Habitat suitability and anthropogenic correlates of Neotropical river otter (Lontra

longicaudis) distribution
Author(s): Jonatan J. Gomez, Juan I. Túnez, Natalia Fracassi, and Marcelo H. Cassini
Source: Journal of Mammalogy, 95(4):824-833. 2014.
Published By: American Society of Mammalogists
DOI: http://dx.doi.org/10.1644/13-MAMM-A-265
URL: http://www.bioone.org/doi/full/10.1644/13-MAMM-A-265

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and
environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published
by nonprofit societies, associations, museums, institutions, and presses.
Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of
BioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use.
Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries
or rights and permissions requests should be directed to the individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research
libraries, and research funders in the common goal of maximizing access to critical research.
Journal of Mammalogy, 95(4):824–833, 2014

Habitat suitability and anthropogenic correlates of Neotropical river

otter (Lontra longicaudis) distribution
JONATAN J. GOMEZ,* JUAN I. TÚNEZ, NATALIA FRACASSI, AND MARCELO H. CASSINI
Grupo de Estudios en Ecologı́a de Mamı́feros, Departamento de Ciencias Básicas, Universidad Nacional de Luján, Rutas
5 y 7, 6700 Luján, Argentina (JJG, JIT, MHC)
Laboratorio de Biologı́a del Comportamiento, Instituto de Biologı́a y Medicina Experimental, Obligado 2490, 1463
Buenos Aires, Argentina (JJG, MHC)
Área Recursos Naturales, Estación Experimental Agropecuaria Delta, Instituto Nacional de Tecnologı́a Agropecuaria,
Paraná de las Palmas y Canal L. Comas, Buenos Aires, Argentina (NF)
* Correspondent: gomezjonatanjose@yahoo.com.ar

The Neotropical river otter Lontra longicaudis is a top predator in many South and Central American aquatic
freshwater systems. Its current category in the International Union for Conservation of Nature is ‘‘data deficient,’’
which makes it imperative to determine the appropriate conservation status. We applied species distribution
models to build a map of habitat suitabilities, and to identify possible anthropogenic factors that affect the
presence of L. longicaudis in the Lower Delta of the Paraná River in the Southern Cone of South America.
Presence/absence of L. longicaudis was obtained using 3 methodologies (sign surveys, camera traps, and
interviews) and 15 environmental predictors. Habitat suitability was higher in areas with polders built for
forestry, and lower in areas with human settlements and boat traffic. At present, geographic isolation and control
of access on private land and reserves appear to be effective at protecting wildlife in the Paraná Delta. Our study
demonstrates that species distribution models can be used for rapidly evaluating potential threats to wildlife.

Key words: flooding pulse, poaching, wetland, wildlife conservation

Ó 2014 American Society of Mammalogists
DOI: 10.1644/13-MAMM-A-265

Otters live in particularly fragile habitats such as wetlands, even to a certain degree of water pollution, at least in the short
coastal marine, and riverine ecosystems. All of these systems term. Kruuk (2006) also found that hunting can be an important
are often unstable due to natural causes and anthropogenic threat locally.
effects. For example, droughts, floods, and tides regularly or The Neotropical river otter, Lontra longicaudis, is a top
sporadically change the appearance and functioning of predator in many South and Central American aquatic
wetlands and coastal seas (Brinson 1993; Davis and Fitzgerald freshwater systems. It is distributed from northern Mexico
2004). Superimposed on these natural variations are anthropo- (328N) to Argentina (328S) and is adapted to a wide variety of
genic effects. Water pollution, regulation of flows, the aquatic habitats, including seashores and elevations up to 3,000
structural degradation of the coast, poaching, and human m (Blacher 1987; Emmons and Feer 1997). Its diet is based on
presence are some of the known factors affecting wildlife in fish but it can include aquatic invertebrates such as crabs,
coastal and riverine environments (Davis and Fitzgerald 2004; crayfish and mussels, and terrestrial prey including mammals,
Erwin 2009). birds, and insects (Pardini and Trajano 1999; Quadros and
Otters are predators located at the top of aquatic environment Monteiro-Filho et al. 2001; Chemes et al. 2010). Home range
food webs and have large territories and habitat requirements size varies from 3 to 8 km (Gallo–Reynoso 1989). It is
(Parera 1996; Kruuk 2006). Most otter species are listed as probably the only Latin American otter species that is not at
endangered on the International Union for Conservation of risk of extinction. However, its current category is ‘‘data
Nature (IUCN) Red List. Kruuk (2006) conducted the most in- deficient’’ with the IUCN (Waldemarin and Alvarez 2008),
depth review on the ecology and conservation of otters to date.
He concluded that the most important overall threat to otters is
the impact of a reduction in food availability. Otters can be
tolerant to human presence, structural habitat degradation, and www.mammalogy.org
824
August 2014 GOMEZ ET AL.—NEOTROPICAL RIVER OTTER DISTRIBUTION 825

