
in Birds: the Study of Monogamy'. (Ed J. M. Black.) pp.
73-90. (Oxford University Press: Oxford.)
Wilson, E. O. (1975). ‘Sociobiology: The New Synthesis’.
(Harvard University Press: Cambridge.)
Woodward, G., Ebenman, B., Emmerson, M., Montoya, J.
M., Olesen, J. M., Valido, A., and Warren, P. H. (2005).
Body size in ecological networks. Trends in Ecology and
Evolution 20, 402–409. doi:10.1016/j.tree.2005.04.005
Zahavi, A. (1975). Mate selection –A selection for a handi-
cap. Journal of Theoretical Biology 53, 205–214.
Zuur, A. F., Leno, E. N., Walker, N. J., Saveliev, A. A., and
Smith, G. M. (2009). ‘Mixed Effects Models and
Extensions in Ecology with R’. (Springer: New York.)
EMU - AUSTRAL ORNITHOLOGY 11