which makes it imperative to collect biological information on (Malvárez 1997), and is located approximately between
the species to assess the appropriate conservation status. 32805 0 S and 58830 0 W, and between 34829 0 S and 60848 0 W.
Species distribution models (SDMs) are associative models The Lower Delta is divided in 2 portions, 1 located in Entre
relating occurrence or abundance data at known locations of Rios Province, the other in Buenos Aires Province. The south
individual species (distribution data) to information on the borders a populated region of the Pampas plain with several
environmental characteristics of those locations (modified from major towns, including the Buenos Aires suburbs, one of the
Elith and Leathwick 2009). Several publications have reviewed 20 largest cities of the world. Our study area comprised the
the available SDMs (Guisan and Thuiller 2005; Heikkinen et Buenos Aires portion of the Lower Delta of the Paraná River
al. 2006; Elith and Leathwick 2009; Cassini 2011a, 2013). (Fig. 1). We divided it into a total of 114 cells of 0.05 3 0.058,
These reviews found that SDMs have been used with good approximately 5.5 km long, which approaches the average size
results to characterize the natural distributions of species and of the home range of L. longicaudis (3–8 km—Ortega et al.
that this information has been applied to investigate a variety of 2012, 6.29 6 5.96 km—Trinca et al. 2013).
scientific and applied issues. SDMs have 2 main uses: From a biogeographical point of view, this region has a high
identifying predictors or key factors of the environment that number of species of flora and fauna, more than expected for
affect species distributions, and predicting distributions in new similar-latitude continental areas (Burkart 1957; Quintana
scenarios, assuming that the variables included in the model are 1999). Many of these species are carried by the Paraná and
relevant factors. When SDMs are used for the latter purpose, Uruguay rivers from subtropical areas and coexist with those
their output is normally a habitat suitability map. The 1st from neighboring temperate areas, giving the region a unique
approach has a more theoretical focus, and considers the causal profile (Malvárez 1997). Its hydrological regime is expressed
drivers of species distributions. The second approach is through a fluvial–tidal gradient in the NW-SE direction, which,
influenced by the strong demand for mapped products for together with the presence of different landscape patterns,
use in conservation and land management. In recent years, determines a marked environmental heterogeneity (Kandus et
several authors (Papes and Gaubert 2007; Sepúlveda et al. al. 2003). The islands of the Lower Delta have a bucket shape,
2009; Morueta-Holmes et al. 2010; Wilting et al. 2010; Cassini surrounded by a peripheral levee enclosing a depressed central
2011b; Jackson and Robertson 2011, Gomez and Cassini area. A rich riparian forest dominated the borders, but was
2013b) proposed and implemented SDMs as a useful tool for replaced on many islands by commercial forests of Populus
ranking key factors or threats to endangered or vulnerable spp. and Salix spp. (Borodowski and Suárez 2005). Central
species. Most current ranking of threats conducted within areas represented freshwater marshes, with low species
IUCN Specialist Groups still relies on the subjective richness, dominated by graminoid and ‘‘ceibo’’ Erythrina
perspectives of workshop attendees or individual experts. crista-galli forests (Kandus et al. 2003).
SDMs are ideal tools for incorporating theoretical and Otter presence and environmental predictors.—Data on the
mathematical rigor to the ranking threat process, because they presence of L. longicaudis was obtained using 3
are relatively easy and fast to implement, can be used with methodologies: sign surveys (feces, footprints, burrows, and
different levels of knowledge about the species in question, and skins collected by local people), camera traps, and interviews.
are particularly suitable for use at the geographical scale for Field surveys and interviews provided information on the
which the IUCN Red List is designed. distribution of Neotropical river otters in 83 (65 presences/18
The objectives of our study were to build a map of habitat absences) of the 114 cells in which we had divided the study
suitability distribution, and identify possible anthropogenic area. Most signs were obtained during 2 field campaigns
factors that affect the presence of Neotropical river otters in the conducted between December 2000 and March 2011, including
Lower Delta of the Paraná River, which stretches through the 5 locations provided by B. Lartigau. Most areas were
final 300 km of the Paraná Basin, in the Southern Cone of accessible by water, allowing random selection of cells.
South America. We estimated the distribution of the species Thirty-eight camera traps were distributed in 13 cells from
from field surveys and data were analyzed by applying 2008 to 2011 as part of an additional study on the impact of
conventional SDMs. The analysis included anthropogenic forestry on wildlife, so they were distributed at random within
variables related to habitat structure (e.g., polders built or this type of habitat. A total of 3,897 trap days was conducted.
dams for forestry), human presence (density of settlements, We conducted semistructured interviews (Sinzogan et al. 2008;
boat traffic), land use (forestry), and pollution (chemical Anadón et al. 2009) with the inhabitants of the Lower Delta
parameters of water such as nitrates and phosphates). Island. A total of 224 interviews was conducted. They provided
information on several social and environmental factors and on
the presence or absence of otters in the neighborhood. Most
MATERIALS AND METHODS interviews were conducted with people arriving from all parts
Study area.—The Paraná River delta is a large wetland of the Delta to the central port, so they also were a
located in the east-central region of Argentina, which stretches representative sample of the study area.
through the final 300 km of the Paraná Basin, the 2nd largest Following Guisan and Zimmermann (2000), we were
basin in South America (Bonetto 1986). The Lower Delta, the especially careful in the selection of environmental predictors.
terminal area of this delta, covers approximately 17,500 km2 We preferred to use fewer variables with high biological
826 JOURNAL OF MAMMALOGY Vol. 95, No. 4

FIG. 1.—Study area, the Lower Delta of the Paraná River (Bonaerense Delta), and spatial reference points. Note: EEA, Estacion Experimental
Agropecuaria (Agricultural Experiment Station); INTA, National Institute for Agricultural Technology.

significance rather than using a large number of less important resolution) we drew a cross from the vertices of the square
variables. Sources of information varied, with data obtained formed by each cell of the study area. The widths of rivers
from satellite images, interviews, and web pages. We did not and streams were measured every time they were
include any climatic variables in the analysis, because these intersected by the cross. The mean width for each cell
showed minimal variation between cells in the study area. In was incorporated to the analyses.
each of the 114 cells, the following 15 environmental 4. Settlement density (settlements/km2): Georeferenced
predictors were measured: points of buildings and settlements in the study area were
obtained from different sources: human settlement shapes
1. Topography (m above sea level): Obtained from the were obtained from Argentinean Geographic Institute
website of the National Geophysical Data Center ETOPO1 (2011), gazetteer data (DIVA-GIS 2011), satellite imagery
through a global 1-arc-min surface model. from Google Earth (2010–2011 images), interviews with
2. River length (m): The shapes of permanent rivers were local people, and web pages about tourism and recreation
obtained from the Argentinean Geographic Institute in the study area. The number of settlements per cell was
(2011).This shape was provided by Grupo de Sensores divided by cell surface area.
Remotos de la Universidad Nacional de Luján (Argenti- 5. Polder cover (%): In each cell we calculated the area
na). ArcView v.3.2 software (ESRI 2008) was used to occupied by polders and expressed this as a percentage of
generate the buffer, from which meters of rivers and the total area. Polder shape was obtained from Gaute et al.
streams in each cell were estimated. (2007).
3. River width (m): Using a set of satellite imagery from 6. Boat traffic: We counted the number of boats visible to the
Google Earth (Google 2010 and 2011, 500-m eye-height naked eye in each cell in 2 available sets of satellite
August 2014 GOMEZ ET AL.—NEOTROPICAL RIVER OTTER DISTRIBUTION 827

images (years 2010–2011) for the study area, obtained regression (Logit). The first 2 models only use the presence of
from Google Earth with a resolution of 500-m eye height otters as the dependent variable, whereas the 3rd uses a
(height from the surface). This method for measuring boat dichotomous dependent variable of 1s and 0s (presences and
traffic has the advantages of being easy to complete and absences, respectively). First, these models were used to
capable of covering a large area; however, it cannot associate data on presence/absence of otters with the 15
capture temporal variability in boat traffic (for example, an independent variables in the sampled cells, and then used to
increase in tourism during the summer and weekends). obtain the probabilities of occurrence for the 114 cells into
7. Tree cover (%): Vegetation coverage data were obtained which the whole study area was divided. The independent
from DIVA-GIS (2011). Cover of perennial and deciduous variables were transformed into ASCII format for ESRI, for
trees were discriminated, so the type assigned to each cell more fluid management of information, and ASCII files were
was that with the highest percent cover in a 150-m buffer manipulated by the software ArcView 3.2 (ESRI 2008), DIVA-
around permanent rivers. GIS 7.3.0 (DIVA-GIS 2010), and GLOBAL MAPPER 11
8. Forestry (%): Data were obtained from a digital map of the (Blue Marble Geographics 2011).
Delta Forest Plantations produced by the GIS and Forest For MaxEnt analysis we used a Java environment,
Inventory Area of the Ministry of Agriculture, Livestock generating a comma separated values (CVS) file for presence
and Fisheries of Argentina in 2010. We calculated the area data and a package of dependent variables in ASCII format.
occupied by forest plantations in each cell using a 150-m The following settings were used in MaxEnt v.3.3.2 (Phillips et
buffer. This area was transformed into a percentage of the al. 2010) to produce the model: automatic feature selection,
total buffer area in each cell. regularization multiplier at unity, and maximum 500 iterations.
9. Floodplain area (%): Originally resampled to 30 s onto BIOMAPPER software 4.0 (Hirzel et al. 2002) was used for
spatial country-level grid (DIVA-GIS 2011), these data ENFA. A set of independent variables in the denomination of
were reclassified in 2 categories: floodplain area and free IDRISI Raster Format (RST), as well as the dependent variable
land (the remaining area). in a Boolean format (required by the model), were generated by
10. Protected area (%): The protected areas that were partially DIVA-GIS. The variables were standardized and the model
or totally included in the study area were: a UNESCO was developed with the mean as a measure of distance. A
Biosphere Reserve called ‘‘Delta del Paraná’’ designated in distance geometric-mean algorithm was used for habitat
2000 by UNESCO, Botija Island (Organismo Provincial suitability computations (Hirzel and Arlettaz 2003). This
para el Desarrollo Sustentable), and Otamendi National algorithm makes no assumption about the shape of the species
Reserve (Administración de Parques Nacionales). The distribution, but the density of observations must be represen-
percentage of surface occupied by protected areas in each tative of the species niche. The logistic regression was
cell was incorporated in the analysis. For the latter reserve, performed using STATISTICA 8 (StataCorp 2003). To
only two small portions of its area were included within evaluate the role of each environmental variable after the
the Paraná Delta ecoregion (Fig. 2b), although it is regression analyses we used the Wald statistic for all effects in
expected that its management affects the surrounding area. the model, including types I and III of the likelihood ratio test;
11. Suspended solids (mg/l). and Spearman correlations between the environmental vari-
12. Nitrates (mg/l). ables and the 0–1 data on otter distribution.
13. Phosphates (lg/l). Modules were applied after algebraic manipulation of
14. Coliforms (most probable number/100 ml). DIVA-GIS grids to obtain a habitat suitability map. For
15. Dissolved oxygen (mg/l) in waters from rivers and MaxEnt, ENFA, and Logit, we used a random tests percentage
streams.
of 25%. Model fit was evaluated using the area under the curve
Estimates of 5 water-quality measurements (11–15) were (AUC) from the receiver operating characteristic curve (ROC),
obtained from Gomez and Cassini (pers. comm.), who which is the relationship between the sensitivity and the false-
extrapolated to the entire Lower Delta data of previous studies positive fraction (Woodward 1999).
collected in several points of the region. Extrapolation was
done using multiple regressions and a database of these
parameters. Reflectance was obtained from Landsat Thematic RESULTS
Mapper(TM) and Enhanced Thematic Mapper Plus (ETMþ) The values of j r j were less than 0.75 for any pair of
images. comparisons, so we used all variables in the models (Table 1).
To reduce the effects of multicollinearity between environ- Forty one (63%) cells were positive for the presence of this
mental variables we applied the criterion used by Dormann et species. We found a significant overlap between the presences
al. (2012), Syfert et al. (2013), and Kramer-Schadt (2013), that determined by interviews and the presences determined by
variables should be eliminated from the final analysis when camera-trap data (60.42%). When applied to otter training data,
Pearson’s j r j . 0.75. the logit regression model had a better fit to the data according
Species distribution models.—We applied 3 SDMs: to the ROC curve (AUC ¼ 0.89, SD ¼ 0.01) than ENFA (AUC
maximum entropy (MaxEnt), ecological niche factor analysis ¼ 0.44, SD ¼ 0.05) and MaxEnt (AUC ¼ 0.37, SD ¼ 0.08). Due
(ENFA), and a generalized linear model based on a logistic to the low AUC values obtained for ENFA and MaxEnt
828 JOURNAL OF MAMMALOGY Vol. 95, No. 4

FIG. 2.—a) Distribution of the probability of occurrence of otters obtained from logistic regression; b) distribution of the environmental
variables most related to L. longicaudis occurrence in the Lower Delta of the Paraná River. We can also see the Biosphere Reserve and the grid
(0.058 3 0.058) that divided the study area.
August 2014 GOMEZ ET AL.—NEOTROPICAL RIVER OTTER DISTRIBUTION 829

TABLE 1.—Pearson’s correlations j r j between environmental variables. All paired comparisons j r j ,, 0.75. Var ¼ variable; the numbering
corresponds to the methodology.

Var 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 1 0.24 0.14 0.09 0.39 0.19 0.17 0.18 0.24 0.01 0.22 0.19 0.02 0.12 0.00
2 1 0.32 0.33 0.08 0.13 0.38 0.08 0.00 0.17 0.07 0.14 0.05 0.01 0.04
3 1 0.26 0.28 0.17 0.15 0.09 0.05 0.17 0.09 0.28 0.13 0.03 0.35
4 1 0.20 0.28 0.33 0.34 0.06 0.14 0.06 0.23 0.11 0.10 0.07
5 1 0.50 0.35 0.13 0.20 0.14 0.52 0.19 0.13 0.15 0.30
6 1 0.29 0.07 0.03 0.04 0.08 0.18 0.15 0.01 0.30
7 1 0.08 0.14 0.16 0.09 0.08 0.06 0.11 0.18
8 1 0.21 0.17 0.01 0.03 0.06 0.01 0.04
9 1 0.28 0.25 0.08 0.08 0.16 0.05
10 1 0.23 0.24 0.14 0.01 0.03
11 1 0.02 0.22 0.22 0.19
12 1 0.06 0.07 0.21
13 1 0.35 0.08
14 1 0.30
15 1

models, we only used results provided by the logit regression large rivers. There were 2 very wide rivers that were positive
analysis. for presence of the species. When these 2 points were removed
The map of habitat suitabilities for the Lower Delta of the the statistical significance disappeared.
Paraná River showed 2 trends: the central zone and those areas A simple additional statistical analysis was conducted by
closest to the northern border showed the highest probabilities applying Spearman rank correlation analyses, which showed
of occurrence (Fig. 2a). that polders, settlements, and boat traffic were significantly
Taking into account the contribution of each variable correlated with presence/absence of otters (Table 2). Habitat
separately, the regression analysis indicated that polder cover suitability for L. longicaudis increased with polder cover and
was the only significant variable, according to a test of all decreased with human settlements (Fig. 3). These trends can be
effects (Table 2), whereas river width and settlement density established visually by comparing the distribution of habitat
were the other significant variables according to a likelihood suitabilities (Fig. 2a) with the distribution of environmental
variables (Fig. 2b).
type I test. The effect of river width appeared to be an artifact
There were no significant relationships between biodiversity
of an unavoidable skew in the amount of data for small and
reserve distribution and the presence of otters. However, visual
TABLE 2.—Multiple regressions and Spearman correlations between
environmental variables and the distribution of Lontra longicaudis in
the Lower Paraná River delta. Significant statistics (P , 0.05) appear
in italics.

Type I Type III

test test
All Chi- Chi-
Environmental effects Log- square Log- square Spearman
variable Walda likelihood testa likelihood testa rank
Intercept 0.10 38.01
1. Topography 1.71 29.03 0.34 26.43 2.10 0.08
2. River length 0.83 30.21 0.04 25.82 0.87 0.00
3. River width 0.31 35.65 4.73 25.60 0.43 0.13
4. Settlement density 1.57 33.24 4.82 26.38 2.00 0.30
5. Polder cover 3.90 30.68 4.41 27.91 5.06 0.30
6. Boat traffic 0.15 29.97 0.49 25.45 0.15 0.37
7. Tree cover 0.85 32.88 0.71 25.82 0.88 0.211
8. Forestry 1.61 30.23 0.88 26.29 1.82 0.14
9. Floodplain area 0.11 29.51 0.32 25.44 0.12 0.13
10. Protected area 0.53 29.20 0.63 25.74 0.73 0.08
11. Suspended solids 0.39 29.67 0.59 25.58 0.40 0.10
12. Nitrates 1.91 28.00 2.05 26.39 2.03 0.25
13. Phosphates 2.76 26.77 2.46 26.92 3.08 0.14
14. Coliforms 2.14 25.38 2.39 26.58 2.39 0.19
FIG. 3.—Response curve of the probability distribution of L.
15. Dissolved oxygen 0.00 26.58 0.40 25.38 0.00 0.03
longicaudis as a function of each of the most significant independent
a
For all variables, d.f. ¼ 1 variables.
830 JOURNAL OF MAMMALOGY Vol. 95, No. 4

inspection of Fig. 2 suggests that habitat suitability tended to ized by cyclical or unpredictable floods (Junk et al. 1989).
increase in some protected areas. In the largest, the UNESCO However, more generalist species that are capable of using a
Biosphere reserve, the buffer zone and a transition area showed diverse range of ecosystems are expected to prefer more stable
low probability of use by otters (Fig. 2b). aquatic habitats. Like most freshwater otters, L. longicaudis is
a semiaquatic carnivore that constructs burrows and has a
social organization based on territorial defense, whose
DISCUSSION territories are scent marked (Kruuk 2006). Floods are likely
We applied 3 species distribution models to the distribution to modify otter behavior by eliminating marks, flooding
of the Neotropical river otter (L. longicaudis) in the Buenos burrows, and causing major disturbance in the environment.
Aires Lower Delta of the Paraná River. These models have 4 For example, on the river banks of the Atlantic forest of
advantages compared with just mapping existing results, as southeastern Brazil, Pardini and Trajano (1999) found that
they allow association of the dependent variables with many these otters more frequently used shelters in higher locations,
environmental factors simultaneously, identify the most probably because they were less vulnerable to flooding. It is
relevant environmental factors affecting the distribution, permit likely that protection provided by polders minimizes these
extrapolation of the probabilities of occurrences to nonsampled effects.
areas, and provide maps that can be used in the design of action The 2nd potential benefit of polders may be reduced
plans for wildlife conservation. The distribution of otters in the poaching. This illegal activity is by far the most important
Paraná Delta was related mainly to 2 types of anthropogenic threat to marsh deer (D’Alessio et al. 2002). Control of hunting
influences on the environment, polders built mainly for appears to be effective only in the nucleus of protected areas.
forestry, and human presence/settlements. Habitat suitability Local people hunt regularly and other hunters enter the Paraná
increased with surface cover by polders. In the Lower Delta of Delta from the large cities located on the southern border of the
the Paraná River, D’Alessio et al. (2002) and Fracassi and reserve. In our interviews, 95 local people informed us that
Somma (2010) found that marsh deer, Blastocerus dichotomus, there was hunting in the neighborhood (81% positive response
were frequently found in polders; they attributed this finding to for poaching), whereas only 22 indicated the contrary. In this
the protection provided by this anthropogenic habitat against context, access control exerted by the owners of private land
flood and poaching. Similarly, maximal bird diversity was used for forestry probably acts as an effective limitation for
found in abandoned forestations partially reinvaded by native poaching in the Paraná Delta. Hunting probably also explains
vegetation of natural or artificial polders (Quintana 1999; why L. longicaudis habitat suitability decreases closer to
Quintana et al. 2002). These authors suggested that landscapes human settlements, considering that most otter species are
with polders are composed of a diverse mosaic of habitats tolerant of human presence and to a considerable degree of
(different shapes, species, and plantation ages; different degradation of habitat structure (Kruuk 2006).
production systems and forestry channels), which permit a The 3rd hypothesis postulates that forestry increases habitat
large quantity of species to use the islands. Malzoff et al. connectivity because of the many channels built for water
(2012) found that abundance of guans, Penelope obscura, is management (Taylor et al. 1993). In the only previous study
mainly related to the available area of secondary woodland, but conducted on L. longicaudis in the Paraná Delta, Cabrera
to a lesser extent also to the presence of mature Salix (2006) found that these otters can make extensive use of
forestations, where this emblematic species of the Delta region channels, especially in autumn and winter. The importance of
would find more food and shelter than in other types of connectivity for otter population structure has been highlighted
landscapes. in previous studies using genetic techniques. For example,
There are at least 3 possible explanations for a positive Latch et al. (2008) found genetic discontinuities in North
association between otter presence and polders: areas with American river otters L. canadensis in a region of Louisiana
polders might provide more stable environments due to that were interpreted as a result of limited dispersal (Latch et al.
protection against flooding than some natural areas of the 2008). The opposite was found with L. longicaudis in the
delta; they are better protected from poaching; or they increase Lacantun River system of Chiapas, Mexico, where high levels
landscape connectivity. The functioning and structure of the of landscape connectivity were estimated using similar genetic
lower Paraná River delta are influenced by periodic flooding techniques (Ortega et al. 2012).
originating from the northwest through the high discharge rates It is critical to realize that the benefits obtained in altered
of the Paraná River, and rising tides and winds from the habitats, such as those associated with an increase in stability,
southeast (‘‘sudestadas’’—Kandus et al. 2011). These inputs of habitat diversity, or protection against poaching, are short-term
water, sediments, and organisms create flood pulses (Junk et al. benefits. Forestry activity causes significant habitat alterations
1989), which vary according to climate processes defined by not only because it changes the composition of the plant
the general circulation of the atmosphere and anomalies, such community, but also the geomorphology of the islands and the
as the El Niño–Southern Oscillation. The effect of these pulses dynamics of flooding (Kandus et al. 2011). In the long run,
on the biology of wildlife varies among taxonomic groups, and changes in the dynamics of these wetlands damage the entire
many of the organisms that colonize river floodplains, deltas, native biota. Recently Fracassi et al. (2013) have proposed
and estuaries are adapted to a hydrological regime character- biodiversity conservation strategies for Delta forest plantations
August 2014 GOMEZ ET AL.—NEOTROPICAL RIVER OTTER DISTRIBUTION 831

that will allow sustainable management of forestry without any identificar posibles factores antrópicos que influyen en la
significant impact on water quality and biodiversity. presencia de L. longicaudis en el Bajo Delta del Rı́o Paraná
The results for protected areas deserve attention, even en el Cono Sur de América del Sur. Se obtuvo la presencia /
though no statistically significant association with habitat ausencia de L. longicaudis utilizando tres métodos (se midieron
suitability for otters was found. There were areas with legal las encuestas signos, trampas de cámaras y entrevistas), y 15
protection that matched high suitability zones, such as the predictores ambientales. La idoneidad del hábitat fue mayor en
Botija. Similarly, those cells with maximal habitat suitability las zonas con diques construidos para la silvicultura y baja en las
along the southern border of the Lower Delta were those within zonas con asentamientos humanos y tráfico de embarcaciones.
or close to the Otamendi Reserve, suggesting that this protected En la actualidad, el aislamiento geográfico y el control en
area has a positive influence on biodiversity conservation in the terrenos privados y las reservas parecen ser eficaces en la
nearby portions of the Delta. As most UNESCO Biosphere protección de la fauna en el delta del Paraná. Este estudio
reserves, it consists of 3 zones with different levels of muestra que los modelos de distribución de especies se pueden
protection. There was only 1 guard for 88,625 ha, who utilizar para evaluar rápidamente las posibles amenazas a la vida
concentrated his vigilance on the core sector (pers. comm.), silvestre.
which in turn is the only 1 of the 3 zones with high suitability
for otters. These examples demonstrate the importance for the ACKNOWLEDGMENTS
future of the Paraná Delta to have an effective management of
legally established protected areas. This work was supported by grants from Consejo Nacional de
We incorporated potential pollutants as environmental Investigaciones Cientı́ficas y Técnicas (CONICET, Argentina, and PIP
0367 Res. 845 /1), the University of Luján (Fondos Finalidad 3.5), and
variables because the study area is close to highly contaminated
the National Institute of Agricultural Technology. MHC and JIT are
areas (such as the city of Buenos Aires), and otters are known
researchers of the CONICET. B. Lartigau kindly provided data on 5
to be affected by pollution (Kruuk 2006). However, we did not locations of otters. M. Gauté provided the information on the polders.
find a correlation between chemical composition of water and
otter distribution. Because we conducted a gross analysis using
occurrence data and at relatively large scale, we cannot rule out LITERATURE CITED
the possibility that the effect of pollution occurs at a finer scale ANADÓN, J. D., A. GIMÉNEZ, R., BALLESTAR AND I. PÉREZ. 2009.
or affects other aspects of otter biology, such as diseases or Evaluation of local ecological knowledge as a method for collecting
population density. Unfortunately, there were no data on fish extensive data on animal abundance. Conservation Biology 23:617–
62.
distribution in the study area; thus we were unable to test for
ARGENTINEAN GEOGRAPHIC INSTITUTE. 2011. http://ign.gob.ar/. Ac-
the effect of food availability on L. longicaudis.
cessed 20 December 2010.
Our study demonstrates that SDMs can be used as an
BLACHER, C. 1987. Ocorrência e preservação de Lutra longicaudis no
effective method for rapidly evaluating potential threats to litoral de Santa Catarina. Boletim FBCN 22:105–117.
wildlife: habitat suitability for otters was higher in areas with BLUE MARBLE GEOGRAPHICS. 2011. Global Mapper 11. http://www.
polders built for forestry and lower in areas with human bluemarblegeo.com/products/global-mapper.php. Accessed 1 Feb-
settlements. Typical conservation biology models such as ruary 2010.
population viability analyses (PVA) are robust tools but require BONETTO, A. 1986. The Paraná River system. Pp. 541–544 in The
large amounts of data that are normally difficult or costly to ecology of river systems (B. R. Davis and K. F. Walker, eds.). Junk
collect, particularly in countries where financial and institu- Publication, Dordrecht, The Netherlands.
tional resources are limited. Additionally, PVA are normally BORODOWSKI, F., AND R. SUÁREZ. 2005. Caracterización forestal de la
applicable to small scales, but conservation strategies must be región del Delta del Paraná. SAGPyA, Buenos Aires, Argentina.
applied frequently to large scales. SDMs are correlative models BRINSON, M. M. 1993. Changes in the functioning of wetlands along
that are rarely used to test hypotheses or explain causations, environmental gradients. Wetlands 13:65–74.
and are normally used with limited data collected unsystem- BURKART, A. 1957. Ojeada sinóptica sobre la vegetación del Delta del
Rı́o Paraná. Darwiniana 11:457–561.
atically. However, they can be especially useful as exploratory
CABRERA, S. 2006. Uso y selección de hábitat del lobito de rı́o, Lontra
techniques, to generate scenarios and objective hypotheses that
longicaudis, en una forestación del Bajo Delta del Paraná. Tesis de
can guide regulatory agencies to identify main sources of licenciatura, Universidad de Buenos Aires Facultad de Ciencia
threats to wildlife. Exactas y Naturales, Argentina.
CASSINI, M. H. 2011a. Ecological principles of species distribution
RESUMEN models: the habitat matching rule. Journal of Biogeography
38:2057–2065.
El lobito de rı́o Lontra longicaudis es un depredador acuático CASSINI, M. H. 2011b. Ranking threats using species distribution
de muchos sistemas de agua dulce de América Central y del Sur. models in the IUCN Red List assessment process. Biodiversity and
Su categorı́a actual en la Unión Internacional para la Conservation 20:3689–3692.
Conservación de la Naturaleza es ‘‘datos insuficientes’’, lo que CASSINI M. H. 2013. Distribution ecology—from individual habitat
hace imprescindible determinar el estado de conservación use to species biogeographical range. Springer, New York.
adecuado. Se aplicaron modelos de distribución de especies CHEMES, S. B., A. R. GIRAUDO, AND G. GIL. 2010. Dieta de Lontra
para construir un mapa de idoneidades de hábitat, y para longicaudis (Carnivora, Mustelidae) en el Parque Nacional El Rey
832 JOURNAL OF MAMMALOGY Vol. 95, No. 4

(Salta, Argentina) y su comparación con otras poblaciones de la HIRZEL, A. H., AND R. ARLETTAZ. 2003. Modelling habitat suitability
cuenca del Paraná. Mastozoologı́a Neotropical 17:19–29. for complex species distributions by the environmental-distance
D’ALESSIO, S., D. VARELA, B. LARTIGAU, F. GAGLIARDI, G. APRILE, AND geometric mean. Environmental Management 32: 614–623.
C. MONACO. 2002. Informe final primera etapa proyecto ciervo de HIRZEL, A. H., J. HAUSSER, D. CHESSEL, AND N. PERRIN. 2002.
los Pantanos. Asociación para la Conservación y el Estudio de la Ecological-niche factor analysis: How to compute habitat-suitability
Naturaleza, Buenos Aires, Argentina. maps without absence data? Ecology 83:2027–2036. http://www2.
DAVIS, R. A., AND D. M. FITZGERALD. 2004. Beaches and coasts. unil.ch/biomapper/products.html. Accessed 12 March 2010.
Blackwell Publishing, Oxford, United Kingdom. JACKSON, C. R., AND M. P. ROBERTSON. 2011. Predicting the potential
DIVA-GIS. 2010. DIVA-GIS 7.3.0. http://www.diva-gis.org/. Accessed distribution of an endangered cryptic subterranean mammal from
23 February 2010. few occurrence records. Journal for Nature Conservation 19:87–94.
DIVA-GIS. 2011. Free spatial data. http://www.diva-gis.org/Data. JUNK, W. J., P. B. BAYLEY, AND R. E. SPARKS. 1989. The flood pulse
Accessed 25 February 2011. concept in river floodplain systems. Journal of Fisheries and
DORMANN C. F., ET AL. 2012. Collinearity: a review of methods to deal Aquatic Sciences 106:110–127.
with it and a simulation study evaluating their performance. KANDUS, P., A. I. MALVÁREZ, AND N. MADANES. 2003. Estudio de las
Ecography 35:1–20. comunidades de plantas herbáceas de las islas bonaerenses del Bajo
ELITH, J., AND J. R. LEATHWICK. 2009. Species distribution models: Delta del Rı́o Paraná (Argentina). Darwiniana Nueva Serie 41:1–16.
ecological explanation and prediction across space and time. KANDUS, P., P. MINOTTI, AND M. BORRO. 2011. Contribuciones al
Annual Review of Ecology, Evolution, and Systematics 40:677– conocimiento de los humedales del Delta del Rı́o Paraná. San
697. Martı́n: Universidad Nacional de Gral. San Martı́n. UNSAM,
EMMONS, L., AND F. FEER. 1997. Neotropical rainforest mammals: a Argentina.
field guide. University of Chicago Press, Chicago, Illinois. KRAMER-SCHADT. 2013. The importance of correcting for sampling
ERWIN, K. L. 2009. Wetlands and global climate change: the role of bias in maxent species distribution models. Diversity and
wetland restoration in a changing world. Wetlands Ecology and Distributions 19:1366–1379.
KRUUK, H. 2006. Otters: ecology, behaviour and conservation. Oxford
Management 17:71–84.
University Press, Oxford, United Kingdom.
ESRI. 2008. ArcView 3.2. http://www.esri.com/. Accessed 14 March
LATCH, E. K., D. G. SCOGNAMILLO, J. A. FIKE, M. J. CHAMBERLAIN, AND
2008.
O. E. RHODES. 2008. Deciphering ecological barriers to North
FRACASSI, N., R. QUINTANA, J. PEREIRA AND G. MUJICA. 2013. Protocolo
American river otter (Lontra canadensis) gene flow in the Lousiana
para la definición de Estrategias de Conservación de la Biodiversi-
landscape. Journal of Heredity 99:265–274.
dad en plantaciones forestales de Salicáceas del Bajo Delta del
MALVÁREZ, A. I. 1997. Las comunidades vegetales del Delta del rı́o
Paraná. Ediciones INTA.
Paraná. Su relación con factores ambientales y patrones de paisaje.
FRACASSI, N., AND D. SOMMA. 2010. Participatory action research
Tesis doctoral. Universidad de Buenos Aires, Buenos Aires,
concerning the landscape use by a native cervid in a wetland of the
Argentina.
Plata Basin, Argentina. IUFRO Landscape Ecology International
MALZOF, S. L., M. L. BOLKOVIC, J. J. THOMPSON, AND R. D. QUINTANA.
Conference, Bragança, Portugal.
2012. Habitat occupancy of the Dusky-legged Guan in the lower
GALLO–REYNOSO, J. P. 1989. Distribución y estado actual de la nutria o delta of the Paraná River, Argentina. Bird Conservation Interna-
perro de agua (Lutra longicaudis annectens, Major, 1897) en la tional 23:83–90.
Sierra Madre del Sur, México. Tesis de maestrı́a, Facultad de MORUETA-HOLME, N., C. FLØJGAARD, AND K. C. SVENNING. 2010.
Ciencias, Universidad Nacional Autónoma de México, México, Climate change risks and conservation implications for a threatened
D.F. small-range mammal species. PLoS One 5:e10360.
GAUTE, M., M. NICOLÁS, E. BORODOWSKI, AND C. DI BELLA. 2007. ORTEGA, J., D. NAVARRETE, AND J. E. MALDONADO. 2012. Non–invasive
Elaboración de un sistema de información geográfica para el sampling of endangered neotropical river otters reveals high levels
monitoreo de pólderes en el Bajo Delta del Paraná, durante el of dispersion in the Lacantun River System of Chiapas, Mexico.
perı́odo 1985–2005. in Teledetección-Hacia un mejor entendimien- Animal Biodiversity and Conservation 35:59–69.
to de la dinámica global y regional (G. Martı́n, ed.). PAPES , M., AND P. GAUBERT. 2007. Modelling ecological niches from
GOMEZ, J. J. AND M. H. CASSINI. 2013b. Analysis of environmental low numbers of occurrences: assessment of the conservation status
correlates of sexual segregation in northern elephant seals using of poorly known viverrids (Mammalia, Carnivora) across two
species distribution models. Marine Biology 161(2):481–487. continents. Diversity and Distributions 13:890–902.
GOOGLE. 2010. Google Earth. http://earth.google.com/. Accessed 20 PARDINI, R., AND E. TRAJANO. 1999. Use of shelters by the Neotropical
December 2010. river otter (Lontra longicaudis) in an Atlantic forest stream,
GOOGLE. 2011. Google Earth. http://earth.google.com/. Accessed 30 southeastern Brazil. Journal of Mammalogy 80:600–610.
December 2011. PARERA, A. 1996. Estimating river otter Lutra longicaudis population
GUISAN, A., AND W. THUILLER. 2005. Predicting species distribution: in Iberá lagoon using a direct sightings methodology. IUCN Otter
offering more than simple habitat models. Ecology Letters 8:993– Specialist Group Bulletin 13:77–83.
1009. PHILLIPS, S. J., M. DUDÍK, AND R. E. SCHAPIRE. 2010. MaxEnt v.3.3.2.
GUISAN, A., AND N. E. ZIMMERMANN. 2000. Predictive habitat http://www.cs.princeton.edu/~schapire/maxent/. Accessed 01 Feb-
distribution models in ecology. Ecological Modelling 135:147–186. ruary 2010.
HEIKKINEN, R. K., M. LUOTO, M. B. ARAÚJO, R. VIRKKALA, W. QUADROS, J., AND E. L. MONTEIRO-FILHO. 2001. Diet of the neotropical
THUILLER, AND M. T. SYKES. 2006. Methods and uncertainties in otter, Lontra longicaudis, in an Atlantic forest area, Santa Catarina
bioclimatic envelope modelling under climate change. Progress in State, southern Brazil. Studies on Neotropical Fauna and Environ-
Physical Geography 30:751–777. ment 36:15–21.
August 2014 GOMEZ ET AL.—NEOTROPICAL RIVER OTTER DISTRIBUTION 833

QUINTANA, R. D. 1999. Relación entre la estructura del paisaje en un TAYLOR, P. D., L. FAHRIG, K. HENEIN, AND G. MERRIAM. 1993.
humedal y la fauna silvestre: el carpincho (Hydrochaeris hydro- Connectivity is a vital element of landscape structure. Oikos
chaeris) como caso de estudio. Pp. 185–204 in Tópicos sobre 68:571–572.
humedales subtropicales y templados de Sudamérica (A. I. TRINCA, C. S., C. F. JAEGER, AND E. EIZIRIK. 2013. Molecular ecology
Malvárez, ed.). ORCyT – MAB/ UNESCO, Montevideo, Uruguay. of the Neotropical otter (Lontra longicaudis): non-invasive
QUINTANA, R. D., R. BÓ, AND F. KALESNIK. 2002. La vegetación y la sampling yields insights into local population dynamics. Biological
fauna silvestre de la porción terminal de la Cuenca del Plata. Journal of the Linnean Society 109:932–948.
Consideraciones biogeográficas y ecológicas. Pp. 99–124 in El Rı́o UNESCO. 2000. Décima reunión del Consejo Internacional de
de la Plata como territorio (J. M. Borthagaray, ed.). FADU, Coordinación del MAB. Parı́s, France.
FURBAN e Infinito, Bs. As., Argentina. WALDEMARIN, H. F., AND R. ALVAREZ. 2008. Lontra longicaudis. In
SEPÚLVEDA, M. A., ET AL. 2009. Landscape features and crustacean
IUCN 2013. IUCN red list of threatened species. Version 2013.1.
prey as predictors of the Southern river otter distribution in Chile.
www.iucnredlist.org. Accessed 9 July 2013.
Animal Conservation 12:522–530.
WILTING, A., ET AL. 2010. Modelling the species distribution of flat-
SINZOGAN, A. A. C., P. VAN MELE, AND J. F. VAYSSIERES. 2008.
headed cats (Prionailurus planiceps), an endangered South-East
Implications of on-farm research for local knowledge related to fruit
flies and the weaver ant Oecophylla longinoda in mango Asian small felid. PloS one 5:e9612.
production. International Journal of Pest Management 54:241–246. WOODWARD, M. 1999. Epidemiology: study design and data analysis.
STATACORP. 2003. STATISTICA 8. StataCorp LP, College Station, Chapman & Hall, London, United Kingdom.
Texas.
SYFERT, M. M., M. J. SMITH, AND D. A. COOMES. 2013. The effects of Submitted 18 October 2013. Accepted 10 March 2014.
sampling bias and model complexity on the predictive performance
of MaxEnt species distribution models. PLoS ONE 8:e55158. Associate Editor was Marcus V. Vieira.